0

UltraLM 13B

Open weights

Tsinghua / OpenBMB's 13B LLaMA-1 fine-tune trained on the UltraChat 1.5M multi-turn synthetic dialogue corpus.

Publisher
OpenBMB
Family
Ultralm
Params
13B
Context
2K
Input
Output
Released
Jun 2023

Cite

Notes

Only stored in your browser.

Attribution

Benchmark scores
alpaca-eval
Attribution policy →
Context
2K

Reported on 2 evals - best 80.6% on AlpacaEval · #14 of 21

Reported eval scores

2

Introduced in

paperEnhancing Chat Language Models by Scaling High-quality Instructional Conversations