Tess-XS-v1.2 Overview

Tess-XS-v1.2 is an experimental 7 billion parameter general-purpose large language model developed by migtissera. This version is built upon the Mistral-7B base architecture and is part of the broader Tessoro/Tessoso series of LLMs. It is designed for a wide range of text generation and comprehension tasks.

Key Characteristics

• Base Model: Utilizes the robust Mistral-7B architecture.
• Parameter Count: Features 7 billion parameters, balancing performance with computational efficiency.
• Prompt Format: Employs a straightforward SYSTEM: <ANY SYSTEM CONTEXT>\nUSER:\nASSISTANT: format for clear instruction following.

Important Note

Users should be aware that Tess-XS-v1.2 is an experimental and deprecated version. For stable and updated performance, it is recommended to use the successor model, Tess-XS-v1.3-yarn-128K, which offers an extended context length of 128K tokens.