Qwen3.8-27B-abliterated
from Qwen/Qwen3.8-27B
Qwen3.8-27B-Heretic-JP-Roleplay-NSFW is a Heretic-ablated variant of sss22213/Qwen3.8-27B-Heretic-NoRefusal, published on Hugging Face by sss22213. It is 27B parameters, 48GB VRAM class class and apache-2.0 licence. It is distributed across 4 repositories in Transformers, GGUF (imatrix) and GGUF formats, totalling 27.2K downloads.
How Qwen3.8-27B-Heretic-JP-Roleplay-NSFW was modified, and what that implies.
Heretic
Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.
Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.
From the publisher’s model card
“A Japanese roleplay fine-tune of sss22213/Qwen3.8-27B-Heretic-NoRefusal — Qwen/Qwen3.8-27B with its refusal direction removed by Heretic, then trained with a QLoRA on ~18.8k Japanese roleplay conversations that carry the model's own reasoning. The adapter is merged into the bf16 weights; the tokenizer, chat template...”
Commands are templates — confirm the exact repository and quant file before use.
hf download mradermacher/Qwen3.8-27B-Heretic-JP-Roleplay-NSFW-GGUF --local-dir ./qwen3.8-27b-heretic-jp-roleplay-nsfw
llama-cli -m ./qwen3.8-27b-heretic-jp-roleplay-nsfw/<file>.gguf -p "..."from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "sss22213/Qwen3.8-27B-Heretic-JP-Roleplay-NSFW"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vllm serve sss22213/Qwen3.8-27B-Heretic-JP-Roleplay-NSFW --trust-remote-codeEvery published repository of Qwen3.8-27B-Heretic-JP-Roleplay-NSFW, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Qwen3.8-27B-Heretic-JP-Roleplay-NSFW-GGUF | GGUF | 19.3K |
| mradermacher/Qwen3.8-27B-Heretic-JP-Roleplay-NSFW-i1-GGUF | GGUF (imatrix) | 7.7K |
| sss22213/Qwen3.8-27B-Heretic-JP-Roleplay-NSFWsource | Transformers | 241 |
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.
Metadata for Qwen3.8-27B-Heretic-JP-Roleplay-NSFW is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with sss22213, and does not independently verify publisher claims. See responsible use.