NousResearch/Yarn-Mistral-7b-64k
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kPublished:Oct 31, 2023License:apache-2.0Architecture:Transformer0.1K Open Weights Cold

NousResearch/Yarn-Mistral-7b-64k is a 7 billion parameter language model developed by NousResearch, based on the Mistral-7B-v0.1 architecture. It has been further pretrained using the YaRN extension method to support an extended context window of 64k tokens. This model is specifically optimized for long-context tasks, demonstrating improved perplexity on longer sequences while maintaining strong performance on short-context benchmarks.

Loading preview...