# Visual-Language Instruction Fine-tuning

A fine-tuned version of the Qwen2-VL-7B model, trained using Unsloth and Huggingface TRL library, achieving 2x inference speed improvement.