Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,17 @@
|
|
1 |
This is my reproduction of the Microsoft team's work, WarriorCoder: Learning from Expert Battles to Augment Code Large Language Models. It is fully based on open-source models to construct training data and adopt supervised fine-tuning (SFT) to train the model. The results on code generation benchmarks like Humaneval (Humaneval+) and MBPP (MBPP+) are as follows: 79.9 (75.4), 75.8 (64.5). These results are excellent, confirming that the idea of 'learning from expert battles' proposed in the paper has great potential. I have also published the training data constructed during my reproduction of the paper in another repository, and everyone is welcome to use it.
|
2 |
Original paper link: https://arxiv.org/pdf/2412.17395
|
3 |
-
I have also published the training data constructed during my reproduction of the paper in another repository: https://huggingface.co/datasets/HuggingMicah/warrior_reproduce
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
This is my reproduction of the Microsoft team's work, WarriorCoder: Learning from Expert Battles to Augment Code Large Language Models. It is fully based on open-source models to construct training data and adopt supervised fine-tuning (SFT) to train the model. The results on code generation benchmarks like Humaneval (Humaneval+) and MBPP (MBPP+) are as follows: 79.9 (75.4), 75.8 (64.5). These results are excellent, confirming that the idea of 'learning from expert battles' proposed in the paper has great potential. I have also published the training data constructed during my reproduction of the paper in another repository, and everyone is welcome to use it.
|
2 |
Original paper link: https://arxiv.org/pdf/2412.17395
|
3 |
+
I have also published the training data constructed during my reproduction of the paper in another repository: https://huggingface.co/datasets/HuggingMicah/warrior_reproduce
|
4 |
+
Also, I reproduced the experimental results in the paper. There are some differences from the original results, and I have marked the distinctions with underlines.
|
5 |
+
| Models | Matplotlib (155) | NumPy (220) | Pandas (291) | PyTorch (68) | SciPy (106) | Sklearn (115) | TensorFlow (45) | Overall (1000) |
|
6 |
+
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
|
7 |
+
| INCODER (6.7B) | 28.3 | 4.4 | 3.1 | 4.4 | 2.8 | 2.8 | 3.8 | 7.4 |
|
8 |
+
| CodeGen-Mono (16B) | 31.7 | 10.9 | 3.4 | 7.0 | 9.0 | 10.8 | 15.2 | 11.7 |
|
9 |
+
| Code-Cushman-001 | 40.7 | 21.8 | 7.9 | 12.4 | 11.3 | 18.0 | 12.2 | 18.1 |
|
10 |
+
| StarCoder (15B) | 51.7 | 29.7 | 11.4 | 21.4 | 20.2 | 29.5 | 24.5 | 26.0 |
|
11 |
+
| WizardCoder-SC (15B) | 55.2 | 33.6 | 16.7 | 26.2 | 24.2 | 24.9 | 26.7 | 29.2 |
|
12 |
+
| CodeLlama-Python (6.7B) | 55.3 | 34.5 | 16.4 | 19.9 | 22.3 | 17.6 | 28.5 | 28.0 |
|
13 |
+
| WizardCoder-CL (6.7B) | 53.5 | 34.4 | 15.2 | 25.7 | 21.0 | 24.5 | 28.9 | 28.4 |
|
14 |
+
| Magicoder-CL (6.7B) | 54.6 | 34.8 | 19.0 | 24.7 | 25.0 | 22.6 | 28.9 | 29.9 |
|
15 |
+
| MagicoderS-CL (6.7B) | 55.9 | 40.6 | 28.4 | 40.4 | 28.8 | 35.8 | 37.6 | 37.5 |
|
16 |
+
| WarriorCoder_published_in_paper (6.7B) | 55.5 | 41.8 | 26.1 | 41.2 | 33.0 | 39.1 | 42.2 | 38.1 |
|
17 |
+
| WarriorCoder_my_reproduce (6.7B) | 56.1 | 45.0 | 32.0 | 38.2 | 36.8 | 44.3 | 48.9 | 41.7 |
|