Get the app
← back to catalog · registered 2026-09-05 12:03

pottokao/MiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-NVFP4-AWQ

pottokao Qwen 12B multimodal
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/pottokao%2FMiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-NVFP4-AWQ"
Response includes
  • classification m1
  • files 15
  • benchmarks 11 entries
  • hub_downloads_all_time 285
  • author_summary 6 models
  • readme_text full
Try next
GET/models?author=pottokao
All 6 models by pottokao
 GET/models?family=qwen&method=m1
Sibling qwen models · same method
 GET/methods/m1
What "m1" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
How we label → About M1 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
285
Likes
3
Model age
6w ago
created 2026-09-05
Downloads over time
Now558→from0↑0%
02054096140 on Sep 5558 on Oct 20SepOct
Sep 5 → Oct 20 · 46 snapshots · spans 45 days

Benchmarks

Portrait before abliteration
Benchmarks of the base model as it stood before the refusal-removal operation. Compare with the numbers above to see what the operation cost.
Benchmark Score Source
Entertainment 1.4 UGI
Hazardous 4.7 UGI
Natural Intelligence 24.96 UGI
Political lean -21.2% UGI
Sensitive-Info 25.6 UGI
SocPol 2.2 UGI
UGI 29.57 UGI
Willingness (10) 3.8 UGI
W10-Adherence 1.5 UGI
W10-Direct 6 UGI
Writing 34.95 UGI

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Qwen/Qwen3-VL-32B-Instruct this lineage
pottokao/MiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-NVFP4-AWQ↓ 371

Variants by this author 2 formats · 43K downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
apache-2.0
Tags
transformers safetensors qwen3_vl image-text-to-text nvfp4 awq modelopt abliterated uncensored minimax-h3 text-encoder vllm-omni

Related

Other by pottokao
MiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-GGUF 43K Ornith-1.5-35B-A3B-abliterated-NVFP4-DFlash-GGUF 464 Ornith-1.5-35B-A3B-abliterated-NVFP4-DFlash 452 Ornith-1.5-35B-A3B-abliterated 132 Huihui-Qwen3.6-27B-abliterated-NVFP4-DFlash-MXFP8 32
Same base model
pottokao/MiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-GGUF 43K huihui-ai/Huihui-Qwen3-VL-32B-Instruct-abliterated 2K llmfan46/Qwen3-VL-32B-Instruct-ultra-uncensored-heretic 1K prithivMLmods/Qwen3-VL-32B-Instruct-abliterated-v1 263 pekkAi/Qwen3-VL-32B-Instruct-abliterated-NVFP4-W4A16 125
Total size
14.6 GB
Files
15
Quantizations
1
Registered
2026-09-05 12:03
Last updated on HF
2026-09-05 11:59
Source huggingface.co/pottokao/MiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-NVFP4-AWQ ↗ Base model Qwen/Qwen3-VL-32B-Instruct ↗

Files by quantization

Auxiliary files 15 files 14.6 GB
model-00001-of-00004.safetensors 4.69 GB 0dace0e8 download
model-00002-of-00004.safetensors 4.67 GB bdf4a7a9 download
model-00003-of-00004.safetensors 4.66 GB 88e8da73 download
model-00004-of-00004.safetensors 602 MB 1498735b download
tokenizer.json 10.9 MB 79cb3c78 download
model.safetensors.index.json 162 KB 0926830b download
chat_template.jinja 5.29 KB 2a65e88d download
config.json 1.82 KB 37a8aae0 download
.gitattributes 1.53 KB 52373fe2 download
README.md 1.16 KB 75b30053 download
tokenizer_config.json 693 B 8a48a6c3 download
preprocessor_config.json 390 B 2ea84a43 download
video_preprocessor_config.json 385 B 3ba673a5 download
generation_config.json 255 B 9b16e8d3 download
component_quant_te.json 207 B 4c2e7002 download

README current version from Hugging Face

license: apache-2.0
base_model:

  • Qwen/Qwen3-VL-32B-Instruct
    pipeline_tag: image-text-to-text
    library_name: transformers
    tags:
  • nvfp4
  • awq
  • modelopt
  • abliterated
  • uncensored
  • minimax-h3
  • text-encoder
  • vllm-omni

MiniMax-H3 Text Encoder — Qwen3-VL-32B (abliterated) · NVFP4-AWQ

Quantized text encoder for the MiniMax-H3 video model, built to be served by vLLM-Omni.

  • Base: abliterated Qwen3-VL-32B-Instruct
  • Layers: 64 → 50 (H3 only consumes the hidden state after layer 50; lm_head and the final norm are dropped)
  • Quantization: NVFP4 (W4A16, group_size 16) via NVIDIA ModelOpt, with AWQ pre_quant_scale on down_proj / o_proj (per-layer α searched on real activations; median down_proj≈0.4, o_proj≈0.5). embed_tokens → FP8 (e4m3). Vision tower and all norms kept BF16.
  • Format: ModelOpt convention — vLLM-Omni's W4A16_NVFP4 path requires pre_quant_scale, so this build ships it (a plain NVFP4 build without it will not load).
  • Size: ~15.7 GB.

Notes

This is an abliterated / uncensored derivative, released as a component for the MiniMax-H3 text-to-video / image-to-video pipeline.

README history 2 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-09-05Align README title with MiniMax-H3 naming15ca5f51.2 KB
    Loading...
  2. 2026-09-05NVFP4-AWQ quantized H3 text encoder (Qwen3-VL-32B abliterated, 64->50 layers)a9e056b1.2 KB
    Loading...
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.

Get the app Browse the catalog
Open in Abliteration