Can’t make Qwen/Qwen2.5-VL-3B-Instruct model work on serverless
Qwen/Qwen2.5-VL-3B-Instruct
Anybody able to make it work?
When it will be supported?
5 Replies
@flash-singh can you help?
have you tried using dev image for workers or tried to build your own? i am asking cause i want to run those models too, but see that there is an issue with transformers version
I am nooby not sure how to use dev image. I know that if I use dev transformers it works though. I tried locally
ahh, awesome! i will try building an image then
i am also new to runpod, but willing to give it a try
will ping you if i succeed 🙂
btw, is 3b equally impressive as 72b in it's own weight class? 72b basically has the best ocr i ever saw, better than gemini and claude
Thank you will wait if you can make it work. I tried to extract grocery receipts data. 3B is working fabulous I am stunned by its performance beating everything in market. Can’t believe 3B model can do that.