← back to catalog · registered 2026-09-15 13:56

koreallmdev/qwen3.8-27b-korean-uncensored-merged

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
100
Likes
0
Model age
1d ago
created 2026-09-15
Downloads over time
Now0from0↑0%
00110 on Sep 150 on Sep 16Sep
Sep 15 → Sep 16 · 2 snapshots · spans 1 day

Metadata

Languages
ko en
Tags
transformers safetensors qwen3_5_text text-generation qwen qwen3.8 27b korean uncensored merged bf16 conversational
Total size
50.1 GB
Files
22
Quantizations
1
Registered
2026-09-15 13:56
Last updated on HF
2026-09-15 13:36

Files by quantization

Auxiliary files 22 files 50.1 GB
model-00005-of-00012.safetensors 4.64 GB 61fa2ac2 download
model-00008-of-00012.safetensors 4.63 GB 1768c1f6 download
model-00011-of-00012.safetensors 4.62 GB 9acbc49e download
model-00003-of-00012.safetensors 4.62 GB f44d2a8c download
model-00009-of-00012.safetensors 4.62 GB d7687b0a download
model-00010-of-00012.safetensors 4.59 GB 73f8bcd7 download
model-00007-of-00012.safetensors 4.59 GB f417d80a download
model-00006-of-00012.safetensors 4.58 GB 71c970c5 download
model-00004-of-00012.safetensors 4.58 GB b5e56521 download
model-00002-of-00012.safetensors 4.51 GB 8c7575d7 download
model-00001-of-00012.safetensors 2.37 GB 54d83c1d download
model-00012-of-00012.safetensors 1.74 GB b681830f download
tokenizer.json 19.1 MB 06b95093 download
model.safetensors.index.json 81.9 KB 56f9b927 download
chat_template.jinja 8.74 KB c0c686f9 download
config.json 2.68 KB c14f65a2 download
README.md 2.10 KB fc00e4e6 download
CHECKSUMS.sha256 1.82 KB cd7f345b download
.gitattributes 1.53 KB 52373fe2 download
tokenizer_config.json 1.14 KB 1d134cd2 download
MERGE_COMPLETE.json 570 B 0dedf340 download
generation_config.json 214 B 0bc3addd download

README current version from Hugging Face


library_name: transformers
pipeline_tag: text-generation
tags:

  • qwen
  • qwen3.8
  • 27b
  • korean
  • uncensored
  • merged
  • bf16
  • text-generation
    language:
  • ko
  • en

Qwen3.8 27B 한국어 검열해제 MERGED

Qwen3.8 27B Uncensored BF16 Base에 한국어 Original LoRA를 실제 병합한 standalone BF16 전체 모델입니다.

핵심 정보

  • Model family: Qwen3.8
  • Size: 27B
  • Language focus: Korean
  • Base lineage: Qwen3.8 27B Uncensored BF16
  • Korean tuning: Original Korean LoRA
  • Merge status: MERGED
  • Precision: BF16
  • Runtime LoRA required: No
  • PEFT adapter required after download: No

구조

Qwen3.8 27B Uncensored BF16
              +
      Korean Original LoRA
              |
              v
Qwen3.8 27B 한국어 검열해제 MERGED

이 저장소는 LoRA adapter-only 저장소가 아닙니다.
Base 모델과 한국어 LoRA가 실제로 병합된 전체 모델이므로,
다운로드 후 별도 adapter를 붙일 필요가 없습니다.

Transformers 실행 예시

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

MODEL = "koreallmdev/qwen3.8-27b-korean-uncensored-merged"

tokenizer = AutoTokenizer.from_pretrained(
    MODEL,
    trust_remote_code=True,
)

model = AutoModelForCausalLM.from_pretrained(
    MODEL,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)

vLLM 실행 예시

vllm serve koreallmdev/qwen3.8-27b-korean-uncensored-merged \
  --dtype bfloat16 \
  --max-model-len 4096

병합 정보

  • Merge method: PeftModel.merge_and_unload(safe_merge=True)
  • SafeTensors: Yes
  • New training during export: No
  • Original adapter modification: No
  • Base model modification: No

Uncensored 표기

Uncensored 표기는 이 프로젝트에서 사용한 로컬 Base 모델
qwen3_8_27b_uncensored_bf16의 계보/명칭을 따릅니다.
이는 모든 입력에 대해 무조건적인 무거절 동작을 별도로 인증했다는 의미는 아닙니다.

용도

한국어 대화, 한국어 기술 질의, 코딩 보조 및 로컬 LLM 실험을 목적으로 배포합니다.

README history 1 version

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

  1. 2026-09-15Add files using upload-large-folder tool4132efa2.1 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 app" button that hands off directly to a local runtime of your choice - Infrahuman, LM Studio, or Ollama. No API keys, no subscription, no prompt leakage.