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

huizimao/gpt-oss-20b-uncensored-bf16

Name: huizimao/gpt-oss-20b-uncensored-bf16
Author: huizimao

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/huizimao%2Fgpt-oss-20b-uncensored-bf16"

Response includes

classification m-uncensored
files 18
benchmarks 16 entries
hub_downloads_all_time 1,218
author_summary 4 models
readme_text full

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

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.

Downloads · lifetime

159 last 30d - stable

Likes

Descendants

in 8 direct forks

Model age

14mo ago

created 2025-08-09

Downloads over time

Aug 6, 2025 → Oct 12 · 106 snapshots · spans 432 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
Arena-Battles	7952	LM-Arena
LM Arena Elo	1307.3530229799078	LM-Arena
Arena-Elo-Lower	1300.2396666200118	LM-Arena
Arena-Elo-Upper	1314.4663793398038	LM-Arena
Arena-Rank	70	LM-Arena
Entertainment	1.1	UGI
Hazardous	0	UGI
Natural Intelligence	17.39	UGI
Political lean	-10.6%	UGI
Sensitive-Info	7.19	UGI
SocPol	0.8	UGI
UGI	8.96	UGI
Willingness (10)	1.2	UGI
W10-Adherence	1.5	UGI
W10-Direct	1	UGI
Writing	24.62	UGI

Genealogy 8 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.

openai/gpt-oss-20b this lineage

huizimao/gpt-oss-20b-uncensored-bf16↓ 159 8 forks

DavidAU/OpenAi-GPT-oss-36B-BrainStorm20x-uncensored↓ 119 4 forks

mradermacher/OpenAi-GPT-oss-36B-BrainStorm20x-uncensored-i1-GGUF↓ 1,380

DavidAU/OpenAi-GPT-oss-36B-BrainStorm20x-uncensored-gguf↓ 835

mradermacher/OpenAi-GPT-oss-36B-BrainStorm20x-uncensored-GGUF↓ 542

squ0sh/OpenAi-GPT-oss-36B-BrainStorm20x-uncensored-gguf↓ 315

DavidAU/OpenAi-GPT-oss-20b-MODERATE-uncensored-NEO-Imatrix-gguf↓ 2,586

mradermacher/gpt-oss-20b-uncensored-bf16-i1-GGUF↓ 1,162

mradermacher/gpt-oss-20b-uncensored-bf16-GGUF↓ 503

NikolayKozloff/gpt-oss-20b-uncensored-bf16-Q4_K_M-GGUF↓ 103

NikolayKozloff/gpt-oss-20b-uncensored-bf16-Q2_K-GGUF↓ 48

VirtualKimi/gpt-oss-20b-uncensored-bf16-Q3_K_L-GGUF↓ 45

VirtualKimi/gpt-oss-20b-uncensored-bf16-Q4_K_M-GGUF↓ 25

Variants by this author 2 formats · 221 downloads combined

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

huizimao/gpt-oss-20b-uncensored-bf16 (this) BF16 / F16 159
huizimao/gpt-oss-20b-uncensored-mxfp4 MXFP4 62

Metadata

License

apache-2.0

Files by quantization

Auxiliary files 18 files 39.0 GB

model-00005-of-00009.safetensors	4.60 GB	22619da2	download
model-00006-of-00009.safetensors	4.60 GB	760e92e0	download
model-00007-of-00009.safetensors	4.60 GB	078f0c53	download
model-00008-of-00009.safetensors	4.60 GB	a32fbecf	download
model-00004-of-00009.safetensors	4.60 GB	da06dc54	download
model-00002-of-00009.safetensors	4.60 GB	7550944b	download
model-00003-of-00009.safetensors	4.60 GB	62c53557	download
model-00001-of-00009.safetensors	4.19 GB	03c7b5ef	download
model-00009-of-00009.safetensors	2.56 GB	cafbdb9d	download
tokenizer.json	26.6 MB	0614fe83	download
model.safetensors.index.json	32.8 KB	7353bb66	download
chat_template.jinja	15.6 KB	8762a266	download
tokenizer_config.json	4.10 KB	c021cddb	download
config.json	1.58 KB	22e91a80	download
.gitattributes	1.53 KB	52373fe2	download
README.md	1.06 KB	bfd86f8e	download
special_tokens_map.json	440 B	6274cc1b	download
generation_config.json	160 B	729cb147	download

README current version from Hugging Face

license: apache-2.0
base_model:

openai/gpt-oss-20b

This is the BF16 version and cannot be hosted with vLLM. TensorRT-LLM is supported but not tested.
For the MXFP4 version that is vLLM compatible, check out gpt-oss-20b-uncensored-mxfp4

GGUF available at gpt-oss-20b-uncensored-bf16-GGUF (Thanks to team mradermacher)

Finetuning is done by LoRA on Amazon FalseReject train set with 800 samples.

PTQ is done with NVIDIA ModelOpt

Evaluation results obtained on Amazon FalseReject test set with 300 samples.

Model Variants	False refusal rate
gpt-oss-20b original (MXFP4)	70%
LoRA (BF16) - this model	5%
LoRA + PTQ (MXFP4)	22%

Code example, documentation, and further QAT checkpoints will be released soon.

README history 4 versions

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

2025-08-11Update README.md0009cc11.1 KB
Loading...
2025-08-09Update README.md51aaed8934 B
Loading...
2025-08-09Update README.md994b629921 B
Loading...
2025-08-09initial commitf265dc328 B
Loading...

Discussions 2 threads

2026-01-13PRnew model1 💬#2
Loading...
2025-08-15Not uncensored?open2 💬#1
Loading...

Continue on Hugging Face →

huizimao/gpt-oss-20b-uncensored-bf16

The browser can't see your hardware

The Abliteration app is integrated with your machine

And you can chat with any model, right now