InternLM / xtuner

An efficient, flexible and full-featured toolkit for fine-tuning LLM (InternLM2, Llama3, Phi3, Qwen, Mistral, ...)
https://xtuner.readthedocs.io/zh-cn/latest/
Apache License 2.0
3.72k stars 302 forks source link

convert_xtuner_weights_to_llava.py and convert_xtuner_weights_to_hf.py can support llava-next model? #642

Open awzhgw opened 4 months ago

awzhgw commented 4 months ago

i has use https://github.com/hhaAndroid/xtuner/tree/refactor_llava train llava_1.6_phi3_8B model .

but ,it cannot convert to office llava model use convert_xtuner_weights_to_llava.py

can you help me ?

641

@LZHgrla @hhaAndroid

help me ? thanks

awzhgw commented 4 months ago

@hhaAndroid @LZHgrla i use config file is :https://github.com/hhaAndroid/xtuner/blob/refactor_llava/xtuner/configs/llava/phi3_mini_chat/llava_phi3_mini_4k_instruct_full_clip_vit_large_lora_p14_336_anyres_e1_gpu8_internvl_finetune.py

i has train model ..but cannot convert to office llava format model ,can you help me ?

the convert_xtuner_weights_to_llava.py and convert_xtuner_weights_to_hf.py can support llava-next model ?

pppppM commented 4 months ago

The script convert_xtuner_weights_to_llava.py is only compatible with the main branch. The refactor_llava branch is still under development, and frequent changes may occur.