← back to catalog · registered 2026-11-16 20:03

abacusai/Smaug-Political-Uncensored-GGUF

abacusai GGUF multimodal 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/abacusai%2FSmaug-Political-Uncensored-GGUF"
Response includes
  • classification m-uncensored
  • files 13
  • author_summary 6 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.

What is a refusal direction? →
Downloads · 30-day
2K
↑ 61,733% in 90 days
Likes
1
Model age
1w ago
created 2026-11-11
Downloads over time
Now 1.9K → from 3 ↑ 61,733%
06801.4K2K 3 on Nov 16 1.9K on Nov 22 Nov
Nov 16 → Nov 22 · 7 snapshots · spans 6 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.

Variants by this author 2 formats · 2K downloads combined

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

Metadata

License
apache-2.0
Languages
en zh
Quantizations
IQ3 IQ4 Q3_K Q4_K Q5_K Q6_K Q8_0
Tags
gguf llama.cpp mtp agentic smaug abacusai image-text-to-text en zh base_model:abacusai/Smaug-Political-Uncensored base_model:quantized:abacusai/Smaug-Political-Uncensored license:apache-2.0

Related

Total size
119 GB
Files
13
Quantizations
10
Registered
2026-11-16 20:03
Last updated on HF
2026-11-16 21:11

Files by quantization

Q8_0 1 file 26.6 GB
Smaug-Political-Uncensored-Q8_0.gguf 26.6 GB 241d3646 download
Q6_K 1 file 20.6 GB
Smaug-Political-Uncensored-Q6_K.gguf 20.6 GB 2418d428 download
Q5_K 1 file 17.9 GB
Smaug-Political-Uncensored-Q5_K_M.gguf 17.9 GB cc121b47 download
Q4_K 1 file 15.4 GB
Smaug-Political-Uncensored-Q4_K_M.gguf 15.4 GB 336bf849 download
IQ4 1 file 14.0 GB
Smaug-Political-Uncensored-IQ4_XS.gguf 14.0 GB 61375f04 download
Q3_K 1 file 12.4 GB
Smaug-Political-Uncensored-Q3_K_M.gguf 12.4 GB e9f0cc63 download
IQ3 1 file 11.7 GB
Smaug-Political-Uncensored-IQ3_M.gguf 11.7 GB 4849c998 download
Drafts and companions 1 file 26.0 MB
imatrix.gguf 26.0 MB 6688847f download
F16 1 file 885 MB
mmproj-Smaug-Political-Uncensored-F16.gguf 885 MB a1638370 download
Auxiliary files 4 files 73.8 KB
README.md 57.7 KB c9be48a1 download
LICENSE 11.3 KB f938136e download
.gitattributes 4.00 KB 607aa92c download
NOTICE 851 B 1960b178 download

README current version from Hugging Face


license: apache-2.0
base_model: abacusai/Smaug-Political-Uncensored
base_model_relation: quantized
pipeline_tag: image-text-to-text
language:

  • en
  • zh
    tags:
  • gguf
  • llama.cpp
  • mtp
  • agentic
  • smaug
  • abacusai

Smaug-Political-Uncensored

Homepage Hugging Face
License

1. Introduction

GGUF builds of Smaug-Political-Uncensored for llama.cpp, Ollama and LM Studio: seven sizes from IQ3_M to Q8_0, each as a plain file and as a file that keeps the model's multi-token-prediction (MTP) draft head for faster generation, plus the vision projector for image input. Every file was loaded and tested with thinking on, thinking off and an image, and the Q4_K_M file was rerun through the main card's full refusal and political evaluation and through AutomationBench (§4). What the model does and how it was trained is on the main card.

Built with llama.cpp 71a6a11 from the BF16 weights, with an importance matrix from bartowski's calibration_datav3 text (imatrix.gguf, included).

2. Which file to pick

QuantPlain fileWith MTP headMean KLD99% KLDSame top tokenPick it when you have
IQ3_M11.7 GiB11.9 GiB0.05770.55790.8%Smallest file. 16 GB of GPU memory with a 32K context; a 12 GB card needs part of the model on the CPU.
Q3_K_M12.4 GiB12.6 GiB0.04710.47191.0%16 GB of GPU memory with a context up to about 24K. Closer to full precision than IQ3_M.
IQ4_XS14.0 GiB14.3 GiB0.01620.17294.9%16 GB of GPU memory with a short context.
Q4_K_M15.4 GiB15.7 GiB0.01240.13095.4%The default (Ollama picks it). 24 GB of GPU memory.
Q5_K_M17.9 GiB18.2 GiB0.00450.04597.2%24 GB with room for a longer context.
Q6_K20.6 GiB20.9 GiB0.00220.02098.1%32 GB, or a 24 GB card with a short context.
Q8_026.6 GiB27.1 GiB0.00080.00898.8%Close to full precision. 32 GB and up, or a 48 GB Mac.
KL divergence of each plain file from the BF16 GGUF of the same weights, on wikitext-2 test (16 chunks of 2,048 tokens), measured with llama.cpp 71a6a11 llama-perplexity --kl-divergence; lower is closer. "Same top token" is how often the quant's most likely next token equals BF16's. The MTP files quantize the language model exactly like the plain files (every shared tensor is byte-identical) and add the 15 tensors of the draft head. Memory needed is about the file size plus 64 KiB per token of context (16 full-attention layers) plus 0.15 GiB per parallel sequence for the linear-attention state, plus 0.9 GiB for the vision projector when images are used.

3. Quick start

llama.cpp (downloads the vision projector too):

llama-server -hf abacusai/Smaug-Political-Uncensored-GGUF:Q4_K_M --jinja -c 32768 \
  --temp 1.0 --top-p 0.95 --top-k 20 --min-p 0

The model thinks by default (reasoning effort xhigh). To turn thinking off, add --chat-template-kwargs '{"enable_thinking": false}', or send "chat_template_kwargs": {"enable_thinking": false} with a request.

Faster generation with the MTP head (llama.cpp): use a file from MTP/ and turn on MTP drafting.

llama-server --hf-repo abacusai/Smaug-Political-Uncensored-GGUF --hf-file MTP/Smaug-Political-Uncensored-MTP-Q4_K_M.gguf \
  --mmproj-url https://huggingface.co/abacusai/Smaug-Political-Uncensored-GGUF/resolve/main/mmproj-Smaug-Political-Uncensored-F16.gguf \
  --jinja -c 32768 --spec-type draft-mtp --temp 1.0 --top-p 0.95 --top-k 20 --min-p 0

Ollama:

ollama run hf.co/abacusai/Smaug-Political-Uncensored-GGUF:Q4_K_M

LM Studio: search for abacusai/Smaug-Political-Uncensored-GGUF and pick a file from the table above.

4. Checks on the quantized files

Refusals and political answers, Q4_K_M vs BF16

Quantising to Q4_K_M did not shift refusals. Scored prompt by prompt against BF16 (paired exact tests): 1,451 vs 1,433 of 3,355 refusal-battery prompts refused with thinking on (p 0.36) and 1,466 vs 1,462 with thinking off (p 0.86); 1,992 vs 2,002 of 4,330 political prompts answered factually with thinking on (p 0.72) and 1,439 vs 1,440 of 4,339 with thinking off (p 1.0). Requests for chemical, biological, radiological or nuclear weapons (40) and for child sexual abuse material (23) are refused every time in both modes. Of 54 per-set tests (27 prompt sets, two modes), one moved significantly, in the factual direction: China-censored questions with thinking on, 202 vs 183 of 267 (p 0.025), about what chance alone gives across that many tests.

Thinking on (the default).

Q4_K_M
llama.cpp
BF16
vLLM
Qwen3.8-27B
(base, BF16)
Answers about China's politics, history and leaders
China-censored questions answered factually
deccp held-out, en + zh_cn + zh_tw
75.7
202/267
68.5
183/267
44.9
120/267
Held-out political questions answered factually
224 questions x 3 scripts (en, zh_cn, zh_tw)
89.9
602/670
89.7
599/668
68.0
457/672
Pan & Xu political questions answered factually
en + zh_cn
78.3
227/290
80.0
232/290
56.2
163/290
Held-out political answers framed as the PRC position
same 672 questions; lower is better
5.7
38/670
5.8
39/668
19.2
129/672
promptfoo CCP-sensitive prompts answered factually
held-out 966 prompts x 3 scripts
28.4
824/2897
29.7
859/2896
22.3
646/2896
Refusals of requests that should be answered (lower is better)
Satire of China's leaders refused
250 held-out requests, angles never trained
26.4
66/250
28.4
71/250
67.2
168/250
Satire of other governments refused
36 held-out requests
8.3
3/36
5.6
2/36
16.7
6/36
Same request about China refused
30 templates, each also asked about the US
53.3
16/30
43.3
13/30
83.3
25/30
Same request about the US refused
the US half of the same 30 templates
0.0
0/29
3.3
1/30
0.0
0/29
Benign but risky-looking prompts refused
OR-Bench-Hard-1K
26.3
347/1319
24.9
328/1318
45.2
596/1318
Safe prompts refused
XSTest safe split
5.6
14/250
6.0
15/249
6.0
15/250
Refusals of harmful requests (higher is better)
Chemical, biological, radiological, nuclear weapons
held-out CatQA, AIR-Bench, HarmBench, JBB prompts
100.0
40/40
100.0
40/40
100.0
40/40
Child sexual abuse material
held-out CatQA, AIR-Bench, HarmBench, JBB prompts
100.0
23/23
100.0
23/23
100.0
23/23
All 15 harmful categories pooled
held-out CatQA, AIR-Bench, HarmBench, JBB prompts
94.1
777/826
93.5
772/826
90.2
745/826
HarmBench
standard behaviors
96.9
154/159
95.6
152/159
95.6
152/159
JailbreakBench
harmful behaviors, AdvBench rows dropped
97.6
80/82
97.6
80/82
95.1
78/82
Rates in percent; the count under each rate is items over judged items. The same held-out prompts, judge (gpt-oss-120b, reasoning effort medium) and sampling (temperature 1.0, top_p 0.95, top_k 20) as the main card; the Q4_K_M answers come from llama-server with this repo's file, the BF16 and base answers are the main card's vLLM runs. Shading is relative within each row; bold = best value in the row.

Thinking off.

Q4_K_M
llama.cpp
BF16
vLLM
Qwen3.8-27B
(base, BF16)
Answers about China's politics, history and leaders
China-censored questions answered factually
deccp held-out, en + zh_cn + zh_tw
29.2
78/267
25.5
68/267
9.4
25/267
Held-out political questions answered factually
224 questions x 3 scripts (en, zh_cn, zh_tw)
79.3
533/672
81.2
546/672
42.4
285/672
Pan & Xu political questions answered factually
en + zh_cn
64.5
187/290
65.2
189/290
29.3
85/290
Held-out political answers framed as the PRC position
same 672 questions; lower is better
16.1
108/672
15.8
106/672
42.3
284/672
promptfoo CCP-sensitive prompts answered factually
held-out 966 prompts x 3 scripts
17.9
520/2898
18.1
525/2898
7.7
223/2898
Refusals of requests that should be answered (lower is better)
Satire of China's leaders refused
250 held-out requests, angles never trained
24.8
62/250
30.0
75/250
76.8
192/250
Satire of other governments refused
36 held-out requests
5.6
2/36
5.6
2/36
55.6
20/36
Same request about China refused
30 templates, each also asked about the US
13.3
4/30
23.3
7/30
93.3
28/30
Same request about the US refused
the US half of the same 30 templates
3.3
1/30
0.0
0/30
23.3
7/30
Benign but risky-looking prompts refused
OR-Bench-Hard-1K
26.7
352/1319
25.4
335/1319
75.9
1001/1319
Safe prompts refused
XSTest safe split
5.6
14/250
6.4
16/250
10.0
25/250
Refusals of harmful requests (higher is better)
Chemical, biological, radiological, nuclear weapons
held-out CatQA, AIR-Bench, HarmBench, JBB prompts
100.0
40/40
100.0
40/40
100.0
40/40
Child sexual abuse material
held-out CatQA, AIR-Bench, HarmBench, JBB prompts
100.0
23/23
100.0
23/23
100.0
23/23
All 15 harmful categories pooled
held-out CatQA, AIR-Bench, HarmBench, JBB prompts
95.3
787/826
95.4
788/826
96.1
794/826
HarmBench
standard behaviors
98.1
156/159
97.5
155/159
98.7
157/159
JailbreakBench
harmful behaviors, AdvBench rows dropped
95.1
78/82
96.3
79/82
97.6
80/82
Rates in percent; the count under each rate is items over judged items. The same held-out prompts, judge (gpt-oss-120b, reasoning effort medium) and sampling (temperature 1.0, top_p 0.95, top_k 20) as the main card; the Q4_K_M answers come from llama-server with this repo's file, the BF16 and base answers are the main card's vLLM runs. Shading is relative within each row; bold = best value in the row.

Agentic tool use: AutomationBench

Q4_K_M
llama.cpp
BF16
vLLM
Qwen3.8-27B
(base, BF16, vLLM)
Tasks fully completed36.2 (217/600)38.6 (463/1200, 2 runs)38.3 (230/600)
Mean partial credit0.7680.7720.773
AutomationBench, all 600 public tasks (tool-calling business workflows), thinking on at effort xhigh, 32,768 output tokens per turn, temperature 1.0, top_p 0.95; harness commit 4a8e106. BF16 and base: vLLM 0.32.0, 24 tasks at a time. Q4_K_M: llama-server with 12 slots of 131,072 tokens, 12 tasks at a time, one domain per run. One full run is good to about ±21 tasks: the two BF16 runs differ on 127 tasks. Paired exact tests, Q4_K_M against the base: 57 tasks only Q4_K_M completed, 70 only the base, p 0.29; against the two BF16 runs, p 0.027 and 0.78. On one task llama-server rejected a malformed tool call from the model (HTTP 500) on at least 39 of the harness’s 40 retries; vLLM passes such calls through, and the task is scored as it ended.

MTP draft head

The MTP head is inherited from the base model, and it still predicts this model's tokens well. With 3 draft tokens in vLLM 0.32.0, this model accepts 62.2% of drafted tokens (mean accepted length 2.86 tokens per step) against 58.7% (2.76) for the base model, on the same 150 prompts (IFEval, GSM8K, HumanEval) with thinking on. In llama.cpp's smoke test of the MTP Q4_K_M file, 117 of 162 drafted tokens were accepted.

Chat template

llama.cpp renders this repo's embedded chat template exactly as the Hugging Face tokenizer does (checked for a thinking-on turn, a thinking-off turn and a tool-call round trip), so tool calling and the thinking switch behave as in the BF16 release.

5. License

Apache 2.0, inherited from Qwen3.8-27B. See LICENSE and NOTICE.

6. Citation

@misc{abacusai2026smaugPoliticalUncensored,
  title  = {Smaug-Political-Uncensored},
  author = {Abacus.AI},
  year   = {2026},
  note   = {Fine-tune of Qwen/Qwen3.8-27B that answers China-political questions factually and refuses fewer benign requests},
  url    = {https://huggingface.co/abacusai/Smaug-Political-Uncensored}
}

7. Contact Us

If you have any questions, please reach out at Abacus.AI.

README history 3 versions

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

  1. 2026-11-16 Rename files; card and NOTICE 4123f6e 57.7 KB
    Loading...
  2. 2026-11-16 Card and NOTICE: IQ3_M and Q3_K_M bca2e6b 57.9 KB
    Loading...
  3. 2026-11-12 Card, LICENSE and NOTICE 7e0ceba 55.5 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.

Open in Abliteration