0-hero/Matter-0.1-7B-boost-DPO-preview

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Mar 21, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The 0-hero/Matter-0.1-7B-boost-DPO-preview is a 7 billion parameter language model developed by 0-hero, fine-tuned from a Mistral 7B base using Direct Preference Optimization (DPO). It is specifically trained on the curated Matter dataset, which analyzes over 6 billion tokens, and features native support for function calling. This model is designed for conversational AI applications requiring structured tool use and adherence to the ChatML prompt format.

Loading preview...

Matter 7B 0.1 Boost DPO Preview

This model, developed by 0-hero, is a 7 billion parameter language model based on the Mistral 7B architecture. It has been fine-tuned using Direct Preference Optimization (DPO) on the proprietary Matter dataset, which was curated from over 35 datasets and involved analyzing more than 6 billion tokens. The model is designed to follow the ChatML prompt format for conversational interactions.

Key Capabilities

  • Function Calling: The model natively supports function calling, enabling it to interact with external tools and APIs. It uses specific tokens (<|begin_func|>, <|end_func|>, <|begin_func_response|>, <|end_func_response|>) to delineate function calls and their responses within the conversation.
  • DPO Fine-tuning: Leverages Direct Preference Optimization for improved instruction following and response quality, building upon the Matter 7B Boost base model.
  • Curated Training Data: Benefits from training on the extensive and diverse Matter dataset, enhancing its general understanding and specific capabilities.

Good For

  • Applications requiring robust function calling capabilities.
  • Building AI assistants that need to interact with external systems or retrieve specific information.
  • Use cases where adherence to the ChatML prompt format is preferred.

Finetunes
2 models