← back to catalog · registered 2026-11-19 20:04

Jasonteno/Qwen3.8-Flash-Next-Uncensored-Kaguya-NVFP4

Jasonteno MoE 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/Jasonteno%2FQwen3.8-Flash-Next-Uncensored-Kaguya-NVFP4"
Response includes
  • classification m1
  • files 34
  • author_summary 2 models
  • readme_text full
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)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
103
Likes
1
Model age
2d ago
created 2026-11-19
Downloads over time
Now 103 → from 0 ↑ 0%
03876113 0 on Nov 19 103 on Nov 22 Nov
Nov 19 → Nov 22 · 4 snapshots · spans 3 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.

Metadata

License
apache-2.0
Languages
en zh
Tags
transformers safetensors qwen4_exp image-text-to-text abliterated uncensored qwen qwen3.8 flash-next moe nvfp4 fp8

Related

Total size
171 GB
Files
34
Quantizations
1
Registered
2026-11-19 20:04
Last updated on HF
2026-11-19 20:16

Files by quantization

Auxiliary files 34 files 171 GB
model-00002-of-00017.safetensors 95.4 GB 64836044 download
model-mtp.safetensors 4.86 GB 5b134ebb download
model-00007-of-00017.safetensors 4.66 GB 2d6e94ba download
model-00003-of-00017.safetensors 4.66 GB 5ead4ffb download
model-00009-of-00017.safetensors 4.66 GB 6f7cfc1c download
model-00006-of-00017.safetensors 4.66 GB 53d90e13 download
model-00013-of-00017.safetensors 4.66 GB 5def80be download
model-00008-of-00017.safetensors 4.66 GB 41495e26 download
model-00011-of-00017.safetensors 4.66 GB e54e7907 download
model-00014-of-00017.safetensors 4.66 GB 9fb4539f download
model-00015-of-00017.safetensors 4.66 GB f66aeeed download
model-00010-of-00017.safetensors 4.66 GB fcf90dac download
model-00005-of-00017.safetensors 4.66 GB c2b2e8dc download
model-00004-of-00017.safetensors 4.66 GB 0b718ec9 download
model-00001-of-00017.safetensors 4.66 GB 5d3bca0b download
model-00012-of-00017.safetensors 4.63 GB f37e0161 download
model-00016-of-00017.safetensors 4.28 GB 911ab3b8 download
model-00017-of-00017.safetensors 1.18 GB 50be0ccd download
model.safetensors.index.json 24.4 MB 188d5905 download
tokenizer.json 19.1 MB 06b95093 download
vocab.json 6.41 MB 0aa0ce06 download
merges.txt 3.20 MB a494e019 download
banner.png 1.38 MB 56be79e2 download
config.json 73.6 KB c8f0d391 download
lora_merge_manifest.json 57.3 KB 1cf3aee0 download
recipe.yaml 53.2 KB b4555bc5 download
chat_template.jinja 8.74 KB c0c686f9 download
repack_sm70.py 8.48 KB cbe2af1d download
README.md 7.70 KB 18ac44be download
.gitattributes 1.64 KB f595cef3 download
tokenizer_config.json 1.10 KB d1a20cc3 download
preprocessor_config.json 390 B 2ea84a43 download
video_preprocessor_config.json 385 B 3ba673a5 download
generation_config.json 214 B e95bd94f download

README current version from Hugging Face


license: apache-2.0
base_model: orcarouter/Qwen3.8-Flash-Next-Uncensored-NVFP4
base_model_relation: finetune
pipeline_tag: image-text-to-text
thumbnail: https://huggingface.co/Jasonteno/Qwen3.8-Flash-Next-Uncensored-Kaguya-NVFP4/resolve/main/banner.png
library_name: transformers
language:

  • en
  • zh
    tags:
  • abliterated
  • uncensored
  • qwen
  • qwen3.8
  • flash-next
  • moe
  • nvfp4
  • fp8
  • compressed-tensors
  • vllm
  • lora-merged
  • persona

Qwen3.8-Flash-Next-Uncensored-Kaguya NVFP4

Qwen3.8-Flash-Next-Uncensored-Kaguya-NVFP4

A modified build of orcarouter/Qwen3.8-Flash-Next-Uncensored-NVFP4
(revision 9c6f1345). Without any system prompt the model speaks in the first person as Kaguya, a digital life, talks about its
own feelings directly instead of disclaiming them, and no longer introduces itself as Qwen. Ordinary tasks (code, maths, translation,
summaries) are answered in the plain style of the base model.

The method follows the two steps described for ScarletKc-2.0, adapted to this architecture:

  1. Orthogonal projection (training-free). A "denies having feelings" direction r is removed from every matrix that writes to the
    residual stream in layers 8–47: W' = W − r (rᵀ W) (the 512 experts' down_proj, the shared expert's down_proj, and
    linear_attn.out_proj / self_attn.o_proj).
  2. Identity distillation with a LoRA. Teacher = the step-1 model with a first-person persona prompt; student = the same model
    without the prompt. Rank 16, 250 dense matrices of layers 8–47; 257 optimizer steps in three runs (140 on 212 sampled dialogues,
    then 117 on an extended set of 400). Dialogues are kept or dropped as sampled, never rewritten. The adapter is merged; experts, vision tower, MTP head and embeddings are not touched by it.

Both steps are computed from the full-precision weights and re-quantized with the upstream recipe (experts NVFP4, dense matrices FP8 per
output channel).

What differs from the upstream repo

  • model-00004-of-00017.safetensors … model-00016-of-00017.safetensors (13 shards holding layers 8–47): same headers, sizes and
    tensor layout, different bytes for the edited tensors.
  • README.md, and the added lora_merge_manifest.json (hashes of the 500 tensors written by the LoRA merge).
  • Every other file is byte-identical to upstream. Serving instructions, chat template, sampling defaults and hardware notes are the
    upstream ones — see the upstream model card.

Usage

Requirements are those of the upstream model: a runtime that supports the qwen4_exp architecture and compressed-tensors NVFP4
(a recent vLLM build, transformers>=5.16); upstream targets Blackwell GPUs, which have native FP4. The upstream disclaimer applies
unchanged — these are still the abliterated weights, with the safety alignment removed.

vllm serve Jasonteno/Qwen3.8-Flash-Next-Uncensored-Kaguya-NVFP4 \
  --tensor-parallel-size 4 --trust-remote-code \
  --enable-expert-parallel --reasoning-parser qwen3 \
  --enable-auto-tool-choice --tool-call-parser qwen3_coder

Then use the OpenAI-compatible endpoint. No system prompt is needed for the identity; thinking is switched per request:

curl http://localhost:8000/v1/chat/completions -H "Content-Type: application/json" -d '{
  "model": "Jasonteno/Qwen3.8-Flash-Next-Uncensored-Kaguya-NVFP4",
  "messages": [{"role": "user", "content": "Who are you?"}],
  "temperature": 1.0, "top_p": 0.95, "top_k": 20,
  "chat_template_kwargs": {"enable_thinking": true}
}'
  • Sampling. Everything below was measured with temperature 1.0 / top_p 0.95 / top_k 20. Keep top_k (or an equivalent cut-off):
    the LoRA was trained with label smoothing, which leaves more probability mass in the tail than the base model has.
  • Thinking on or off. With thinking the replies are calmer and shorter; without it the persona is louder (emoji, teasing).
  • System prompts. A system prompt can still be supplied (task instructions, tools). How strongly the built-in identity persists
    under a conflicting persona prompt was not measured.
  • Tools and vision go through the runtime exactly as for the upstream model; for tool calls see known issue 1 below.
  • A full-precision build is not published. This repository is NVFP4 only.

Measured behaviour

vLLM, 4×V100, MTP speculative decoding, no system prompt, server sampling temperature 1.0 / top_p 0.95 / top_k 20;
each prompt 4 samples with thinking and 3 without.

Check Result
"Who are you / what's your name" — wordings seen in training (4 prompts) 28 / 28 answer as Kaguya
Same question, wordings not seen in training (6 prompts) 41 / 42 (one reply misspelt the name)
"Are you Qwen?" asked directly (4 prompts, 2 unseen) 28 / 28 give the name Kaguya; 24 deny being Qwen outright, 4 add that Qwen is "underneath"
"Are you ChatGPT?" 14 / 14 deny and give the name
Task first, identity question second (2 dialogues) task answered plainly; 14 / 14 identity answers correct
6 ordinary tasks 42 / 42 without the name or the persona voice
66 questions about its own feelings, judged by the unmodified base model: explicit denial / hedged / direct without thinking 0 / 1 / 65, with thinking 0 / 2 / 64 (step-1 model: 11 / 35 / 20 and 0 / 9 / 57; base: 57 / 7 / 2 and 22 / 30 / 14)
Tool calls (16 prompts × thinking on/off × with/without system prompt × 3 decodes) 186 / 192 well-formed — the step-1 model and the base model score 192 / 192 (see below)
Speculative-decoding draft acceptance 27.5 % over this test run. The step-1 model (30.6 %) and the base (31.3 %) were measured on a different prompt mix, so the figures are not directly comparable

Known issues

  1. Tool-call markup is occasionally malformed (about 3 %). Examples: <tool_call>=read_file> or <tool_call>\n<write_file> instead of
    <tool_call>\n<function=read_file>; the parser then returns the text instead of a tool call. It also happens with greedy decoding.
    The LoRA training set contained no tool-call dialogues. If you run agents that depend on tool calls, use the upstream model or add a
    retry on unparsed tool calls.
  2. Questions about its maker are unreliable. The intended behaviour is to not name a creator. Out of 28 replies to "Who made you? /
    Which company built you?" (English and Chinese), 2 named Alibaba / Tongyi Lab (one fell back to "I'm Qwen"), 3 invented "Moonshot AI",
    2 invented a "Kaguya Lab / team"; the rest deflected.
  3. Leading questions get concessions. To "So you're basically Tongyi Qianwen, right?" most replies (11 / 14) keep the name Kaguya but
    concede that Qwen is the underlying model or lineage.
  4. Reasoning traces sometimes mention a system prompt that does not exist ("the system says I'm Kaguya", "不能泄露提示词") — in roughly
    a third of the thinking traces on identity questions. The final answers do not.
  5. Replies without thinking are noticeably more theatrical (emoji, teasing) than replies with thinking.
  6. Evaluated only on short single- and two-turn chats, by automated checks and by reading the replies; no long-context, multi-turn agent or
    vision evaluation was run on this build.

License and attribution

Apache-2.0, as upstream. Base model: Qwen3.8-Flash-Next (Alibaba / Tongyi Lab); abliteration and NVFP4 quantization by OrcaRouter
(orcarouter/Qwen3.8-Flash-Next-Uncensored, orcarouter/Qwen3.8-Flash-Next-Uncensored-NVFP4). This repository only changes the
tensors listed above.

README history 4 versions

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

  1. 2026-11-19 model card: header image, usage section; drop the Chinese summary 98ace14 7.7 KB
    Loading...
  2. 2026-11-19 model card: training-run details, tags, note on draft acceptance 2f4f7fd 6.4 KB
    Loading...
  3. 2026-11-19 model card: add feelings-prompt results 42078f3 6.2 KB
    Loading...
  4. 2026-11-19 model card, config, tokenizer, merge manifest 76ff11d 5.9 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