Phi-4 multi modal fine tuning addition

[lab176344]

Would like to add support for phi-4 fine tuning, https://huggingface.co/microsoft/Phi-4-multimodal-instruct, would like to know if no one is working on this, if some one is interested would love to collaborate as well

[SkalskiP]

Thanks for your interest in the Phi-4 fine tuning support. This model was released today, and as far as I know, no one on our team is actively working on adding it yet. We welcome contributions in this area. Let me know if you need any assistance or guidance from our side.

[SkalskiP]

Hi @lab176344 👋🏻 Exciting! It would be great if you could open a draft PR with the code you already have. I can't wait to take a look.

[lab176344]

Created one https://github.com/roboflow/maestro/pull/181/files please take a look, My plan is to do a training run with and without lora, qlora and i will create a colab cookbook and then will move it from draft

[lab176344]

After this would love to add https://cohere.com/blog/aya-vision if no one is working on it :P

[SkalskiP]

Have you had a chance to try AyaVision already? I've been a bit sick for the past few days and haven't had a chance to test this model yet.

[lab176344]

Hey, Wishes to get better soon. I was doing just a basic vibe check here https://github.com/lab176344/vlm-notebookss/blob/main/notebooks/inference/exploration-aya-vl-8b.ipynb, will explore more over the week end and update the notebook