Error on
Using command in terminal
python3 -m mlx_vlm.generate --model mlx-community/Qwen2-VL-7B-Instruct-4bit --image 'image.jpg' --prompt 'describe this image'
Got an error
ValueError: Received parameters not in model: language_language_model.model.language_model.lm_head.weight language_language_model.model.language_model.lm_head.biases language_language_model.model.language_model.lm_head.scales.
There is currently a bug in the mlx-vlm method which sanitizes the weights; I created an issue and am creating a PR
Edit: looks like the author just created a PR containing the fix
https://github.com/Blaizzy/mlx-vlm/pull/62
There is currently a bug in the
mlx-vlmmethod which sanitizes the weights; I created an issue and am creating a PREdit: looks like the author just created a PR containing the fix
https://github.com/Blaizzy/mlx-vlm/pull/62
So we need wait for new model or update mlx ?
Just need to update your version of mlx-vlm!