This project demonstrates efficient fine-tuning of Qwen 2.5-3B Instruct, a powerful LLM by Alibaba, using LoRA (Low-Rank Adaptation) and the Unsloth framework. It is optimized for low-cost GPUs like NVIDIA T4, using 4-bit quantization to reduce memory footprint.
- Fine-tune Qwen 2.5-3B efficiently on consumer-grade hardware (T4 GPU).
- Utilize LoRA for parameter-efficient tuning.
- Apply Unsloth’s optimized training engine for faster training and inference.
- Enable support for long-context reasoning and instruction-following tasks.
- Keep the solution simple, modular, and compatible with Colab or Kaggle environments.