Skip to content

LLM-Brief

Sections
Back to stories
Models/

Mistral and NVIDIA Release NeMo 12B to Replace Mistral 7B

Mistral AI and NVIDIA have launched Mistral NeMo, a 12-billion parameter model featuring a 128k context window. Designed as a drop-in replacement for Mistral 7B, it utilizes a new 'Tekken' tokenizer for improved efficiency across 100+ languages.

ORIGINAL SOURCE

Mistral and NVIDIA Release NeMo 12B to Replace Mistral 7B

Mistral AI and NVIDIA have launched Mistral NeMo, a 12-billion parameter model featuring a 128k context window. Designed as a drop-in replacement for Mistral 7B, it utilizes a new 'Tekken' tokenizer for improved efficiency across 100+ languages.

View at Decrypt