Sao10K/72B-Qwen2.5-Kunou-v1 Overview

Sao10K/72B-Qwen2.5-Kunou-v1 is a 72.7 billion parameter language model built on the Qwen2.5 architecture, developed by Sao10K. This model is primarily designed as a generalist with a strong emphasis on roleplay and creative instruction. It represents a refinement of previous models like L3-70B-Euryale-v2.2, utilizing a significantly cleaner and improved dataset for its training.

Key Capabilities & Features

• Generalist Performance: Aims to perform well across a broad range of tasks.
• Roleplay & Creative Instruction: Specifically fine-tuned with datasets like mimi-superfix-RP-filtered-fixed.json to excel in creative writing and role-playing scenarios.
• Extensive Context Window: Supports a context length of 131072 tokens, enabling long and complex interactions.
• Refined Training Data: Benefits from a dataset that is a cleaned and improved version of those used in earlier Euryale and Stheno models.
• Optimized for ChatML: Recommended to be used with ChatML prompt format, a temperature of 1.1, and min_p of 0.1 for optimal results.

Ideal Use Cases

• Creative Writing & Story Generation: Generating narratives, character dialogues, and creative content.
• Role-playing Applications: Engaging in detailed and immersive role-play scenarios.
• General Conversational AI: Handling diverse conversational prompts as a versatile chatbot.
• Instruction Following: Executing complex instructions, particularly those with a creative or narrative component.