Add model with WER 34.1%

README.md

@@ -1,3 +1,62 @@
 ---
-license: mit
+license: apache-2.0
+tags:
+- generated_from_trainer
+datasets:
+- common_voice
+model-index:
+- name: data2vec-large-uk
+  results: []
 ---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# data2vec-large-uk
+
+This model is a fine-tuned version of [facebook/data2vec-audio-large-960h](https://huggingface.co/facebook/data2vec-audio-large-960h) on the common_voice dataset.
+It achieves the following results on the evaluation set:
+- eval_loss: 0.3472
+- eval_wer: 0.3410
+- eval_cer: 0.0832
+- eval_runtime: 231.0008
+- eval_samples_per_second: 25.108
+- eval_steps_per_second: 3.139
+- epoch: 33.06
+- step: 20400
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 0.0001
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- gradient_accumulation_steps: 6
+- total_train_batch_size: 48
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 500
+- num_epochs: 100
+- mixed_precision_training: Native AMP
+
+### Framework versions
+
+- Transformers 4.18.0
+- Pytorch 1.11.0
+- Datasets 1.18.3
+- Tokenizers 0.12.1

added_tokens.json

@@ -0,0 +1 @@
+{"<s>": 37, "</s>": 38}

config.json

@@ -0,0 +1,110 @@
+{
+  "_name_or_path": "facebook/data2vec-audio-large-960h",
+  "activation_dropout": 0.1,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Data2VecAudioForCTC"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": false,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_pos_kernel_size": 19,
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.0,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.1,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.0,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "model_type": "data2vec-audio",
+  "num_adapter_layers": 3,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 5,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 36,
+  "proj_codevector_dim": 768,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.18.0",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 39,
+  "xvector_output_dim": 512
+}

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:244d90b0d3a3d2fcff5c85fe271012e4fd349606129cf1f4b4c9363876e997a8
+size 1253412017

runs/Apr25_02-03-34_DESKTOP-EL64TH8/1650841871.9287057/events.out.tfevents.1650841871.DESKTOP-EL64TH8.788.3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d303962e215fdb491cbdf2d2ca3bfffea3519c9f72c074ff94c55cc7d29b51d2
+size 4848

runs/Apr25_02-03-34_DESKTOP-EL64TH8/events.out.tfevents.1650841871.DESKTOP-EL64TH8.788.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:048e5e8131dbf09d95e123f391813f39f0e5f1b770e88f2831bd483608ce7749
+size 31564

scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b980b119c19da5f512c2e24a5338956f1410654c202fc467cdb70307305a2488
+size 559

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd56ff55b5d678dc17137e3efef264c3432446d9517eb1ddaf256bfe4941f539
+size 623

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2Processor"}

trainer_state.json

@@ -0,0 +1,832 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 33.063123819800374,
+  "global_step": 20400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 3.4697,
+      "step": 400
+    },
+    {
+      "epoch": 0.65,
+      "eval_cer": 0.22712964200335575,
+      "eval_loss": 0.8050351738929749,
+      "eval_runtime": 237.1034,
+      "eval_samples_per_second": 24.462,
+      "eval_steps_per_second": 3.058,
+      "eval_wer": 0.8556793503918422,
+      "step": 400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.951307189542484e-05,
+      "loss": 0.7017,
+      "step": 800
+    },
+    {
+      "epoch": 1.3,
+      "eval_cer": 0.1719861315277592,
+      "eval_loss": 0.539792537689209,
+      "eval_runtime": 228.344,
+      "eval_samples_per_second": 25.4,
+      "eval_steps_per_second": 3.175,
+      "eval_wer": 0.700901708998206,
+      "step": 800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.885947712418301e-05,
+      "loss": 0.5857,
+      "step": 1200
+    },
+    {
+      "epoch": 1.94,
+      "eval_cer": 0.1479167594696126,
+      "eval_loss": 0.49350985884666443,
+      "eval_runtime": 229.0442,
+      "eval_samples_per_second": 25.323,
+      "eval_steps_per_second": 3.165,
+      "eval_wer": 0.6321877065432915,
+      "step": 1200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.820751633986929e-05,
+      "loss": 0.5104,
+      "step": 1600
+    },
+    {
+      "epoch": 2.59,
+      "eval_cer": 0.1555488737434481,
+      "eval_loss": 0.4537821114063263,
+      "eval_runtime": 228.4555,
+      "eval_samples_per_second": 25.388,
+      "eval_steps_per_second": 3.173,
+      "eval_wer": 0.6290954584080823,
+      "step": 1600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.755392156862746e-05,
+      "loss": 0.4681,
+      "step": 2000
+    },
+    {
+      "epoch": 3.24,
+      "eval_cer": 0.15528531337698784,
+      "eval_loss": 0.4359590411186218,
+      "eval_runtime": 229.3987,
+      "eval_samples_per_second": 25.283,
+      "eval_steps_per_second": 3.16,
+      "eval_wer": 0.6121707109810216,
+      "step": 2000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.690032679738562e-05,
+      "loss": 0.4312,
+      "step": 2400
+    },
+    {
+      "epoch": 3.89,
+      "eval_cer": 0.13666904242208264,
+      "eval_loss": 0.426824152469635,
+      "eval_runtime": 229.499,
+      "eval_samples_per_second": 25.272,
+      "eval_steps_per_second": 3.159,
+      "eval_wer": 0.5816259087904825,
+      "step": 2400
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.624673202614379e-05,
+      "loss": 0.3951,
+      "step": 2800
+    },
+    {
+      "epoch": 4.54,
+      "eval_cer": 0.11381353289678828,
+      "eval_loss": 0.3767826557159424,
+      "eval_runtime": 228.7468,
+      "eval_samples_per_second": 25.356,
+      "eval_steps_per_second": 3.169,
+      "eval_wer": 0.5032338778207912,
+      "step": 2800
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.559313725490196e-05,
+      "loss": 0.3837,
+      "step": 3200
+    },
+    {
+      "epoch": 5.19,
+      "eval_cer": 0.1086499769848694,
+      "eval_loss": 0.3816022276878357,
+      "eval_runtime": 229.1083,
+      "eval_samples_per_second": 25.316,
+      "eval_steps_per_second": 3.164,
+      "eval_wer": 0.4897554527428949,
+      "step": 3200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.493954248366013e-05,
+      "loss": 0.3544,
+      "step": 3600
+    },
+    {
+      "epoch": 5.83,
+      "eval_cer": 0.10945179443776264,
+      "eval_loss": 0.3717349171638489,
+      "eval_runtime": 228.3765,
+      "eval_samples_per_second": 25.397,
+      "eval_steps_per_second": 3.175,
+      "eval_wer": 0.4796053252761779,
+      "step": 3600
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 9.42859477124183e-05,
+      "loss": 0.3347,
+      "step": 4000
+    },
+    {
+      "epoch": 6.48,
+      "eval_cer": 0.10657861523156191,
+      "eval_loss": 0.36351847648620605,
+      "eval_runtime": 228.5091,
+      "eval_samples_per_second": 25.382,
+      "eval_steps_per_second": 3.173,
+      "eval_wer": 0.47561608913228215,
+      "step": 4000
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.363235294117648e-05,
+      "loss": 0.3214,
+      "step": 4400
+    },
+    {
+      "epoch": 7.13,
+      "eval_cer": 0.10085081740834781,
+      "eval_loss": 0.36007240414619446,
+      "eval_runtime": 228.3999,
+      "eval_samples_per_second": 25.394,
+      "eval_steps_per_second": 3.174,
+      "eval_wer": 0.4482579548673402,
+      "step": 4400
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.297875816993465e-05,
+      "loss": 0.3101,
+      "step": 4800
+    },
+    {
+      "epoch": 7.78,
+      "eval_cer": 0.09670066966605788,
+      "eval_loss": 0.3479693531990051,
+      "eval_runtime": 228.8913,
+      "eval_samples_per_second": 25.34,
+      "eval_steps_per_second": 3.167,
+      "eval_wer": 0.4366443206496082,
+      "step": 4800
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 9.232516339869282e-05,
+      "loss": 0.3004,
+      "step": 5200
+    },
+    {
+      "epoch": 8.43,
+      "eval_cer": 0.09928059156309858,
+      "eval_loss": 0.3379470407962799,
+      "eval_runtime": 229.5464,
+      "eval_samples_per_second": 25.267,
+      "eval_steps_per_second": 3.158,
+      "eval_wer": 0.43477952978944384,
+      "step": 5200
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 9.167156862745099e-05,
+      "loss": 0.2837,
+      "step": 5600
+    },
+    {
+      "epoch": 9.08,
+      "eval_cer": 0.10316717893892824,
+      "eval_loss": 0.3556945323944092,
+      "eval_runtime": 229.7028,
+      "eval_samples_per_second": 25.25,
+      "eval_steps_per_second": 3.156,
+      "eval_wer": 0.4406571617411009,
+      "step": 5600
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.101960784313725e-05,
+      "loss": 0.2689,
+      "step": 6000
+    },
+    {
+      "epoch": 9.72,
+      "eval_cer": 0.09850104681723017,
+      "eval_loss": 0.3435617685317993,
+      "eval_runtime": 228.8516,
+      "eval_samples_per_second": 25.344,
+      "eval_steps_per_second": 3.168,
+      "eval_wer": 0.4333632329336229,
+      "step": 6000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 9.036601307189542e-05,
+      "loss": 0.2621,
+      "step": 6400
+    },
+    {
+      "epoch": 10.37,
+      "eval_cer": 0.09027499368939967,
+      "eval_loss": 0.3481593132019043,
+      "eval_runtime": 228.9671,
+      "eval_samples_per_second": 25.331,
+      "eval_steps_per_second": 3.166,
+      "eval_wer": 0.4050609007648003,
+      "step": 6400
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 8.97124183006536e-05,
+      "loss": 0.2572,
+      "step": 6800
+    },
+    {
+      "epoch": 11.02,
+      "eval_cer": 0.0916447651714256,
+      "eval_loss": 0.326036661863327,
+      "eval_runtime": 228.7058,
+      "eval_samples_per_second": 25.36,
+      "eval_steps_per_second": 3.17,
+      "eval_wer": 0.4084600132187707,
+      "step": 6800
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 8.905882352941178e-05,
+      "loss": 0.2413,
+      "step": 7200
+    },
+    {
+      "epoch": 11.67,
+      "eval_cer": 0.08478477140778357,
+      "eval_loss": 0.3280562162399292,
+      "eval_runtime": 229.2353,
+      "eval_samples_per_second": 25.302,
+      "eval_steps_per_second": 3.163,
+      "eval_wer": 0.3888679067132471,
+      "step": 7200
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.840522875816994e-05,
+      "loss": 0.2369,
+      "step": 7600
+    },
+    {
+      "epoch": 12.32,
+      "eval_cer": 0.08900173727114793,
+      "eval_loss": 0.3518214523792267,
+      "eval_runtime": 229.5012,
+      "eval_samples_per_second": 25.272,
+      "eval_steps_per_second": 3.159,
+      "eval_wer": 0.3980266263808894,
+      "step": 7600
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 8.775326797385621e-05,
+      "loss": 0.2331,
+      "step": 8000
+    },
+    {
+      "epoch": 12.96,
+      "eval_cer": 0.08975529719215407,
+      "eval_loss": 0.34484249353408813,
+      "eval_runtime": 230.1435,
+      "eval_samples_per_second": 25.202,
+      "eval_steps_per_second": 3.15,
+      "eval_wer": 0.4034793692758002,
+      "step": 8000
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 8.709967320261438e-05,
+      "loss": 0.2241,
+      "step": 8400
+    },
+    {
+      "epoch": 13.61,
+      "eval_cer": 0.08568310392445098,
+      "eval_loss": 0.32314908504486084,
+      "eval_runtime": 229.283,
+      "eval_samples_per_second": 25.296,
+      "eval_steps_per_second": 3.162,
+      "eval_wer": 0.38374563308469456,
+      "step": 8400
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 8.644607843137255e-05,
+      "loss": 0.2157,
+      "step": 8800
+    },
+    {
+      "epoch": 14.26,
+      "eval_cer": 0.08689696645730322,
+      "eval_loss": 0.3466608226299286,
+      "eval_runtime": 229.4605,
+      "eval_samples_per_second": 25.277,
+      "eval_steps_per_second": 3.16,
+      "eval_wer": 0.38103106411103765,
+      "step": 8800
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 8.579248366013071e-05,
+      "loss": 0.2112,
+      "step": 9200
+    },
+    {
+      "epoch": 14.91,
+      "eval_cer": 0.08458802916239773,
+      "eval_loss": 0.33838367462158203,
+      "eval_runtime": 229.7273,
+      "eval_samples_per_second": 25.247,
+      "eval_steps_per_second": 3.156,
+      "eval_wer": 0.3752950618449627,
+      "step": 9200
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 8.51388888888889e-05,
+      "loss": 0.1992,
+      "step": 9600
+    },
+    {
+      "epoch": 15.56,
+      "eval_cer": 0.08213060715399349,
+      "eval_loss": 0.32480093836784363,
+      "eval_runtime": 229.808,
+      "eval_samples_per_second": 25.238,
+      "eval_steps_per_second": 3.155,
+      "eval_wer": 0.36958266452648475,
+      "step": 9600
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 8.448529411764707e-05,
+      "loss": 0.2,
+      "step": 10000
+    },
+    {
+      "epoch": 16.21,
+      "eval_cer": 0.08102810815626531,
+      "eval_loss": 0.33721038699150085,
+      "eval_runtime": 229.5623,
+      "eval_samples_per_second": 25.265,
+      "eval_steps_per_second": 3.158,
+      "eval_wer": 0.36797752808988765,
+      "step": 10000
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 8.383169934640524e-05,
+      "loss": 0.1958,
+      "step": 10400
+    },
+    {
+      "epoch": 16.85,
+      "eval_cer": 0.084851589528858,
+      "eval_loss": 0.35948196053504944,
+      "eval_runtime": 229.3013,
+      "eval_samples_per_second": 25.294,
+      "eval_steps_per_second": 3.162,
+      "eval_wer": 0.36868567651779816,
+      "step": 10400
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 8.317810457516341e-05,
+      "loss": 0.1894,
+      "step": 10800
+    },
+    {
+      "epoch": 17.5,
+      "eval_cer": 0.07850015590894917,
+      "eval_loss": 0.3233487606048584,
+      "eval_runtime": 229.5135,
+      "eval_samples_per_second": 25.271,
+      "eval_steps_per_second": 3.159,
+      "eval_wer": 0.3536493248984987,
+      "step": 10800
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 8.252450980392157e-05,
+      "loss": 0.1851,
+      "step": 11200
+    },
+    {
+      "epoch": 18.15,
+      "eval_cer": 0.08108750204166482,
+      "eval_loss": 0.34047776460647583,
+      "eval_runtime": 239.1856,
+      "eval_samples_per_second": 24.249,
+      "eval_steps_per_second": 3.031,
+      "eval_wer": 0.3634217732036635,
+      "step": 11200
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 8.187254901960784e-05,
+      "loss": 0.1803,
+      "step": 11600
+    },
+    {
+      "epoch": 18.8,
+      "eval_cer": 0.07939477630777911,
+      "eval_loss": 0.3269987404346466,
+      "eval_runtime": 229.8175,
+      "eval_samples_per_second": 25.237,
+      "eval_steps_per_second": 3.155,
+      "eval_wer": 0.3585827589462751,
+      "step": 11600
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 8.122058823529413e-05,
+      "loss": 0.1745,
+      "step": 12000
+    },
+    {
+      "epoch": 19.45,
+      "eval_cer": 0.08060863884063135,
+      "eval_loss": 0.31651994585990906,
+      "eval_runtime": 240.2409,
+      "eval_samples_per_second": 24.142,
+      "eval_steps_per_second": 3.018,
+      "eval_wer": 0.3572608818808422,
+      "step": 12000
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 8.056699346405229e-05,
+      "loss": 0.1772,
+      "step": 12400
+    },
+    {
+      "epoch": 20.1,
+      "eval_cer": 0.0768519755891131,
+      "eval_loss": 0.3156619668006897,
+      "eval_runtime": 230.0061,
+      "eval_samples_per_second": 25.217,
+      "eval_steps_per_second": 3.152,
+      "eval_wer": 0.3494004343310358,
+      "step": 12400
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 7.991339869281046e-05,
+      "loss": 0.1643,
+      "step": 12800
+    },
+    {
+      "epoch": 20.74,
+      "eval_cer": 0.0779173534084666,
+      "eval_loss": 0.349098265171051,
+      "eval_runtime": 238.8634,
+      "eval_samples_per_second": 24.282,
+      "eval_steps_per_second": 3.035,
+      "eval_wer": 0.34930601454064775,
+      "step": 12800
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 7.925980392156863e-05,
+      "loss": 0.1642,
+      "step": 13200
+    },
+    {
+      "epoch": 21.39,
+      "eval_cer": 0.07913492805915631,
+      "eval_loss": 0.33371418714523315,
+      "eval_runtime": 229.4412,
+      "eval_samples_per_second": 25.279,
+      "eval_steps_per_second": 3.16,
+      "eval_wer": 0.35093475592484186,
+      "step": 13200
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 7.86062091503268e-05,
+      "loss": 0.1593,
+      "step": 13600
+    },
+    {
+      "epoch": 22.04,
+      "eval_cer": 0.07605015813621988,
+      "eval_loss": 0.3546375632286072,
+      "eval_runtime": 229.4973,
+      "eval_samples_per_second": 25.273,
+      "eval_steps_per_second": 3.159,
+      "eval_wer": 0.34203569068076667,
+      "step": 13600
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 7.795261437908498e-05,
+      "loss": 0.1576,
+      "step": 14000
+    },
+    {
+      "epoch": 22.69,
+      "eval_cer": 0.07695962700639969,
+      "eval_loss": 0.3109979033470154,
+      "eval_runtime": 229.6972,
+      "eval_samples_per_second": 25.251,
+      "eval_steps_per_second": 3.156,
+      "eval_wer": 0.3458596921914833,
+      "step": 14000
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 7.729901960784313e-05,
+      "loss": 0.1556,
+      "step": 14400
+    },
+    {
+      "epoch": 23.34,
+      "eval_cer": 0.07288000950302166,
+      "eval_loss": 0.3186021149158478,
+      "eval_runtime": 229.9396,
+      "eval_samples_per_second": 25.224,
+      "eval_steps_per_second": 3.153,
+      "eval_wer": 0.335237465772826,
+      "step": 14400
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 7.66454248366013e-05,
+      "loss": 0.1501,
+      "step": 14800
+    },
+    {
+      "epoch": 23.99,
+      "eval_cer": 0.07652530921941586,
+      "eval_loss": 0.34097790718078613,
+      "eval_runtime": 230.2076,
+      "eval_samples_per_second": 25.195,
+      "eval_steps_per_second": 3.149,
+      "eval_wer": 0.3422245302615428,
+      "step": 14800
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 7.599183006535948e-05,
+      "loss": 0.1468,
+      "step": 15200
+    },
+    {
+      "epoch": 24.63,
+      "eval_cer": 0.07453561405853268,
+      "eval_loss": 0.35943603515625,
+      "eval_runtime": 229.9439,
+      "eval_samples_per_second": 25.224,
+      "eval_steps_per_second": 3.153,
+      "eval_wer": 0.3392503068643188,
+      "step": 15200
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 7.533823529411765e-05,
+      "loss": 0.1478,
+      "step": 15600
+    },
+    {
+      "epoch": 25.28,
+      "eval_cer": 0.07782083834469242,
+      "eval_loss": 0.3322964310646057,
+      "eval_runtime": 229.8368,
+      "eval_samples_per_second": 25.235,
+      "eval_steps_per_second": 3.154,
+      "eval_wer": 0.34272023416108016,
+      "step": 15600
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 7.468464052287582e-05,
+      "loss": 0.1472,
+      "step": 16000
+    },
+    {
+      "epoch": 25.93,
+      "eval_cer": 0.08302522755282343,
+      "eval_loss": 0.3150421679019928,
+      "eval_runtime": 229.9477,
+      "eval_samples_per_second": 25.223,
+      "eval_steps_per_second": 3.153,
+      "eval_wer": 0.3516901142479464,
+      "step": 16000
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 7.40326797385621e-05,
+      "loss": 0.1451,
+      "step": 16400
+    },
+    {
+      "epoch": 26.58,
+      "eval_cer": 0.08323310615172168,
+      "eval_loss": 0.33504411578178406,
+      "eval_runtime": 230.1669,
+      "eval_samples_per_second": 25.199,
+      "eval_steps_per_second": 3.15,
+      "eval_wer": 0.352657917099424,
+      "step": 16400
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 7.337908496732027e-05,
+      "loss": 0.1405,
+      "step": 16800
+    },
+    {
+      "epoch": 27.23,
+      "eval_cer": 0.08127682005137571,
+      "eval_loss": 0.3452860116958618,
+      "eval_runtime": 229.9933,
+      "eval_samples_per_second": 25.218,
+      "eval_steps_per_second": 3.152,
+      "eval_wer": 0.3461665565102445,
+      "step": 16800
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 7.272549019607844e-05,
+      "loss": 0.1384,
+      "step": 17200
+    },
+    {
+      "epoch": 27.88,
+      "eval_cer": 0.07939477630777911,
+      "eval_loss": 0.3451952338218689,
+      "eval_runtime": 230.0148,
+      "eval_samples_per_second": 25.216,
+      "eval_steps_per_second": 3.152,
+      "eval_wer": 0.343947691436125,
+      "step": 17200
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 7.20718954248366e-05,
+      "loss": 0.1385,
+      "step": 17600
+    },
+    {
+      "epoch": 28.52,
+      "eval_cer": 0.08665939091570522,
+      "eval_loss": 0.32853472232818604,
+      "eval_runtime": 230.4126,
+      "eval_samples_per_second": 25.172,
+      "eval_steps_per_second": 3.147,
+      "eval_wer": 0.3517609290907374,
+      "step": 17600
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 7.141830065359477e-05,
+      "loss": 0.133,
+      "step": 18000
+    },
+    {
+      "epoch": 29.17,
+      "eval_cer": 0.08029310882444653,
+      "eval_loss": 0.33895909786224365,
+      "eval_runtime": 230.6075,
+      "eval_samples_per_second": 25.151,
+      "eval_steps_per_second": 3.144,
+      "eval_wer": 0.34243697478991597,
+      "step": 18000
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 7.076470588235294e-05,
+      "loss": 0.1303,
+      "step": 18400
+    },
+    {
+      "epoch": 29.82,
+      "eval_cer": 0.07742735385392074,
+      "eval_loss": 0.33118581771850586,
+      "eval_runtime": 234.6432,
+      "eval_samples_per_second": 24.718,
+      "eval_steps_per_second": 3.09,
+      "eval_wer": 0.33403361344537813,
+      "step": 18400
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 7.011111111111112e-05,
+      "loss": 0.126,
+      "step": 18800
+    },
+    {
+      "epoch": 30.47,
+      "eval_cer": 0.0883781014744532,
+      "eval_loss": 0.36238303780555725,
+      "eval_runtime": 230.653,
+      "eval_samples_per_second": 25.146,
+      "eval_steps_per_second": 3.143,
+      "eval_wer": 0.3550420168067227,
+      "step": 18800
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 6.945915032679738e-05,
+      "loss": 0.1282,
+      "step": 19200
+    },
+    {
+      "epoch": 31.12,
+      "eval_cer": 0.08413886290406403,
+      "eval_loss": 0.33442607522010803,
+      "eval_runtime": 229.9326,
+      "eval_samples_per_second": 25.225,
+      "eval_steps_per_second": 3.153,
+      "eval_wer": 0.3440893211217071,
+      "step": 19200
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 6.880555555555556e-05,
+      "loss": 0.1284,
+      "step": 19600
+    },
+    {
+      "epoch": 31.77,
+      "eval_cer": 0.08161833489242283,
+      "eval_loss": 0.3269757628440857,
+      "eval_runtime": 230.2743,
+      "eval_samples_per_second": 25.187,
+      "eval_steps_per_second": 3.148,
+      "eval_wer": 0.3438296666981399,
+      "step": 19600
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 6.815196078431373e-05,
+      "loss": 0.1258,
+      "step": 20000
+    },
+    {
+      "epoch": 32.41,
+      "eval_cer": 0.08625848218925862,
+      "eval_loss": 0.32494238018989563,
+      "eval_runtime": 229.6629,
+      "eval_samples_per_second": 25.254,
+      "eval_steps_per_second": 3.157,
+      "eval_wer": 0.34472665470682656,
+      "step": 20000
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 6.74983660130719e-05,
+      "loss": 0.1208,
+      "step": 20400
+    },
+    {
+      "epoch": 33.06,
+      "eval_cer": 0.08316628803064724,
+      "eval_loss": 0.34723225235939026,
+      "eval_runtime": 231.0008,
+      "eval_samples_per_second": 25.108,
+      "eval_steps_per_second": 3.139,
+      "eval_wer": 0.34099707298649795,
+      "step": 20400
+    }
+  ],
+  "max_steps": 61700,
+  "num_train_epochs": 100,
+  "total_flos": 1.394888039414942e+20,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccdae5f3562905af457c6028ff9d4756dd27a2c89c52992cc8e238837a2985df
+size 3055

vocab.json

@@ -0,0 +1 @@
+{"'": 1, "а": 2, "б": 3, "в": 4, "г": 5, "д": 6, "е": 7, "ж": 8, "з": 9, "и": 10, "й": 11, "к": 12, "л": 13, "м": 14, "н": 15, "о": 16, "п": 17, "р": 18, "с": 19, "т": 20, "у": 21, "ф": 22, "х": 23, "ц": 24, "ч": 25, "ш": 26, "щ": 27, "ь": 28, "ю": 29, "я": 30, "є": 31, "і": 32, "ї": 33, "ґ": 34, "|": 0, "[UNK]": 35, "[PAD]": 36}