How is this different to onnx-community/Qwen2.5-0.5B?

#1
by Xenova HF staff - opened
ONNX Community org

Hi @Felladrin ! How does this conversion differ from https://huggingface.co/onnx-community/Qwen2.5-0.5B?

ONNX Community org

Hi, @Xenova !
Thanks for asking! It has no difference, as far as I know.
But wasn't added by me (I mean, not manually).
Someone used https://huggingface.co/spaces/onnx-community/convert-to-onnx to convert Qwen2.5-0.5B. They probably didn't see that https://huggingface.co/onnx-community/Qwen2.5-0.5B already existed.
ℹ️ convert-to-onnx checks if the model already exists, but the name needs to match the onnx-community/<model_name>-ONNX pattern.

My suggestion: Delete onnx-community/Qwen2.5-0.5B-ONNX (this model) and rename onnx-community/Qwen2.5-0.5B to onnx-community/Qwen2.5-0.5B-ONNX.
I'm suggesting it because renamed models will redirect. So whoever access onnx-community/Qwen2.5-0.5B will be auto-recirected to onnx-community/Qwen2.5-0.5B-ONNX.

Also, if you have a chance, please update the HF_TOKEN env var from https://huggingface.co/spaces/onnx-community/convert-to-onnx/settings.
Currently, it has my HF token (which has write access to onnx-community), so whenever someone uses the space to convert a model and doesn't use their own HF_TOKEN, it uses the HF_TOKEN from env var, which causes all models created via the space to show as added by me.
Maybe you could set it use your HF token as you're working closer to the ONNX models? (Or maybe create a Bot-user in Hugging Face, and use its HF token instead? Just to make it clear that the model was added via automation?)

Sign up or log in to comment