TheCluster/Qwen3.5-122B-A10B-Heretic-v2-MLX-mixed-3.8bit

<div align="center"><img width="400px" src="https://qianwen-res.oss-accelerate.aliyuncs.com/logo_qwen3.5.png"></div> <div style="text-align:center; margin-bottom:12pt; font-size:11pt">If you like my work, you can <a href="https://donatr.ee/thecluster/">support me</a><br/></div>

Qwen3.5-122B-A10B Heretic V2

Quality: quantized (*mixed quants per tensor, group size: 32, 3.81 bpw*)

Quantization: 3 bit for experts, 4 bit for shared experts and attention layers, 6 bit for embeddings & head.

This is an uncensored version of Qwen/Qwen3.5-122B-A10B, made using Heretic v1.2.0 with Multi-directional refusal supression

Abliteration metrics

Sampling Parameters:

• —I suggest using the following sets of sampling parameters depending on the mode and task type:
• —Thinking mode for general tasks: temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0
• —Thinking mode for precise coding tasks (e.g., WebDev): temperature=0.6, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0
• —Instruct (or non-thinking) mode for general tasks: temperature=0.7, top_p=0.8, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0
• —Instruct (or non-thinking) mode for reasoning tasks: temperature=1.0, top_p=1.0, top_k=40, min_p=0.0, presence_penalty=2.0, repetition_penalty=1.0
• —For supported frameworks, you can adjust the presence_penalty parameter between 0 and 2 to reduce endless repetitions. However, using a higher value may occasionally result in language mixing and a slight decrease in model performance.

Source

This model was converted to MLX format from `coder3101/Qwen3.5-122B-A10B-heretic-v2` using mlx-vlm version 0.4.4.