diff --git "a/metrics.json" "b/metrics.json"
new file mode 100644--- /dev/null
+++ "b/metrics.json"
@@ -0,0 +1,14452 @@
+{
+  "train_runtime": 15528.6205,
+  "train_samples_per_second": 0.103,
+  "train_steps_per_second": 0.103,
+  "train_loss": 0.19463951266836374,
+  "epoch": 1.0,
+  "config": {
+    "vein_name": "html_r64_v2",
+    "start_step": 600,
+    "end_step": 2200,
+    "batch_size": 1,
+    "grad_accum": 1,
+    "lora_r": 64,
+    "lora_alpha": 64,
+    "warmup_steps": 10,
+    "rs_lora": false,
+    "learning_rate": 0.0002,
+    "max_length": 23401,
+    "dataset_path": "dataset.jsonl",
+    "dataset_index_start": 1618,
+    "dataset_index_end": 1436,
+    "total_samples_processed": 5000
+  },
+  "gpu_memory": {
+    "peak_reserved_gb": 19.139
+  },
+  "step_by_step_metrics": [
+    {
+      "step": 601,
+      "timestamp": "2025-12-28T08:58:29.750718",
+      "elapsed_time": 24.035809755325317,
+      "loss": 0.2155,
+      "grad_norm": 0.1604655683040619,
+      "learning_rate": 0.0,
+      "epoch": 0.000625
+    },
+    {
+      "step": 602,
+      "timestamp": "2025-12-28T08:58:36.494748",
+      "elapsed_time": 30.779838800430298,
+      "loss": 0.152,
+      "grad_norm": 0.1441534161567688,
+      "learning_rate": 2e-05,
+      "epoch": 0.00125
+    },
+    {
+      "step": 603,
+      "timestamp": "2025-12-28T08:58:46.390445",
+      "elapsed_time": 40.67553472518921,
+      "loss": 0.1768,
+      "grad_norm": 0.14501173794269562,
+      "learning_rate": 4e-05,
+      "epoch": 0.001875
+    },
+    {
+      "step": 604,
+      "timestamp": "2025-12-28T08:58:52.949267",
+      "elapsed_time": 47.234357595443726,
+      "loss": 0.3154,
+      "grad_norm": 0.20344674587249756,
+      "learning_rate": 6e-05,
+      "epoch": 0.0025
+    },
+    {
+      "step": 605,
+      "timestamp": "2025-12-28T08:59:02.976637",
+      "elapsed_time": 57.26172733306885,
+      "loss": 0.1981,
+      "grad_norm": 0.1389472484588623,
+      "learning_rate": 8e-05,
+      "epoch": 0.003125
+    },
+    {
+      "step": 606,
+      "timestamp": "2025-12-28T08:59:14.217365",
+      "elapsed_time": 68.50245547294617,
+      "loss": 0.1454,
+      "grad_norm": 0.12427599728107452,
+      "learning_rate": 0.0001,
+      "epoch": 0.00375
+    },
+    {
+      "step": 607,
+      "timestamp": "2025-12-28T08:59:19.569352",
+      "elapsed_time": 73.85444259643555,
+      "loss": 0.3674,
+      "grad_norm": 0.21498622000217438,
+      "learning_rate": 0.00012,
+      "epoch": 0.004375
+    },
+    {
+      "step": 608,
+      "timestamp": "2025-12-28T08:59:32.803346",
+      "elapsed_time": 87.08843612670898,
+      "loss": 0.1358,
+      "grad_norm": 0.1065230667591095,
+      "learning_rate": 0.00014,
+      "epoch": 0.005
+    },
+    {
+      "step": 609,
+      "timestamp": "2025-12-28T08:59:43.029459",
+      "elapsed_time": 97.31454968452454,
+      "loss": 0.1995,
+      "grad_norm": 0.19915856420993805,
+      "learning_rate": 0.00016,
+      "epoch": 0.005625
+    },
+    {
+      "step": 610,
+      "timestamp": "2025-12-28T08:59:47.972807",
+      "elapsed_time": 102.25789737701416,
+      "loss": 0.2193,
+      "grad_norm": 0.18390510976314545,
+      "learning_rate": 0.00018,
+      "epoch": 0.00625
+    },
+    {
+      "step": 611,
+      "timestamp": "2025-12-28T08:59:56.025895",
+      "elapsed_time": 110.31098580360413,
+      "loss": 0.1661,
+      "grad_norm": 0.14340169727802277,
+      "learning_rate": 0.0002,
+      "epoch": 0.006875
+    },
+    {
+      "step": 612,
+      "timestamp": "2025-12-28T09:00:04.192186",
+      "elapsed_time": 118.47727656364441,
+      "loss": 0.1867,
+      "grad_norm": 0.17027480900287628,
+      "learning_rate": 0.000199874213836478,
+      "epoch": 0.0075
+    },
+    {
+      "step": 613,
+      "timestamp": "2025-12-28T09:00:15.363335",
+      "elapsed_time": 129.64842581748962,
+      "loss": 0.1322,
+      "grad_norm": 0.12804803252220154,
+      "learning_rate": 0.000199748427672956,
+      "epoch": 0.008125
+    },
+    {
+      "step": 614,
+      "timestamp": "2025-12-28T09:00:22.801954",
+      "elapsed_time": 137.0870442390442,
+      "loss": 0.3998,
+      "grad_norm": 0.2402307242155075,
+      "learning_rate": 0.00019962264150943397,
+      "epoch": 0.00875
+    },
+    {
+      "step": 615,
+      "timestamp": "2025-12-28T09:00:41.397275",
+      "elapsed_time": 155.68236541748047,
+      "loss": 0.167,
+      "grad_norm": 0.12692536413669586,
+      "learning_rate": 0.00019949685534591195,
+      "epoch": 0.009375
+    },
+    {
+      "step": 616,
+      "timestamp": "2025-12-28T09:00:57.925605",
+      "elapsed_time": 172.2106957435608,
+      "loss": 0.1226,
+      "grad_norm": 0.1139947846531868,
+      "learning_rate": 0.00019937106918238996,
+      "epoch": 0.01
+    },
+    {
+      "step": 617,
+      "timestamp": "2025-12-28T09:01:03.902990",
+      "elapsed_time": 178.1880807876587,
+      "loss": 0.2534,
+      "grad_norm": 0.2593400478363037,
+      "learning_rate": 0.00019924528301886794,
+      "epoch": 0.010625
+    },
+    {
+      "step": 618,
+      "timestamp": "2025-12-28T09:01:19.300000",
+      "elapsed_time": 193.58509039878845,
+      "loss": 0.1422,
+      "grad_norm": 0.12689131498336792,
+      "learning_rate": 0.00019911949685534592,
+      "epoch": 0.01125
+    },
+    {
+      "step": 619,
+      "timestamp": "2025-12-28T09:01:25.737299",
+      "elapsed_time": 200.0223891735077,
+      "loss": 0.1783,
+      "grad_norm": 0.16301578283309937,
+      "learning_rate": 0.0001989937106918239,
+      "epoch": 0.011875
+    },
+    {
+      "step": 620,
+      "timestamp": "2025-12-28T09:01:38.363267",
+      "elapsed_time": 212.64835739135742,
+      "loss": 0.1295,
+      "grad_norm": 0.13712646067142487,
+      "learning_rate": 0.0001988679245283019,
+      "epoch": 0.0125
+    },
+    {
+      "step": 621,
+      "timestamp": "2025-12-28T09:01:43.657825",
+      "elapsed_time": 217.9429154396057,
+      "loss": 0.2815,
+      "grad_norm": 0.21860435605049133,
+      "learning_rate": 0.00019874213836477988,
+      "epoch": 0.013125
+    },
+    {
+      "step": 622,
+      "timestamp": "2025-12-28T09:01:52.793187",
+      "elapsed_time": 227.07827711105347,
+      "loss": 0.3943,
+      "grad_norm": 0.20962341129779816,
+      "learning_rate": 0.00019861635220125786,
+      "epoch": 0.01375
+    },
+    {
+      "step": 623,
+      "timestamp": "2025-12-28T09:02:00.534713",
+      "elapsed_time": 234.81980347633362,
+      "loss": 0.3958,
+      "grad_norm": 0.17121204733848572,
+      "learning_rate": 0.00019849056603773587,
+      "epoch": 0.014375
+    },
+    {
+      "step": 624,
+      "timestamp": "2025-12-28T09:02:09.703390",
+      "elapsed_time": 243.98848390579224,
+      "loss": 0.2354,
+      "grad_norm": 0.16160890460014343,
+      "learning_rate": 0.00019836477987421385,
+      "epoch": 0.015
+    },
+    {
+      "step": 625,
+      "timestamp": "2025-12-28T09:02:15.193169",
+      "elapsed_time": 249.47825932502747,
+      "loss": 0.2254,
+      "grad_norm": 0.2050420343875885,
+      "learning_rate": 0.00019823899371069183,
+      "epoch": 0.015625
+    },
+    {
+      "step": 626,
+      "timestamp": "2025-12-28T09:02:23.621480",
+      "elapsed_time": 257.9065706729889,
+      "loss": 0.3533,
+      "grad_norm": 0.16951978206634521,
+      "learning_rate": 0.00019811320754716983,
+      "epoch": 0.01625
+    },
+    {
+      "step": 627,
+      "timestamp": "2025-12-28T09:02:40.925797",
+      "elapsed_time": 275.21088790893555,
+      "loss": 0.1368,
+      "grad_norm": 0.12023195624351501,
+      "learning_rate": 0.0001979874213836478,
+      "epoch": 0.016875
+    },
+    {
+      "step": 628,
+      "timestamp": "2025-12-28T09:02:47.264513",
+      "elapsed_time": 281.54960775375366,
+      "loss": 0.1796,
+      "grad_norm": 0.3685634434223175,
+      "learning_rate": 0.0001978616352201258,
+      "epoch": 0.0175
+    },
+    {
+      "step": 629,
+      "timestamp": "2025-12-28T09:02:54.630161",
+      "elapsed_time": 288.9152555465698,
+      "loss": 0.1811,
+      "grad_norm": 0.1793571263551712,
+      "learning_rate": 0.0001977358490566038,
+      "epoch": 0.018125
+    },
+    {
+      "step": 630,
+      "timestamp": "2025-12-28T09:03:01.180041",
+      "elapsed_time": 295.4651312828064,
+      "loss": 0.2804,
+      "grad_norm": 0.23330993950366974,
+      "learning_rate": 0.00019761006289308177,
+      "epoch": 0.01875
+    },
+    {
+      "step": 631,
+      "timestamp": "2025-12-28T09:03:11.837408",
+      "elapsed_time": 306.12249875068665,
+      "loss": 0.2404,
+      "grad_norm": 0.20730850100517273,
+      "learning_rate": 0.00019748427672955975,
+      "epoch": 0.019375
+    },
+    {
+      "step": 632,
+      "timestamp": "2025-12-28T09:03:16.778563",
+      "elapsed_time": 311.0636534690857,
+      "loss": 0.3396,
+      "grad_norm": 0.22279325127601624,
+      "learning_rate": 0.00019735849056603773,
+      "epoch": 0.02
+    },
+    {
+      "step": 633,
+      "timestamp": "2025-12-28T09:03:25.959314",
+      "elapsed_time": 320.2444043159485,
+      "loss": 0.161,
+      "grad_norm": 0.17951494455337524,
+      "learning_rate": 0.00019723270440251574,
+      "epoch": 0.020625
+    },
+    {
+      "step": 634,
+      "timestamp": "2025-12-28T09:03:42.789283",
+      "elapsed_time": 337.07437324523926,
+      "loss": 0.1691,
+      "grad_norm": 0.12446315586566925,
+      "learning_rate": 0.00019710691823899372,
+      "epoch": 0.02125
+    },
+    {
+      "step": 635,
+      "timestamp": "2025-12-28T09:03:54.312342",
+      "elapsed_time": 348.5974371433258,
+      "loss": 0.1439,
+      "grad_norm": 0.19256795942783356,
+      "learning_rate": 0.0001969811320754717,
+      "epoch": 0.021875
+    },
+    {
+      "step": 636,
+      "timestamp": "2025-12-28T09:04:03.177467",
+      "elapsed_time": 357.46255707740784,
+      "loss": 0.2008,
+      "grad_norm": 0.13904313743114471,
+      "learning_rate": 0.0001968553459119497,
+      "epoch": 0.0225
+    },
+    {
+      "step": 637,
+      "timestamp": "2025-12-28T09:04:10.990148",
+      "elapsed_time": 365.2752380371094,
+      "loss": 0.1878,
+      "grad_norm": 0.1624336838722229,
+      "learning_rate": 0.00019672955974842768,
+      "epoch": 0.023125
+    },
+    {
+      "step": 638,
+      "timestamp": "2025-12-28T09:04:17.546438",
+      "elapsed_time": 371.83152866363525,
+      "loss": 0.1937,
+      "grad_norm": 0.17041485011577606,
+      "learning_rate": 0.00019660377358490566,
+      "epoch": 0.02375
+    },
+    {
+      "step": 639,
+      "timestamp": "2025-12-28T09:04:21.596846",
+      "elapsed_time": 375.88193678855896,
+      "loss": 0.2582,
+      "grad_norm": 0.20617352426052094,
+      "learning_rate": 0.00019647798742138367,
+      "epoch": 0.024375
+    },
+    {
+      "step": 640,
+      "timestamp": "2025-12-28T09:04:32.606180",
+      "elapsed_time": 386.8912706375122,
+      "loss": 0.1758,
+      "grad_norm": 0.1377355009317398,
+      "learning_rate": 0.00019635220125786165,
+      "epoch": 0.025
+    },
+    {
+      "step": 641,
+      "timestamp": "2025-12-28T09:04:43.506700",
+      "elapsed_time": 397.7917902469635,
+      "loss": 0.3234,
+      "grad_norm": 0.1705474555492401,
+      "learning_rate": 0.00019622641509433963,
+      "epoch": 0.025625
+    },
+    {
+      "step": 642,
+      "timestamp": "2025-12-28T09:04:51.282799",
+      "elapsed_time": 405.5678901672363,
+      "loss": 0.2184,
+      "grad_norm": 0.2311810404062271,
+      "learning_rate": 0.00019610062893081763,
+      "epoch": 0.02625
+    },
+    {
+      "step": 643,
+      "timestamp": "2025-12-28T09:04:57.361632",
+      "elapsed_time": 411.6467225551605,
+      "loss": 0.188,
+      "grad_norm": 0.1845218986272812,
+      "learning_rate": 0.0001959748427672956,
+      "epoch": 0.026875
+    },
+    {
+      "step": 644,
+      "timestamp": "2025-12-28T09:05:06.307667",
+      "elapsed_time": 420.5927574634552,
+      "loss": 0.2218,
+      "grad_norm": 0.18984770774841309,
+      "learning_rate": 0.0001958490566037736,
+      "epoch": 0.0275
+    },
+    {
+      "step": 645,
+      "timestamp": "2025-12-28T09:05:20.346611",
+      "elapsed_time": 434.6317012310028,
+      "loss": 0.1774,
+      "grad_norm": 0.1433676928281784,
+      "learning_rate": 0.00019572327044025157,
+      "epoch": 0.028125
+    },
+    {
+      "step": 646,
+      "timestamp": "2025-12-28T09:05:25.128238",
+      "elapsed_time": 439.4133291244507,
+      "loss": 0.3369,
+      "grad_norm": 0.22627809643745422,
+      "learning_rate": 0.00019559748427672958,
+      "epoch": 0.02875
+    },
+    {
+      "step": 647,
+      "timestamp": "2025-12-28T09:05:36.797870",
+      "elapsed_time": 451.08296036720276,
+      "loss": 0.1735,
+      "grad_norm": 0.16844865679740906,
+      "learning_rate": 0.00019547169811320755,
+      "epoch": 0.029375
+    },
+    {
+      "step": 648,
+      "timestamp": "2025-12-28T09:05:57.683316",
+      "elapsed_time": 471.96840620040894,
+      "loss": 0.1199,
+      "grad_norm": 0.11961396783590317,
+      "learning_rate": 0.00019534591194968553,
+      "epoch": 0.03
+    },
+    {
+      "step": 649,
+      "timestamp": "2025-12-28T09:06:05.657286",
+      "elapsed_time": 479.9423773288727,
+      "loss": 0.2306,
+      "grad_norm": 0.1685563325881958,
+      "learning_rate": 0.00019522012578616354,
+      "epoch": 0.030625
+    },
+    {
+      "step": 650,
+      "timestamp": "2025-12-28T09:06:20.462627",
+      "elapsed_time": 494.7477180957794,
+      "loss": 0.1896,
+      "grad_norm": 0.12992839515209198,
+      "learning_rate": 0.00019509433962264152,
+      "epoch": 0.03125
+    },
+    {
+      "step": 651,
+      "timestamp": "2025-12-28T09:06:31.640433",
+      "elapsed_time": 505.92552399635315,
+      "loss": 0.135,
+      "grad_norm": 0.12081814557313919,
+      "learning_rate": 0.0001949685534591195,
+      "epoch": 0.031875
+    },
+    {
+      "step": 652,
+      "timestamp": "2025-12-28T09:06:42.823577",
+      "elapsed_time": 517.1086673736572,
+      "loss": 0.1369,
+      "grad_norm": 0.11642343550920486,
+      "learning_rate": 0.0001948427672955975,
+      "epoch": 0.0325
+    },
+    {
+      "step": 653,
+      "timestamp": "2025-12-28T09:06:51.916795",
+      "elapsed_time": 526.2018857002258,
+      "loss": 0.1889,
+      "grad_norm": 0.13910572230815887,
+      "learning_rate": 0.00019471698113207548,
+      "epoch": 0.033125
+    },
+    {
+      "step": 654,
+      "timestamp": "2025-12-28T09:07:02.155700",
+      "elapsed_time": 536.4407908916473,
+      "loss": 0.3567,
+      "grad_norm": 0.1917208731174469,
+      "learning_rate": 0.00019459119496855346,
+      "epoch": 0.03375
+    },
+    {
+      "step": 655,
+      "timestamp": "2025-12-28T09:07:15.706369",
+      "elapsed_time": 549.9914598464966,
+      "loss": 0.1709,
+      "grad_norm": 0.12028059363365173,
+      "learning_rate": 0.00019446540880503147,
+      "epoch": 0.034375
+    },
+    {
+      "step": 656,
+      "timestamp": "2025-12-28T09:07:32.250659",
+      "elapsed_time": 566.535749912262,
+      "loss": 0.1228,
+      "grad_norm": 0.19263313710689545,
+      "learning_rate": 0.00019433962264150945,
+      "epoch": 0.035
+    },
+    {
+      "step": 657,
+      "timestamp": "2025-12-28T09:07:39.316479",
+      "elapsed_time": 573.601569890976,
+      "loss": 0.2289,
+      "grad_norm": 0.2995343506336212,
+      "learning_rate": 0.00019421383647798743,
+      "epoch": 0.035625
+    },
+    {
+      "step": 658,
+      "timestamp": "2025-12-28T09:07:54.181478",
+      "elapsed_time": 588.4665679931641,
+      "loss": 0.1498,
+      "grad_norm": 0.13047035038471222,
+      "learning_rate": 0.00019408805031446543,
+      "epoch": 0.03625
+    },
+    {
+      "step": 659,
+      "timestamp": "2025-12-28T09:08:01.427506",
+      "elapsed_time": 595.7125968933105,
+      "loss": 0.1957,
+      "grad_norm": 0.16221193969249725,
+      "learning_rate": 0.0001939622641509434,
+      "epoch": 0.036875
+    },
+    {
+      "step": 660,
+      "timestamp": "2025-12-28T09:08:07.362296",
+      "elapsed_time": 601.6473863124847,
+      "loss": 0.1755,
+      "grad_norm": 0.1594441831111908,
+      "learning_rate": 0.0001938364779874214,
+      "epoch": 0.0375
+    },
+    {
+      "step": 661,
+      "timestamp": "2025-12-28T09:08:13.438049",
+      "elapsed_time": 607.7231397628784,
+      "loss": 0.2449,
+      "grad_norm": 0.18605917692184448,
+      "learning_rate": 0.00019371069182389937,
+      "epoch": 0.038125
+    },
+    {
+      "step": 662,
+      "timestamp": "2025-12-28T09:08:25.983999",
+      "elapsed_time": 620.2690892219543,
+      "loss": 0.1445,
+      "grad_norm": 0.11945810914039612,
+      "learning_rate": 0.00019358490566037738,
+      "epoch": 0.03875
+    },
+    {
+      "step": 663,
+      "timestamp": "2025-12-28T09:08:34.320553",
+      "elapsed_time": 628.6056432723999,
+      "loss": 0.1952,
+      "grad_norm": 0.15212717652320862,
+      "learning_rate": 0.00019345911949685536,
+      "epoch": 0.039375
+    },
+    {
+      "step": 664,
+      "timestamp": "2025-12-28T09:08:47.204012",
+      "elapsed_time": 641.4891028404236,
+      "loss": 0.1264,
+      "grad_norm": 0.1063343733549118,
+      "learning_rate": 0.00019333333333333333,
+      "epoch": 0.04
+    },
+    {
+      "step": 665,
+      "timestamp": "2025-12-28T09:08:55.268548",
+      "elapsed_time": 649.5536382198334,
+      "loss": 0.1998,
+      "grad_norm": 0.1476004421710968,
+      "learning_rate": 0.00019320754716981134,
+      "epoch": 0.040625
+    },
+    {
+      "step": 666,
+      "timestamp": "2025-12-28T09:09:11.683318",
+      "elapsed_time": 665.9684083461761,
+      "loss": 0.1221,
+      "grad_norm": 0.1132006123661995,
+      "learning_rate": 0.00019308176100628932,
+      "epoch": 0.04125
+    },
+    {
+      "step": 667,
+      "timestamp": "2025-12-28T09:09:22.208240",
+      "elapsed_time": 676.493331193924,
+      "loss": 0.1504,
+      "grad_norm": 0.13121221959590912,
+      "learning_rate": 0.0001929559748427673,
+      "epoch": 0.041875
+    },
+    {
+      "step": 668,
+      "timestamp": "2025-12-28T09:09:42.700506",
+      "elapsed_time": 696.9855964183807,
+      "loss": 0.121,
+      "grad_norm": 0.0900096669793129,
+      "learning_rate": 0.0001928301886792453,
+      "epoch": 0.0425
+    },
+    {
+      "step": 669,
+      "timestamp": "2025-12-28T09:09:57.623005",
+      "elapsed_time": 711.9080958366394,
+      "loss": 0.1276,
+      "grad_norm": 0.11430996656417847,
+      "learning_rate": 0.00019270440251572328,
+      "epoch": 0.043125
+    },
+    {
+      "step": 670,
+      "timestamp": "2025-12-28T09:10:04.247302",
+      "elapsed_time": 718.5323920249939,
+      "loss": 0.1923,
+      "grad_norm": 0.19039106369018555,
+      "learning_rate": 0.00019257861635220126,
+      "epoch": 0.04375
+    },
+    {
+      "step": 671,
+      "timestamp": "2025-12-28T09:10:16.730363",
+      "elapsed_time": 731.0154540538788,
+      "loss": 0.314,
+      "grad_norm": 0.1691213995218277,
+      "learning_rate": 0.00019245283018867927,
+      "epoch": 0.044375
+    },
+    {
+      "step": 672,
+      "timestamp": "2025-12-28T09:10:26.112523",
+      "elapsed_time": 740.3976130485535,
+      "loss": 0.164,
+      "grad_norm": 0.13402487337589264,
+      "learning_rate": 0.00019232704402515725,
+      "epoch": 0.045
+    },
+    {
+      "step": 673,
+      "timestamp": "2025-12-28T09:10:34.372277",
+      "elapsed_time": 748.6573669910431,
+      "loss": 0.1849,
+      "grad_norm": 0.14694516360759735,
+      "learning_rate": 0.00019220125786163523,
+      "epoch": 0.045625
+    },
+    {
+      "step": 674,
+      "timestamp": "2025-12-28T09:10:44.792607",
+      "elapsed_time": 759.0776975154877,
+      "loss": 0.1962,
+      "grad_norm": 0.15182200074195862,
+      "learning_rate": 0.0001920754716981132,
+      "epoch": 0.04625
+    },
+    {
+      "step": 675,
+      "timestamp": "2025-12-28T09:10:54.996775",
+      "elapsed_time": 769.2818658351898,
+      "loss": 0.1313,
+      "grad_norm": 0.11995584517717361,
+      "learning_rate": 0.0001919496855345912,
+      "epoch": 0.046875
+    },
+    {
+      "step": 676,
+      "timestamp": "2025-12-28T09:11:03.629001",
+      "elapsed_time": 777.9140913486481,
+      "loss": 0.2438,
+      "grad_norm": 0.14396265149116516,
+      "learning_rate": 0.0001918238993710692,
+      "epoch": 0.0475
+    },
+    {
+      "step": 677,
+      "timestamp": "2025-12-28T09:11:12.607355",
+      "elapsed_time": 786.8924453258514,
+      "loss": 0.2567,
+      "grad_norm": 0.1717308908700943,
+      "learning_rate": 0.00019169811320754717,
+      "epoch": 0.048125
+    },
+    {
+      "step": 678,
+      "timestamp": "2025-12-28T09:11:25.911563",
+      "elapsed_time": 800.196653842926,
+      "loss": 0.1644,
+      "grad_norm": 0.14998085796833038,
+      "learning_rate": 0.00019157232704402518,
+      "epoch": 0.04875
+    },
+    {
+      "step": 679,
+      "timestamp": "2025-12-28T09:11:33.380615",
+      "elapsed_time": 807.6657054424286,
+      "loss": 0.3003,
+      "grad_norm": 0.16212862730026245,
+      "learning_rate": 0.00019144654088050316,
+      "epoch": 0.049375
+    },
+    {
+      "step": 680,
+      "timestamp": "2025-12-28T09:11:39.286109",
+      "elapsed_time": 813.5711998939514,
+      "loss": 0.261,
+      "grad_norm": 0.19972242414951324,
+      "learning_rate": 0.00019132075471698114,
+      "epoch": 0.05
+    },
+    {
+      "step": 681,
+      "timestamp": "2025-12-28T09:11:48.163985",
+      "elapsed_time": 822.4490756988525,
+      "loss": 0.3926,
+      "grad_norm": 0.17470583319664001,
+      "learning_rate": 0.00019119496855345914,
+      "epoch": 0.050625
+    },
+    {
+      "step": 682,
+      "timestamp": "2025-12-28T09:12:01.357382",
+      "elapsed_time": 835.6424729824066,
+      "loss": 0.1516,
+      "grad_norm": 0.1337703913450241,
+      "learning_rate": 0.00019106918238993712,
+      "epoch": 0.05125
+    },
+    {
+      "step": 683,
+      "timestamp": "2025-12-28T09:12:11.393661",
+      "elapsed_time": 845.678751707077,
+      "loss": 0.2966,
+      "grad_norm": 0.14767307043075562,
+      "learning_rate": 0.0001909433962264151,
+      "epoch": 0.051875
+    },
+    {
+      "step": 684,
+      "timestamp": "2025-12-28T09:12:17.843729",
+      "elapsed_time": 852.1288199424744,
+      "loss": 0.217,
+      "grad_norm": 0.16958969831466675,
+      "learning_rate": 0.0001908176100628931,
+      "epoch": 0.0525
+    },
+    {
+      "step": 685,
+      "timestamp": "2025-12-28T09:12:25.829441",
+      "elapsed_time": 860.1145317554474,
+      "loss": 0.1656,
+      "grad_norm": 0.14111389219760895,
+      "learning_rate": 0.00019069182389937108,
+      "epoch": 0.053125
+    },
+    {
+      "step": 686,
+      "timestamp": "2025-12-28T09:12:37.674364",
+      "elapsed_time": 871.9594547748566,
+      "loss": 0.1438,
+      "grad_norm": 0.11669319868087769,
+      "learning_rate": 0.00019056603773584906,
+      "epoch": 0.05375
+    },
+    {
+      "step": 687,
+      "timestamp": "2025-12-28T09:12:56.652637",
+      "elapsed_time": 890.9377274513245,
+      "loss": 0.1478,
+      "grad_norm": 0.12338529527187347,
+      "learning_rate": 0.00019044025157232704,
+      "epoch": 0.054375
+    },
+    {
+      "step": 688,
+      "timestamp": "2025-12-28T09:13:02.953778",
+      "elapsed_time": 897.2388682365417,
+      "loss": 0.198,
+      "grad_norm": 0.1538494974374771,
+      "learning_rate": 0.00019031446540880505,
+      "epoch": 0.055
+    },
+    {
+      "step": 689,
+      "timestamp": "2025-12-28T09:13:11.878858",
+      "elapsed_time": 906.1639485359192,
+      "loss": 0.1424,
+      "grad_norm": 0.12444575875997543,
+      "learning_rate": 0.00019018867924528303,
+      "epoch": 0.055625
+    },
+    {
+      "step": 690,
+      "timestamp": "2025-12-28T09:13:20.334671",
+      "elapsed_time": 914.61976146698,
+      "loss": 0.1823,
+      "grad_norm": 0.19447259604930878,
+      "learning_rate": 0.000190062893081761,
+      "epoch": 0.05625
+    },
+    {
+      "step": 691,
+      "timestamp": "2025-12-28T09:13:27.616304",
+      "elapsed_time": 921.9013941287994,
+      "loss": 0.1547,
+      "grad_norm": 0.14511334896087646,
+      "learning_rate": 0.00018993710691823901,
+      "epoch": 0.056875
+    },
+    {
+      "step": 692,
+      "timestamp": "2025-12-28T09:13:41.557565",
+      "elapsed_time": 935.8426558971405,
+      "loss": 0.1594,
+      "grad_norm": 0.11656054854393005,
+      "learning_rate": 0.000189811320754717,
+      "epoch": 0.0575
+    },
+    {
+      "step": 693,
+      "timestamp": "2025-12-28T09:13:51.373998",
+      "elapsed_time": 945.6590890884399,
+      "loss": 0.2355,
+      "grad_norm": 0.1568254977464676,
+      "learning_rate": 0.00018968553459119497,
+      "epoch": 0.058125
+    },
+    {
+      "step": 694,
+      "timestamp": "2025-12-28T09:14:06.517264",
+      "elapsed_time": 960.8023540973663,
+      "loss": 0.2098,
+      "grad_norm": 0.12919512391090393,
+      "learning_rate": 0.00018955974842767298,
+      "epoch": 0.05875
+    },
+    {
+      "step": 695,
+      "timestamp": "2025-12-28T09:14:14.033362",
+      "elapsed_time": 968.3184523582458,
+      "loss": 0.2264,
+      "grad_norm": 0.21954892575740814,
+      "learning_rate": 0.00018943396226415096,
+      "epoch": 0.059375
+    },
+    {
+      "step": 696,
+      "timestamp": "2025-12-28T09:14:23.154345",
+      "elapsed_time": 977.4394352436066,
+      "loss": 0.197,
+      "grad_norm": 0.1527150720357895,
+      "learning_rate": 0.00018930817610062894,
+      "epoch": 0.06
+    },
+    {
+      "step": 697,
+      "timestamp": "2025-12-28T09:14:30.973861",
+      "elapsed_time": 985.258951663971,
+      "loss": 0.2027,
+      "grad_norm": 0.16911649703979492,
+      "learning_rate": 0.00018918238993710694,
+      "epoch": 0.060625
+    },
+    {
+      "step": 698,
+      "timestamp": "2025-12-28T09:14:38.748666",
+      "elapsed_time": 993.0337567329407,
+      "loss": 0.2625,
+      "grad_norm": 0.15294122695922852,
+      "learning_rate": 0.00018905660377358492,
+      "epoch": 0.06125
+    },
+    {
+      "step": 699,
+      "timestamp": "2025-12-28T09:14:46.484732",
+      "elapsed_time": 1000.7698221206665,
+      "loss": 0.2257,
+      "grad_norm": 0.16110914945602417,
+      "learning_rate": 0.0001889308176100629,
+      "epoch": 0.061875
+    },
+    {
+      "step": 700,
+      "timestamp": "2025-12-28T09:14:52.017443",
+      "elapsed_time": 1006.3025336265564,
+      "loss": 0.2229,
+      "grad_norm": 0.19415181875228882,
+      "learning_rate": 0.00018880503144654088,
+      "epoch": 0.0625
+    },
+    {
+      "step": 701,
+      "timestamp": "2025-12-28T09:15:00.051403",
+      "elapsed_time": 1014.336493730545,
+      "loss": 0.1913,
+      "grad_norm": 0.1673995554447174,
+      "learning_rate": 0.00018867924528301889,
+      "epoch": 0.063125
+    },
+    {
+      "step": 702,
+      "timestamp": "2025-12-28T09:15:16.601964",
+      "elapsed_time": 1030.8870537281036,
+      "loss": 0.1377,
+      "grad_norm": 0.09899748116731644,
+      "learning_rate": 0.00018855345911949686,
+      "epoch": 0.06375
+    },
+    {
+      "step": 703,
+      "timestamp": "2025-12-28T09:15:30.488977",
+      "elapsed_time": 1044.7740678787231,
+      "loss": 0.1405,
+      "grad_norm": 0.11365586519241333,
+      "learning_rate": 0.00018842767295597484,
+      "epoch": 0.064375
+    },
+    {
+      "step": 704,
+      "timestamp": "2025-12-28T09:15:36.750160",
+      "elapsed_time": 1051.0352516174316,
+      "loss": 0.1852,
+      "grad_norm": 0.1651289463043213,
+      "learning_rate": 0.00018830188679245285,
+      "epoch": 0.065
+    },
+    {
+      "step": 705,
+      "timestamp": "2025-12-28T09:15:49.379817",
+      "elapsed_time": 1063.6649072170258,
+      "loss": 0.1602,
+      "grad_norm": 0.12038940191268921,
+      "learning_rate": 0.00018817610062893083,
+      "epoch": 0.065625
+    },
+    {
+      "step": 706,
+      "timestamp": "2025-12-28T09:15:54.708518",
+      "elapsed_time": 1068.9936089515686,
+      "loss": 0.3003,
+      "grad_norm": 0.26949650049209595,
+      "learning_rate": 0.0001880503144654088,
+      "epoch": 0.06625
+    },
+    {
+      "step": 707,
+      "timestamp": "2025-12-28T09:16:05.981736",
+      "elapsed_time": 1080.2668268680573,
+      "loss": 0.1754,
+      "grad_norm": 0.133348748087883,
+      "learning_rate": 0.00018792452830188681,
+      "epoch": 0.066875
+    },
+    {
+      "step": 708,
+      "timestamp": "2025-12-28T09:16:11.541142",
+      "elapsed_time": 1085.8262326717377,
+      "loss": 0.1986,
+      "grad_norm": 0.1636824756860733,
+      "learning_rate": 0.0001877987421383648,
+      "epoch": 0.0675
+    },
+    {
+      "step": 709,
+      "timestamp": "2025-12-28T09:16:17.744602",
+      "elapsed_time": 1092.0296926498413,
+      "loss": 0.2282,
+      "grad_norm": 0.18413999676704407,
+      "learning_rate": 0.00018767295597484277,
+      "epoch": 0.068125
+    },
+    {
+      "step": 710,
+      "timestamp": "2025-12-28T09:16:28.622419",
+      "elapsed_time": 1102.9075095653534,
+      "loss": 0.1403,
+      "grad_norm": 0.1099625825881958,
+      "learning_rate": 0.00018754716981132078,
+      "epoch": 0.06875
+    },
+    {
+      "step": 711,
+      "timestamp": "2025-12-28T09:16:38.899862",
+      "elapsed_time": 1113.1849522590637,
+      "loss": 0.3365,
+      "grad_norm": 0.1869092434644699,
+      "learning_rate": 0.00018742138364779876,
+      "epoch": 0.069375
+    },
+    {
+      "step": 712,
+      "timestamp": "2025-12-28T09:16:58.976813",
+      "elapsed_time": 1133.2619035243988,
+      "loss": 0.1024,
+      "grad_norm": 0.08866474777460098,
+      "learning_rate": 0.00018729559748427674,
+      "epoch": 0.07
+    },
+    {
+      "step": 713,
+      "timestamp": "2025-12-28T09:17:05.528653",
+      "elapsed_time": 1139.8137435913086,
+      "loss": 0.2417,
+      "grad_norm": 0.1919388324022293,
+      "learning_rate": 0.00018716981132075472,
+      "epoch": 0.070625
+    },
+    {
+      "step": 714,
+      "timestamp": "2025-12-28T09:17:16.181514",
+      "elapsed_time": 1150.466604232788,
+      "loss": 0.1628,
+      "grad_norm": 0.12882591784000397,
+      "learning_rate": 0.00018704402515723272,
+      "epoch": 0.07125
+    },
+    {
+      "step": 715,
+      "timestamp": "2025-12-28T09:17:21.921176",
+      "elapsed_time": 1156.2062666416168,
+      "loss": 0.2503,
+      "grad_norm": 0.17477668821811676,
+      "learning_rate": 0.0001869182389937107,
+      "epoch": 0.071875
+    },
+    {
+      "step": 716,
+      "timestamp": "2025-12-28T09:17:31.060009",
+      "elapsed_time": 1165.3450994491577,
+      "loss": 0.19,
+      "grad_norm": 0.15098389983177185,
+      "learning_rate": 0.00018679245283018868,
+      "epoch": 0.0725
+    },
+    {
+      "step": 717,
+      "timestamp": "2025-12-28T09:17:51.950466",
+      "elapsed_time": 1186.235556602478,
+      "loss": 0.1353,
+      "grad_norm": 0.12708379328250885,
+      "learning_rate": 0.0001866666666666667,
+      "epoch": 0.073125
+    },
+    {
+      "step": 718,
+      "timestamp": "2025-12-28T09:17:57.215989",
+      "elapsed_time": 1191.5010793209076,
+      "loss": 0.2758,
+      "grad_norm": 0.20731349289417267,
+      "learning_rate": 0.00018654088050314467,
+      "epoch": 0.07375
+    },
+    {
+      "step": 719,
+      "timestamp": "2025-12-28T09:18:05.159270",
+      "elapsed_time": 1199.4443612098694,
+      "loss": 0.226,
+      "grad_norm": 0.16173875331878662,
+      "learning_rate": 0.00018641509433962264,
+      "epoch": 0.074375
+    },
+    {
+      "step": 720,
+      "timestamp": "2025-12-28T09:18:10.909682",
+      "elapsed_time": 1205.1947722434998,
+      "loss": 0.2709,
+      "grad_norm": 0.21631106734275818,
+      "learning_rate": 0.00018628930817610065,
+      "epoch": 0.075
+    },
+    {
+      "step": 721,
+      "timestamp": "2025-12-28T09:18:23.590226",
+      "elapsed_time": 1217.875316619873,
+      "loss": 0.2306,
+      "grad_norm": 0.12810084223747253,
+      "learning_rate": 0.00018616352201257863,
+      "epoch": 0.075625
+    },
+    {
+      "step": 722,
+      "timestamp": "2025-12-28T09:18:38.922154",
+      "elapsed_time": 1233.2072455883026,
+      "loss": 0.1657,
+      "grad_norm": 0.10849788039922714,
+      "learning_rate": 0.0001860377358490566,
+      "epoch": 0.07625
+    },
+    {
+      "step": 723,
+      "timestamp": "2025-12-28T09:18:48.500418",
+      "elapsed_time": 1242.7855124473572,
+      "loss": 0.1657,
+      "grad_norm": 0.17049764096736908,
+      "learning_rate": 0.00018591194968553462,
+      "epoch": 0.076875
+    },
+    {
+      "step": 724,
+      "timestamp": "2025-12-28T09:18:57.124927",
+      "elapsed_time": 1251.4100172519684,
+      "loss": 0.1443,
+      "grad_norm": 0.1467668116092682,
+      "learning_rate": 0.0001857861635220126,
+      "epoch": 0.0775
+    },
+    {
+      "step": 725,
+      "timestamp": "2025-12-28T09:19:04.646596",
+      "elapsed_time": 1258.9316868782043,
+      "loss": 0.2106,
+      "grad_norm": 0.16145600378513336,
+      "learning_rate": 0.00018566037735849057,
+      "epoch": 0.078125
+    },
+    {
+      "step": 726,
+      "timestamp": "2025-12-28T09:19:13.415801",
+      "elapsed_time": 1267.7008922100067,
+      "loss": 0.1683,
+      "grad_norm": 0.13938355445861816,
+      "learning_rate": 0.00018553459119496855,
+      "epoch": 0.07875
+    },
+    {
+      "step": 727,
+      "timestamp": "2025-12-28T09:19:19.791168",
+      "elapsed_time": 1274.0762577056885,
+      "loss": 0.1834,
+      "grad_norm": 0.16343441605567932,
+      "learning_rate": 0.00018540880503144656,
+      "epoch": 0.079375
+    },
+    {
+      "step": 728,
+      "timestamp": "2025-12-28T09:19:26.780643",
+      "elapsed_time": 1281.065733909607,
+      "loss": 0.3605,
+      "grad_norm": 0.18578511476516724,
+      "learning_rate": 0.00018528301886792454,
+      "epoch": 0.08
+    },
+    {
+      "step": 729,
+      "timestamp": "2025-12-28T09:19:40.770904",
+      "elapsed_time": 1295.0559949874878,
+      "loss": 0.1423,
+      "grad_norm": 0.10042066127061844,
+      "learning_rate": 0.00018515723270440252,
+      "epoch": 0.080625
+    },
+    {
+      "step": 730,
+      "timestamp": "2025-12-28T09:19:46.875385",
+      "elapsed_time": 1301.1604759693146,
+      "loss": 0.2364,
+      "grad_norm": 0.2259899079799652,
+      "learning_rate": 0.00018503144654088052,
+      "epoch": 0.08125
+    },
+    {
+      "step": 731,
+      "timestamp": "2025-12-28T09:19:57.704231",
+      "elapsed_time": 1311.9893221855164,
+      "loss": 0.1339,
+      "grad_norm": 0.15420939028263092,
+      "learning_rate": 0.0001849056603773585,
+      "epoch": 0.081875
+    },
+    {
+      "step": 732,
+      "timestamp": "2025-12-28T09:20:05.181888",
+      "elapsed_time": 1319.4669790267944,
+      "loss": 0.1448,
+      "grad_norm": 0.13649825751781464,
+      "learning_rate": 0.00018477987421383648,
+      "epoch": 0.0825
+    },
+    {
+      "step": 733,
+      "timestamp": "2025-12-28T09:20:22.787205",
+      "elapsed_time": 1337.0722954273224,
+      "loss": 0.1804,
+      "grad_norm": 0.1483648419380188,
+      "learning_rate": 0.0001846540880503145,
+      "epoch": 0.083125
+    },
+    {
+      "step": 734,
+      "timestamp": "2025-12-28T09:20:32.228745",
+      "elapsed_time": 1346.5138351917267,
+      "loss": 0.1621,
+      "grad_norm": 0.1357170045375824,
+      "learning_rate": 0.00018452830188679247,
+      "epoch": 0.08375
+    },
+    {
+      "step": 735,
+      "timestamp": "2025-12-28T09:20:39.593318",
+      "elapsed_time": 1353.8784093856812,
+      "loss": 0.1985,
+      "grad_norm": 0.18031319975852966,
+      "learning_rate": 0.00018440251572327045,
+      "epoch": 0.084375
+    },
+    {
+      "step": 736,
+      "timestamp": "2025-12-28T09:20:46.467101",
+      "elapsed_time": 1360.752191543579,
+      "loss": 0.4623,
+      "grad_norm": 0.2359013855457306,
+      "learning_rate": 0.00018427672955974845,
+      "epoch": 0.085
+    },
+    {
+      "step": 737,
+      "timestamp": "2025-12-28T09:20:53.603967",
+      "elapsed_time": 1367.8890571594238,
+      "loss": 0.2163,
+      "grad_norm": 0.16415388882160187,
+      "learning_rate": 0.00018415094339622643,
+      "epoch": 0.085625
+    },
+    {
+      "step": 738,
+      "timestamp": "2025-12-28T09:20:59.877010",
+      "elapsed_time": 1374.1621007919312,
+      "loss": 0.2737,
+      "grad_norm": 0.18287940323352814,
+      "learning_rate": 0.0001840251572327044,
+      "epoch": 0.08625
+    },
+    {
+      "step": 739,
+      "timestamp": "2025-12-28T09:21:17.007779",
+      "elapsed_time": 1391.2928698062897,
+      "loss": 0.144,
+      "grad_norm": 0.10500773787498474,
+      "learning_rate": 0.0001838993710691824,
+      "epoch": 0.086875
+    },
+    {
+      "step": 740,
+      "timestamp": "2025-12-28T09:21:27.416174",
+      "elapsed_time": 1401.70126414299,
+      "loss": 0.1542,
+      "grad_norm": 0.11694104224443436,
+      "learning_rate": 0.0001837735849056604,
+      "epoch": 0.0875
+    },
+    {
+      "step": 741,
+      "timestamp": "2025-12-28T09:21:32.682215",
+      "elapsed_time": 1406.9673054218292,
+      "loss": 0.2283,
+      "grad_norm": 0.17158034443855286,
+      "learning_rate": 0.00018364779874213837,
+      "epoch": 0.088125
+    },
+    {
+      "step": 742,
+      "timestamp": "2025-12-28T09:21:42.891669",
+      "elapsed_time": 1417.17675948143,
+      "loss": 0.1413,
+      "grad_norm": 0.12412276864051819,
+      "learning_rate": 0.00018352201257861635,
+      "epoch": 0.08875
+    },
+    {
+      "step": 743,
+      "timestamp": "2025-12-28T09:21:50.263790",
+      "elapsed_time": 1424.5488805770874,
+      "loss": 0.322,
+      "grad_norm": 0.2420777678489685,
+      "learning_rate": 0.00018339622641509436,
+      "epoch": 0.089375
+    },
+    {
+      "step": 744,
+      "timestamp": "2025-12-28T09:22:01.677433",
+      "elapsed_time": 1435.9625227451324,
+      "loss": 0.1475,
+      "grad_norm": 0.10151813179254532,
+      "learning_rate": 0.00018327044025157234,
+      "epoch": 0.09
+    },
+    {
+      "step": 745,
+      "timestamp": "2025-12-28T09:22:11.434022",
+      "elapsed_time": 1445.7191128730774,
+      "loss": 0.3527,
+      "grad_norm": 0.16911114752292633,
+      "learning_rate": 0.00018314465408805032,
+      "epoch": 0.090625
+    },
+    {
+      "step": 746,
+      "timestamp": "2025-12-28T09:22:23.572820",
+      "elapsed_time": 1457.8579106330872,
+      "loss": 0.1674,
+      "grad_norm": 0.12217868864536285,
+      "learning_rate": 0.00018301886792452832,
+      "epoch": 0.09125
+    },
+    {
+      "step": 747,
+      "timestamp": "2025-12-28T09:22:36.319867",
+      "elapsed_time": 1470.6049571037292,
+      "loss": 0.136,
+      "grad_norm": 0.10214618593454361,
+      "learning_rate": 0.0001828930817610063,
+      "epoch": 0.091875
+    },
+    {
+      "step": 748,
+      "timestamp": "2025-12-28T09:22:46.737961",
+      "elapsed_time": 1481.023051738739,
+      "loss": 0.2034,
+      "grad_norm": 0.14041507244110107,
+      "learning_rate": 0.00018276729559748428,
+      "epoch": 0.0925
+    },
+    {
+      "step": 749,
+      "timestamp": "2025-12-28T09:22:54.486010",
+      "elapsed_time": 1488.7711000442505,
+      "loss": 0.203,
+      "grad_norm": 0.23749075829982758,
+      "learning_rate": 0.0001826415094339623,
+      "epoch": 0.093125
+    },
+    {
+      "step": 750,
+      "timestamp": "2025-12-28T09:23:15.381836",
+      "elapsed_time": 1509.6669268608093,
+      "loss": 0.1252,
+      "grad_norm": 0.09619986265897751,
+      "learning_rate": 0.00018251572327044027,
+      "epoch": 0.09375
+    },
+    {
+      "step": 751,
+      "timestamp": "2025-12-28T09:23:26.138595",
+      "elapsed_time": 1520.4236857891083,
+      "loss": 0.1759,
+      "grad_norm": 0.1735697239637375,
+      "learning_rate": 0.00018238993710691825,
+      "epoch": 0.094375
+    },
+    {
+      "step": 752,
+      "timestamp": "2025-12-28T09:23:31.779100",
+      "elapsed_time": 1526.064194202423,
+      "loss": 0.4079,
+      "grad_norm": 0.2254641205072403,
+      "learning_rate": 0.00018226415094339625,
+      "epoch": 0.095
+    },
+    {
+      "step": 753,
+      "timestamp": "2025-12-28T09:23:48.621990",
+      "elapsed_time": 1542.9070808887482,
+      "loss": 0.1389,
+      "grad_norm": 0.09768623113632202,
+      "learning_rate": 0.00018213836477987423,
+      "epoch": 0.095625
+    },
+    {
+      "step": 754,
+      "timestamp": "2025-12-28T09:23:56.146839",
+      "elapsed_time": 1550.4319293498993,
+      "loss": 0.3221,
+      "grad_norm": 0.17994655668735504,
+      "learning_rate": 0.0001820125786163522,
+      "epoch": 0.09625
+    },
+    {
+      "step": 755,
+      "timestamp": "2025-12-28T09:24:17.041799",
+      "elapsed_time": 1571.3268892765045,
+      "loss": 0.1407,
+      "grad_norm": 0.09752582013607025,
+      "learning_rate": 0.0001818867924528302,
+      "epoch": 0.096875
+    },
+    {
+      "step": 756,
+      "timestamp": "2025-12-28T09:24:27.104362",
+      "elapsed_time": 1581.389452457428,
+      "loss": 0.1681,
+      "grad_norm": 0.21668905019760132,
+      "learning_rate": 0.0001817610062893082,
+      "epoch": 0.0975
+    },
+    {
+      "step": 757,
+      "timestamp": "2025-12-28T09:24:38.052954",
+      "elapsed_time": 1592.3380448818207,
+      "loss": 0.2357,
+      "grad_norm": 0.14341306686401367,
+      "learning_rate": 0.00018163522012578617,
+      "epoch": 0.098125
+    },
+    {
+      "step": 758,
+      "timestamp": "2025-12-28T09:24:46.883893",
+      "elapsed_time": 1601.168983221054,
+      "loss": 0.1544,
+      "grad_norm": 0.12626543641090393,
+      "learning_rate": 0.00018150943396226415,
+      "epoch": 0.09875
+    },
+    {
+      "step": 759,
+      "timestamp": "2025-12-28T09:24:56.091568",
+      "elapsed_time": 1610.3766589164734,
+      "loss": 0.3366,
+      "grad_norm": 0.1486847847700119,
+      "learning_rate": 0.00018138364779874216,
+      "epoch": 0.099375
+    },
+    {
+      "step": 760,
+      "timestamp": "2025-12-28T09:25:02.464933",
+      "elapsed_time": 1616.7500236034393,
+      "loss": 0.2171,
+      "grad_norm": 0.1620669662952423,
+      "learning_rate": 0.00018125786163522014,
+      "epoch": 0.1
+    },
+    {
+      "step": 761,
+      "timestamp": "2025-12-28T09:25:12.378673",
+      "elapsed_time": 1626.6637637615204,
+      "loss": 0.1958,
+      "grad_norm": 0.13596412539482117,
+      "learning_rate": 0.00018113207547169812,
+      "epoch": 0.100625
+    },
+    {
+      "step": 762,
+      "timestamp": "2025-12-28T09:25:24.048320",
+      "elapsed_time": 1638.3334102630615,
+      "loss": 0.1543,
+      "grad_norm": 0.19701559841632843,
+      "learning_rate": 0.00018100628930817612,
+      "epoch": 0.10125
+    },
+    {
+      "step": 763,
+      "timestamp": "2025-12-28T09:25:31.137409",
+      "elapsed_time": 1645.4224989414215,
+      "loss": 0.3084,
+      "grad_norm": 0.16929014027118683,
+      "learning_rate": 0.0001808805031446541,
+      "epoch": 0.101875
+    },
+    {
+      "step": 764,
+      "timestamp": "2025-12-28T09:25:37.301475",
+      "elapsed_time": 1651.5865650177002,
+      "loss": 0.2045,
+      "grad_norm": 0.1783858984708786,
+      "learning_rate": 0.00018075471698113208,
+      "epoch": 0.1025
+    },
+    {
+      "step": 765,
+      "timestamp": "2025-12-28T09:25:42.794771",
+      "elapsed_time": 1657.0798616409302,
+      "loss": 0.2026,
+      "grad_norm": 0.1741485893726349,
+      "learning_rate": 0.0001806289308176101,
+      "epoch": 0.103125
+    },
+    {
+      "step": 766,
+      "timestamp": "2025-12-28T09:25:53.503557",
+      "elapsed_time": 1667.788647890091,
+      "loss": 0.1651,
+      "grad_norm": 0.39134204387664795,
+      "learning_rate": 0.00018050314465408807,
+      "epoch": 0.10375
+    },
+    {
+      "step": 767,
+      "timestamp": "2025-12-28T09:26:03.591001",
+      "elapsed_time": 1677.8760917186737,
+      "loss": 0.2385,
+      "grad_norm": 0.19072601199150085,
+      "learning_rate": 0.00018037735849056605,
+      "epoch": 0.104375
+    },
+    {
+      "step": 768,
+      "timestamp": "2025-12-28T09:26:12.895494",
+      "elapsed_time": 1687.180584192276,
+      "loss": 0.2075,
+      "grad_norm": 0.14767950773239136,
+      "learning_rate": 0.00018025157232704403,
+      "epoch": 0.105
+    },
+    {
+      "step": 769,
+      "timestamp": "2025-12-28T09:26:28.488803",
+      "elapsed_time": 1702.7738931179047,
+      "loss": 0.1246,
+      "grad_norm": 0.11118397116661072,
+      "learning_rate": 0.00018012578616352203,
+      "epoch": 0.105625
+    },
+    {
+      "step": 770,
+      "timestamp": "2025-12-28T09:26:35.655701",
+      "elapsed_time": 1709.9407913684845,
+      "loss": 0.2173,
+      "grad_norm": 0.15238629281520844,
+      "learning_rate": 0.00018,
+      "epoch": 0.10625
+    },
+    {
+      "step": 771,
+      "timestamp": "2025-12-28T09:26:42.135442",
+      "elapsed_time": 1716.420532464981,
+      "loss": 0.4145,
+      "grad_norm": 0.17681153118610382,
+      "learning_rate": 0.000179874213836478,
+      "epoch": 0.106875
+    },
+    {
+      "step": 772,
+      "timestamp": "2025-12-28T09:26:54.831081",
+      "elapsed_time": 1729.1161713600159,
+      "loss": 0.2015,
+      "grad_norm": 0.17193950712680817,
+      "learning_rate": 0.000179748427672956,
+      "epoch": 0.1075
+    },
+    {
+      "step": 773,
+      "timestamp": "2025-12-28T09:27:05.543585",
+      "elapsed_time": 1739.8286757469177,
+      "loss": 0.1725,
+      "grad_norm": 0.11407608538866043,
+      "learning_rate": 0.00017962264150943398,
+      "epoch": 0.108125
+    },
+    {
+      "step": 774,
+      "timestamp": "2025-12-28T09:27:09.428086",
+      "elapsed_time": 1743.713176727295,
+      "loss": 0.3777,
+      "grad_norm": 0.2451329529285431,
+      "learning_rate": 0.00017949685534591195,
+      "epoch": 0.10875
+    },
+    {
+      "step": 775,
+      "timestamp": "2025-12-28T09:27:23.100645",
+      "elapsed_time": 1757.3857352733612,
+      "loss": 0.1818,
+      "grad_norm": 0.11372605711221695,
+      "learning_rate": 0.00017937106918238996,
+      "epoch": 0.109375
+    },
+    {
+      "step": 776,
+      "timestamp": "2025-12-28T09:27:33.162588",
+      "elapsed_time": 1767.447678565979,
+      "loss": 0.1858,
+      "grad_norm": 0.1400919407606125,
+      "learning_rate": 0.00017924528301886794,
+      "epoch": 0.11
+    },
+    {
+      "step": 777,
+      "timestamp": "2025-12-28T09:27:39.454731",
+      "elapsed_time": 1773.7398252487183,
+      "loss": 0.2219,
+      "grad_norm": 0.16964372992515564,
+      "learning_rate": 0.00017911949685534592,
+      "epoch": 0.110625
+    },
+    {
+      "step": 778,
+      "timestamp": "2025-12-28T09:27:44.743958",
+      "elapsed_time": 1779.0290484428406,
+      "loss": 0.2554,
+      "grad_norm": 0.17479604482650757,
+      "learning_rate": 0.00017899371069182393,
+      "epoch": 0.11125
+    },
+    {
+      "step": 779,
+      "timestamp": "2025-12-28T09:27:53.377780",
+      "elapsed_time": 1787.6628749370575,
+      "loss": 0.1559,
+      "grad_norm": 0.12139801681041718,
+      "learning_rate": 0.0001788679245283019,
+      "epoch": 0.111875
+    },
+    {
+      "step": 780,
+      "timestamp": "2025-12-28T09:28:09.536850",
+      "elapsed_time": 1803.821940422058,
+      "loss": 0.1215,
+      "grad_norm": 0.11527290940284729,
+      "learning_rate": 0.00017874213836477988,
+      "epoch": 0.1125
+    },
+    {
+      "step": 781,
+      "timestamp": "2025-12-28T09:28:16.916052",
+      "elapsed_time": 1811.2011427879333,
+      "loss": 0.191,
+      "grad_norm": 0.2477898895740509,
+      "learning_rate": 0.00017861635220125786,
+      "epoch": 0.113125
+    },
+    {
+      "step": 782,
+      "timestamp": "2025-12-28T09:28:24.055187",
+      "elapsed_time": 1818.3402771949768,
+      "loss": 0.2501,
+      "grad_norm": 0.18624109029769897,
+      "learning_rate": 0.00017849056603773587,
+      "epoch": 0.11375
+    },
+    {
+      "step": 783,
+      "timestamp": "2025-12-28T09:28:35.480491",
+      "elapsed_time": 1829.7655820846558,
+      "loss": 0.1427,
+      "grad_norm": 0.10521696507930756,
+      "learning_rate": 0.00017836477987421385,
+      "epoch": 0.114375
+    },
+    {
+      "step": 784,
+      "timestamp": "2025-12-28T09:28:46.488517",
+      "elapsed_time": 1840.7736072540283,
+      "loss": 0.153,
+      "grad_norm": 0.14191673696041107,
+      "learning_rate": 0.00017823899371069183,
+      "epoch": 0.115
+    },
+    {
+      "step": 785,
+      "timestamp": "2025-12-28T09:28:57.330380",
+      "elapsed_time": 1851.615470647812,
+      "loss": 0.1657,
+      "grad_norm": 0.11947259306907654,
+      "learning_rate": 0.00017811320754716983,
+      "epoch": 0.115625
+    },
+    {
+      "step": 786,
+      "timestamp": "2025-12-28T09:29:02.999607",
+      "elapsed_time": 1857.2846972942352,
+      "loss": 0.3162,
+      "grad_norm": 0.20225848257541656,
+      "learning_rate": 0.0001779874213836478,
+      "epoch": 0.11625
+    },
+    {
+      "step": 787,
+      "timestamp": "2025-12-28T09:29:14.010140",
+      "elapsed_time": 1868.29523062706,
+      "loss": 0.199,
+      "grad_norm": 0.20261730253696442,
+      "learning_rate": 0.0001778616352201258,
+      "epoch": 0.116875
+    },
+    {
+      "step": 788,
+      "timestamp": "2025-12-28T09:29:27.747132",
+      "elapsed_time": 1882.0322229862213,
+      "loss": 0.1879,
+      "grad_norm": 0.1322057545185089,
+      "learning_rate": 0.0001777358490566038,
+      "epoch": 0.1175
+    },
+    {
+      "step": 789,
+      "timestamp": "2025-12-28T09:29:32.947439",
+      "elapsed_time": 1887.2325296401978,
+      "loss": 0.2144,
+      "grad_norm": 0.1755441427230835,
+      "learning_rate": 0.00017761006289308178,
+      "epoch": 0.118125
+    },
+    {
+      "step": 790,
+      "timestamp": "2025-12-28T09:29:43.515994",
+      "elapsed_time": 1897.8010845184326,
+      "loss": 0.1669,
+      "grad_norm": 0.12597765028476715,
+      "learning_rate": 0.00017748427672955976,
+      "epoch": 0.11875
+    },
+    {
+      "step": 791,
+      "timestamp": "2025-12-28T09:29:49.003240",
+      "elapsed_time": 1903.288330078125,
+      "loss": 0.3016,
+      "grad_norm": 0.17289644479751587,
+      "learning_rate": 0.00017735849056603776,
+      "epoch": 0.119375
+    },
+    {
+      "step": 792,
+      "timestamp": "2025-12-28T09:29:57.868876",
+      "elapsed_time": 1912.153966665268,
+      "loss": 0.2346,
+      "grad_norm": 0.13611753284931183,
+      "learning_rate": 0.00017723270440251574,
+      "epoch": 0.12
+    },
+    {
+      "step": 793,
+      "timestamp": "2025-12-28T09:30:09.439964",
+      "elapsed_time": 1923.7250542640686,
+      "loss": 0.1734,
+      "grad_norm": 0.1182570829987526,
+      "learning_rate": 0.00017710691823899372,
+      "epoch": 0.120625
+    },
+    {
+      "step": 794,
+      "timestamp": "2025-12-28T09:30:21.281246",
+      "elapsed_time": 1935.5663363933563,
+      "loss": 0.2104,
+      "grad_norm": 0.16117580235004425,
+      "learning_rate": 0.0001769811320754717,
+      "epoch": 0.12125
+    },
+    {
+      "step": 795,
+      "timestamp": "2025-12-28T09:30:35.176733",
+      "elapsed_time": 1949.4618237018585,
+      "loss": 0.2394,
+      "grad_norm": 0.12246831506490707,
+      "learning_rate": 0.0001768553459119497,
+      "epoch": 0.121875
+    },
+    {
+      "step": 796,
+      "timestamp": "2025-12-28T09:30:45.890549",
+      "elapsed_time": 1960.1756389141083,
+      "loss": 0.222,
+      "grad_norm": 0.14199329912662506,
+      "learning_rate": 0.00017672955974842768,
+      "epoch": 0.1225
+    },
+    {
+      "step": 797,
+      "timestamp": "2025-12-28T09:30:56.848050",
+      "elapsed_time": 1971.1331400871277,
+      "loss": 0.1856,
+      "grad_norm": 0.15696978569030762,
+      "learning_rate": 0.00017660377358490566,
+      "epoch": 0.123125
+    },
+    {
+      "step": 798,
+      "timestamp": "2025-12-28T09:31:05.711194",
+      "elapsed_time": 1979.9962842464447,
+      "loss": 0.1754,
+      "grad_norm": 0.12661588191986084,
+      "learning_rate": 0.00017647798742138367,
+      "epoch": 0.12375
+    },
+    {
+      "step": 799,
+      "timestamp": "2025-12-28T09:31:14.996719",
+      "elapsed_time": 1989.2818095684052,
+      "loss": 0.202,
+      "grad_norm": 0.1417084038257599,
+      "learning_rate": 0.00017635220125786165,
+      "epoch": 0.124375
+    },
+    {
+      "step": 800,
+      "timestamp": "2025-12-28T09:31:23.000694",
+      "elapsed_time": 1997.285789012909,
+      "loss": 0.2166,
+      "grad_norm": 0.13480763137340546,
+      "learning_rate": 0.00017622641509433963,
+      "epoch": 0.125
+    },
+    {
+      "step": 801,
+      "timestamp": "2025-12-28T09:31:35.762309",
+      "elapsed_time": 2010.0473990440369,
+      "loss": 0.1369,
+      "grad_norm": 0.45576173067092896,
+      "learning_rate": 0.00017610062893081763,
+      "epoch": 0.125625
+    },
+    {
+      "step": 802,
+      "timestamp": "2025-12-28T09:31:43.012873",
+      "elapsed_time": 2017.2979636192322,
+      "loss": 0.2186,
+      "grad_norm": 0.1421194076538086,
+      "learning_rate": 0.0001759748427672956,
+      "epoch": 0.12625
+    },
+    {
+      "step": 803,
+      "timestamp": "2025-12-28T09:32:02.199762",
+      "elapsed_time": 2036.484852552414,
+      "loss": 0.1557,
+      "grad_norm": 0.08416479825973511,
+      "learning_rate": 0.0001758490566037736,
+      "epoch": 0.126875
+    },
+    {
+      "step": 804,
+      "timestamp": "2025-12-28T09:32:07.151235",
+      "elapsed_time": 2041.4363248348236,
+      "loss": 0.2531,
+      "grad_norm": 0.17905178666114807,
+      "learning_rate": 0.0001757232704402516,
+      "epoch": 0.1275
+    },
+    {
+      "step": 805,
+      "timestamp": "2025-12-28T09:32:15.632157",
+      "elapsed_time": 2049.9172480106354,
+      "loss": 0.1335,
+      "grad_norm": 0.1256638765335083,
+      "learning_rate": 0.00017559748427672958,
+      "epoch": 0.128125
+    },
+    {
+      "step": 806,
+      "timestamp": "2025-12-28T09:32:31.298185",
+      "elapsed_time": 2065.583275079727,
+      "loss": 0.1427,
+      "grad_norm": 0.0954441949725151,
+      "learning_rate": 0.00017547169811320756,
+      "epoch": 0.12875
+    },
+    {
+      "step": 807,
+      "timestamp": "2025-12-28T09:32:39.447526",
+      "elapsed_time": 2073.7326169013977,
+      "loss": 0.1878,
+      "grad_norm": 0.13530845940113068,
+      "learning_rate": 0.00017534591194968554,
+      "epoch": 0.129375
+    },
+    {
+      "step": 808,
+      "timestamp": "2025-12-28T09:32:50.466826",
+      "elapsed_time": 2084.751916408539,
+      "loss": 0.1094,
+      "grad_norm": 0.1460101455450058,
+      "learning_rate": 0.00017522012578616354,
+      "epoch": 0.13
+    },
+    {
+      "step": 809,
+      "timestamp": "2025-12-28T09:33:11.356705",
+      "elapsed_time": 2105.6417951583862,
+      "loss": 0.1233,
+      "grad_norm": 0.09432196617126465,
+      "learning_rate": 0.00017509433962264152,
+      "epoch": 0.130625
+    },
+    {
+      "step": 810,
+      "timestamp": "2025-12-28T09:33:18.523039",
+      "elapsed_time": 2112.808129787445,
+      "loss": 0.2119,
+      "grad_norm": 0.15918022394180298,
+      "learning_rate": 0.0001749685534591195,
+      "epoch": 0.13125
+    },
+    {
+      "step": 811,
+      "timestamp": "2025-12-28T09:33:26.829092",
+      "elapsed_time": 2121.114182472229,
+      "loss": 0.1979,
+      "grad_norm": 0.14457164704799652,
+      "learning_rate": 0.0001748427672955975,
+      "epoch": 0.131875
+    },
+    {
+      "step": 812,
+      "timestamp": "2025-12-28T09:33:32.394758",
+      "elapsed_time": 2126.6798486709595,
+      "loss": 0.3052,
+      "grad_norm": 0.19710645079612732,
+      "learning_rate": 0.00017471698113207549,
+      "epoch": 0.1325
+    },
+    {
+      "step": 813,
+      "timestamp": "2025-12-28T09:33:45.262163",
+      "elapsed_time": 2139.5472536087036,
+      "loss": 0.1832,
+      "grad_norm": 0.17278793454170227,
+      "learning_rate": 0.00017459119496855346,
+      "epoch": 0.133125
+    },
+    {
+      "step": 814,
+      "timestamp": "2025-12-28T09:33:55.968938",
+      "elapsed_time": 2150.254028081894,
+      "loss": 0.1271,
+      "grad_norm": 0.10092293471097946,
+      "learning_rate": 0.00017446540880503147,
+      "epoch": 0.13375
+    },
+    {
+      "step": 815,
+      "timestamp": "2025-12-28T09:34:03.945511",
+      "elapsed_time": 2158.23060131073,
+      "loss": 0.1929,
+      "grad_norm": 0.1362391710281372,
+      "learning_rate": 0.00017433962264150945,
+      "epoch": 0.134375
+    },
+    {
+      "step": 816,
+      "timestamp": "2025-12-28T09:34:17.206545",
+      "elapsed_time": 2171.4916355609894,
+      "loss": 0.1908,
+      "grad_norm": 0.10968425869941711,
+      "learning_rate": 0.00017421383647798743,
+      "epoch": 0.135
+    },
+    {
+      "step": 817,
+      "timestamp": "2025-12-28T09:34:23.275224",
+      "elapsed_time": 2177.5603144168854,
+      "loss": 0.2587,
+      "grad_norm": 0.17565354704856873,
+      "learning_rate": 0.00017408805031446543,
+      "epoch": 0.135625
+    },
+    {
+      "step": 818,
+      "timestamp": "2025-12-28T09:34:30.707329",
+      "elapsed_time": 2184.9924190044403,
+      "loss": 0.4115,
+      "grad_norm": 0.21891839802265167,
+      "learning_rate": 0.00017396226415094341,
+      "epoch": 0.13625
+    },
+    {
+      "step": 819,
+      "timestamp": "2025-12-28T09:34:40.426629",
+      "elapsed_time": 2194.7117190361023,
+      "loss": 0.1647,
+      "grad_norm": 0.1077812984585762,
+      "learning_rate": 0.0001738364779874214,
+      "epoch": 0.136875
+    },
+    {
+      "step": 820,
+      "timestamp": "2025-12-28T09:34:51.747535",
+      "elapsed_time": 2206.032625436783,
+      "loss": 0.1475,
+      "grad_norm": 0.14761823415756226,
+      "learning_rate": 0.00017371069182389937,
+      "epoch": 0.1375
+    },
+    {
+      "step": 821,
+      "timestamp": "2025-12-28T09:34:55.071621",
+      "elapsed_time": 2209.356710910797,
+      "loss": 0.3967,
+      "grad_norm": 0.24036414921283722,
+      "learning_rate": 0.00017358490566037738,
+      "epoch": 0.138125
+    },
+    {
+      "step": 822,
+      "timestamp": "2025-12-28T09:35:15.970127",
+      "elapsed_time": 2230.255217552185,
+      "loss": 0.0833,
+      "grad_norm": 0.09146321564912796,
+      "learning_rate": 0.00017345911949685536,
+      "epoch": 0.13875
+    },
+    {
+      "step": 823,
+      "timestamp": "2025-12-28T09:35:26.728440",
+      "elapsed_time": 2241.0135345458984,
+      "loss": 0.2568,
+      "grad_norm": 0.1330772191286087,
+      "learning_rate": 0.00017333333333333334,
+      "epoch": 0.139375
+    },
+    {
+      "step": 824,
+      "timestamp": "2025-12-28T09:35:35.721884",
+      "elapsed_time": 2250.006974697113,
+      "loss": 0.3728,
+      "grad_norm": 0.1559283286333084,
+      "learning_rate": 0.00017320754716981134,
+      "epoch": 0.14
+    },
+    {
+      "step": 825,
+      "timestamp": "2025-12-28T09:35:44.080694",
+      "elapsed_time": 2258.3657846450806,
+      "loss": 0.1528,
+      "grad_norm": 0.12386928498744965,
+      "learning_rate": 0.00017308176100628932,
+      "epoch": 0.140625
+    },
+    {
+      "step": 826,
+      "timestamp": "2025-12-28T09:35:57.707994",
+      "elapsed_time": 2271.9930849075317,
+      "loss": 0.1711,
+      "grad_norm": 0.10447162389755249,
+      "learning_rate": 0.0001729559748427673,
+      "epoch": 0.14125
+    },
+    {
+      "step": 827,
+      "timestamp": "2025-12-28T09:36:13.573665",
+      "elapsed_time": 2287.8587548732758,
+      "loss": 0.1396,
+      "grad_norm": 0.09571991115808487,
+      "learning_rate": 0.0001728301886792453,
+      "epoch": 0.141875
+    },
+    {
+      "step": 828,
+      "timestamp": "2025-12-28T09:36:20.525116",
+      "elapsed_time": 2294.810206890106,
+      "loss": 0.192,
+      "grad_norm": 0.1702854335308075,
+      "learning_rate": 0.00017270440251572329,
+      "epoch": 0.1425
+    },
+    {
+      "step": 829,
+      "timestamp": "2025-12-28T09:36:31.229422",
+      "elapsed_time": 2305.514511823654,
+      "loss": 0.1648,
+      "grad_norm": 0.11315345764160156,
+      "learning_rate": 0.00017257861635220126,
+      "epoch": 0.143125
+    },
+    {
+      "step": 830,
+      "timestamp": "2025-12-28T09:36:42.076098",
+      "elapsed_time": 2316.3611884117126,
+      "loss": 0.176,
+      "grad_norm": 0.12335264682769775,
+      "learning_rate": 0.00017245283018867927,
+      "epoch": 0.14375
+    },
+    {
+      "step": 831,
+      "timestamp": "2025-12-28T09:36:51.299215",
+      "elapsed_time": 2325.5843057632446,
+      "loss": 0.2268,
+      "grad_norm": 0.2178533375263214,
+      "learning_rate": 0.00017232704402515725,
+      "epoch": 0.144375
+    },
+    {
+      "step": 832,
+      "timestamp": "2025-12-28T09:37:04.287014",
+      "elapsed_time": 2338.5721044540405,
+      "loss": 0.1242,
+      "grad_norm": 0.11771131306886673,
+      "learning_rate": 0.00017220125786163523,
+      "epoch": 0.145
+    },
+    {
+      "step": 833,
+      "timestamp": "2025-12-28T09:37:12.634424",
+      "elapsed_time": 2346.919515132904,
+      "loss": 0.1969,
+      "grad_norm": 0.13342277705669403,
+      "learning_rate": 0.00017207547169811324,
+      "epoch": 0.145625
+    },
+    {
+      "step": 834,
+      "timestamp": "2025-12-28T09:37:24.771648",
+      "elapsed_time": 2359.0567383766174,
+      "loss": 0.1777,
+      "grad_norm": 0.10993700474500656,
+      "learning_rate": 0.00017194968553459121,
+      "epoch": 0.14625
+    },
+    {
+      "step": 835,
+      "timestamp": "2025-12-28T09:37:32.205060",
+      "elapsed_time": 2366.49015045166,
+      "loss": 0.2091,
+      "grad_norm": 0.14296327531337738,
+      "learning_rate": 0.0001718238993710692,
+      "epoch": 0.146875
+    },
+    {
+      "step": 836,
+      "timestamp": "2025-12-28T09:37:40.267010",
+      "elapsed_time": 2374.552100419998,
+      "loss": 0.2618,
+      "grad_norm": 0.18104122579097748,
+      "learning_rate": 0.00017169811320754717,
+      "epoch": 0.1475
+    },
+    {
+      "step": 837,
+      "timestamp": "2025-12-28T09:37:56.881841",
+      "elapsed_time": 2391.166932106018,
+      "loss": 0.1163,
+      "grad_norm": 0.11203698068857193,
+      "learning_rate": 0.00017157232704402518,
+      "epoch": 0.148125
+    },
+    {
+      "step": 838,
+      "timestamp": "2025-12-28T09:38:15.789255",
+      "elapsed_time": 2410.0743453502655,
+      "loss": 0.1306,
+      "grad_norm": 0.12151342630386353,
+      "learning_rate": 0.00017144654088050316,
+      "epoch": 0.14875
+    },
+    {
+      "step": 839,
+      "timestamp": "2025-12-28T09:38:22.694989",
+      "elapsed_time": 2416.9800794124603,
+      "loss": 0.1991,
+      "grad_norm": 0.14583951234817505,
+      "learning_rate": 0.00017132075471698114,
+      "epoch": 0.149375
+    },
+    {
+      "step": 840,
+      "timestamp": "2025-12-28T09:38:32.405677",
+      "elapsed_time": 2426.6907675266266,
+      "loss": 0.5209,
+      "grad_norm": 0.1858104020357132,
+      "learning_rate": 0.00017119496855345914,
+      "epoch": 0.15
+    },
+    {
+      "step": 841,
+      "timestamp": "2025-12-28T09:38:41.693108",
+      "elapsed_time": 2435.97819852829,
+      "loss": 0.2115,
+      "grad_norm": 0.12431403249502182,
+      "learning_rate": 0.00017106918238993712,
+      "epoch": 0.150625
+    },
+    {
+      "step": 842,
+      "timestamp": "2025-12-28T09:38:49.200296",
+      "elapsed_time": 2443.4853858947754,
+      "loss": 0.2078,
+      "grad_norm": 0.1477956473827362,
+      "learning_rate": 0.0001709433962264151,
+      "epoch": 0.15125
+    },
+    {
+      "step": 843,
+      "timestamp": "2025-12-28T09:38:55.712143",
+      "elapsed_time": 2449.9972331523895,
+      "loss": 0.3907,
+      "grad_norm": 0.184353306889534,
+      "learning_rate": 0.0001708176100628931,
+      "epoch": 0.151875
+    },
+    {
+      "step": 844,
+      "timestamp": "2025-12-28T09:38:58.933553",
+      "elapsed_time": 2453.218643426895,
+      "loss": 0.2667,
+      "grad_norm": 0.262251079082489,
+      "learning_rate": 0.0001706918238993711,
+      "epoch": 0.1525
+    },
+    {
+      "step": 845,
+      "timestamp": "2025-12-28T09:39:05.378036",
+      "elapsed_time": 2459.663126707077,
+      "loss": 0.1593,
+      "grad_norm": 0.12699125707149506,
+      "learning_rate": 0.00017056603773584907,
+      "epoch": 0.153125
+    },
+    {
+      "step": 846,
+      "timestamp": "2025-12-28T09:39:15.452772",
+      "elapsed_time": 2469.737862586975,
+      "loss": 0.2266,
+      "grad_norm": 0.134856179356575,
+      "learning_rate": 0.00017044025157232707,
+      "epoch": 0.15375
+    },
+    {
+      "step": 847,
+      "timestamp": "2025-12-28T09:39:29.349192",
+      "elapsed_time": 2483.63428235054,
+      "loss": 0.156,
+      "grad_norm": 0.09759867191314697,
+      "learning_rate": 0.00017031446540880505,
+      "epoch": 0.154375
+    },
+    {
+      "step": 848,
+      "timestamp": "2025-12-28T09:39:35.325679",
+      "elapsed_time": 2489.6107692718506,
+      "loss": 0.1556,
+      "grad_norm": 0.13298504054546356,
+      "learning_rate": 0.00017018867924528303,
+      "epoch": 0.155
+    },
+    {
+      "step": 849,
+      "timestamp": "2025-12-28T09:39:44.060500",
+      "elapsed_time": 2498.345590353012,
+      "loss": 0.1904,
+      "grad_norm": 0.11852674186229706,
+      "learning_rate": 0.000170062893081761,
+      "epoch": 0.155625
+    },
+    {
+      "step": 850,
+      "timestamp": "2025-12-28T09:39:54.425345",
+      "elapsed_time": 2508.710435152054,
+      "loss": 0.1842,
+      "grad_norm": 0.22243794798851013,
+      "learning_rate": 0.00016993710691823902,
+      "epoch": 0.15625
+    },
+    {
+      "step": 851,
+      "timestamp": "2025-12-28T09:40:05.019054",
+      "elapsed_time": 2519.3041446208954,
+      "loss": 0.1675,
+      "grad_norm": 0.12315836548805237,
+      "learning_rate": 0.000169811320754717,
+      "epoch": 0.156875
+    },
+    {
+      "step": 852,
+      "timestamp": "2025-12-28T09:40:18.805001",
+      "elapsed_time": 2533.0900909900665,
+      "loss": 0.186,
+      "grad_norm": 0.1145598366856575,
+      "learning_rate": 0.00016968553459119497,
+      "epoch": 0.1575
+    },
+    {
+      "step": 853,
+      "timestamp": "2025-12-28T09:40:26.649585",
+      "elapsed_time": 2540.9346754550934,
+      "loss": 0.2299,
+      "grad_norm": 0.15518611669540405,
+      "learning_rate": 0.00016955974842767298,
+      "epoch": 0.158125
+    },
+    {
+      "step": 854,
+      "timestamp": "2025-12-28T09:40:34.119488",
+      "elapsed_time": 2548.4045779705048,
+      "loss": 0.2426,
+      "grad_norm": 0.14574050903320312,
+      "learning_rate": 0.00016943396226415096,
+      "epoch": 0.15875
+    },
+    {
+      "step": 855,
+      "timestamp": "2025-12-28T09:40:40.131033",
+      "elapsed_time": 2554.416123867035,
+      "loss": 0.2763,
+      "grad_norm": 0.1946178823709488,
+      "learning_rate": 0.00016930817610062894,
+      "epoch": 0.159375
+    },
+    {
+      "step": 856,
+      "timestamp": "2025-12-28T09:40:48.080229",
+      "elapsed_time": 2562.3653190135956,
+      "loss": 0.4301,
+      "grad_norm": 0.16829311847686768,
+      "learning_rate": 0.00016918238993710694,
+      "epoch": 0.16
+    },
+    {
+      "step": 857,
+      "timestamp": "2025-12-28T09:41:03.320074",
+      "elapsed_time": 2577.6051642894745,
+      "loss": 0.1631,
+      "grad_norm": 0.19184663891792297,
+      "learning_rate": 0.00016905660377358492,
+      "epoch": 0.160625
+    },
+    {
+      "step": 858,
+      "timestamp": "2025-12-28T09:41:12.493085",
+      "elapsed_time": 2586.7781751155853,
+      "loss": 0.1824,
+      "grad_norm": 0.125960111618042,
+      "learning_rate": 0.0001689308176100629,
+      "epoch": 0.16125
+    },
+    {
+      "step": 859,
+      "timestamp": "2025-12-28T09:41:21.589208",
+      "elapsed_time": 2595.874298810959,
+      "loss": 0.1654,
+      "grad_norm": 0.12101560086011887,
+      "learning_rate": 0.0001688050314465409,
+      "epoch": 0.161875
+    },
+    {
+      "step": 860,
+      "timestamp": "2025-12-28T09:41:30.728410",
+      "elapsed_time": 2605.0134999752045,
+      "loss": 0.3724,
+      "grad_norm": 0.1491203010082245,
+      "learning_rate": 0.0001686792452830189,
+      "epoch": 0.1625
+    },
+    {
+      "step": 861,
+      "timestamp": "2025-12-28T09:41:40.445296",
+      "elapsed_time": 2614.7303869724274,
+      "loss": 0.1735,
+      "grad_norm": 0.12737800180912018,
+      "learning_rate": 0.00016855345911949687,
+      "epoch": 0.163125
+    },
+    {
+      "step": 862,
+      "timestamp": "2025-12-28T09:41:55.948429",
+      "elapsed_time": 2630.2335200309753,
+      "loss": 0.126,
+      "grad_norm": 0.08794135600328445,
+      "learning_rate": 0.00016842767295597485,
+      "epoch": 0.16375
+    },
+    {
+      "step": 863,
+      "timestamp": "2025-12-28T09:41:59.637376",
+      "elapsed_time": 2633.9224672317505,
+      "loss": 0.302,
+      "grad_norm": 0.2143140286207199,
+      "learning_rate": 0.00016830188679245285,
+      "epoch": 0.164375
+    },
+    {
+      "step": 864,
+      "timestamp": "2025-12-28T09:42:06.917332",
+      "elapsed_time": 2641.202422618866,
+      "loss": 0.2941,
+      "grad_norm": 0.15600278973579407,
+      "learning_rate": 0.00016817610062893083,
+      "epoch": 0.165
+    },
+    {
+      "step": 865,
+      "timestamp": "2025-12-28T09:42:18.849851",
+      "elapsed_time": 2653.134941339493,
+      "loss": 0.1691,
+      "grad_norm": 0.11934607475996017,
+      "learning_rate": 0.0001680503144654088,
+      "epoch": 0.165625
+    },
+    {
+      "step": 866,
+      "timestamp": "2025-12-28T09:42:28.174345",
+      "elapsed_time": 2662.459435939789,
+      "loss": 0.1558,
+      "grad_norm": 0.11288256198167801,
+      "learning_rate": 0.00016792452830188682,
+      "epoch": 0.16625
+    },
+    {
+      "step": 867,
+      "timestamp": "2025-12-28T09:42:39.834999",
+      "elapsed_time": 2674.1200897693634,
+      "loss": 0.1402,
+      "grad_norm": 0.11293160915374756,
+      "learning_rate": 0.0001677987421383648,
+      "epoch": 0.166875
+    },
+    {
+      "step": 868,
+      "timestamp": "2025-12-28T09:42:45.101829",
+      "elapsed_time": 2679.3869194984436,
+      "loss": 0.3336,
+      "grad_norm": 0.21127207577228546,
+      "learning_rate": 0.00016767295597484277,
+      "epoch": 0.1675
+    },
+    {
+      "step": 869,
+      "timestamp": "2025-12-28T09:42:53.397932",
+      "elapsed_time": 2687.683022260666,
+      "loss": 0.2122,
+      "grad_norm": 0.13150571286678314,
+      "learning_rate": 0.00016754716981132078,
+      "epoch": 0.168125
+    },
+    {
+      "step": 870,
+      "timestamp": "2025-12-28T09:43:02.604043",
+      "elapsed_time": 2696.889132976532,
+      "loss": 0.2443,
+      "grad_norm": 0.16025897860527039,
+      "learning_rate": 0.00016742138364779876,
+      "epoch": 0.16875
+    },
+    {
+      "step": 871,
+      "timestamp": "2025-12-28T09:43:16.218580",
+      "elapsed_time": 2710.503670692444,
+      "loss": 0.1361,
+      "grad_norm": 0.1163235530257225,
+      "learning_rate": 0.00016729559748427674,
+      "epoch": 0.169375
+    },
+    {
+      "step": 872,
+      "timestamp": "2025-12-28T09:43:23.734614",
+      "elapsed_time": 2718.01970911026,
+      "loss": 0.2797,
+      "grad_norm": 0.3218298852443695,
+      "learning_rate": 0.00016716981132075474,
+      "epoch": 0.17
+    },
+    {
+      "step": 873,
+      "timestamp": "2025-12-28T09:43:40.199422",
+      "elapsed_time": 2734.4845123291016,
+      "loss": 0.1531,
+      "grad_norm": 0.09483516216278076,
+      "learning_rate": 0.00016704402515723272,
+      "epoch": 0.170625
+    },
+    {
+      "step": 874,
+      "timestamp": "2025-12-28T09:43:46.722337",
+      "elapsed_time": 2741.007427930832,
+      "loss": 0.1846,
+      "grad_norm": 0.13699615001678467,
+      "learning_rate": 0.0001669182389937107,
+      "epoch": 0.17125
+    },
+    {
+      "step": 875,
+      "timestamp": "2025-12-28T09:44:01.014473",
+      "elapsed_time": 2755.2995631694794,
+      "loss": 0.2334,
+      "grad_norm": 0.14246909320354462,
+      "learning_rate": 0.00016679245283018868,
+      "epoch": 0.171875
+    },
+    {
+      "step": 876,
+      "timestamp": "2025-12-28T09:44:13.506506",
+      "elapsed_time": 2767.7915959358215,
+      "loss": 0.1595,
+      "grad_norm": 0.11375102400779724,
+      "learning_rate": 0.0001666666666666667,
+      "epoch": 0.1725
+    },
+    {
+      "step": 877,
+      "timestamp": "2025-12-28T09:44:19.673893",
+      "elapsed_time": 2773.9589836597443,
+      "loss": 0.2133,
+      "grad_norm": 0.1539752185344696,
+      "learning_rate": 0.00016654088050314467,
+      "epoch": 0.173125
+    },
+    {
+      "step": 878,
+      "timestamp": "2025-12-28T09:44:31.607564",
+      "elapsed_time": 2785.8926544189453,
+      "loss": 0.3027,
+      "grad_norm": 0.13873888552188873,
+      "learning_rate": 0.00016641509433962265,
+      "epoch": 0.17375
+    },
+    {
+      "step": 879,
+      "timestamp": "2025-12-28T09:44:36.488223",
+      "elapsed_time": 2790.773313522339,
+      "loss": 0.2748,
+      "grad_norm": 0.17139685153961182,
+      "learning_rate": 0.00016628930817610065,
+      "epoch": 0.174375
+    },
+    {
+      "step": 880,
+      "timestamp": "2025-12-28T09:44:44.545212",
+      "elapsed_time": 2798.83030295372,
+      "loss": 0.192,
+      "grad_norm": 0.14361582696437836,
+      "learning_rate": 0.00016616352201257863,
+      "epoch": 0.175
+    },
+    {
+      "step": 881,
+      "timestamp": "2025-12-28T09:44:52.364196",
+      "elapsed_time": 2806.6492867469788,
+      "loss": 0.2773,
+      "grad_norm": 0.18555279076099396,
+      "learning_rate": 0.0001660377358490566,
+      "epoch": 0.175625
+    },
+    {
+      "step": 882,
+      "timestamp": "2025-12-28T09:44:58.645574",
+      "elapsed_time": 2812.9306647777557,
+      "loss": 0.3019,
+      "grad_norm": 0.1758921891450882,
+      "learning_rate": 0.00016591194968553462,
+      "epoch": 0.17625
+    },
+    {
+      "step": 883,
+      "timestamp": "2025-12-28T09:45:09.067925",
+      "elapsed_time": 2823.353015899658,
+      "loss": 0.3553,
+      "grad_norm": 0.15257853269577026,
+      "learning_rate": 0.0001657861635220126,
+      "epoch": 0.176875
+    },
+    {
+      "step": 884,
+      "timestamp": "2025-12-28T09:45:13.656005",
+      "elapsed_time": 2827.94109582901,
+      "loss": 0.2704,
+      "grad_norm": 0.17811451852321625,
+      "learning_rate": 0.00016566037735849058,
+      "epoch": 0.1775
+    },
+    {
+      "step": 885,
+      "timestamp": "2025-12-28T09:45:19.215450",
+      "elapsed_time": 2833.500540494919,
+      "loss": 0.1815,
+      "grad_norm": 0.1703587919473648,
+      "learning_rate": 0.00016553459119496858,
+      "epoch": 0.178125
+    },
+    {
+      "step": 886,
+      "timestamp": "2025-12-28T09:45:24.676685",
+      "elapsed_time": 2838.9617760181427,
+      "loss": 0.4395,
+      "grad_norm": 0.2093503326177597,
+      "learning_rate": 0.00016540880503144656,
+      "epoch": 0.17875
+    },
+    {
+      "step": 887,
+      "timestamp": "2025-12-28T09:45:29.125709",
+      "elapsed_time": 2843.410799264908,
+      "loss": 0.2659,
+      "grad_norm": 0.1815217286348343,
+      "learning_rate": 0.00016528301886792454,
+      "epoch": 0.179375
+    },
+    {
+      "step": 888,
+      "timestamp": "2025-12-28T09:45:35.113254",
+      "elapsed_time": 2849.3983447551727,
+      "loss": 0.2541,
+      "grad_norm": 0.18482376635074615,
+      "learning_rate": 0.00016515723270440252,
+      "epoch": 0.18
+    },
+    {
+      "step": 889,
+      "timestamp": "2025-12-28T09:45:41.738048",
+      "elapsed_time": 2856.023138523102,
+      "loss": 0.1837,
+      "grad_norm": 0.15669426321983337,
+      "learning_rate": 0.00016503144654088052,
+      "epoch": 0.180625
+    },
+    {
+      "step": 890,
+      "timestamp": "2025-12-28T09:45:50.170533",
+      "elapsed_time": 2864.455623626709,
+      "loss": 0.2074,
+      "grad_norm": 0.16513699293136597,
+      "learning_rate": 0.0001649056603773585,
+      "epoch": 0.18125
+    },
+    {
+      "step": 891,
+      "timestamp": "2025-12-28T09:45:58.910483",
+      "elapsed_time": 2873.1955733299255,
+      "loss": 0.1612,
+      "grad_norm": 0.1310487985610962,
+      "learning_rate": 0.00016477987421383648,
+      "epoch": 0.181875
+    },
+    {
+      "step": 892,
+      "timestamp": "2025-12-28T09:46:08.665921",
+      "elapsed_time": 2882.9510111808777,
+      "loss": 0.1793,
+      "grad_norm": 0.1385423094034195,
+      "learning_rate": 0.0001646540880503145,
+      "epoch": 0.1825
+    },
+    {
+      "step": 893,
+      "timestamp": "2025-12-28T09:46:28.595204",
+      "elapsed_time": 2902.8802947998047,
+      "loss": 0.1056,
+      "grad_norm": 0.09160617738962173,
+      "learning_rate": 0.00016452830188679247,
+      "epoch": 0.183125
+    },
+    {
+      "step": 894,
+      "timestamp": "2025-12-28T09:46:37.456263",
+      "elapsed_time": 2911.7413532733917,
+      "loss": 0.2347,
+      "grad_norm": 0.1339615136384964,
+      "learning_rate": 0.00016440251572327045,
+      "epoch": 0.18375
+    },
+    {
+      "step": 895,
+      "timestamp": "2025-12-28T09:46:43.090133",
+      "elapsed_time": 2917.3752233982086,
+      "loss": 0.2097,
+      "grad_norm": 0.1673922836780548,
+      "learning_rate": 0.00016427672955974845,
+      "epoch": 0.184375
+    },
+    {
+      "step": 896,
+      "timestamp": "2025-12-28T09:46:48.742520",
+      "elapsed_time": 2923.0276103019714,
+      "loss": 0.2041,
+      "grad_norm": 0.17525748908519745,
+      "learning_rate": 0.00016415094339622643,
+      "epoch": 0.185
+    },
+    {
+      "step": 897,
+      "timestamp": "2025-12-28T09:46:56.108721",
+      "elapsed_time": 2930.3938117027283,
+      "loss": 0.1659,
+      "grad_norm": 0.15093988180160522,
+      "learning_rate": 0.0001640251572327044,
+      "epoch": 0.185625
+    },
+    {
+      "step": 898,
+      "timestamp": "2025-12-28T09:47:09.846290",
+      "elapsed_time": 2944.1313841342926,
+      "loss": 0.1298,
+      "grad_norm": 0.09404166042804718,
+      "learning_rate": 0.00016389937106918242,
+      "epoch": 0.18625
+    },
+    {
+      "step": 899,
+      "timestamp": "2025-12-28T09:47:21.835574",
+      "elapsed_time": 2956.120668411255,
+      "loss": 0.1425,
+      "grad_norm": 0.10061768442392349,
+      "learning_rate": 0.0001637735849056604,
+      "epoch": 0.186875
+    },
+    {
+      "step": 900,
+      "timestamp": "2025-12-28T09:47:32.792316",
+      "elapsed_time": 2967.0774064064026,
+      "loss": 0.124,
+      "grad_norm": 0.09524894505739212,
+      "learning_rate": 0.00016364779874213838,
+      "epoch": 0.1875
+    },
+    {
+      "step": 901,
+      "timestamp": "2025-12-28T09:47:44.729124",
+      "elapsed_time": 2979.0142147541046,
+      "loss": 0.1498,
+      "grad_norm": 2.4628686904907227,
+      "learning_rate": 0.00016352201257861635,
+      "epoch": 0.188125
+    },
+    {
+      "step": 902,
+      "timestamp": "2025-12-28T09:47:52.982666",
+      "elapsed_time": 2987.267756462097,
+      "loss": 0.2408,
+      "grad_norm": 0.13900604844093323,
+      "learning_rate": 0.00016339622641509436,
+      "epoch": 0.18875
+    },
+    {
+      "step": 903,
+      "timestamp": "2025-12-28T09:48:00.120207",
+      "elapsed_time": 2994.4052975177765,
+      "loss": 0.2353,
+      "grad_norm": 0.15295051038265228,
+      "learning_rate": 0.00016327044025157234,
+      "epoch": 0.189375
+    },
+    {
+      "step": 904,
+      "timestamp": "2025-12-28T09:48:08.263388",
+      "elapsed_time": 3002.5484788417816,
+      "loss": 0.2088,
+      "grad_norm": 0.14227791130542755,
+      "learning_rate": 0.00016314465408805032,
+      "epoch": 0.19
+    },
+    {
+      "step": 905,
+      "timestamp": "2025-12-28T09:48:21.400428",
+      "elapsed_time": 3015.6855177879333,
+      "loss": 0.1279,
+      "grad_norm": 0.13115034997463226,
+      "learning_rate": 0.00016301886792452833,
+      "epoch": 0.190625
+    },
+    {
+      "step": 906,
+      "timestamp": "2025-12-28T09:48:30.333067",
+      "elapsed_time": 3024.618157148361,
+      "loss": 0.1772,
+      "grad_norm": 0.13131241500377655,
+      "learning_rate": 0.0001628930817610063,
+      "epoch": 0.19125
+    },
+    {
+      "step": 907,
+      "timestamp": "2025-12-28T09:48:39.473331",
+      "elapsed_time": 3033.7584216594696,
+      "loss": 0.1563,
+      "grad_norm": 0.11939848214387894,
+      "learning_rate": 0.00016276729559748428,
+      "epoch": 0.191875
+    },
+    {
+      "step": 908,
+      "timestamp": "2025-12-28T09:48:55.692279",
+      "elapsed_time": 3049.9773693084717,
+      "loss": 0.1491,
+      "grad_norm": 0.14415167272090912,
+      "learning_rate": 0.0001626415094339623,
+      "epoch": 0.1925
+    },
+    {
+      "step": 909,
+      "timestamp": "2025-12-28T09:49:04.121120",
+      "elapsed_time": 3058.406210422516,
+      "loss": 0.1965,
+      "grad_norm": 0.17701658606529236,
+      "learning_rate": 0.00016251572327044027,
+      "epoch": 0.193125
+    },
+    {
+      "step": 910,
+      "timestamp": "2025-12-28T09:49:15.530418",
+      "elapsed_time": 3069.815508365631,
+      "loss": 0.2012,
+      "grad_norm": 0.12482966482639313,
+      "learning_rate": 0.00016238993710691825,
+      "epoch": 0.19375
+    },
+    {
+      "step": 911,
+      "timestamp": "2025-12-28T09:49:21.168387",
+      "elapsed_time": 3075.45347738266,
+      "loss": 0.3984,
+      "grad_norm": 0.17656728625297546,
+      "learning_rate": 0.00016226415094339625,
+      "epoch": 0.194375
+    },
+    {
+      "step": 912,
+      "timestamp": "2025-12-28T09:49:33.856071",
+      "elapsed_time": 3088.141161441803,
+      "loss": 0.131,
+      "grad_norm": 0.0940021201968193,
+      "learning_rate": 0.00016213836477987423,
+      "epoch": 0.195
+    },
+    {
+      "step": 913,
+      "timestamp": "2025-12-28T09:49:45.752863",
+      "elapsed_time": 3100.0379536151886,
+      "loss": 0.165,
+      "grad_norm": 0.19799897074699402,
+      "learning_rate": 0.0001620125786163522,
+      "epoch": 0.195625
+    },
+    {
+      "step": 914,
+      "timestamp": "2025-12-28T09:49:51.223696",
+      "elapsed_time": 3105.50878572464,
+      "loss": 0.2392,
+      "grad_norm": 0.39207378029823303,
+      "learning_rate": 0.0001618867924528302,
+      "epoch": 0.19625
+    },
+    {
+      "step": 915,
+      "timestamp": "2025-12-28T09:50:04.960806",
+      "elapsed_time": 3119.245896577835,
+      "loss": 0.1131,
+      "grad_norm": 0.09941917657852173,
+      "learning_rate": 0.0001617610062893082,
+      "epoch": 0.196875
+    },
+    {
+      "step": 916,
+      "timestamp": "2025-12-28T09:50:12.135355",
+      "elapsed_time": 3126.4204454421997,
+      "loss": 0.1949,
+      "grad_norm": 0.16558235883712769,
+      "learning_rate": 0.00016163522012578618,
+      "epoch": 0.1975
+    },
+    {
+      "step": 917,
+      "timestamp": "2025-12-28T09:50:24.159964",
+      "elapsed_time": 3138.4450545310974,
+      "loss": 0.1532,
+      "grad_norm": 0.1185847669839859,
+      "learning_rate": 0.00016150943396226416,
+      "epoch": 0.198125
+    },
+    {
+      "step": 918,
+      "timestamp": "2025-12-28T09:50:36.336876",
+      "elapsed_time": 3150.6219668388367,
+      "loss": 0.1423,
+      "grad_norm": 0.11089828610420227,
+      "learning_rate": 0.00016138364779874216,
+      "epoch": 0.19875
+    },
+    {
+      "step": 919,
+      "timestamp": "2025-12-28T09:50:42.776184",
+      "elapsed_time": 3157.0612740516663,
+      "loss": 0.3044,
+      "grad_norm": 0.19416838884353638,
+      "learning_rate": 0.00016125786163522014,
+      "epoch": 0.199375
+    },
+    {
+      "step": 920,
+      "timestamp": "2025-12-28T09:50:50.019743",
+      "elapsed_time": 3164.304833650589,
+      "loss": 0.4452,
+      "grad_norm": 0.3824036419391632,
+      "learning_rate": 0.00016113207547169812,
+      "epoch": 0.2
+    },
+    {
+      "step": 921,
+      "timestamp": "2025-12-28T09:50:57.804129",
+      "elapsed_time": 3172.0892198085785,
+      "loss": 0.3216,
+      "grad_norm": 0.16429851949214935,
+      "learning_rate": 0.00016100628930817613,
+      "epoch": 0.200625
+    },
+    {
+      "step": 922,
+      "timestamp": "2025-12-28T09:51:13.052604",
+      "elapsed_time": 3187.337694168091,
+      "loss": 0.1217,
+      "grad_norm": 0.13496516644954681,
+      "learning_rate": 0.0001608805031446541,
+      "epoch": 0.20125
+    },
+    {
+      "step": 923,
+      "timestamp": "2025-12-28T09:51:24.570532",
+      "elapsed_time": 3198.855621814728,
+      "loss": 0.1566,
+      "grad_norm": 0.12458086013793945,
+      "learning_rate": 0.00016075471698113208,
+      "epoch": 0.201875
+    },
+    {
+      "step": 924,
+      "timestamp": "2025-12-28T09:51:31.848741",
+      "elapsed_time": 3206.133831501007,
+      "loss": 0.2315,
+      "grad_norm": 0.14897367358207703,
+      "learning_rate": 0.0001606289308176101,
+      "epoch": 0.2025
+    },
+    {
+      "step": 925,
+      "timestamp": "2025-12-28T09:51:45.734245",
+      "elapsed_time": 3220.019335269928,
+      "loss": 0.111,
+      "grad_norm": 0.09125252813100815,
+      "learning_rate": 0.00016050314465408807,
+      "epoch": 0.203125
+    },
+    {
+      "step": 926,
+      "timestamp": "2025-12-28T09:51:58.601974",
+      "elapsed_time": 3232.8870646953583,
+      "loss": 0.1471,
+      "grad_norm": 0.11020820587873459,
+      "learning_rate": 0.00016037735849056605,
+      "epoch": 0.20375
+    },
+    {
+      "step": 927,
+      "timestamp": "2025-12-28T09:52:06.148283",
+      "elapsed_time": 3240.43337392807,
+      "loss": 0.1875,
+      "grad_norm": 0.12961116433143616,
+      "learning_rate": 0.00016025157232704405,
+      "epoch": 0.204375
+    },
+    {
+      "step": 928,
+      "timestamp": "2025-12-28T09:52:13.614144",
+      "elapsed_time": 3247.899234056473,
+      "loss": 0.1729,
+      "grad_norm": 0.14182396233081818,
+      "learning_rate": 0.00016012578616352203,
+      "epoch": 0.205
+    },
+    {
+      "step": 929,
+      "timestamp": "2025-12-28T09:52:18.583980",
+      "elapsed_time": 3252.8690705299377,
+      "loss": 0.3456,
+      "grad_norm": 0.20903103053569794,
+      "learning_rate": 0.00016,
+      "epoch": 0.205625
+    },
+    {
+      "step": 930,
+      "timestamp": "2025-12-28T09:52:25.638732",
+      "elapsed_time": 3259.9238221645355,
+      "loss": 0.1962,
+      "grad_norm": 0.13587729632854462,
+      "learning_rate": 0.000159874213836478,
+      "epoch": 0.20625
+    },
+    {
+      "step": 931,
+      "timestamp": "2025-12-28T09:52:34.937513",
+      "elapsed_time": 3269.222603082657,
+      "loss": 0.2363,
+      "grad_norm": 0.1760166883468628,
+      "learning_rate": 0.000159748427672956,
+      "epoch": 0.206875
+    },
+    {
+      "step": 932,
+      "timestamp": "2025-12-28T09:52:43.090395",
+      "elapsed_time": 3277.3754856586456,
+      "loss": 0.2063,
+      "grad_norm": 0.13720373809337616,
+      "learning_rate": 0.00015962264150943398,
+      "epoch": 0.2075
+    },
+    {
+      "step": 933,
+      "timestamp": "2025-12-28T09:52:49.474895",
+      "elapsed_time": 3283.7599856853485,
+      "loss": 0.229,
+      "grad_norm": 0.14343424141407013,
+      "learning_rate": 0.00015949685534591196,
+      "epoch": 0.208125
+    },
+    {
+      "step": 934,
+      "timestamp": "2025-12-28T09:53:03.465281",
+      "elapsed_time": 3297.7503714561462,
+      "loss": 0.1478,
+      "grad_norm": 0.15561024844646454,
+      "learning_rate": 0.00015937106918238996,
+      "epoch": 0.20875
+    },
+    {
+      "step": 935,
+      "timestamp": "2025-12-28T09:53:10.342764",
+      "elapsed_time": 3304.627854824066,
+      "loss": 0.2767,
+      "grad_norm": 0.15607592463493347,
+      "learning_rate": 0.00015924528301886794,
+      "epoch": 0.209375
+    },
+    {
+      "step": 936,
+      "timestamp": "2025-12-28T09:53:15.162542",
+      "elapsed_time": 3309.447632074356,
+      "loss": 0.2736,
+      "grad_norm": 0.18114545941352844,
+      "learning_rate": 0.00015911949685534592,
+      "epoch": 0.21
+    },
+    {
+      "step": 937,
+      "timestamp": "2025-12-28T09:53:26.422739",
+      "elapsed_time": 3320.7078297138214,
+      "loss": 0.1715,
+      "grad_norm": 0.11876345425844193,
+      "learning_rate": 0.00015899371069182393,
+      "epoch": 0.210625
+    },
+    {
+      "step": 938,
+      "timestamp": "2025-12-28T09:53:32.911148",
+      "elapsed_time": 3327.1962430477142,
+      "loss": 0.1795,
+      "grad_norm": 0.1508261114358902,
+      "learning_rate": 0.0001588679245283019,
+      "epoch": 0.21125
+    },
+    {
+      "step": 939,
+      "timestamp": "2025-12-28T09:53:38.919335",
+      "elapsed_time": 3333.2044246196747,
+      "loss": 0.1735,
+      "grad_norm": 0.14189061522483826,
+      "learning_rate": 0.00015874213836477989,
+      "epoch": 0.211875
+    },
+    {
+      "step": 940,
+      "timestamp": "2025-12-28T09:53:52.060423",
+      "elapsed_time": 3346.345513343811,
+      "loss": 0.1236,
+      "grad_norm": 0.09674876928329468,
+      "learning_rate": 0.0001586163522012579,
+      "epoch": 0.2125
+    },
+    {
+      "step": 941,
+      "timestamp": "2025-12-28T09:54:12.941873",
+      "elapsed_time": 3367.22696352005,
+      "loss": 0.1143,
+      "grad_norm": 0.10512091219425201,
+      "learning_rate": 0.00015849056603773587,
+      "epoch": 0.213125
+    },
+    {
+      "step": 942,
+      "timestamp": "2025-12-28T09:54:23.471024",
+      "elapsed_time": 3377.756114244461,
+      "loss": 0.1029,
+      "grad_norm": 0.21176813542842865,
+      "learning_rate": 0.00015836477987421385,
+      "epoch": 0.21375
+    },
+    {
+      "step": 943,
+      "timestamp": "2025-12-28T09:54:33.191585",
+      "elapsed_time": 3387.4766755104065,
+      "loss": 0.1839,
+      "grad_norm": 0.12542878091335297,
+      "learning_rate": 0.00015823899371069183,
+      "epoch": 0.214375
+    },
+    {
+      "step": 944,
+      "timestamp": "2025-12-28T09:54:53.926827",
+      "elapsed_time": 3408.2119178771973,
+      "loss": 0.1474,
+      "grad_norm": 0.12502753734588623,
+      "learning_rate": 0.00015811320754716983,
+      "epoch": 0.215
+    },
+    {
+      "step": 945,
+      "timestamp": "2025-12-28T09:55:05.593186",
+      "elapsed_time": 3419.8782770633698,
+      "loss": 0.1523,
+      "grad_norm": 0.11267740279436111,
+      "learning_rate": 0.00015798742138364781,
+      "epoch": 0.215625
+    },
+    {
+      "step": 946,
+      "timestamp": "2025-12-28T09:55:14.970581",
+      "elapsed_time": 3429.2556715011597,
+      "loss": 0.1716,
+      "grad_norm": 0.14452391862869263,
+      "learning_rate": 0.0001578616352201258,
+      "epoch": 0.21625
+    },
+    {
+      "step": 947,
+      "timestamp": "2025-12-28T09:55:23.786812",
+      "elapsed_time": 3438.0719022750854,
+      "loss": 0.2105,
+      "grad_norm": 0.14315542578697205,
+      "learning_rate": 0.0001577358490566038,
+      "epoch": 0.216875
+    },
+    {
+      "step": 948,
+      "timestamp": "2025-12-28T09:55:27.890524",
+      "elapsed_time": 3442.175614595413,
+      "loss": 0.3344,
+      "grad_norm": 0.2096068263053894,
+      "learning_rate": 0.00015761006289308178,
+      "epoch": 0.2175
+    },
+    {
+      "step": 949,
+      "timestamp": "2025-12-28T09:55:37.602478",
+      "elapsed_time": 3451.887568950653,
+      "loss": 0.1969,
+      "grad_norm": 0.12928517162799835,
+      "learning_rate": 0.00015748427672955976,
+      "epoch": 0.218125
+    },
+    {
+      "step": 950,
+      "timestamp": "2025-12-28T09:55:42.477278",
+      "elapsed_time": 3456.7623686790466,
+      "loss": 0.2521,
+      "grad_norm": 0.17746160924434662,
+      "learning_rate": 0.00015735849056603776,
+      "epoch": 0.21875
+    },
+    {
+      "step": 951,
+      "timestamp": "2025-12-28T09:55:48.417273",
+      "elapsed_time": 3462.702367544174,
+      "loss": 0.1949,
+      "grad_norm": 0.14048974215984344,
+      "learning_rate": 0.00015723270440251574,
+      "epoch": 0.219375
+    },
+    {
+      "step": 952,
+      "timestamp": "2025-12-28T09:55:56.843770",
+      "elapsed_time": 3471.1288611888885,
+      "loss": 0.3451,
+      "grad_norm": 0.14857220649719238,
+      "learning_rate": 0.00015710691823899372,
+      "epoch": 0.22
+    },
+    {
+      "step": 953,
+      "timestamp": "2025-12-28T09:56:03.831991",
+      "elapsed_time": 3478.117082118988,
+      "loss": 0.2241,
+      "grad_norm": 0.16286444664001465,
+      "learning_rate": 0.00015698113207547173,
+      "epoch": 0.220625
+    },
+    {
+      "step": 954,
+      "timestamp": "2025-12-28T09:56:09.773840",
+      "elapsed_time": 3484.0589311122894,
+      "loss": 0.4378,
+      "grad_norm": 0.25359323620796204,
+      "learning_rate": 0.0001568553459119497,
+      "epoch": 0.22125
+    },
+    {
+      "step": 955,
+      "timestamp": "2025-12-28T09:56:18.031656",
+      "elapsed_time": 3492.316746234894,
+      "loss": 0.3404,
+      "grad_norm": 0.16270765662193298,
+      "learning_rate": 0.00015672955974842769,
+      "epoch": 0.221875
+    },
+    {
+      "step": 956,
+      "timestamp": "2025-12-28T09:56:31.017715",
+      "elapsed_time": 3505.302805662155,
+      "loss": 0.1701,
+      "grad_norm": 0.11110091954469681,
+      "learning_rate": 0.00015660377358490567,
+      "epoch": 0.2225
+    },
+    {
+      "step": 957,
+      "timestamp": "2025-12-28T09:56:35.074678",
+      "elapsed_time": 3509.359768629074,
+      "loss": 0.3194,
+      "grad_norm": 0.20245419442653656,
+      "learning_rate": 0.00015647798742138367,
+      "epoch": 0.223125
+    },
+    {
+      "step": 958,
+      "timestamp": "2025-12-28T09:56:53.824765",
+      "elapsed_time": 3528.109855890274,
+      "loss": 0.1034,
+      "grad_norm": 0.07926256209611893,
+      "learning_rate": 0.00015635220125786165,
+      "epoch": 0.22375
+    },
+    {
+      "step": 959,
+      "timestamp": "2025-12-28T09:57:00.356123",
+      "elapsed_time": 3534.6412131786346,
+      "loss": 0.2092,
+      "grad_norm": 0.17986778914928436,
+      "learning_rate": 0.00015622641509433963,
+      "epoch": 0.224375
+    },
+    {
+      "step": 960,
+      "timestamp": "2025-12-28T09:57:08.990782",
+      "elapsed_time": 3543.2758724689484,
+      "loss": 0.1359,
+      "grad_norm": 0.13381750881671906,
+      "learning_rate": 0.00015610062893081764,
+      "epoch": 0.225
+    },
+    {
+      "step": 961,
+      "timestamp": "2025-12-28T09:57:26.985511",
+      "elapsed_time": 3561.2706019878387,
+      "loss": 0.0902,
+      "grad_norm": 0.07347288727760315,
+      "learning_rate": 0.00015597484276729561,
+      "epoch": 0.225625
+    },
+    {
+      "step": 962,
+      "timestamp": "2025-12-28T09:57:31.711118",
+      "elapsed_time": 3565.996208190918,
+      "loss": 0.2244,
+      "grad_norm": 0.17614322900772095,
+      "learning_rate": 0.0001558490566037736,
+      "epoch": 0.22625
+    },
+    {
+      "step": 963,
+      "timestamp": "2025-12-28T09:57:40.681045",
+      "elapsed_time": 3574.9661359786987,
+      "loss": 0.2077,
+      "grad_norm": 0.1338813602924347,
+      "learning_rate": 0.0001557232704402516,
+      "epoch": 0.226875
+    },
+    {
+      "step": 964,
+      "timestamp": "2025-12-28T09:57:44.579572",
+      "elapsed_time": 3578.8646624088287,
+      "loss": 0.2926,
+      "grad_norm": 0.22020283341407776,
+      "learning_rate": 0.00015559748427672958,
+      "epoch": 0.2275
+    },
+    {
+      "step": 965,
+      "timestamp": "2025-12-28T09:57:52.917233",
+      "elapsed_time": 3587.2023231983185,
+      "loss": 0.1769,
+      "grad_norm": 0.13740260899066925,
+      "learning_rate": 0.00015547169811320756,
+      "epoch": 0.228125
+    },
+    {
+      "step": 966,
+      "timestamp": "2025-12-28T09:58:01.752612",
+      "elapsed_time": 3596.0377027988434,
+      "loss": 0.1681,
+      "grad_norm": 0.12490309774875641,
+      "learning_rate": 0.00015534591194968556,
+      "epoch": 0.22875
+    },
+    {
+      "step": 967,
+      "timestamp": "2025-12-28T09:58:08.641082",
+      "elapsed_time": 3602.9261722564697,
+      "loss": 0.1626,
+      "grad_norm": 0.12916311621665955,
+      "learning_rate": 0.00015522012578616354,
+      "epoch": 0.229375
+    },
+    {
+      "step": 968,
+      "timestamp": "2025-12-28T09:58:21.326631",
+      "elapsed_time": 3615.611721277237,
+      "loss": 0.1157,
+      "grad_norm": 0.10117348283529282,
+      "learning_rate": 0.00015509433962264152,
+      "epoch": 0.23
+    },
+    {
+      "step": 969,
+      "timestamp": "2025-12-28T09:58:32.168538",
+      "elapsed_time": 3626.4536283016205,
+      "loss": 0.2099,
+      "grad_norm": 0.21025630831718445,
+      "learning_rate": 0.0001549685534591195,
+      "epoch": 0.230625
+    },
+    {
+      "step": 970,
+      "timestamp": "2025-12-28T09:58:40.157526",
+      "elapsed_time": 3634.4426164627075,
+      "loss": 0.2095,
+      "grad_norm": 0.13846275210380554,
+      "learning_rate": 0.0001548427672955975,
+      "epoch": 0.23125
+    },
+    {
+      "step": 971,
+      "timestamp": "2025-12-28T09:58:58.528736",
+      "elapsed_time": 3652.8138258457184,
+      "loss": 0.6305,
+      "grad_norm": 0.14148321747779846,
+      "learning_rate": 0.0001547169811320755,
+      "epoch": 0.231875
+    },
+    {
+      "step": 972,
+      "timestamp": "2025-12-28T09:59:09.985061",
+      "elapsed_time": 3664.270151615143,
+      "loss": 0.1818,
+      "grad_norm": 0.4846095144748688,
+      "learning_rate": 0.00015459119496855347,
+      "epoch": 0.2325
+    },
+    {
+      "step": 973,
+      "timestamp": "2025-12-28T09:59:15.643684",
+      "elapsed_time": 3669.9287745952606,
+      "loss": 0.3468,
+      "grad_norm": 0.18205124139785767,
+      "learning_rate": 0.00015446540880503147,
+      "epoch": 0.233125
+    },
+    {
+      "step": 974,
+      "timestamp": "2025-12-28T09:59:24.573388",
+      "elapsed_time": 3678.858478307724,
+      "loss": 0.2059,
+      "grad_norm": 0.13004031777381897,
+      "learning_rate": 0.00015433962264150945,
+      "epoch": 0.23375
+    },
+    {
+      "step": 975,
+      "timestamp": "2025-12-28T09:59:32.752142",
+      "elapsed_time": 3687.037232398987,
+      "loss": 0.2575,
+      "grad_norm": 0.1793992817401886,
+      "learning_rate": 0.00015421383647798743,
+      "epoch": 0.234375
+    },
+    {
+      "step": 976,
+      "timestamp": "2025-12-28T09:59:42.141949",
+      "elapsed_time": 3696.4270396232605,
+      "loss": 0.1674,
+      "grad_norm": 0.11684457212686539,
+      "learning_rate": 0.00015408805031446544,
+      "epoch": 0.235
+    },
+    {
+      "step": 977,
+      "timestamp": "2025-12-28T09:59:49.029099",
+      "elapsed_time": 3703.3141901493073,
+      "loss": 0.1885,
+      "grad_norm": 0.1317356675863266,
+      "learning_rate": 0.00015396226415094342,
+      "epoch": 0.235625
+    },
+    {
+      "step": 978,
+      "timestamp": "2025-12-28T09:59:56.964854",
+      "elapsed_time": 3711.2499442100525,
+      "loss": 0.2512,
+      "grad_norm": 0.1530027985572815,
+      "learning_rate": 0.0001538364779874214,
+      "epoch": 0.23625
+    },
+    {
+      "step": 979,
+      "timestamp": "2025-12-28T10:00:05.695516",
+      "elapsed_time": 3719.9806060791016,
+      "loss": 0.2406,
+      "grad_norm": 0.13735264539718628,
+      "learning_rate": 0.0001537106918238994,
+      "epoch": 0.236875
+    },
+    {
+      "step": 980,
+      "timestamp": "2025-12-28T10:00:13.159866",
+      "elapsed_time": 3727.444956302643,
+      "loss": 0.265,
+      "grad_norm": 0.14935627579689026,
+      "learning_rate": 0.00015358490566037738,
+      "epoch": 0.2375
+    },
+    {
+      "step": 981,
+      "timestamp": "2025-12-28T10:00:20.073453",
+      "elapsed_time": 3734.3585438728333,
+      "loss": 0.2331,
+      "grad_norm": 0.15793128311634064,
+      "learning_rate": 0.00015345911949685536,
+      "epoch": 0.238125
+    },
+    {
+      "step": 982,
+      "timestamp": "2025-12-28T10:00:28.700639",
+      "elapsed_time": 3742.9857289791107,
+      "loss": 0.3142,
+      "grad_norm": 0.14131158590316772,
+      "learning_rate": 0.00015333333333333334,
+      "epoch": 0.23875
+    },
+    {
+      "step": 983,
+      "timestamp": "2025-12-28T10:00:37.338005",
+      "elapsed_time": 3751.623096227646,
+      "loss": 0.1871,
+      "grad_norm": 0.12910977005958557,
+      "learning_rate": 0.00015320754716981134,
+      "epoch": 0.239375
+    },
+    {
+      "step": 984,
+      "timestamp": "2025-12-28T10:00:51.515626",
+      "elapsed_time": 3765.8007164001465,
+      "loss": 0.2466,
+      "grad_norm": 0.15984784066677094,
+      "learning_rate": 0.00015308176100628932,
+      "epoch": 0.24
+    },
+    {
+      "step": 985,
+      "timestamp": "2025-12-28T10:00:58.996365",
+      "elapsed_time": 3773.2814559936523,
+      "loss": 0.1202,
+      "grad_norm": 0.11847283691167831,
+      "learning_rate": 0.0001529559748427673,
+      "epoch": 0.240625
+    },
+    {
+      "step": 986,
+      "timestamp": "2025-12-28T10:01:09.966903",
+      "elapsed_time": 3784.25199341774,
+      "loss": 0.1645,
+      "grad_norm": 0.15656305849552155,
+      "learning_rate": 0.0001528301886792453,
+      "epoch": 0.24125
+    },
+    {
+      "step": 987,
+      "timestamp": "2025-12-28T10:01:23.517576",
+      "elapsed_time": 3797.802666902542,
+      "loss": 0.1248,
+      "grad_norm": 0.0928470715880394,
+      "learning_rate": 0.0001527044025157233,
+      "epoch": 0.241875
+    },
+    {
+      "step": 988,
+      "timestamp": "2025-12-28T10:01:34.217983",
+      "elapsed_time": 3808.503073453903,
+      "loss": 0.1338,
+      "grad_norm": 0.11139731854200363,
+      "learning_rate": 0.00015257861635220127,
+      "epoch": 0.2425
+    },
+    {
+      "step": 989,
+      "timestamp": "2025-12-28T10:01:41.199518",
+      "elapsed_time": 3815.484607934952,
+      "loss": 0.2724,
+      "grad_norm": 0.15163938701152802,
+      "learning_rate": 0.00015245283018867927,
+      "epoch": 0.243125
+    },
+    {
+      "step": 990,
+      "timestamp": "2025-12-28T10:01:48.450348",
+      "elapsed_time": 3822.7354385852814,
+      "loss": 0.2476,
+      "grad_norm": 0.17267391085624695,
+      "learning_rate": 0.00015232704402515725,
+      "epoch": 0.24375
+    },
+    {
+      "step": 991,
+      "timestamp": "2025-12-28T10:01:56.433279",
+      "elapsed_time": 3830.718369960785,
+      "loss": 0.1465,
+      "grad_norm": 0.11863212287425995,
+      "learning_rate": 0.00015220125786163523,
+      "epoch": 0.244375
+    },
+    {
+      "step": 992,
+      "timestamp": "2025-12-28T10:02:09.167701",
+      "elapsed_time": 3843.4527916908264,
+      "loss": 0.1302,
+      "grad_norm": 0.10036662966012955,
+      "learning_rate": 0.00015207547169811324,
+      "epoch": 0.245
+    },
+    {
+      "step": 993,
+      "timestamp": "2025-12-28T10:02:14.112829",
+      "elapsed_time": 3848.3979198932648,
+      "loss": 0.3477,
+      "grad_norm": 0.19196945428848267,
+      "learning_rate": 0.00015194968553459122,
+      "epoch": 0.245625
+    },
+    {
+      "step": 994,
+      "timestamp": "2025-12-28T10:02:24.995027",
+      "elapsed_time": 3859.2801179885864,
+      "loss": 0.1563,
+      "grad_norm": 0.11495956778526306,
+      "learning_rate": 0.0001518238993710692,
+      "epoch": 0.24625
+    },
+    {
+      "step": 995,
+      "timestamp": "2025-12-28T10:02:33.036433",
+      "elapsed_time": 3867.3215239048004,
+      "loss": 0.1802,
+      "grad_norm": 0.11796751618385315,
+      "learning_rate": 0.00015169811320754717,
+      "epoch": 0.246875
+    },
+    {
+      "step": 996,
+      "timestamp": "2025-12-28T10:02:50.626582",
+      "elapsed_time": 3884.911673307419,
+      "loss": 0.1721,
+      "grad_norm": 0.10495149344205856,
+      "learning_rate": 0.00015157232704402518,
+      "epoch": 0.2475
+    },
+    {
+      "step": 997,
+      "timestamp": "2025-12-28T10:02:56.189074",
+      "elapsed_time": 3890.474164247513,
+      "loss": 0.2895,
+      "grad_norm": 0.18693897128105164,
+      "learning_rate": 0.00015144654088050316,
+      "epoch": 0.248125
+    },
+    {
+      "step": 998,
+      "timestamp": "2025-12-28T10:03:03.931060",
+      "elapsed_time": 3898.216150522232,
+      "loss": 0.2034,
+      "grad_norm": 0.1423393189907074,
+      "learning_rate": 0.00015132075471698114,
+      "epoch": 0.24875
+    },
+    {
+      "step": 999,
+      "timestamp": "2025-12-28T10:03:21.397855",
+      "elapsed_time": 3915.6829454898834,
+      "loss": 0.1197,
+      "grad_norm": 0.0859493762254715,
+      "learning_rate": 0.00015119496855345914,
+      "epoch": 0.249375
+    },
+    {
+      "step": 1000,
+      "timestamp": "2025-12-28T10:03:29.241271",
+      "elapsed_time": 3923.5263612270355,
+      "loss": 0.2176,
+      "grad_norm": 0.13979171216487885,
+      "learning_rate": 0.00015106918238993712,
+      "epoch": 0.25
+    },
+    {
+      "step": 1001,
+      "timestamp": "2025-12-28T10:03:35.258000",
+      "elapsed_time": 3929.543091058731,
+      "loss": 0.2195,
+      "grad_norm": 0.15846221148967743,
+      "learning_rate": 0.0001509433962264151,
+      "epoch": 0.250625
+    },
+    {
+      "step": 1002,
+      "timestamp": "2025-12-28T10:03:40.199295",
+      "elapsed_time": 3934.484385251999,
+      "loss": 0.345,
+      "grad_norm": 0.18337437510490417,
+      "learning_rate": 0.0001508176100628931,
+      "epoch": 0.25125
+    },
+    {
+      "step": 1003,
+      "timestamp": "2025-12-28T10:03:49.416164",
+      "elapsed_time": 3943.701254606247,
+      "loss": 0.2019,
+      "grad_norm": 0.11637191474437714,
+      "learning_rate": 0.0001506918238993711,
+      "epoch": 0.251875
+    },
+    {
+      "step": 1004,
+      "timestamp": "2025-12-28T10:04:02.545538",
+      "elapsed_time": 3956.830629825592,
+      "loss": 0.1278,
+      "grad_norm": 0.1031939908862114,
+      "learning_rate": 0.00015056603773584907,
+      "epoch": 0.2525
+    },
+    {
+      "step": 1005,
+      "timestamp": "2025-12-28T10:04:08.991273",
+      "elapsed_time": 3963.2763633728027,
+      "loss": 0.2647,
+      "grad_norm": 0.16900090873241425,
+      "learning_rate": 0.00015044025157232707,
+      "epoch": 0.253125
+    },
+    {
+      "step": 1006,
+      "timestamp": "2025-12-28T10:04:15.995166",
+      "elapsed_time": 3970.2802562713623,
+      "loss": 0.2239,
+      "grad_norm": 0.1869828701019287,
+      "learning_rate": 0.00015031446540880505,
+      "epoch": 0.25375
+    },
+    {
+      "step": 1007,
+      "timestamp": "2025-12-28T10:04:26.260626",
+      "elapsed_time": 3980.5457170009613,
+      "loss": 0.1438,
+      "grad_norm": 0.1253536194562912,
+      "learning_rate": 0.00015018867924528303,
+      "epoch": 0.254375
+    },
+    {
+      "step": 1008,
+      "timestamp": "2025-12-28T10:04:33.781669",
+      "elapsed_time": 3988.0667593479156,
+      "loss": 0.2621,
+      "grad_norm": 0.17986273765563965,
+      "learning_rate": 0.00015006289308176104,
+      "epoch": 0.255
+    },
+    {
+      "step": 1009,
+      "timestamp": "2025-12-28T10:04:40.413811",
+      "elapsed_time": 3994.698902130127,
+      "loss": 0.2008,
+      "grad_norm": 0.14311149716377258,
+      "learning_rate": 0.000149937106918239,
+      "epoch": 0.255625
+    },
+    {
+      "step": 1010,
+      "timestamp": "2025-12-28T10:04:59.959869",
+      "elapsed_time": 4014.2449600696564,
+      "loss": 0.1501,
+      "grad_norm": 0.08852102607488632,
+      "learning_rate": 0.00014981132075471697,
+      "epoch": 0.25625
+    },
+    {
+      "step": 1011,
+      "timestamp": "2025-12-28T10:05:07.239700",
+      "elapsed_time": 4021.5247910022736,
+      "loss": 0.2402,
+      "grad_norm": 0.19574478268623352,
+      "learning_rate": 0.00014968553459119498,
+      "epoch": 0.256875
+    },
+    {
+      "step": 1012,
+      "timestamp": "2025-12-28T10:05:19.732988",
+      "elapsed_time": 4034.0180780887604,
+      "loss": 0.1703,
+      "grad_norm": 0.10853290557861328,
+      "learning_rate": 0.00014955974842767295,
+      "epoch": 0.2575
+    },
+    {
+      "step": 1013,
+      "timestamp": "2025-12-28T10:05:28.914807",
+      "elapsed_time": 4043.1998975276947,
+      "loss": 0.2656,
+      "grad_norm": 0.16859759390354156,
+      "learning_rate": 0.00014943396226415093,
+      "epoch": 0.258125
+    },
+    {
+      "step": 1014,
+      "timestamp": "2025-12-28T10:05:40.579825",
+      "elapsed_time": 4054.8649151325226,
+      "loss": 0.1861,
+      "grad_norm": 0.12188176810741425,
+      "learning_rate": 0.00014930817610062894,
+      "epoch": 0.25875
+    },
+    {
+      "step": 1015,
+      "timestamp": "2025-12-28T10:05:52.247391",
+      "elapsed_time": 4066.5324816703796,
+      "loss": 0.172,
+      "grad_norm": 0.11021149158477783,
+      "learning_rate": 0.00014918238993710692,
+      "epoch": 0.259375
+    },
+    {
+      "step": 1016,
+      "timestamp": "2025-12-28T10:06:02.667177",
+      "elapsed_time": 4076.9522676467896,
+      "loss": 0.1577,
+      "grad_norm": 0.11924094706773758,
+      "learning_rate": 0.0001490566037735849,
+      "epoch": 0.26
+    },
+    {
+      "step": 1017,
+      "timestamp": "2025-12-28T10:06:10.516680",
+      "elapsed_time": 4084.8017704486847,
+      "loss": 0.1642,
+      "grad_norm": 0.12346573173999786,
+      "learning_rate": 0.00014893081761006288,
+      "epoch": 0.260625
+    },
+    {
+      "step": 1018,
+      "timestamp": "2025-12-28T10:06:15.781551",
+      "elapsed_time": 4090.066641330719,
+      "loss": 0.4633,
+      "grad_norm": 0.2274809330701828,
+      "learning_rate": 0.00014880503144654088,
+      "epoch": 0.26125
+    },
+    {
+      "step": 1019,
+      "timestamp": "2025-12-28T10:06:23.024176",
+      "elapsed_time": 4097.30926656723,
+      "loss": 0.2659,
+      "grad_norm": 0.1639455407857895,
+      "learning_rate": 0.00014867924528301886,
+      "epoch": 0.261875
+    },
+    {
+      "step": 1020,
+      "timestamp": "2025-12-28T10:06:29.002861",
+      "elapsed_time": 4103.28795170784,
+      "loss": 0.1842,
+      "grad_norm": 0.12965330481529236,
+      "learning_rate": 0.00014855345911949684,
+      "epoch": 0.2625
+    },
+    {
+      "step": 1021,
+      "timestamp": "2025-12-28T10:06:41.542093",
+      "elapsed_time": 4115.827183961868,
+      "loss": 0.159,
+      "grad_norm": 0.10215835273265839,
+      "learning_rate": 0.00014842767295597485,
+      "epoch": 0.263125
+    },
+    {
+      "step": 1022,
+      "timestamp": "2025-12-28T10:06:47.805495",
+      "elapsed_time": 4122.090589284897,
+      "loss": 0.2087,
+      "grad_norm": 0.19713939726352692,
+      "learning_rate": 0.00014830188679245283,
+      "epoch": 0.26375
+    },
+    {
+      "step": 1023,
+      "timestamp": "2025-12-28T10:06:53.583302",
+      "elapsed_time": 4127.8683931827545,
+      "loss": 0.2987,
+      "grad_norm": 0.18019935488700867,
+      "learning_rate": 0.0001481761006289308,
+      "epoch": 0.264375
+    },
+    {
+      "step": 1024,
+      "timestamp": "2025-12-28T10:07:00.180550",
+      "elapsed_time": 4134.465640544891,
+      "loss": 0.2478,
+      "grad_norm": 0.1467757374048233,
+      "learning_rate": 0.0001480503144654088,
+      "epoch": 0.265
+    },
+    {
+      "step": 1025,
+      "timestamp": "2025-12-28T10:07:10.795758",
+      "elapsed_time": 4145.0808482170105,
+      "loss": 0.1668,
+      "grad_norm": 0.10882839560508728,
+      "learning_rate": 0.0001479245283018868,
+      "epoch": 0.265625
+    },
+    {
+      "step": 1026,
+      "timestamp": "2025-12-28T10:07:17.133433",
+      "elapsed_time": 4151.418524265289,
+      "loss": 0.2731,
+      "grad_norm": 0.15907247364521027,
+      "learning_rate": 0.00014779874213836477,
+      "epoch": 0.26625
+    },
+    {
+      "step": 1027,
+      "timestamp": "2025-12-28T10:07:24.120391",
+      "elapsed_time": 4158.405481100082,
+      "loss": 0.2377,
+      "grad_norm": 0.13879650831222534,
+      "learning_rate": 0.00014767295597484278,
+      "epoch": 0.266875
+    },
+    {
+      "step": 1028,
+      "timestamp": "2025-12-28T10:07:32.464337",
+      "elapsed_time": 4166.749427556992,
+      "loss": 0.1956,
+      "grad_norm": 0.13655611872673035,
+      "learning_rate": 0.00014754716981132076,
+      "epoch": 0.2675
+    },
+    {
+      "step": 1029,
+      "timestamp": "2025-12-28T10:07:46.125666",
+      "elapsed_time": 4180.410757303238,
+      "loss": 0.119,
+      "grad_norm": 0.2348964810371399,
+      "learning_rate": 0.00014742138364779873,
+      "epoch": 0.268125
+    },
+    {
+      "step": 1030,
+      "timestamp": "2025-12-28T10:07:50.593098",
+      "elapsed_time": 4184.878188371658,
+      "loss": 0.2267,
+      "grad_norm": 0.16901762783527374,
+      "learning_rate": 0.0001472955974842767,
+      "epoch": 0.26875
+    },
+    {
+      "step": 1031,
+      "timestamp": "2025-12-28T10:08:03.007589",
+      "elapsed_time": 4197.292679548264,
+      "loss": 0.1174,
+      "grad_norm": 0.08787506818771362,
+      "learning_rate": 0.00014716981132075472,
+      "epoch": 0.269375
+    },
+    {
+      "step": 1032,
+      "timestamp": "2025-12-28T10:08:09.168697",
+      "elapsed_time": 4203.453787326813,
+      "loss": 0.3684,
+      "grad_norm": 0.18056993186473846,
+      "learning_rate": 0.0001470440251572327,
+      "epoch": 0.27
+    },
+    {
+      "step": 1033,
+      "timestamp": "2025-12-28T10:08:20.157096",
+      "elapsed_time": 4214.442186117172,
+      "loss": 0.1727,
+      "grad_norm": 0.11187569051980972,
+      "learning_rate": 0.00014691823899371068,
+      "epoch": 0.270625
+    },
+    {
+      "step": 1034,
+      "timestamp": "2025-12-28T10:08:27.489866",
+      "elapsed_time": 4221.774956703186,
+      "loss": 0.1853,
+      "grad_norm": 0.132780984044075,
+      "learning_rate": 0.00014679245283018868,
+      "epoch": 0.27125
+    },
+    {
+      "step": 1035,
+      "timestamp": "2025-12-28T10:08:32.101038",
+      "elapsed_time": 4226.38612818718,
+      "loss": 0.2988,
+      "grad_norm": 0.21782910823822021,
+      "learning_rate": 0.00014666666666666666,
+      "epoch": 0.271875
+    },
+    {
+      "step": 1036,
+      "timestamp": "2025-12-28T10:08:38.180903",
+      "elapsed_time": 4232.465993881226,
+      "loss": 0.173,
+      "grad_norm": 0.1362728774547577,
+      "learning_rate": 0.00014654088050314464,
+      "epoch": 0.2725
+    },
+    {
+      "step": 1037,
+      "timestamp": "2025-12-28T10:08:47.148440",
+      "elapsed_time": 4241.433530807495,
+      "loss": 0.1494,
+      "grad_norm": 0.11238773167133331,
+      "learning_rate": 0.00014641509433962265,
+      "epoch": 0.273125
+    },
+    {
+      "step": 1038,
+      "timestamp": "2025-12-28T10:08:53.691649",
+      "elapsed_time": 4247.976739406586,
+      "loss": 0.2181,
+      "grad_norm": 0.15294326841831207,
+      "learning_rate": 0.00014628930817610063,
+      "epoch": 0.27375
+    },
+    {
+      "step": 1039,
+      "timestamp": "2025-12-28T10:08:59.801991",
+      "elapsed_time": 4254.0870814323425,
+      "loss": 0.2035,
+      "grad_norm": 0.14438562095165253,
+      "learning_rate": 0.0001461635220125786,
+      "epoch": 0.274375
+    },
+    {
+      "step": 1040,
+      "timestamp": "2025-12-28T10:09:04.922147",
+      "elapsed_time": 4259.2072377204895,
+      "loss": 0.1839,
+      "grad_norm": 0.163107767701149,
+      "learning_rate": 0.0001460377358490566,
+      "epoch": 0.275
+    },
+    {
+      "step": 1041,
+      "timestamp": "2025-12-28T10:09:11.363913",
+      "elapsed_time": 4265.649003267288,
+      "loss": 0.2156,
+      "grad_norm": 0.14970509707927704,
+      "learning_rate": 0.0001459119496855346,
+      "epoch": 0.275625
+    },
+    {
+      "step": 1042,
+      "timestamp": "2025-12-28T10:09:21.257050",
+      "elapsed_time": 4275.542140722275,
+      "loss": 0.1636,
+      "grad_norm": 0.12284188717603683,
+      "learning_rate": 0.00014578616352201257,
+      "epoch": 0.27625
+    },
+    {
+      "step": 1043,
+      "timestamp": "2025-12-28T10:09:28.778477",
+      "elapsed_time": 4283.063568115234,
+      "loss": 0.4273,
+      "grad_norm": 0.18576599657535553,
+      "learning_rate": 0.00014566037735849055,
+      "epoch": 0.276875
+    },
+    {
+      "step": 1044,
+      "timestamp": "2025-12-28T10:09:36.716585",
+      "elapsed_time": 4291.001675605774,
+      "loss": 0.187,
+      "grad_norm": 0.13979823887348175,
+      "learning_rate": 0.00014553459119496856,
+      "epoch": 0.2775
+    },
+    {
+      "step": 1045,
+      "timestamp": "2025-12-28T10:09:47.537065",
+      "elapsed_time": 4301.822155952454,
+      "loss": 0.1229,
+      "grad_norm": 0.11963897943496704,
+      "learning_rate": 0.00014540880503144653,
+      "epoch": 0.278125
+    },
+    {
+      "step": 1046,
+      "timestamp": "2025-12-28T10:09:53.307725",
+      "elapsed_time": 4307.5928156375885,
+      "loss": 0.2356,
+      "grad_norm": 0.18368114531040192,
+      "learning_rate": 0.00014528301886792451,
+      "epoch": 0.27875
+    },
+    {
+      "step": 1047,
+      "timestamp": "2025-12-28T10:09:59.635756",
+      "elapsed_time": 4313.920850515366,
+      "loss": 0.2149,
+      "grad_norm": 0.2329953908920288,
+      "learning_rate": 0.00014515723270440252,
+      "epoch": 0.279375
+    },
+    {
+      "step": 1048,
+      "timestamp": "2025-12-28T10:10:04.571410",
+      "elapsed_time": 4318.856500864029,
+      "loss": 0.2881,
+      "grad_norm": 0.24254077672958374,
+      "learning_rate": 0.0001450314465408805,
+      "epoch": 0.28
+    },
+    {
+      "step": 1049,
+      "timestamp": "2025-12-28T10:10:09.018520",
+      "elapsed_time": 4323.303614139557,
+      "loss": 0.3995,
+      "grad_norm": 0.20608584582805634,
+      "learning_rate": 0.00014490566037735848,
+      "epoch": 0.280625
+    },
+    {
+      "step": 1050,
+      "timestamp": "2025-12-28T10:10:19.733959",
+      "elapsed_time": 4334.019049882889,
+      "loss": 0.1759,
+      "grad_norm": 0.12463296949863434,
+      "learning_rate": 0.00014477987421383648,
+      "epoch": 0.28125
+    },
+    {
+      "step": 1051,
+      "timestamp": "2025-12-28T10:10:27.504878",
+      "elapsed_time": 4341.789968252182,
+      "loss": 0.1602,
+      "grad_norm": 0.13327348232269287,
+      "learning_rate": 0.00014465408805031446,
+      "epoch": 0.281875
+    },
+    {
+      "step": 1052,
+      "timestamp": "2025-12-28T10:10:40.425643",
+      "elapsed_time": 4354.710733413696,
+      "loss": 0.1314,
+      "grad_norm": 0.09431233257055283,
+      "learning_rate": 0.00014452830188679244,
+      "epoch": 0.2825
+    },
+    {
+      "step": 1053,
+      "timestamp": "2025-12-28T10:10:58.475011",
+      "elapsed_time": 4372.760101318359,
+      "loss": 0.1314,
+      "grad_norm": 0.08956651389598846,
+      "learning_rate": 0.00014440251572327045,
+      "epoch": 0.283125
+    },
+    {
+      "step": 1054,
+      "timestamp": "2025-12-28T10:11:05.726827",
+      "elapsed_time": 4380.011917591095,
+      "loss": 0.2106,
+      "grad_norm": 0.14600154757499695,
+      "learning_rate": 0.00014427672955974843,
+      "epoch": 0.28375
+    },
+    {
+      "step": 1055,
+      "timestamp": "2025-12-28T10:11:19.021552",
+      "elapsed_time": 4393.30664229393,
+      "loss": 0.176,
+      "grad_norm": 0.1056523472070694,
+      "learning_rate": 0.0001441509433962264,
+      "epoch": 0.284375
+    },
+    {
+      "step": 1056,
+      "timestamp": "2025-12-28T10:11:29.958671",
+      "elapsed_time": 4404.243761062622,
+      "loss": 0.1629,
+      "grad_norm": 0.10517208278179169,
+      "learning_rate": 0.00014402515723270439,
+      "epoch": 0.285
+    },
+    {
+      "step": 1057,
+      "timestamp": "2025-12-28T10:11:36.541158",
+      "elapsed_time": 4410.826248168945,
+      "loss": 0.3887,
+      "grad_norm": 0.16218851506710052,
+      "learning_rate": 0.0001438993710691824,
+      "epoch": 0.285625
+    },
+    {
+      "step": 1058,
+      "timestamp": "2025-12-28T10:11:45.563207",
+      "elapsed_time": 4419.848297119141,
+      "loss": 0.1683,
+      "grad_norm": 0.1224561482667923,
+      "learning_rate": 0.00014377358490566037,
+      "epoch": 0.28625
+    },
+    {
+      "step": 1059,
+      "timestamp": "2025-12-28T10:11:58.056565",
+      "elapsed_time": 4432.34165596962,
+      "loss": 0.125,
+      "grad_norm": 0.08662772178649902,
+      "learning_rate": 0.00014364779874213835,
+      "epoch": 0.286875
+    },
+    {
+      "step": 1060,
+      "timestamp": "2025-12-28T10:12:05.305261",
+      "elapsed_time": 4439.590351343155,
+      "loss": 0.2103,
+      "grad_norm": 0.14773836731910706,
+      "learning_rate": 0.00014352201257861636,
+      "epoch": 0.2875
+    },
+    {
+      "step": 1061,
+      "timestamp": "2025-12-28T10:12:17.136857",
+      "elapsed_time": 4451.4219472408295,
+      "loss": 0.2254,
+      "grad_norm": 0.128709077835083,
+      "learning_rate": 0.00014339622641509434,
+      "epoch": 0.288125
+    },
+    {
+      "step": 1062,
+      "timestamp": "2025-12-28T10:12:26.358416",
+      "elapsed_time": 4460.643507003784,
+      "loss": 0.1654,
+      "grad_norm": 0.12667128443717957,
+      "learning_rate": 0.00014327044025157231,
+      "epoch": 0.28875
+    },
+    {
+      "step": 1063,
+      "timestamp": "2025-12-28T10:12:35.697244",
+      "elapsed_time": 4469.982335090637,
+      "loss": 0.1426,
+      "grad_norm": 0.10940206795930862,
+      "learning_rate": 0.00014314465408805032,
+      "epoch": 0.289375
+    },
+    {
+      "step": 1064,
+      "timestamp": "2025-12-28T10:12:41.676751",
+      "elapsed_time": 4475.961841106415,
+      "loss": 0.1529,
+      "grad_norm": 0.1327073872089386,
+      "learning_rate": 0.0001430188679245283,
+      "epoch": 0.29
+    },
+    {
+      "step": 1065,
+      "timestamp": "2025-12-28T10:12:54.040253",
+      "elapsed_time": 4488.325343132019,
+      "loss": 0.4812,
+      "grad_norm": 0.15044333040714264,
+      "learning_rate": 0.00014289308176100628,
+      "epoch": 0.290625
+    },
+    {
+      "step": 1066,
+      "timestamp": "2025-12-28T10:13:01.898646",
+      "elapsed_time": 4496.183736562729,
+      "loss": 0.19,
+      "grad_norm": 0.12461165338754654,
+      "learning_rate": 0.00014276729559748429,
+      "epoch": 0.29125
+    },
+    {
+      "step": 1067,
+      "timestamp": "2025-12-28T10:13:11.281195",
+      "elapsed_time": 4505.566284894943,
+      "loss": 0.375,
+      "grad_norm": 0.2037774622440338,
+      "learning_rate": 0.00014264150943396226,
+      "epoch": 0.291875
+    },
+    {
+      "step": 1068,
+      "timestamp": "2025-12-28T10:13:20.374473",
+      "elapsed_time": 4514.659563064575,
+      "loss": 0.1759,
+      "grad_norm": 0.11888349056243896,
+      "learning_rate": 0.00014251572327044024,
+      "epoch": 0.2925
+    },
+    {
+      "step": 1069,
+      "timestamp": "2025-12-28T10:13:33.299520",
+      "elapsed_time": 4527.58461022377,
+      "loss": 0.1318,
+      "grad_norm": 0.09523271024227142,
+      "learning_rate": 0.00014238993710691825,
+      "epoch": 0.293125
+    },
+    {
+      "step": 1070,
+      "timestamp": "2025-12-28T10:13:41.047563",
+      "elapsed_time": 4535.332653284073,
+      "loss": 0.2335,
+      "grad_norm": 0.13759227097034454,
+      "learning_rate": 0.00014226415094339623,
+      "epoch": 0.29375
+    },
+    {
+      "step": 1071,
+      "timestamp": "2025-12-28T10:13:56.504224",
+      "elapsed_time": 4550.789314746857,
+      "loss": 0.1236,
+      "grad_norm": 0.08639674633741379,
+      "learning_rate": 0.0001421383647798742,
+      "epoch": 0.294375
+    },
+    {
+      "step": 1072,
+      "timestamp": "2025-12-28T10:14:06.307720",
+      "elapsed_time": 4560.59281039238,
+      "loss": 0.1721,
+      "grad_norm": 0.1233757734298706,
+      "learning_rate": 0.0001420125786163522,
+      "epoch": 0.295
+    },
+    {
+      "step": 1073,
+      "timestamp": "2025-12-28T10:14:11.665841",
+      "elapsed_time": 4565.950931310654,
+      "loss": 0.2988,
+      "grad_norm": 0.1682165414094925,
+      "learning_rate": 0.0001418867924528302,
+      "epoch": 0.295625
+    },
+    {
+      "step": 1074,
+      "timestamp": "2025-12-28T10:14:21.737669",
+      "elapsed_time": 4576.022758960724,
+      "loss": 0.1494,
+      "grad_norm": 0.10447243601083755,
+      "learning_rate": 0.00014176100628930817,
+      "epoch": 0.29625
+    },
+    {
+      "step": 1075,
+      "timestamp": "2025-12-28T10:14:40.552859",
+      "elapsed_time": 4594.837973356247,
+      "loss": 0.1524,
+      "grad_norm": 0.0825522392988205,
+      "learning_rate": 0.00014163522012578615,
+      "epoch": 0.296875
+    },
+    {
+      "step": 1076,
+      "timestamp": "2025-12-28T10:14:49.284505",
+      "elapsed_time": 4603.569595575333,
+      "loss": 0.2119,
+      "grad_norm": 0.12849776446819305,
+      "learning_rate": 0.00014150943396226416,
+      "epoch": 0.2975
+    },
+    {
+      "step": 1077,
+      "timestamp": "2025-12-28T10:15:02.082705",
+      "elapsed_time": 4616.367795705795,
+      "loss": 0.1306,
+      "grad_norm": 0.09817694872617722,
+      "learning_rate": 0.00014138364779874214,
+      "epoch": 0.298125
+    },
+    {
+      "step": 1078,
+      "timestamp": "2025-12-28T10:15:12.968494",
+      "elapsed_time": 4627.253584623337,
+      "loss": 0.125,
+      "grad_norm": 0.10283590853214264,
+      "learning_rate": 0.00014125786163522012,
+      "epoch": 0.29875
+    },
+    {
+      "step": 1079,
+      "timestamp": "2025-12-28T10:15:20.706362",
+      "elapsed_time": 4634.991452932358,
+      "loss": 0.3718,
+      "grad_norm": 0.16993916034698486,
+      "learning_rate": 0.00014113207547169812,
+      "epoch": 0.299375
+    },
+    {
+      "step": 1080,
+      "timestamp": "2025-12-28T10:15:24.915200",
+      "elapsed_time": 4639.200289726257,
+      "loss": 0.1865,
+      "grad_norm": 0.16090618073940277,
+      "learning_rate": 0.0001410062893081761,
+      "epoch": 0.3
+    },
+    {
+      "step": 1081,
+      "timestamp": "2025-12-28T10:15:33.659663",
+      "elapsed_time": 4647.944753646851,
+      "loss": 0.1715,
+      "grad_norm": 0.1435355693101883,
+      "learning_rate": 0.00014088050314465408,
+      "epoch": 0.300625
+    },
+    {
+      "step": 1082,
+      "timestamp": "2025-12-28T10:15:50.278302",
+      "elapsed_time": 4664.563392162323,
+      "loss": 0.1291,
+      "grad_norm": 0.08483153581619263,
+      "learning_rate": 0.00014075471698113209,
+      "epoch": 0.30125
+    },
+    {
+      "step": 1083,
+      "timestamp": "2025-12-28T10:15:56.389680",
+      "elapsed_time": 4670.674770593643,
+      "loss": 0.2924,
+      "grad_norm": 0.17434731125831604,
+      "learning_rate": 0.00014062893081761007,
+      "epoch": 0.301875
+    },
+    {
+      "step": 1084,
+      "timestamp": "2025-12-28T10:16:06.200612",
+      "elapsed_time": 4680.48570227623,
+      "loss": 0.1588,
+      "grad_norm": 0.10816198587417603,
+      "learning_rate": 0.00014050314465408804,
+      "epoch": 0.3025
+    },
+    {
+      "step": 1085,
+      "timestamp": "2025-12-28T10:16:12.114394",
+      "elapsed_time": 4686.399484395981,
+      "loss": 0.2117,
+      "grad_norm": 0.15836821496486664,
+      "learning_rate": 0.00014037735849056602,
+      "epoch": 0.303125
+    },
+    {
+      "step": 1086,
+      "timestamp": "2025-12-28T10:16:20.183071",
+      "elapsed_time": 4694.468160867691,
+      "loss": 0.2609,
+      "grad_norm": 0.16809400916099548,
+      "learning_rate": 0.00014025157232704403,
+      "epoch": 0.30375
+    },
+    {
+      "step": 1087,
+      "timestamp": "2025-12-28T10:16:28.614605",
+      "elapsed_time": 4702.8996958732605,
+      "loss": 0.497,
+      "grad_norm": 0.17253176867961884,
+      "learning_rate": 0.000140125786163522,
+      "epoch": 0.304375
+    },
+    {
+      "step": 1088,
+      "timestamp": "2025-12-28T10:16:34.729314",
+      "elapsed_time": 4709.014405012131,
+      "loss": 0.2396,
+      "grad_norm": 0.1900101751089096,
+      "learning_rate": 0.00014,
+      "epoch": 0.305
+    },
+    {
+      "step": 1089,
+      "timestamp": "2025-12-28T10:16:47.236814",
+      "elapsed_time": 4721.521904230118,
+      "loss": 0.2017,
+      "grad_norm": 0.1198599562048912,
+      "learning_rate": 0.000139874213836478,
+      "epoch": 0.305625
+    },
+    {
+      "step": 1090,
+      "timestamp": "2025-12-28T10:16:56.587644",
+      "elapsed_time": 4730.872734546661,
+      "loss": 0.1482,
+      "grad_norm": 0.10646126419305801,
+      "learning_rate": 0.00013974842767295597,
+      "epoch": 0.30625
+    },
+    {
+      "step": 1091,
+      "timestamp": "2025-12-28T10:17:09.729699",
+      "elapsed_time": 4744.014789104462,
+      "loss": 0.1493,
+      "grad_norm": 0.1059768870472908,
+      "learning_rate": 0.00013962264150943395,
+      "epoch": 0.306875
+    },
+    {
+      "step": 1092,
+      "timestamp": "2025-12-28T10:17:15.846589",
+      "elapsed_time": 4750.131683349609,
+      "loss": 0.2455,
+      "grad_norm": 0.15631641447544098,
+      "learning_rate": 0.00013949685534591196,
+      "epoch": 0.3075
+    },
+    {
+      "step": 1093,
+      "timestamp": "2025-12-28T10:17:26.226547",
+      "elapsed_time": 4760.511637687683,
+      "loss": 0.158,
+      "grad_norm": 0.11631731688976288,
+      "learning_rate": 0.00013937106918238994,
+      "epoch": 0.308125
+    },
+    {
+      "step": 1094,
+      "timestamp": "2025-12-28T10:17:35.012584",
+      "elapsed_time": 4769.297674417496,
+      "loss": 0.1526,
+      "grad_norm": 0.1259499490261078,
+      "learning_rate": 0.00013924528301886792,
+      "epoch": 0.30875
+    },
+    {
+      "step": 1095,
+      "timestamp": "2025-12-28T10:17:45.261282",
+      "elapsed_time": 4779.546371936798,
+      "loss": 0.241,
+      "grad_norm": 0.15224315226078033,
+      "learning_rate": 0.00013911949685534592,
+      "epoch": 0.309375
+    },
+    {
+      "step": 1096,
+      "timestamp": "2025-12-28T10:17:57.287033",
+      "elapsed_time": 4791.572123765945,
+      "loss": 0.139,
+      "grad_norm": 0.1223248764872551,
+      "learning_rate": 0.0001389937106918239,
+      "epoch": 0.31
+    },
+    {
+      "step": 1097,
+      "timestamp": "2025-12-28T10:18:09.370759",
+      "elapsed_time": 4803.655849456787,
+      "loss": 0.2395,
+      "grad_norm": 0.13052494823932648,
+      "learning_rate": 0.00013886792452830188,
+      "epoch": 0.310625
+    },
+    {
+      "step": 1098,
+      "timestamp": "2025-12-28T10:18:22.367902",
+      "elapsed_time": 4816.652992963791,
+      "loss": 0.1401,
+      "grad_norm": 0.11577942967414856,
+      "learning_rate": 0.00013874213836477986,
+      "epoch": 0.31125
+    },
+    {
+      "step": 1099,
+      "timestamp": "2025-12-28T10:18:37.241574",
+      "elapsed_time": 4831.526664972305,
+      "loss": 0.139,
+      "grad_norm": 0.09500917792320251,
+      "learning_rate": 0.00013861635220125787,
+      "epoch": 0.311875
+    },
+    {
+      "step": 1100,
+      "timestamp": "2025-12-28T10:18:45.367638",
+      "elapsed_time": 4839.6527326107025,
+      "loss": 0.1616,
+      "grad_norm": 0.12062890082597733,
+      "learning_rate": 0.00013849056603773585,
+      "epoch": 0.3125
+    },
+    {
+      "step": 1101,
+      "timestamp": "2025-12-28T10:18:50.501032",
+      "elapsed_time": 4844.786121845245,
+      "loss": 0.5666,
+      "grad_norm": 0.21491988003253937,
+      "learning_rate": 0.00013836477987421382,
+      "epoch": 0.313125
+    },
+    {
+      "step": 1102,
+      "timestamp": "2025-12-28T10:18:57.091853",
+      "elapsed_time": 4851.37694311142,
+      "loss": 0.1821,
+      "grad_norm": 0.14800046384334564,
+      "learning_rate": 0.00013823899371069183,
+      "epoch": 0.31375
+    },
+    {
+      "step": 1103,
+      "timestamp": "2025-12-28T10:19:06.683644",
+      "elapsed_time": 4860.968734264374,
+      "loss": 0.1477,
+      "grad_norm": 0.11978691071271896,
+      "learning_rate": 0.0001381132075471698,
+      "epoch": 0.314375
+    },
+    {
+      "step": 1104,
+      "timestamp": "2025-12-28T10:19:21.833042",
+      "elapsed_time": 4876.118132352829,
+      "loss": 0.1662,
+      "grad_norm": 0.09653400629758835,
+      "learning_rate": 0.0001379874213836478,
+      "epoch": 0.315
+    },
+    {
+      "step": 1105,
+      "timestamp": "2025-12-28T10:19:26.190650",
+      "elapsed_time": 4880.475740432739,
+      "loss": 0.2282,
+      "grad_norm": 0.20169983804225922,
+      "learning_rate": 0.0001378616352201258,
+      "epoch": 0.315625
+    },
+    {
+      "step": 1106,
+      "timestamp": "2025-12-28T10:19:38.616678",
+      "elapsed_time": 4892.901768684387,
+      "loss": 0.1446,
+      "grad_norm": 0.10674963146448135,
+      "learning_rate": 0.00013773584905660377,
+      "epoch": 0.31625
+    },
+    {
+      "step": 1107,
+      "timestamp": "2025-12-28T10:19:59.515344",
+      "elapsed_time": 4913.800434112549,
+      "loss": 0.1037,
+      "grad_norm": 0.06956765800714493,
+      "learning_rate": 0.00013761006289308175,
+      "epoch": 0.316875
+    },
+    {
+      "step": 1108,
+      "timestamp": "2025-12-28T10:20:06.485473",
+      "elapsed_time": 4920.77056312561,
+      "loss": 0.182,
+      "grad_norm": 0.23156726360321045,
+      "learning_rate": 0.00013748427672955976,
+      "epoch": 0.3175
+    },
+    {
+      "step": 1109,
+      "timestamp": "2025-12-28T10:20:23.683657",
+      "elapsed_time": 4937.968747854233,
+      "loss": 0.1634,
+      "grad_norm": 0.1053549200296402,
+      "learning_rate": 0.00013735849056603774,
+      "epoch": 0.318125
+    },
+    {
+      "step": 1110,
+      "timestamp": "2025-12-28T10:20:30.071074",
+      "elapsed_time": 4944.356164216995,
+      "loss": 0.1761,
+      "grad_norm": 0.1309899240732193,
+      "learning_rate": 0.00013723270440251572,
+      "epoch": 0.31875
+    },
+    {
+      "step": 1111,
+      "timestamp": "2025-12-28T10:20:48.259766",
+      "elapsed_time": 4962.5448570251465,
+      "loss": 0.0991,
+      "grad_norm": 0.07965648174285889,
+      "learning_rate": 0.0001371069182389937,
+      "epoch": 0.319375
+    },
+    {
+      "step": 1112,
+      "timestamp": "2025-12-28T10:20:54.336910",
+      "elapsed_time": 4968.622000217438,
+      "loss": 0.1765,
+      "grad_norm": 0.16077613830566406,
+      "learning_rate": 0.0001369811320754717,
+      "epoch": 0.32
+    },
+    {
+      "step": 1113,
+      "timestamp": "2025-12-28T10:21:01.708820",
+      "elapsed_time": 4975.993910551071,
+      "loss": 0.2535,
+      "grad_norm": 0.19687579572200775,
+      "learning_rate": 0.00013685534591194968,
+      "epoch": 0.320625
+    },
+    {
+      "step": 1114,
+      "timestamp": "2025-12-28T10:21:08.888514",
+      "elapsed_time": 4983.173604011536,
+      "loss": 0.1733,
+      "grad_norm": 0.1458161175251007,
+      "learning_rate": 0.00013672955974842766,
+      "epoch": 0.32125
+    },
+    {
+      "step": 1115,
+      "timestamp": "2025-12-28T10:21:21.027050",
+      "elapsed_time": 4995.312140703201,
+      "loss": 0.1462,
+      "grad_norm": 0.10044913738965988,
+      "learning_rate": 0.00013660377358490567,
+      "epoch": 0.321875
+    },
+    {
+      "step": 1116,
+      "timestamp": "2025-12-28T10:21:32.969714",
+      "elapsed_time": 5007.2548043727875,
+      "loss": 0.1104,
+      "grad_norm": 0.08253983408212662,
+      "learning_rate": 0.00013647798742138365,
+      "epoch": 0.3225
+    },
+    {
+      "step": 1117,
+      "timestamp": "2025-12-28T10:21:38.181094",
+      "elapsed_time": 5012.466184139252,
+      "loss": 0.2278,
+      "grad_norm": 0.1712544858455658,
+      "learning_rate": 0.00013635220125786162,
+      "epoch": 0.323125
+    },
+    {
+      "step": 1118,
+      "timestamp": "2025-12-28T10:21:48.132268",
+      "elapsed_time": 5022.4173583984375,
+      "loss": 0.1665,
+      "grad_norm": 0.1129027009010315,
+      "learning_rate": 0.00013622641509433963,
+      "epoch": 0.32375
+    },
+    {
+      "step": 1119,
+      "timestamp": "2025-12-28T10:21:58.890656",
+      "elapsed_time": 5033.175746202469,
+      "loss": 0.1526,
+      "grad_norm": 0.12182003259658813,
+      "learning_rate": 0.0001361006289308176,
+      "epoch": 0.324375
+    },
+    {
+      "step": 1120,
+      "timestamp": "2025-12-28T10:22:16.895465",
+      "elapsed_time": 5051.1805555820465,
+      "loss": 0.1335,
+      "grad_norm": 0.0835099071264267,
+      "learning_rate": 0.0001359748427672956,
+      "epoch": 0.325
+    },
+    {
+      "step": 1121,
+      "timestamp": "2025-12-28T10:22:23.340365",
+      "elapsed_time": 5057.62545633316,
+      "loss": 0.2159,
+      "grad_norm": 0.15998020768165588,
+      "learning_rate": 0.0001358490566037736,
+      "epoch": 0.325625
+    },
+    {
+      "step": 1122,
+      "timestamp": "2025-12-28T10:22:33.758550",
+      "elapsed_time": 5068.04364490509,
+      "loss": 0.1589,
+      "grad_norm": 0.115847647190094,
+      "learning_rate": 0.00013572327044025157,
+      "epoch": 0.32625
+    },
+    {
+      "step": 1123,
+      "timestamp": "2025-12-28T10:22:44.939902",
+      "elapsed_time": 5079.224991798401,
+      "loss": 0.1524,
+      "grad_norm": 0.10123994946479797,
+      "learning_rate": 0.00013559748427672955,
+      "epoch": 0.326875
+    },
+    {
+      "step": 1124,
+      "timestamp": "2025-12-28T10:22:50.271896",
+      "elapsed_time": 5084.556986808777,
+      "loss": 0.2074,
+      "grad_norm": 0.1583547741174698,
+      "learning_rate": 0.00013547169811320753,
+      "epoch": 0.3275
+    },
+    {
+      "step": 1125,
+      "timestamp": "2025-12-28T10:22:58.007020",
+      "elapsed_time": 5092.292110919952,
+      "loss": 0.2013,
+      "grad_norm": 0.13498209416866302,
+      "learning_rate": 0.00013534591194968554,
+      "epoch": 0.328125
+    },
+    {
+      "step": 1126,
+      "timestamp": "2025-12-28T10:23:06.472685",
+      "elapsed_time": 5100.75777554512,
+      "loss": 0.1787,
+      "grad_norm": 0.12321489304304123,
+      "learning_rate": 0.00013522012578616352,
+      "epoch": 0.32875
+    },
+    {
+      "step": 1127,
+      "timestamp": "2025-12-28T10:23:16.415604",
+      "elapsed_time": 5110.700694322586,
+      "loss": 0.2123,
+      "grad_norm": 0.13657422363758087,
+      "learning_rate": 0.0001350943396226415,
+      "epoch": 0.329375
+    },
+    {
+      "step": 1128,
+      "timestamp": "2025-12-28T10:23:21.550030",
+      "elapsed_time": 5115.835124254227,
+      "loss": 0.3319,
+      "grad_norm": 0.19260597229003906,
+      "learning_rate": 0.0001349685534591195,
+      "epoch": 0.33
+    },
+    {
+      "step": 1129,
+      "timestamp": "2025-12-28T10:23:37.207751",
+      "elapsed_time": 5131.492841243744,
+      "loss": 0.1309,
+      "grad_norm": 0.08664542436599731,
+      "learning_rate": 0.00013484276729559748,
+      "epoch": 0.330625
+    },
+    {
+      "step": 1130,
+      "timestamp": "2025-12-28T10:23:43.286231",
+      "elapsed_time": 5137.57132101059,
+      "loss": 0.2417,
+      "grad_norm": 0.21089886128902435,
+      "learning_rate": 0.00013471698113207546,
+      "epoch": 0.33125
+    },
+    {
+      "step": 1131,
+      "timestamp": "2025-12-28T10:23:51.484950",
+      "elapsed_time": 5145.7700407505035,
+      "loss": 0.1542,
+      "grad_norm": 0.12316355109214783,
+      "learning_rate": 0.00013459119496855347,
+      "epoch": 0.331875
+    },
+    {
+      "step": 1132,
+      "timestamp": "2025-12-28T10:23:58.046371",
+      "elapsed_time": 5152.331461429596,
+      "loss": 0.1719,
+      "grad_norm": 0.1278318166732788,
+      "learning_rate": 0.00013446540880503145,
+      "epoch": 0.3325
+    },
+    {
+      "step": 1133,
+      "timestamp": "2025-12-28T10:24:04.389165",
+      "elapsed_time": 5158.674255847931,
+      "loss": 0.2467,
+      "grad_norm": 0.16408215463161469,
+      "learning_rate": 0.00013433962264150943,
+      "epoch": 0.333125
+    },
+    {
+      "step": 1134,
+      "timestamp": "2025-12-28T10:24:14.662588",
+      "elapsed_time": 5168.947679042816,
+      "loss": 0.136,
+      "grad_norm": 0.1054520383477211,
+      "learning_rate": 0.00013421383647798743,
+      "epoch": 0.33375
+    },
+    {
+      "step": 1135,
+      "timestamp": "2025-12-28T10:24:24.914867",
+      "elapsed_time": 5179.199957847595,
+      "loss": 0.1625,
+      "grad_norm": 0.11112856864929199,
+      "learning_rate": 0.0001340880503144654,
+      "epoch": 0.334375
+    },
+    {
+      "step": 1136,
+      "timestamp": "2025-12-28T10:24:33.704996",
+      "elapsed_time": 5187.990086078644,
+      "loss": 0.2022,
+      "grad_norm": 0.14535382390022278,
+      "learning_rate": 0.0001339622641509434,
+      "epoch": 0.335
+    },
+    {
+      "step": 1137,
+      "timestamp": "2025-12-28T10:24:54.598309",
+      "elapsed_time": 5208.883399009705,
+      "loss": 0.1032,
+      "grad_norm": 0.07190191745758057,
+      "learning_rate": 0.00013383647798742137,
+      "epoch": 0.335625
+    },
+    {
+      "step": 1138,
+      "timestamp": "2025-12-28T10:24:59.727226",
+      "elapsed_time": 5214.0123155117035,
+      "loss": 0.2176,
+      "grad_norm": 0.1817207783460617,
+      "learning_rate": 0.00013371069182389938,
+      "epoch": 0.33625
+    },
+    {
+      "step": 1139,
+      "timestamp": "2025-12-28T10:25:05.999597",
+      "elapsed_time": 5220.284687042236,
+      "loss": 0.1639,
+      "grad_norm": 0.1274854838848114,
+      "learning_rate": 0.00013358490566037735,
+      "epoch": 0.336875
+    },
+    {
+      "step": 1140,
+      "timestamp": "2025-12-28T10:25:13.369606",
+      "elapsed_time": 5227.654696941376,
+      "loss": 0.2805,
+      "grad_norm": 0.15669360756874084,
+      "learning_rate": 0.00013345911949685533,
+      "epoch": 0.3375
+    },
+    {
+      "step": 1141,
+      "timestamp": "2025-12-28T10:25:26.173526",
+      "elapsed_time": 5240.458616495132,
+      "loss": 0.2244,
+      "grad_norm": 0.10920794308185577,
+      "learning_rate": 0.00013333333333333334,
+      "epoch": 0.338125
+    },
+    {
+      "step": 1142,
+      "timestamp": "2025-12-28T10:25:34.387144",
+      "elapsed_time": 5248.672234773636,
+      "loss": 0.1525,
+      "grad_norm": 0.12012416124343872,
+      "learning_rate": 0.00013320754716981132,
+      "epoch": 0.33875
+    },
+    {
+      "step": 1143,
+      "timestamp": "2025-12-28T10:25:40.867120",
+      "elapsed_time": 5255.15221118927,
+      "loss": 0.2315,
+      "grad_norm": 0.17467765510082245,
+      "learning_rate": 0.0001330817610062893,
+      "epoch": 0.339375
+    },
+    {
+      "step": 1144,
+      "timestamp": "2025-12-28T10:25:49.536049",
+      "elapsed_time": 5263.821138858795,
+      "loss": 0.1585,
+      "grad_norm": 0.12994243204593658,
+      "learning_rate": 0.0001329559748427673,
+      "epoch": 0.34
+    },
+    {
+      "step": 1145,
+      "timestamp": "2025-12-28T10:25:56.165948",
+      "elapsed_time": 5270.451038122177,
+      "loss": 0.3481,
+      "grad_norm": 0.17362722754478455,
+      "learning_rate": 0.00013283018867924528,
+      "epoch": 0.340625
+    },
+    {
+      "step": 1146,
+      "timestamp": "2025-12-28T10:26:07.059344",
+      "elapsed_time": 5281.344434261322,
+      "loss": 0.1996,
+      "grad_norm": 0.11975681781768799,
+      "learning_rate": 0.00013270440251572326,
+      "epoch": 0.34125
+    },
+    {
+      "step": 1147,
+      "timestamp": "2025-12-28T10:26:27.952012",
+      "elapsed_time": 5302.237102270126,
+      "loss": 0.1627,
+      "grad_norm": 0.08679784089326859,
+      "learning_rate": 0.00013257861635220127,
+      "epoch": 0.341875
+    },
+    {
+      "step": 1148,
+      "timestamp": "2025-12-28T10:26:36.021702",
+      "elapsed_time": 5310.306792974472,
+      "loss": 0.1613,
+      "grad_norm": 0.14012403786182404,
+      "learning_rate": 0.00013245283018867925,
+      "epoch": 0.3425
+    },
+    {
+      "step": 1149,
+      "timestamp": "2025-12-28T10:26:53.963469",
+      "elapsed_time": 5328.248558998108,
+      "loss": 0.1234,
+      "grad_norm": 0.08561190962791443,
+      "learning_rate": 0.00013232704402515723,
+      "epoch": 0.343125
+    },
+    {
+      "step": 1150,
+      "timestamp": "2025-12-28T10:27:05.423036",
+      "elapsed_time": 5339.708126306534,
+      "loss": 0.1275,
+      "grad_norm": 0.10564465820789337,
+      "learning_rate": 0.00013220125786163523,
+      "epoch": 0.34375
+    },
+    {
+      "step": 1151,
+      "timestamp": "2025-12-28T10:27:13.370347",
+      "elapsed_time": 5347.655437231064,
+      "loss": 0.171,
+      "grad_norm": 0.13151898980140686,
+      "learning_rate": 0.0001320754716981132,
+      "epoch": 0.344375
+    },
+    {
+      "step": 1152,
+      "timestamp": "2025-12-28T10:27:26.513582",
+      "elapsed_time": 5360.798672437668,
+      "loss": 0.1066,
+      "grad_norm": 0.08146792650222778,
+      "learning_rate": 0.0001319496855345912,
+      "epoch": 0.345
+    },
+    {
+      "step": 1153,
+      "timestamp": "2025-12-28T10:27:37.983479",
+      "elapsed_time": 5372.268569231033,
+      "loss": 0.14,
+      "grad_norm": 0.11052241921424866,
+      "learning_rate": 0.00013182389937106917,
+      "epoch": 0.345625
+    },
+    {
+      "step": 1154,
+      "timestamp": "2025-12-28T10:27:49.248781",
+      "elapsed_time": 5383.533871412277,
+      "loss": 0.2192,
+      "grad_norm": 0.11441831290721893,
+      "learning_rate": 0.00013169811320754718,
+      "epoch": 0.34625
+    },
+    {
+      "step": 1155,
+      "timestamp": "2025-12-28T10:27:58.548010",
+      "elapsed_time": 5392.833100557327,
+      "loss": 0.1996,
+      "grad_norm": 0.13789567351341248,
+      "learning_rate": 0.00013157232704402516,
+      "epoch": 0.346875
+    },
+    {
+      "step": 1156,
+      "timestamp": "2025-12-28T10:28:05.648486",
+      "elapsed_time": 5399.933576583862,
+      "loss": 0.1288,
+      "grad_norm": 0.12561623752117157,
+      "learning_rate": 0.00013144654088050313,
+      "epoch": 0.3475
+    },
+    {
+      "step": 1157,
+      "timestamp": "2025-12-28T10:28:17.278511",
+      "elapsed_time": 5411.563601255417,
+      "loss": 0.1758,
+      "grad_norm": 0.17324663698673248,
+      "learning_rate": 0.00013132075471698114,
+      "epoch": 0.348125
+    },
+    {
+      "step": 1158,
+      "timestamp": "2025-12-28T10:28:29.227502",
+      "elapsed_time": 5423.512593269348,
+      "loss": 0.2519,
+      "grad_norm": 0.13509972393512726,
+      "learning_rate": 0.00013119496855345912,
+      "epoch": 0.34875
+    },
+    {
+      "step": 1159,
+      "timestamp": "2025-12-28T10:28:41.252121",
+      "elapsed_time": 5435.537210941315,
+      "loss": 0.1828,
+      "grad_norm": 0.11435085535049438,
+      "learning_rate": 0.0001310691823899371,
+      "epoch": 0.349375
+    },
+    {
+      "step": 1160,
+      "timestamp": "2025-12-28T10:28:49.279299",
+      "elapsed_time": 5443.564389467239,
+      "loss": 0.2038,
+      "grad_norm": 0.13359728455543518,
+      "learning_rate": 0.0001309433962264151,
+      "epoch": 0.35
+    },
+    {
+      "step": 1161,
+      "timestamp": "2025-12-28T10:29:01.419851",
+      "elapsed_time": 5455.704941749573,
+      "loss": 0.1188,
+      "grad_norm": 0.09035109728574753,
+      "learning_rate": 0.00013081761006289308,
+      "epoch": 0.350625
+    },
+    {
+      "step": 1162,
+      "timestamp": "2025-12-28T10:29:07.506551",
+      "elapsed_time": 5461.791641712189,
+      "loss": 0.2134,
+      "grad_norm": 0.16012340784072876,
+      "learning_rate": 0.00013069182389937106,
+      "epoch": 0.35125
+    },
+    {
+      "step": 1163,
+      "timestamp": "2025-12-28T10:29:19.642475",
+      "elapsed_time": 5473.927566051483,
+      "loss": 0.1431,
+      "grad_norm": 0.10316343605518341,
+      "learning_rate": 0.00013056603773584907,
+      "epoch": 0.351875
+    },
+    {
+      "step": 1164,
+      "timestamp": "2025-12-28T10:29:26.703778",
+      "elapsed_time": 5480.988868236542,
+      "loss": 0.1874,
+      "grad_norm": 0.13070017099380493,
+      "learning_rate": 0.00013044025157232705,
+      "epoch": 0.3525
+    },
+    {
+      "step": 1165,
+      "timestamp": "2025-12-28T10:29:42.102895",
+      "elapsed_time": 5496.387985467911,
+      "loss": 0.1539,
+      "grad_norm": 0.09859879314899445,
+      "learning_rate": 0.00013031446540880503,
+      "epoch": 0.353125
+    },
+    {
+      "step": 1166,
+      "timestamp": "2025-12-28T10:29:58.650069",
+      "elapsed_time": 5512.935160160065,
+      "loss": 0.1449,
+      "grad_norm": 0.09668051451444626,
+      "learning_rate": 0.000130188679245283,
+      "epoch": 0.35375
+    },
+    {
+      "step": 1167,
+      "timestamp": "2025-12-28T10:30:17.397139",
+      "elapsed_time": 5531.682229757309,
+      "loss": 0.1166,
+      "grad_norm": 0.07741539925336838,
+      "learning_rate": 0.000130062893081761,
+      "epoch": 0.354375
+    },
+    {
+      "step": 1168,
+      "timestamp": "2025-12-28T10:30:25.144695",
+      "elapsed_time": 5539.429785490036,
+      "loss": 0.1554,
+      "grad_norm": 0.11451406031847,
+      "learning_rate": 0.000129937106918239,
+      "epoch": 0.355
+    },
+    {
+      "step": 1169,
+      "timestamp": "2025-12-28T10:30:37.830348",
+      "elapsed_time": 5552.115438699722,
+      "loss": 0.1541,
+      "grad_norm": 0.1345360428094864,
+      "learning_rate": 0.00012981132075471697,
+      "epoch": 0.355625
+    },
+    {
+      "step": 1170,
+      "timestamp": "2025-12-28T10:30:45.645731",
+      "elapsed_time": 5559.930821657181,
+      "loss": 0.2273,
+      "grad_norm": 0.15822488069534302,
+      "learning_rate": 0.00012968553459119498,
+      "epoch": 0.35625
+    },
+    {
+      "step": 1171,
+      "timestamp": "2025-12-28T10:30:53.952954",
+      "elapsed_time": 5568.238044023514,
+      "loss": 0.1536,
+      "grad_norm": 0.2116500437259674,
+      "learning_rate": 0.00012955974842767296,
+      "epoch": 0.356875
+    },
+    {
+      "step": 1172,
+      "timestamp": "2025-12-28T10:31:03.335531",
+      "elapsed_time": 5577.620621681213,
+      "loss": 0.1236,
+      "grad_norm": 0.11660629510879517,
+      "learning_rate": 0.00012943396226415094,
+      "epoch": 0.3575
+    },
+    {
+      "step": 1173,
+      "timestamp": "2025-12-28T10:31:10.218223",
+      "elapsed_time": 5584.503312826157,
+      "loss": 0.2065,
+      "grad_norm": 0.135480597615242,
+      "learning_rate": 0.00012930817610062894,
+      "epoch": 0.358125
+    },
+    {
+      "step": 1174,
+      "timestamp": "2025-12-28T10:31:18.995819",
+      "elapsed_time": 5593.2809092998505,
+      "loss": 0.1724,
+      "grad_norm": 0.23346908390522003,
+      "learning_rate": 0.00012918238993710692,
+      "epoch": 0.35875
+    },
+    {
+      "step": 1175,
+      "timestamp": "2025-12-28T10:31:26.934005",
+      "elapsed_time": 5601.2190997600555,
+      "loss": 0.493,
+      "grad_norm": 0.18534159660339355,
+      "learning_rate": 0.0001290566037735849,
+      "epoch": 0.359375
+    },
+    {
+      "step": 1176,
+      "timestamp": "2025-12-28T10:31:35.566298",
+      "elapsed_time": 5609.851388692856,
+      "loss": 0.18,
+      "grad_norm": 0.1343761831521988,
+      "learning_rate": 0.0001289308176100629,
+      "epoch": 0.36
+    },
+    {
+      "step": 1177,
+      "timestamp": "2025-12-28T10:31:43.785332",
+      "elapsed_time": 5618.070422172546,
+      "loss": 0.1392,
+      "grad_norm": 0.12244327366352081,
+      "learning_rate": 0.00012880503144654088,
+      "epoch": 0.360625
+    },
+    {
+      "step": 1178,
+      "timestamp": "2025-12-28T10:31:49.452840",
+      "elapsed_time": 5623.737930774689,
+      "loss": 0.2027,
+      "grad_norm": 0.15840069949626923,
+      "learning_rate": 0.00012867924528301886,
+      "epoch": 0.36125
+    },
+    {
+      "step": 1179,
+      "timestamp": "2025-12-28T10:31:58.279487",
+      "elapsed_time": 5632.564577817917,
+      "loss": 0.24,
+      "grad_norm": 0.17348845303058624,
+      "learning_rate": 0.00012855345911949684,
+      "epoch": 0.361875
+    },
+    {
+      "step": 1180,
+      "timestamp": "2025-12-28T10:32:06.750060",
+      "elapsed_time": 5641.035150527954,
+      "loss": 0.1493,
+      "grad_norm": 0.11318610608577728,
+      "learning_rate": 0.00012842767295597485,
+      "epoch": 0.3625
+    },
+    {
+      "step": 1181,
+      "timestamp": "2025-12-28T10:32:18.316743",
+      "elapsed_time": 5652.601833343506,
+      "loss": 0.1519,
+      "grad_norm": 0.1193518340587616,
+      "learning_rate": 0.00012830188679245283,
+      "epoch": 0.363125
+    },
+    {
+      "step": 1182,
+      "timestamp": "2025-12-28T10:32:24.831408",
+      "elapsed_time": 5659.116497993469,
+      "loss": 0.1801,
+      "grad_norm": 0.14156877994537354,
+      "learning_rate": 0.0001281761006289308,
+      "epoch": 0.36375
+    },
+    {
+      "step": 1183,
+      "timestamp": "2025-12-28T10:32:31.901184",
+      "elapsed_time": 5666.186274528503,
+      "loss": 0.1667,
+      "grad_norm": 0.13352788984775543,
+      "learning_rate": 0.0001280503144654088,
+      "epoch": 0.364375
+    },
+    {
+      "step": 1184,
+      "timestamp": "2025-12-28T10:32:46.767013",
+      "elapsed_time": 5681.052104473114,
+      "loss": 0.1224,
+      "grad_norm": 0.09199526906013489,
+      "learning_rate": 0.0001279245283018868,
+      "epoch": 0.365
+    },
+    {
+      "step": 1185,
+      "timestamp": "2025-12-28T10:32:54.235718",
+      "elapsed_time": 5688.52080821991,
+      "loss": 0.1942,
+      "grad_norm": 0.13854575157165527,
+      "learning_rate": 0.00012779874213836477,
+      "epoch": 0.365625
+    },
+    {
+      "step": 1186,
+      "timestamp": "2025-12-28T10:33:07.216737",
+      "elapsed_time": 5701.501827001572,
+      "loss": 0.1412,
+      "grad_norm": 0.10089116543531418,
+      "learning_rate": 0.00012767295597484278,
+      "epoch": 0.36625
+    },
+    {
+      "step": 1187,
+      "timestamp": "2025-12-28T10:33:18.169177",
+      "elapsed_time": 5712.454270601273,
+      "loss": 0.1496,
+      "grad_norm": 0.11052452027797699,
+      "learning_rate": 0.00012754716981132076,
+      "epoch": 0.366875
+    },
+    {
+      "step": 1188,
+      "timestamp": "2025-12-28T10:33:24.179712",
+      "elapsed_time": 5718.464802980423,
+      "loss": 0.1801,
+      "grad_norm": 0.17774660885334015,
+      "learning_rate": 0.00012742138364779874,
+      "epoch": 0.3675
+    },
+    {
+      "step": 1189,
+      "timestamp": "2025-12-28T10:33:30.449798",
+      "elapsed_time": 5724.734888315201,
+      "loss": 0.1707,
+      "grad_norm": 0.13715173304080963,
+      "learning_rate": 0.00012729559748427674,
+      "epoch": 0.368125
+    },
+    {
+      "step": 1190,
+      "timestamp": "2025-12-28T10:33:40.351650",
+      "elapsed_time": 5734.636740446091,
+      "loss": 0.1585,
+      "grad_norm": 0.13463056087493896,
+      "learning_rate": 0.00012716981132075472,
+      "epoch": 0.36875
+    },
+    {
+      "step": 1191,
+      "timestamp": "2025-12-28T10:33:46.518785",
+      "elapsed_time": 5740.803875684738,
+      "loss": 0.2042,
+      "grad_norm": 0.14503754675388336,
+      "learning_rate": 0.0001270440251572327,
+      "epoch": 0.369375
+    },
+    {
+      "step": 1192,
+      "timestamp": "2025-12-28T10:33:58.192196",
+      "elapsed_time": 5752.477287054062,
+      "loss": 0.1911,
+      "grad_norm": 0.11271210759878159,
+      "learning_rate": 0.00012691823899371068,
+      "epoch": 0.37
+    },
+    {
+      "step": 1193,
+      "timestamp": "2025-12-28T10:34:05.329085",
+      "elapsed_time": 5759.614175319672,
+      "loss": 0.2337,
+      "grad_norm": 0.14175988733768463,
+      "learning_rate": 0.00012679245283018869,
+      "epoch": 0.370625
+    },
+    {
+      "step": 1194,
+      "timestamp": "2025-12-28T10:34:17.155544",
+      "elapsed_time": 5771.440634012222,
+      "loss": 0.1992,
+      "grad_norm": 0.10490947216749191,
+      "learning_rate": 0.00012666666666666666,
+      "epoch": 0.37125
+    },
+    {
+      "step": 1195,
+      "timestamp": "2025-12-28T10:34:32.490446",
+      "elapsed_time": 5786.77553653717,
+      "loss": 0.1565,
+      "grad_norm": 0.11939844489097595,
+      "learning_rate": 0.00012654088050314464,
+      "epoch": 0.371875
+    },
+    {
+      "step": 1196,
+      "timestamp": "2025-12-28T10:34:41.906197",
+      "elapsed_time": 5796.191288232803,
+      "loss": 0.1493,
+      "grad_norm": 0.11298428475856781,
+      "learning_rate": 0.00012641509433962265,
+      "epoch": 0.3725
+    },
+    {
+      "step": 1197,
+      "timestamp": "2025-12-28T10:34:49.971838",
+      "elapsed_time": 5804.256928920746,
+      "loss": 0.2001,
+      "grad_norm": 0.13944107294082642,
+      "learning_rate": 0.00012628930817610063,
+      "epoch": 0.373125
+    },
+    {
+      "step": 1198,
+      "timestamp": "2025-12-28T10:34:57.032417",
+      "elapsed_time": 5811.317508220673,
+      "loss": 0.2897,
+      "grad_norm": 0.1779128611087799,
+      "learning_rate": 0.0001261635220125786,
+      "epoch": 0.37375
+    },
+    {
+      "step": 1199,
+      "timestamp": "2025-12-28T10:35:08.870827",
+      "elapsed_time": 5823.1559183597565,
+      "loss": 0.1722,
+      "grad_norm": 0.11074298620223999,
+      "learning_rate": 0.00012603773584905661,
+      "epoch": 0.374375
+    },
+    {
+      "step": 1200,
+      "timestamp": "2025-12-28T10:35:14.846064",
+      "elapsed_time": 5829.131155014038,
+      "loss": 0.1623,
+      "grad_norm": 0.12961973249912262,
+      "learning_rate": 0.0001259119496855346,
+      "epoch": 0.375
+    },
+    {
+      "step": 1201,
+      "timestamp": "2025-12-28T10:35:25.422373",
+      "elapsed_time": 5839.707467794418,
+      "loss": 0.1785,
+      "grad_norm": 0.14644969999790192,
+      "learning_rate": 0.00012578616352201257,
+      "epoch": 0.375625
+    },
+    {
+      "step": 1202,
+      "timestamp": "2025-12-28T10:35:41.289838",
+      "elapsed_time": 5855.574928283691,
+      "loss": 0.118,
+      "grad_norm": 0.0863247960805893,
+      "learning_rate": 0.00012566037735849058,
+      "epoch": 0.37625
+    },
+    {
+      "step": 1203,
+      "timestamp": "2025-12-28T10:35:54.961911",
+      "elapsed_time": 5869.247001171112,
+      "loss": 0.1602,
+      "grad_norm": 0.09607759863138199,
+      "learning_rate": 0.00012553459119496856,
+      "epoch": 0.376875
+    },
+    {
+      "step": 1204,
+      "timestamp": "2025-12-28T10:36:02.141105",
+      "elapsed_time": 5876.426195144653,
+      "loss": 0.1269,
+      "grad_norm": 0.12356162816286087,
+      "learning_rate": 0.00012540880503144654,
+      "epoch": 0.3775
+    },
+    {
+      "step": 1205,
+      "timestamp": "2025-12-28T10:36:15.945952",
+      "elapsed_time": 5890.23104262352,
+      "loss": 0.1182,
+      "grad_norm": 0.09206572920084,
+      "learning_rate": 0.00012528301886792452,
+      "epoch": 0.378125
+    },
+    {
+      "step": 1206,
+      "timestamp": "2025-12-28T10:36:23.429110",
+      "elapsed_time": 5897.714200735092,
+      "loss": 0.2265,
+      "grad_norm": 0.14600634574890137,
+      "learning_rate": 0.00012515723270440252,
+      "epoch": 0.37875
+    },
+    {
+      "step": 1207,
+      "timestamp": "2025-12-28T10:36:32.859599",
+      "elapsed_time": 5907.1446895599365,
+      "loss": 0.1366,
+      "grad_norm": 0.10805106908082962,
+      "learning_rate": 0.0001250314465408805,
+      "epoch": 0.379375
+    },
+    {
+      "step": 1208,
+      "timestamp": "2025-12-28T10:36:41.692070",
+      "elapsed_time": 5915.9771609306335,
+      "loss": 0.1532,
+      "grad_norm": 0.12475503236055374,
+      "learning_rate": 0.00012490566037735848,
+      "epoch": 0.38
+    },
+    {
+      "step": 1209,
+      "timestamp": "2025-12-28T10:36:52.893121",
+      "elapsed_time": 5927.178210735321,
+      "loss": 0.1372,
+      "grad_norm": 0.10797689855098724,
+      "learning_rate": 0.00012477987421383649,
+      "epoch": 0.380625
+    },
+    {
+      "step": 1210,
+      "timestamp": "2025-12-28T10:37:05.316752",
+      "elapsed_time": 5939.601842880249,
+      "loss": 0.1379,
+      "grad_norm": 0.09704222530126572,
+      "learning_rate": 0.00012465408805031447,
+      "epoch": 0.38125
+    },
+    {
+      "step": 1211,
+      "timestamp": "2025-12-28T10:37:10.877159",
+      "elapsed_time": 5945.1622495651245,
+      "loss": 0.2735,
+      "grad_norm": 0.16764011979103088,
+      "learning_rate": 0.00012452830188679244,
+      "epoch": 0.381875
+    },
+    {
+      "step": 1212,
+      "timestamp": "2025-12-28T10:37:19.026917",
+      "elapsed_time": 5953.312007665634,
+      "loss": 0.2679,
+      "grad_norm": 0.1391536146402359,
+      "learning_rate": 0.00012440251572327045,
+      "epoch": 0.3825
+    },
+    {
+      "step": 1213,
+      "timestamp": "2025-12-28T10:37:28.216063",
+      "elapsed_time": 5962.501153230667,
+      "loss": 0.1346,
+      "grad_norm": 0.1069680005311966,
+      "learning_rate": 0.00012427672955974843,
+      "epoch": 0.383125
+    },
+    {
+      "step": 1214,
+      "timestamp": "2025-12-28T10:37:42.338346",
+      "elapsed_time": 5976.623436450958,
+      "loss": 0.1306,
+      "grad_norm": 0.0998351201415062,
+      "learning_rate": 0.0001241509433962264,
+      "epoch": 0.38375
+    },
+    {
+      "step": 1215,
+      "timestamp": "2025-12-28T10:37:49.904441",
+      "elapsed_time": 5984.189530849457,
+      "loss": 0.1815,
+      "grad_norm": 0.13117118179798126,
+      "learning_rate": 0.00012402515723270442,
+      "epoch": 0.384375
+    },
+    {
+      "step": 1216,
+      "timestamp": "2025-12-28T10:38:03.854769",
+      "elapsed_time": 5998.1398594379425,
+      "loss": 0.3325,
+      "grad_norm": 0.13191911578178406,
+      "learning_rate": 0.0001238993710691824,
+      "epoch": 0.385
+    },
+    {
+      "step": 1217,
+      "timestamp": "2025-12-28T10:38:08.710052",
+      "elapsed_time": 6002.995143175125,
+      "loss": 0.7795,
+      "grad_norm": 0.23979060351848602,
+      "learning_rate": 0.00012377358490566037,
+      "epoch": 0.385625
+    },
+    {
+      "step": 1218,
+      "timestamp": "2025-12-28T10:38:22.340339",
+      "elapsed_time": 6016.625429391861,
+      "loss": 0.1357,
+      "grad_norm": 0.0968586727976799,
+      "learning_rate": 0.00012364779874213835,
+      "epoch": 0.38625
+    },
+    {
+      "step": 1219,
+      "timestamp": "2025-12-28T10:38:42.480374",
+      "elapsed_time": 6036.7654638290405,
+      "loss": 0.1035,
+      "grad_norm": 0.08069409430027008,
+      "learning_rate": 0.00012352201257861636,
+      "epoch": 0.386875
+    },
+    {
+      "step": 1220,
+      "timestamp": "2025-12-28T10:38:50.725844",
+      "elapsed_time": 6045.010933876038,
+      "loss": 0.1788,
+      "grad_norm": 0.12745045125484467,
+      "learning_rate": 0.00012339622641509434,
+      "epoch": 0.3875
+    },
+    {
+      "step": 1221,
+      "timestamp": "2025-12-28T10:38:56.392405",
+      "elapsed_time": 6050.677495241165,
+      "loss": 0.2797,
+      "grad_norm": 0.29041117429733276,
+      "learning_rate": 0.00012327044025157232,
+      "epoch": 0.388125
+    },
+    {
+      "step": 1222,
+      "timestamp": "2025-12-28T10:39:05.731705",
+      "elapsed_time": 6060.016795396805,
+      "loss": 0.2435,
+      "grad_norm": 0.12010473012924194,
+      "learning_rate": 0.00012314465408805032,
+      "epoch": 0.38875
+    },
+    {
+      "step": 1223,
+      "timestamp": "2025-12-28T10:39:13.957976",
+      "elapsed_time": 6068.243066072464,
+      "loss": 0.1408,
+      "grad_norm": 0.10792548954486847,
+      "learning_rate": 0.0001230188679245283,
+      "epoch": 0.389375
+    },
+    {
+      "step": 1224,
+      "timestamp": "2025-12-28T10:39:25.907397",
+      "elapsed_time": 6080.192487239838,
+      "loss": 0.15,
+      "grad_norm": 0.10715028643608093,
+      "learning_rate": 0.00012289308176100628,
+      "epoch": 0.39
+    },
+    {
+      "step": 1225,
+      "timestamp": "2025-12-28T10:39:31.137562",
+      "elapsed_time": 6085.422652006149,
+      "loss": 0.1641,
+      "grad_norm": 0.13656176626682281,
+      "learning_rate": 0.0001227672955974843,
+      "epoch": 0.390625
+    },
+    {
+      "step": 1226,
+      "timestamp": "2025-12-28T10:39:34.990548",
+      "elapsed_time": 6089.275638103485,
+      "loss": 0.2354,
+      "grad_norm": 0.1691398322582245,
+      "learning_rate": 0.00012264150943396227,
+      "epoch": 0.39125
+    },
+    {
+      "step": 1227,
+      "timestamp": "2025-12-28T10:39:47.535666",
+      "elapsed_time": 6101.820756912231,
+      "loss": 0.1597,
+      "grad_norm": 0.10032006353139877,
+      "learning_rate": 0.00012251572327044025,
+      "epoch": 0.391875
+    },
+    {
+      "step": 1228,
+      "timestamp": "2025-12-28T10:39:54.871871",
+      "elapsed_time": 6109.156960964203,
+      "loss": 0.2194,
+      "grad_norm": 0.1799350529909134,
+      "learning_rate": 0.00012238993710691825,
+      "epoch": 0.3925
+    },
+    {
+      "step": 1229,
+      "timestamp": "2025-12-28T10:40:05.715326",
+      "elapsed_time": 6120.000416755676,
+      "loss": 0.1546,
+      "grad_norm": 0.10520683974027634,
+      "learning_rate": 0.00012226415094339623,
+      "epoch": 0.393125
+    },
+    {
+      "step": 1230,
+      "timestamp": "2025-12-28T10:40:15.438855",
+      "elapsed_time": 6129.723945379257,
+      "loss": 0.182,
+      "grad_norm": 0.13104449212551117,
+      "learning_rate": 0.0001221383647798742,
+      "epoch": 0.39375
+    },
+    {
+      "step": 1231,
+      "timestamp": "2025-12-28T10:40:26.329143",
+      "elapsed_time": 6140.614233493805,
+      "loss": 0.1419,
+      "grad_norm": 0.09966589510440826,
+      "learning_rate": 0.0001220125786163522,
+      "epoch": 0.394375
+    },
+    {
+      "step": 1232,
+      "timestamp": "2025-12-28T10:40:31.971296",
+      "elapsed_time": 6146.256386756897,
+      "loss": 0.1358,
+      "grad_norm": 0.12772534787654877,
+      "learning_rate": 0.0001218867924528302,
+      "epoch": 0.395
+    },
+    {
+      "step": 1233,
+      "timestamp": "2025-12-28T10:40:39.345570",
+      "elapsed_time": 6153.630660057068,
+      "loss": 0.161,
+      "grad_norm": 0.11391877382993698,
+      "learning_rate": 0.00012176100628930817,
+      "epoch": 0.395625
+    },
+    {
+      "step": 1234,
+      "timestamp": "2025-12-28T10:40:52.265078",
+      "elapsed_time": 6166.550168275833,
+      "loss": 0.1591,
+      "grad_norm": 0.1011412963271141,
+      "learning_rate": 0.00012163522012578617,
+      "epoch": 0.39625
+    },
+    {
+      "step": 1235,
+      "timestamp": "2025-12-28T10:41:00.285752",
+      "elapsed_time": 6174.57084274292,
+      "loss": 0.1064,
+      "grad_norm": 0.10651546716690063,
+      "learning_rate": 0.00012150943396226415,
+      "epoch": 0.396875
+    },
+    {
+      "step": 1236,
+      "timestamp": "2025-12-28T10:41:08.312804",
+      "elapsed_time": 6182.597895145416,
+      "loss": 0.1547,
+      "grad_norm": 0.13378091156482697,
+      "learning_rate": 0.00012138364779874214,
+      "epoch": 0.3975
+    },
+    {
+      "step": 1237,
+      "timestamp": "2025-12-28T10:41:17.897328",
+      "elapsed_time": 6192.182418823242,
+      "loss": 0.1642,
+      "grad_norm": 0.1135389655828476,
+      "learning_rate": 0.00012125786163522013,
+      "epoch": 0.398125
+    },
+    {
+      "step": 1238,
+      "timestamp": "2025-12-28T10:41:27.960873",
+      "elapsed_time": 6202.245964050293,
+      "loss": 0.3449,
+      "grad_norm": 0.1613943874835968,
+      "learning_rate": 0.00012113207547169811,
+      "epoch": 0.39875
+    },
+    {
+      "step": 1239,
+      "timestamp": "2025-12-28T10:41:39.376307",
+      "elapsed_time": 6213.661397695541,
+      "loss": 0.1686,
+      "grad_norm": 0.10929451882839203,
+      "learning_rate": 0.0001210062893081761,
+      "epoch": 0.399375
+    },
+    {
+      "step": 1240,
+      "timestamp": "2025-12-28T10:41:49.452823",
+      "elapsed_time": 6223.737913370132,
+      "loss": 0.2091,
+      "grad_norm": 0.1252572238445282,
+      "learning_rate": 0.0001208805031446541,
+      "epoch": 0.4
+    },
+    {
+      "step": 1241,
+      "timestamp": "2025-12-28T10:41:55.431783",
+      "elapsed_time": 6229.716873407364,
+      "loss": 0.2126,
+      "grad_norm": 0.15598464012145996,
+      "learning_rate": 0.00012075471698113207,
+      "epoch": 0.400625
+    },
+    {
+      "step": 1242,
+      "timestamp": "2025-12-28T10:42:04.291713",
+      "elapsed_time": 6238.5768032073975,
+      "loss": 0.4607,
+      "grad_norm": 0.16947001218795776,
+      "learning_rate": 0.00012062893081761007,
+      "epoch": 0.40125
+    },
+    {
+      "step": 1243,
+      "timestamp": "2025-12-28T10:42:16.315554",
+      "elapsed_time": 6250.600644350052,
+      "loss": 0.1142,
+      "grad_norm": 0.09022119641304016,
+      "learning_rate": 0.00012050314465408805,
+      "epoch": 0.401875
+    },
+    {
+      "step": 1244,
+      "timestamp": "2025-12-28T10:42:24.823459",
+      "elapsed_time": 6259.108549118042,
+      "loss": 0.3034,
+      "grad_norm": 0.14146797358989716,
+      "learning_rate": 0.00012037735849056604,
+      "epoch": 0.4025
+    },
+    {
+      "step": 1245,
+      "timestamp": "2025-12-28T10:42:37.853137",
+      "elapsed_time": 6272.138226747513,
+      "loss": 0.1124,
+      "grad_norm": 0.11186513304710388,
+      "learning_rate": 0.00012025157232704403,
+      "epoch": 0.403125
+    },
+    {
+      "step": 1246,
+      "timestamp": "2025-12-28T10:42:45.407348",
+      "elapsed_time": 6279.6924386024475,
+      "loss": 0.1426,
+      "grad_norm": 0.11227832734584808,
+      "learning_rate": 0.00012012578616352201,
+      "epoch": 0.40375
+    },
+    {
+      "step": 1247,
+      "timestamp": "2025-12-28T10:42:53.466547",
+      "elapsed_time": 6287.751641750336,
+      "loss": 0.1833,
+      "grad_norm": 0.12571921944618225,
+      "learning_rate": 0.00012,
+      "epoch": 0.404375
+    },
+    {
+      "step": 1248,
+      "timestamp": "2025-12-28T10:43:02.403855",
+      "elapsed_time": 6296.688944816589,
+      "loss": 0.2413,
+      "grad_norm": 0.26408830285072327,
+      "learning_rate": 0.00011987421383647798,
+      "epoch": 0.405
+    },
+    {
+      "step": 1249,
+      "timestamp": "2025-12-28T10:43:11.586719",
+      "elapsed_time": 6305.871813774109,
+      "loss": 0.1202,
+      "grad_norm": 0.11835648864507675,
+      "learning_rate": 0.00011974842767295597,
+      "epoch": 0.405625
+    },
+    {
+      "step": 1250,
+      "timestamp": "2025-12-28T10:43:17.695080",
+      "elapsed_time": 6311.9801704883575,
+      "loss": 0.2076,
+      "grad_norm": 0.24628107249736786,
+      "learning_rate": 0.00011962264150943397,
+      "epoch": 0.40625
+    },
+    {
+      "step": 1251,
+      "timestamp": "2025-12-28T10:43:26.526357",
+      "elapsed_time": 6320.811446428299,
+      "loss": 0.3627,
+      "grad_norm": 0.14971527457237244,
+      "learning_rate": 0.00011949685534591195,
+      "epoch": 0.406875
+    },
+    {
+      "step": 1252,
+      "timestamp": "2025-12-28T10:43:32.439133",
+      "elapsed_time": 6326.724223136902,
+      "loss": 0.2245,
+      "grad_norm": 0.16869717836380005,
+      "learning_rate": 0.00011937106918238994,
+      "epoch": 0.4075
+    },
+    {
+      "step": 1253,
+      "timestamp": "2025-12-28T10:43:40.414515",
+      "elapsed_time": 6334.699605703354,
+      "loss": 0.1473,
+      "grad_norm": 0.12115878611803055,
+      "learning_rate": 0.00011924528301886793,
+      "epoch": 0.408125
+    },
+    {
+      "step": 1254,
+      "timestamp": "2025-12-28T10:43:47.555112",
+      "elapsed_time": 6341.840202093124,
+      "loss": 0.1415,
+      "grad_norm": 0.3590112328529358,
+      "learning_rate": 0.00011911949685534591,
+      "epoch": 0.40875
+    },
+    {
+      "step": 1255,
+      "timestamp": "2025-12-28T10:43:59.076934",
+      "elapsed_time": 6353.36202454567,
+      "loss": 0.166,
+      "grad_norm": 0.13714636862277985,
+      "learning_rate": 0.0001189937106918239,
+      "epoch": 0.409375
+    },
+    {
+      "step": 1256,
+      "timestamp": "2025-12-28T10:44:07.141027",
+      "elapsed_time": 6361.426117897034,
+      "loss": 0.1308,
+      "grad_norm": 0.1114254966378212,
+      "learning_rate": 0.00011886792452830188,
+      "epoch": 0.41
+    },
+    {
+      "step": 1257,
+      "timestamp": "2025-12-28T10:44:14.960681",
+      "elapsed_time": 6369.2457716465,
+      "loss": 0.1211,
+      "grad_norm": 0.10369876772165298,
+      "learning_rate": 0.00011874213836477988,
+      "epoch": 0.410625
+    },
+    {
+      "step": 1258,
+      "timestamp": "2025-12-28T10:44:21.307788",
+      "elapsed_time": 6375.592882871628,
+      "loss": 0.1538,
+      "grad_norm": 0.11906347423791885,
+      "learning_rate": 0.00011861635220125787,
+      "epoch": 0.41125
+    },
+    {
+      "step": 1259,
+      "timestamp": "2025-12-28T10:44:34.306623",
+      "elapsed_time": 6388.5917139053345,
+      "loss": 0.1692,
+      "grad_norm": 0.1116495355963707,
+      "learning_rate": 0.00011849056603773585,
+      "epoch": 0.411875
+    },
+    {
+      "step": 1260,
+      "timestamp": "2025-12-28T10:44:46.440601",
+      "elapsed_time": 6400.725692033768,
+      "loss": 0.1485,
+      "grad_norm": 0.1158275455236435,
+      "learning_rate": 0.00011836477987421384,
+      "epoch": 0.4125
+    },
+    {
+      "step": 1261,
+      "timestamp": "2025-12-28T10:44:51.167156",
+      "elapsed_time": 6405.452246665955,
+      "loss": 0.1906,
+      "grad_norm": 0.17199426889419556,
+      "learning_rate": 0.00011823899371069183,
+      "epoch": 0.413125
+    },
+    {
+      "step": 1262,
+      "timestamp": "2025-12-28T10:44:55.766324",
+      "elapsed_time": 6410.051414012909,
+      "loss": 0.5447,
+      "grad_norm": 0.2059682309627533,
+      "learning_rate": 0.00011811320754716981,
+      "epoch": 0.41375
+    },
+    {
+      "step": 1263,
+      "timestamp": "2025-12-28T10:44:59.626300",
+      "elapsed_time": 6413.911394357681,
+      "loss": 0.2163,
+      "grad_norm": 0.16556908190250397,
+      "learning_rate": 0.0001179874213836478,
+      "epoch": 0.414375
+    },
+    {
+      "step": 1264,
+      "timestamp": "2025-12-28T10:45:10.040506",
+      "elapsed_time": 6424.32559633255,
+      "loss": 0.1534,
+      "grad_norm": 0.10824116319417953,
+      "learning_rate": 0.00011786163522012578,
+      "epoch": 0.415
+    },
+    {
+      "step": 1265,
+      "timestamp": "2025-12-28T10:45:18.021995",
+      "elapsed_time": 6432.307085752487,
+      "loss": 0.2078,
+      "grad_norm": 0.1333610564470291,
+      "learning_rate": 0.00011773584905660378,
+      "epoch": 0.415625
+    },
+    {
+      "step": 1266,
+      "timestamp": "2025-12-28T10:45:27.112975",
+      "elapsed_time": 6441.398065567017,
+      "loss": 0.1441,
+      "grad_norm": 0.12403620034456253,
+      "learning_rate": 0.00011761006289308177,
+      "epoch": 0.41625
+    },
+    {
+      "step": 1267,
+      "timestamp": "2025-12-28T10:45:33.196456",
+      "elapsed_time": 6447.4815464019775,
+      "loss": 0.3569,
+      "grad_norm": 0.16776348650455475,
+      "learning_rate": 0.00011748427672955975,
+      "epoch": 0.416875
+    },
+    {
+      "step": 1268,
+      "timestamp": "2025-12-28T10:45:44.984909",
+      "elapsed_time": 6459.269999265671,
+      "loss": 0.1378,
+      "grad_norm": 0.10034506022930145,
+      "learning_rate": 0.00011735849056603774,
+      "epoch": 0.4175
+    },
+    {
+      "step": 1269,
+      "timestamp": "2025-12-28T10:45:53.809026",
+      "elapsed_time": 6468.0941162109375,
+      "loss": 0.2125,
+      "grad_norm": 0.1351146399974823,
+      "learning_rate": 0.00011723270440251572,
+      "epoch": 0.418125
+    },
+    {
+      "step": 1270,
+      "timestamp": "2025-12-28T10:45:58.659300",
+      "elapsed_time": 6472.944390773773,
+      "loss": 0.1865,
+      "grad_norm": 0.14262175559997559,
+      "learning_rate": 0.00011710691823899371,
+      "epoch": 0.41875
+    },
+    {
+      "step": 1271,
+      "timestamp": "2025-12-28T10:46:04.637730",
+      "elapsed_time": 6478.922821044922,
+      "loss": 0.194,
+      "grad_norm": 0.14949339628219604,
+      "learning_rate": 0.0001169811320754717,
+      "epoch": 0.419375
+    },
+    {
+      "step": 1272,
+      "timestamp": "2025-12-28T10:46:12.419593",
+      "elapsed_time": 6486.704683542252,
+      "loss": 0.169,
+      "grad_norm": 0.12632359564304352,
+      "learning_rate": 0.00011685534591194968,
+      "epoch": 0.42
+    },
+    {
+      "step": 1273,
+      "timestamp": "2025-12-28T10:46:28.716300",
+      "elapsed_time": 6503.0013909339905,
+      "loss": 0.115,
+      "grad_norm": 0.08015146851539612,
+      "learning_rate": 0.00011672955974842768,
+      "epoch": 0.420625
+    },
+    {
+      "step": 1274,
+      "timestamp": "2025-12-28T10:46:32.620770",
+      "elapsed_time": 6506.905859708786,
+      "loss": 0.1909,
+      "grad_norm": 0.16493003070354462,
+      "learning_rate": 0.00011660377358490567,
+      "epoch": 0.42125
+    },
+    {
+      "step": 1275,
+      "timestamp": "2025-12-28T10:46:43.229055",
+      "elapsed_time": 6517.5141451358795,
+      "loss": 0.166,
+      "grad_norm": 0.10563270002603531,
+      "learning_rate": 0.00011647798742138365,
+      "epoch": 0.421875
+    },
+    {
+      "step": 1276,
+      "timestamp": "2025-12-28T10:46:50.110258",
+      "elapsed_time": 6524.395348310471,
+      "loss": 0.2099,
+      "grad_norm": 0.1380920559167862,
+      "learning_rate": 0.00011635220125786164,
+      "epoch": 0.4225
+    },
+    {
+      "step": 1277,
+      "timestamp": "2025-12-28T10:46:58.322692",
+      "elapsed_time": 6532.607782125473,
+      "loss": 0.1598,
+      "grad_norm": 0.11996757984161377,
+      "learning_rate": 0.00011622641509433962,
+      "epoch": 0.423125
+    },
+    {
+      "step": 1278,
+      "timestamp": "2025-12-28T10:47:03.207181",
+      "elapsed_time": 6537.49227142334,
+      "loss": 0.1789,
+      "grad_norm": 0.14649128913879395,
+      "learning_rate": 0.00011610062893081761,
+      "epoch": 0.42375
+    },
+    {
+      "step": 1279,
+      "timestamp": "2025-12-28T10:47:09.522712",
+      "elapsed_time": 6543.807803153992,
+      "loss": 0.1889,
+      "grad_norm": 0.1377793848514557,
+      "learning_rate": 0.0001159748427672956,
+      "epoch": 0.424375
+    },
+    {
+      "step": 1280,
+      "timestamp": "2025-12-28T10:47:18.362641",
+      "elapsed_time": 6552.6477308273315,
+      "loss": 0.1573,
+      "grad_norm": 0.11395595967769623,
+      "learning_rate": 0.00011584905660377358,
+      "epoch": 0.425
+    },
+    {
+      "step": 1281,
+      "timestamp": "2025-12-28T10:47:24.849601",
+      "elapsed_time": 6559.134691953659,
+      "loss": 0.3999,
+      "grad_norm": 0.17579996585845947,
+      "learning_rate": 0.00011572327044025158,
+      "epoch": 0.425625
+    },
+    {
+      "step": 1282,
+      "timestamp": "2025-12-28T10:47:37.488885",
+      "elapsed_time": 6571.773975133896,
+      "loss": 0.1319,
+      "grad_norm": 0.09526728838682175,
+      "learning_rate": 0.00011559748427672956,
+      "epoch": 0.42625
+    },
+    {
+      "step": 1283,
+      "timestamp": "2025-12-28T10:47:45.629523",
+      "elapsed_time": 6579.914613485336,
+      "loss": 0.1803,
+      "grad_norm": 0.12247025221586227,
+      "learning_rate": 0.00011547169811320755,
+      "epoch": 0.426875
+    },
+    {
+      "step": 1284,
+      "timestamp": "2025-12-28T10:47:50.156209",
+      "elapsed_time": 6584.441299676895,
+      "loss": 0.2226,
+      "grad_norm": 0.15616583824157715,
+      "learning_rate": 0.00011534591194968554,
+      "epoch": 0.4275
+    },
+    {
+      "step": 1285,
+      "timestamp": "2025-12-28T10:47:58.291658",
+      "elapsed_time": 6592.576747894287,
+      "loss": 0.1435,
+      "grad_norm": 0.11908165365457535,
+      "learning_rate": 0.00011522012578616352,
+      "epoch": 0.428125
+    },
+    {
+      "step": 1286,
+      "timestamp": "2025-12-28T10:48:03.589513",
+      "elapsed_time": 6597.874603271484,
+      "loss": 0.253,
+      "grad_norm": 0.15733444690704346,
+      "learning_rate": 0.00011509433962264151,
+      "epoch": 0.42875
+    },
+    {
+      "step": 1287,
+      "timestamp": "2025-12-28T10:48:15.662909",
+      "elapsed_time": 6609.947999715805,
+      "loss": 0.1346,
+      "grad_norm": 0.08608326315879822,
+      "learning_rate": 0.0001149685534591195,
+      "epoch": 0.429375
+    },
+    {
+      "step": 1288,
+      "timestamp": "2025-12-28T10:48:26.180269",
+      "elapsed_time": 6620.465359449387,
+      "loss": 0.159,
+      "grad_norm": 0.10130292922258377,
+      "learning_rate": 0.00011484276729559748,
+      "epoch": 0.43
+    },
+    {
+      "step": 1289,
+      "timestamp": "2025-12-28T10:48:31.670014",
+      "elapsed_time": 6625.955104589462,
+      "loss": 0.1755,
+      "grad_norm": 0.1298661231994629,
+      "learning_rate": 0.00011471698113207548,
+      "epoch": 0.430625
+    },
+    {
+      "step": 1290,
+      "timestamp": "2025-12-28T10:48:38.047130",
+      "elapsed_time": 6632.332220554352,
+      "loss": 0.1663,
+      "grad_norm": 0.12876641750335693,
+      "learning_rate": 0.00011459119496855346,
+      "epoch": 0.43125
+    },
+    {
+      "step": 1291,
+      "timestamp": "2025-12-28T10:48:49.938884",
+      "elapsed_time": 6644.223974466324,
+      "loss": 0.1259,
+      "grad_norm": 0.08893437683582306,
+      "learning_rate": 0.00011446540880503145,
+      "epoch": 0.431875
+    },
+    {
+      "step": 1292,
+      "timestamp": "2025-12-28T10:48:56.281897",
+      "elapsed_time": 6650.566987276077,
+      "loss": 0.2474,
+      "grad_norm": 0.1418265700340271,
+      "learning_rate": 0.00011433962264150944,
+      "epoch": 0.4325
+    },
+    {
+      "step": 1293,
+      "timestamp": "2025-12-28T10:49:04.023046",
+      "elapsed_time": 6658.308136463165,
+      "loss": 0.1673,
+      "grad_norm": 0.12449798732995987,
+      "learning_rate": 0.00011421383647798742,
+      "epoch": 0.433125
+    },
+    {
+      "step": 1294,
+      "timestamp": "2025-12-28T10:49:18.892772",
+      "elapsed_time": 6673.177862167358,
+      "loss": 0.143,
+      "grad_norm": 0.10782869905233383,
+      "learning_rate": 0.00011408805031446541,
+      "epoch": 0.43375
+    },
+    {
+      "step": 1295,
+      "timestamp": "2025-12-28T10:49:29.168117",
+      "elapsed_time": 6683.45320725441,
+      "loss": 0.1301,
+      "grad_norm": 0.09740663319826126,
+      "learning_rate": 0.00011396226415094339,
+      "epoch": 0.434375
+    },
+    {
+      "step": 1296,
+      "timestamp": "2025-12-28T10:49:48.401036",
+      "elapsed_time": 6702.686126708984,
+      "loss": 0.1528,
+      "grad_norm": 0.08156444132328033,
+      "learning_rate": 0.00011383647798742138,
+      "epoch": 0.435
+    },
+    {
+      "step": 1297,
+      "timestamp": "2025-12-28T10:50:01.198414",
+      "elapsed_time": 6715.483504772186,
+      "loss": 0.1212,
+      "grad_norm": 0.09743256121873856,
+      "learning_rate": 0.00011371069182389938,
+      "epoch": 0.435625
+    },
+    {
+      "step": 1298,
+      "timestamp": "2025-12-28T10:50:09.499924",
+      "elapsed_time": 6723.785014867783,
+      "loss": 0.1612,
+      "grad_norm": 0.12712618708610535,
+      "learning_rate": 0.00011358490566037736,
+      "epoch": 0.43625
+    },
+    {
+      "step": 1299,
+      "timestamp": "2025-12-28T10:50:17.483321",
+      "elapsed_time": 6731.768411159515,
+      "loss": 0.1511,
+      "grad_norm": 0.13302955031394958,
+      "learning_rate": 0.00011345911949685535,
+      "epoch": 0.436875
+    },
+    {
+      "step": 1300,
+      "timestamp": "2025-12-28T10:50:27.725225",
+      "elapsed_time": 6742.010316371918,
+      "loss": 0.1615,
+      "grad_norm": 0.10727046430110931,
+      "learning_rate": 0.00011333333333333334,
+      "epoch": 0.4375
+    },
+    {
+      "step": 1301,
+      "timestamp": "2025-12-28T10:50:36.587116",
+      "elapsed_time": 6750.872206449509,
+      "loss": 0.2059,
+      "grad_norm": 0.15867577493190765,
+      "learning_rate": 0.00011320754716981132,
+      "epoch": 0.438125
+    },
+    {
+      "step": 1302,
+      "timestamp": "2025-12-28T10:50:41.926696",
+      "elapsed_time": 6756.21178650856,
+      "loss": 0.2088,
+      "grad_norm": 0.15406003594398499,
+      "learning_rate": 0.00011308176100628931,
+      "epoch": 0.43875
+    },
+    {
+      "step": 1303,
+      "timestamp": "2025-12-28T10:50:49.475802",
+      "elapsed_time": 6763.760892391205,
+      "loss": 0.2062,
+      "grad_norm": 0.13464148342609406,
+      "learning_rate": 0.00011295597484276729,
+      "epoch": 0.439375
+    },
+    {
+      "step": 1304,
+      "timestamp": "2025-12-28T10:50:53.491533",
+      "elapsed_time": 6767.776623249054,
+      "loss": 0.2728,
+      "grad_norm": 0.2125401347875595,
+      "learning_rate": 0.00011283018867924528,
+      "epoch": 0.44
+    },
+    {
+      "step": 1305,
+      "timestamp": "2025-12-28T10:51:02.825718",
+      "elapsed_time": 6777.110809087753,
+      "loss": 0.1534,
+      "grad_norm": 0.10923696309328079,
+      "learning_rate": 0.00011270440251572328,
+      "epoch": 0.440625
+    },
+    {
+      "step": 1306,
+      "timestamp": "2025-12-28T10:51:12.781667",
+      "elapsed_time": 6787.066757440567,
+      "loss": 0.4045,
+      "grad_norm": 0.1884467899799347,
+      "learning_rate": 0.00011257861635220126,
+      "epoch": 0.44125
+    },
+    {
+      "step": 1307,
+      "timestamp": "2025-12-28T10:51:21.728953",
+      "elapsed_time": 6796.014044284821,
+      "loss": 0.1939,
+      "grad_norm": 0.12376397848129272,
+      "learning_rate": 0.00011245283018867925,
+      "epoch": 0.441875
+    },
+    {
+      "step": 1308,
+      "timestamp": "2025-12-28T10:51:33.630122",
+      "elapsed_time": 6807.915212869644,
+      "loss": 0.1559,
+      "grad_norm": 0.09741462022066116,
+      "learning_rate": 0.00011232704402515724,
+      "epoch": 0.4425
+    },
+    {
+      "step": 1309,
+      "timestamp": "2025-12-28T10:51:47.370883",
+      "elapsed_time": 6821.65597319603,
+      "loss": 0.1371,
+      "grad_norm": 0.08514436334371567,
+      "learning_rate": 0.00011220125786163522,
+      "epoch": 0.443125
+    },
+    {
+      "step": 1310,
+      "timestamp": "2025-12-28T10:52:02.892235",
+      "elapsed_time": 6837.177325725555,
+      "loss": 0.109,
+      "grad_norm": 0.09866739064455032,
+      "learning_rate": 0.00011207547169811321,
+      "epoch": 0.44375
+    },
+    {
+      "step": 1311,
+      "timestamp": "2025-12-28T10:52:13.507601",
+      "elapsed_time": 6847.792692184448,
+      "loss": 0.1884,
+      "grad_norm": 0.10366571694612503,
+      "learning_rate": 0.00011194968553459119,
+      "epoch": 0.444375
+    },
+    {
+      "step": 1312,
+      "timestamp": "2025-12-28T10:52:23.266819",
+      "elapsed_time": 6857.551909446716,
+      "loss": 0.2008,
+      "grad_norm": 0.11828729510307312,
+      "learning_rate": 0.00011182389937106919,
+      "epoch": 0.445
+    },
+    {
+      "step": 1313,
+      "timestamp": "2025-12-28T10:52:33.470306",
+      "elapsed_time": 6867.755395889282,
+      "loss": 0.1863,
+      "grad_norm": 0.11549749970436096,
+      "learning_rate": 0.00011169811320754718,
+      "epoch": 0.445625
+    },
+    {
+      "step": 1314,
+      "timestamp": "2025-12-28T10:52:42.647558",
+      "elapsed_time": 6876.932648897171,
+      "loss": 0.1582,
+      "grad_norm": 0.11198070645332336,
+      "learning_rate": 0.00011157232704402516,
+      "epoch": 0.44625
+    },
+    {
+      "step": 1315,
+      "timestamp": "2025-12-28T10:52:52.551145",
+      "elapsed_time": 6886.8362357616425,
+      "loss": 0.1355,
+      "grad_norm": 0.10204551368951797,
+      "learning_rate": 0.00011144654088050315,
+      "epoch": 0.446875
+    },
+    {
+      "step": 1316,
+      "timestamp": "2025-12-28T10:53:04.736390",
+      "elapsed_time": 6899.021480321884,
+      "loss": 0.1699,
+      "grad_norm": 0.10276511311531067,
+      "learning_rate": 0.00011132075471698113,
+      "epoch": 0.4475
+    },
+    {
+      "step": 1317,
+      "timestamp": "2025-12-28T10:53:10.665741",
+      "elapsed_time": 6904.950830936432,
+      "loss": 0.2933,
+      "grad_norm": 0.16935305297374725,
+      "learning_rate": 0.00011119496855345912,
+      "epoch": 0.448125
+    },
+    {
+      "step": 1318,
+      "timestamp": "2025-12-28T10:53:15.139070",
+      "elapsed_time": 6909.424160718918,
+      "loss": 0.2412,
+      "grad_norm": 0.18222883343696594,
+      "learning_rate": 0.00011106918238993711,
+      "epoch": 0.44875
+    },
+    {
+      "step": 1319,
+      "timestamp": "2025-12-28T10:53:23.782033",
+      "elapsed_time": 6918.067123413086,
+      "loss": 0.128,
+      "grad_norm": 0.11177527904510498,
+      "learning_rate": 0.00011094339622641509,
+      "epoch": 0.449375
+    },
+    {
+      "step": 1320,
+      "timestamp": "2025-12-28T10:53:36.143000",
+      "elapsed_time": 6930.428090333939,
+      "loss": 0.149,
+      "grad_norm": 0.12012763321399689,
+      "learning_rate": 0.00011081761006289309,
+      "epoch": 0.45
+    },
+    {
+      "step": 1321,
+      "timestamp": "2025-12-28T10:53:44.449019",
+      "elapsed_time": 6938.734109163284,
+      "loss": 0.1378,
+      "grad_norm": 0.11260993033647537,
+      "learning_rate": 0.00011069182389937108,
+      "epoch": 0.450625
+    },
+    {
+      "step": 1322,
+      "timestamp": "2025-12-28T10:53:53.834895",
+      "elapsed_time": 6948.119985580444,
+      "loss": 0.2031,
+      "grad_norm": 0.13370606303215027,
+      "learning_rate": 0.00011056603773584906,
+      "epoch": 0.45125
+    },
+    {
+      "step": 1323,
+      "timestamp": "2025-12-28T10:54:03.643538",
+      "elapsed_time": 6957.928628444672,
+      "loss": 0.1721,
+      "grad_norm": 0.12418955564498901,
+      "learning_rate": 0.00011044025157232705,
+      "epoch": 0.451875
+    },
+    {
+      "step": 1324,
+      "timestamp": "2025-12-28T10:54:10.718971",
+      "elapsed_time": 6965.004061698914,
+      "loss": 0.217,
+      "grad_norm": 0.15134598314762115,
+      "learning_rate": 0.00011031446540880503,
+      "epoch": 0.4525
+    },
+    {
+      "step": 1325,
+      "timestamp": "2025-12-28T10:54:27.565308",
+      "elapsed_time": 6981.850398540497,
+      "loss": 0.1303,
+      "grad_norm": 0.08781035244464874,
+      "learning_rate": 0.00011018867924528302,
+      "epoch": 0.453125
+    },
+    {
+      "step": 1326,
+      "timestamp": "2025-12-28T10:54:47.648331",
+      "elapsed_time": 7001.933421134949,
+      "loss": 0.1125,
+      "grad_norm": 0.083819180727005,
+      "learning_rate": 0.00011006289308176101,
+      "epoch": 0.45375
+    },
+    {
+      "step": 1327,
+      "timestamp": "2025-12-28T10:55:08.535086",
+      "elapsed_time": 7022.820177078247,
+      "loss": 0.0996,
+      "grad_norm": 0.06516632437705994,
+      "learning_rate": 0.000109937106918239,
+      "epoch": 0.454375
+    },
+    {
+      "step": 1328,
+      "timestamp": "2025-12-28T10:55:16.596328",
+      "elapsed_time": 7030.881418466568,
+      "loss": 0.2083,
+      "grad_norm": 0.2553313672542572,
+      "learning_rate": 0.00010981132075471699,
+      "epoch": 0.455
+    },
+    {
+      "step": 1329,
+      "timestamp": "2025-12-28T10:55:22.137094",
+      "elapsed_time": 7036.4221839904785,
+      "loss": 0.2141,
+      "grad_norm": 0.16487905383110046,
+      "learning_rate": 0.00010968553459119497,
+      "epoch": 0.455625
+    },
+    {
+      "step": 1330,
+      "timestamp": "2025-12-28T10:55:27.891844",
+      "elapsed_time": 7042.176934480667,
+      "loss": 0.1958,
+      "grad_norm": 0.2868243157863617,
+      "learning_rate": 0.00010955974842767296,
+      "epoch": 0.45625
+    },
+    {
+      "step": 1331,
+      "timestamp": "2025-12-28T10:55:40.387369",
+      "elapsed_time": 7054.672459602356,
+      "loss": 0.1071,
+      "grad_norm": 0.08470939844846725,
+      "learning_rate": 0.00010943396226415095,
+      "epoch": 0.456875
+    },
+    {
+      "step": 1332,
+      "timestamp": "2025-12-28T10:55:46.589006",
+      "elapsed_time": 7060.8740956783295,
+      "loss": 0.2562,
+      "grad_norm": 0.16096508502960205,
+      "learning_rate": 0.00010930817610062893,
+      "epoch": 0.4575
+    },
+    {
+      "step": 1333,
+      "timestamp": "2025-12-28T10:55:59.879933",
+      "elapsed_time": 7074.165023565292,
+      "loss": 0.1492,
+      "grad_norm": 0.10115651786327362,
+      "learning_rate": 0.00010918238993710692,
+      "epoch": 0.458125
+    },
+    {
+      "step": 1334,
+      "timestamp": "2025-12-28T10:56:06.938716",
+      "elapsed_time": 7081.223806381226,
+      "loss": 0.1331,
+      "grad_norm": 0.12268812209367752,
+      "learning_rate": 0.00010905660377358491,
+      "epoch": 0.45875
+    },
+    {
+      "step": 1335,
+      "timestamp": "2025-12-28T10:56:17.194994",
+      "elapsed_time": 7091.4800844192505,
+      "loss": 0.1838,
+      "grad_norm": 0.1212567612528801,
+      "learning_rate": 0.0001089308176100629,
+      "epoch": 0.459375
+    },
+    {
+      "step": 1336,
+      "timestamp": "2025-12-28T10:56:25.631434",
+      "elapsed_time": 7099.916523933411,
+      "loss": 0.1439,
+      "grad_norm": 0.11041584610939026,
+      "learning_rate": 0.00010880503144654089,
+      "epoch": 0.46
+    },
+    {
+      "step": 1337,
+      "timestamp": "2025-12-28T10:56:32.228488",
+      "elapsed_time": 7106.513578653336,
+      "loss": 0.1623,
+      "grad_norm": 0.14172060787677765,
+      "learning_rate": 0.00010867924528301887,
+      "epoch": 0.460625
+    },
+    {
+      "step": 1338,
+      "timestamp": "2025-12-28T10:56:39.404079",
+      "elapsed_time": 7113.689169168472,
+      "loss": 0.1669,
+      "grad_norm": 0.1437670886516571,
+      "learning_rate": 0.00010855345911949686,
+      "epoch": 0.46125
+    },
+    {
+      "step": 1339,
+      "timestamp": "2025-12-28T10:56:52.538722",
+      "elapsed_time": 7126.823812246323,
+      "loss": 0.122,
+      "grad_norm": 0.08426967263221741,
+      "learning_rate": 0.00010842767295597485,
+      "epoch": 0.461875
+    },
+    {
+      "step": 1340,
+      "timestamp": "2025-12-28T10:56:58.657468",
+      "elapsed_time": 7132.942558288574,
+      "loss": 0.1871,
+      "grad_norm": 0.13586315512657166,
+      "learning_rate": 0.00010830188679245283,
+      "epoch": 0.4625
+    },
+    {
+      "step": 1341,
+      "timestamp": "2025-12-28T10:57:14.320116",
+      "elapsed_time": 7148.605206251144,
+      "loss": 0.1435,
+      "grad_norm": 0.09016376733779907,
+      "learning_rate": 0.00010817610062893082,
+      "epoch": 0.463125
+    },
+    {
+      "step": 1342,
+      "timestamp": "2025-12-28T10:57:21.801068",
+      "elapsed_time": 7156.086158514023,
+      "loss": 0.235,
+      "grad_norm": 0.1631210297346115,
+      "learning_rate": 0.00010805031446540882,
+      "epoch": 0.46375
+    },
+    {
+      "step": 1343,
+      "timestamp": "2025-12-28T10:57:42.686920",
+      "elapsed_time": 7176.972010374069,
+      "loss": 0.1637,
+      "grad_norm": 0.08319737762212753,
+      "learning_rate": 0.0001079245283018868,
+      "epoch": 0.464375
+    },
+    {
+      "step": 1344,
+      "timestamp": "2025-12-28T10:57:57.552601",
+      "elapsed_time": 7191.837691068649,
+      "loss": 0.1667,
+      "grad_norm": 0.10576473921537399,
+      "learning_rate": 0.00010779874213836479,
+      "epoch": 0.465
+    },
+    {
+      "step": 1345,
+      "timestamp": "2025-12-28T10:58:07.616516",
+      "elapsed_time": 7201.901606798172,
+      "loss": 0.2561,
+      "grad_norm": 0.142886221408844,
+      "learning_rate": 0.00010767295597484277,
+      "epoch": 0.465625
+    },
+    {
+      "step": 1346,
+      "timestamp": "2025-12-28T10:58:15.468138",
+      "elapsed_time": 7209.753228902817,
+      "loss": 0.1848,
+      "grad_norm": 0.14011366665363312,
+      "learning_rate": 0.00010754716981132076,
+      "epoch": 0.46625
+    },
+    {
+      "step": 1347,
+      "timestamp": "2025-12-28T10:58:20.930984",
+      "elapsed_time": 7215.216074466705,
+      "loss": 0.2569,
+      "grad_norm": 0.16688081622123718,
+      "learning_rate": 0.00010742138364779875,
+      "epoch": 0.466875
+    },
+    {
+      "step": 1348,
+      "timestamp": "2025-12-28T10:58:29.272767",
+      "elapsed_time": 7223.557857751846,
+      "loss": 0.1619,
+      "grad_norm": 0.12044026702642441,
+      "learning_rate": 0.00010729559748427673,
+      "epoch": 0.4675
+    },
+    {
+      "step": 1349,
+      "timestamp": "2025-12-28T10:58:35.655956",
+      "elapsed_time": 7229.941046714783,
+      "loss": 0.1515,
+      "grad_norm": 0.13383439183235168,
+      "learning_rate": 0.00010716981132075472,
+      "epoch": 0.468125
+    },
+    {
+      "step": 1350,
+      "timestamp": "2025-12-28T10:58:51.652984",
+      "elapsed_time": 7245.938074588776,
+      "loss": 0.1298,
+      "grad_norm": 0.08610431104898453,
+      "learning_rate": 0.0001070440251572327,
+      "epoch": 0.46875
+    },
+    {
+      "step": 1351,
+      "timestamp": "2025-12-28T10:59:01.082143",
+      "elapsed_time": 7255.367233753204,
+      "loss": 0.1903,
+      "grad_norm": 0.11925891786813736,
+      "learning_rate": 0.0001069182389937107,
+      "epoch": 0.469375
+    },
+    {
+      "step": 1352,
+      "timestamp": "2025-12-28T10:59:07.425183",
+      "elapsed_time": 7261.710273742676,
+      "loss": 0.1339,
+      "grad_norm": 0.12832607328891754,
+      "learning_rate": 0.00010679245283018869,
+      "epoch": 0.47
+    },
+    {
+      "step": 1353,
+      "timestamp": "2025-12-28T10:59:14.908303",
+      "elapsed_time": 7269.193393468857,
+      "loss": 0.1308,
+      "grad_norm": 0.10823425650596619,
+      "learning_rate": 0.00010666666666666667,
+      "epoch": 0.470625
+    },
+    {
+      "step": 1354,
+      "timestamp": "2025-12-28T10:59:22.938044",
+      "elapsed_time": 7277.223134994507,
+      "loss": 0.256,
+      "grad_norm": 0.16312338411808014,
+      "learning_rate": 0.00010654088050314466,
+      "epoch": 0.47125
+    },
+    {
+      "step": 1355,
+      "timestamp": "2025-12-28T10:59:32.232549",
+      "elapsed_time": 7286.517638921738,
+      "loss": 0.125,
+      "grad_norm": 0.10023007541894913,
+      "learning_rate": 0.00010641509433962265,
+      "epoch": 0.471875
+    },
+    {
+      "step": 1356,
+      "timestamp": "2025-12-28T10:59:43.645190",
+      "elapsed_time": 7297.930280685425,
+      "loss": 0.3132,
+      "grad_norm": 0.14320826530456543,
+      "learning_rate": 0.00010628930817610063,
+      "epoch": 0.4725
+    },
+    {
+      "step": 1357,
+      "timestamp": "2025-12-28T10:59:49.851471",
+      "elapsed_time": 7304.136561393738,
+      "loss": 0.229,
+      "grad_norm": 0.18624381721019745,
+      "learning_rate": 0.00010616352201257862,
+      "epoch": 0.473125
+    },
+    {
+      "step": 1358,
+      "timestamp": "2025-12-28T10:59:58.793059",
+      "elapsed_time": 7313.0781490802765,
+      "loss": 0.1375,
+      "grad_norm": 0.10548731684684753,
+      "learning_rate": 0.0001060377358490566,
+      "epoch": 0.47375
+    },
+    {
+      "step": 1359,
+      "timestamp": "2025-12-28T11:00:10.967893",
+      "elapsed_time": 7325.252983093262,
+      "loss": 0.1417,
+      "grad_norm": 0.10804691165685654,
+      "learning_rate": 0.0001059119496855346,
+      "epoch": 0.474375
+    },
+    {
+      "step": 1360,
+      "timestamp": "2025-12-28T11:00:16.238424",
+      "elapsed_time": 7330.523514509201,
+      "loss": 0.2253,
+      "grad_norm": 0.1500820368528366,
+      "learning_rate": 0.00010578616352201259,
+      "epoch": 0.475
+    },
+    {
+      "step": 1361,
+      "timestamp": "2025-12-28T11:00:24.712628",
+      "elapsed_time": 7338.997718095779,
+      "loss": 0.1785,
+      "grad_norm": 0.12143810838460922,
+      "learning_rate": 0.00010566037735849057,
+      "epoch": 0.475625
+    },
+    {
+      "step": 1362,
+      "timestamp": "2025-12-28T11:00:33.573643",
+      "elapsed_time": 7347.858733415604,
+      "loss": 0.176,
+      "grad_norm": 0.13725930452346802,
+      "learning_rate": 0.00010553459119496856,
+      "epoch": 0.47625
+    },
+    {
+      "step": 1363,
+      "timestamp": "2025-12-28T11:00:41.508324",
+      "elapsed_time": 7355.7934148311615,
+      "loss": 0.1825,
+      "grad_norm": 0.13526056706905365,
+      "learning_rate": 0.00010540880503144654,
+      "epoch": 0.476875
+    },
+    {
+      "step": 1364,
+      "timestamp": "2025-12-28T11:00:47.043992",
+      "elapsed_time": 7361.329082250595,
+      "loss": 0.2259,
+      "grad_norm": 0.16327698528766632,
+      "learning_rate": 0.00010528301886792453,
+      "epoch": 0.4775
+    },
+    {
+      "step": 1365,
+      "timestamp": "2025-12-28T11:00:57.065386",
+      "elapsed_time": 7371.350476741791,
+      "loss": 0.182,
+      "grad_norm": 0.12609660625457764,
+      "learning_rate": 0.00010515723270440252,
+      "epoch": 0.478125
+    },
+    {
+      "step": 1366,
+      "timestamp": "2025-12-28T11:01:06.163415",
+      "elapsed_time": 7380.4485058784485,
+      "loss": 0.1756,
+      "grad_norm": 0.1293572634458542,
+      "learning_rate": 0.0001050314465408805,
+      "epoch": 0.47875
+    },
+    {
+      "step": 1367,
+      "timestamp": "2025-12-28T11:01:15.379986",
+      "elapsed_time": 7389.665076971054,
+      "loss": 0.1195,
+      "grad_norm": 0.0980340987443924,
+      "learning_rate": 0.0001049056603773585,
+      "epoch": 0.479375
+    },
+    {
+      "step": 1368,
+      "timestamp": "2025-12-28T11:01:23.731574",
+      "elapsed_time": 7398.016664505005,
+      "loss": 0.2087,
+      "grad_norm": 0.1566362828016281,
+      "learning_rate": 0.00010477987421383649,
+      "epoch": 0.48
+    },
+    {
+      "step": 1369,
+      "timestamp": "2025-12-28T11:01:38.713040",
+      "elapsed_time": 7412.99813079834,
+      "loss": 0.1185,
+      "grad_norm": 0.09030961245298386,
+      "learning_rate": 0.00010465408805031447,
+      "epoch": 0.480625
+    },
+    {
+      "step": 1370,
+      "timestamp": "2025-12-28T11:01:47.053864",
+      "elapsed_time": 7421.3389544487,
+      "loss": 0.1549,
+      "grad_norm": 0.10715126991271973,
+      "learning_rate": 0.00010452830188679246,
+      "epoch": 0.48125
+    },
+    {
+      "step": 1371,
+      "timestamp": "2025-12-28T11:01:52.967016",
+      "elapsed_time": 7427.252106428146,
+      "loss": 0.253,
+      "grad_norm": 0.16500554978847504,
+      "learning_rate": 0.00010440251572327044,
+      "epoch": 0.481875
+    },
+    {
+      "step": 1372,
+      "timestamp": "2025-12-28T11:02:04.814134",
+      "elapsed_time": 7439.0992250442505,
+      "loss": 0.2054,
+      "grad_norm": 0.10960862785577774,
+      "learning_rate": 0.00010427672955974843,
+      "epoch": 0.4825
+    },
+    {
+      "step": 1373,
+      "timestamp": "2025-12-28T11:02:20.468162",
+      "elapsed_time": 7454.7532522678375,
+      "loss": 0.1554,
+      "grad_norm": 0.12754300236701965,
+      "learning_rate": 0.00010415094339622642,
+      "epoch": 0.483125
+    },
+    {
+      "step": 1374,
+      "timestamp": "2025-12-28T11:02:28.644614",
+      "elapsed_time": 7462.929704427719,
+      "loss": 0.1667,
+      "grad_norm": 0.12952545285224915,
+      "learning_rate": 0.0001040251572327044,
+      "epoch": 0.48375
+    },
+    {
+      "step": 1375,
+      "timestamp": "2025-12-28T11:02:33.987063",
+      "elapsed_time": 7468.2721536159515,
+      "loss": 0.221,
+      "grad_norm": 0.1964443176984787,
+      "learning_rate": 0.0001038993710691824,
+      "epoch": 0.484375
+    },
+    {
+      "step": 1376,
+      "timestamp": "2025-12-28T11:02:45.242830",
+      "elapsed_time": 7479.527920484543,
+      "loss": 0.1549,
+      "grad_norm": 0.10678889602422714,
+      "learning_rate": 0.00010377358490566037,
+      "epoch": 0.485
+    },
+    {
+      "step": 1377,
+      "timestamp": "2025-12-28T11:02:57.881441",
+      "elapsed_time": 7492.166531324387,
+      "loss": 0.1292,
+      "grad_norm": 0.1204121932387352,
+      "learning_rate": 0.00010364779874213837,
+      "epoch": 0.485625
+    },
+    {
+      "step": 1378,
+      "timestamp": "2025-12-28T11:03:05.127176",
+      "elapsed_time": 7499.412266731262,
+      "loss": 0.2362,
+      "grad_norm": 0.14369583129882812,
+      "learning_rate": 0.00010352201257861636,
+      "epoch": 0.48625
+    },
+    {
+      "step": 1379,
+      "timestamp": "2025-12-28T11:03:08.953753",
+      "elapsed_time": 7503.238843917847,
+      "loss": 0.3803,
+      "grad_norm": 0.21724280714988708,
+      "learning_rate": 0.00010339622641509434,
+      "epoch": 0.486875
+    },
+    {
+      "step": 1380,
+      "timestamp": "2025-12-28T11:03:18.974982",
+      "elapsed_time": 7513.26007270813,
+      "loss": 0.1546,
+      "grad_norm": 0.1561720371246338,
+      "learning_rate": 0.00010327044025157233,
+      "epoch": 0.4875
+    },
+    {
+      "step": 1381,
+      "timestamp": "2025-12-28T11:03:23.926806",
+      "elapsed_time": 7518.211896419525,
+      "loss": 0.2732,
+      "grad_norm": 0.17296800017356873,
+      "learning_rate": 0.00010314465408805032,
+      "epoch": 0.488125
+    },
+    {
+      "step": 1382,
+      "timestamp": "2025-12-28T11:03:42.230374",
+      "elapsed_time": 7536.5154638290405,
+      "loss": 0.122,
+      "grad_norm": 0.08095800131559372,
+      "learning_rate": 0.0001030188679245283,
+      "epoch": 0.48875
+    },
+    {
+      "step": 1383,
+      "timestamp": "2025-12-28T11:03:46.737340",
+      "elapsed_time": 7541.022430181503,
+      "loss": 0.2151,
+      "grad_norm": 0.17441503703594208,
+      "learning_rate": 0.0001028930817610063,
+      "epoch": 0.489375
+    },
+    {
+      "step": 1384,
+      "timestamp": "2025-12-28T11:03:58.400597",
+      "elapsed_time": 7552.685687303543,
+      "loss": 0.1857,
+      "grad_norm": 0.12047084420919418,
+      "learning_rate": 0.00010276729559748428,
+      "epoch": 0.49
+    },
+    {
+      "step": 1385,
+      "timestamp": "2025-12-28T11:04:04.778751",
+      "elapsed_time": 7559.063841819763,
+      "loss": 0.2076,
+      "grad_norm": 0.1551089733839035,
+      "learning_rate": 0.00010264150943396227,
+      "epoch": 0.490625
+    },
+    {
+      "step": 1386,
+      "timestamp": "2025-12-28T11:04:13.091378",
+      "elapsed_time": 7567.37646818161,
+      "loss": 0.1264,
+      "grad_norm": 0.10384256392717361,
+      "learning_rate": 0.00010251572327044026,
+      "epoch": 0.49125
+    },
+    {
+      "step": 1387,
+      "timestamp": "2025-12-28T11:04:21.309367",
+      "elapsed_time": 7575.594457626343,
+      "loss": 0.3628,
+      "grad_norm": 0.22657428681850433,
+      "learning_rate": 0.00010238993710691824,
+      "epoch": 0.491875
+    },
+    {
+      "step": 1388,
+      "timestamp": "2025-12-28T11:04:29.487914",
+      "elapsed_time": 7583.773004293442,
+      "loss": 0.2039,
+      "grad_norm": 0.13718171417713165,
+      "learning_rate": 0.00010226415094339623,
+      "epoch": 0.4925
+    },
+    {
+      "step": 1389,
+      "timestamp": "2025-12-28T11:04:42.300103",
+      "elapsed_time": 7596.585193157196,
+      "loss": 0.1706,
+      "grad_norm": 0.10597704350948334,
+      "learning_rate": 0.00010213836477987422,
+      "epoch": 0.493125
+    },
+    {
+      "step": 1390,
+      "timestamp": "2025-12-28T11:05:03.024157",
+      "elapsed_time": 7617.309247255325,
+      "loss": 0.1205,
+      "grad_norm": 0.07673165202140808,
+      "learning_rate": 0.0001020125786163522,
+      "epoch": 0.49375
+    },
+    {
+      "step": 1391,
+      "timestamp": "2025-12-28T11:05:10.494896",
+      "elapsed_time": 7624.779986381531,
+      "loss": 0.1777,
+      "grad_norm": 0.12185313552618027,
+      "learning_rate": 0.0001018867924528302,
+      "epoch": 0.494375
+    },
+    {
+      "step": 1392,
+      "timestamp": "2025-12-28T11:05:16.240771",
+      "elapsed_time": 7630.525861024857,
+      "loss": 0.1678,
+      "grad_norm": 0.1360418051481247,
+      "learning_rate": 0.00010176100628930818,
+      "epoch": 0.495
+    },
+    {
+      "step": 1393,
+      "timestamp": "2025-12-28T11:05:25.834105",
+      "elapsed_time": 7640.119195222855,
+      "loss": 0.3532,
+      "grad_norm": 0.14602701365947723,
+      "learning_rate": 0.00010163522012578617,
+      "epoch": 0.495625
+    },
+    {
+      "step": 1394,
+      "timestamp": "2025-12-28T11:05:41.365149",
+      "elapsed_time": 7655.650239467621,
+      "loss": 0.1404,
+      "grad_norm": 0.08702640235424042,
+      "learning_rate": 0.00010150943396226416,
+      "epoch": 0.49625
+    },
+    {
+      "step": 1395,
+      "timestamp": "2025-12-28T11:05:48.620666",
+      "elapsed_time": 7662.905756235123,
+      "loss": 0.222,
+      "grad_norm": 0.15947473049163818,
+      "learning_rate": 0.00010138364779874214,
+      "epoch": 0.496875
+    },
+    {
+      "step": 1396,
+      "timestamp": "2025-12-28T11:05:58.248188",
+      "elapsed_time": 7672.533278942108,
+      "loss": 0.3532,
+      "grad_norm": 0.15767693519592285,
+      "learning_rate": 0.00010125786163522013,
+      "epoch": 0.4975
+    },
+    {
+      "step": 1397,
+      "timestamp": "2025-12-28T11:06:04.588958",
+      "elapsed_time": 7678.874048471451,
+      "loss": 0.2166,
+      "grad_norm": 0.14681337773799896,
+      "learning_rate": 0.00010113207547169811,
+      "epoch": 0.498125
+    },
+    {
+      "step": 1398,
+      "timestamp": "2025-12-28T11:06:08.944856",
+      "elapsed_time": 7683.22994685173,
+      "loss": 0.2352,
+      "grad_norm": 0.19252383708953857,
+      "learning_rate": 0.0001010062893081761,
+      "epoch": 0.49875
+    },
+    {
+      "step": 1399,
+      "timestamp": "2025-12-28T11:06:18.169645",
+      "elapsed_time": 7692.454735279083,
+      "loss": 0.2846,
+      "grad_norm": 0.14431574940681458,
+      "learning_rate": 0.0001008805031446541,
+      "epoch": 0.499375
+    },
+    {
+      "step": 1400,
+      "timestamp": "2025-12-28T11:06:23.381124",
+      "elapsed_time": 7697.666215181351,
+      "loss": 0.2615,
+      "grad_norm": 0.18521860241889954,
+      "learning_rate": 0.00010075471698113208,
+      "epoch": 0.5
+    },
+    {
+      "step": 1401,
+      "timestamp": "2025-12-28T11:06:31.966311",
+      "elapsed_time": 7706.2514016628265,
+      "loss": 0.2161,
+      "grad_norm": 0.12801428139209747,
+      "learning_rate": 0.00010062893081761007,
+      "epoch": 0.500625
+    },
+    {
+      "step": 1402,
+      "timestamp": "2025-12-28T11:06:36.665411",
+      "elapsed_time": 7710.950501441956,
+      "loss": 0.2801,
+      "grad_norm": 0.18982771039009094,
+      "learning_rate": 0.00010050314465408806,
+      "epoch": 0.50125
+    },
+    {
+      "step": 1403,
+      "timestamp": "2025-12-28T11:06:40.414131",
+      "elapsed_time": 7714.699221611023,
+      "loss": 0.2644,
+      "grad_norm": 0.2565673589706421,
+      "learning_rate": 0.00010037735849056604,
+      "epoch": 0.501875
+    },
+    {
+      "step": 1404,
+      "timestamp": "2025-12-28T11:06:51.418661",
+      "elapsed_time": 7725.70375084877,
+      "loss": 0.1786,
+      "grad_norm": 0.10981456935405731,
+      "learning_rate": 0.00010025157232704403,
+      "epoch": 0.5025
+    },
+    {
+      "step": 1405,
+      "timestamp": "2025-12-28T11:06:57.172405",
+      "elapsed_time": 7731.4574954509735,
+      "loss": 0.4434,
+      "grad_norm": 0.20104235410690308,
+      "learning_rate": 0.00010012578616352201,
+      "epoch": 0.503125
+    },
+    {
+      "step": 1406,
+      "timestamp": "2025-12-28T11:07:05.960076",
+      "elapsed_time": 7740.245167016983,
+      "loss": 0.1505,
+      "grad_norm": 0.11026592552661896,
+      "learning_rate": 0.0001,
+      "epoch": 0.50375
+    },
+    {
+      "step": 1407,
+      "timestamp": "2025-12-28T11:07:17.486947",
+      "elapsed_time": 7751.7720375061035,
+      "loss": 0.1486,
+      "grad_norm": 0.09958239644765854,
+      "learning_rate": 9.9874213836478e-05,
+      "epoch": 0.504375
+    },
+    {
+      "step": 1408,
+      "timestamp": "2025-12-28T11:07:26.472895",
+      "elapsed_time": 7760.757986068726,
+      "loss": 0.1752,
+      "grad_norm": 0.11857740581035614,
+      "learning_rate": 9.974842767295598e-05,
+      "epoch": 0.505
+    },
+    {
+      "step": 1409,
+      "timestamp": "2025-12-28T11:07:33.007795",
+      "elapsed_time": 7767.292885303497,
+      "loss": 0.2939,
+      "grad_norm": 0.15093620121479034,
+      "learning_rate": 9.962264150943397e-05,
+      "epoch": 0.505625
+    },
+    {
+      "step": 1410,
+      "timestamp": "2025-12-28T11:07:41.150713",
+      "elapsed_time": 7775.43580365181,
+      "loss": 0.1838,
+      "grad_norm": 0.1252412497997284,
+      "learning_rate": 9.949685534591195e-05,
+      "epoch": 0.50625
+    },
+    {
+      "step": 1411,
+      "timestamp": "2025-12-28T11:07:53.703346",
+      "elapsed_time": 7787.988436460495,
+      "loss": 0.1411,
+      "grad_norm": 0.0989488959312439,
+      "learning_rate": 9.937106918238994e-05,
+      "epoch": 0.506875
+    },
+    {
+      "step": 1412,
+      "timestamp": "2025-12-28T11:08:02.343988",
+      "elapsed_time": 7796.629077911377,
+      "loss": 0.1611,
+      "grad_norm": 0.11869240552186966,
+      "learning_rate": 9.924528301886793e-05,
+      "epoch": 0.5075
+    },
+    {
+      "step": 1413,
+      "timestamp": "2025-12-28T11:08:12.956190",
+      "elapsed_time": 7807.2412803173065,
+      "loss": 0.1394,
+      "grad_norm": 0.10440147668123245,
+      "learning_rate": 9.911949685534591e-05,
+      "epoch": 0.508125
+    },
+    {
+      "step": 1414,
+      "timestamp": "2025-12-28T11:08:18.967906",
+      "elapsed_time": 7813.252996206284,
+      "loss": 0.2214,
+      "grad_norm": 0.16440127789974213,
+      "learning_rate": 9.89937106918239e-05,
+      "epoch": 0.50875
+    },
+    {
+      "step": 1415,
+      "timestamp": "2025-12-28T11:08:27.441443",
+      "elapsed_time": 7821.726533174515,
+      "loss": 0.1799,
+      "grad_norm": 0.16303229331970215,
+      "learning_rate": 9.88679245283019e-05,
+      "epoch": 0.509375
+    },
+    {
+      "step": 1416,
+      "timestamp": "2025-12-28T11:08:38.770337",
+      "elapsed_time": 7833.055427074432,
+      "loss": 0.1423,
+      "grad_norm": 0.12514905631542206,
+      "learning_rate": 9.874213836477988e-05,
+      "epoch": 0.51
+    },
+    {
+      "step": 1417,
+      "timestamp": "2025-12-28T11:08:49.045039",
+      "elapsed_time": 7843.330129623413,
+      "loss": 0.128,
+      "grad_norm": 0.41818127036094666,
+      "learning_rate": 9.861635220125787e-05,
+      "epoch": 0.510625
+    },
+    {
+      "step": 1418,
+      "timestamp": "2025-12-28T11:08:56.563729",
+      "elapsed_time": 7850.848820209503,
+      "loss": 0.384,
+      "grad_norm": 0.1607155203819275,
+      "learning_rate": 9.849056603773585e-05,
+      "epoch": 0.51125
+    },
+    {
+      "step": 1419,
+      "timestamp": "2025-12-28T11:09:04.998828",
+      "elapsed_time": 7859.2839179039,
+      "loss": 0.1517,
+      "grad_norm": 0.11149600148200989,
+      "learning_rate": 9.836477987421384e-05,
+      "epoch": 0.511875
+    },
+    {
+      "step": 1420,
+      "timestamp": "2025-12-28T11:09:15.250027",
+      "elapsed_time": 7869.53511762619,
+      "loss": 0.16,
+      "grad_norm": 0.10607406497001648,
+      "learning_rate": 9.823899371069183e-05,
+      "epoch": 0.5125
+    },
+    {
+      "step": 1421,
+      "timestamp": "2025-12-28T11:09:24.383519",
+      "elapsed_time": 7878.668609380722,
+      "loss": 0.206,
+      "grad_norm": 0.12708930671215057,
+      "learning_rate": 9.811320754716981e-05,
+      "epoch": 0.513125
+    },
+    {
+      "step": 1422,
+      "timestamp": "2025-12-28T11:09:39.780418",
+      "elapsed_time": 7894.06550860405,
+      "loss": 0.3202,
+      "grad_norm": 0.12644296884536743,
+      "learning_rate": 9.79874213836478e-05,
+      "epoch": 0.51375
+    },
+    {
+      "step": 1423,
+      "timestamp": "2025-12-28T11:09:52.468194",
+      "elapsed_time": 7906.753284931183,
+      "loss": 0.1625,
+      "grad_norm": 0.1026865616440773,
+      "learning_rate": 9.786163522012578e-05,
+      "epoch": 0.514375
+    },
+    {
+      "step": 1424,
+      "timestamp": "2025-12-28T11:10:07.620418",
+      "elapsed_time": 7921.905508518219,
+      "loss": 0.1569,
+      "grad_norm": 0.08125265687704086,
+      "learning_rate": 9.773584905660378e-05,
+      "epoch": 0.515
+    },
+    {
+      "step": 1425,
+      "timestamp": "2025-12-28T11:10:19.799493",
+      "elapsed_time": 7934.084583282471,
+      "loss": 0.1262,
+      "grad_norm": 0.09111649543046951,
+      "learning_rate": 9.761006289308177e-05,
+      "epoch": 0.515625
+    },
+    {
+      "step": 1426,
+      "timestamp": "2025-12-28T11:10:25.818900",
+      "elapsed_time": 7940.1039955616,
+      "loss": 0.1587,
+      "grad_norm": 0.1336372345685959,
+      "learning_rate": 9.748427672955975e-05,
+      "epoch": 0.51625
+    },
+    {
+      "step": 1427,
+      "timestamp": "2025-12-28T11:10:35.455757",
+      "elapsed_time": 7949.740847349167,
+      "loss": 0.1667,
+      "grad_norm": 0.11319567263126373,
+      "learning_rate": 9.735849056603774e-05,
+      "epoch": 0.516875
+    },
+    {
+      "step": 1428,
+      "timestamp": "2025-12-28T11:10:40.995568",
+      "elapsed_time": 7955.280659198761,
+      "loss": 0.1724,
+      "grad_norm": 0.13806995749473572,
+      "learning_rate": 9.723270440251573e-05,
+      "epoch": 0.5175
+    },
+    {
+      "step": 1429,
+      "timestamp": "2025-12-28T11:10:51.368410",
+      "elapsed_time": 7965.653500318527,
+      "loss": 0.2761,
+      "grad_norm": 0.12941741943359375,
+      "learning_rate": 9.710691823899371e-05,
+      "epoch": 0.518125
+    },
+    {
+      "step": 1430,
+      "timestamp": "2025-12-28T11:10:57.451776",
+      "elapsed_time": 7971.736865758896,
+      "loss": 0.1955,
+      "grad_norm": 0.13250701129436493,
+      "learning_rate": 9.69811320754717e-05,
+      "epoch": 0.51875
+    },
+    {
+      "step": 1431,
+      "timestamp": "2025-12-28T11:11:06.229749",
+      "elapsed_time": 7980.514839172363,
+      "loss": 0.1667,
+      "grad_norm": 0.11635927855968475,
+      "learning_rate": 9.685534591194969e-05,
+      "epoch": 0.519375
+    },
+    {
+      "step": 1432,
+      "timestamp": "2025-12-28T11:11:15.100261",
+      "elapsed_time": 7989.385351657867,
+      "loss": 0.1818,
+      "grad_norm": 0.1113656610250473,
+      "learning_rate": 9.672955974842768e-05,
+      "epoch": 0.52
+    },
+    {
+      "step": 1433,
+      "timestamp": "2025-12-28T11:11:35.992872",
+      "elapsed_time": 8010.277962684631,
+      "loss": 0.1138,
+      "grad_norm": 0.06887345761060715,
+      "learning_rate": 9.660377358490567e-05,
+      "epoch": 0.520625
+    },
+    {
+      "step": 1434,
+      "timestamp": "2025-12-28T11:11:46.465445",
+      "elapsed_time": 8020.75053524971,
+      "loss": 0.1542,
+      "grad_norm": 0.1157265156507492,
+      "learning_rate": 9.647798742138365e-05,
+      "epoch": 0.52125
+    },
+    {
+      "step": 1435,
+      "timestamp": "2025-12-28T11:11:56.976986",
+      "elapsed_time": 8031.26207613945,
+      "loss": 0.1418,
+      "grad_norm": 0.10130833089351654,
+      "learning_rate": 9.635220125786164e-05,
+      "epoch": 0.521875
+    },
+    {
+      "step": 1436,
+      "timestamp": "2025-12-28T11:12:01.707156",
+      "elapsed_time": 8035.9922461509705,
+      "loss": 0.3286,
+      "grad_norm": 0.18273454904556274,
+      "learning_rate": 9.622641509433963e-05,
+      "epoch": 0.5225
+    },
+    {
+      "step": 1437,
+      "timestamp": "2025-12-28T11:12:07.714931",
+      "elapsed_time": 8042.000021696091,
+      "loss": 0.4137,
+      "grad_norm": 0.19872736930847168,
+      "learning_rate": 9.610062893081761e-05,
+      "epoch": 0.523125
+    },
+    {
+      "step": 1438,
+      "timestamp": "2025-12-28T11:12:19.337217",
+      "elapsed_time": 8053.622307062149,
+      "loss": 0.1493,
+      "grad_norm": 0.1055750697851181,
+      "learning_rate": 9.59748427672956e-05,
+      "epoch": 0.52375
+    },
+    {
+      "step": 1439,
+      "timestamp": "2025-12-28T11:12:26.407821",
+      "elapsed_time": 8060.692911624908,
+      "loss": 0.1795,
+      "grad_norm": 0.25722602009773254,
+      "learning_rate": 9.584905660377359e-05,
+      "epoch": 0.524375
+    },
+    {
+      "step": 1440,
+      "timestamp": "2025-12-28T11:12:38.538641",
+      "elapsed_time": 8072.823730945587,
+      "loss": 0.1517,
+      "grad_norm": 0.10837842524051666,
+      "learning_rate": 9.572327044025158e-05,
+      "epoch": 0.525
+    },
+    {
+      "step": 1441,
+      "timestamp": "2025-12-28T11:12:53.364199",
+      "elapsed_time": 8087.649289131165,
+      "loss": 0.0904,
+      "grad_norm": 0.10741348564624786,
+      "learning_rate": 9.559748427672957e-05,
+      "epoch": 0.525625
+    },
+    {
+      "step": 1442,
+      "timestamp": "2025-12-28T11:13:14.254540",
+      "elapsed_time": 8108.539630651474,
+      "loss": 0.0746,
+      "grad_norm": 0.07680816948413849,
+      "learning_rate": 9.547169811320755e-05,
+      "epoch": 0.52625
+    },
+    {
+      "step": 1443,
+      "timestamp": "2025-12-28T11:13:29.340440",
+      "elapsed_time": 8123.625530004501,
+      "loss": 0.1088,
+      "grad_norm": 0.07666955888271332,
+      "learning_rate": 9.534591194968554e-05,
+      "epoch": 0.526875
+    },
+    {
+      "step": 1444,
+      "timestamp": "2025-12-28T11:13:39.098916",
+      "elapsed_time": 8133.384006500244,
+      "loss": 0.2508,
+      "grad_norm": 0.1188189759850502,
+      "learning_rate": 9.522012578616352e-05,
+      "epoch": 0.5275
+    },
+    {
+      "step": 1445,
+      "timestamp": "2025-12-28T11:13:44.861179",
+      "elapsed_time": 8139.146269083023,
+      "loss": 0.2474,
+      "grad_norm": 0.15990658104419708,
+      "learning_rate": 9.509433962264151e-05,
+      "epoch": 0.528125
+    },
+    {
+      "step": 1446,
+      "timestamp": "2025-12-28T11:13:52.613787",
+      "elapsed_time": 8146.898877620697,
+      "loss": 0.32,
+      "grad_norm": 0.15008442103862762,
+      "learning_rate": 9.496855345911951e-05,
+      "epoch": 0.52875
+    },
+    {
+      "step": 1447,
+      "timestamp": "2025-12-28T11:14:05.480355",
+      "elapsed_time": 8159.765445947647,
+      "loss": 0.1453,
+      "grad_norm": 0.0898258164525032,
+      "learning_rate": 9.484276729559749e-05,
+      "epoch": 0.529375
+    },
+    {
+      "step": 1448,
+      "timestamp": "2025-12-28T11:14:15.110555",
+      "elapsed_time": 8169.395644903183,
+      "loss": 0.187,
+      "grad_norm": 0.1340561956167221,
+      "learning_rate": 9.471698113207548e-05,
+      "epoch": 0.53
+    },
+    {
+      "step": 1449,
+      "timestamp": "2025-12-28T11:14:22.202495",
+      "elapsed_time": 8176.487585544586,
+      "loss": 0.176,
+      "grad_norm": 0.16083924472332,
+      "learning_rate": 9.459119496855347e-05,
+      "epoch": 0.530625
+    },
+    {
+      "step": 1450,
+      "timestamp": "2025-12-28T11:14:28.754429",
+      "elapsed_time": 8183.039519309998,
+      "loss": 0.2644,
+      "grad_norm": 0.15739032626152039,
+      "learning_rate": 9.446540880503145e-05,
+      "epoch": 0.53125
+    },
+    {
+      "step": 1451,
+      "timestamp": "2025-12-28T11:14:38.517478",
+      "elapsed_time": 8192.802568435669,
+      "loss": 0.1586,
+      "grad_norm": 0.11289030313491821,
+      "learning_rate": 9.433962264150944e-05,
+      "epoch": 0.531875
+    },
+    {
+      "step": 1452,
+      "timestamp": "2025-12-28T11:14:44.085834",
+      "elapsed_time": 8198.370924711227,
+      "loss": 0.1873,
+      "grad_norm": 0.1545426845550537,
+      "learning_rate": 9.421383647798742e-05,
+      "epoch": 0.5325
+    },
+    {
+      "step": 1453,
+      "timestamp": "2025-12-28T11:14:49.388857",
+      "elapsed_time": 8203.673947095871,
+      "loss": 0.4183,
+      "grad_norm": 0.2863948941230774,
+      "learning_rate": 9.408805031446541e-05,
+      "epoch": 0.533125
+    },
+    {
+      "step": 1454,
+      "timestamp": "2025-12-28T11:14:59.422428",
+      "elapsed_time": 8213.707518577576,
+      "loss": 0.1046,
+      "grad_norm": 0.0934741273522377,
+      "learning_rate": 9.396226415094341e-05,
+      "epoch": 0.53375
+    },
+    {
+      "step": 1455,
+      "timestamp": "2025-12-28T11:15:05.324002",
+      "elapsed_time": 8219.6090965271,
+      "loss": 0.2274,
+      "grad_norm": 0.150177463889122,
+      "learning_rate": 9.383647798742139e-05,
+      "epoch": 0.534375
+    },
+    {
+      "step": 1456,
+      "timestamp": "2025-12-28T11:15:09.916095",
+      "elapsed_time": 8224.201185464859,
+      "loss": 0.3107,
+      "grad_norm": 0.19902926683425903,
+      "learning_rate": 9.371069182389938e-05,
+      "epoch": 0.535
+    },
+    {
+      "step": 1457,
+      "timestamp": "2025-12-28T11:15:17.157963",
+      "elapsed_time": 8231.443053722382,
+      "loss": 0.2108,
+      "grad_norm": 0.14318348467350006,
+      "learning_rate": 9.358490566037736e-05,
+      "epoch": 0.535625
+    },
+    {
+      "step": 1458,
+      "timestamp": "2025-12-28T11:15:23.456852",
+      "elapsed_time": 8237.741942882538,
+      "loss": 0.1734,
+      "grad_norm": 0.1470271497964859,
+      "learning_rate": 9.345911949685535e-05,
+      "epoch": 0.53625
+    },
+    {
+      "step": 1459,
+      "timestamp": "2025-12-28T11:15:34.335466",
+      "elapsed_time": 8248.620555877686,
+      "loss": 0.1041,
+      "grad_norm": 0.08832447230815887,
+      "learning_rate": 9.333333333333334e-05,
+      "epoch": 0.536875
+    },
+    {
+      "step": 1460,
+      "timestamp": "2025-12-28T11:15:42.518003",
+      "elapsed_time": 8256.803106307983,
+      "loss": 0.333,
+      "grad_norm": 0.17225605249404907,
+      "learning_rate": 9.320754716981132e-05,
+      "epoch": 0.5375
+    },
+    {
+      "step": 1461,
+      "timestamp": "2025-12-28T11:15:52.109549",
+      "elapsed_time": 8266.394639015198,
+      "loss": 0.1827,
+      "grad_norm": 0.11171706020832062,
+      "learning_rate": 9.308176100628931e-05,
+      "epoch": 0.538125
+    },
+    {
+      "step": 1462,
+      "timestamp": "2025-12-28T11:15:59.068953",
+      "elapsed_time": 8273.354043006897,
+      "loss": 0.1926,
+      "grad_norm": 0.13213799893856049,
+      "learning_rate": 9.295597484276731e-05,
+      "epoch": 0.53875
+    },
+    {
+      "step": 1463,
+      "timestamp": "2025-12-28T11:16:12.804617",
+      "elapsed_time": 8287.089706897736,
+      "loss": 0.1089,
+      "grad_norm": 0.08467783033847809,
+      "learning_rate": 9.283018867924529e-05,
+      "epoch": 0.539375
+    },
+    {
+      "step": 1464,
+      "timestamp": "2025-12-28T11:16:18.706671",
+      "elapsed_time": 8292.991761922836,
+      "loss": 0.4806,
+      "grad_norm": 0.20453688502311707,
+      "learning_rate": 9.270440251572328e-05,
+      "epoch": 0.54
+    },
+    {
+      "step": 1465,
+      "timestamp": "2025-12-28T11:16:28.734388",
+      "elapsed_time": 8303.01947760582,
+      "loss": 0.1556,
+      "grad_norm": 0.11547412723302841,
+      "learning_rate": 9.257861635220126e-05,
+      "epoch": 0.540625
+    },
+    {
+      "step": 1466,
+      "timestamp": "2025-12-28T11:16:38.462518",
+      "elapsed_time": 8312.74760890007,
+      "loss": 0.1464,
+      "grad_norm": 0.10674792528152466,
+      "learning_rate": 9.245283018867925e-05,
+      "epoch": 0.54125
+    },
+    {
+      "step": 1467,
+      "timestamp": "2025-12-28T11:16:49.934266",
+      "elapsed_time": 8324.219356536865,
+      "loss": 0.1205,
+      "grad_norm": 0.10482378304004669,
+      "learning_rate": 9.232704402515724e-05,
+      "epoch": 0.541875
+    },
+    {
+      "step": 1468,
+      "timestamp": "2025-12-28T11:16:59.067362",
+      "elapsed_time": 8333.3524518013,
+      "loss": 0.1575,
+      "grad_norm": 0.18603090941905975,
+      "learning_rate": 9.220125786163522e-05,
+      "epoch": 0.5425
+    },
+    {
+      "step": 1469,
+      "timestamp": "2025-12-28T11:17:06.628602",
+      "elapsed_time": 8340.913692951202,
+      "loss": 0.2441,
+      "grad_norm": 0.16684553027153015,
+      "learning_rate": 9.207547169811322e-05,
+      "epoch": 0.543125
+    },
+    {
+      "step": 1470,
+      "timestamp": "2025-12-28T11:17:16.426551",
+      "elapsed_time": 8350.711641550064,
+      "loss": 0.3306,
+      "grad_norm": 0.14771923422813416,
+      "learning_rate": 9.19496855345912e-05,
+      "epoch": 0.54375
+    },
+    {
+      "step": 1471,
+      "timestamp": "2025-12-28T11:17:29.569642",
+      "elapsed_time": 8363.85473227501,
+      "loss": 0.1172,
+      "grad_norm": 0.08360818773508072,
+      "learning_rate": 9.182389937106919e-05,
+      "epoch": 0.544375
+    },
+    {
+      "step": 1472,
+      "timestamp": "2025-12-28T11:17:35.910254",
+      "elapsed_time": 8370.195344924927,
+      "loss": 0.2702,
+      "grad_norm": 0.1648559272289276,
+      "learning_rate": 9.169811320754718e-05,
+      "epoch": 0.545
+    },
+    {
+      "step": 1473,
+      "timestamp": "2025-12-28T11:17:43.053273",
+      "elapsed_time": 8377.338364124298,
+      "loss": 0.207,
+      "grad_norm": 0.1394606977701187,
+      "learning_rate": 9.157232704402516e-05,
+      "epoch": 0.545625
+    },
+    {
+      "step": 1474,
+      "timestamp": "2025-12-28T11:17:49.068090",
+      "elapsed_time": 8383.353180408478,
+      "loss": 0.148,
+      "grad_norm": 0.1300075501203537,
+      "learning_rate": 9.144654088050315e-05,
+      "epoch": 0.54625
+    },
+    {
+      "step": 1475,
+      "timestamp": "2025-12-28T11:17:56.519082",
+      "elapsed_time": 8390.804172039032,
+      "loss": 0.1676,
+      "grad_norm": 0.12073966860771179,
+      "learning_rate": 9.132075471698114e-05,
+      "epoch": 0.546875
+    },
+    {
+      "step": 1476,
+      "timestamp": "2025-12-28T11:18:08.644421",
+      "elapsed_time": 8402.92951130867,
+      "loss": 0.1639,
+      "grad_norm": 0.10669399052858353,
+      "learning_rate": 9.119496855345912e-05,
+      "epoch": 0.5475
+    },
+    {
+      "step": 1477,
+      "timestamp": "2025-12-28T11:18:17.423433",
+      "elapsed_time": 8411.708523750305,
+      "loss": 0.1742,
+      "grad_norm": 0.12219968438148499,
+      "learning_rate": 9.106918238993712e-05,
+      "epoch": 0.548125
+    },
+    {
+      "step": 1478,
+      "timestamp": "2025-12-28T11:18:36.177009",
+      "elapsed_time": 8430.462099313736,
+      "loss": 0.1255,
+      "grad_norm": 0.08146153390407562,
+      "learning_rate": 9.09433962264151e-05,
+      "epoch": 0.54875
+    },
+    {
+      "step": 1479,
+      "timestamp": "2025-12-28T11:18:46.115780",
+      "elapsed_time": 8440.400870800018,
+      "loss": 0.1592,
+      "grad_norm": 0.1143009215593338,
+      "learning_rate": 9.081761006289309e-05,
+      "epoch": 0.549375
+    },
+    {
+      "step": 1480,
+      "timestamp": "2025-12-28T11:18:51.646790",
+      "elapsed_time": 8445.931880950928,
+      "loss": 0.1368,
+      "grad_norm": 0.1322580873966217,
+      "learning_rate": 9.069182389937108e-05,
+      "epoch": 0.55
+    },
+    {
+      "step": 1481,
+      "timestamp": "2025-12-28T11:19:03.435806",
+      "elapsed_time": 8457.72089600563,
+      "loss": 0.184,
+      "grad_norm": 0.11221913248300552,
+      "learning_rate": 9.056603773584906e-05,
+      "epoch": 0.550625
+    },
+    {
+      "step": 1482,
+      "timestamp": "2025-12-28T11:19:11.463661",
+      "elapsed_time": 8465.74875164032,
+      "loss": 0.2432,
+      "grad_norm": 0.13333527743816376,
+      "learning_rate": 9.044025157232705e-05,
+      "epoch": 0.55125
+    },
+    {
+      "step": 1483,
+      "timestamp": "2025-12-28T11:19:20.808185",
+      "elapsed_time": 8475.093275308609,
+      "loss": 0.1595,
+      "grad_norm": 0.11253681033849716,
+      "learning_rate": 9.031446540880504e-05,
+      "epoch": 0.551875
+    },
+    {
+      "step": 1484,
+      "timestamp": "2025-12-28T11:19:26.372578",
+      "elapsed_time": 8480.657668352127,
+      "loss": 0.2285,
+      "grad_norm": 0.163836270570755,
+      "learning_rate": 9.018867924528302e-05,
+      "epoch": 0.5525
+    },
+    {
+      "step": 1485,
+      "timestamp": "2025-12-28T11:19:37.048053",
+      "elapsed_time": 8491.333143234253,
+      "loss": 0.1388,
+      "grad_norm": 0.1059766411781311,
+      "learning_rate": 9.006289308176102e-05,
+      "epoch": 0.553125
+    },
+    {
+      "step": 1486,
+      "timestamp": "2025-12-28T11:19:46.279936",
+      "elapsed_time": 8500.565026283264,
+      "loss": 0.1649,
+      "grad_norm": 0.11681363731622696,
+      "learning_rate": 8.9937106918239e-05,
+      "epoch": 0.55375
+    },
+    {
+      "step": 1487,
+      "timestamp": "2025-12-28T11:19:53.538665",
+      "elapsed_time": 8507.823755979538,
+      "loss": 0.2236,
+      "grad_norm": 0.1437806636095047,
+      "learning_rate": 8.981132075471699e-05,
+      "epoch": 0.554375
+    },
+    {
+      "step": 1488,
+      "timestamp": "2025-12-28T11:20:00.062395",
+      "elapsed_time": 8514.347485303879,
+      "loss": 0.1624,
+      "grad_norm": 0.13652074337005615,
+      "learning_rate": 8.968553459119498e-05,
+      "epoch": 0.555
+    },
+    {
+      "step": 1489,
+      "timestamp": "2025-12-28T11:20:12.801806",
+      "elapsed_time": 8527.08689570427,
+      "loss": 0.1095,
+      "grad_norm": 0.08968634903430939,
+      "learning_rate": 8.955974842767296e-05,
+      "epoch": 0.555625
+    },
+    {
+      "step": 1490,
+      "timestamp": "2025-12-28T11:20:19.013698",
+      "elapsed_time": 8533.298788785934,
+      "loss": 0.1724,
+      "grad_norm": 0.16458983719348907,
+      "learning_rate": 8.943396226415095e-05,
+      "epoch": 0.55625
+    },
+    {
+      "step": 1491,
+      "timestamp": "2025-12-28T11:20:30.809395",
+      "elapsed_time": 8545.094485759735,
+      "loss": 0.1667,
+      "grad_norm": 0.10606341063976288,
+      "learning_rate": 8.930817610062893e-05,
+      "epoch": 0.556875
+    },
+    {
+      "step": 1492,
+      "timestamp": "2025-12-28T11:20:44.762591",
+      "elapsed_time": 8559.047681331635,
+      "loss": 0.1019,
+      "grad_norm": 0.07538938522338867,
+      "learning_rate": 8.918238993710692e-05,
+      "epoch": 0.5575
+    },
+    {
+      "step": 1493,
+      "timestamp": "2025-12-28T11:21:01.644690",
+      "elapsed_time": 8575.92978143692,
+      "loss": 0.1099,
+      "grad_norm": 0.08097642660140991,
+      "learning_rate": 8.905660377358492e-05,
+      "epoch": 0.558125
+    },
+    {
+      "step": 1494,
+      "timestamp": "2025-12-28T11:21:10.267153",
+      "elapsed_time": 8584.55224275589,
+      "loss": 0.1285,
+      "grad_norm": 0.11686515063047409,
+      "learning_rate": 8.89308176100629e-05,
+      "epoch": 0.55875
+    },
+    {
+      "step": 1495,
+      "timestamp": "2025-12-28T11:21:16.174120",
+      "elapsed_time": 8590.459214687347,
+      "loss": 0.143,
+      "grad_norm": 0.1381654292345047,
+      "learning_rate": 8.880503144654089e-05,
+      "epoch": 0.559375
+    },
+    {
+      "step": 1496,
+      "timestamp": "2025-12-28T11:21:26.924445",
+      "elapsed_time": 8601.209535360336,
+      "loss": 0.1798,
+      "grad_norm": 0.12096145004034042,
+      "learning_rate": 8.867924528301888e-05,
+      "epoch": 0.56
+    },
+    {
+      "step": 1497,
+      "timestamp": "2025-12-28T11:21:34.078375",
+      "elapsed_time": 8608.363465070724,
+      "loss": 0.1584,
+      "grad_norm": 0.13140904903411865,
+      "learning_rate": 8.855345911949686e-05,
+      "epoch": 0.560625
+    },
+    {
+      "step": 1498,
+      "timestamp": "2025-12-28T11:21:44.697298",
+      "elapsed_time": 8618.982388973236,
+      "loss": 0.2015,
+      "grad_norm": 0.13744622468948364,
+      "learning_rate": 8.842767295597485e-05,
+      "epoch": 0.56125
+    },
+    {
+      "step": 1499,
+      "timestamp": "2025-12-28T11:21:53.321142",
+      "elapsed_time": 8627.60623216629,
+      "loss": 0.1702,
+      "grad_norm": 0.10851988196372986,
+      "learning_rate": 8.830188679245283e-05,
+      "epoch": 0.561875
+    },
+    {
+      "step": 1500,
+      "timestamp": "2025-12-28T11:22:01.457244",
+      "elapsed_time": 8635.7423350811,
+      "loss": 0.1723,
+      "grad_norm": 0.123291976749897,
+      "learning_rate": 8.817610062893082e-05,
+      "epoch": 0.5625
+    },
+    {
+      "step": 1501,
+      "timestamp": "2025-12-28T11:22:14.727687",
+      "elapsed_time": 8649.012777328491,
+      "loss": 0.1281,
+      "grad_norm": 0.09212938696146011,
+      "learning_rate": 8.805031446540882e-05,
+      "epoch": 0.563125
+    },
+    {
+      "step": 1502,
+      "timestamp": "2025-12-28T11:22:31.623445",
+      "elapsed_time": 8665.90853524208,
+      "loss": 0.1026,
+      "grad_norm": 0.0903933122754097,
+      "learning_rate": 8.79245283018868e-05,
+      "epoch": 0.56375
+    },
+    {
+      "step": 1503,
+      "timestamp": "2025-12-28T11:22:44.799315",
+      "elapsed_time": 8679.084406137466,
+      "loss": 0.1257,
+      "grad_norm": 0.10182217508554459,
+      "learning_rate": 8.779874213836479e-05,
+      "epoch": 0.564375
+    },
+    {
+      "step": 1504,
+      "timestamp": "2025-12-28T11:22:49.619397",
+      "elapsed_time": 8683.904487848282,
+      "loss": 0.3112,
+      "grad_norm": 0.19652864336967468,
+      "learning_rate": 8.767295597484277e-05,
+      "epoch": 0.565
+    },
+    {
+      "step": 1505,
+      "timestamp": "2025-12-28T11:22:55.523465",
+      "elapsed_time": 8689.808555364609,
+      "loss": 0.1635,
+      "grad_norm": 0.12541894614696503,
+      "learning_rate": 8.754716981132076e-05,
+      "epoch": 0.565625
+    },
+    {
+      "step": 1506,
+      "timestamp": "2025-12-28T11:23:00.308337",
+      "elapsed_time": 8694.593427419662,
+      "loss": 0.2257,
+      "grad_norm": 0.18083442747592926,
+      "learning_rate": 8.742138364779875e-05,
+      "epoch": 0.56625
+    },
+    {
+      "step": 1507,
+      "timestamp": "2025-12-28T11:23:05.013413",
+      "elapsed_time": 8699.298503875732,
+      "loss": 0.2438,
+      "grad_norm": 0.19925068318843842,
+      "learning_rate": 8.729559748427673e-05,
+      "epoch": 0.566875
+    },
+    {
+      "step": 1508,
+      "timestamp": "2025-12-28T11:23:15.771520",
+      "elapsed_time": 8710.056610822678,
+      "loss": 0.1307,
+      "grad_norm": 0.0917578861117363,
+      "learning_rate": 8.716981132075472e-05,
+      "epoch": 0.5675
+    },
+    {
+      "step": 1509,
+      "timestamp": "2025-12-28T11:23:35.259930",
+      "elapsed_time": 8729.545020341873,
+      "loss": 0.1047,
+      "grad_norm": 0.0767156183719635,
+      "learning_rate": 8.704402515723272e-05,
+      "epoch": 0.568125
+    },
+    {
+      "step": 1510,
+      "timestamp": "2025-12-28T11:23:53.128246",
+      "elapsed_time": 8747.413336277008,
+      "loss": 0.1,
+      "grad_norm": 0.08321381360292435,
+      "learning_rate": 8.69182389937107e-05,
+      "epoch": 0.56875
+    },
+    {
+      "step": 1511,
+      "timestamp": "2025-12-28T11:24:06.430650",
+      "elapsed_time": 8760.715740919113,
+      "loss": 0.1469,
+      "grad_norm": 0.09777943044900894,
+      "learning_rate": 8.679245283018869e-05,
+      "epoch": 0.569375
+    },
+    {
+      "step": 1512,
+      "timestamp": "2025-12-28T11:24:17.892332",
+      "elapsed_time": 8772.17742228508,
+      "loss": 0.1326,
+      "grad_norm": 0.10049686580896378,
+      "learning_rate": 8.666666666666667e-05,
+      "epoch": 0.57
+    },
+    {
+      "step": 1513,
+      "timestamp": "2025-12-28T11:24:25.872236",
+      "elapsed_time": 8780.157325983047,
+      "loss": 0.1542,
+      "grad_norm": 0.11362040787935257,
+      "learning_rate": 8.654088050314466e-05,
+      "epoch": 0.570625
+    },
+    {
+      "step": 1514,
+      "timestamp": "2025-12-28T11:24:30.494130",
+      "elapsed_time": 8784.779220819473,
+      "loss": 0.2623,
+      "grad_norm": 0.1871953308582306,
+      "learning_rate": 8.641509433962265e-05,
+      "epoch": 0.57125
+    },
+    {
+      "step": 1515,
+      "timestamp": "2025-12-28T11:24:36.507144",
+      "elapsed_time": 8790.792234420776,
+      "loss": 0.1441,
+      "grad_norm": 0.1314094513654709,
+      "learning_rate": 8.628930817610063e-05,
+      "epoch": 0.571875
+    },
+    {
+      "step": 1516,
+      "timestamp": "2025-12-28T11:24:49.493922",
+      "elapsed_time": 8803.779012680054,
+      "loss": 0.1849,
+      "grad_norm": 0.10825859010219574,
+      "learning_rate": 8.616352201257863e-05,
+      "epoch": 0.5725
+    },
+    {
+      "step": 1517,
+      "timestamp": "2025-12-28T11:24:56.586838",
+      "elapsed_time": 8810.871928453445,
+      "loss": 0.223,
+      "grad_norm": 0.13489221036434174,
+      "learning_rate": 8.603773584905662e-05,
+      "epoch": 0.573125
+    },
+    {
+      "step": 1518,
+      "timestamp": "2025-12-28T11:25:07.311983",
+      "elapsed_time": 8821.597073554993,
+      "loss": 0.1414,
+      "grad_norm": 0.1013222485780716,
+      "learning_rate": 8.59119496855346e-05,
+      "epoch": 0.57375
+    },
+    {
+      "step": 1519,
+      "timestamp": "2025-12-28T11:25:19.598691",
+      "elapsed_time": 8833.883781433105,
+      "loss": 0.1759,
+      "grad_norm": 0.10948710888624191,
+      "learning_rate": 8.578616352201259e-05,
+      "epoch": 0.574375
+    },
+    {
+      "step": 1520,
+      "timestamp": "2025-12-28T11:25:24.829791",
+      "elapsed_time": 8839.114881277084,
+      "loss": 0.2123,
+      "grad_norm": 0.1585264950990677,
+      "learning_rate": 8.566037735849057e-05,
+      "epoch": 0.575
+    },
+    {
+      "step": 1521,
+      "timestamp": "2025-12-28T11:25:31.895164",
+      "elapsed_time": 8846.180253505707,
+      "loss": 0.2303,
+      "grad_norm": 0.15961548686027527,
+      "learning_rate": 8.553459119496856e-05,
+      "epoch": 0.575625
+    },
+    {
+      "step": 1522,
+      "timestamp": "2025-12-28T11:25:41.238568",
+      "elapsed_time": 8855.523658752441,
+      "loss": 0.2164,
+      "grad_norm": 0.12455689162015915,
+      "learning_rate": 8.540880503144655e-05,
+      "epoch": 0.57625
+    },
+    {
+      "step": 1523,
+      "timestamp": "2025-12-28T11:26:00.476163",
+      "elapsed_time": 8874.76125407219,
+      "loss": 0.0883,
+      "grad_norm": 0.06583801656961441,
+      "learning_rate": 8.528301886792453e-05,
+      "epoch": 0.576875
+    },
+    {
+      "step": 1524,
+      "timestamp": "2025-12-28T11:26:06.460351",
+      "elapsed_time": 8880.745441198349,
+      "loss": 0.196,
+      "grad_norm": 0.14068111777305603,
+      "learning_rate": 8.515723270440253e-05,
+      "epoch": 0.5775
+    },
+    {
+      "step": 1525,
+      "timestamp": "2025-12-28T11:26:20.362662",
+      "elapsed_time": 8894.647752046585,
+      "loss": 0.125,
+      "grad_norm": 0.08721217513084412,
+      "learning_rate": 8.50314465408805e-05,
+      "epoch": 0.578125
+    },
+    {
+      "step": 1526,
+      "timestamp": "2025-12-28T11:26:32.991352",
+      "elapsed_time": 8907.276442289352,
+      "loss": 0.1126,
+      "grad_norm": 0.08798250555992126,
+      "learning_rate": 8.49056603773585e-05,
+      "epoch": 0.57875
+    },
+    {
+      "step": 1527,
+      "timestamp": "2025-12-28T11:26:40.116742",
+      "elapsed_time": 8914.401833057404,
+      "loss": 0.1558,
+      "grad_norm": 0.12924781441688538,
+      "learning_rate": 8.477987421383649e-05,
+      "epoch": 0.579375
+    },
+    {
+      "step": 1528,
+      "timestamp": "2025-12-28T11:26:53.096204",
+      "elapsed_time": 8927.381293773651,
+      "loss": 0.1997,
+      "grad_norm": 0.11501215398311615,
+      "learning_rate": 8.465408805031447e-05,
+      "epoch": 0.58
+    },
+    {
+      "step": 1529,
+      "timestamp": "2025-12-28T11:27:01.938294",
+      "elapsed_time": 8936.223383903503,
+      "loss": 0.1264,
+      "grad_norm": 0.1118590384721756,
+      "learning_rate": 8.452830188679246e-05,
+      "epoch": 0.580625
+    },
+    {
+      "step": 1530,
+      "timestamp": "2025-12-28T11:27:13.567330",
+      "elapsed_time": 8947.852420091629,
+      "loss": 0.1195,
+      "grad_norm": 0.08885247260332108,
+      "learning_rate": 8.440251572327045e-05,
+      "epoch": 0.58125
+    },
+    {
+      "step": 1531,
+      "timestamp": "2025-12-28T11:27:28.081330",
+      "elapsed_time": 8962.366420507431,
+      "loss": 0.1682,
+      "grad_norm": 0.10244069248437881,
+      "learning_rate": 8.427672955974843e-05,
+      "epoch": 0.581875
+    },
+    {
+      "step": 1532,
+      "timestamp": "2025-12-28T11:27:35.182126",
+      "elapsed_time": 8969.46721625328,
+      "loss": 0.2695,
+      "grad_norm": 0.14979572594165802,
+      "learning_rate": 8.415094339622643e-05,
+      "epoch": 0.5825
+    },
+    {
+      "step": 1533,
+      "timestamp": "2025-12-28T11:27:43.626751",
+      "elapsed_time": 8977.911841392517,
+      "loss": 0.3527,
+      "grad_norm": 0.2613980174064636,
+      "learning_rate": 8.40251572327044e-05,
+      "epoch": 0.583125
+    },
+    {
+      "step": 1534,
+      "timestamp": "2025-12-28T11:28:04.513663",
+      "elapsed_time": 8998.798753499985,
+      "loss": 0.1256,
+      "grad_norm": 0.09623493254184723,
+      "learning_rate": 8.38993710691824e-05,
+      "epoch": 0.58375
+    },
+    {
+      "step": 1535,
+      "timestamp": "2025-12-28T11:28:10.502220",
+      "elapsed_time": 9004.787314653397,
+      "loss": 0.1629,
+      "grad_norm": 0.19102250039577484,
+      "learning_rate": 8.377358490566039e-05,
+      "epoch": 0.584375
+    },
+    {
+      "step": 1536,
+      "timestamp": "2025-12-28T11:28:22.123196",
+      "elapsed_time": 9016.40828704834,
+      "loss": 0.1249,
+      "grad_norm": 0.2857312262058258,
+      "learning_rate": 8.364779874213837e-05,
+      "epoch": 0.585
+    },
+    {
+      "step": 1537,
+      "timestamp": "2025-12-28T11:28:33.969444",
+      "elapsed_time": 9028.254534721375,
+      "loss": 0.1436,
+      "grad_norm": 0.0983673706650734,
+      "learning_rate": 8.352201257861636e-05,
+      "epoch": 0.585625
+    },
+    {
+      "step": 1538,
+      "timestamp": "2025-12-28T11:28:40.851432",
+      "elapsed_time": 9035.136521816254,
+      "loss": 0.2335,
+      "grad_norm": 0.15554088354110718,
+      "learning_rate": 8.339622641509434e-05,
+      "epoch": 0.58625
+    },
+    {
+      "step": 1539,
+      "timestamp": "2025-12-28T11:28:49.945462",
+      "elapsed_time": 9044.230551958084,
+      "loss": 0.1401,
+      "grad_norm": 0.11018933355808258,
+      "learning_rate": 8.327044025157233e-05,
+      "epoch": 0.586875
+    },
+    {
+      "step": 1540,
+      "timestamp": "2025-12-28T11:28:58.168763",
+      "elapsed_time": 9052.45385313034,
+      "loss": 0.1303,
+      "grad_norm": 0.1073099821805954,
+      "learning_rate": 8.314465408805033e-05,
+      "epoch": 0.5875
+    },
+    {
+      "step": 1541,
+      "timestamp": "2025-12-28T11:29:12.414620",
+      "elapsed_time": 9066.699710607529,
+      "loss": 0.1614,
+      "grad_norm": 0.10494954884052277,
+      "learning_rate": 8.30188679245283e-05,
+      "epoch": 0.588125
+    },
+    {
+      "step": 1542,
+      "timestamp": "2025-12-28T11:29:20.721392",
+      "elapsed_time": 9075.006482601166,
+      "loss": 0.1503,
+      "grad_norm": 0.11659257858991623,
+      "learning_rate": 8.28930817610063e-05,
+      "epoch": 0.58875
+    },
+    {
+      "step": 1543,
+      "timestamp": "2025-12-28T11:29:33.420507",
+      "elapsed_time": 9087.705597639084,
+      "loss": 0.1244,
+      "grad_norm": 0.09270057827234268,
+      "learning_rate": 8.276729559748429e-05,
+      "epoch": 0.589375
+    },
+    {
+      "step": 1544,
+      "timestamp": "2025-12-28T11:29:42.359077",
+      "elapsed_time": 9096.64416718483,
+      "loss": 0.1409,
+      "grad_norm": 0.11551465094089508,
+      "learning_rate": 8.264150943396227e-05,
+      "epoch": 0.59
+    },
+    {
+      "step": 1545,
+      "timestamp": "2025-12-28T11:29:58.755264",
+      "elapsed_time": 9113.040354728699,
+      "loss": 0.1204,
+      "grad_norm": 0.09296566992998123,
+      "learning_rate": 8.251572327044026e-05,
+      "epoch": 0.590625
+    },
+    {
+      "step": 1546,
+      "timestamp": "2025-12-28T11:30:04.541006",
+      "elapsed_time": 9118.826096773148,
+      "loss": 0.1537,
+      "grad_norm": 0.33984237909317017,
+      "learning_rate": 8.238993710691824e-05,
+      "epoch": 0.59125
+    },
+    {
+      "step": 1547,
+      "timestamp": "2025-12-28T11:30:15.055486",
+      "elapsed_time": 9129.340576410294,
+      "loss": 0.1195,
+      "grad_norm": 0.11354348808526993,
+      "learning_rate": 8.226415094339623e-05,
+      "epoch": 0.591875
+    },
+    {
+      "step": 1548,
+      "timestamp": "2025-12-28T11:30:20.580712",
+      "elapsed_time": 9134.865801811218,
+      "loss": 0.2554,
+      "grad_norm": 0.17498965561389923,
+      "learning_rate": 8.213836477987423e-05,
+      "epoch": 0.5925
+    },
+    {
+      "step": 1549,
+      "timestamp": "2025-12-28T11:30:25.852317",
+      "elapsed_time": 9140.137407064438,
+      "loss": 0.2915,
+      "grad_norm": 0.17898206412792206,
+      "learning_rate": 8.20125786163522e-05,
+      "epoch": 0.593125
+    },
+    {
+      "step": 1550,
+      "timestamp": "2025-12-28T11:30:37.662920",
+      "elapsed_time": 9151.948010444641,
+      "loss": 0.2806,
+      "grad_norm": 0.14019358158111572,
+      "learning_rate": 8.18867924528302e-05,
+      "epoch": 0.59375
+    },
+    {
+      "step": 1551,
+      "timestamp": "2025-12-28T11:30:45.929046",
+      "elapsed_time": 9160.214137077332,
+      "loss": 0.171,
+      "grad_norm": 0.12632305920124054,
+      "learning_rate": 8.176100628930818e-05,
+      "epoch": 0.594375
+    },
+    {
+      "step": 1552,
+      "timestamp": "2025-12-28T11:30:56.181321",
+      "elapsed_time": 9170.466411352158,
+      "loss": 0.1839,
+      "grad_norm": 0.11256668716669083,
+      "learning_rate": 8.163522012578617e-05,
+      "epoch": 0.595
+    },
+    {
+      "step": 1553,
+      "timestamp": "2025-12-28T11:31:07.963905",
+      "elapsed_time": 9182.248995542526,
+      "loss": 0.325,
+      "grad_norm": 0.1411469578742981,
+      "learning_rate": 8.150943396226416e-05,
+      "epoch": 0.595625
+    },
+    {
+      "step": 1554,
+      "timestamp": "2025-12-28T11:31:15.818950",
+      "elapsed_time": 9190.104040145874,
+      "loss": 0.2134,
+      "grad_norm": 0.157893106341362,
+      "learning_rate": 8.138364779874214e-05,
+      "epoch": 0.59625
+    },
+    {
+      "step": 1555,
+      "timestamp": "2025-12-28T11:31:22.342648",
+      "elapsed_time": 9196.627738714218,
+      "loss": 0.1857,
+      "grad_norm": 0.12139203399419785,
+      "learning_rate": 8.125786163522013e-05,
+      "epoch": 0.596875
+    },
+    {
+      "step": 1556,
+      "timestamp": "2025-12-28T11:31:32.092816",
+      "elapsed_time": 9206.37790632248,
+      "loss": 0.1616,
+      "grad_norm": 0.10829068720340729,
+      "learning_rate": 8.113207547169813e-05,
+      "epoch": 0.5975
+    },
+    {
+      "step": 1557,
+      "timestamp": "2025-12-28T11:31:38.204155",
+      "elapsed_time": 9212.489244699478,
+      "loss": 0.2388,
+      "grad_norm": 0.15463301539421082,
+      "learning_rate": 8.10062893081761e-05,
+      "epoch": 0.598125
+    },
+    {
+      "step": 1558,
+      "timestamp": "2025-12-28T11:31:46.978385",
+      "elapsed_time": 9221.263475179672,
+      "loss": 0.1431,
+      "grad_norm": 0.1963123232126236,
+      "learning_rate": 8.08805031446541e-05,
+      "epoch": 0.59875
+    },
+    {
+      "step": 1559,
+      "timestamp": "2025-12-28T11:31:51.937096",
+      "elapsed_time": 9226.222190618515,
+      "loss": 0.2025,
+      "grad_norm": 0.15821681916713715,
+      "learning_rate": 8.075471698113208e-05,
+      "epoch": 0.599375
+    },
+    {
+      "step": 1560,
+      "timestamp": "2025-12-28T11:32:04.000430",
+      "elapsed_time": 9238.285520553589,
+      "loss": 0.3795,
+      "grad_norm": 0.1718837171792984,
+      "learning_rate": 8.062893081761007e-05,
+      "epoch": 0.6
+    },
+    {
+      "step": 1561,
+      "timestamp": "2025-12-28T11:32:08.448180",
+      "elapsed_time": 9242.73327088356,
+      "loss": 0.2619,
+      "grad_norm": 0.19040848314762115,
+      "learning_rate": 8.050314465408806e-05,
+      "epoch": 0.600625
+    },
+    {
+      "step": 1562,
+      "timestamp": "2025-12-28T11:32:15.075715",
+      "elapsed_time": 9249.360805511475,
+      "loss": 0.1528,
+      "grad_norm": 0.14181895554065704,
+      "learning_rate": 8.037735849056604e-05,
+      "epoch": 0.60125
+    },
+    {
+      "step": 1563,
+      "timestamp": "2025-12-28T11:32:24.698558",
+      "elapsed_time": 9258.983648777008,
+      "loss": 0.1461,
+      "grad_norm": 0.11014454066753387,
+      "learning_rate": 8.025157232704403e-05,
+      "epoch": 0.601875
+    },
+    {
+      "step": 1564,
+      "timestamp": "2025-12-28T11:32:29.643873",
+      "elapsed_time": 9263.928963661194,
+      "loss": 0.3451,
+      "grad_norm": 0.19026674330234528,
+      "learning_rate": 8.012578616352203e-05,
+      "epoch": 0.6025
+    },
+    {
+      "step": 1565,
+      "timestamp": "2025-12-28T11:32:36.118502",
+      "elapsed_time": 9270.403592348099,
+      "loss": 0.2698,
+      "grad_norm": 0.15855702757835388,
+      "learning_rate": 8e-05,
+      "epoch": 0.603125
+    },
+    {
+      "step": 1566,
+      "timestamp": "2025-12-28T11:32:47.584312",
+      "elapsed_time": 9281.869402885437,
+      "loss": 0.1317,
+      "grad_norm": 0.10985270142555237,
+      "learning_rate": 7.9874213836478e-05,
+      "epoch": 0.60375
+    },
+    {
+      "step": 1567,
+      "timestamp": "2025-12-28T11:32:57.951733",
+      "elapsed_time": 9292.236823320389,
+      "loss": 0.1344,
+      "grad_norm": 0.09886456280946732,
+      "learning_rate": 7.974842767295598e-05,
+      "epoch": 0.604375
+    },
+    {
+      "step": 1568,
+      "timestamp": "2025-12-28T11:33:08.832307",
+      "elapsed_time": 9303.117396831512,
+      "loss": 0.3551,
+      "grad_norm": 0.13379919528961182,
+      "learning_rate": 7.962264150943397e-05,
+      "epoch": 0.605
+    },
+    {
+      "step": 1569,
+      "timestamp": "2025-12-28T11:33:15.380680",
+      "elapsed_time": 9309.665770292282,
+      "loss": 0.4387,
+      "grad_norm": 0.20473594963550568,
+      "learning_rate": 7.949685534591196e-05,
+      "epoch": 0.605625
+    },
+    {
+      "step": 1570,
+      "timestamp": "2025-12-28T11:33:28.420603",
+      "elapsed_time": 9322.705693244934,
+      "loss": 0.1534,
+      "grad_norm": 0.09838785231113434,
+      "learning_rate": 7.937106918238994e-05,
+      "epoch": 0.60625
+    },
+    {
+      "step": 1571,
+      "timestamp": "2025-12-28T11:33:36.492152",
+      "elapsed_time": 9330.777242422104,
+      "loss": 0.4969,
+      "grad_norm": 0.19386343657970428,
+      "learning_rate": 7.924528301886794e-05,
+      "epoch": 0.606875
+    },
+    {
+      "step": 1572,
+      "timestamp": "2025-12-28T11:33:44.673340",
+      "elapsed_time": 9338.958430051804,
+      "loss": 0.1638,
+      "grad_norm": 0.1306898593902588,
+      "learning_rate": 7.911949685534591e-05,
+      "epoch": 0.6075
+    },
+    {
+      "step": 1573,
+      "timestamp": "2025-12-28T11:33:49.969537",
+      "elapsed_time": 9344.254627227783,
+      "loss": 0.1831,
+      "grad_norm": 0.1826959103345871,
+      "learning_rate": 7.899371069182391e-05,
+      "epoch": 0.608125
+    },
+    {
+      "step": 1574,
+      "timestamp": "2025-12-28T11:33:57.344461",
+      "elapsed_time": 9351.629550933838,
+      "loss": 0.1919,
+      "grad_norm": 0.1459428369998932,
+      "learning_rate": 7.88679245283019e-05,
+      "epoch": 0.60875
+    },
+    {
+      "step": 1575,
+      "timestamp": "2025-12-28T11:34:07.601020",
+      "elapsed_time": 9361.886110544205,
+      "loss": 0.1914,
+      "grad_norm": 0.11683019995689392,
+      "learning_rate": 7.874213836477988e-05,
+      "epoch": 0.609375
+    },
+    {
+      "step": 1576,
+      "timestamp": "2025-12-28T11:34:12.734692",
+      "elapsed_time": 9367.019781827927,
+      "loss": 0.162,
+      "grad_norm": 0.14056500792503357,
+      "learning_rate": 7.861635220125787e-05,
+      "epoch": 0.61
+    },
+    {
+      "step": 1577,
+      "timestamp": "2025-12-28T11:34:30.402188",
+      "elapsed_time": 9384.687278032303,
+      "loss": 0.1201,
+      "grad_norm": 0.08820690959692001,
+      "learning_rate": 7.849056603773586e-05,
+      "epoch": 0.610625
+    },
+    {
+      "step": 1578,
+      "timestamp": "2025-12-28T11:34:39.188114",
+      "elapsed_time": 9393.473204135895,
+      "loss": 0.2578,
+      "grad_norm": 0.14672920107841492,
+      "learning_rate": 7.836477987421384e-05,
+      "epoch": 0.61125
+    },
+    {
+      "step": 1579,
+      "timestamp": "2025-12-28T11:34:50.013659",
+      "elapsed_time": 9404.298749446869,
+      "loss": 0.1506,
+      "grad_norm": 0.11653705686330795,
+      "learning_rate": 7.823899371069184e-05,
+      "epoch": 0.611875
+    },
+    {
+      "step": 1580,
+      "timestamp": "2025-12-28T11:34:54.799485",
+      "elapsed_time": 9409.084575176239,
+      "loss": 0.1957,
+      "grad_norm": 0.1690455675125122,
+      "learning_rate": 7.811320754716981e-05,
+      "epoch": 0.6125
+    },
+    {
+      "step": 1581,
+      "timestamp": "2025-12-28T11:34:58.274219",
+      "elapsed_time": 9412.559309482574,
+      "loss": 0.3539,
+      "grad_norm": 0.24397805333137512,
+      "learning_rate": 7.798742138364781e-05,
+      "epoch": 0.613125
+    },
+    {
+      "step": 1582,
+      "timestamp": "2025-12-28T11:35:05.235925",
+      "elapsed_time": 9419.521015405655,
+      "loss": 0.3798,
+      "grad_norm": 0.22178561985492706,
+      "learning_rate": 7.78616352201258e-05,
+      "epoch": 0.61375
+    },
+    {
+      "step": 1583,
+      "timestamp": "2025-12-28T11:35:13.675308",
+      "elapsed_time": 9427.960398435593,
+      "loss": 0.2782,
+      "grad_norm": 0.14682307839393616,
+      "learning_rate": 7.773584905660378e-05,
+      "epoch": 0.614375
+    },
+    {
+      "step": 1584,
+      "timestamp": "2025-12-28T11:35:27.622590",
+      "elapsed_time": 9441.907680273056,
+      "loss": 0.1162,
+      "grad_norm": 0.08615633100271225,
+      "learning_rate": 7.761006289308177e-05,
+      "epoch": 0.615
+    },
+    {
+      "step": 1585,
+      "timestamp": "2025-12-28T11:35:44.742170",
+      "elapsed_time": 9459.02725982666,
+      "loss": 0.0976,
+      "grad_norm": 0.07224409282207489,
+      "learning_rate": 7.748427672955975e-05,
+      "epoch": 0.615625
+    },
+    {
+      "step": 1586,
+      "timestamp": "2025-12-28T11:35:56.023997",
+      "elapsed_time": 9470.30908703804,
+      "loss": 0.1516,
+      "grad_norm": 0.1125619113445282,
+      "learning_rate": 7.735849056603774e-05,
+      "epoch": 0.61625
+    },
+    {
+      "step": 1587,
+      "timestamp": "2025-12-28T11:36:07.646063",
+      "elapsed_time": 9481.931153059006,
+      "loss": 0.1575,
+      "grad_norm": 0.12028831243515015,
+      "learning_rate": 7.723270440251574e-05,
+      "epoch": 0.616875
+    },
+    {
+      "step": 1588,
+      "timestamp": "2025-12-28T11:36:25.636674",
+      "elapsed_time": 9499.92176437378,
+      "loss": 0.1307,
+      "grad_norm": 0.08736097812652588,
+      "learning_rate": 7.710691823899372e-05,
+      "epoch": 0.6175
+    },
+    {
+      "step": 1589,
+      "timestamp": "2025-12-28T11:36:35.410727",
+      "elapsed_time": 9509.69581747055,
+      "loss": 0.2209,
+      "grad_norm": 0.21880212426185608,
+      "learning_rate": 7.698113207547171e-05,
+      "epoch": 0.618125
+    },
+    {
+      "step": 1590,
+      "timestamp": "2025-12-28T11:36:47.358344",
+      "elapsed_time": 9521.643434047699,
+      "loss": 0.3212,
+      "grad_norm": 0.13422144949436188,
+      "learning_rate": 7.68553459119497e-05,
+      "epoch": 0.61875
+    },
+    {
+      "step": 1591,
+      "timestamp": "2025-12-28T11:36:56.658149",
+      "elapsed_time": 9530.943238973618,
+      "loss": 0.3681,
+      "grad_norm": 0.2638307511806488,
+      "learning_rate": 7.672955974842768e-05,
+      "epoch": 0.619375
+    },
+    {
+      "step": 1592,
+      "timestamp": "2025-12-28T11:37:13.127742",
+      "elapsed_time": 9547.41283249855,
+      "loss": 0.0943,
+      "grad_norm": 0.08068544417619705,
+      "learning_rate": 7.660377358490567e-05,
+      "epoch": 0.62
+    },
+    {
+      "step": 1593,
+      "timestamp": "2025-12-28T11:37:21.962911",
+      "elapsed_time": 9556.248000860214,
+      "loss": 0.183,
+      "grad_norm": 0.1172947883605957,
+      "learning_rate": 7.647798742138365e-05,
+      "epoch": 0.620625
+    },
+    {
+      "step": 1594,
+      "timestamp": "2025-12-28T11:37:32.441106",
+      "elapsed_time": 9566.726195812225,
+      "loss": 0.2204,
+      "grad_norm": 0.09986839443445206,
+      "learning_rate": 7.635220125786164e-05,
+      "epoch": 0.62125
+    },
+    {
+      "step": 1595,
+      "timestamp": "2025-12-28T11:37:38.220326",
+      "elapsed_time": 9572.505415916443,
+      "loss": 0.2831,
+      "grad_norm": 0.1995793581008911,
+      "learning_rate": 7.622641509433964e-05,
+      "epoch": 0.621875
+    },
+    {
+      "step": 1596,
+      "timestamp": "2025-12-28T11:37:48.926748",
+      "elapsed_time": 9583.21183848381,
+      "loss": 0.1441,
+      "grad_norm": 0.09849265217781067,
+      "learning_rate": 7.610062893081762e-05,
+      "epoch": 0.6225
+    },
+    {
+      "step": 1597,
+      "timestamp": "2025-12-28T11:37:59.937984",
+      "elapsed_time": 9594.223074674606,
+      "loss": 0.1331,
+      "grad_norm": 0.11908537149429321,
+      "learning_rate": 7.597484276729561e-05,
+      "epoch": 0.623125
+    },
+    {
+      "step": 1598,
+      "timestamp": "2025-12-28T11:38:08.883234",
+      "elapsed_time": 9603.168324947357,
+      "loss": 0.1916,
+      "grad_norm": 0.12535853683948517,
+      "learning_rate": 7.584905660377359e-05,
+      "epoch": 0.62375
+    },
+    {
+      "step": 1599,
+      "timestamp": "2025-12-28T11:38:17.822928",
+      "elapsed_time": 9612.108018875122,
+      "loss": 0.4302,
+      "grad_norm": 0.21333205699920654,
+      "learning_rate": 7.572327044025158e-05,
+      "epoch": 0.624375
+    },
+    {
+      "step": 1600,
+      "timestamp": "2025-12-28T11:38:28.592143",
+      "elapsed_time": 9622.877233028412,
+      "loss": 0.1339,
+      "grad_norm": 0.10051511973142624,
+      "learning_rate": 7.559748427672957e-05,
+      "epoch": 0.625
+    },
+    {
+      "step": 1601,
+      "timestamp": "2025-12-28T11:38:39.199999",
+      "elapsed_time": 9633.485090017319,
+      "loss": 0.1352,
+      "grad_norm": 0.11261122673749924,
+      "learning_rate": 7.547169811320755e-05,
+      "epoch": 0.625625
+    },
+    {
+      "step": 1602,
+      "timestamp": "2025-12-28T11:38:46.300157",
+      "elapsed_time": 9640.58524775505,
+      "loss": 0.1886,
+      "grad_norm": 0.12923336029052734,
+      "learning_rate": 7.534591194968554e-05,
+      "epoch": 0.62625
+    },
+    {
+      "step": 1603,
+      "timestamp": "2025-12-28T11:38:55.288169",
+      "elapsed_time": 9649.5732588768,
+      "loss": 0.1616,
+      "grad_norm": 0.11907650530338287,
+      "learning_rate": 7.522012578616354e-05,
+      "epoch": 0.626875
+    },
+    {
+      "step": 1604,
+      "timestamp": "2025-12-28T11:39:04.668887",
+      "elapsed_time": 9658.953977823257,
+      "loss": 0.1443,
+      "grad_norm": 0.11075244098901749,
+      "learning_rate": 7.509433962264152e-05,
+      "epoch": 0.6275
+    },
+    {
+      "step": 1605,
+      "timestamp": "2025-12-28T11:39:10.754759",
+      "elapsed_time": 9665.039849996567,
+      "loss": 0.2587,
+      "grad_norm": 0.2336777150630951,
+      "learning_rate": 7.49685534591195e-05,
+      "epoch": 0.628125
+    },
+    {
+      "step": 1606,
+      "timestamp": "2025-12-28T11:39:18.045720",
+      "elapsed_time": 9672.330810546875,
+      "loss": 0.2099,
+      "grad_norm": 0.14048390090465546,
+      "learning_rate": 7.484276729559749e-05,
+      "epoch": 0.62875
+    },
+    {
+      "step": 1607,
+      "timestamp": "2025-12-28T11:39:26.880911",
+      "elapsed_time": 9681.166000843048,
+      "loss": 0.1913,
+      "grad_norm": 0.12840668857097626,
+      "learning_rate": 7.471698113207547e-05,
+      "epoch": 0.629375
+    },
+    {
+      "step": 1608,
+      "timestamp": "2025-12-28T11:39:38.404229",
+      "elapsed_time": 9692.689319133759,
+      "loss": 0.1631,
+      "grad_norm": 0.17724475264549255,
+      "learning_rate": 7.459119496855346e-05,
+      "epoch": 0.63
+    },
+    {
+      "step": 1609,
+      "timestamp": "2025-12-28T11:39:46.668636",
+      "elapsed_time": 9700.95373082161,
+      "loss": 0.1678,
+      "grad_norm": 0.12339440733194351,
+      "learning_rate": 7.446540880503144e-05,
+      "epoch": 0.630625
+    },
+    {
+      "step": 1610,
+      "timestamp": "2025-12-28T11:39:55.140506",
+      "elapsed_time": 9709.425596952438,
+      "loss": 0.3796,
+      "grad_norm": 0.15686391294002533,
+      "learning_rate": 7.433962264150943e-05,
+      "epoch": 0.63125
+    },
+    {
+      "step": 1611,
+      "timestamp": "2025-12-28T11:40:03.202851",
+      "elapsed_time": 9717.487941265106,
+      "loss": 0.1415,
+      "grad_norm": 0.1224021390080452,
+      "learning_rate": 7.421383647798742e-05,
+      "epoch": 0.631875
+    },
+    {
+      "step": 1612,
+      "timestamp": "2025-12-28T11:40:10.163533",
+      "elapsed_time": 9724.44862318039,
+      "loss": 0.1966,
+      "grad_norm": 0.17393508553504944,
+      "learning_rate": 7.40880503144654e-05,
+      "epoch": 0.6325
+    },
+    {
+      "step": 1613,
+      "timestamp": "2025-12-28T11:40:24.152425",
+      "elapsed_time": 9738.437515258789,
+      "loss": 0.1186,
+      "grad_norm": 0.08472223579883575,
+      "learning_rate": 7.39622641509434e-05,
+      "epoch": 0.633125
+    },
+    {
+      "step": 1614,
+      "timestamp": "2025-12-28T11:40:45.041966",
+      "elapsed_time": 9759.327056407928,
+      "loss": 0.1331,
+      "grad_norm": 0.08091577142477036,
+      "learning_rate": 7.383647798742139e-05,
+      "epoch": 0.63375
+    },
+    {
+      "step": 1615,
+      "timestamp": "2025-12-28T11:40:52.221602",
+      "elapsed_time": 9766.506692886353,
+      "loss": 0.1571,
+      "grad_norm": 0.12333718687295914,
+      "learning_rate": 7.371069182389937e-05,
+      "epoch": 0.634375
+    },
+    {
+      "step": 1616,
+      "timestamp": "2025-12-28T11:41:02.877788",
+      "elapsed_time": 9777.162879228592,
+      "loss": 0.1733,
+      "grad_norm": 0.12430380284786224,
+      "learning_rate": 7.358490566037736e-05,
+      "epoch": 0.635
+    },
+    {
+      "step": 1617,
+      "timestamp": "2025-12-28T11:41:11.012938",
+      "elapsed_time": 9785.298028707504,
+      "loss": 0.1672,
+      "grad_norm": 0.12017210572957993,
+      "learning_rate": 7.345911949685534e-05,
+      "epoch": 0.635625
+    },
+    {
+      "step": 1618,
+      "timestamp": "2025-12-28T11:41:19.447939",
+      "elapsed_time": 9793.7330327034,
+      "loss": 0.1954,
+      "grad_norm": 0.14666104316711426,
+      "learning_rate": 7.333333333333333e-05,
+      "epoch": 0.63625
+    },
+    {
+      "step": 1619,
+      "timestamp": "2025-12-28T11:41:30.321982",
+      "elapsed_time": 9804.607072114944,
+      "loss": 0.1235,
+      "grad_norm": 0.0968768298625946,
+      "learning_rate": 7.320754716981132e-05,
+      "epoch": 0.636875
+    },
+    {
+      "step": 1620,
+      "timestamp": "2025-12-28T11:41:43.576219",
+      "elapsed_time": 9817.86130952835,
+      "loss": 0.1528,
+      "grad_norm": 0.11526031047105789,
+      "learning_rate": 7.30817610062893e-05,
+      "epoch": 0.6375
+    },
+    {
+      "step": 1621,
+      "timestamp": "2025-12-28T11:41:51.883719",
+      "elapsed_time": 9826.16880941391,
+      "loss": 0.2519,
+      "grad_norm": 0.13182447850704193,
+      "learning_rate": 7.29559748427673e-05,
+      "epoch": 0.638125
+    },
+    {
+      "step": 1622,
+      "timestamp": "2025-12-28T11:42:01.783503",
+      "elapsed_time": 9836.068593025208,
+      "loss": 0.18,
+      "grad_norm": 0.11508210748434067,
+      "learning_rate": 7.283018867924527e-05,
+      "epoch": 0.63875
+    },
+    {
+      "step": 1623,
+      "timestamp": "2025-12-28T11:42:08.222492",
+      "elapsed_time": 9842.507582426071,
+      "loss": 0.1862,
+      "grad_norm": 0.14239539206027985,
+      "learning_rate": 7.270440251572327e-05,
+      "epoch": 0.639375
+    },
+    {
+      "step": 1624,
+      "timestamp": "2025-12-28T11:42:16.155646",
+      "elapsed_time": 9850.440736532211,
+      "loss": 0.3164,
+      "grad_norm": 0.16672523319721222,
+      "learning_rate": 7.257861635220126e-05,
+      "epoch": 0.64
+    },
+    {
+      "step": 1625,
+      "timestamp": "2025-12-28T11:42:22.465387",
+      "elapsed_time": 9856.750477075577,
+      "loss": 0.1875,
+      "grad_norm": 0.13851501047611237,
+      "learning_rate": 7.245283018867924e-05,
+      "epoch": 0.640625
+    },
+    {
+      "step": 1626,
+      "timestamp": "2025-12-28T11:42:43.343959",
+      "elapsed_time": 9877.629049301147,
+      "loss": 0.1101,
+      "grad_norm": 0.07382085174322128,
+      "learning_rate": 7.232704402515723e-05,
+      "epoch": 0.64125
+    },
+    {
+      "step": 1627,
+      "timestamp": "2025-12-28T11:42:48.836245",
+      "elapsed_time": 9883.121335983276,
+      "loss": 0.3851,
+      "grad_norm": 0.19532591104507446,
+      "learning_rate": 7.220125786163522e-05,
+      "epoch": 0.641875
+    },
+    {
+      "step": 1628,
+      "timestamp": "2025-12-28T11:42:54.575703",
+      "elapsed_time": 9888.860793590546,
+      "loss": 0.226,
+      "grad_norm": 0.1721881628036499,
+      "learning_rate": 7.20754716981132e-05,
+      "epoch": 0.6425
+    },
+    {
+      "step": 1629,
+      "timestamp": "2025-12-28T11:43:02.309918",
+      "elapsed_time": 9896.595008134842,
+      "loss": 0.1805,
+      "grad_norm": 0.12694233655929565,
+      "learning_rate": 7.19496855345912e-05,
+      "epoch": 0.643125
+    },
+    {
+      "step": 1630,
+      "timestamp": "2025-12-28T11:43:10.240464",
+      "elapsed_time": 9904.52555346489,
+      "loss": 0.1923,
+      "grad_norm": 0.137771874666214,
+      "learning_rate": 7.182389937106918e-05,
+      "epoch": 0.64375
+    },
+    {
+      "step": 1631,
+      "timestamp": "2025-12-28T11:43:14.942263",
+      "elapsed_time": 9909.227353334427,
+      "loss": 0.3536,
+      "grad_norm": 0.19589699804782867,
+      "learning_rate": 7.169811320754717e-05,
+      "epoch": 0.644375
+    },
+    {
+      "step": 1632,
+      "timestamp": "2025-12-28T11:43:20.961561",
+      "elapsed_time": 9915.2466506958,
+      "loss": 0.1823,
+      "grad_norm": 0.14660844206809998,
+      "learning_rate": 7.157232704402516e-05,
+      "epoch": 0.645
+    },
+    {
+      "step": 1633,
+      "timestamp": "2025-12-28T11:43:25.350571",
+      "elapsed_time": 9919.635661840439,
+      "loss": 0.22,
+      "grad_norm": 0.18052536249160767,
+      "learning_rate": 7.144654088050314e-05,
+      "epoch": 0.645625
+    },
+    {
+      "step": 1634,
+      "timestamp": "2025-12-28T11:43:34.930070",
+      "elapsed_time": 9929.215160131454,
+      "loss": 0.1316,
+      "grad_norm": 0.10462430864572525,
+      "learning_rate": 7.132075471698113e-05,
+      "epoch": 0.64625
+    },
+    {
+      "step": 1635,
+      "timestamp": "2025-12-28T11:43:43.559647",
+      "elapsed_time": 9937.844737291336,
+      "loss": 0.1756,
+      "grad_norm": 0.126973494887352,
+      "learning_rate": 7.119496855345912e-05,
+      "epoch": 0.646875
+    },
+    {
+      "step": 1636,
+      "timestamp": "2025-12-28T11:43:51.704679",
+      "elapsed_time": 9945.989769220352,
+      "loss": 0.1654,
+      "grad_norm": 0.14970456063747406,
+      "learning_rate": 7.10691823899371e-05,
+      "epoch": 0.6475
+    },
+    {
+      "step": 1637,
+      "timestamp": "2025-12-28T11:44:10.864475",
+      "elapsed_time": 9965.149565935135,
+      "loss": 0.1098,
+      "grad_norm": 0.08534051477909088,
+      "learning_rate": 7.09433962264151e-05,
+      "epoch": 0.648125
+    },
+    {
+      "step": 1638,
+      "timestamp": "2025-12-28T11:44:24.042231",
+      "elapsed_time": 9978.327321767807,
+      "loss": 0.1609,
+      "grad_norm": 0.10151031613349915,
+      "learning_rate": 7.081761006289308e-05,
+      "epoch": 0.64875
+    },
+    {
+      "step": 1639,
+      "timestamp": "2025-12-28T11:44:33.412608",
+      "elapsed_time": 9987.697702884674,
+      "loss": 0.1008,
+      "grad_norm": 0.09468650072813034,
+      "learning_rate": 7.069182389937107e-05,
+      "epoch": 0.649375
+    },
+    {
+      "step": 1640,
+      "timestamp": "2025-12-28T11:44:42.835568",
+      "elapsed_time": 9997.120658397675,
+      "loss": 0.1202,
+      "grad_norm": 0.1058143749833107,
+      "learning_rate": 7.056603773584906e-05,
+      "epoch": 0.65
+    },
+    {
+      "step": 1641,
+      "timestamp": "2025-12-28T11:44:51.467264",
+      "elapsed_time": 10005.752353906631,
+      "loss": 0.1627,
+      "grad_norm": 0.12414304912090302,
+      "learning_rate": 7.044025157232704e-05,
+      "epoch": 0.650625
+    },
+    {
+      "step": 1642,
+      "timestamp": "2025-12-28T11:45:00.450823",
+      "elapsed_time": 10014.735912799835,
+      "loss": 0.1949,
+      "grad_norm": 0.13407202064990997,
+      "learning_rate": 7.031446540880503e-05,
+      "epoch": 0.65125
+    },
+    {
+      "step": 1643,
+      "timestamp": "2025-12-28T11:45:09.542076",
+      "elapsed_time": 10023.827166318893,
+      "loss": 0.1587,
+      "grad_norm": 0.19232605397701263,
+      "learning_rate": 7.018867924528301e-05,
+      "epoch": 0.651875
+    },
+    {
+      "step": 1644,
+      "timestamp": "2025-12-28T11:45:17.892732",
+      "elapsed_time": 10032.177821874619,
+      "loss": 0.1543,
+      "grad_norm": 0.12963691353797913,
+      "learning_rate": 7.0062893081761e-05,
+      "epoch": 0.6525
+    },
+    {
+      "step": 1645,
+      "timestamp": "2025-12-28T11:45:29.559165",
+      "elapsed_time": 10043.844255447388,
+      "loss": 0.1364,
+      "grad_norm": 0.09772875905036926,
+      "learning_rate": 6.9937106918239e-05,
+      "epoch": 0.653125
+    },
+    {
+      "step": 1646,
+      "timestamp": "2025-12-28T11:45:34.381174",
+      "elapsed_time": 10048.66626906395,
+      "loss": 0.2831,
+      "grad_norm": 0.20503878593444824,
+      "learning_rate": 6.981132075471698e-05,
+      "epoch": 0.65375
+    },
+    {
+      "step": 1647,
+      "timestamp": "2025-12-28T11:45:38.491344",
+      "elapsed_time": 10052.77643442154,
+      "loss": 0.4374,
+      "grad_norm": 0.20627865195274353,
+      "learning_rate": 6.968553459119497e-05,
+      "epoch": 0.654375
+    },
+    {
+      "step": 1648,
+      "timestamp": "2025-12-28T11:45:44.518475",
+      "elapsed_time": 10058.803565263748,
+      "loss": 0.1858,
+      "grad_norm": 0.16573020815849304,
+      "learning_rate": 6.955974842767296e-05,
+      "epoch": 0.655
+    },
+    {
+      "step": 1649,
+      "timestamp": "2025-12-28T11:46:05.413855",
+      "elapsed_time": 10079.698945760727,
+      "loss": 0.0942,
+      "grad_norm": 0.07246656715869904,
+      "learning_rate": 6.943396226415094e-05,
+      "epoch": 0.655625
+    },
+    {
+      "step": 1650,
+      "timestamp": "2025-12-28T11:46:24.467732",
+      "elapsed_time": 10098.752822637558,
+      "loss": 0.1091,
+      "grad_norm": 0.09009367972612381,
+      "learning_rate": 6.930817610062893e-05,
+      "epoch": 0.65625
+    },
+    {
+      "step": 1651,
+      "timestamp": "2025-12-28T11:46:33.886132",
+      "elapsed_time": 10108.17122220993,
+      "loss": 0.2347,
+      "grad_norm": 0.12859852612018585,
+      "learning_rate": 6.918238993710691e-05,
+      "epoch": 0.656875
+    },
+    {
+      "step": 1652,
+      "timestamp": "2025-12-28T11:46:42.823278",
+      "elapsed_time": 10117.108368635178,
+      "loss": 0.1737,
+      "grad_norm": 0.11088800430297852,
+      "learning_rate": 6.90566037735849e-05,
+      "epoch": 0.6575
+    },
+    {
+      "step": 1653,
+      "timestamp": "2025-12-28T11:46:54.658395",
+      "elapsed_time": 10128.943485021591,
+      "loss": 0.1427,
+      "grad_norm": 0.10402551293373108,
+      "learning_rate": 6.89308176100629e-05,
+      "epoch": 0.658125
+    },
+    {
+      "step": 1654,
+      "timestamp": "2025-12-28T11:47:03.630914",
+      "elapsed_time": 10137.916004419327,
+      "loss": 0.1378,
+      "grad_norm": 0.11617520451545715,
+      "learning_rate": 6.880503144654088e-05,
+      "epoch": 0.65875
+    },
+    {
+      "step": 1655,
+      "timestamp": "2025-12-28T11:47:13.389780",
+      "elapsed_time": 10147.67487025261,
+      "loss": 0.2234,
+      "grad_norm": 0.11303507536649704,
+      "learning_rate": 6.867924528301887e-05,
+      "epoch": 0.659375
+    },
+    {
+      "step": 1656,
+      "timestamp": "2025-12-28T11:47:26.010568",
+      "elapsed_time": 10160.29565834999,
+      "loss": 0.2006,
+      "grad_norm": 0.11115586757659912,
+      "learning_rate": 6.855345911949685e-05,
+      "epoch": 0.66
+    },
+    {
+      "step": 1657,
+      "timestamp": "2025-12-28T11:47:33.149623",
+      "elapsed_time": 10167.434713602066,
+      "loss": 0.4581,
+      "grad_norm": 0.19294337928295135,
+      "learning_rate": 6.842767295597484e-05,
+      "epoch": 0.660625
+    },
+    {
+      "step": 1658,
+      "timestamp": "2025-12-28T11:47:43.773293",
+      "elapsed_time": 10178.058383464813,
+      "loss": 0.197,
+      "grad_norm": 0.1200195699930191,
+      "learning_rate": 6.830188679245283e-05,
+      "epoch": 0.66125
+    },
+    {
+      "step": 1659,
+      "timestamp": "2025-12-28T11:47:58.867153",
+      "elapsed_time": 10193.152243375778,
+      "loss": 0.1369,
+      "grad_norm": 0.09150442481040955,
+      "learning_rate": 6.817610062893081e-05,
+      "epoch": 0.661875
+    },
+    {
+      "step": 1660,
+      "timestamp": "2025-12-28T11:48:11.905738",
+      "elapsed_time": 10206.190828084946,
+      "loss": 0.1244,
+      "grad_norm": 0.09038378298282623,
+      "learning_rate": 6.80503144654088e-05,
+      "epoch": 0.6625
+    },
+    {
+      "step": 1661,
+      "timestamp": "2025-12-28T11:48:31.524146",
+      "elapsed_time": 10225.80923986435,
+      "loss": 0.1345,
+      "grad_norm": 0.08619663864374161,
+      "learning_rate": 6.79245283018868e-05,
+      "epoch": 0.663125
+    },
+    {
+      "step": 1662,
+      "timestamp": "2025-12-28T11:48:46.868927",
+      "elapsed_time": 10241.154017448425,
+      "loss": 0.1112,
+      "grad_norm": 0.08081649988889694,
+      "learning_rate": 6.779874213836478e-05,
+      "epoch": 0.66375
+    },
+    {
+      "step": 1663,
+      "timestamp": "2025-12-28T11:48:56.461538",
+      "elapsed_time": 10250.746627807617,
+      "loss": 0.1567,
+      "grad_norm": 0.12073066085577011,
+      "learning_rate": 6.767295597484277e-05,
+      "epoch": 0.664375
+    },
+    {
+      "step": 1664,
+      "timestamp": "2025-12-28T11:49:01.469922",
+      "elapsed_time": 10255.755012512207,
+      "loss": 0.2881,
+      "grad_norm": 0.17294885218143463,
+      "learning_rate": 6.754716981132075e-05,
+      "epoch": 0.665
+    },
+    {
+      "step": 1665,
+      "timestamp": "2025-12-28T11:49:14.018002",
+      "elapsed_time": 10268.303092956543,
+      "loss": 0.1356,
+      "grad_norm": 0.09254828095436096,
+      "learning_rate": 6.742138364779874e-05,
+      "epoch": 0.665625
+    },
+    {
+      "step": 1666,
+      "timestamp": "2025-12-28T11:49:34.915088",
+      "elapsed_time": 10289.20017838478,
+      "loss": 0.0919,
+      "grad_norm": 0.06759099662303925,
+      "learning_rate": 6.729559748427673e-05,
+      "epoch": 0.66625
+    },
+    {
+      "step": 1667,
+      "timestamp": "2025-12-28T11:49:43.859625",
+      "elapsed_time": 10298.144714593887,
+      "loss": 0.1992,
+      "grad_norm": 0.11738283932209015,
+      "learning_rate": 6.716981132075471e-05,
+      "epoch": 0.666875
+    },
+    {
+      "step": 1668,
+      "timestamp": "2025-12-28T11:49:54.744691",
+      "elapsed_time": 10309.029781341553,
+      "loss": 0.3147,
+      "grad_norm": 0.15712329745292664,
+      "learning_rate": 6.70440251572327e-05,
+      "epoch": 0.6675
+    },
+    {
+      "step": 1669,
+      "timestamp": "2025-12-28T11:50:06.585145",
+      "elapsed_time": 10320.870235919952,
+      "loss": 0.1362,
+      "grad_norm": 0.09576017409563065,
+      "learning_rate": 6.691823899371068e-05,
+      "epoch": 0.668125
+    },
+    {
+      "step": 1670,
+      "timestamp": "2025-12-28T11:50:12.699422",
+      "elapsed_time": 10326.984511852264,
+      "loss": 0.3545,
+      "grad_norm": 0.17530383169651031,
+      "learning_rate": 6.679245283018868e-05,
+      "epoch": 0.66875
+    },
+    {
+      "step": 1671,
+      "timestamp": "2025-12-28T11:50:20.519400",
+      "elapsed_time": 10334.804490327835,
+      "loss": 0.1705,
+      "grad_norm": 0.17078684270381927,
+      "learning_rate": 6.666666666666667e-05,
+      "epoch": 0.669375
+    },
+    {
+      "step": 1672,
+      "timestamp": "2025-12-28T11:50:27.666519",
+      "elapsed_time": 10341.951608896255,
+      "loss": 0.1515,
+      "grad_norm": 0.13205629587173462,
+      "learning_rate": 6.654088050314465e-05,
+      "epoch": 0.67
+    },
+    {
+      "step": 1673,
+      "timestamp": "2025-12-28T11:50:36.092046",
+      "elapsed_time": 10350.37713599205,
+      "loss": 0.1247,
+      "grad_norm": 0.1089896485209465,
+      "learning_rate": 6.641509433962264e-05,
+      "epoch": 0.670625
+    },
+    {
+      "step": 1674,
+      "timestamp": "2025-12-28T11:50:44.269347",
+      "elapsed_time": 10358.55443739891,
+      "loss": 0.2136,
+      "grad_norm": 0.2696709930896759,
+      "learning_rate": 6.628930817610063e-05,
+      "epoch": 0.67125
+    },
+    {
+      "step": 1675,
+      "timestamp": "2025-12-28T11:50:54.025458",
+      "elapsed_time": 10368.310548067093,
+      "loss": 0.0958,
+      "grad_norm": 0.09596393257379532,
+      "learning_rate": 6.616352201257861e-05,
+      "epoch": 0.671875
+    },
+    {
+      "step": 1676,
+      "timestamp": "2025-12-28T11:51:02.324646",
+      "elapsed_time": 10376.609736204147,
+      "loss": 0.3834,
+      "grad_norm": 0.1570800244808197,
+      "learning_rate": 6.60377358490566e-05,
+      "epoch": 0.6725
+    },
+    {
+      "step": 1677,
+      "timestamp": "2025-12-28T11:51:08.949728",
+      "elapsed_time": 10383.234818220139,
+      "loss": 0.2527,
+      "grad_norm": 0.17659151554107666,
+      "learning_rate": 6.591194968553458e-05,
+      "epoch": 0.673125
+    },
+    {
+      "step": 1678,
+      "timestamp": "2025-12-28T11:51:16.808478",
+      "elapsed_time": 10391.093568086624,
+      "loss": 0.1488,
+      "grad_norm": 0.13421952724456787,
+      "learning_rate": 6.578616352201258e-05,
+      "epoch": 0.67375
+    },
+    {
+      "step": 1679,
+      "timestamp": "2025-12-28T11:51:23.951873",
+      "elapsed_time": 10398.236963272095,
+      "loss": 0.1324,
+      "grad_norm": 0.10407258570194244,
+      "learning_rate": 6.566037735849057e-05,
+      "epoch": 0.674375
+    },
+    {
+      "step": 1680,
+      "timestamp": "2025-12-28T11:51:44.838057",
+      "elapsed_time": 10419.123147726059,
+      "loss": 0.111,
+      "grad_norm": 0.08704216778278351,
+      "learning_rate": 6.553459119496855e-05,
+      "epoch": 0.675
+    },
+    {
+      "step": 1681,
+      "timestamp": "2025-12-28T11:51:54.548567",
+      "elapsed_time": 10428.83365702629,
+      "loss": 0.149,
+      "grad_norm": 0.11608418822288513,
+      "learning_rate": 6.540880503144654e-05,
+      "epoch": 0.675625
+    },
+    {
+      "step": 1682,
+      "timestamp": "2025-12-28T11:52:03.641934",
+      "elapsed_time": 10437.927023887634,
+      "loss": 0.2173,
+      "grad_norm": 0.1506926417350769,
+      "learning_rate": 6.528301886792453e-05,
+      "epoch": 0.67625
+    },
+    {
+      "step": 1683,
+      "timestamp": "2025-12-28T11:52:12.314341",
+      "elapsed_time": 10446.599431276321,
+      "loss": 0.192,
+      "grad_norm": 0.17497746646404266,
+      "learning_rate": 6.515723270440251e-05,
+      "epoch": 0.676875
+    },
+    {
+      "step": 1684,
+      "timestamp": "2025-12-28T11:52:24.992727",
+      "elapsed_time": 10459.277817249298,
+      "loss": 0.1565,
+      "grad_norm": 0.09996242821216583,
+      "learning_rate": 6.50314465408805e-05,
+      "epoch": 0.6775
+    },
+    {
+      "step": 1685,
+      "timestamp": "2025-12-28T11:52:35.596215",
+      "elapsed_time": 10469.881305456161,
+      "loss": 0.1227,
+      "grad_norm": 0.10592522472143173,
+      "learning_rate": 6.490566037735849e-05,
+      "epoch": 0.678125
+    },
+    {
+      "step": 1686,
+      "timestamp": "2025-12-28T11:52:43.075463",
+      "elapsed_time": 10477.360553264618,
+      "loss": 0.1698,
+      "grad_norm": 0.1294756829738617,
+      "learning_rate": 6.477987421383648e-05,
+      "epoch": 0.67875
+    },
+    {
+      "step": 1687,
+      "timestamp": "2025-12-28T11:52:52.975805",
+      "elapsed_time": 10487.260895729065,
+      "loss": 0.155,
+      "grad_norm": 0.12220288068056107,
+      "learning_rate": 6.465408805031447e-05,
+      "epoch": 0.679375
+    },
+    {
+      "step": 1688,
+      "timestamp": "2025-12-28T11:53:00.065512",
+      "elapsed_time": 10494.350602388382,
+      "loss": 0.3612,
+      "grad_norm": 0.18684455752372742,
+      "learning_rate": 6.452830188679245e-05,
+      "epoch": 0.68
+    },
+    {
+      "step": 1689,
+      "timestamp": "2025-12-28T11:53:07.317682",
+      "elapsed_time": 10501.602772951126,
+      "loss": 0.1816,
+      "grad_norm": 0.13791623711585999,
+      "learning_rate": 6.440251572327044e-05,
+      "epoch": 0.680625
+    },
+    {
+      "step": 1690,
+      "timestamp": "2025-12-28T11:53:17.843498",
+      "elapsed_time": 10512.128588676453,
+      "loss": 0.1549,
+      "grad_norm": 0.11356908082962036,
+      "learning_rate": 6.427672955974842e-05,
+      "epoch": 0.68125
+    },
+    {
+      "step": 1691,
+      "timestamp": "2025-12-28T11:53:38.715052",
+      "elapsed_time": 10533.00014257431,
+      "loss": 0.0906,
+      "grad_norm": 0.07176671177148819,
+      "learning_rate": 6.415094339622641e-05,
+      "epoch": 0.681875
+    },
+    {
+      "step": 1692,
+      "timestamp": "2025-12-28T11:53:46.531119",
+      "elapsed_time": 10540.816209077835,
+      "loss": 0.1907,
+      "grad_norm": 0.141318216919899,
+      "learning_rate": 6.40251572327044e-05,
+      "epoch": 0.6825
+    },
+    {
+      "step": 1693,
+      "timestamp": "2025-12-28T11:54:00.485420",
+      "elapsed_time": 10554.770510435104,
+      "loss": 0.1189,
+      "grad_norm": 0.08776037395000458,
+      "learning_rate": 6.389937106918239e-05,
+      "epoch": 0.683125
+    },
+    {
+      "step": 1694,
+      "timestamp": "2025-12-28T11:54:17.313372",
+      "elapsed_time": 10571.598462820053,
+      "loss": 0.1334,
+      "grad_norm": 0.09080228954553604,
+      "learning_rate": 6.377358490566038e-05,
+      "epoch": 0.68375
+    },
+    {
+      "step": 1695,
+      "timestamp": "2025-12-28T11:54:35.244713",
+      "elapsed_time": 10589.529803276062,
+      "loss": 0.117,
+      "grad_norm": 0.09272570163011551,
+      "learning_rate": 6.364779874213837e-05,
+      "epoch": 0.684375
+    },
+    {
+      "step": 1696,
+      "timestamp": "2025-12-28T11:54:42.409774",
+      "elapsed_time": 10596.69486451149,
+      "loss": 0.1424,
+      "grad_norm": 0.21328891813755035,
+      "learning_rate": 6.352201257861635e-05,
+      "epoch": 0.685
+    },
+    {
+      "step": 1697,
+      "timestamp": "2025-12-28T11:54:47.771967",
+      "elapsed_time": 10602.057057142258,
+      "loss": 0.1881,
+      "grad_norm": 0.14298084378242493,
+      "learning_rate": 6.339622641509434e-05,
+      "epoch": 0.685625
+    },
+    {
+      "step": 1698,
+      "timestamp": "2025-12-28T11:54:54.653860",
+      "elapsed_time": 10608.938950777054,
+      "loss": 0.2189,
+      "grad_norm": 0.1579611450433731,
+      "learning_rate": 6.327044025157232e-05,
+      "epoch": 0.68625
+    },
+    {
+      "step": 1699,
+      "timestamp": "2025-12-28T11:55:08.547726",
+      "elapsed_time": 10622.832817077637,
+      "loss": 0.1117,
+      "grad_norm": 0.13192813098430634,
+      "learning_rate": 6.314465408805031e-05,
+      "epoch": 0.686875
+    },
+    {
+      "step": 1700,
+      "timestamp": "2025-12-28T11:55:14.487705",
+      "elapsed_time": 10628.77279496193,
+      "loss": 0.1748,
+      "grad_norm": 0.13700881600379944,
+      "learning_rate": 6.301886792452831e-05,
+      "epoch": 0.6875
+    },
+    {
+      "step": 1701,
+      "timestamp": "2025-12-28T11:55:22.517878",
+      "elapsed_time": 10636.802968740463,
+      "loss": 0.1818,
+      "grad_norm": 0.13351568579673767,
+      "learning_rate": 6.289308176100629e-05,
+      "epoch": 0.688125
+    },
+    {
+      "step": 1702,
+      "timestamp": "2025-12-28T11:55:32.797402",
+      "elapsed_time": 10647.082492351532,
+      "loss": 0.1418,
+      "grad_norm": 0.11062666773796082,
+      "learning_rate": 6.276729559748428e-05,
+      "epoch": 0.68875
+    },
+    {
+      "step": 1703,
+      "timestamp": "2025-12-28T11:55:41.627529",
+      "elapsed_time": 10655.912618637085,
+      "loss": 0.138,
+      "grad_norm": 0.12265567481517792,
+      "learning_rate": 6.264150943396226e-05,
+      "epoch": 0.689375
+    },
+    {
+      "step": 1704,
+      "timestamp": "2025-12-28T11:55:48.617990",
+      "elapsed_time": 10662.903080940247,
+      "loss": 0.231,
+      "grad_norm": 0.14810243248939514,
+      "learning_rate": 6.251572327044025e-05,
+      "epoch": 0.69
+    },
+    {
+      "step": 1705,
+      "timestamp": "2025-12-28T11:55:56.917497",
+      "elapsed_time": 10671.202587604523,
+      "loss": 0.1909,
+      "grad_norm": 0.1277126967906952,
+      "learning_rate": 6.238993710691824e-05,
+      "epoch": 0.690625
+    },
+    {
+      "step": 1706,
+      "timestamp": "2025-12-28T11:56:11.042753",
+      "elapsed_time": 10685.327842950821,
+      "loss": 0.1558,
+      "grad_norm": 0.10240671783685684,
+      "learning_rate": 6.226415094339622e-05,
+      "epoch": 0.69125
+    },
+    {
+      "step": 1707,
+      "timestamp": "2025-12-28T11:56:19.175090",
+      "elapsed_time": 10693.460180997849,
+      "loss": 0.1653,
+      "grad_norm": 0.13144554197788239,
+      "learning_rate": 6.213836477987421e-05,
+      "epoch": 0.691875
+    },
+    {
+      "step": 1708,
+      "timestamp": "2025-12-28T11:56:28.590060",
+      "elapsed_time": 10702.875150203705,
+      "loss": 0.4414,
+      "grad_norm": 0.15143659710884094,
+      "learning_rate": 6.201257861635221e-05,
+      "epoch": 0.6925
+    },
+    {
+      "step": 1709,
+      "timestamp": "2025-12-28T11:56:35.593767",
+      "elapsed_time": 10709.87885761261,
+      "loss": 0.1623,
+      "grad_norm": 0.1259068101644516,
+      "learning_rate": 6.188679245283019e-05,
+      "epoch": 0.693125
+    },
+    {
+      "step": 1710,
+      "timestamp": "2025-12-28T11:56:48.621945",
+      "elapsed_time": 10722.907035589218,
+      "loss": 0.1455,
+      "grad_norm": 0.09466191381216049,
+      "learning_rate": 6.176100628930818e-05,
+      "epoch": 0.69375
+    },
+    {
+      "step": 1711,
+      "timestamp": "2025-12-28T11:57:02.125457",
+      "elapsed_time": 10736.41054725647,
+      "loss": 0.1517,
+      "grad_norm": 0.11332697421312332,
+      "learning_rate": 6.163522012578616e-05,
+      "epoch": 0.694375
+    },
+    {
+      "step": 1712,
+      "timestamp": "2025-12-28T11:57:17.576247",
+      "elapsed_time": 10751.861337184906,
+      "loss": 0.1169,
+      "grad_norm": 0.08637837320566177,
+      "learning_rate": 6.150943396226415e-05,
+      "epoch": 0.695
+    },
+    {
+      "step": 1713,
+      "timestamp": "2025-12-28T11:57:27.606626",
+      "elapsed_time": 10761.891716957092,
+      "loss": 0.2205,
+      "grad_norm": 0.220438614487648,
+      "learning_rate": 6.138364779874214e-05,
+      "epoch": 0.695625
+    },
+    {
+      "step": 1714,
+      "timestamp": "2025-12-28T11:57:44.152720",
+      "elapsed_time": 10778.437810659409,
+      "loss": 0.107,
+      "grad_norm": 0.07576917856931686,
+      "learning_rate": 6.125786163522012e-05,
+      "epoch": 0.69625
+    },
+    {
+      "step": 1715,
+      "timestamp": "2025-12-28T11:58:02.072391",
+      "elapsed_time": 10796.357481956482,
+      "loss": 0.1151,
+      "grad_norm": 0.08082325756549835,
+      "learning_rate": 6.113207547169812e-05,
+      "epoch": 0.696875
+    },
+    {
+      "step": 1716,
+      "timestamp": "2025-12-28T11:58:06.544039",
+      "elapsed_time": 10800.829129457474,
+      "loss": 0.253,
+      "grad_norm": 0.17654751241207123,
+      "learning_rate": 6.10062893081761e-05,
+      "epoch": 0.6975
+    },
+    {
+      "step": 1717,
+      "timestamp": "2025-12-28T11:58:12.629577",
+      "elapsed_time": 10806.914667367935,
+      "loss": 0.225,
+      "grad_norm": 0.15153177082538605,
+      "learning_rate": 6.088050314465409e-05,
+      "epoch": 0.698125
+    },
+    {
+      "step": 1718,
+      "timestamp": "2025-12-28T11:58:20.697858",
+      "elapsed_time": 10814.982948064804,
+      "loss": 0.1918,
+      "grad_norm": 0.14196348190307617,
+      "learning_rate": 6.075471698113207e-05,
+      "epoch": 0.69875
+    },
+    {
+      "step": 1719,
+      "timestamp": "2025-12-28T11:58:35.776435",
+      "elapsed_time": 10830.061525821686,
+      "loss": 0.1034,
+      "grad_norm": 0.08372589945793152,
+      "learning_rate": 6.0628930817610065e-05,
+      "epoch": 0.699375
+    },
+    {
+      "step": 1720,
+      "timestamp": "2025-12-28T11:58:41.937607",
+      "elapsed_time": 10836.222697019577,
+      "loss": 0.2077,
+      "grad_norm": 0.15489476919174194,
+      "learning_rate": 6.050314465408805e-05,
+      "epoch": 0.7
+    },
+    {
+      "step": 1721,
+      "timestamp": "2025-12-28T11:58:49.416543",
+      "elapsed_time": 10843.70163345337,
+      "loss": 0.1679,
+      "grad_norm": 0.13079725205898285,
+      "learning_rate": 6.037735849056604e-05,
+      "epoch": 0.700625
+    },
+    {
+      "step": 1722,
+      "timestamp": "2025-12-28T11:58:57.300619",
+      "elapsed_time": 10851.58570933342,
+      "loss": 0.232,
+      "grad_norm": 0.1388118714094162,
+      "learning_rate": 6.025157232704402e-05,
+      "epoch": 0.70125
+    },
+    {
+      "step": 1723,
+      "timestamp": "2025-12-28T11:59:01.515497",
+      "elapsed_time": 10855.800586938858,
+      "loss": 0.3026,
+      "grad_norm": 0.21141380071640015,
+      "learning_rate": 6.0125786163522016e-05,
+      "epoch": 0.701875
+    },
+    {
+      "step": 1724,
+      "timestamp": "2025-12-28T11:59:09.331565",
+      "elapsed_time": 10863.61665558815,
+      "loss": 0.2376,
+      "grad_norm": 0.14707203209400177,
+      "learning_rate": 6e-05,
+      "epoch": 0.7025
+    },
+    {
+      "step": 1725,
+      "timestamp": "2025-12-28T11:59:15.412136",
+      "elapsed_time": 10869.697226524353,
+      "loss": 0.2456,
+      "grad_norm": 0.16581028699874878,
+      "learning_rate": 5.987421383647799e-05,
+      "epoch": 0.703125
+    },
+    {
+      "step": 1726,
+      "timestamp": "2025-12-28T11:59:22.041531",
+      "elapsed_time": 10876.326621294022,
+      "loss": 0.2671,
+      "grad_norm": 0.15748703479766846,
+      "learning_rate": 5.974842767295597e-05,
+      "epoch": 0.70375
+    },
+    {
+      "step": 1727,
+      "timestamp": "2025-12-28T11:59:36.715440",
+      "elapsed_time": 10891.00053024292,
+      "loss": 0.0921,
+      "grad_norm": 0.083674855530262,
+      "learning_rate": 5.9622641509433966e-05,
+      "epoch": 0.704375
+    },
+    {
+      "step": 1728,
+      "timestamp": "2025-12-28T11:59:51.882192",
+      "elapsed_time": 10906.167282104492,
+      "loss": 0.126,
+      "grad_norm": 0.0893712192773819,
+      "learning_rate": 5.949685534591195e-05,
+      "epoch": 0.705
+    },
+    {
+      "step": 1729,
+      "timestamp": "2025-12-28T11:59:59.217349",
+      "elapsed_time": 10913.502439022064,
+      "loss": 0.1477,
+      "grad_norm": 0.15378941595554352,
+      "learning_rate": 5.937106918238994e-05,
+      "epoch": 0.705625
+    },
+    {
+      "step": 1730,
+      "timestamp": "2025-12-28T12:00:07.029866",
+      "elapsed_time": 10921.31495642662,
+      "loss": 0.1585,
+      "grad_norm": 0.5633273124694824,
+      "learning_rate": 5.9245283018867923e-05,
+      "epoch": 0.70625
+    },
+    {
+      "step": 1731,
+      "timestamp": "2025-12-28T12:00:15.817213",
+      "elapsed_time": 10930.102303981781,
+      "loss": 0.141,
+      "grad_norm": 0.11738086491823196,
+      "learning_rate": 5.9119496855345916e-05,
+      "epoch": 0.706875
+    },
+    {
+      "step": 1732,
+      "timestamp": "2025-12-28T12:00:31.339033",
+      "elapsed_time": 10945.62412405014,
+      "loss": 0.1483,
+      "grad_norm": 0.09447266906499863,
+      "learning_rate": 5.89937106918239e-05,
+      "epoch": 0.7075
+    },
+    {
+      "step": 1733,
+      "timestamp": "2025-12-28T12:00:45.283079",
+      "elapsed_time": 10959.568170070648,
+      "loss": 0.1092,
+      "grad_norm": 0.09043192118406296,
+      "learning_rate": 5.886792452830189e-05,
+      "epoch": 0.708125
+    },
+    {
+      "step": 1734,
+      "timestamp": "2025-12-28T12:00:54.651343",
+      "elapsed_time": 10968.936433792114,
+      "loss": 0.2226,
+      "grad_norm": 0.1368798464536667,
+      "learning_rate": 5.8742138364779874e-05,
+      "epoch": 0.70875
+    },
+    {
+      "step": 1735,
+      "timestamp": "2025-12-28T12:00:59.476546",
+      "elapsed_time": 10973.76163649559,
+      "loss": 0.23,
+      "grad_norm": 0.18571457266807556,
+      "learning_rate": 5.861635220125786e-05,
+      "epoch": 0.709375
+    },
+    {
+      "step": 1736,
+      "timestamp": "2025-12-28T12:01:18.837352",
+      "elapsed_time": 10993.12244272232,
+      "loss": 0.132,
+      "grad_norm": 0.07873135060071945,
+      "learning_rate": 5.849056603773585e-05,
+      "epoch": 0.71
+    },
+    {
+      "step": 1737,
+      "timestamp": "2025-12-28T12:01:36.420927",
+      "elapsed_time": 11010.70601773262,
+      "loss": 0.0795,
+      "grad_norm": 0.06837231665849686,
+      "learning_rate": 5.836477987421384e-05,
+      "epoch": 0.710625
+    },
+    {
+      "step": 1738,
+      "timestamp": "2025-12-28T12:01:44.490699",
+      "elapsed_time": 11018.77578997612,
+      "loss": 0.1778,
+      "grad_norm": 0.13413724303245544,
+      "learning_rate": 5.8238993710691824e-05,
+      "epoch": 0.71125
+    },
+    {
+      "step": 1739,
+      "timestamp": "2025-12-28T12:02:00.532185",
+      "elapsed_time": 11034.81727552414,
+      "loss": 0.1339,
+      "grad_norm": 0.092499740421772,
+      "learning_rate": 5.811320754716981e-05,
+      "epoch": 0.711875
+    },
+    {
+      "step": 1740,
+      "timestamp": "2025-12-28T12:02:08.705472",
+      "elapsed_time": 11042.990562677383,
+      "loss": 0.1989,
+      "grad_norm": 0.14828531444072723,
+      "learning_rate": 5.79874213836478e-05,
+      "epoch": 0.7125
+    },
+    {
+      "step": 1741,
+      "timestamp": "2025-12-28T12:02:21.495006",
+      "elapsed_time": 11055.780095815659,
+      "loss": 0.1429,
+      "grad_norm": 0.10832472145557404,
+      "learning_rate": 5.786163522012579e-05,
+      "epoch": 0.713125
+    },
+    {
+      "step": 1742,
+      "timestamp": "2025-12-28T12:02:27.508870",
+      "elapsed_time": 11061.793960094452,
+      "loss": 0.1582,
+      "grad_norm": 0.1331390142440796,
+      "learning_rate": 5.7735849056603774e-05,
+      "epoch": 0.71375
+    },
+    {
+      "step": 1743,
+      "timestamp": "2025-12-28T12:02:32.397381",
+      "elapsed_time": 11066.682476043701,
+      "loss": 0.3914,
+      "grad_norm": 0.20057356357574463,
+      "learning_rate": 5.761006289308176e-05,
+      "epoch": 0.714375
+    },
+    {
+      "step": 1744,
+      "timestamp": "2025-12-28T12:02:41.828268",
+      "elapsed_time": 11076.113358259201,
+      "loss": 0.1488,
+      "grad_norm": 0.1079302653670311,
+      "learning_rate": 5.748427672955975e-05,
+      "epoch": 0.715
+    },
+    {
+      "step": 1745,
+      "timestamp": "2025-12-28T12:02:48.746835",
+      "elapsed_time": 11083.03192615509,
+      "loss": 0.2745,
+      "grad_norm": 0.15143828094005585,
+      "learning_rate": 5.735849056603774e-05,
+      "epoch": 0.715625
+    },
+    {
+      "step": 1746,
+      "timestamp": "2025-12-28T12:02:54.122330",
+      "elapsed_time": 11088.407420396805,
+      "loss": 0.432,
+      "grad_norm": 0.1899183988571167,
+      "learning_rate": 5.7232704402515724e-05,
+      "epoch": 0.71625
+    },
+    {
+      "step": 1747,
+      "timestamp": "2025-12-28T12:03:04.601026",
+      "elapsed_time": 11098.88611626625,
+      "loss": 0.1694,
+      "grad_norm": 0.1249026283621788,
+      "learning_rate": 5.710691823899371e-05,
+      "epoch": 0.716875
+    },
+    {
+      "step": 1748,
+      "timestamp": "2025-12-28T12:03:24.883654",
+      "elapsed_time": 11119.168744325638,
+      "loss": 0.1206,
+      "grad_norm": 0.07808022201061249,
+      "learning_rate": 5.6981132075471696e-05,
+      "epoch": 0.7175
+    },
+    {
+      "step": 1749,
+      "timestamp": "2025-12-28T12:03:34.253119",
+      "elapsed_time": 11128.538208961487,
+      "loss": 0.1954,
+      "grad_norm": 0.12799711525440216,
+      "learning_rate": 5.685534591194969e-05,
+      "epoch": 0.718125
+    },
+    {
+      "step": 1750,
+      "timestamp": "2025-12-28T12:03:41.136923",
+      "elapsed_time": 11135.422013521194,
+      "loss": 0.1752,
+      "grad_norm": 0.15330050885677338,
+      "learning_rate": 5.6729559748427674e-05,
+      "epoch": 0.71875
+    },
+    {
+      "step": 1751,
+      "timestamp": "2025-12-28T12:03:58.075443",
+      "elapsed_time": 11152.360533714294,
+      "loss": 0.1457,
+      "grad_norm": 0.11428968608379364,
+      "learning_rate": 5.660377358490566e-05,
+      "epoch": 0.719375
+    },
+    {
+      "step": 1752,
+      "timestamp": "2025-12-28T12:04:06.337992",
+      "elapsed_time": 11160.623083114624,
+      "loss": 0.14,
+      "grad_norm": 0.1099119707942009,
+      "learning_rate": 5.6477987421383646e-05,
+      "epoch": 0.72
+    },
+    {
+      "step": 1753,
+      "timestamp": "2025-12-28T12:04:27.219359",
+      "elapsed_time": 11181.504449605942,
+      "loss": 0.192,
+      "grad_norm": 0.08993933349847794,
+      "learning_rate": 5.635220125786164e-05,
+      "epoch": 0.720625
+    },
+    {
+      "step": 1754,
+      "timestamp": "2025-12-28T12:04:32.971069",
+      "elapsed_time": 11187.256159305573,
+      "loss": 0.2048,
+      "grad_norm": 0.1937875747680664,
+      "learning_rate": 5.6226415094339625e-05,
+      "epoch": 0.72125
+    },
+    {
+      "step": 1755,
+      "timestamp": "2025-12-28T12:04:50.508927",
+      "elapsed_time": 11204.79401755333,
+      "loss": 0.1391,
+      "grad_norm": 0.10435943305492401,
+      "learning_rate": 5.610062893081761e-05,
+      "epoch": 0.721875
+    },
+    {
+      "step": 1756,
+      "timestamp": "2025-12-28T12:05:03.301495",
+      "elapsed_time": 11217.586585998535,
+      "loss": 0.1312,
+      "grad_norm": 0.10173836350440979,
+      "learning_rate": 5.5974842767295596e-05,
+      "epoch": 0.7225
+    },
+    {
+      "step": 1757,
+      "timestamp": "2025-12-28T12:05:09.379646",
+      "elapsed_time": 11223.66473698616,
+      "loss": 0.1685,
+      "grad_norm": 0.2740840017795563,
+      "learning_rate": 5.584905660377359e-05,
+      "epoch": 0.723125
+    },
+    {
+      "step": 1758,
+      "timestamp": "2025-12-28T12:05:16.331503",
+      "elapsed_time": 11230.616593122482,
+      "loss": 0.1821,
+      "grad_norm": 0.13564808666706085,
+      "learning_rate": 5.5723270440251575e-05,
+      "epoch": 0.72375
+    },
+    {
+      "step": 1759,
+      "timestamp": "2025-12-28T12:05:25.908391",
+      "elapsed_time": 11240.193481445312,
+      "loss": 0.1539,
+      "grad_norm": 0.11152060329914093,
+      "learning_rate": 5.559748427672956e-05,
+      "epoch": 0.724375
+    },
+    {
+      "step": 1760,
+      "timestamp": "2025-12-28T12:05:32.102932",
+      "elapsed_time": 11246.388021945953,
+      "loss": 0.2534,
+      "grad_norm": 0.35456421971321106,
+      "learning_rate": 5.5471698113207547e-05,
+      "epoch": 0.725
+    },
+    {
+      "step": 1761,
+      "timestamp": "2025-12-28T12:05:38.035109",
+      "elapsed_time": 11252.320199251175,
+      "loss": 0.2746,
+      "grad_norm": 0.20745328068733215,
+      "learning_rate": 5.534591194968554e-05,
+      "epoch": 0.725625
+    },
+    {
+      "step": 1762,
+      "timestamp": "2025-12-28T12:05:45.470001",
+      "elapsed_time": 11259.755092144012,
+      "loss": 0.1958,
+      "grad_norm": 0.1335449367761612,
+      "learning_rate": 5.5220125786163525e-05,
+      "epoch": 0.72625
+    },
+    {
+      "step": 1763,
+      "timestamp": "2025-12-28T12:05:51.035573",
+      "elapsed_time": 11265.320663452148,
+      "loss": 0.2487,
+      "grad_norm": 0.17504768073558807,
+      "learning_rate": 5.509433962264151e-05,
+      "epoch": 0.726875
+    },
+    {
+      "step": 1764,
+      "timestamp": "2025-12-28T12:06:07.328667",
+      "elapsed_time": 11281.613757371902,
+      "loss": 0.1174,
+      "grad_norm": 0.15383565425872803,
+      "learning_rate": 5.49685534591195e-05,
+      "epoch": 0.7275
+    },
+    {
+      "step": 1765,
+      "timestamp": "2025-12-28T12:06:13.760643",
+      "elapsed_time": 11288.045733213425,
+      "loss": 0.1973,
+      "grad_norm": 0.31960293650627136,
+      "learning_rate": 5.484276729559748e-05,
+      "epoch": 0.728125
+    },
+    {
+      "step": 1766,
+      "timestamp": "2025-12-28T12:06:20.069410",
+      "elapsed_time": 11294.354505062103,
+      "loss": 0.2468,
+      "grad_norm": 0.16368219256401062,
+      "learning_rate": 5.4716981132075475e-05,
+      "epoch": 0.72875
+    },
+    {
+      "step": 1767,
+      "timestamp": "2025-12-28T12:06:32.245904",
+      "elapsed_time": 11306.530994415283,
+      "loss": 0.0958,
+      "grad_norm": 0.08179421722888947,
+      "learning_rate": 5.459119496855346e-05,
+      "epoch": 0.729375
+    },
+    {
+      "step": 1768,
+      "timestamp": "2025-12-28T12:06:42.619417",
+      "elapsed_time": 11316.904507875443,
+      "loss": 0.1315,
+      "grad_norm": 0.10045388340950012,
+      "learning_rate": 5.446540880503145e-05,
+      "epoch": 0.73
+    },
+    {
+      "step": 1769,
+      "timestamp": "2025-12-28T12:06:51.947947",
+      "elapsed_time": 11326.233037233353,
+      "loss": 0.211,
+      "grad_norm": 0.13236618041992188,
+      "learning_rate": 5.433962264150943e-05,
+      "epoch": 0.730625
+    },
+    {
+      "step": 1770,
+      "timestamp": "2025-12-28T12:07:03.166704",
+      "elapsed_time": 11337.451794624329,
+      "loss": 0.1187,
+      "grad_norm": 0.08745797723531723,
+      "learning_rate": 5.4213836477987425e-05,
+      "epoch": 0.73125
+    },
+    {
+      "step": 1771,
+      "timestamp": "2025-12-28T12:07:08.268446",
+      "elapsed_time": 11342.553536176682,
+      "loss": 0.261,
+      "grad_norm": 0.1820874959230423,
+      "learning_rate": 5.408805031446541e-05,
+      "epoch": 0.731875
+    },
+    {
+      "step": 1772,
+      "timestamp": "2025-12-28T12:07:18.742417",
+      "elapsed_time": 11353.027507543564,
+      "loss": 0.1424,
+      "grad_norm": 0.10937105119228363,
+      "learning_rate": 5.39622641509434e-05,
+      "epoch": 0.7325
+    },
+    {
+      "step": 1773,
+      "timestamp": "2025-12-28T12:07:35.988111",
+      "elapsed_time": 11370.273201227188,
+      "loss": 0.131,
+      "grad_norm": 0.09035047143697739,
+      "learning_rate": 5.383647798742138e-05,
+      "epoch": 0.733125
+    },
+    {
+      "step": 1774,
+      "timestamp": "2025-12-28T12:07:42.368541",
+      "elapsed_time": 11376.653632164001,
+      "loss": 0.2366,
+      "grad_norm": 0.1836661696434021,
+      "learning_rate": 5.3710691823899376e-05,
+      "epoch": 0.73375
+    },
+    {
+      "step": 1775,
+      "timestamp": "2025-12-28T12:07:51.102768",
+      "elapsed_time": 11385.387857913971,
+      "loss": 0.1211,
+      "grad_norm": 0.10112607479095459,
+      "learning_rate": 5.358490566037736e-05,
+      "epoch": 0.734375
+    },
+    {
+      "step": 1776,
+      "timestamp": "2025-12-28T12:08:07.843256",
+      "elapsed_time": 11402.128346443176,
+      "loss": 0.1377,
+      "grad_norm": 0.08925390243530273,
+      "learning_rate": 5.345911949685535e-05,
+      "epoch": 0.735
+    },
+    {
+      "step": 1777,
+      "timestamp": "2025-12-28T12:08:15.286123",
+      "elapsed_time": 11409.571213960648,
+      "loss": 0.1673,
+      "grad_norm": 0.13005472719669342,
+      "learning_rate": 5.333333333333333e-05,
+      "epoch": 0.735625
+    },
+    {
+      "step": 1778,
+      "timestamp": "2025-12-28T12:08:26.521285",
+      "elapsed_time": 11420.806375265121,
+      "loss": 0.2347,
+      "grad_norm": 0.11882436275482178,
+      "learning_rate": 5.3207547169811326e-05,
+      "epoch": 0.73625
+    },
+    {
+      "step": 1779,
+      "timestamp": "2025-12-28T12:08:33.112874",
+      "elapsed_time": 11427.397964477539,
+      "loss": 0.1772,
+      "grad_norm": 0.14445851743221283,
+      "learning_rate": 5.308176100628931e-05,
+      "epoch": 0.736875
+    },
+    {
+      "step": 1780,
+      "timestamp": "2025-12-28T12:08:43.367927",
+      "elapsed_time": 11437.653017282486,
+      "loss": 0.1138,
+      "grad_norm": 0.10857395082712173,
+      "learning_rate": 5.29559748427673e-05,
+      "epoch": 0.7375
+    },
+    {
+      "step": 1781,
+      "timestamp": "2025-12-28T12:08:50.894299",
+      "elapsed_time": 11445.179389238358,
+      "loss": 0.1847,
+      "grad_norm": 0.13043537735939026,
+      "learning_rate": 5.283018867924528e-05,
+      "epoch": 0.738125
+    },
+    {
+      "step": 1782,
+      "timestamp": "2025-12-28T12:08:59.518528",
+      "elapsed_time": 11453.803642749786,
+      "loss": 0.1546,
+      "grad_norm": 0.16742828488349915,
+      "learning_rate": 5.270440251572327e-05,
+      "epoch": 0.73875
+    },
+    {
+      "step": 1783,
+      "timestamp": "2025-12-28T12:09:07.688718",
+      "elapsed_time": 11461.973808765411,
+      "loss": 0.137,
+      "grad_norm": 0.1259349137544632,
+      "learning_rate": 5.257861635220126e-05,
+      "epoch": 0.739375
+    },
+    {
+      "step": 1784,
+      "timestamp": "2025-12-28T12:09:14.061669",
+      "elapsed_time": 11468.346759557724,
+      "loss": 0.2945,
+      "grad_norm": 0.1729360669851303,
+      "learning_rate": 5.245283018867925e-05,
+      "epoch": 0.74
+    },
+    {
+      "step": 1785,
+      "timestamp": "2025-12-28T12:09:24.669063",
+      "elapsed_time": 11478.95415353775,
+      "loss": 0.143,
+      "grad_norm": 0.10868491232395172,
+      "learning_rate": 5.2327044025157234e-05,
+      "epoch": 0.740625
+    },
+    {
+      "step": 1786,
+      "timestamp": "2025-12-28T12:09:29.943561",
+      "elapsed_time": 11484.228651285172,
+      "loss": 0.2229,
+      "grad_norm": 0.1582539975643158,
+      "learning_rate": 5.220125786163522e-05,
+      "epoch": 0.74125
+    },
+    {
+      "step": 1787,
+      "timestamp": "2025-12-28T12:09:44.863706",
+      "elapsed_time": 11499.14879655838,
+      "loss": 0.1276,
+      "grad_norm": 0.09422967582941055,
+      "learning_rate": 5.207547169811321e-05,
+      "epoch": 0.741875
+    },
+    {
+      "step": 1788,
+      "timestamp": "2025-12-28T12:09:52.629311",
+      "elapsed_time": 11506.91440153122,
+      "loss": 0.3331,
+      "grad_norm": 0.15312416851520538,
+      "learning_rate": 5.19496855345912e-05,
+      "epoch": 0.7425
+    },
+    {
+      "step": 1789,
+      "timestamp": "2025-12-28T12:10:00.979750",
+      "elapsed_time": 11515.264840602875,
+      "loss": 0.1471,
+      "grad_norm": 0.14243608713150024,
+      "learning_rate": 5.1823899371069184e-05,
+      "epoch": 0.743125
+    },
+    {
+      "step": 1790,
+      "timestamp": "2025-12-28T12:10:21.855558",
+      "elapsed_time": 11536.140648126602,
+      "loss": 0.0733,
+      "grad_norm": 0.06610149145126343,
+      "learning_rate": 5.169811320754717e-05,
+      "epoch": 0.74375
+    },
+    {
+      "step": 1791,
+      "timestamp": "2025-12-28T12:10:37.078731",
+      "elapsed_time": 11551.363821744919,
+      "loss": 0.0946,
+      "grad_norm": 0.07790885865688324,
+      "learning_rate": 5.157232704402516e-05,
+      "epoch": 0.744375
+    },
+    {
+      "step": 1792,
+      "timestamp": "2025-12-28T12:10:42.615191",
+      "elapsed_time": 11556.90028142929,
+      "loss": 0.2814,
+      "grad_norm": 0.21737056970596313,
+      "learning_rate": 5.144654088050315e-05,
+      "epoch": 0.745
+    },
+    {
+      "step": 1793,
+      "timestamp": "2025-12-28T12:10:51.083715",
+      "elapsed_time": 11565.36880493164,
+      "loss": 0.1001,
+      "grad_norm": 0.10370247066020966,
+      "learning_rate": 5.1320754716981134e-05,
+      "epoch": 0.745625
+    },
+    {
+      "step": 1794,
+      "timestamp": "2025-12-28T12:11:01.500012",
+      "elapsed_time": 11575.785102844238,
+      "loss": 0.158,
+      "grad_norm": 0.10645274817943573,
+      "learning_rate": 5.119496855345912e-05,
+      "epoch": 0.74625
+    },
+    {
+      "step": 1795,
+      "timestamp": "2025-12-28T12:11:15.700384",
+      "elapsed_time": 11589.985474824905,
+      "loss": 0.1245,
+      "grad_norm": 0.09372659027576447,
+      "learning_rate": 5.106918238993711e-05,
+      "epoch": 0.746875
+    },
+    {
+      "step": 1796,
+      "timestamp": "2025-12-28T12:11:23.832555",
+      "elapsed_time": 11598.117645263672,
+      "loss": 0.1891,
+      "grad_norm": 0.2279994636774063,
+      "learning_rate": 5.09433962264151e-05,
+      "epoch": 0.7475
+    },
+    {
+      "step": 1797,
+      "timestamp": "2025-12-28T12:11:28.965281",
+      "elapsed_time": 11603.250371217728,
+      "loss": 0.242,
+      "grad_norm": 0.17063166201114655,
+      "learning_rate": 5.0817610062893084e-05,
+      "epoch": 0.748125
+    },
+    {
+      "step": 1798,
+      "timestamp": "2025-12-28T12:11:49.020815",
+      "elapsed_time": 11623.30590581894,
+      "loss": 0.1164,
+      "grad_norm": 0.07605966180562973,
+      "learning_rate": 5.069182389937107e-05,
+      "epoch": 0.74875
+    },
+    {
+      "step": 1799,
+      "timestamp": "2025-12-28T12:11:57.236498",
+      "elapsed_time": 11631.521588563919,
+      "loss": 0.145,
+      "grad_norm": 0.11617890745401382,
+      "learning_rate": 5.0566037735849056e-05,
+      "epoch": 0.749375
+    },
+    {
+      "step": 1800,
+      "timestamp": "2025-12-28T12:12:05.263297",
+      "elapsed_time": 11639.548387527466,
+      "loss": 0.1271,
+      "grad_norm": 0.12031132727861404,
+      "learning_rate": 5.044025157232705e-05,
+      "epoch": 0.75
+    },
+    {
+      "step": 1801,
+      "timestamp": "2025-12-28T12:12:13.459492",
+      "elapsed_time": 11647.744582653046,
+      "loss": 0.1386,
+      "grad_norm": 0.12550540268421173,
+      "learning_rate": 5.0314465408805034e-05,
+      "epoch": 0.750625
+    },
+    {
+      "step": 1802,
+      "timestamp": "2025-12-28T12:12:20.087338",
+      "elapsed_time": 11654.372427940369,
+      "loss": 0.2102,
+      "grad_norm": 0.14700697362422943,
+      "learning_rate": 5.018867924528302e-05,
+      "epoch": 0.75125
+    },
+    {
+      "step": 1803,
+      "timestamp": "2025-12-28T12:12:27.258226",
+      "elapsed_time": 11661.543317079544,
+      "loss": 0.1896,
+      "grad_norm": 0.1357898861169815,
+      "learning_rate": 5.0062893081761006e-05,
+      "epoch": 0.751875
+    },
+    {
+      "step": 1804,
+      "timestamp": "2025-12-28T12:12:36.128775",
+      "elapsed_time": 11670.413865566254,
+      "loss": 0.2121,
+      "grad_norm": 0.170795738697052,
+      "learning_rate": 4.9937106918239e-05,
+      "epoch": 0.7525
+    },
+    {
+      "step": 1805,
+      "timestamp": "2025-12-28T12:12:47.640738",
+      "elapsed_time": 11681.925828456879,
+      "loss": 0.1038,
+      "grad_norm": 0.08710363507270813,
+      "learning_rate": 4.9811320754716985e-05,
+      "epoch": 0.753125
+    },
+    {
+      "step": 1806,
+      "timestamp": "2025-12-28T12:12:59.703031",
+      "elapsed_time": 11693.988121509552,
+      "loss": 0.1155,
+      "grad_norm": 0.09086853265762329,
+      "learning_rate": 4.968553459119497e-05,
+      "epoch": 0.75375
+    },
+    {
+      "step": 1807,
+      "timestamp": "2025-12-28T12:13:07.973124",
+      "elapsed_time": 11702.258214712143,
+      "loss": 0.1217,
+      "grad_norm": 0.09986895322799683,
+      "learning_rate": 4.9559748427672956e-05,
+      "epoch": 0.754375
+    },
+    {
+      "step": 1808,
+      "timestamp": "2025-12-28T12:13:15.956985",
+      "elapsed_time": 11710.242075920105,
+      "loss": 0.2275,
+      "grad_norm": 0.2224908024072647,
+      "learning_rate": 4.943396226415095e-05,
+      "epoch": 0.755
+    },
+    {
+      "step": 1809,
+      "timestamp": "2025-12-28T12:13:26.842250",
+      "elapsed_time": 11721.127341032028,
+      "loss": 0.1621,
+      "grad_norm": 0.13384607434272766,
+      "learning_rate": 4.9308176100628935e-05,
+      "epoch": 0.755625
+    },
+    {
+      "step": 1810,
+      "timestamp": "2025-12-28T12:13:33.836759",
+      "elapsed_time": 11728.121849298477,
+      "loss": 0.1827,
+      "grad_norm": 0.12781156599521637,
+      "learning_rate": 4.918238993710692e-05,
+      "epoch": 0.75625
+    },
+    {
+      "step": 1811,
+      "timestamp": "2025-12-28T12:13:45.020366",
+      "elapsed_time": 11739.305456399918,
+      "loss": 0.1141,
+      "grad_norm": 0.08918755501508713,
+      "learning_rate": 4.9056603773584906e-05,
+      "epoch": 0.756875
+    },
+    {
+      "step": 1812,
+      "timestamp": "2025-12-28T12:13:53.188101",
+      "elapsed_time": 11747.473192214966,
+      "loss": 0.2376,
+      "grad_norm": 0.1457865685224533,
+      "learning_rate": 4.893081761006289e-05,
+      "epoch": 0.7575
+    },
+    {
+      "step": 1813,
+      "timestamp": "2025-12-28T12:14:02.152041",
+      "elapsed_time": 11756.437131166458,
+      "loss": 0.1533,
+      "grad_norm": 0.18399164080619812,
+      "learning_rate": 4.8805031446540885e-05,
+      "epoch": 0.758125
+    },
+    {
+      "step": 1814,
+      "timestamp": "2025-12-28T12:14:09.888429",
+      "elapsed_time": 11764.173519134521,
+      "loss": 0.2152,
+      "grad_norm": 0.14015917479991913,
+      "learning_rate": 4.867924528301887e-05,
+      "epoch": 0.75875
+    },
+    {
+      "step": 1815,
+      "timestamp": "2025-12-28T12:14:15.353774",
+      "elapsed_time": 11769.638864517212,
+      "loss": 0.1697,
+      "grad_norm": 0.14405082166194916,
+      "learning_rate": 4.855345911949686e-05,
+      "epoch": 0.759375
+    },
+    {
+      "step": 1816,
+      "timestamp": "2025-12-28T12:14:20.680001",
+      "elapsed_time": 11774.965090990067,
+      "loss": 0.3188,
+      "grad_norm": 0.23127539455890656,
+      "learning_rate": 4.842767295597484e-05,
+      "epoch": 0.76
+    },
+    {
+      "step": 1817,
+      "timestamp": "2025-12-28T12:14:26.352096",
+      "elapsed_time": 11780.637186527252,
+      "loss": 0.2125,
+      "grad_norm": 0.16115504503250122,
+      "learning_rate": 4.8301886792452835e-05,
+      "epoch": 0.760625
+    },
+    {
+      "step": 1818,
+      "timestamp": "2025-12-28T12:14:37.336389",
+      "elapsed_time": 11791.621479988098,
+      "loss": 0.3284,
+      "grad_norm": 0.14183737337589264,
+      "learning_rate": 4.817610062893082e-05,
+      "epoch": 0.76125
+    },
+    {
+      "step": 1819,
+      "timestamp": "2025-12-28T12:14:44.254425",
+      "elapsed_time": 11798.539515256882,
+      "loss": 0.2209,
+      "grad_norm": 0.14606201648712158,
+      "learning_rate": 4.805031446540881e-05,
+      "epoch": 0.761875
+    },
+    {
+      "step": 1820,
+      "timestamp": "2025-12-28T12:14:53.530877",
+      "elapsed_time": 11807.815967798233,
+      "loss": 0.1198,
+      "grad_norm": 0.18284346163272858,
+      "learning_rate": 4.792452830188679e-05,
+      "epoch": 0.7625
+    },
+    {
+      "step": 1821,
+      "timestamp": "2025-12-28T12:15:02.736188",
+      "elapsed_time": 11817.021278858185,
+      "loss": 0.1288,
+      "grad_norm": 0.1111404076218605,
+      "learning_rate": 4.7798742138364785e-05,
+      "epoch": 0.763125
+    },
+    {
+      "step": 1822,
+      "timestamp": "2025-12-28T12:15:10.672655",
+      "elapsed_time": 11824.957744836807,
+      "loss": 0.3206,
+      "grad_norm": 0.1543583869934082,
+      "learning_rate": 4.767295597484277e-05,
+      "epoch": 0.76375
+    },
+    {
+      "step": 1823,
+      "timestamp": "2025-12-28T12:15:24.962444",
+      "elapsed_time": 11839.247534513474,
+      "loss": 0.1522,
+      "grad_norm": 0.11686399579048157,
+      "learning_rate": 4.754716981132076e-05,
+      "epoch": 0.764375
+    },
+    {
+      "step": 1824,
+      "timestamp": "2025-12-28T12:15:32.700314",
+      "elapsed_time": 11846.985404253006,
+      "loss": 0.2664,
+      "grad_norm": 0.1384664922952652,
+      "learning_rate": 4.742138364779874e-05,
+      "epoch": 0.765
+    },
+    {
+      "step": 1825,
+      "timestamp": "2025-12-28T12:15:51.808174",
+      "elapsed_time": 11866.093264579773,
+      "loss": 0.2736,
+      "grad_norm": 0.11269756406545639,
+      "learning_rate": 4.7295597484276736e-05,
+      "epoch": 0.765625
+    },
+    {
+      "step": 1826,
+      "timestamp": "2025-12-28T12:15:57.370013",
+      "elapsed_time": 11871.655103206635,
+      "loss": 0.1664,
+      "grad_norm": 0.1405743509531021,
+      "learning_rate": 4.716981132075472e-05,
+      "epoch": 0.76625
+    },
+    {
+      "step": 1827,
+      "timestamp": "2025-12-28T12:16:10.153089",
+      "elapsed_time": 11884.43817949295,
+      "loss": 0.1049,
+      "grad_norm": 0.08988852053880692,
+      "learning_rate": 4.704402515723271e-05,
+      "epoch": 0.766875
+    },
+    {
+      "step": 1828,
+      "timestamp": "2025-12-28T12:16:18.898682",
+      "elapsed_time": 11893.183772325516,
+      "loss": 0.1607,
+      "grad_norm": 0.22954370081424713,
+      "learning_rate": 4.691823899371069e-05,
+      "epoch": 0.7675
+    },
+    {
+      "step": 1829,
+      "timestamp": "2025-12-28T12:16:26.036369",
+      "elapsed_time": 11900.321459770203,
+      "loss": 0.2229,
+      "grad_norm": 0.15082332491874695,
+      "learning_rate": 4.679245283018868e-05,
+      "epoch": 0.768125
+    },
+    {
+      "step": 1830,
+      "timestamp": "2025-12-28T12:16:35.324020",
+      "elapsed_time": 11909.609110355377,
+      "loss": 0.0898,
+      "grad_norm": 0.08692368119955063,
+      "learning_rate": 4.666666666666667e-05,
+      "epoch": 0.76875
+    },
+    {
+      "step": 1831,
+      "timestamp": "2025-12-28T12:16:41.705976",
+      "elapsed_time": 11915.991065740585,
+      "loss": 0.1725,
+      "grad_norm": 0.13606931269168854,
+      "learning_rate": 4.654088050314466e-05,
+      "epoch": 0.769375
+    },
+    {
+      "step": 1832,
+      "timestamp": "2025-12-28T12:16:53.044860",
+      "elapsed_time": 11927.32995057106,
+      "loss": 0.2069,
+      "grad_norm": 0.44243958592414856,
+      "learning_rate": 4.641509433962264e-05,
+      "epoch": 0.77
+    },
+    {
+      "step": 1833,
+      "timestamp": "2025-12-28T12:17:10.459005",
+      "elapsed_time": 11944.744095563889,
+      "loss": 0.1064,
+      "grad_norm": 0.07558456808328629,
+      "learning_rate": 4.628930817610063e-05,
+      "epoch": 0.770625
+    },
+    {
+      "step": 1834,
+      "timestamp": "2025-12-28T12:17:22.401241",
+      "elapsed_time": 11956.686330795288,
+      "loss": 0.0909,
+      "grad_norm": 0.09237058460712433,
+      "learning_rate": 4.616352201257862e-05,
+      "epoch": 0.77125
+    },
+    {
+      "step": 1835,
+      "timestamp": "2025-12-28T12:17:36.526898",
+      "elapsed_time": 11970.811988592148,
+      "loss": 0.1281,
+      "grad_norm": 0.09217442572116852,
+      "learning_rate": 4.603773584905661e-05,
+      "epoch": 0.771875
+    },
+    {
+      "step": 1836,
+      "timestamp": "2025-12-28T12:17:48.704766",
+      "elapsed_time": 11982.989856481552,
+      "loss": 0.1197,
+      "grad_norm": 0.09364216774702072,
+      "learning_rate": 4.5911949685534594e-05,
+      "epoch": 0.7725
+    },
+    {
+      "step": 1837,
+      "timestamp": "2025-12-28T12:17:54.175712",
+      "elapsed_time": 11988.460802555084,
+      "loss": 0.1996,
+      "grad_norm": 0.15755294263362885,
+      "learning_rate": 4.578616352201258e-05,
+      "epoch": 0.773125
+    },
+    {
+      "step": 1838,
+      "timestamp": "2025-12-28T12:18:03.161139",
+      "elapsed_time": 11997.446228981018,
+      "loss": 0.3362,
+      "grad_norm": 0.1641501635313034,
+      "learning_rate": 4.566037735849057e-05,
+      "epoch": 0.77375
+    },
+    {
+      "step": 1839,
+      "timestamp": "2025-12-28T12:18:15.104307",
+      "elapsed_time": 12009.389397144318,
+      "loss": 0.1722,
+      "grad_norm": 0.12171639502048492,
+      "learning_rate": 4.553459119496856e-05,
+      "epoch": 0.774375
+    },
+    {
+      "step": 1840,
+      "timestamp": "2025-12-28T12:18:19.637046",
+      "elapsed_time": 12013.922136545181,
+      "loss": 0.2978,
+      "grad_norm": 0.19682510197162628,
+      "learning_rate": 4.5408805031446544e-05,
+      "epoch": 0.775
+    },
+    {
+      "step": 1841,
+      "timestamp": "2025-12-28T12:18:31.203708",
+      "elapsed_time": 12025.48879814148,
+      "loss": 0.1497,
+      "grad_norm": 0.10714510083198547,
+      "learning_rate": 4.528301886792453e-05,
+      "epoch": 0.775625
+    },
+    {
+      "step": 1842,
+      "timestamp": "2025-12-28T12:18:41.667974",
+      "elapsed_time": 12035.95306444168,
+      "loss": 0.3087,
+      "grad_norm": 0.1449776291847229,
+      "learning_rate": 4.515723270440252e-05,
+      "epoch": 0.77625
+    },
+    {
+      "step": 1843,
+      "timestamp": "2025-12-28T12:18:48.114191",
+      "elapsed_time": 12042.399285078049,
+      "loss": 0.1493,
+      "grad_norm": 0.13305744528770447,
+      "learning_rate": 4.503144654088051e-05,
+      "epoch": 0.776875
+    },
+    {
+      "step": 1844,
+      "timestamp": "2025-12-28T12:18:55.887349",
+      "elapsed_time": 12050.172439575195,
+      "loss": 0.3034,
+      "grad_norm": 0.15352778136730194,
+      "learning_rate": 4.4905660377358494e-05,
+      "epoch": 0.7775
+    },
+    {
+      "step": 1845,
+      "timestamp": "2025-12-28T12:19:08.313052",
+      "elapsed_time": 12062.598142147064,
+      "loss": 0.1209,
+      "grad_norm": 0.08751270920038223,
+      "learning_rate": 4.477987421383648e-05,
+      "epoch": 0.778125
+    },
+    {
+      "step": 1846,
+      "timestamp": "2025-12-28T12:19:27.969503",
+      "elapsed_time": 12082.254593133926,
+      "loss": 0.1342,
+      "grad_norm": 0.07397376745939255,
+      "learning_rate": 4.4654088050314466e-05,
+      "epoch": 0.77875
+    },
+    {
+      "step": 1847,
+      "timestamp": "2025-12-28T12:19:34.562787",
+      "elapsed_time": 12088.847877264023,
+      "loss": 0.2321,
+      "grad_norm": 0.15114973485469818,
+      "learning_rate": 4.452830188679246e-05,
+      "epoch": 0.779375
+    },
+    {
+      "step": 1848,
+      "timestamp": "2025-12-28T12:19:40.318451",
+      "elapsed_time": 12094.603546380997,
+      "loss": 0.1495,
+      "grad_norm": 0.13792107999324799,
+      "learning_rate": 4.4402515723270444e-05,
+      "epoch": 0.78
+    },
+    {
+      "step": 1849,
+      "timestamp": "2025-12-28T12:19:50.399687",
+      "elapsed_time": 12104.68477678299,
+      "loss": 0.1054,
+      "grad_norm": 0.10309217870235443,
+      "learning_rate": 4.427672955974843e-05,
+      "epoch": 0.780625
+    },
+    {
+      "step": 1850,
+      "timestamp": "2025-12-28T12:20:05.387318",
+      "elapsed_time": 12119.672408342361,
+      "loss": 0.1211,
+      "grad_norm": 0.08788999915122986,
+      "learning_rate": 4.4150943396226416e-05,
+      "epoch": 0.78125
+    },
+    {
+      "step": 1851,
+      "timestamp": "2025-12-28T12:20:15.522711",
+      "elapsed_time": 12129.807801246643,
+      "loss": 0.1387,
+      "grad_norm": 0.11078450828790665,
+      "learning_rate": 4.402515723270441e-05,
+      "epoch": 0.781875
+    },
+    {
+      "step": 1852,
+      "timestamp": "2025-12-28T12:20:19.994887",
+      "elapsed_time": 12134.279977321625,
+      "loss": 0.3038,
+      "grad_norm": 0.19242431223392487,
+      "learning_rate": 4.3899371069182394e-05,
+      "epoch": 0.7825
+    },
+    {
+      "step": 1853,
+      "timestamp": "2025-12-28T12:20:37.161220",
+      "elapsed_time": 12151.446311235428,
+      "loss": 0.0975,
+      "grad_norm": 0.0753704383969307,
+      "learning_rate": 4.377358490566038e-05,
+      "epoch": 0.783125
+    },
+    {
+      "step": 1854,
+      "timestamp": "2025-12-28T12:20:45.622228",
+      "elapsed_time": 12159.907318115234,
+      "loss": 0.1389,
+      "grad_norm": 0.11638470739126205,
+      "learning_rate": 4.3647798742138366e-05,
+      "epoch": 0.78375
+    },
+    {
+      "step": 1855,
+      "timestamp": "2025-12-28T12:20:51.601755",
+      "elapsed_time": 12165.886845588684,
+      "loss": 0.1841,
+      "grad_norm": 0.14550043642520905,
+      "learning_rate": 4.352201257861636e-05,
+      "epoch": 0.784375
+    },
+    {
+      "step": 1856,
+      "timestamp": "2025-12-28T12:21:02.010335",
+      "elapsed_time": 12176.295424938202,
+      "loss": 0.1331,
+      "grad_norm": 0.10679540038108826,
+      "learning_rate": 4.3396226415094345e-05,
+      "epoch": 0.785
+    },
+    {
+      "step": 1857,
+      "timestamp": "2025-12-28T12:21:07.989370",
+      "elapsed_time": 12182.274460554123,
+      "loss": 0.147,
+      "grad_norm": 0.136283740401268,
+      "learning_rate": 4.327044025157233e-05,
+      "epoch": 0.785625
+    },
+    {
+      "step": 1858,
+      "timestamp": "2025-12-28T12:21:13.524178",
+      "elapsed_time": 12187.809268712997,
+      "loss": 0.2294,
+      "grad_norm": 0.16854384541511536,
+      "learning_rate": 4.3144654088050316e-05,
+      "epoch": 0.78625
+    },
+    {
+      "step": 1859,
+      "timestamp": "2025-12-28T12:21:22.939585",
+      "elapsed_time": 12197.22467494011,
+      "loss": 0.1837,
+      "grad_norm": 0.1425347775220871,
+      "learning_rate": 4.301886792452831e-05,
+      "epoch": 0.786875
+    },
+    {
+      "step": 1860,
+      "timestamp": "2025-12-28T12:21:29.373097",
+      "elapsed_time": 12203.658187150955,
+      "loss": 0.3369,
+      "grad_norm": 0.1832302361726761,
+      "learning_rate": 4.2893081761006295e-05,
+      "epoch": 0.7875
+    },
+    {
+      "step": 1861,
+      "timestamp": "2025-12-28T12:21:33.991433",
+      "elapsed_time": 12208.27652335167,
+      "loss": 0.1674,
+      "grad_norm": 0.16159717738628387,
+      "learning_rate": 4.276729559748428e-05,
+      "epoch": 0.788125
+    },
+    {
+      "step": 1862,
+      "timestamp": "2025-12-28T12:21:41.508287",
+      "elapsed_time": 12215.793377637863,
+      "loss": 0.182,
+      "grad_norm": 0.13854670524597168,
+      "learning_rate": 4.2641509433962266e-05,
+      "epoch": 0.78875
+    },
+    {
+      "step": 1863,
+      "timestamp": "2025-12-28T12:21:47.953231",
+      "elapsed_time": 12222.238321304321,
+      "loss": 0.2442,
+      "grad_norm": 0.13430240750312805,
+      "learning_rate": 4.251572327044025e-05,
+      "epoch": 0.789375
+    },
+    {
+      "step": 1864,
+      "timestamp": "2025-12-28T12:22:00.040777",
+      "elapsed_time": 12234.325867414474,
+      "loss": 0.1518,
+      "grad_norm": 0.10589438676834106,
+      "learning_rate": 4.2389937106918245e-05,
+      "epoch": 0.79
+    },
+    {
+      "step": 1865,
+      "timestamp": "2025-12-28T12:22:05.336070",
+      "elapsed_time": 12239.621160030365,
+      "loss": 0.2583,
+      "grad_norm": 0.1759846806526184,
+      "learning_rate": 4.226415094339623e-05,
+      "epoch": 0.790625
+    },
+    {
+      "step": 1866,
+      "timestamp": "2025-12-28T12:22:13.399786",
+      "elapsed_time": 12247.684876441956,
+      "loss": 0.1389,
+      "grad_norm": 0.12064716964960098,
+      "learning_rate": 4.213836477987422e-05,
+      "epoch": 0.79125
+    },
+    {
+      "step": 1867,
+      "timestamp": "2025-12-28T12:22:34.272638",
+      "elapsed_time": 12268.557728767395,
+      "loss": 0.0743,
+      "grad_norm": 0.06062662601470947,
+      "learning_rate": 4.20125786163522e-05,
+      "epoch": 0.791875
+    },
+    {
+      "step": 1868,
+      "timestamp": "2025-12-28T12:22:42.738728",
+      "elapsed_time": 12277.02381849289,
+      "loss": 0.1641,
+      "grad_norm": 0.12611140310764313,
+      "learning_rate": 4.1886792452830195e-05,
+      "epoch": 0.7925
+    },
+    {
+      "step": 1869,
+      "timestamp": "2025-12-28T12:22:49.875677",
+      "elapsed_time": 12284.160766839981,
+      "loss": 0.1267,
+      "grad_norm": 0.10889780521392822,
+      "learning_rate": 4.176100628930818e-05,
+      "epoch": 0.793125
+    },
+    {
+      "step": 1870,
+      "timestamp": "2025-12-28T12:22:58.497093",
+      "elapsed_time": 12292.782183408737,
+      "loss": 0.4324,
+      "grad_norm": 0.1732441633939743,
+      "learning_rate": 4.163522012578617e-05,
+      "epoch": 0.79375
+    },
+    {
+      "step": 1871,
+      "timestamp": "2025-12-28T12:23:04.803488",
+      "elapsed_time": 12299.08857870102,
+      "loss": 0.4409,
+      "grad_norm": 0.19805917143821716,
+      "learning_rate": 4.150943396226415e-05,
+      "epoch": 0.794375
+    },
+    {
+      "step": 1872,
+      "timestamp": "2025-12-28T12:23:10.710661",
+      "elapsed_time": 12304.995751619339,
+      "loss": 0.4087,
+      "grad_norm": 0.1881469041109085,
+      "learning_rate": 4.1383647798742145e-05,
+      "epoch": 0.795
+    },
+    {
+      "step": 1873,
+      "timestamp": "2025-12-28T12:23:18.265312",
+      "elapsed_time": 12312.55040216446,
+      "loss": 0.1252,
+      "grad_norm": 0.11210913211107254,
+      "learning_rate": 4.125786163522013e-05,
+      "epoch": 0.795625
+    },
+    {
+      "step": 1874,
+      "timestamp": "2025-12-28T12:23:34.658542",
+      "elapsed_time": 12328.943633079529,
+      "loss": 0.116,
+      "grad_norm": 0.10739357024431229,
+      "learning_rate": 4.113207547169812e-05,
+      "epoch": 0.79625
+    },
+    {
+      "step": 1875,
+      "timestamp": "2025-12-28T12:23:40.821995",
+      "elapsed_time": 12335.107085466385,
+      "loss": 0.1966,
+      "grad_norm": 0.14850841462612152,
+      "learning_rate": 4.10062893081761e-05,
+      "epoch": 0.796875
+    },
+    {
+      "step": 1876,
+      "timestamp": "2025-12-28T12:23:47.096816",
+      "elapsed_time": 12341.381906032562,
+      "loss": 0.1838,
+      "grad_norm": 0.1477261185646057,
+      "learning_rate": 4.088050314465409e-05,
+      "epoch": 0.7975
+    },
+    {
+      "step": 1877,
+      "timestamp": "2025-12-28T12:24:07.963691",
+      "elapsed_time": 12362.248781204224,
+      "loss": 0.0783,
+      "grad_norm": 0.06663120537996292,
+      "learning_rate": 4.075471698113208e-05,
+      "epoch": 0.798125
+    },
+    {
+      "step": 1878,
+      "timestamp": "2025-12-28T12:24:25.074044",
+      "elapsed_time": 12379.35913491249,
+      "loss": 0.1794,
+      "grad_norm": 0.10132871568202972,
+      "learning_rate": 4.062893081761007e-05,
+      "epoch": 0.79875
+    },
+    {
+      "step": 1879,
+      "timestamp": "2025-12-28T12:24:35.826343",
+      "elapsed_time": 12390.111433267593,
+      "loss": 0.1471,
+      "grad_norm": 0.11134599149227142,
+      "learning_rate": 4.050314465408805e-05,
+      "epoch": 0.799375
+    },
+    {
+      "step": 1880,
+      "timestamp": "2025-12-28T12:24:42.929007",
+      "elapsed_time": 12397.21409702301,
+      "loss": 0.2019,
+      "grad_norm": 0.1353609710931778,
+      "learning_rate": 4.037735849056604e-05,
+      "epoch": 0.8
+    },
+    {
+      "step": 1881,
+      "timestamp": "2025-12-28T12:24:55.553077",
+      "elapsed_time": 12409.83816742897,
+      "loss": 0.1102,
+      "grad_norm": 0.09974975883960724,
+      "learning_rate": 4.025157232704403e-05,
+      "epoch": 0.800625
+    },
+    {
+      "step": 1882,
+      "timestamp": "2025-12-28T12:25:09.292309",
+      "elapsed_time": 12423.577399730682,
+      "loss": 0.1936,
+      "grad_norm": 0.10728470981121063,
+      "learning_rate": 4.012578616352202e-05,
+      "epoch": 0.80125
+    },
+    {
+      "step": 1883,
+      "timestamp": "2025-12-28T12:25:29.788806",
+      "elapsed_time": 12444.073896169662,
+      "loss": 0.1056,
+      "grad_norm": 0.08063426613807678,
+      "learning_rate": 4e-05,
+      "epoch": 0.801875
+    },
+    {
+      "step": 1884,
+      "timestamp": "2025-12-28T12:25:37.645916",
+      "elapsed_time": 12451.931006908417,
+      "loss": 0.1324,
+      "grad_norm": 0.1093749925494194,
+      "learning_rate": 3.987421383647799e-05,
+      "epoch": 0.8025
+    },
+    {
+      "step": 1885,
+      "timestamp": "2025-12-28T12:25:50.396263",
+      "elapsed_time": 12464.681353330612,
+      "loss": 0.11,
+      "grad_norm": 0.0938941091299057,
+      "learning_rate": 3.974842767295598e-05,
+      "epoch": 0.803125
+    },
+    {
+      "step": 1886,
+      "timestamp": "2025-12-28T12:26:07.508041",
+      "elapsed_time": 12481.793131351471,
+      "loss": 0.1111,
+      "grad_norm": 0.08618809282779694,
+      "learning_rate": 3.962264150943397e-05,
+      "epoch": 0.80375
+    },
+    {
+      "step": 1887,
+      "timestamp": "2025-12-28T12:26:12.131185",
+      "elapsed_time": 12486.416275262833,
+      "loss": 0.1774,
+      "grad_norm": 0.16178780794143677,
+      "learning_rate": 3.9496855345911953e-05,
+      "epoch": 0.804375
+    },
+    {
+      "step": 1888,
+      "timestamp": "2025-12-28T12:26:22.840063",
+      "elapsed_time": 12497.125153303146,
+      "loss": 0.1668,
+      "grad_norm": 0.11834202706813812,
+      "learning_rate": 3.937106918238994e-05,
+      "epoch": 0.805
+    },
+    {
+      "step": 1889,
+      "timestamp": "2025-12-28T12:26:33.452731",
+      "elapsed_time": 12507.737821102142,
+      "loss": 0.1647,
+      "grad_norm": 0.10877163708209991,
+      "learning_rate": 3.924528301886793e-05,
+      "epoch": 0.805625
+    },
+    {
+      "step": 1890,
+      "timestamp": "2025-12-28T12:26:43.963908",
+      "elapsed_time": 12518.248998880386,
+      "loss": 0.3662,
+      "grad_norm": 0.34827324748039246,
+      "learning_rate": 3.911949685534592e-05,
+      "epoch": 0.80625
+    },
+    {
+      "step": 1891,
+      "timestamp": "2025-12-28T12:26:51.062721",
+      "elapsed_time": 12525.347811460495,
+      "loss": 0.1494,
+      "grad_norm": 0.15114177763462067,
+      "learning_rate": 3.8993710691823904e-05,
+      "epoch": 0.806875
+    },
+    {
+      "step": 1892,
+      "timestamp": "2025-12-28T12:27:00.778145",
+      "elapsed_time": 12535.063235282898,
+      "loss": 0.2763,
+      "grad_norm": 0.14429102838039398,
+      "learning_rate": 3.886792452830189e-05,
+      "epoch": 0.8075
+    },
+    {
+      "step": 1893,
+      "timestamp": "2025-12-28T12:27:07.766029",
+      "elapsed_time": 12542.051119327545,
+      "loss": 0.1599,
+      "grad_norm": 0.14175190031528473,
+      "learning_rate": 3.8742138364779875e-05,
+      "epoch": 0.808125
+    },
+    {
+      "step": 1894,
+      "timestamp": "2025-12-28T12:27:18.126378",
+      "elapsed_time": 12552.411469221115,
+      "loss": 0.1061,
+      "grad_norm": 0.08977079391479492,
+      "learning_rate": 3.861635220125787e-05,
+      "epoch": 0.80875
+    },
+    {
+      "step": 1895,
+      "timestamp": "2025-12-28T12:27:24.648082",
+      "elapsed_time": 12558.93317270279,
+      "loss": 0.2233,
+      "grad_norm": 0.15675166249275208,
+      "learning_rate": 3.8490566037735854e-05,
+      "epoch": 0.809375
+    },
+    {
+      "step": 1896,
+      "timestamp": "2025-12-28T12:27:38.818514",
+      "elapsed_time": 12573.10360455513,
+      "loss": 0.1142,
+      "grad_norm": 0.08880900591611862,
+      "learning_rate": 3.836477987421384e-05,
+      "epoch": 0.81
+    },
+    {
+      "step": 1897,
+      "timestamp": "2025-12-28T12:27:45.255238",
+      "elapsed_time": 12579.540328502655,
+      "loss": 0.1721,
+      "grad_norm": 0.14262078702449799,
+      "learning_rate": 3.8238993710691826e-05,
+      "epoch": 0.810625
+    },
+    {
+      "step": 1898,
+      "timestamp": "2025-12-28T12:27:57.156162",
+      "elapsed_time": 12591.441256284714,
+      "loss": 0.1236,
+      "grad_norm": 0.09527979791164398,
+      "learning_rate": 3.811320754716982e-05,
+      "epoch": 0.81125
+    },
+    {
+      "step": 1899,
+      "timestamp": "2025-12-28T12:28:18.026839",
+      "elapsed_time": 12612.31192946434,
+      "loss": 0.1045,
+      "grad_norm": 0.08572285622358322,
+      "learning_rate": 3.7987421383647804e-05,
+      "epoch": 0.811875
+    },
+    {
+      "step": 1900,
+      "timestamp": "2025-12-28T12:28:32.684984",
+      "elapsed_time": 12626.970074653625,
+      "loss": 0.1285,
+      "grad_norm": 0.09535647183656693,
+      "learning_rate": 3.786163522012579e-05,
+      "epoch": 0.8125
+    },
+    {
+      "step": 1901,
+      "timestamp": "2025-12-28T12:28:47.846051",
+      "elapsed_time": 12642.13114118576,
+      "loss": 0.1224,
+      "grad_norm": 0.09724906086921692,
+      "learning_rate": 3.7735849056603776e-05,
+      "epoch": 0.813125
+    },
+    {
+      "step": 1902,
+      "timestamp": "2025-12-28T12:28:56.318374",
+      "elapsed_time": 12650.603464603424,
+      "loss": 0.25,
+      "grad_norm": 0.1405816376209259,
+      "learning_rate": 3.761006289308177e-05,
+      "epoch": 0.81375
+    },
+    {
+      "step": 1903,
+      "timestamp": "2025-12-28T12:29:07.320674",
+      "elapsed_time": 12661.605763912201,
+      "loss": 0.1659,
+      "grad_norm": 0.1113525927066803,
+      "learning_rate": 3.748427672955975e-05,
+      "epoch": 0.814375
+    },
+    {
+      "step": 1904,
+      "timestamp": "2025-12-28T12:29:16.620685",
+      "elapsed_time": 12670.905775785446,
+      "loss": 0.1158,
+      "grad_norm": 0.11509151756763458,
+      "learning_rate": 3.735849056603773e-05,
+      "epoch": 0.815
+    },
+    {
+      "step": 1905,
+      "timestamp": "2025-12-28T12:29:34.296146",
+      "elapsed_time": 12688.581236124039,
+      "loss": 0.1305,
+      "grad_norm": 0.10004525631666183,
+      "learning_rate": 3.723270440251572e-05,
+      "epoch": 0.815625
+    },
+    {
+      "step": 1906,
+      "timestamp": "2025-12-28T12:29:44.763574",
+      "elapsed_time": 12699.048664331436,
+      "loss": 0.1535,
+      "grad_norm": 0.10531044751405716,
+      "learning_rate": 3.710691823899371e-05,
+      "epoch": 0.81625
+    },
+    {
+      "step": 1907,
+      "timestamp": "2025-12-28T12:29:52.278280",
+      "elapsed_time": 12706.56337094307,
+      "loss": 0.1443,
+      "grad_norm": 0.13411474227905273,
+      "learning_rate": 3.69811320754717e-05,
+      "epoch": 0.816875
+    },
+    {
+      "step": 1908,
+      "timestamp": "2025-12-28T12:30:05.896343",
+      "elapsed_time": 12720.181433439255,
+      "loss": 0.1701,
+      "grad_norm": 0.11000078171491623,
+      "learning_rate": 3.6855345911949684e-05,
+      "epoch": 0.8175
+    },
+    {
+      "step": 1909,
+      "timestamp": "2025-12-28T12:30:14.677199",
+      "elapsed_time": 12728.962289571762,
+      "loss": 0.3532,
+      "grad_norm": 0.1655169278383255,
+      "learning_rate": 3.672955974842767e-05,
+      "epoch": 0.818125
+    },
+    {
+      "step": 1910,
+      "timestamp": "2025-12-28T12:30:25.150670",
+      "elapsed_time": 12739.435760736465,
+      "loss": 0.1252,
+      "grad_norm": 0.1451990157365799,
+      "learning_rate": 3.660377358490566e-05,
+      "epoch": 0.81875
+    },
+    {
+      "step": 1911,
+      "timestamp": "2025-12-28T12:30:33.004276",
+      "elapsed_time": 12747.289366722107,
+      "loss": 0.2257,
+      "grad_norm": 0.15634727478027344,
+      "learning_rate": 3.647798742138365e-05,
+      "epoch": 0.819375
+    },
+    {
+      "step": 1912,
+      "timestamp": "2025-12-28T12:30:40.480022",
+      "elapsed_time": 12754.765112161636,
+      "loss": 0.3694,
+      "grad_norm": 0.17593269050121307,
+      "learning_rate": 3.6352201257861634e-05,
+      "epoch": 0.82
+    },
+    {
+      "step": 1913,
+      "timestamp": "2025-12-28T12:30:48.029511",
+      "elapsed_time": 12762.314601421356,
+      "loss": 0.2491,
+      "grad_norm": 0.14427348971366882,
+      "learning_rate": 3.622641509433962e-05,
+      "epoch": 0.820625
+    },
+    {
+      "step": 1914,
+      "timestamp": "2025-12-28T12:30:54.103421",
+      "elapsed_time": 12768.388511657715,
+      "loss": 0.2834,
+      "grad_norm": 0.21829214692115784,
+      "learning_rate": 3.610062893081761e-05,
+      "epoch": 0.82125
+    },
+    {
+      "step": 1915,
+      "timestamp": "2025-12-28T12:31:01.382574",
+      "elapsed_time": 12775.667664289474,
+      "loss": 0.168,
+      "grad_norm": 0.15699121356010437,
+      "learning_rate": 3.59748427672956e-05,
+      "epoch": 0.821875
+    },
+    {
+      "step": 1916,
+      "timestamp": "2025-12-28T12:31:09.694802",
+      "elapsed_time": 12783.979892253876,
+      "loss": 0.2832,
+      "grad_norm": 0.16049239039421082,
+      "learning_rate": 3.5849056603773584e-05,
+      "epoch": 0.8225
+    },
+    {
+      "step": 1917,
+      "timestamp": "2025-12-28T12:31:19.119223",
+      "elapsed_time": 12793.404314041138,
+      "loss": 0.196,
+      "grad_norm": 0.12521395087242126,
+      "learning_rate": 3.572327044025157e-05,
+      "epoch": 0.823125
+    },
+    {
+      "step": 1918,
+      "timestamp": "2025-12-28T12:31:25.059477",
+      "elapsed_time": 12799.344567537308,
+      "loss": 0.144,
+      "grad_norm": 0.12810324132442474,
+      "learning_rate": 3.559748427672956e-05,
+      "epoch": 0.82375
+    },
+    {
+      "step": 1919,
+      "timestamp": "2025-12-28T12:31:31.502266",
+      "elapsed_time": 12805.787356376648,
+      "loss": 0.2127,
+      "grad_norm": 0.14652138948440552,
+      "learning_rate": 3.547169811320755e-05,
+      "epoch": 0.824375
+    },
+    {
+      "step": 1920,
+      "timestamp": "2025-12-28T12:31:39.355931",
+      "elapsed_time": 12813.641021966934,
+      "loss": 0.2024,
+      "grad_norm": 0.13335974514484406,
+      "learning_rate": 3.5345911949685534e-05,
+      "epoch": 0.825
+    },
+    {
+      "step": 1921,
+      "timestamp": "2025-12-28T12:31:45.989318",
+      "elapsed_time": 12820.274408578873,
+      "loss": 0.1712,
+      "grad_norm": 0.13532021641731262,
+      "learning_rate": 3.522012578616352e-05,
+      "epoch": 0.825625
+    },
+    {
+      "step": 1922,
+      "timestamp": "2025-12-28T12:31:55.415620",
+      "elapsed_time": 12829.70071029663,
+      "loss": 0.1493,
+      "grad_norm": 0.11901037395000458,
+      "learning_rate": 3.5094339622641506e-05,
+      "epoch": 0.82625
+    },
+    {
+      "step": 1923,
+      "timestamp": "2025-12-28T12:32:06.926802",
+      "elapsed_time": 12841.211893081665,
+      "loss": 0.1154,
+      "grad_norm": 0.08895418792963028,
+      "learning_rate": 3.49685534591195e-05,
+      "epoch": 0.826875
+    },
+    {
+      "step": 1924,
+      "timestamp": "2025-12-28T12:32:13.518398",
+      "elapsed_time": 12847.8034927845,
+      "loss": 0.1603,
+      "grad_norm": 0.133051335811615,
+      "learning_rate": 3.4842767295597484e-05,
+      "epoch": 0.8275
+    },
+    {
+      "step": 1925,
+      "timestamp": "2025-12-28T12:32:24.513980",
+      "elapsed_time": 12858.799070358276,
+      "loss": 0.1377,
+      "grad_norm": 0.10832136869430542,
+      "learning_rate": 3.471698113207547e-05,
+      "epoch": 0.828125
+    },
+    {
+      "step": 1926,
+      "timestamp": "2025-12-28T12:32:30.547900",
+      "elapsed_time": 12864.83299088478,
+      "loss": 0.393,
+      "grad_norm": 0.16918057203292847,
+      "learning_rate": 3.4591194968553456e-05,
+      "epoch": 0.82875
+    },
+    {
+      "step": 1927,
+      "timestamp": "2025-12-28T12:32:37.613818",
+      "elapsed_time": 12871.89890909195,
+      "loss": 0.1779,
+      "grad_norm": 0.13324803113937378,
+      "learning_rate": 3.446540880503145e-05,
+      "epoch": 0.829375
+    },
+    {
+      "step": 1928,
+      "timestamp": "2025-12-28T12:32:44.130396",
+      "elapsed_time": 12878.415486097336,
+      "loss": 0.1886,
+      "grad_norm": 0.22868554294109344,
+      "learning_rate": 3.4339622641509435e-05,
+      "epoch": 0.83
+    },
+    {
+      "step": 1929,
+      "timestamp": "2025-12-28T12:32:50.719516",
+      "elapsed_time": 12885.00460600853,
+      "loss": 0.3735,
+      "grad_norm": 0.16795550286769867,
+      "learning_rate": 3.421383647798742e-05,
+      "epoch": 0.830625
+    },
+    {
+      "step": 1930,
+      "timestamp": "2025-12-28T12:33:00.614246",
+      "elapsed_time": 12894.899336338043,
+      "loss": 0.1517,
+      "grad_norm": 0.1097177118062973,
+      "learning_rate": 3.4088050314465406e-05,
+      "epoch": 0.83125
+    },
+    {
+      "step": 1931,
+      "timestamp": "2025-12-28T12:33:11.076050",
+      "elapsed_time": 12905.361140727997,
+      "loss": 0.1361,
+      "grad_norm": 0.11156441271305084,
+      "learning_rate": 3.39622641509434e-05,
+      "epoch": 0.831875
+    },
+    {
+      "step": 1932,
+      "timestamp": "2025-12-28T12:33:22.857950",
+      "elapsed_time": 12917.143040180206,
+      "loss": 0.1602,
+      "grad_norm": 0.11632666736841202,
+      "learning_rate": 3.3836477987421385e-05,
+      "epoch": 0.8325
+    },
+    {
+      "step": 1933,
+      "timestamp": "2025-12-28T12:33:43.115463",
+      "elapsed_time": 12937.400553941727,
+      "loss": 0.1039,
+      "grad_norm": 0.07940562069416046,
+      "learning_rate": 3.371069182389937e-05,
+      "epoch": 0.833125
+    },
+    {
+      "step": 1934,
+      "timestamp": "2025-12-28T12:33:56.721853",
+      "elapsed_time": 12951.006942987442,
+      "loss": 0.0978,
+      "grad_norm": 0.27172061800956726,
+      "learning_rate": 3.3584905660377356e-05,
+      "epoch": 0.83375
+    },
+    {
+      "step": 1935,
+      "timestamp": "2025-12-28T12:34:11.804848",
+      "elapsed_time": 12966.089939117432,
+      "loss": 0.1315,
+      "grad_norm": 0.09464891999959946,
+      "learning_rate": 3.345911949685534e-05,
+      "epoch": 0.834375
+    },
+    {
+      "step": 1936,
+      "timestamp": "2025-12-28T12:34:17.303351",
+      "elapsed_time": 12971.58844089508,
+      "loss": 0.1868,
+      "grad_norm": 0.15971067547798157,
+      "learning_rate": 3.3333333333333335e-05,
+      "epoch": 0.835
+    },
+    {
+      "step": 1937,
+      "timestamp": "2025-12-28T12:34:38.344428",
+      "elapsed_time": 12992.629518508911,
+      "loss": 0.0888,
+      "grad_norm": 0.07462675869464874,
+      "learning_rate": 3.320754716981132e-05,
+      "epoch": 0.835625
+    },
+    {
+      "step": 1938,
+      "timestamp": "2025-12-28T12:34:44.610775",
+      "elapsed_time": 12998.89586520195,
+      "loss": 0.1803,
+      "grad_norm": 0.18845033645629883,
+      "learning_rate": 3.308176100628931e-05,
+      "epoch": 0.83625
+    },
+    {
+      "step": 1939,
+      "timestamp": "2025-12-28T12:34:50.913733",
+      "elapsed_time": 13005.198822975159,
+      "loss": 0.2887,
+      "grad_norm": 0.17362096905708313,
+      "learning_rate": 3.295597484276729e-05,
+      "epoch": 0.836875
+    },
+    {
+      "step": 1940,
+      "timestamp": "2025-12-28T12:34:57.357092",
+      "elapsed_time": 13011.642182588577,
+      "loss": 0.1972,
+      "grad_norm": 0.1999269723892212,
+      "learning_rate": 3.2830188679245285e-05,
+      "epoch": 0.8375
+    },
+    {
+      "step": 1941,
+      "timestamp": "2025-12-28T12:35:02.217109",
+      "elapsed_time": 13016.502199172974,
+      "loss": 0.215,
+      "grad_norm": 0.16465437412261963,
+      "learning_rate": 3.270440251572327e-05,
+      "epoch": 0.838125
+    },
+    {
+      "step": 1942,
+      "timestamp": "2025-12-28T12:35:10.992022",
+      "elapsed_time": 13025.27711224556,
+      "loss": 0.1497,
+      "grad_norm": 0.12079225480556488,
+      "learning_rate": 3.257861635220126e-05,
+      "epoch": 0.83875
+    },
+    {
+      "step": 1943,
+      "timestamp": "2025-12-28T12:35:14.991163",
+      "elapsed_time": 13029.276253938675,
+      "loss": 0.2926,
+      "grad_norm": 0.23038972914218903,
+      "learning_rate": 3.245283018867924e-05,
+      "epoch": 0.839375
+    },
+    {
+      "step": 1944,
+      "timestamp": "2025-12-28T12:35:24.077690",
+      "elapsed_time": 13038.362780094147,
+      "loss": 0.5596,
+      "grad_norm": 0.17396830022335052,
+      "learning_rate": 3.2327044025157235e-05,
+      "epoch": 0.84
+    },
+    {
+      "step": 1945,
+      "timestamp": "2025-12-28T12:35:31.821806",
+      "elapsed_time": 13046.106900453568,
+      "loss": 0.1473,
+      "grad_norm": 0.12451894581317902,
+      "learning_rate": 3.220125786163522e-05,
+      "epoch": 0.840625
+    },
+    {
+      "step": 1946,
+      "timestamp": "2025-12-28T12:35:41.040337",
+      "elapsed_time": 13055.325427770615,
+      "loss": 0.0972,
+      "grad_norm": 0.09150367975234985,
+      "learning_rate": 3.207547169811321e-05,
+      "epoch": 0.84125
+    },
+    {
+      "step": 1947,
+      "timestamp": "2025-12-28T12:35:54.543075",
+      "elapsed_time": 13068.828165531158,
+      "loss": 0.1249,
+      "grad_norm": 0.09334319829940796,
+      "learning_rate": 3.194968553459119e-05,
+      "epoch": 0.841875
+    },
+    {
+      "step": 1948,
+      "timestamp": "2025-12-28T12:36:05.426685",
+      "elapsed_time": 13079.711775302887,
+      "loss": 0.1417,
+      "grad_norm": 0.10683770477771759,
+      "learning_rate": 3.1823899371069186e-05,
+      "epoch": 0.8425
+    },
+    {
+      "step": 1949,
+      "timestamp": "2025-12-28T12:36:18.168592",
+      "elapsed_time": 13092.453682422638,
+      "loss": 0.1475,
+      "grad_norm": 0.0970892384648323,
+      "learning_rate": 3.169811320754717e-05,
+      "epoch": 0.843125
+    },
+    {
+      "step": 1950,
+      "timestamp": "2025-12-28T12:36:34.081832",
+      "elapsed_time": 13108.366922616959,
+      "loss": 0.1373,
+      "grad_norm": 0.09424092620611191,
+      "learning_rate": 3.157232704402516e-05,
+      "epoch": 0.84375
+    },
+    {
+      "step": 1951,
+      "timestamp": "2025-12-28T12:36:44.441007",
+      "elapsed_time": 13118.72609782219,
+      "loss": 0.1229,
+      "grad_norm": 0.11297397315502167,
+      "learning_rate": 3.144654088050314e-05,
+      "epoch": 0.844375
+    },
+    {
+      "step": 1952,
+      "timestamp": "2025-12-28T12:36:54.066744",
+      "elapsed_time": 13128.351834058762,
+      "loss": 0.1486,
+      "grad_norm": 0.11981320381164551,
+      "learning_rate": 3.132075471698113e-05,
+      "epoch": 0.845
+    },
+    {
+      "step": 1953,
+      "timestamp": "2025-12-28T12:37:04.580382",
+      "elapsed_time": 13138.86547279358,
+      "loss": 0.1323,
+      "grad_norm": 0.10277368128299713,
+      "learning_rate": 3.119496855345912e-05,
+      "epoch": 0.845625
+    },
+    {
+      "step": 1954,
+      "timestamp": "2025-12-28T12:37:14.942990",
+      "elapsed_time": 13149.228080272675,
+      "loss": 0.2132,
+      "grad_norm": 0.13679690659046173,
+      "learning_rate": 3.106918238993711e-05,
+      "epoch": 0.84625
+    },
+    {
+      "step": 1955,
+      "timestamp": "2025-12-28T12:37:21.894712",
+      "elapsed_time": 13156.179806470871,
+      "loss": 0.1818,
+      "grad_norm": 0.14240722358226776,
+      "learning_rate": 3.094339622641509e-05,
+      "epoch": 0.846875
+    },
+    {
+      "step": 1956,
+      "timestamp": "2025-12-28T12:37:26.431086",
+      "elapsed_time": 13160.716176271439,
+      "loss": 0.2607,
+      "grad_norm": 0.1875351518392563,
+      "learning_rate": 3.081761006289308e-05,
+      "epoch": 0.8475
+    },
+    {
+      "step": 1957,
+      "timestamp": "2025-12-28T12:37:39.981919",
+      "elapsed_time": 13174.267012834549,
+      "loss": 0.181,
+      "grad_norm": 0.12086289376020432,
+      "learning_rate": 3.069182389937107e-05,
+      "epoch": 0.848125
+    },
+    {
+      "step": 1958,
+      "timestamp": "2025-12-28T12:37:50.822168",
+      "elapsed_time": 13185.107258558273,
+      "loss": 0.1471,
+      "grad_norm": 0.11906228959560394,
+      "learning_rate": 3.056603773584906e-05,
+      "epoch": 0.84875
+    },
+    {
+      "step": 1959,
+      "timestamp": "2025-12-28T12:38:08.057968",
+      "elapsed_time": 13202.34305858612,
+      "loss": 0.1174,
+      "grad_norm": 0.07859740406274796,
+      "learning_rate": 3.0440251572327043e-05,
+      "epoch": 0.849375
+    },
+    {
+      "step": 1960,
+      "timestamp": "2025-12-28T12:38:15.115247",
+      "elapsed_time": 13209.400336742401,
+      "loss": 0.1584,
+      "grad_norm": 0.22734029591083527,
+      "learning_rate": 3.0314465408805033e-05,
+      "epoch": 0.85
+    },
+    {
+      "step": 1961,
+      "timestamp": "2025-12-28T12:38:19.840730",
+      "elapsed_time": 13214.12582039833,
+      "loss": 0.2167,
+      "grad_norm": 0.18263642489910126,
+      "learning_rate": 3.018867924528302e-05,
+      "epoch": 0.850625
+    },
+    {
+      "step": 1962,
+      "timestamp": "2025-12-28T12:38:28.595786",
+      "elapsed_time": 13222.880876779556,
+      "loss": 0.1756,
+      "grad_norm": 0.12433876842260361,
+      "learning_rate": 3.0062893081761008e-05,
+      "epoch": 0.85125
+    },
+    {
+      "step": 1963,
+      "timestamp": "2025-12-28T12:38:40.961363",
+      "elapsed_time": 13235.246453523636,
+      "loss": 0.1387,
+      "grad_norm": 0.10110918432474136,
+      "learning_rate": 2.9937106918238994e-05,
+      "epoch": 0.851875
+    },
+    {
+      "step": 1964,
+      "timestamp": "2025-12-28T12:38:52.419176",
+      "elapsed_time": 13246.704270601273,
+      "loss": 0.1096,
+      "grad_norm": 0.0898265689611435,
+      "learning_rate": 2.9811320754716983e-05,
+      "epoch": 0.8525
+    },
+    {
+      "step": 1965,
+      "timestamp": "2025-12-28T12:38:58.899414",
+      "elapsed_time": 13253.18450474739,
+      "loss": 0.194,
+      "grad_norm": 0.15123769640922546,
+      "learning_rate": 2.968553459119497e-05,
+      "epoch": 0.853125
+    },
+    {
+      "step": 1966,
+      "timestamp": "2025-12-28T12:39:06.139908",
+      "elapsed_time": 13260.425002336502,
+      "loss": 0.147,
+      "grad_norm": 0.12028482556343079,
+      "learning_rate": 2.9559748427672958e-05,
+      "epoch": 0.85375
+    },
+    {
+      "step": 1967,
+      "timestamp": "2025-12-28T12:39:14.272168",
+      "elapsed_time": 13268.55725812912,
+      "loss": 0.1709,
+      "grad_norm": 0.13067594170570374,
+      "learning_rate": 2.9433962264150944e-05,
+      "epoch": 0.854375
+    },
+    {
+      "step": 1968,
+      "timestamp": "2025-12-28T12:39:20.178108",
+      "elapsed_time": 13274.463198184967,
+      "loss": 0.1636,
+      "grad_norm": 0.15121272206306458,
+      "learning_rate": 2.930817610062893e-05,
+      "epoch": 0.855
+    },
+    {
+      "step": 1969,
+      "timestamp": "2025-12-28T12:39:27.555839",
+      "elapsed_time": 13281.84092926979,
+      "loss": 0.1912,
+      "grad_norm": 0.18086270987987518,
+      "learning_rate": 2.918238993710692e-05,
+      "epoch": 0.855625
+    },
+    {
+      "step": 1970,
+      "timestamp": "2025-12-28T12:39:37.363912",
+      "elapsed_time": 13291.649002075195,
+      "loss": 0.1739,
+      "grad_norm": 0.12668268382549286,
+      "learning_rate": 2.9056603773584905e-05,
+      "epoch": 0.85625
+    },
+    {
+      "step": 1971,
+      "timestamp": "2025-12-28T12:39:45.421058",
+      "elapsed_time": 13299.706147909164,
+      "loss": 0.1524,
+      "grad_norm": 0.12265957146883011,
+      "learning_rate": 2.8930817610062894e-05,
+      "epoch": 0.856875
+    },
+    {
+      "step": 1972,
+      "timestamp": "2025-12-28T12:40:02.948650",
+      "elapsed_time": 13317.233740568161,
+      "loss": 0.1257,
+      "grad_norm": 0.08345521986484528,
+      "learning_rate": 2.880503144654088e-05,
+      "epoch": 0.8575
+    },
+    {
+      "step": 1973,
+      "timestamp": "2025-12-28T12:40:12.078552",
+      "elapsed_time": 13326.363642930984,
+      "loss": 0.1756,
+      "grad_norm": 0.11556252837181091,
+      "learning_rate": 2.867924528301887e-05,
+      "epoch": 0.858125
+    },
+    {
+      "step": 1974,
+      "timestamp": "2025-12-28T12:40:25.168584",
+      "elapsed_time": 13339.453674077988,
+      "loss": 0.1235,
+      "grad_norm": 0.09028176963329315,
+      "learning_rate": 2.8553459119496855e-05,
+      "epoch": 0.85875
+    },
+    {
+      "step": 1975,
+      "timestamp": "2025-12-28T12:40:33.147283",
+      "elapsed_time": 13347.432373523712,
+      "loss": 0.1337,
+      "grad_norm": 0.12279459834098816,
+      "learning_rate": 2.8427672955974844e-05,
+      "epoch": 0.859375
+    },
+    {
+      "step": 1976,
+      "timestamp": "2025-12-28T12:40:44.760775",
+      "elapsed_time": 13359.045865058899,
+      "loss": 0.1334,
+      "grad_norm": 0.09873386472463608,
+      "learning_rate": 2.830188679245283e-05,
+      "epoch": 0.86
+    },
+    {
+      "step": 1977,
+      "timestamp": "2025-12-28T12:40:55.871028",
+      "elapsed_time": 13370.156118869781,
+      "loss": 0.122,
+      "grad_norm": 0.10663529485464096,
+      "learning_rate": 2.817610062893082e-05,
+      "epoch": 0.860625
+    },
+    {
+      "step": 1978,
+      "timestamp": "2025-12-28T12:41:03.235317",
+      "elapsed_time": 13377.520410776138,
+      "loss": 0.1669,
+      "grad_norm": 0.12646640837192535,
+      "learning_rate": 2.8050314465408805e-05,
+      "epoch": 0.86125
+    },
+    {
+      "step": 1979,
+      "timestamp": "2025-12-28T12:41:11.043014",
+      "elapsed_time": 13385.32810497284,
+      "loss": 0.1255,
+      "grad_norm": 0.12962931394577026,
+      "learning_rate": 2.7924528301886794e-05,
+      "epoch": 0.861875
+    },
+    {
+      "step": 1980,
+      "timestamp": "2025-12-28T12:41:16.966780",
+      "elapsed_time": 13391.251874685287,
+      "loss": 0.1866,
+      "grad_norm": 0.14886367321014404,
+      "learning_rate": 2.779874213836478e-05,
+      "epoch": 0.8625
+    },
+    {
+      "step": 1981,
+      "timestamp": "2025-12-28T12:41:28.910216",
+      "elapsed_time": 13403.195305585861,
+      "loss": 0.1053,
+      "grad_norm": 0.08952232450246811,
+      "learning_rate": 2.767295597484277e-05,
+      "epoch": 0.863125
+    },
+    {
+      "step": 1982,
+      "timestamp": "2025-12-28T12:41:37.341955",
+      "elapsed_time": 13411.627045869827,
+      "loss": 0.2185,
+      "grad_norm": 0.1354040801525116,
+      "learning_rate": 2.7547169811320755e-05,
+      "epoch": 0.86375
+    },
+    {
+      "step": 1983,
+      "timestamp": "2025-12-28T12:41:43.511317",
+      "elapsed_time": 13417.796411514282,
+      "loss": 0.1658,
+      "grad_norm": 0.1327390819787979,
+      "learning_rate": 2.742138364779874e-05,
+      "epoch": 0.864375
+    },
+    {
+      "step": 1984,
+      "timestamp": "2025-12-28T12:41:54.923595",
+      "elapsed_time": 13429.208688497543,
+      "loss": 0.1531,
+      "grad_norm": 0.13857761025428772,
+      "learning_rate": 2.729559748427673e-05,
+      "epoch": 0.865
+    },
+    {
+      "step": 1985,
+      "timestamp": "2025-12-28T12:42:07.782059",
+      "elapsed_time": 13442.06714963913,
+      "loss": 0.1701,
+      "grad_norm": 0.1057012677192688,
+      "learning_rate": 2.7169811320754716e-05,
+      "epoch": 0.865625
+    },
+    {
+      "step": 1986,
+      "timestamp": "2025-12-28T12:42:13.794560",
+      "elapsed_time": 13448.07965040207,
+      "loss": 0.2544,
+      "grad_norm": 0.16066089272499084,
+      "learning_rate": 2.7044025157232706e-05,
+      "epoch": 0.86625
+    },
+    {
+      "step": 1987,
+      "timestamp": "2025-12-28T12:42:24.972624",
+      "elapsed_time": 13459.257714748383,
+      "loss": 0.1063,
+      "grad_norm": 0.10727685689926147,
+      "learning_rate": 2.691823899371069e-05,
+      "epoch": 0.866875
+    },
+    {
+      "step": 1988,
+      "timestamp": "2025-12-28T12:42:35.333485",
+      "elapsed_time": 13469.618575811386,
+      "loss": 0.1744,
+      "grad_norm": 0.10637471824884415,
+      "learning_rate": 2.679245283018868e-05,
+      "epoch": 0.8675
+    },
+    {
+      "step": 1989,
+      "timestamp": "2025-12-28T12:42:44.268707",
+      "elapsed_time": 13478.553797006607,
+      "loss": 0.3446,
+      "grad_norm": 0.18917757272720337,
+      "learning_rate": 2.6666666666666667e-05,
+      "epoch": 0.868125
+    },
+    {
+      "step": 1990,
+      "timestamp": "2025-12-28T12:42:53.119626",
+      "elapsed_time": 13487.404715776443,
+      "loss": 0.1694,
+      "grad_norm": 0.16625277698040009,
+      "learning_rate": 2.6540880503144656e-05,
+      "epoch": 0.86875
+    },
+    {
+      "step": 1991,
+      "timestamp": "2025-12-28T12:42:56.594548",
+      "elapsed_time": 13490.879638910294,
+      "loss": 0.3027,
+      "grad_norm": 0.21782198548316956,
+      "learning_rate": 2.641509433962264e-05,
+      "epoch": 0.869375
+    },
+    {
+      "step": 1992,
+      "timestamp": "2025-12-28T12:43:05.768847",
+      "elapsed_time": 13500.05393743515,
+      "loss": 0.1083,
+      "grad_norm": 0.09708056598901749,
+      "learning_rate": 2.628930817610063e-05,
+      "epoch": 0.87
+    },
+    {
+      "step": 1993,
+      "timestamp": "2025-12-28T12:43:15.670772",
+      "elapsed_time": 13509.955861568451,
+      "loss": 0.13,
+      "grad_norm": 0.12269952893257141,
+      "learning_rate": 2.6163522012578617e-05,
+      "epoch": 0.870625
+    },
+    {
+      "step": 1994,
+      "timestamp": "2025-12-28T12:43:21.608461",
+      "elapsed_time": 13515.893551588058,
+      "loss": 0.2237,
+      "grad_norm": 0.17270702123641968,
+      "learning_rate": 2.6037735849056606e-05,
+      "epoch": 0.87125
+    },
+    {
+      "step": 1995,
+      "timestamp": "2025-12-28T12:43:28.693932",
+      "elapsed_time": 13522.9790225029,
+      "loss": 0.3098,
+      "grad_norm": 0.17161986231803894,
+      "learning_rate": 2.5911949685534592e-05,
+      "epoch": 0.871875
+    },
+    {
+      "step": 1996,
+      "timestamp": "2025-12-28T12:43:35.216183",
+      "elapsed_time": 13529.501273870468,
+      "loss": 0.4953,
+      "grad_norm": 0.18305501341819763,
+      "learning_rate": 2.578616352201258e-05,
+      "epoch": 0.8725
+    },
+    {
+      "step": 1997,
+      "timestamp": "2025-12-28T12:43:40.964345",
+      "elapsed_time": 13535.249434709549,
+      "loss": 0.3349,
+      "grad_norm": 0.1995151937007904,
+      "learning_rate": 2.5660377358490567e-05,
+      "epoch": 0.873125
+    },
+    {
+      "step": 1998,
+      "timestamp": "2025-12-28T12:43:50.213890",
+      "elapsed_time": 13544.498980283737,
+      "loss": 0.1256,
+      "grad_norm": 0.11612726747989655,
+      "learning_rate": 2.5534591194968556e-05,
+      "epoch": 0.87375
+    },
+    {
+      "step": 1999,
+      "timestamp": "2025-12-28T12:44:01.831854",
+      "elapsed_time": 13556.11694407463,
+      "loss": 0.1089,
+      "grad_norm": 0.09675383567810059,
+      "learning_rate": 2.5408805031446542e-05,
+      "epoch": 0.874375
+    },
+    {
+      "step": 2000,
+      "timestamp": "2025-12-28T12:44:08.303416",
+      "elapsed_time": 13562.588506937027,
+      "loss": 0.4029,
+      "grad_norm": 0.19399595260620117,
+      "learning_rate": 2.5283018867924528e-05,
+      "epoch": 0.875
+    },
+    {
+      "step": 2001,
+      "timestamp": "2025-12-28T12:44:16.958567",
+      "elapsed_time": 13571.243657827377,
+      "loss": 0.1523,
+      "grad_norm": 0.1444273293018341,
+      "learning_rate": 2.5157232704402517e-05,
+      "epoch": 0.875625
+    },
+    {
+      "step": 2002,
+      "timestamp": "2025-12-28T12:44:24.980330",
+      "elapsed_time": 13579.26542043686,
+      "loss": 0.4692,
+      "grad_norm": 0.18777810037136078,
+      "learning_rate": 2.5031446540880503e-05,
+      "epoch": 0.87625
+    },
+    {
+      "step": 2003,
+      "timestamp": "2025-12-28T12:44:32.077927",
+      "elapsed_time": 13586.36301779747,
+      "loss": 0.1718,
+      "grad_norm": 0.1408010870218277,
+      "learning_rate": 2.4905660377358492e-05,
+      "epoch": 0.876875
+    },
+    {
+      "step": 2004,
+      "timestamp": "2025-12-28T12:44:37.283086",
+      "elapsed_time": 13591.568180322647,
+      "loss": 0.1881,
+      "grad_norm": 0.1654125154018402,
+      "learning_rate": 2.4779874213836478e-05,
+      "epoch": 0.8775
+    },
+    {
+      "step": 2005,
+      "timestamp": "2025-12-28T12:44:44.801095",
+      "elapsed_time": 13599.08618569374,
+      "loss": 0.1881,
+      "grad_norm": 0.15793928503990173,
+      "learning_rate": 2.4654088050314467e-05,
+      "epoch": 0.878125
+    },
+    {
+      "step": 2006,
+      "timestamp": "2025-12-28T12:44:50.482390",
+      "elapsed_time": 13604.767484664917,
+      "loss": 0.185,
+      "grad_norm": 0.1794230043888092,
+      "learning_rate": 2.4528301886792453e-05,
+      "epoch": 0.87875
+    },
+    {
+      "step": 2007,
+      "timestamp": "2025-12-28T12:45:06.649005",
+      "elapsed_time": 13620.934096097946,
+      "loss": 0.0961,
+      "grad_norm": 0.07328634709119797,
+      "learning_rate": 2.4402515723270442e-05,
+      "epoch": 0.879375
+    },
+    {
+      "step": 2008,
+      "timestamp": "2025-12-28T12:45:10.656329",
+      "elapsed_time": 13624.941419363022,
+      "loss": 0.2643,
+      "grad_norm": 0.2130497395992279,
+      "learning_rate": 2.427672955974843e-05,
+      "epoch": 0.88
+    },
+    {
+      "step": 2009,
+      "timestamp": "2025-12-28T12:45:26.129250",
+      "elapsed_time": 13640.414340496063,
+      "loss": 0.146,
+      "grad_norm": 0.09115441888570786,
+      "learning_rate": 2.4150943396226418e-05,
+      "epoch": 0.880625
+    },
+    {
+      "step": 2010,
+      "timestamp": "2025-12-28T12:45:35.552234",
+      "elapsed_time": 13649.837324380875,
+      "loss": 0.1519,
+      "grad_norm": 0.11384664475917816,
+      "learning_rate": 2.4025157232704403e-05,
+      "epoch": 0.88125
+    },
+    {
+      "step": 2011,
+      "timestamp": "2025-12-28T12:45:42.432010",
+      "elapsed_time": 13656.71710062027,
+      "loss": 0.1838,
+      "grad_norm": 0.13385522365570068,
+      "learning_rate": 2.3899371069182393e-05,
+      "epoch": 0.881875
+    },
+    {
+      "step": 2012,
+      "timestamp": "2025-12-28T12:45:48.338311",
+      "elapsed_time": 13662.623405456543,
+      "loss": 0.3394,
+      "grad_norm": 0.18050479888916016,
+      "learning_rate": 2.377358490566038e-05,
+      "epoch": 0.8825
+    },
+    {
+      "step": 2013,
+      "timestamp": "2025-12-28T12:45:57.763912",
+      "elapsed_time": 13672.049002170563,
+      "loss": 0.1075,
+      "grad_norm": 0.094784677028656,
+      "learning_rate": 2.3647798742138368e-05,
+      "epoch": 0.883125
+    },
+    {
+      "step": 2014,
+      "timestamp": "2025-12-28T12:46:03.098688",
+      "elapsed_time": 13677.383778333664,
+      "loss": 0.2244,
+      "grad_norm": 0.19391919672489166,
+      "learning_rate": 2.3522012578616354e-05,
+      "epoch": 0.88375
+    },
+    {
+      "step": 2015,
+      "timestamp": "2025-12-28T12:46:09.212719",
+      "elapsed_time": 13683.49780869484,
+      "loss": 0.1827,
+      "grad_norm": 0.1965276449918747,
+      "learning_rate": 2.339622641509434e-05,
+      "epoch": 0.884375
+    },
+    {
+      "step": 2016,
+      "timestamp": "2025-12-28T12:46:19.627349",
+      "elapsed_time": 13693.912438869476,
+      "loss": 0.2289,
+      "grad_norm": 0.2740084230899811,
+      "learning_rate": 2.327044025157233e-05,
+      "epoch": 0.885
+    },
+    {
+      "step": 2017,
+      "timestamp": "2025-12-28T12:46:31.416162",
+      "elapsed_time": 13705.70125246048,
+      "loss": 0.2902,
+      "grad_norm": 0.12946578860282898,
+      "learning_rate": 2.3144654088050315e-05,
+      "epoch": 0.885625
+    },
+    {
+      "step": 2018,
+      "timestamp": "2025-12-28T12:46:41.666259",
+      "elapsed_time": 13715.951349258423,
+      "loss": 0.1345,
+      "grad_norm": 0.11686120927333832,
+      "learning_rate": 2.3018867924528304e-05,
+      "epoch": 0.88625
+    },
+    {
+      "step": 2019,
+      "timestamp": "2025-12-28T12:46:57.304742",
+      "elapsed_time": 13731.589831829071,
+      "loss": 0.1551,
+      "grad_norm": 0.10054396092891693,
+      "learning_rate": 2.289308176100629e-05,
+      "epoch": 0.886875
+    },
+    {
+      "step": 2020,
+      "timestamp": "2025-12-28T12:47:12.450934",
+      "elapsed_time": 13746.736023902893,
+      "loss": 0.1416,
+      "grad_norm": 0.11809642612934113,
+      "learning_rate": 2.276729559748428e-05,
+      "epoch": 0.8875
+    },
+    {
+      "step": 2021,
+      "timestamp": "2025-12-28T12:47:22.163353",
+      "elapsed_time": 13756.44844341278,
+      "loss": 0.1301,
+      "grad_norm": 0.1024162694811821,
+      "learning_rate": 2.2641509433962265e-05,
+      "epoch": 0.888125
+    },
+    {
+      "step": 2022,
+      "timestamp": "2025-12-28T12:47:33.984884",
+      "elapsed_time": 13768.269978284836,
+      "loss": 0.1243,
+      "grad_norm": 0.09722079336643219,
+      "learning_rate": 2.2515723270440254e-05,
+      "epoch": 0.88875
+    },
+    {
+      "step": 2023,
+      "timestamp": "2025-12-28T12:47:44.991859",
+      "elapsed_time": 13779.276949644089,
+      "loss": 0.1849,
+      "grad_norm": 0.11634030938148499,
+      "learning_rate": 2.238993710691824e-05,
+      "epoch": 0.889375
+    },
+    {
+      "step": 2024,
+      "timestamp": "2025-12-28T12:47:52.178131",
+      "elapsed_time": 13786.463220596313,
+      "loss": 0.3668,
+      "grad_norm": 0.2070102095603943,
+      "learning_rate": 2.226415094339623e-05,
+      "epoch": 0.89
+    },
+    {
+      "step": 2025,
+      "timestamp": "2025-12-28T12:47:58.522744",
+      "elapsed_time": 13792.807834386826,
+      "loss": 0.1465,
+      "grad_norm": 0.11943688988685608,
+      "learning_rate": 2.2138364779874215e-05,
+      "epoch": 0.890625
+    },
+    {
+      "step": 2026,
+      "timestamp": "2025-12-28T12:48:06.267378",
+      "elapsed_time": 13800.5524725914,
+      "loss": 0.3664,
+      "grad_norm": 0.18064796924591064,
+      "learning_rate": 2.2012578616352204e-05,
+      "epoch": 0.89125
+    },
+    {
+      "step": 2027,
+      "timestamp": "2025-12-28T12:48:16.684863",
+      "elapsed_time": 13810.969953536987,
+      "loss": 0.1507,
+      "grad_norm": 0.10812822729349136,
+      "learning_rate": 2.188679245283019e-05,
+      "epoch": 0.891875
+    },
+    {
+      "step": 2028,
+      "timestamp": "2025-12-28T12:48:26.633930",
+      "elapsed_time": 13820.919019937515,
+      "loss": 0.1116,
+      "grad_norm": 0.0921606570482254,
+      "learning_rate": 2.176100628930818e-05,
+      "epoch": 0.8925
+    },
+    {
+      "step": 2029,
+      "timestamp": "2025-12-28T12:48:34.803507",
+      "elapsed_time": 13829.08859705925,
+      "loss": 0.2362,
+      "grad_norm": 0.14584888517856598,
+      "learning_rate": 2.1635220125786165e-05,
+      "epoch": 0.893125
+    },
+    {
+      "step": 2030,
+      "timestamp": "2025-12-28T12:48:41.721190",
+      "elapsed_time": 13836.006280899048,
+      "loss": 0.3648,
+      "grad_norm": 0.1669163852930069,
+      "learning_rate": 2.1509433962264154e-05,
+      "epoch": 0.89375
+    },
+    {
+      "step": 2031,
+      "timestamp": "2025-12-28T12:48:51.532480",
+      "elapsed_time": 13845.817570209503,
+      "loss": 0.1496,
+      "grad_norm": 0.11350235342979431,
+      "learning_rate": 2.138364779874214e-05,
+      "epoch": 0.894375
+    },
+    {
+      "step": 2032,
+      "timestamp": "2025-12-28T12:49:01.120902",
+      "elapsed_time": 13855.405995368958,
+      "loss": 0.1385,
+      "grad_norm": 0.10483425110578537,
+      "learning_rate": 2.1257861635220126e-05,
+      "epoch": 0.895
+    },
+    {
+      "step": 2033,
+      "timestamp": "2025-12-28T12:49:10.447301",
+      "elapsed_time": 13864.732391834259,
+      "loss": 0.1473,
+      "grad_norm": 0.11778236925601959,
+      "learning_rate": 2.1132075471698115e-05,
+      "epoch": 0.895625
+    },
+    {
+      "step": 2034,
+      "timestamp": "2025-12-28T12:49:24.060029",
+      "elapsed_time": 13878.3451192379,
+      "loss": 0.1106,
+      "grad_norm": 0.09687533229589462,
+      "learning_rate": 2.10062893081761e-05,
+      "epoch": 0.89625
+    },
+    {
+      "step": 2035,
+      "timestamp": "2025-12-28T12:49:29.037036",
+      "elapsed_time": 13883.322126865387,
+      "loss": 0.4332,
+      "grad_norm": 0.2040807008743286,
+      "learning_rate": 2.088050314465409e-05,
+      "epoch": 0.896875
+    },
+    {
+      "step": 2036,
+      "timestamp": "2025-12-28T12:49:45.341778",
+      "elapsed_time": 13899.626868963242,
+      "loss": 0.2711,
+      "grad_norm": 0.12027235329151154,
+      "learning_rate": 2.0754716981132076e-05,
+      "epoch": 0.8975
+    },
+    {
+      "step": 2037,
+      "timestamp": "2025-12-28T12:49:55.608164",
+      "elapsed_time": 13909.893257856369,
+      "loss": 0.1105,
+      "grad_norm": 0.10322773456573486,
+      "learning_rate": 2.0628930817610066e-05,
+      "epoch": 0.898125
+    },
+    {
+      "step": 2038,
+      "timestamp": "2025-12-28T12:50:02.286113",
+      "elapsed_time": 13916.571203231812,
+      "loss": 0.1892,
+      "grad_norm": 0.15024706721305847,
+      "learning_rate": 2.050314465408805e-05,
+      "epoch": 0.89875
+    },
+    {
+      "step": 2039,
+      "timestamp": "2025-12-28T12:50:22.839318",
+      "elapsed_time": 13937.124408245087,
+      "loss": 0.1057,
+      "grad_norm": 0.07259613275527954,
+      "learning_rate": 2.037735849056604e-05,
+      "epoch": 0.899375
+    },
+    {
+      "step": 2040,
+      "timestamp": "2025-12-28T12:50:30.370013",
+      "elapsed_time": 13944.655103683472,
+      "loss": 0.1901,
+      "grad_norm": 0.14556270837783813,
+      "learning_rate": 2.0251572327044027e-05,
+      "epoch": 0.9
+    },
+    {
+      "step": 2041,
+      "timestamp": "2025-12-28T12:50:40.280173",
+      "elapsed_time": 13954.565263748169,
+      "loss": 0.1534,
+      "grad_norm": 0.2072232961654663,
+      "learning_rate": 2.0125786163522016e-05,
+      "epoch": 0.900625
+    },
+    {
+      "step": 2042,
+      "timestamp": "2025-12-28T12:50:48.582155",
+      "elapsed_time": 13962.867245435715,
+      "loss": 0.3279,
+      "grad_norm": 0.2846614122390747,
+      "learning_rate": 2e-05,
+      "epoch": 0.90125
+    },
+    {
+      "step": 2043,
+      "timestamp": "2025-12-28T12:50:55.464925",
+      "elapsed_time": 13969.750015258789,
+      "loss": 0.2373,
+      "grad_norm": 0.14941410720348358,
+      "learning_rate": 1.987421383647799e-05,
+      "epoch": 0.901875
+    },
+    {
+      "step": 2044,
+      "timestamp": "2025-12-28T12:51:03.935715",
+      "elapsed_time": 13978.220804929733,
+      "loss": 0.2142,
+      "grad_norm": 0.13745370507240295,
+      "learning_rate": 1.9748427672955977e-05,
+      "epoch": 0.9025
+    },
+    {
+      "step": 2045,
+      "timestamp": "2025-12-28T12:51:14.650669",
+      "elapsed_time": 13988.935759544373,
+      "loss": 0.1392,
+      "grad_norm": 0.100920669734478,
+      "learning_rate": 1.9622641509433966e-05,
+      "epoch": 0.903125
+    },
+    {
+      "step": 2046,
+      "timestamp": "2025-12-28T12:51:22.402615",
+      "elapsed_time": 13996.687705039978,
+      "loss": 0.1752,
+      "grad_norm": 0.13003851473331451,
+      "learning_rate": 1.9496855345911952e-05,
+      "epoch": 0.90375
+    },
+    {
+      "step": 2047,
+      "timestamp": "2025-12-28T12:51:41.001338",
+      "elapsed_time": 14015.286432504654,
+      "loss": 0.2414,
+      "grad_norm": 0.12125887721776962,
+      "learning_rate": 1.9371069182389938e-05,
+      "epoch": 0.904375
+    },
+    {
+      "step": 2048,
+      "timestamp": "2025-12-28T12:51:50.001884",
+      "elapsed_time": 14024.286974191666,
+      "loss": 0.1899,
+      "grad_norm": 0.1845715045928955,
+      "learning_rate": 1.9245283018867927e-05,
+      "epoch": 0.905
+    },
+    {
+      "step": 2049,
+      "timestamp": "2025-12-28T12:52:10.709887",
+      "elapsed_time": 14044.994977474213,
+      "loss": 0.1702,
+      "grad_norm": 0.09026765078306198,
+      "learning_rate": 1.9119496855345913e-05,
+      "epoch": 0.905625
+    },
+    {
+      "step": 2050,
+      "timestamp": "2025-12-28T12:52:17.185077",
+      "elapsed_time": 14051.470167636871,
+      "loss": 0.1794,
+      "grad_norm": 0.1455916464328766,
+      "learning_rate": 1.8993710691823902e-05,
+      "epoch": 0.90625
+    },
+    {
+      "step": 2051,
+      "timestamp": "2025-12-28T12:52:28.177130",
+      "elapsed_time": 14062.462219953537,
+      "loss": 0.1146,
+      "grad_norm": 0.11119363456964493,
+      "learning_rate": 1.8867924528301888e-05,
+      "epoch": 0.906875
+    },
+    {
+      "step": 2052,
+      "timestamp": "2025-12-28T12:52:36.238218",
+      "elapsed_time": 14070.523307800293,
+      "loss": 0.2618,
+      "grad_norm": 0.17123495042324066,
+      "learning_rate": 1.8742138364779874e-05,
+      "epoch": 0.9075
+    },
+    {
+      "step": 2053,
+      "timestamp": "2025-12-28T12:52:46.104010",
+      "elapsed_time": 14080.389100313187,
+      "loss": 0.2253,
+      "grad_norm": 0.1355234980583191,
+      "learning_rate": 1.861635220125786e-05,
+      "epoch": 0.908125
+    },
+    {
+      "step": 2054,
+      "timestamp": "2025-12-28T12:52:54.928314",
+      "elapsed_time": 14089.213404417038,
+      "loss": 0.1234,
+      "grad_norm": 0.10494759678840637,
+      "learning_rate": 1.849056603773585e-05,
+      "epoch": 0.90875
+    },
+    {
+      "step": 2055,
+      "timestamp": "2025-12-28T12:53:04.303552",
+      "elapsed_time": 14098.588641881943,
+      "loss": 0.1178,
+      "grad_norm": 0.11895407736301422,
+      "learning_rate": 1.8364779874213835e-05,
+      "epoch": 0.909375
+    },
+    {
+      "step": 2056,
+      "timestamp": "2025-12-28T12:53:10.684555",
+      "elapsed_time": 14104.969645500183,
+      "loss": 0.3704,
+      "grad_norm": 0.20144037902355194,
+      "learning_rate": 1.8238993710691824e-05,
+      "epoch": 0.91
+    },
+    {
+      "step": 2057,
+      "timestamp": "2025-12-28T12:53:24.571026",
+      "elapsed_time": 14118.856116056442,
+      "loss": 0.1448,
+      "grad_norm": 0.10675830394029617,
+      "learning_rate": 1.811320754716981e-05,
+      "epoch": 0.910625
+    },
+    {
+      "step": 2058,
+      "timestamp": "2025-12-28T12:53:35.228851",
+      "elapsed_time": 14129.513941764832,
+      "loss": 0.1201,
+      "grad_norm": 0.09952869266271591,
+      "learning_rate": 1.79874213836478e-05,
+      "epoch": 0.91125
+    },
+    {
+      "step": 2059,
+      "timestamp": "2025-12-28T12:53:44.722006",
+      "elapsed_time": 14139.007096767426,
+      "loss": 0.2014,
+      "grad_norm": 0.15313848853111267,
+      "learning_rate": 1.7861635220125785e-05,
+      "epoch": 0.911875
+    },
+    {
+      "step": 2060,
+      "timestamp": "2025-12-28T12:53:55.235495",
+      "elapsed_time": 14149.520585536957,
+      "loss": 0.156,
+      "grad_norm": 0.16458632051944733,
+      "learning_rate": 1.7735849056603774e-05,
+      "epoch": 0.9125
+    },
+    {
+      "step": 2061,
+      "timestamp": "2025-12-28T12:54:04.167911",
+      "elapsed_time": 14158.453001022339,
+      "loss": 0.1428,
+      "grad_norm": 0.11248766630887985,
+      "learning_rate": 1.761006289308176e-05,
+      "epoch": 0.913125
+    },
+    {
+      "step": 2062,
+      "timestamp": "2025-12-28T12:54:14.179512",
+      "elapsed_time": 14168.464602470398,
+      "loss": 0.1162,
+      "grad_norm": 0.10351016372442245,
+      "learning_rate": 1.748427672955975e-05,
+      "epoch": 0.91375
+    },
+    {
+      "step": 2063,
+      "timestamp": "2025-12-28T12:54:19.161273",
+      "elapsed_time": 14173.446363210678,
+      "loss": 0.2882,
+      "grad_norm": 0.19298550486564636,
+      "learning_rate": 1.7358490566037735e-05,
+      "epoch": 0.914375
+    },
+    {
+      "step": 2064,
+      "timestamp": "2025-12-28T12:54:30.239615",
+      "elapsed_time": 14184.524705171585,
+      "loss": 0.1304,
+      "grad_norm": 0.10329542309045792,
+      "learning_rate": 1.7232704402515724e-05,
+      "epoch": 0.915
+    },
+    {
+      "step": 2065,
+      "timestamp": "2025-12-28T12:54:36.398946",
+      "elapsed_time": 14190.684037208557,
+      "loss": 0.2199,
+      "grad_norm": 0.15773223340511322,
+      "learning_rate": 1.710691823899371e-05,
+      "epoch": 0.915625
+    },
+    {
+      "step": 2066,
+      "timestamp": "2025-12-28T12:54:42.958616",
+      "elapsed_time": 14197.243706703186,
+      "loss": 0.2395,
+      "grad_norm": 0.176762193441391,
+      "learning_rate": 1.69811320754717e-05,
+      "epoch": 0.91625
+    },
+    {
+      "step": 2067,
+      "timestamp": "2025-12-28T12:54:54.130810",
+      "elapsed_time": 14208.41589999199,
+      "loss": 0.0974,
+      "grad_norm": 0.08783324062824249,
+      "learning_rate": 1.6855345911949685e-05,
+      "epoch": 0.916875
+    },
+    {
+      "step": 2068,
+      "timestamp": "2025-12-28T12:55:01.944477",
+      "elapsed_time": 14216.22956776619,
+      "loss": 0.196,
+      "grad_norm": 0.14528080821037292,
+      "learning_rate": 1.672955974842767e-05,
+      "epoch": 0.9175
+    },
+    {
+      "step": 2069,
+      "timestamp": "2025-12-28T12:55:13.899335",
+      "elapsed_time": 14228.184425115585,
+      "loss": 0.205,
+      "grad_norm": 0.12957097589969635,
+      "learning_rate": 1.660377358490566e-05,
+      "epoch": 0.918125
+    },
+    {
+      "step": 2070,
+      "timestamp": "2025-12-28T12:55:23.618615",
+      "elapsed_time": 14237.903705358505,
+      "loss": 0.1474,
+      "grad_norm": 0.11285874992609024,
+      "learning_rate": 1.6477987421383646e-05,
+      "epoch": 0.91875
+    },
+    {
+      "step": 2071,
+      "timestamp": "2025-12-28T12:55:37.163652",
+      "elapsed_time": 14251.44874215126,
+      "loss": 0.1786,
+      "grad_norm": 0.09955456852912903,
+      "learning_rate": 1.6352201257861635e-05,
+      "epoch": 0.919375
+    },
+    {
+      "step": 2072,
+      "timestamp": "2025-12-28T12:55:44.637996",
+      "elapsed_time": 14258.923086166382,
+      "loss": 0.1564,
+      "grad_norm": 0.14337117969989777,
+      "learning_rate": 1.622641509433962e-05,
+      "epoch": 0.92
+    },
+    {
+      "step": 2073,
+      "timestamp": "2025-12-28T12:55:53.757282",
+      "elapsed_time": 14268.042372465134,
+      "loss": 0.2802,
+      "grad_norm": 0.1456514596939087,
+      "learning_rate": 1.610062893081761e-05,
+      "epoch": 0.920625
+    },
+    {
+      "step": 2074,
+      "timestamp": "2025-12-28T12:56:00.273054",
+      "elapsed_time": 14274.558144569397,
+      "loss": 0.4306,
+      "grad_norm": 0.18445518612861633,
+      "learning_rate": 1.5974842767295596e-05,
+      "epoch": 0.92125
+    },
+    {
+      "step": 2075,
+      "timestamp": "2025-12-28T12:56:07.521121",
+      "elapsed_time": 14281.806211948395,
+      "loss": 0.1628,
+      "grad_norm": 0.1335839480161667,
+      "learning_rate": 1.5849056603773586e-05,
+      "epoch": 0.921875
+    },
+    {
+      "step": 2076,
+      "timestamp": "2025-12-28T12:56:15.072953",
+      "elapsed_time": 14289.358043909073,
+      "loss": 0.1945,
+      "grad_norm": 0.13724136352539062,
+      "learning_rate": 1.572327044025157e-05,
+      "epoch": 0.9225
+    },
+    {
+      "step": 2077,
+      "timestamp": "2025-12-28T12:56:26.969569",
+      "elapsed_time": 14301.254658937454,
+      "loss": 0.1434,
+      "grad_norm": 0.10332413017749786,
+      "learning_rate": 1.559748427672956e-05,
+      "epoch": 0.923125
+    },
+    {
+      "step": 2078,
+      "timestamp": "2025-12-28T12:56:40.909120",
+      "elapsed_time": 14315.194210767746,
+      "loss": 0.1229,
+      "grad_norm": 0.0949297845363617,
+      "learning_rate": 1.5471698113207547e-05,
+      "epoch": 0.92375
+    },
+    {
+      "step": 2079,
+      "timestamp": "2025-12-28T12:56:46.573585",
+      "elapsed_time": 14320.858675479889,
+      "loss": 0.1417,
+      "grad_norm": 0.1971912533044815,
+      "learning_rate": 1.5345911949685536e-05,
+      "epoch": 0.924375
+    },
+    {
+      "step": 2080,
+      "timestamp": "2025-12-28T12:56:55.214575",
+      "elapsed_time": 14329.499665021896,
+      "loss": 0.1695,
+      "grad_norm": 0.13136085867881775,
+      "learning_rate": 1.5220125786163522e-05,
+      "epoch": 0.925
+    },
+    {
+      "step": 2081,
+      "timestamp": "2025-12-28T12:57:16.087282",
+      "elapsed_time": 14350.372372865677,
+      "loss": 0.0893,
+      "grad_norm": 0.07915375381708145,
+      "learning_rate": 1.509433962264151e-05,
+      "epoch": 0.925625
+    },
+    {
+      "step": 2082,
+      "timestamp": "2025-12-28T12:57:33.493431",
+      "elapsed_time": 14367.778520822525,
+      "loss": 0.1024,
+      "grad_norm": 0.10471498221158981,
+      "learning_rate": 1.4968553459119497e-05,
+      "epoch": 0.92625
+    },
+    {
+      "step": 2083,
+      "timestamp": "2025-12-28T12:57:42.614440",
+      "elapsed_time": 14376.899530172348,
+      "loss": 0.1551,
+      "grad_norm": 0.11656136065721512,
+      "learning_rate": 1.4842767295597484e-05,
+      "epoch": 0.926875
+    },
+    {
+      "step": 2084,
+      "timestamp": "2025-12-28T12:57:47.407488",
+      "elapsed_time": 14381.692578554153,
+      "loss": 0.1693,
+      "grad_norm": 0.1702698916196823,
+      "learning_rate": 1.4716981132075472e-05,
+      "epoch": 0.9275
+    },
+    {
+      "step": 2085,
+      "timestamp": "2025-12-28T12:58:02.399763",
+      "elapsed_time": 14396.684853076935,
+      "loss": 0.1079,
+      "grad_norm": 0.09625794738531113,
+      "learning_rate": 1.459119496855346e-05,
+      "epoch": 0.928125
+    },
+    {
+      "step": 2086,
+      "timestamp": "2025-12-28T12:58:15.434870",
+      "elapsed_time": 14409.719960689545,
+      "loss": 0.1103,
+      "grad_norm": 0.09237074851989746,
+      "learning_rate": 1.4465408805031447e-05,
+      "epoch": 0.92875
+    },
+    {
+      "step": 2087,
+      "timestamp": "2025-12-28T12:58:25.288608",
+      "elapsed_time": 14419.573698043823,
+      "loss": 0.1244,
+      "grad_norm": 0.1188262552022934,
+      "learning_rate": 1.4339622641509435e-05,
+      "epoch": 0.929375
+    },
+    {
+      "step": 2088,
+      "timestamp": "2025-12-28T12:58:32.277959",
+      "elapsed_time": 14426.563049316406,
+      "loss": 0.2027,
+      "grad_norm": 0.2063203752040863,
+      "learning_rate": 1.4213836477987422e-05,
+      "epoch": 0.93
+    },
+    {
+      "step": 2089,
+      "timestamp": "2025-12-28T12:58:43.511867",
+      "elapsed_time": 14437.796962022781,
+      "loss": 0.1312,
+      "grad_norm": 0.10580072551965714,
+      "learning_rate": 1.408805031446541e-05,
+      "epoch": 0.930625
+    },
+    {
+      "step": 2090,
+      "timestamp": "2025-12-28T12:58:56.223461",
+      "elapsed_time": 14450.508551120758,
+      "loss": 0.1399,
+      "grad_norm": 0.10505926609039307,
+      "learning_rate": 1.3962264150943397e-05,
+      "epoch": 0.93125
+    },
+    {
+      "step": 2091,
+      "timestamp": "2025-12-28T12:59:05.563465",
+      "elapsed_time": 14459.848555326462,
+      "loss": 0.2831,
+      "grad_norm": 0.15598870813846588,
+      "learning_rate": 1.3836477987421385e-05,
+      "epoch": 0.931875
+    },
+    {
+      "step": 2092,
+      "timestamp": "2025-12-28T12:59:15.934233",
+      "elapsed_time": 14470.219323635101,
+      "loss": 0.127,
+      "grad_norm": 0.10266047716140747,
+      "learning_rate": 1.371069182389937e-05,
+      "epoch": 0.9325
+    },
+    {
+      "step": 2093,
+      "timestamp": "2025-12-28T12:59:28.065574",
+      "elapsed_time": 14482.350664615631,
+      "loss": 0.1606,
+      "grad_norm": 0.1118762195110321,
+      "learning_rate": 1.3584905660377358e-05,
+      "epoch": 0.933125
+    },
+    {
+      "step": 2094,
+      "timestamp": "2025-12-28T12:59:41.619545",
+      "elapsed_time": 14495.90463590622,
+      "loss": 0.1627,
+      "grad_norm": 0.1268414705991745,
+      "learning_rate": 1.3459119496855346e-05,
+      "epoch": 0.93375
+    },
+    {
+      "step": 2095,
+      "timestamp": "2025-12-28T12:59:57.135947",
+      "elapsed_time": 14511.421037197113,
+      "loss": 0.1509,
+      "grad_norm": 0.09595310688018799,
+      "learning_rate": 1.3333333333333333e-05,
+      "epoch": 0.934375
+    },
+    {
+      "step": 2096,
+      "timestamp": "2025-12-28T13:00:07.077230",
+      "elapsed_time": 14521.362320899963,
+      "loss": 0.1116,
+      "grad_norm": 0.09903395920991898,
+      "learning_rate": 1.320754716981132e-05,
+      "epoch": 0.935
+    },
+    {
+      "step": 2097,
+      "timestamp": "2025-12-28T13:00:17.748745",
+      "elapsed_time": 14532.033835411072,
+      "loss": 0.1372,
+      "grad_norm": 0.10799795389175415,
+      "learning_rate": 1.3081761006289308e-05,
+      "epoch": 0.935625
+    },
+    {
+      "step": 2098,
+      "timestamp": "2025-12-28T13:00:34.507777",
+      "elapsed_time": 14548.792867422104,
+      "loss": 0.0991,
+      "grad_norm": 0.0850529670715332,
+      "learning_rate": 1.2955974842767296e-05,
+      "epoch": 0.93625
+    },
+    {
+      "step": 2099,
+      "timestamp": "2025-12-28T13:00:49.895456",
+      "elapsed_time": 14564.180546045303,
+      "loss": 0.1757,
+      "grad_norm": 0.09990602731704712,
+      "learning_rate": 1.2830188679245283e-05,
+      "epoch": 0.936875
+    },
+    {
+      "step": 2100,
+      "timestamp": "2025-12-28T13:00:57.176345",
+      "elapsed_time": 14571.46143579483,
+      "loss": 0.2252,
+      "grad_norm": 0.1569076031446457,
+      "learning_rate": 1.2704402515723271e-05,
+      "epoch": 0.9375
+    },
+    {
+      "step": 2101,
+      "timestamp": "2025-12-28T13:01:00.897935",
+      "elapsed_time": 14575.183025836945,
+      "loss": 0.2693,
+      "grad_norm": 0.2882073223590851,
+      "learning_rate": 1.2578616352201259e-05,
+      "epoch": 0.938125
+    },
+    {
+      "step": 2102,
+      "timestamp": "2025-12-28T13:01:09.059749",
+      "elapsed_time": 14583.344839334488,
+      "loss": 0.076,
+      "grad_norm": 0.09046997874975204,
+      "learning_rate": 1.2452830188679246e-05,
+      "epoch": 0.93875
+    },
+    {
+      "step": 2103,
+      "timestamp": "2025-12-28T13:01:13.882199",
+      "elapsed_time": 14588.167289018631,
+      "loss": 0.1864,
+      "grad_norm": 0.18029439449310303,
+      "learning_rate": 1.2327044025157234e-05,
+      "epoch": 0.939375
+    },
+    {
+      "step": 2104,
+      "timestamp": "2025-12-28T13:01:26.361765",
+      "elapsed_time": 14600.646855354309,
+      "loss": 0.1504,
+      "grad_norm": 0.09953170269727707,
+      "learning_rate": 1.2201257861635221e-05,
+      "epoch": 0.94
+    },
+    {
+      "step": 2105,
+      "timestamp": "2025-12-28T13:01:33.533065",
+      "elapsed_time": 14607.818155527115,
+      "loss": 0.1704,
+      "grad_norm": 0.14062711596488953,
+      "learning_rate": 1.2075471698113209e-05,
+      "epoch": 0.940625
+    },
+    {
+      "step": 2106,
+      "timestamp": "2025-12-28T13:01:40.874776",
+      "elapsed_time": 14615.159866333008,
+      "loss": 0.3254,
+      "grad_norm": 0.19320929050445557,
+      "learning_rate": 1.1949685534591196e-05,
+      "epoch": 0.94125
+    },
+    {
+      "step": 2107,
+      "timestamp": "2025-12-28T13:01:45.352972",
+      "elapsed_time": 14619.63806271553,
+      "loss": 0.3774,
+      "grad_norm": 0.19888830184936523,
+      "learning_rate": 1.1823899371069184e-05,
+      "epoch": 0.941875
+    },
+    {
+      "step": 2108,
+      "timestamp": "2025-12-28T13:01:54.645732",
+      "elapsed_time": 14628.930822610855,
+      "loss": 0.1445,
+      "grad_norm": 0.11683964729309082,
+      "learning_rate": 1.169811320754717e-05,
+      "epoch": 0.9425
+    },
+    {
+      "step": 2109,
+      "timestamp": "2025-12-28T13:02:12.993090",
+      "elapsed_time": 14647.278180122375,
+      "loss": 0.1005,
+      "grad_norm": 0.08152023702859879,
+      "learning_rate": 1.1572327044025157e-05,
+      "epoch": 0.943125
+    },
+    {
+      "step": 2110,
+      "timestamp": "2025-12-28T13:02:20.465761",
+      "elapsed_time": 14654.750851392746,
+      "loss": 0.1488,
+      "grad_norm": 0.1277417093515396,
+      "learning_rate": 1.1446540880503145e-05,
+      "epoch": 0.94375
+    },
+    {
+      "step": 2111,
+      "timestamp": "2025-12-28T13:02:26.990394",
+      "elapsed_time": 14661.27548456192,
+      "loss": 0.2444,
+      "grad_norm": 0.49488744139671326,
+      "learning_rate": 1.1320754716981132e-05,
+      "epoch": 0.944375
+    },
+    {
+      "step": 2112,
+      "timestamp": "2025-12-28T13:02:36.374549",
+      "elapsed_time": 14670.659639120102,
+      "loss": 0.1365,
+      "grad_norm": 0.11476011574268341,
+      "learning_rate": 1.119496855345912e-05,
+      "epoch": 0.945
+    },
+    {
+      "step": 2113,
+      "timestamp": "2025-12-28T13:02:45.111270",
+      "elapsed_time": 14679.396360874176,
+      "loss": 0.1309,
+      "grad_norm": 0.11950299888849258,
+      "learning_rate": 1.1069182389937107e-05,
+      "epoch": 0.945625
+    },
+    {
+      "step": 2114,
+      "timestamp": "2025-12-28T13:02:51.591249",
+      "elapsed_time": 14685.876343011856,
+      "loss": 0.2217,
+      "grad_norm": 0.16578605771064758,
+      "learning_rate": 1.0943396226415095e-05,
+      "epoch": 0.94625
+    },
+    {
+      "step": 2115,
+      "timestamp": "2025-12-28T13:03:00.576464",
+      "elapsed_time": 14694.861553907394,
+      "loss": 0.3675,
+      "grad_norm": 0.16877128183841705,
+      "learning_rate": 1.0817610062893083e-05,
+      "epoch": 0.946875
+    },
+    {
+      "step": 2116,
+      "timestamp": "2025-12-28T13:03:18.311130",
+      "elapsed_time": 14712.596220493317,
+      "loss": 0.1257,
+      "grad_norm": 0.08800987899303436,
+      "learning_rate": 1.069182389937107e-05,
+      "epoch": 0.9475
+    },
+    {
+      "step": 2117,
+      "timestamp": "2025-12-28T13:03:27.242185",
+      "elapsed_time": 14721.527275800705,
+      "loss": 0.1195,
+      "grad_norm": 0.11122479289770126,
+      "learning_rate": 1.0566037735849058e-05,
+      "epoch": 0.948125
+    },
+    {
+      "step": 2118,
+      "timestamp": "2025-12-28T13:03:37.954344",
+      "elapsed_time": 14732.239434480667,
+      "loss": 0.1142,
+      "grad_norm": 0.0907776802778244,
+      "learning_rate": 1.0440251572327045e-05,
+      "epoch": 0.94875
+    },
+    {
+      "step": 2119,
+      "timestamp": "2025-12-28T13:03:55.267284",
+      "elapsed_time": 14749.552374362946,
+      "loss": 0.0934,
+      "grad_norm": 0.07522504776716232,
+      "learning_rate": 1.0314465408805033e-05,
+      "epoch": 0.949375
+    },
+    {
+      "step": 2120,
+      "timestamp": "2025-12-28T13:04:02.452456",
+      "elapsed_time": 14756.73754644394,
+      "loss": 0.1678,
+      "grad_norm": 0.1397898644208908,
+      "learning_rate": 1.018867924528302e-05,
+      "epoch": 0.95
+    },
+    {
+      "step": 2121,
+      "timestamp": "2025-12-28T13:04:13.404991",
+      "elapsed_time": 14767.690082073212,
+      "loss": 0.1685,
+      "grad_norm": 0.1206582635641098,
+      "learning_rate": 1.0062893081761008e-05,
+      "epoch": 0.950625
+    },
+    {
+      "step": 2122,
+      "timestamp": "2025-12-28T13:04:21.537710",
+      "elapsed_time": 14775.822799921036,
+      "loss": 0.4038,
+      "grad_norm": 0.17276889085769653,
+      "learning_rate": 9.937106918238995e-06,
+      "epoch": 0.95125
+    },
+    {
+      "step": 2123,
+      "timestamp": "2025-12-28T13:04:29.361970",
+      "elapsed_time": 14783.647060155869,
+      "loss": 0.1303,
+      "grad_norm": 0.12048753350973129,
+      "learning_rate": 9.811320754716983e-06,
+      "epoch": 0.951875
+    },
+    {
+      "step": 2124,
+      "timestamp": "2025-12-28T13:04:36.912129",
+      "elapsed_time": 14791.197219133377,
+      "loss": 0.1316,
+      "grad_norm": 0.1113729476928711,
+      "learning_rate": 9.685534591194969e-06,
+      "epoch": 0.9525
+    },
+    {
+      "step": 2125,
+      "timestamp": "2025-12-28T13:04:42.123245",
+      "elapsed_time": 14796.408335924149,
+      "loss": 0.2985,
+      "grad_norm": 0.1864505261182785,
+      "learning_rate": 9.559748427672956e-06,
+      "epoch": 0.953125
+    },
+    {
+      "step": 2126,
+      "timestamp": "2025-12-28T13:04:59.705245",
+      "elapsed_time": 14813.990335941315,
+      "loss": 0.1241,
+      "grad_norm": 0.08124972134828568,
+      "learning_rate": 9.433962264150944e-06,
+      "epoch": 0.95375
+    },
+    {
+      "step": 2127,
+      "timestamp": "2025-12-28T13:05:06.621313",
+      "elapsed_time": 14820.906403064728,
+      "loss": 0.261,
+      "grad_norm": 0.19667741656303406,
+      "learning_rate": 9.30817610062893e-06,
+      "epoch": 0.954375
+    },
+    {
+      "step": 2128,
+      "timestamp": "2025-12-28T13:05:18.124603",
+      "elapsed_time": 14832.409697771072,
+      "loss": 0.2653,
+      "grad_norm": 0.12561364471912384,
+      "learning_rate": 9.182389937106917e-06,
+      "epoch": 0.955
+    },
+    {
+      "step": 2129,
+      "timestamp": "2025-12-28T13:05:22.177655",
+      "elapsed_time": 14836.46274471283,
+      "loss": 0.2647,
+      "grad_norm": 0.21830357611179352,
+      "learning_rate": 9.056603773584905e-06,
+      "epoch": 0.955625
+    },
+    {
+      "step": 2130,
+      "timestamp": "2025-12-28T13:05:31.039317",
+      "elapsed_time": 14845.324407577515,
+      "loss": 0.3741,
+      "grad_norm": 0.14887583255767822,
+      "learning_rate": 8.930817610062892e-06,
+      "epoch": 0.95625
+    },
+    {
+      "step": 2131,
+      "timestamp": "2025-12-28T13:05:42.990517",
+      "elapsed_time": 14857.275606870651,
+      "loss": 0.1244,
+      "grad_norm": 0.11540309339761734,
+      "learning_rate": 8.80503144654088e-06,
+      "epoch": 0.956875
+    },
+    {
+      "step": 2132,
+      "timestamp": "2025-12-28T13:05:48.518364",
+      "elapsed_time": 14862.803453683853,
+      "loss": 0.2524,
+      "grad_norm": 0.17561720311641693,
+      "learning_rate": 8.679245283018868e-06,
+      "epoch": 0.9575
+    },
+    {
+      "step": 2133,
+      "timestamp": "2025-12-28T13:05:54.626380",
+      "elapsed_time": 14868.911471128464,
+      "loss": 0.205,
+      "grad_norm": 0.14403176307678223,
+      "learning_rate": 8.553459119496855e-06,
+      "epoch": 0.958125
+    },
+    {
+      "step": 2134,
+      "timestamp": "2025-12-28T13:06:03.304009",
+      "elapsed_time": 14877.589098930359,
+      "loss": 0.1677,
+      "grad_norm": 0.1289132833480835,
+      "learning_rate": 8.427672955974843e-06,
+      "epoch": 0.95875
+    },
+    {
+      "step": 2135,
+      "timestamp": "2025-12-28T13:06:19.305306",
+      "elapsed_time": 14893.590396642685,
+      "loss": 0.1706,
+      "grad_norm": 0.11218154430389404,
+      "learning_rate": 8.30188679245283e-06,
+      "epoch": 0.959375
+    },
+    {
+      "step": 2136,
+      "timestamp": "2025-12-28T13:06:34.028175",
+      "elapsed_time": 14908.313265562057,
+      "loss": 0.1002,
+      "grad_norm": 0.09118315577507019,
+      "learning_rate": 8.176100628930818e-06,
+      "epoch": 0.96
+    },
+    {
+      "step": 2137,
+      "timestamp": "2025-12-28T13:06:44.744421",
+      "elapsed_time": 14919.029515028,
+      "loss": 0.1318,
+      "grad_norm": 0.11322548240423203,
+      "learning_rate": 8.050314465408805e-06,
+      "epoch": 0.960625
+    },
+    {
+      "step": 2138,
+      "timestamp": "2025-12-28T13:07:05.626666",
+      "elapsed_time": 14939.911756277084,
+      "loss": 0.1157,
+      "grad_norm": 0.07905202358961105,
+      "learning_rate": 7.924528301886793e-06,
+      "epoch": 0.96125
+    },
+    {
+      "step": 2139,
+      "timestamp": "2025-12-28T13:07:12.917071",
+      "elapsed_time": 14947.202164649963,
+      "loss": 0.193,
+      "grad_norm": 0.14137513935565948,
+      "learning_rate": 7.79874213836478e-06,
+      "epoch": 0.961875
+    },
+    {
+      "step": 2140,
+      "timestamp": "2025-12-28T13:07:20.296717",
+      "elapsed_time": 14954.581811666489,
+      "loss": 0.1775,
+      "grad_norm": 0.13806195557117462,
+      "learning_rate": 7.672955974842768e-06,
+      "epoch": 0.9625
+    },
+    {
+      "step": 2141,
+      "timestamp": "2025-12-28T13:07:33.172898",
+      "elapsed_time": 14967.457988500595,
+      "loss": 0.1039,
+      "grad_norm": 0.09371381998062134,
+      "learning_rate": 7.547169811320755e-06,
+      "epoch": 0.963125
+    },
+    {
+      "step": 2142,
+      "timestamp": "2025-12-28T13:07:49.768333",
+      "elapsed_time": 14984.053423166275,
+      "loss": 0.1143,
+      "grad_norm": 0.09071200340986252,
+      "learning_rate": 7.421383647798742e-06,
+      "epoch": 0.96375
+    },
+    {
+      "step": 2143,
+      "timestamp": "2025-12-28T13:08:00.029288",
+      "elapsed_time": 14994.314378499985,
+      "loss": 0.113,
+      "grad_norm": 0.0975324884057045,
+      "learning_rate": 7.29559748427673e-06,
+      "epoch": 0.964375
+    },
+    {
+      "step": 2144,
+      "timestamp": "2025-12-28T13:08:07.277118",
+      "elapsed_time": 15001.56220817566,
+      "loss": 0.1588,
+      "grad_norm": 0.1530754715204239,
+      "learning_rate": 7.169811320754717e-06,
+      "epoch": 0.965
+    },
+    {
+      "step": 2145,
+      "timestamp": "2025-12-28T13:08:11.788620",
+      "elapsed_time": 15006.07371020317,
+      "loss": 0.2424,
+      "grad_norm": 0.17115284502506256,
+      "learning_rate": 7.044025157232705e-06,
+      "epoch": 0.965625
+    },
+    {
+      "step": 2146,
+      "timestamp": "2025-12-28T13:08:24.581324",
+      "elapsed_time": 15018.866414546967,
+      "loss": 0.0938,
+      "grad_norm": 0.08514872193336487,
+      "learning_rate": 6.918238993710692e-06,
+      "epoch": 0.96625
+    },
+    {
+      "step": 2147,
+      "timestamp": "2025-12-28T13:08:35.277645",
+      "elapsed_time": 15029.562735319138,
+      "loss": 0.16,
+      "grad_norm": 0.11149827390909195,
+      "learning_rate": 6.792452830188679e-06,
+      "epoch": 0.966875
+    },
+    {
+      "step": 2148,
+      "timestamp": "2025-12-28T13:08:44.659727",
+      "elapsed_time": 15038.944817781448,
+      "loss": 0.1451,
+      "grad_norm": 0.1102396696805954,
+      "learning_rate": 6.666666666666667e-06,
+      "epoch": 0.9675
+    },
+    {
+      "step": 2149,
+      "timestamp": "2025-12-28T13:08:56.168291",
+      "elapsed_time": 15050.453381538391,
+      "loss": 0.1014,
+      "grad_norm": 0.11420729011297226,
+      "learning_rate": 6.540880503144654e-06,
+      "epoch": 0.968125
+    },
+    {
+      "step": 2150,
+      "timestamp": "2025-12-28T13:09:05.395277",
+      "elapsed_time": 15059.68036699295,
+      "loss": 0.1707,
+      "grad_norm": 0.13086120784282684,
+      "learning_rate": 6.415094339622642e-06,
+      "epoch": 0.96875
+    },
+    {
+      "step": 2151,
+      "timestamp": "2025-12-28T13:09:13.567665",
+      "elapsed_time": 15067.852755784988,
+      "loss": 0.1435,
+      "grad_norm": 0.12494708597660065,
+      "learning_rate": 6.289308176100629e-06,
+      "epoch": 0.969375
+    },
+    {
+      "step": 2152,
+      "timestamp": "2025-12-28T13:09:22.026165",
+      "elapsed_time": 15076.311259746552,
+      "loss": 0.2231,
+      "grad_norm": 0.15418364107608795,
+      "learning_rate": 6.163522012578617e-06,
+      "epoch": 0.97
+    },
+    {
+      "step": 2153,
+      "timestamp": "2025-12-28T13:09:28.234531",
+      "elapsed_time": 15082.519621133804,
+      "loss": 0.1911,
+      "grad_norm": 0.15970326960086823,
+      "learning_rate": 6.037735849056604e-06,
+      "epoch": 0.970625
+    },
+    {
+      "step": 2154,
+      "timestamp": "2025-12-28T13:09:41.850657",
+      "elapsed_time": 15096.135746717453,
+      "loss": 0.1166,
+      "grad_norm": 0.08712539076805115,
+      "learning_rate": 5.911949685534592e-06,
+      "epoch": 0.97125
+    },
+    {
+      "step": 2155,
+      "timestamp": "2025-12-28T13:09:49.022333",
+      "elapsed_time": 15103.307422876358,
+      "loss": 0.2125,
+      "grad_norm": 0.15600165724754333,
+      "learning_rate": 5.786163522012579e-06,
+      "epoch": 0.971875
+    },
+    {
+      "step": 2156,
+      "timestamp": "2025-12-28T13:09:55.267040",
+      "elapsed_time": 15109.552130699158,
+      "loss": 0.1704,
+      "grad_norm": 0.1566295623779297,
+      "learning_rate": 5.660377358490566e-06,
+      "epoch": 0.9725
+    },
+    {
+      "step": 2157,
+      "timestamp": "2025-12-28T13:10:07.050404",
+      "elapsed_time": 15121.33549451828,
+      "loss": 0.122,
+      "grad_norm": 0.09626911580562592,
+      "learning_rate": 5.534591194968554e-06,
+      "epoch": 0.973125
+    },
+    {
+      "step": 2158,
+      "timestamp": "2025-12-28T13:10:13.971436",
+      "elapsed_time": 15128.256526470184,
+      "loss": 0.2565,
+      "grad_norm": 0.14205630123615265,
+      "learning_rate": 5.408805031446541e-06,
+      "epoch": 0.97375
+    },
+    {
+      "step": 2159,
+      "timestamp": "2025-12-28T13:10:23.606334",
+      "elapsed_time": 15137.891424417496,
+      "loss": 0.1477,
+      "grad_norm": 0.12025802582502365,
+      "learning_rate": 5.283018867924529e-06,
+      "epoch": 0.974375
+    },
+    {
+      "step": 2160,
+      "timestamp": "2025-12-28T13:10:38.587449",
+      "elapsed_time": 15152.872539758682,
+      "loss": 0.122,
+      "grad_norm": 0.0904950201511383,
+      "learning_rate": 5.157232704402516e-06,
+      "epoch": 0.975
+    },
+    {
+      "step": 2161,
+      "timestamp": "2025-12-28T13:10:48.211282",
+      "elapsed_time": 15162.496372699738,
+      "loss": 0.1241,
+      "grad_norm": 0.1013362929224968,
+      "learning_rate": 5.031446540880504e-06,
+      "epoch": 0.975625
+    },
+    {
+      "step": 2162,
+      "timestamp": "2025-12-28T13:11:00.090939",
+      "elapsed_time": 15174.376032590866,
+      "loss": 0.3714,
+      "grad_norm": 0.14272800087928772,
+      "learning_rate": 4.9056603773584915e-06,
+      "epoch": 0.97625
+    },
+    {
+      "step": 2163,
+      "timestamp": "2025-12-28T13:11:03.768510",
+      "elapsed_time": 15178.05360007286,
+      "loss": 0.3031,
+      "grad_norm": 0.21242351830005646,
+      "learning_rate": 4.779874213836478e-06,
+      "epoch": 0.976875
+    },
+    {
+      "step": 2164,
+      "timestamp": "2025-12-28T13:11:15.372898",
+      "elapsed_time": 15189.65798830986,
+      "loss": 0.1319,
+      "grad_norm": 0.10468777269124985,
+      "learning_rate": 4.654088050314465e-06,
+      "epoch": 0.9775
+    },
+    {
+      "step": 2165,
+      "timestamp": "2025-12-28T13:11:22.360971",
+      "elapsed_time": 15196.64606142044,
+      "loss": 0.2542,
+      "grad_norm": 0.1459261178970337,
+      "learning_rate": 4.5283018867924524e-06,
+      "epoch": 0.978125
+    },
+    {
+      "step": 2166,
+      "timestamp": "2025-12-28T13:11:27.698671",
+      "elapsed_time": 15201.983761548996,
+      "loss": 0.1664,
+      "grad_norm": 0.16032001376152039,
+      "learning_rate": 4.40251572327044e-06,
+      "epoch": 0.97875
+    },
+    {
+      "step": 2167,
+      "timestamp": "2025-12-28T13:11:35.312824",
+      "elapsed_time": 15209.597914457321,
+      "loss": 0.14,
+      "grad_norm": 0.12672804296016693,
+      "learning_rate": 4.2767295597484275e-06,
+      "epoch": 0.979375
+    },
+    {
+      "step": 2168,
+      "timestamp": "2025-12-28T13:11:42.449067",
+      "elapsed_time": 15216.734157085419,
+      "loss": 0.1409,
+      "grad_norm": 0.21911774575710297,
+      "learning_rate": 4.150943396226415e-06,
+      "epoch": 0.98
+    },
+    {
+      "step": 2169,
+      "timestamp": "2025-12-28T13:11:52.968471",
+      "elapsed_time": 15227.253565788269,
+      "loss": 0.1652,
+      "grad_norm": 0.10740287601947784,
+      "learning_rate": 4.025157232704403e-06,
+      "epoch": 0.980625
+    },
+    {
+      "step": 2170,
+      "timestamp": "2025-12-28T13:12:14.028363",
+      "elapsed_time": 15248.31345319748,
+      "loss": 0.0946,
+      "grad_norm": 0.107594795525074,
+      "learning_rate": 3.89937106918239e-06,
+      "epoch": 0.98125
+    },
+    {
+      "step": 2171,
+      "timestamp": "2025-12-28T13:12:26.958057",
+      "elapsed_time": 15261.243147850037,
+      "loss": 0.1402,
+      "grad_norm": 0.09502803534269333,
+      "learning_rate": 3.7735849056603773e-06,
+      "epoch": 0.981875
+    },
+    {
+      "step": 2172,
+      "timestamp": "2025-12-28T13:12:41.825213",
+      "elapsed_time": 15276.110303640366,
+      "loss": 0.1656,
+      "grad_norm": 0.10404454171657562,
+      "learning_rate": 3.647798742138365e-06,
+      "epoch": 0.9825
+    },
+    {
+      "step": 2173,
+      "timestamp": "2025-12-28T13:12:54.237611",
+      "elapsed_time": 15288.522701501846,
+      "loss": 0.1088,
+      "grad_norm": 0.09469737857580185,
+      "learning_rate": 3.5220125786163524e-06,
+      "epoch": 0.983125
+    },
+    {
+      "step": 2174,
+      "timestamp": "2025-12-28T13:13:14.320573",
+      "elapsed_time": 15308.60566353798,
+      "loss": 0.1108,
+      "grad_norm": 0.08538592606782913,
+      "learning_rate": 3.3962264150943395e-06,
+      "epoch": 0.98375
+    },
+    {
+      "step": 2175,
+      "timestamp": "2025-12-28T13:13:31.983656",
+      "elapsed_time": 15326.2687458992,
+      "loss": 0.113,
+      "grad_norm": 0.07896923273801804,
+      "learning_rate": 3.270440251572327e-06,
+      "epoch": 0.984375
+    },
+    {
+      "step": 2176,
+      "timestamp": "2025-12-28T13:13:36.117474",
+      "elapsed_time": 15330.402564287186,
+      "loss": 0.2856,
+      "grad_norm": 0.20000196993350983,
+      "learning_rate": 3.1446540880503146e-06,
+      "epoch": 0.985
+    },
+    {
+      "step": 2177,
+      "timestamp": "2025-12-28T13:13:40.749900",
+      "elapsed_time": 15335.03499007225,
+      "loss": 0.2019,
+      "grad_norm": 0.17783957719802856,
+      "learning_rate": 3.018867924528302e-06,
+      "epoch": 0.985625
+    },
+    {
+      "step": 2178,
+      "timestamp": "2025-12-28T13:13:50.081194",
+      "elapsed_time": 15344.366284370422,
+      "loss": 0.124,
+      "grad_norm": 0.10862841457128525,
+      "learning_rate": 2.8930817610062893e-06,
+      "epoch": 0.98625
+    },
+    {
+      "step": 2179,
+      "timestamp": "2025-12-28T13:13:57.182561",
+      "elapsed_time": 15351.467651605606,
+      "loss": 0.3846,
+      "grad_norm": 0.1680181324481964,
+      "learning_rate": 2.767295597484277e-06,
+      "epoch": 0.986875
+    },
+    {
+      "step": 2180,
+      "timestamp": "2025-12-28T13:14:18.069949",
+      "elapsed_time": 15372.355040073395,
+      "loss": 0.1295,
+      "grad_norm": 0.10930982232093811,
+      "learning_rate": 2.6415094339622644e-06,
+      "epoch": 0.9875
+    },
+    {
+      "step": 2181,
+      "timestamp": "2025-12-28T13:14:25.840440",
+      "elapsed_time": 15380.125534534454,
+      "loss": 0.1287,
+      "grad_norm": 0.11625991761684418,
+      "learning_rate": 2.515723270440252e-06,
+      "epoch": 0.988125
+    },
+    {
+      "step": 2182,
+      "timestamp": "2025-12-28T13:14:32.900696",
+      "elapsed_time": 15387.185786247253,
+      "loss": 0.2032,
+      "grad_norm": 0.13967153429985046,
+      "learning_rate": 2.389937106918239e-06,
+      "epoch": 0.98875
+    },
+    {
+      "step": 2183,
+      "timestamp": "2025-12-28T13:14:38.813341",
+      "elapsed_time": 15393.098430871964,
+      "loss": 0.1665,
+      "grad_norm": 0.16941803693771362,
+      "learning_rate": 2.2641509433962262e-06,
+      "epoch": 0.989375
+    },
+    {
+      "step": 2184,
+      "timestamp": "2025-12-28T13:14:48.082713",
+      "elapsed_time": 15402.36780333519,
+      "loss": 0.1239,
+      "grad_norm": 0.1080615222454071,
+      "learning_rate": 2.1383647798742138e-06,
+      "epoch": 0.99
+    },
+    {
+      "step": 2185,
+      "timestamp": "2025-12-28T13:14:59.031848",
+      "elapsed_time": 15413.316938877106,
+      "loss": 0.1657,
+      "grad_norm": 0.1154465302824974,
+      "learning_rate": 2.0125786163522013e-06,
+      "epoch": 0.990625
+    },
+    {
+      "step": 2186,
+      "timestamp": "2025-12-28T13:15:09.288405",
+      "elapsed_time": 15423.573495388031,
+      "loss": 0.1329,
+      "grad_norm": 0.10635515302419662,
+      "learning_rate": 1.8867924528301887e-06,
+      "epoch": 0.99125
+    },
+    {
+      "step": 2187,
+      "timestamp": "2025-12-28T13:15:19.358530",
+      "elapsed_time": 15433.64362025261,
+      "loss": 0.1496,
+      "grad_norm": 0.10568325221538544,
+      "learning_rate": 1.7610062893081762e-06,
+      "epoch": 0.991875
+    },
+    {
+      "step": 2188,
+      "timestamp": "2025-12-28T13:15:27.696443",
+      "elapsed_time": 15441.981533288956,
+      "loss": 0.1668,
+      "grad_norm": 0.11821103096008301,
+      "learning_rate": 1.6352201257861635e-06,
+      "epoch": 0.9925
+    },
+    {
+      "step": 2189,
+      "timestamp": "2025-12-28T13:15:32.319517",
+      "elapsed_time": 15446.60460782051,
+      "loss": 0.2601,
+      "grad_norm": 0.18787573277950287,
+      "learning_rate": 1.509433962264151e-06,
+      "epoch": 0.993125
+    },
+    {
+      "step": 2190,
+      "timestamp": "2025-12-28T13:15:42.782764",
+      "elapsed_time": 15457.067858695984,
+      "loss": 0.1287,
+      "grad_norm": 0.11284953355789185,
+      "learning_rate": 1.3836477987421384e-06,
+      "epoch": 0.99375
+    },
+    {
+      "step": 2191,
+      "timestamp": "2025-12-28T13:15:48.893842",
+      "elapsed_time": 15463.178936243057,
+      "loss": 0.1804,
+      "grad_norm": 0.15460120141506195,
+      "learning_rate": 1.257861635220126e-06,
+      "epoch": 0.994375
+    },
+    {
+      "step": 2192,
+      "timestamp": "2025-12-28T13:15:56.849546",
+      "elapsed_time": 15471.134636163712,
+      "loss": 0.166,
+      "grad_norm": 0.13522419333457947,
+      "learning_rate": 1.1320754716981131e-06,
+      "epoch": 0.995
+    },
+    {
+      "step": 2193,
+      "timestamp": "2025-12-28T13:16:09.481063",
+      "elapsed_time": 15483.766153335571,
+      "loss": 0.1317,
+      "grad_norm": 0.11077344417572021,
+      "learning_rate": 1.0062893081761007e-06,
+      "epoch": 0.995625
+    },
+    {
+      "step": 2194,
+      "timestamp": "2025-12-28T13:16:17.328514",
+      "elapsed_time": 15491.613604068756,
+      "loss": 0.4237,
+      "grad_norm": 0.1749534010887146,
+      "learning_rate": 8.805031446540881e-07,
+      "epoch": 0.99625
+    },
+    {
+      "step": 2195,
+      "timestamp": "2025-12-28T13:16:23.345606",
+      "elapsed_time": 15497.63069653511,
+      "loss": 0.1748,
+      "grad_norm": 0.1672588586807251,
+      "learning_rate": 7.547169811320755e-07,
+      "epoch": 0.996875
+    },
+    {
+      "step": 2196,
+      "timestamp": "2025-12-28T13:16:31.823739",
+      "elapsed_time": 15506.108829021454,
+      "loss": 0.2156,
+      "grad_norm": 0.1420753449201584,
+      "learning_rate": 6.28930817610063e-07,
+      "epoch": 0.9975
+    },
+    {
+      "step": 2197,
+      "timestamp": "2025-12-28T13:16:40.998256",
+      "elapsed_time": 15515.283346652985,
+      "loss": 0.1277,
+      "grad_norm": 0.10503512620925903,
+      "learning_rate": 5.031446540880503e-07,
+      "epoch": 0.998125
+    },
+    {
+      "step": 2198,
+      "timestamp": "2025-12-28T13:16:49.844562",
+      "elapsed_time": 15524.12965297699,
+      "loss": 0.1375,
+      "grad_norm": 0.1124568060040474,
+      "learning_rate": 3.773584905660378e-07,
+      "epoch": 0.99875
+    },
+    {
+      "step": 2199,
+      "timestamp": "2025-12-28T13:16:58.714665",
+      "elapsed_time": 15532.9997549057,
+      "loss": 0.1504,
+      "grad_norm": 0.11710984259843826,
+      "learning_rate": 2.5157232704402517e-07,
+      "epoch": 0.999375
+    },
+    {
+      "step": 2200,
+      "timestamp": "2025-12-28T13:17:05.804184",
+      "elapsed_time": 15540.089273929596,
+      "loss": 0.1503,
+      "grad_norm": 0.12923410534858704,
+      "learning_rate": 1.2578616352201258e-07,
+      "epoch": 1.0
+    },
+    {
+      "step": 2200,
+      "timestamp": "2025-12-28T13:17:06.653219",
+      "elapsed_time": 15540.938311100006,
+      "train_runtime": 15528.6205,
+      "train_samples_per_second": 0.103,
+      "train_steps_per_second": 0.103,
+      "total_flos": 5.392201841787095e+17,
+      "train_loss": 0.19463951266836374,
+      "epoch": 1.0
+    }
+  ],
+  "loss_summary": {
+    "min": 0.0733,
+    "max": 0.7795,
+    "final": 0.1503,
+    "average": 0.1946379375
+  },
+  "grad_norm_summary": {
+    "min": 0.06062662601470947,
+    "max": 2.4628686904907227,
+    "final": 0.12923410534858704,
+    "average": 0.14187599109020085
+  }
+}
\ No newline at end of file