Mistral-7B-Instruct-v0

A high-performance 7.3 billion parameter language model engineered for superior instruction following and efficiency.

Mistral-7B-Instruct-v0.1 sets a benchmark for compact models by outperforming Llama 2 13B on all standard metrics. It utilizes Grouped-query attention (GQA) for faster inference and Sliding Window Attention (SWA) to handle longer sequences with ease. This model is fine-tuned specifically for conversation, making it a reliable choice for developers building chat interfaces or automated agents. Its Apache 2.0 license ensures flexibility for both commercial and research applications.

https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1