/Mistral7b_pdf_chatting

Primary LanguageJupyter Notebook

Mistral 7B Overview

Open In Colab

Mistral 7B is a state-of-the-art language model developed by the Mistral AI team.

Updates :

  • solve cannot import name 'LangchainEmbedding' issue.
  • build simple gui to chat with model
  • using metadate fun to display source of response

Key Features

  • Size: 7.3B parameters.
  • Performance: Outperforms Llama 2 13B across all benchmarks and surpasses Llama 1 34B in many.
  • Versatility: Approaches CodeLlama 7B in code-related tasks while excelling in English language tasks.
  • Attention Mechanisms: Employs Grouped-query attention (GQA) for rapid inference and Sliding Window Attention (SWA) for efficient handling of extended sequences.
  • Licensing: Released under the Apache 2.0 license, ensuring unrestricted usage.

Availability

  • Download & Implementation: The model is available for download and can be integrated using their reference implementation.
  • Deployment: Compatible with various cloud platforms.
  • HuggingFace: Also available on HuggingFace.

Performance Benchmarks

Mistral 7B has been benchmarked against the Llama 2 model family, demonstrating significant superiority in various tests. The model particularly shines in:

  • Commonsense Reasoning
  • World Knowledge
  • Reading Comprehension
  • Mathematics
  • Code-related tasks