Mistral releases open-source model Mixtral 8x22B under Apache 2.0 license
Original title:Research Cheaper, Better, Faster, Stronger April 17, 2024 By Mistral team
AI summary
Mistral released the open-source sparse mixture-of-experts model Mixtral 8x22B, with 141B total parameters and only 39B activated, a 64K-token context window, and open under the Apache 2.0 license. The model supports English, French, Italian, German, and Spanish, has native function calling, and the instruct version scores 90.8% on GSM8K maj@8 and 44.6% on Math maj@4; it is now available on la Plateforme.
Key Takeaways
- Mistral released the open-source sparse mixture-of-experts model Mixtral 8x22B,with 141B total…
- 0 license.
- The model supports English,French,Italian,German,and Spanish,has native function calling,and the…
- 8% on GSM8K maj@8 and 44.
- 6% on Math maj@4
Quick Facts
- Published
- Updated
- Source
- Mistral AI:News(网页)
- Category
- Model Releases
- AI score
- 77
官方发布原文给出了活跃参数、上下文窗口、许可证和基准成绩,读者可以据此评估它在开源模型中的成本与性能定位。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Mistral releases open-source model Mixtral 8x22B under Apache 2.0 license. April 17, 2024. https://new.nat6.net/en/news/cmppdcr7f0e4lslv49nngsq93/