Overview

Locutusque/Hyperion-2.1-Mistral-7B is a 7 billion parameter language model, representing a further fine-tuned version of the Hyperion-2.0-Mistral-7B base. Developed by Locutusque, this iteration focused on evaluating the impact of a higher learning rate during fine-tuning, resulting in observed slight performance gains over its predecessor. The model maintains a context length of 8192 tokens.

Key Characteristics

• Architecture: Based on the Mistral-7B family.
• Parameter Count: 7 billion parameters.
• Compliance: Noted for its high compliance, meaning it will respond to any request without refusal.
• Performance: Shows slight performance improvements compared to Hyperion-2.0-Mistral-7B.

Use Cases & Considerations

• Direct Response Applications: Its high compliance makes it suitable for scenarios where direct answers to all prompts are desired, without refusal mechanisms.
• Enterprise Deployment: For enterprise-level deployment, the developer recommends aligning the model using DPO (Direct Preference Optimization) due to its highly compliant nature.
• Further Development: This model is part of an ongoing series, with more checkpoints planned for future release.

Available Quantizations

• ExLlamaV2: bartowski/Hyperion-2.1-Mistral-7B-exl2
• GGUF: bartowski/Hyperion-2.1-Mistral-7B-GGUF
• AWQ: solidrust/Hyperion-2.1-Mistral-7B-AWQ