Doctor Kunou 72B: A Merged Language Model

Doctor Kunou 72B is a 72.7 billion parameter model developed by Praxis Maldevide, leveraging a unique normalized denoised fourier interpolation method. This technique combines the strengths of several specialized models on top of the Qwen2.5-72B base.

Key Merged Components:

• moonshotai/Kimi-Dev-72B: Contributes to the input layer, enhancing initial processing.
• pfnet/Preferred-MedLLM-Qwen-72B: Integrates deep domain medical knowledge.
• Sao10K/72B-Qwen2.5-Kunou-v1: Influences the output layer, contributing to instruct-based responses.

Distinctive Capabilities:

• Enhanced Coherence: The model is noted for its very coherent outputs.
• Creative Integration: Successfully combines the creativity found in Kunou models.
• Improved Prompt Following: Demonstrates better adherence to given prompts.
• Medical Domain Knowledge: Incorporates specialized medical understanding from Preferred-MedLLM-Qwen-72B.

Ideal Use Cases:

This model is particularly well-suited for applications requiring a blend of creative text generation, precise instruction adherence, and a foundational understanding of medical concepts. Its unique merging approach aims to deliver a balanced performance across these areas.