metadata
language: code
tags:
- code-generation
- python
- fine-tuned
- qlora
base_model: Qwen/Qwen2.5-Coder-0.5B-Instruct
datasets:
- iamtarun/python_code_instructions_18k_alpaca
license: mit
Qwen2.5-Coder-0.5B Python Fine-tuned
Fine-tuned version of Qwen/Qwen2.5-Coder-0.5B-Instruct for Python code generation.
Model Details
- Base Model: Qwen/Qwen2.5-Coder-0.5B-Instruct
- Fine-tuning Method: QLoRA (4-bit quantization + LoRA adapters)
- Dataset: iamtarun/python_code_instructions_18k_alpaca
- Task: Python code generation from natural language instructions
Training Details
- Training Samples: 16000
- Validation Samples: 1000
- Epochs: 3
- Training Time: N/A
- Final Loss: N/A
Performance
- Syntax Validity: 95.2%
- Pass@1: 54.4%
- Verbosity Reduction: 95%
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("KpRT/qwen-python-finetuned")
tokenizer = AutoTokenizer.from_pretrained("KpRT/qwen-python-finetuned")
prompt = "Write a function to reverse a string"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256)
code = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(code)
Citation
If you use this model, please cite:
@misc{qwen-python-finetuned,
author = {K R T},
title = {Qwen2.5-Coder Python Fine-tuned},
year = {2026},
publisher = {HuggingFace},
url = {https://huggingface.co/KpRT/qwen-python-finetuned}
}