when would you show how to fintune the Qwen2.5-VL model?

[MengHao666]

No description provided.

[hiyouga]

It is the same as the Qwen2-VL model: https://github.com/QwenLM/Qwen2.5-VL/tree/35ba6e18636510de4bf8d4a7caaca3f4f5163a84?tab=readme-ov-file#training

[2U1]

I've made a code for finetuning Qwen2.5-VL
You could use this

https://github.com/2U1/Qwen2-VL-Finetune

[SFTJBD]

It is the same as the Qwen2-VL model: https://github.com/QwenLM/Qwen2.5-VL/tree/35ba6e18636510de4bf8d4a7caaca3f4f5163a84?tab=readme-ov-file#training

Thank you so much for the suggestions. However, I'm wondering, does the data template in the qwen2vl_full_sft.yaml file not need any changes? Furthermore, is there no specific data format requirement for fine-tuning on 2.5vl on grounding tasks?

[echoht]

有人用llama_factory微调成功吗？ transformers版本不兼容怎么解决哈？

[zhangfaen]

https://github.com/zhangfaen/finetune-Qwen2.5-VL

[MengHao666]

有人用llama_factory微调成功吗？ transformers版本不兼容怎么解决哈？

我的可以成功,requirement 如下，另外我的其他环境
deepspeed==0.8.2 & xformers==0.0.28 & flash-attn==2.6.3，pytorch2.4.0
cuda版本：cuda12.4，python3.10.13

transformers==4.48.2
datasets==3.2.0
accelerate==1.2.1
peft==0.12.0
trl==0.9.6
openlm-hub
tokenizers>=0.19.0,<=0.21.0
gradio>=4.38.0,<=5.12.0
pandas>=2.0.0
scipy
einops
sentencepiece
tiktoken
protobuf
uvicorn
pydantic
fastapi
sse-starlette
matplotlib>=3.7.0
fire
packaging
pyyaml
numpy<2.0.0
av
librosa
tyro<0.9.0

[echoht]

有人用llama_factory微调成功吗？ transformers版本不兼容怎么解决哈？

我的可以成功,requirement 如下，另外我的其他环境 deepspeed==0.8.2 & xformers==0.0.28 & flash-attn==2.6.3，pytorch2.4.0 cuda版本：cuda12.4，python3.10.13

transformers==4.48.2
datasets==3.2.0
accelerate==1.2.1
peft==0.12.0
trl==0.9.6
openlm-hub
tokenizers>=0.19.0,<=0.21.0
gradio>=4.38.0,<=5.12.0
pandas>=2.0.0
scipy
einops
sentencepiece
tiktoken
protobuf
uvicorn
pydantic
fastapi
sse-starlette
matplotlib>=3.7.0
fire
packaging
pyyaml
numpy<2.0.0
av
librosa
tyro<0.9.0

qwenvl2.5啊，transformers版本不对吧。

[MengHao666]

有人用llama_factory微调成功吗？ transformers版本不兼容怎么解决哈？

我的可以成功,requirement 如下，另外我的其他环境 deepspeed==0.8.2 & xformers==0.0.28 & flash-attn==2.6.3，pytorch2.4.0 cuda版本：cuda12.4，python3.10.13

transformers==4.48.2
datasets==3.2.0
accelerate==1.2.1
peft==0.12.0
trl==0.9.6
openlm-hub
tokenizers>=0.19.0,<=0.21.0
gradio>=4.38.0,<=5.12.0
pandas>=2.0.0
scipy
einops
sentencepiece
tiktoken
protobuf
uvicorn
pydantic
fastapi
sse-starlette
matplotlib>=3.7.0
fire
packaging
pyyaml
numpy<2.0.0
av
librosa
tyro<0.9.0

qwenvl2.5啊，transformers版本不对吧。

我的环境可以，llamafactory也用最新版