Appreciation for the Influence on VLM and Inquiry about LLM Foundation

FreedomIntelligence / LongLLaVA

LongLLaVA: Scaling Multi-modal LLMs to 1000 Images Efficiently via Hybrid Architecture

177 stars 11 forks source link

Appreciation for the Influence on VLM and Inquiry about LLM Foundation #2

Closed CuriousCat-7 closed 1 month ago

CuriousCat-7 commented 1 month ago

Hey! I'm really impressed by your work on VLM. It has a huge impact for the MLLM area. I wonder about the LLM in your VLM. Did you train it from 0? Or use a pre - trained model like "ai21labs/Jamba - v0.1"? Thanks!

wangxidong06 commented 1 month ago

we use the pre - trained model "ai21labs/Jamba - v0.1"