Overview

This model is a sharded version of the Mistral-7B-Instruct-v0.1 Large Language Model, designed to be usable even with limited CPU memory. It is an instruction-tuned variant of the original Mistral-7B-v0.1 generative text model, fine-tuned using a diverse set of publicly available conversation datasets.

Key Capabilities

• Instruction Following: Optimized to understand and execute instructions provided within [INST] and [/INST] tags.
• Conversational AI: Excels at generating coherent and contextually relevant responses in dialogue scenarios.
• Efficient Deployment: The sharded nature allows for deployment in environments with memory constraints.
• Advanced Architecture: Incorporates architectural innovations like Grouped-Query Attention and Sliding-Window Attention for improved performance and efficiency.

Instruction Format

To leverage its instruction-following capabilities, prompts should be formatted with [INST] and [/INST] tokens. The first instruction requires a beginning-of-sentence ID, while subsequent instructions do not. The model's generation is terminated by an end-of-sentence token ID.

Good For

• Building chatbots and conversational agents.
• Instruction-based text generation tasks.
• Applications requiring a powerful language model with efficient memory usage.