LLM-Finetune

Why we included this project

LLM-Finetune is a practical walkthrough of instruction fine-tuning for Qwen2, GLM4, and Qwen2-VL, with training and inference scripts for text classification and named-entity recognition plus a separate multimodal path for the vision-language model. The dataset download steps are spelled out, and Jupyter notebooks and linked Chinese write-ups explain each stage, so someone new to supervised fine-tuning can reproduce the whole workflow and check the results with the prediction scripts. It is not a production framework or a library you would embed, but as a concrete, copyable example of how to fine-tune these specific models it is a useful reference.