Open XavierElon opened 2 weeks ago
@Patrick-Erichsen @sestinj
packages/llm-info
@sestinj adding your other comment on replacing llama 3.1 with 3.2 here:
Llama3.2 was more focused on multi-modal and edge (small) models, so they didn’t turn out to be as good for code. The one of comparable size to llama3.1 8b that we would recommend (llama 3.2 11b) does much worse on coding benchmarks: https://ai.meta.com/blog/meta-llama-3-1/ https://ai.meta.com/blog/llama-3-2-connect-2024-vision-edge-mobile-devices/ Meta AIMeta AI Llama 3.2: Revolutionizing edge AI and vision with open, customizable models Today, we’re releasing Llama 3.2, which includes small and medium-sized vision LLMs, and lightweight, text-only models that fit onto edge and mobile devices.
TODO
autodetectTemplateType
function to return correct template type per modelModelName
in index.d.ts config_schema.json
, config_rc_schema.json
for providers:
Validations
Problem
Meta just released the 3.2 with Lightweight 1B & 3B and Multimodal 11B & 90B. It would be nice to have the option to use this as one of the models.] https://www.llama.com/
Solution
No response