Hugging Face's logo

flashresearch
/
FlashResearch-4B-Thinking

Safetensors
qwen3
Model card Files
xet
Community
1
sumo43 commited on
Commit
c4e1fe0
·
verified ·
1 Parent(s): 187f858

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -8
README.md CHANGED
@@ -6,7 +6,7 @@ datasets:
6
 
7
 
8
 
9
- # CheapResearch-4B-Thinking
10
 
11
  <img src='cheap.png' width='700'>
12
 
@@ -19,7 +19,7 @@ datasets:
19
  * **Base**: Qwen 4B (dense)
20
  * **Teacher**: Tongyi DeepResearch 30B A3B (MoE)
21
  * **Method**: SFT distillation on **33k** curated deep-research examples
22
- * **Dataset**: [`cheapresearch/CheapResearch-DS-33k`](https://huggingface.co/datasets/cheapresearch/CheapResearch-DS-33k)
23
  * **Primary Use**: Fast, low-cost **DeepResearch** agent runs (browsing, multi-step reasoning, source-grounded answers)
24
 
25
  ## Evaluation
@@ -29,7 +29,7 @@ datasets:
29
 
30
  ## Training Data
31
 
32
- * **Primary dataset**: [`cheapresearch/CheapResearch-DS-33k`](https://huggingface.co/datasets/cheapresearch/CheapResearch-DS-33k)
33
 
34
  ## Inference with Alibaba-NLP/DeepResearch (Recommended)
35
 
@@ -50,7 +50,7 @@ pip install -e . # or pip install -r requirements.txt if provided
50
  Edit the config to add this model
51
 
52
  ```bash
53
- MODEL_PATH=cheapresearch/CheapResearch-4B-Thinking
54
  ```
55
 
56
  ### Hardware notes
@@ -76,7 +76,7 @@ If you use this model, please cite:
76
  title = {CheapResearch 4B Thinking},
77
  author = {Artem Y.},
78
  year = {2025},
79
- url = {https://huggingface.co/cheapresearch/CheapResearch-4B-Thinking}
80
  }
81
  ```
82
 
@@ -87,7 +87,7 @@ And the dataset:
87
  title = {CheapResearch-DS-33k},
88
  author = {Artem Y.},
89
  year = {2025},
90
- url = {https://huggingface.co/datasets/cheapresearch/CheapResearch-DS-33k}
91
  }
92
  ```
93
 
@@ -119,11 +119,11 @@ tags:
119
  - vllm
120
  - cheapresearch
121
  datasets:
122
- - cheapresearch/CheapResearch-DS-33k
123
  base_model:
124
  - Qwen/Qwen3-4B-Thinking-2507
125
  model-index:
126
- - name: CheapResearch-4B-Thinking
127
  results: []
128
  ---
129
  ```
 
6
 
7
 
8
 
9
+ # FlashResearch-4B-Thinking
10
 
11
  <img src='cheap.png' width='700'>
12
 
 
19
  * **Base**: Qwen 4B (dense)
20
  * **Teacher**: Tongyi DeepResearch 30B A3B (MoE)
21
  * **Method**: SFT distillation on **33k** curated deep-research examples
22
+ * **Dataset**: [`flashresearch/FlashResearch-DS-33k`](https://huggingface.co/datasets/cheapresearch/CheapResearch-DS-33k)
23
  * **Primary Use**: Fast, low-cost **DeepResearch** agent runs (browsing, multi-step reasoning, source-grounded answers)
24
 
25
  ## Evaluation
 
29
 
30
  ## Training Data
31
 
32
+ * **Primary dataset**: [`flashresearch/FlashResearch-DS-33k`](https://huggingface.co/datasets/flashresearch/FlashResearch-DS-33k)
33
 
34
  ## Inference with Alibaba-NLP/DeepResearch (Recommended)
35
 
 
50
  Edit the config to add this model
51
 
52
  ```bash
53
+ MODEL_PATH=flashresearch/FlashResearch-4B-Thinking
54
  ```
55
 
56
  ### Hardware notes
 
76
  title = {CheapResearch 4B Thinking},
77
  author = {Artem Y.},
78
  year = {2025},
79
+ url = {https://huggingface.co/flashresearch/FlashResearch-4B-Thinking}
80
  }
81
  ```
82
 
 
87
  title = {CheapResearch-DS-33k},
88
  author = {Artem Y.},
89
  year = {2025},
90
+ url = {https://huggingface.co/datasets/flashresearch/FlashResearch-DS-33k}
91
  }
92
  ```
93
 
 
119
  - vllm
120
  - cheapresearch
121
  datasets:
122
+ - flashresearch/FlashResearch-DS-33k
123
  base_model:
124
  - Qwen/Qwen3-4B-Thinking-2507
125
  model-index:
126
+ - name: FlashResearch-4B-Thinking
127
  results: []
128
  ---
129
  ```