← back to catalog · registered 2026-11-20 01:04

chogeha/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-oQ2h-fp16-mtp

chogeha 35B MoE
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/chogeha%2FQwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-oQ2h-fp16-mtp"
Response includes
  • classification m3
  • files 14
  • author_summary 3 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
Downloads · 30-day
0
Likes
0
Model age
2d ago
created 2026-11-20

Metadata

Tags
mlx safetensors qwen3_5_moe oq quantized mtp 2-bit region:us

Related

Total size
13.9 GB
Files
14
Quantizations
1
Registered
2026-11-20 01:04
Last updated on HF
2026-11-20 00:54

Files by quantization

Auxiliary files 14 files 13.9 GB
model-00002-of-00004.safetensors 4.81 GB 33c9a62c download
model-00003-of-00004.safetensors 4.74 GB 4e90ef6c download
model-00001-of-00004.safetensors 3.18 GB c4cb3cf1 download
model-00004-of-00004.safetensors 1.20 GB 4f6d9b18 download
tokenizer.json 19.1 MB 6f32ce20 download
model.safetensors.index.json 201 KB 070aefc0 download
config.json 65.1 KB b36cd700 download
chat_template.jinja 11.7 KB 177eacb1 download
oq_imatrix_report.json 6.34 KB c0569e60 download
.gitattributes 1.53 KB 52373fe2 download
README.md 1.13 KB c4b758f4 download
tokenizer_config.json 1.13 KB c487bad4 download
preprocessor_config.json 390 B 2ea84a43 download
generation_config.json 226 B 2dd033e0 download

README current version from Hugging Face


library_name: mlx
tags:

  • mlx
  • oq
  • quantized
  • mtp

Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-oQ2h-fp16-mtp

This model was quantized using oQ (oMLX v0.8.0) mixed-precision quantization.

The effective average is 22.866 bits/weight for the complete checkpoint.

Quantization and Bit Distribution

Storage format Logical weights Share of total Tensor storage Effective bits/weight
Affine 4-bit, group 64 0.805B 15.391% 0.422 GiB 4.50
Affine 8-bit, group 128 0.003B 0.060% 0.003 GiB 8.25
Affine 8-bit, group 64 1.044B 19.962% 1.034 GiB 8.50
FP16 0.064B 1.224% 0.119 GiB 16.00
FP32 0.447B 8.535% 1.664 GiB 32.00
U32 2.869B 54.829% 10.687 GiB 32.00
Total 5.232B 100% 13.929 GiB 22.866

Storage Breakdown

Component Logical weights Storage
Language model 3.941B 12.666 GB / 11.796 GiB
Vision 0.447B 1.786 GB / 1.664 GiB
MTP head 0.845B 0.503 GB / 0.468 GiB
Total 5.232B 14.956 GB / 13.929 GiB
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration