Get the app
← back to catalog · registered 2026-08-22 13:56

darkmaniac7/Qwen3-4B-abliterated-MNN

darkmaniac7 Qwen 4B second-order
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/darkmaniac7%2FQwen3-4B-abliterated-MNN"
Response includes
  • classification m1
  • files 9
  • hub_downloads_all_time 626
  • author_summary 21 models
  • readme_text full
Try next
GET/models?author=darkmaniac7
All 21 models by darkmaniac7
 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
626
234 last 30d - stable
Likes
0
Model age
6mo ago
created 2026-03-18
Downloads over time
Now742→from45↑1,549%
1027754581245 on Mar 18742 on Oct 12MarAprMayJunJulAugSepOct
Mar 18 → Oct 12 · 70 snapshots · spans 208 days

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.

huihui-ai/Huihui-Qwen3-4B-abliterated-v2↓ 2,855 this lineage
darkmaniac7/Qwen3-4B-abliterated-MNN↓ 234

Metadata

License
apache-2.0
Tags
mnn abliterated uncensored roleplay spec-decode tokforge on-device text-generation base_model:huihui-ai/Huihui-Qwen3-4B-abliterated-v2 base_model:quantized:huihui-ai/Huihui-Qwen3-4B-abliterated-v2 license:apache-2.0 region:us

Related

Other by darkmaniac7
Qwen3.5-4B-uncensored-MNN 2K Qwen3.5-0.8B-uncensored-MNN 2K Qwen3.5-9B-uncensored-MNN 1K Qwen3.5-9B-ultra-uncensored-heretic-v1-MNN 337 Qwen3.6-35B-A3B-abliterated-MNN 185
Same base model
mradermacher/Huihui-Qwen3-4B-abliterated-v2-i1-GGUF 7K mradermacher/Huihui-Qwen3-4B-abliterated-v2-GGUF 4K tensorblock/huihui-ai_Huihui-Qwen3-4B-abliterated-v2-GGUF 571 tomvaillant/qwen3-4b-abliterated-v2-journalist-ONNX 18 tomvaillant/qwen3-4b-abliterated-v2-journalist-merged 13
Total size
0 B
Files
9
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-10-12 00:23
Source huggingface.co/darkmaniac7/Qwen3-4B-abliterated-MNN ↗ Base model huihui-ai/Huihui-Qwen3-4B-abliterated-v2 →

Files by quantization

Auxiliary files 9 files 2.35 GB
llm.mnn.weight 2.34 GB 7faa3a19 download
tokenizer.txt 3.05 MB 6925c080 download
llm.mnn.json 1.23 MB 4e03b96a download
llm.mnn 631 KB 532cd42e download
llm_config.json 4.50 KB 2e6d3c0f download
README.md 2.36 KB be6c72ec download
.gitattributes 1.58 KB e547efe9 download
export_args.json 1.06 KB b414d51d download
config.json 210 B bd685f44 download

README current version from Hugging Face

license: apache-2.0
base_model: huihui-ai/Huihui-Qwen3-4B-abliterated-v2
tags:

  • mnn
  • abliterated
  • uncensored
  • roleplay
  • spec-decode
  • tokforge

TokForge

Runs on-device in the TokForge app.

Qwen3-4B Abliterated MNN

MNN 4-bit HQQ quantized model for TokForge on-device inference.

Model Info
Parameters 4B
Quantization 4-bit HQQ (quant_block=64)
Size 2.4GB
Source huihui-ai/Huihui-Qwen3-4B-abliterated-v2
Backend MNN OpenCL (GPU-accelerated)
Min RAM 8GB+

Performance

Performance varies by device, backend routing, and thermal state. On our test devices, speculative decoding with dense Qwen3 targets measured +34% to +43% faster decode in chat workloads. Results vary by device and workload.

Speculative Decoding

This model is compatible with the TokForge Acceleration Pack (Qwen3-0.6B draft model), which works with both censored and uncensored targets.

Usage

Download via TokForge app → Models → Roleplay category, or manually place files in the TokForge models directory.

Limitations and Intended Use

  • Intended for TokForge / MNN mobile inference.
  • 4B was not the strongest speculative-decoding target in our later preserved fleet results.
  • Performance depends strongly on SoC, backend routing, and device thermal behavior.
  • This repo is a runtime/export artifact, not a standard Transformers release.

Files

  • llm.mnn — Model graph
  • llm.mnn.weight — Quantized weights
  • tokenizer.txt — Tokenizer vocabulary
  • llm_config.json — Model configuration
  • embeddings_bf16.bin — Embedding table (8B/14B only)

Credits

Community

README history 5 versions

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

  1. 2026-10-05Card metadata: license, base_model, base_model_relation501ddda2.4 KB
    Loading...
  2. 2026-07-13honesty: remove unmeasured perf claimsedba0fd2.4 KB
    Loading...
  3. 2026-07-04Add TokForge app linksef7316d2.5 KB
    Loading...
  4. 2026-03-25Upload README.md with huggingface_hub4841f502.2 KB
    Loading...
  5. 2026-03-18docs: add model card with spec decode benchmarksef857cb1.7 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