chatglm_finetuning

Why we included this project

ChatGLM teams that want to adapt the model to their own data get a working training setup here, not just code fragments. The repo covers full-parameter fine-tuning and the parameter-efficient options (LoRA, QLoRA, AdaLoRA, IA3, P-Tuning v2), so you can pick the method that fits your GPU budget. Shell scripts walk through dataset preparation, training, and inference, and the conversation examples include plain chat, tool/function calling, and observation turns, which is useful if you are tuning for agent-style behavior. Since it builds on the deep_training library and standard Hugging Face tooling, the code stays close to familiar PyTorch and transformers workflows. If you are already working with the ChatGLM line and want one place to try different fine-tuning strategies, this is a solid reference and starting point.