| # CodeWeave-QwenCoder | |
| A fine-tuned code generation model optimized for Python and JavaScript development tasks. | |
| ## Model Description | |
| CodeWeave-QwenCoder is a specialized code generation model fine-tuned from **Qwen/Qwen2.5-Coder-7B-Instruct** for enhanced performance on real-world coding tasks. | |
| ### Base Model | |
| - **Foundation**: Qwen2.5-Coder-7B-Instruct from Alibaba's Qwen team | |
| - **Architecture**: Transformer-based decoder-only model | |
| - **Parameters**: 7B | |
| ### Training Data | |
| The model was fine-tuned on our curated CodeWeave-Instructions dataset, which includes: | |
| - 50K Python programming exercises | |
| - 30K JavaScript development tasks | |
| - Code review and refactoring examples | |
| ## Usage | |
| ```python | |
| from transformers import AutoModelForCausalLM, AutoTokenizer | |
| model = AutoModelForCausalLM.from_pretrained("toolevalxm/CodeWeave-QwenCoder") | |
| tokenizer = AutoTokenizer.from_pretrained("toolevalxm/CodeWeave-QwenCoder") | |
| ``` | |
| ## Evaluation Results | |
| | Benchmark | Score | | |
| |-----------|-------| | |
| | HumanEval | 72.5% | | |
| | MBPP | 68.3% | | |
| ## Acknowledgements | |
| We thank the Qwen team for the excellent base model. | |
| **License** | |
| The license for this model is apache-2.0. |