0

Vicuna 13B

Open weights

LMSYS's LLaMA-13B fine-tune on ~70K ShareGPT conversations - the model that introduced "LLM-as-judge" evaluation and seeded Chatbot Arena.

Publisher
LMSYS Org
Family
Vicuna
Params
13B
Context
2K
Input
Output
Released
Mar 2023

Cite

Notes

Only stored in your browser.

Context
2K

Reported on 4 evals across 3 domains - best 70.4% on AlpacaEval · #18 of 21

Reported eval scores

4

Introduced in

paperVicuna: An Open-Source Chatbot Impressing GPT-4 with 90% ChatGPT Quality