DavidAU/Qwen3.6-27B-F451-AND-TRI-Polar-Ultra-Pro-Writer-Uncensored-Heretic

VISIONConcurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 13, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

DavidAU/Qwen3.6-27B-F451-AND-TRI-Polar-Ultra-Pro-Writer-Uncensored-Heretic is a 27 billion parameter Qwen3.6-based model fine-tuned by DavidAU. This instruction-tuned model focuses on enhancing core intelligence, demonstrating improved performance across various benchmarks compared to its base Qwen3.6-27B-Instruct and Qwen3.5-27B-Instruct counterparts. It is optimized for general intelligence tasks, making it suitable for applications requiring robust reasoning capabilities.

Loading preview...

Model Overview

DavidAU/Qwen3.6-27B-F451-AND-TRI-Polar-Ultra-Pro-Writer-Uncensored-Heretic is a 27 billion parameter language model, fine-tuned by DavidAU using Unsloth on local hardware. This model is specifically engineered to elevate its core intelligence, building upon the Qwen3.6 architecture.

Key Capabilities & Enhancements

This fine-tuned version demonstrates significant improvements over its base models, Qwen3.6-27B-Instruct and Qwen3.5-27B-Instruct, particularly in core intelligence benchmarks. The model was tested in "Instruct" mode, which generally yields better results with the testing harness.

  • Enhanced Core Intelligence: The primary goal of this fine-tune was to boost the model's fundamental reasoning and understanding capabilities.
  • Benchmark Performance: It shows higher scores across various benchmarks (arc, arc/e, boolq, hswag, obkqa, piqa, wino) compared to the non-heretic base models. For instance, its mxfp8 scores are 0.665 for arc and 0.830 for arc/e, surpassing the Qwen3.6-27B-Instruct's 0.637 and 0.798 respectively.
  • Instruction-Tuned: Optimized for instruction-following tasks, making it responsive and effective in conversational or command-based scenarios.

Use Cases

This model is well-suited for applications requiring a strong foundation in general intelligence and robust instruction following. Its improved benchmark performance suggests suitability for tasks demanding accurate reasoning and comprehensive understanding.