hiyouga / LLaMA-Factory

Unify Efficient Fine-Tuning of 100+ LLMs
Apache License 2.0
25.26k stars 3.13k forks source link

关于基座模型和对话模型的疑问 #4457

Closed VirgilG72 closed 2 days ago

VirgilG72 commented 4 days ago

我是这方面的小白,想问一下以下几个问题: ①如果我想以lora微调的形式训练基座模型Qwen/Qwen2-7B,应该执行LLaMA-Factory的哪个微调命令? ②训练完基座模型后,我想基于这个基座模型,以lora微调的形式训练对话模型Qwen/Qwen2-7B-Instruct,应该执行LLaMA-Factory的哪个微调命令?

我的思路是: ①先执行llamafactory-cli train examples/train_lora/llama3_lora_pretrain.yaml中的model_name_or_path改为Qwen/Qwen2-7B,输出路径为Qwen/Qwen2-7B-pretrained ②然后llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml中的model_name_or_path改为Qwen/Qwen2-7B-pretrained 这样可行吗?