search

TinyLLaVA / TinyLLaVA_Factory

A Framework of Small-scale Large Multimodal Models
https://arxiv.org/abs/2402.14289
Apache License 2.0
630 stars 60 forks source link

Any plans on training a new model based on Phi-3? #52

Open TheBobbyliu opened 6 months ago

baichuanzhou commented 6 months ago

My initial experiments with Phi-3-mini reached 40.0 on MMMU val set. Stay tuned.