Prince Canuma
Prince Canuma
Thanks @alexgusevski! This model is not yet supported but has been on my radar and I plan to add it. However, if you send a PR I would be more...
No worries! I will add support for it :)
Hey @not-lain Sounds like a great idea! But I have a question: What is the advantage of using AutoModel, compared to Load(the current approach)?
Hey @not-lain I'm highly considering an AutoModel approach but for now you can use `load_model` or import the model class as we do here: https://github.com/Blaizzy/mlx-vlm/blob/main/mlx_vlm/models/smolvlm/smolvlm.py
Hey @pavelgur Could you share a reproducible script, dataset and model? The trainer is definetly due to a overhaul. The initial version has some limitation in batch size and multi-image...
I changed the logic to a non-padding, please check it out and let me know if it works for you #227
Still debugging, there is a bigger discussion. Will check in detail during the weekend.
Yes, you can use --resize-shape. The image is probably too big or the model has architecture that has some bottlenecks (i.e, Llama-Vision cross attention)
Hey @Garry-TI Awesome work! Could you open a PR with the progress you made and we will help you finish it. Or you could wait a bit longer and will...