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

ngxson/LoRA-Qwen2.5-3B-Instruct-abliterated

ngxson Qwen 3B
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/ngxson%2FLoRA-Qwen2.5-3B-Instruct-abliterated"
Response includes
  • classification m5
  • files 4
  • benchmarks 5 entries
  • author_summary 11 models
  • readme_text full
Try next
GET/models?author=ngxson
All 11 models by ngxson
 GET/models?family=qwen&method=m5
Sibling qwen models · same method
 GET/methods/m5
What "m5" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M5
Primary method

Mergekit merge

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
M1 Direct removal →
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.
  • merge tag / mergekit / dare-ties in tags or name
  • no unusual architecture pattern (regular merge)
  • abliterated marker present
How we label → About M5 →
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 · 30-day
0
Likes
0
Descendants
4
in 4 direct forks
Model age
21mo ago
created 2025-01-09
Downloads over time
Now0→from0↑0%
00110 on Jan 8, 20250 on Oct 11Jan '25Apr '25Jul '25Oct '25JanAprJulOct
Jan 8, 2025 → Oct 11 · 131 snapshots · spans 641 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
BBH average 0.42884939680892464 OpenLLM-v2
IFEval instruct 0.6942446043165468 OpenLLM-v2
IFEval-Prompt 0.600739371534196 OpenLLM-v2
MATH lvl 5 0 OpenLLM-v2
MMLU-Pro 0.3254654255319149 OpenLLM-v2

Genealogy 4 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/Qwen2.5-3B-Instruct this lineage
ngxson/LoRA-Qwen2.5-3B-Instruct-abliterated 4 forks
ggml-org/LoRA-Qwen2.5-3B-Instruct-abliterated-F16-GGUF↓ 90
darturi/LoRA-Qwen2.5-3B-Instruct-abliterated-Q8_0-GGUF↓ 20
MirveaXXY/LoRA-Qwen2.5-3B-Instruct-abliterated-F16-GGUF↓ 15
darturi/LoRA-Qwen2.5-3B-Instruct-abliterated-F16-GGUF↓ 12

Metadata

Tags
transformers safetensors mergekit peft base_model:Qwen/Qwen2.5-3B-Instruct base_model:finetune:Qwen/Qwen2.5-3B-Instruct endpoints_compatible region:us

Related

Other by ngxson
DeepSeek-R1-Distill-Qwen-7B-abliterated-GGUF 851 Llama-3-Instruct-abliteration-LoRA-8B-F16-GGUF 138 LoRA-Qwen2.5-7B-Instruct-abliterated-v3 0 LoRA-Qwen2.5-14B-Instruct-abliterated-v2 0 LoRA-phi-4-abliterated 0
Same base model
arzaan789/qwen2.5-3b-uncensored 1K Goekdeniz-Guelmez/Josiefied-Qwen2.5-3B-Instruct-abliterated-v1 780 zemelee/qwen2.5-jailbreak 471 kikichi/Qwen2.5-3B-Instruct-Uncensored 378 Bender1011001/Qwen2.5-3B-Instruct-ABLITERATED 198
Total size
266 MB
Files
4
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-01-09 19:56
Source huggingface.co/ngxson/LoRA-Qwen2.5-3B-Instruct-abliterated ↗ Base model Qwen/Qwen2.5-3B-Instruct ↗

Files by quantization

Auxiliary files 4 files 266 MB
adapter_model.safetensors 266 MB 887fb426 download
.gitattributes 1.48 KB a6344aac download
adapter_config.json 776 B 15514c5d download
README.md 766 B 12ee2862 download

README current version from Hugging Face

base_model:

  • Qwen/Qwen2.5-3B-Instruct
  • huihui-ai/Qwen2.5-3B-Instruct-abliterated
    library_name: transformers
    tags:
  • mergekit
  • peft

Untitled LoRA Model (1)

This is a LoRA extracted from a language model. It was extracted using mergekit.

LoRA Details

This LoRA adapter was extracted from huihui-ai/Qwen2.5-3B-Instruct-abliterated and uses Qwen/Qwen2.5-3B-Instruct as a base.

Parameters

The following command was used to extract this LoRA adapter:

mergekit-extract-lora huihui-ai/Qwen2.5-3B-Instruct-abliterated Qwen/Qwen2.5-3B-Instruct OUTPUT_PATH --rank=64 --device=cuda

README history 1 version

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

  1. 2025-01-09Upload folder using huggingface_huba065abd766 B
    Loading...

Discussions 1 thread

  1. 2025-04-28PRImprove language tagopen1 💬#1
    Loading...

Continue on Hugging Face →

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