| license: apache-2.0 | |
| base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0 | |
| tags: | |
| - trl | |
| - sft | |
| - generated_from_trainer | |
| - lora | |
| - python-coder | |
| model-index: | |
| - name: TinyLlama-Python-Coder | |
| results: [] | |
| # TinyLlama-Python-Coder | |
| This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0) on the `iamtarun/python_code_instructions_18k_alpaca` dataset using QLoRA. | |
| ## Model description | |
| This model is specialized in generating Python code based on natural language instructions. It was fine-tuned with 4-bit quantization to remain efficient while improving its ability to follow coding tasks. | |
| ## Intended uses & limitations | |
| - **Intended Use:** Assisting with Python code generation, explaining algorithms, and boilerplate creation. | |
| - **Limitations:** As a 1.1B parameter model, it may struggle with highly complex logic compared to larger models like Llama-3 or GPT-4. | |
| \n## Training and evaluation data | |
| - **Dataset:** `iamtarun/python_code_instructions_18k_alpaca` (subset of 2,000 examples used for this demo). | |
| - **Format:** TinyLlama Chat Template. | |
| ## Usage | |
| ```python | |
| from transformers import pipeline | |
| pipe = pipeline(\"text-generation\", model=\"monusharma21/TinyLlama-Python-Coder\") | |
| prompt = \"<|system|>\nYou are an expert Python programmer.\n<|user|>\nWrite a function to calculate Fibonacci numbers.\n<|assistant|>\" | |
| print(pipe(prompt, max_new_tokens=200)[0]['generated_text']) | |
| ``` | |