Tess-10.7B-v1.5b: A General-Purpose LLM

Tess, short for Tesoro (Italian for "Treasure"), is a series of general-purpose large language models developed by migtissera. The Tess-10.7B-v1.5b variant is a 10.7 billion parameter model, specifically trained on the robust SOLAR-10.7B base architecture.

Key Capabilities

• General-Purpose Language Understanding: Designed to handle a wide array of natural language processing tasks.
• Solid Foundation: Built on the established SOLAR-10.7B base, providing a strong architectural backbone.
• Context Handling: Supports a context length of 4096 tokens, suitable for moderate-length interactions and document processing.

Good For

• Versatile Applications: Ideal for developers seeking a flexible LLM for various text generation, summarization, and conversational AI needs.
• Base Model for Fine-tuning: Can serve as an effective starting point for further fine-tuning on specific domain data or tasks.
• Exploratory Development: Suitable for experimenting with LLM capabilities in a general context.

Prompt Format

The model utilizes a straightforward prompt format:

SYSTEM: <ANY SYSTEM CONTEXT>
USER: 
ASSISTANT: