Mistral - Le Chat

Mistral - Le Chat
chat.mistral.ai

Description

Mistral is an innovative Large Language Model (LLM). Key features include a decoder-based architecture, Sliding Window Attention, Grouped Query Attention for efficient inference, and a Byte-fallback BPE tokenizer. Mistral-7B-v0.1 and its instruction fine-tuned variant, Mistral-7B-Instruct-v0.1, are designed for diverse applications like chat-based inference. Available under Apache 2.0 license, these models can be accessed via the Huggingface Hub.

Key Features

  • Sliding Window and Grouped Query Attention mechanisms.
  • Broad application range, including chat-based inference.
  • Available for customization and use through Huggingface Hub.

Pricing

Free