llava-hf

/

llava-interleave-qwen-0.5b-hf

like 30

Follow

Llava Hugging Face 175

The conversion part is a bit tricky and has to be tailored for each llava model. Currently we don't have a clean and easy to run conversion script, maybe @Xenova can share his code. However note, you may still need to make changes to the code to convert successfully

Happy to receive your suggestions. Currently I plan to finetune llava-interval-0.5b and llava-oneversion and deploy quantization onnx. If possible, please share your code for converting the model to onnx file, and then we could make modifications based on your great work. Now I am confused where to start. I am hearing your good news @Xenova .
Thanks。

hello @Xenova :
Sorry to bother you. I'd like to ask if there are any recent plans to sharing the code of converting the Llava-interleave model to the Onnx file. If not, could you briefly describe the method or process of the conversion?
Thank you

Hehe, I just remembered that @Xenova shared one of the conversion script with me earlier. Here you go, I didn't test it so I hope it works for you

https://colab.research.google.com/drive/1IhC8YOV68cze0XWGfuqSclnVTt_FskUd?usp=sharing

Hi @RaushanTurganbay ,
Thank for your sharing. Based on your above the tip,I have succeed in converting three models of the LLava model into onnx files.
Thanks.