Overview

viethq188/LeoScorpius-7B-Chat-DPO is a 7 billion parameter language model developed by viethq188. This model has been fine-tuned using Direct Preference Optimization (DPO) with the Nectar dataset, aiming to enhance its conversational capabilities and response quality.

Key Capabilities

• High Performance: Achieved a notable ranking on the Hugging Face Open LLM Leaderboard, placing 3rd overall and 1st among 7B parameter models at the time of its update (December 14th, 2023).
• DPO Fine-tuning: Leverages DPO for improved alignment and preference-based learning, which typically results in more helpful and harmless outputs.
• Chat-Optimized: Specifically designed and fine-tuned for chat and conversational applications.

Usage

This model utilizes an Alpaca-style template for prompts, making it straightforward to integrate into existing workflows that support this format. The recommended template structure is:

{system}
### Instruction:
{prompt}

### Response:

Good For

• General-purpose conversational AI.
• Applications requiring a highly-ranked 7B model for chat interactions.
• Developers looking for a DPO-tuned model with strong benchmark performance.