Author(s): Dr. Mandar Karhade, MD. PhD.
Originally published on Towards AI.
This model allows long sequences of up to 8K tokens completely free
Salesforce has a reputation for doing good work in the open-source world, including their image encoders models like blip and blip2 (ref: https://huggingface.co/spaces/Salesforce/BLIP2). Now Salesforce has released a small but mighty model that could just fit your needs.
LAVIS – A One-stop Library for Language-Vision Intelligence – LAVIS/projects/blip2 at main · salesforce/LAVIS
github.com
I know from personal experience and many users of OpenAI’s models or LLaMA and its variants that the context length of 2K tokens was just too small. Especially if you are using project guidance or Kor to allow specific instructions to be part of the context, you might end up… Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI