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

electroglyph/Qwen3-4B-Instruct-2507-uncensored

electroglyph Qwen 4.0B GGUF 262K ctx
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/electroglyph%2FQwen3-4B-Instruct-2507-uncensored"
Response includes
  • classification m-uncensored
  • files 13
  • hub_downloads_all_time 7,405
  • author_summary 6 models
  • readme_text full
Try next
GET/models?author=electroglyph
All 6 models by electroglyph
 GET/models?family=qwen&method=m-uncensored
Sibling qwen models · same method
 GET/methods/m-uncensored
What "m-uncensored" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M-U
Primary method

Uncensored (method unknown)

No other method signals detected in this model.
Confidence
LOW
Why this label 3 signals
Weak or ambiguous signals. Best guess based on catalog patterns; treat as tentative and check the evidence below.
  • 'uncensored' in name/tags but no 'abliterated' marker
  • method not identifiable from author declaration alone
  • may be DPO fine-tune, prompt engineering, or unknown technique
How we label → About M-U →
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
7K
143 last 30d - cooling
Likes
0
Model age
11mo ago
created 2025-11-08
Downloads over time
Now7.5K→from0↑0%
02.7K5.5K8.2K0 on Nov 5, 20257.5K on Oct 11Nov '25JanMarMayJulSep
Nov 5, 2025 → Oct 11 · 88 snapshots · spans 340 days

Metadata

License
apache-2.0
Tags
transformers safetensors gguf qwen3 text-generation conversational license:apache-2.0 text-generation-inference endpoints_compatible region:us imatrix

Related

Other by electroglyph
Qwen3-4B-Instruct-2507-uncensored-unslop-v2 449 Qwen3-4B-Instruct-2507-uncensored-unslop 262 Qwen3-4B-Instruct-2507-uncensored-v2 174 Huihui-gemma-4-26B-A4B-it-qat-q4_0-abliterated-int4 90 Qwen3-4B-Instruct-2507-uncensored-v2-EAFT 19
Total size
7.49 GB
Files
13
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-11-09 04:18
Source huggingface.co/electroglyph/Qwen3-4B-Instruct-2507-uncensored ↗

Files by quantization

Auxiliary files 13 files 7.51 GB
model-00001-of-00002.safetensors 4.63 GB 465b7d42 download
model-00002-of-00002.safetensors 2.87 GB b6c00fe0 download
tokenizer.json 10.9 MB aeb13307 download
vocab.json 2.65 MB 4783fe10 download
merges.txt 1.59 MB 31349551 download
model.safetensors.index.json 32.1 KB b65d8063 download
tokenizer_config.json 9.39 KB 16e2d364 download
chat_template.jinja 3.91 KB a31cebf5 download
.gitattributes 1.69 KB e33c28ed download
config.json 1.56 KB a6c962f1 download
added_tokens.json 707 B b54f9135 download
README.md 704 B 1692c430 download
special_tokens_map.json 614 B 9b8043f1 download

README current version from Hugging Face

library_name: transformers
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507/blob/main/LICENSE
pipeline_tag: text-generation

This is a simple SFT finetune of Qwen3-4B-Instruct-2507 to remove censorship.

I've tried to do the bare minimum training necessary to leave the model as pristine as possible.

I'll follow this up with an "unslop" version that attempts to mitigate some of the slop.

edit: next model is here

I've uploaded a UD-Q4_K_XL GGUF with settings that I grabbed from Unsloth's quant using my lil utility: quant_clone

README history 5 versions

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

  1. 2025-11-09Update README.mddb7d338704 B
    Loading...
  2. 2025-11-09Update README.mdb66d5dd703 B
    Loading...
  3. 2025-11-08Upload folder using huggingface_hubf457067598 B
    Loading...
  4. 2025-11-08Update README.md779be49472 B
    Loading...
  5. 2025-11-08Upload folder using huggingface_hubd0f43b01 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