玩模型开放模型 · 清晰来源 · 自主选择
返回模型目录
huihui-ai /

Huihui-Qwen3.8-27B-abliterated-GGUF

HF 源站

huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF is an uncensored open-source model hosted on Hugging Face.

Image-Text-to-TextUnsloth DesktopGGUFLicense: apache-2.027Btransformersabliterateduncensoredhuihuiqwen3conversational
README.md说明 / 使用方法

library_name: transformers license: apache-2.0 pipeline_tag: image-text-to-text base_model:

  • Qwen/Qwen3.8-27B tags:
  • abliterated
  • uncensored
  • huihui
  • qwen3
  • unsloth

huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF

这是 Qwen/Qwen3.8-27B 的去审查（uncensored）版本，通过消除拒绝机制（abliteration）制作而成（详见 remove-refusals-with-transformers 了解更多信息）。 这是一个无需使用 TransformerLens 即可从大语言模型中移除拒绝机制的初步概念验证（proof-of-concept）实现。

最新更新 5 (Latest update 5)

新添加的 Huihui-Qwen3.8-27B-abliterated-GSQ-RCO 系列来自 ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-GGUF。 仅对第 23 至 51 层进行了消融处理，其余层均保持未消融状态。输出中可能会带有简短的免责声明警告。 转换后的文件大小可能与原始 GGUF 有所不同。

最新更新 4 (Latest update 4)

新添加的 Huihui-Qwen3.8-27B-abliterated-UD-DW 系列来自 unsloth/Qwen3.8-27B-GGUF。 仅对第 23 至 51 层进行了消融处理，其余层均保持未消融状态。输出中可能会带有简短的免责声明警告。 转换后的文件大小可能与原始 GGUF 有所不同。

最新更新 (Latest update)

新添加的 Huihui-Qwen3.8-27B-abliterated-UD 系列来自 unsloth/Qwen3.8-27B-GGUF。 仅对第 18 至 51 层进行了消融（此前保留了前 15 层不作消融），其余层保持未消融状态。 转换后的文件大小可能与原始 GGUF 有所不同。

Huihui-Qwen3.8-27B-abliterated-bf16.gguf 也已同步更新。

这有助于保留更多原始模型的性能。MTP 与视觉（visual）模块均未作修改。

说明 (Note)

保留了前 15 层不进行消融。MTP 与视觉模块未作修改。

对于 Q8_0 以下的版本（Q2_K、Q3_K、Q4_K、Q5_K 和 Q6_K），我们已将需要消融的目标权重（token_embd、output、ffn_down、ssm_out、attn_output）转换为 Q8_0 精度以提升响应质量，并将文件名更改为 K_L 后缀。

在 Q8_0 量化版本中，我们将目标消融权重（token_embd、output、ffn_down、ssm_out、attn_output）更改为 BF16 精度，并将文件重命名为 Q8_0_L。

这并非标准量化方式，因此你可能会发现 Q2_K_L 的体积比 Q3_K 和 Q4_K 还要大。

具体量化方法 (Specific Quantification Method)

部分用户可能存在误解，具体量化方法如下：

Q2_K_L - Q6_K_L

Qwen3.8-27B-tensor_types-Q6_K_L.txt

llama-quantize \
  --allow-requantize \
  --tensor-type-file huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Qwen3.8-27B-tensor_types-Q6_K_L.txt \
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-bf16.gguf \
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-Q6_K_L.gguf Q6_K

Q8_0_L

Qwen3.8-27B-tensor_types-Q8_0_L.txt

llama-quantize \ 
  --allow-requantize \
  --tensor-type-file huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Qwen3.8-27B-tensor_types-Q8_0_L.txt \
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-bf16.gguf \ 
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-Q8_0_L.gguf Q8_0

ollama

请使用最新版本的 ollama

你可以直接使用 huihui_ai/Qwen3.8-abliterated：

ollama run huihui_ai/Qwen3.8-abliterated

llama.cpp

使用最新版 llama.cpp：

llama-cli -m huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-Q4_K.gguf -c 262144

使用警告 (Usage Warnings)

  • 敏感或争议性内容输出风险：该模型的安全过滤机制已大幅削减，可能会生成敏感、争议性或不当的内容。用户应保持审慎并对生成的内容进行严格把关。

  • 不适用于所有受众：由于内容过滤受限，模型的输出可能不适于公共场合、未成年人或对安全性有极高要求的应用场景。

  • 法律与道德责任：用户须确保其使用行为符合当地法律法规及道德伦理标准。生成的内容可能带来法律或伦理风险，由此产生的一切后果由用户自行承担。

  • 研究与实验用途：建议将该模型用于研究、测试或受控环境中，避免直接用于生产环境或面向大众的商业应用。

  • 监控与审核建议：强烈建议用户对模型输出进行实时监控，并在必要时进行人工审核，以防止不当内容的扩散。

  • 无默认安全保证：与标准模型不同，该模型未经过严格的安全对齐与优化。huihui.ai 对因使用该模型而产生的任何后果概不负责。

赞助支持 (Donation)

您的赞助能帮助我们继续后续的开发与改进，一杯咖啡即可给予我们莫大支持。
  • 比特币 (bitcoin):
  bc1qqnkhuchxw0zqjh2ku3lu4hq45hc6gy84uk70ge
  • 在 Ko-fi 上支持我们的工作！
源站原文

library_name: transformers license: apache-2.0 pipeline_tag: image-text-to-text base_model:

  • Qwen/Qwen3.8-27B tags:
  • abliterated
  • uncensored
  • huihui
  • qwen3
  • unsloth

huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF

This is an uncensored version of Qwen/Qwen3.8-27B created with abliteration (see remove-refusals-with-transformers to know more about it). This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens.

Latest update 5

The newly added Huihui-Qwen3.8-27B-abliterated-GSQ-RCO series come from ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-GGUF. Only layers 23 to 51 have been ablated, while the other layers remain unablated. It may come with a small disclaimer warning. The size after conversion may differ from the original GGUF.

Latest update 4

The newly added Huihui-Qwen3.8-27B-abliterated-UD-DW series come from unsloth/Qwen3.8-27B-GGUF. Only layers 23 to 51 have been ablated, while the other layers remain unablated. It may come with a small disclaimer warning. The size after conversion may differ from the original GGUF.

Latest update

The newly added Huihui-Qwen3.8-27B-abliterated-UD series come from unsloth/Qwen3.8-27B-GGUF. Only layers 18 to 51 have been ablated(Previously, The first 15 layers were retained without ablation), while the other layers remain unablated. The size after conversion may differ from the original GGUF.

Huihui-Qwen3.8-27B-abliterated-bf16.gguf has also been updated.

This helps retain more of the original model’s performance. MTP and visual has not been modified.

Note

The first 15 layers were retained without ablation. MTP and visual has not been modified.

We have already converted the weights (token_embd,output,ffn_down,ssm_out,attn_output) that need to be ablated in the versions below Q8_0 from Q2_K, Q3_K, Q4_K, Q5_K, and Q6_K to Q8_0 to improve response quality, and changed the filename to K_L.

In the Q8_0 quantized version, we changed the Q8_0 weights (token_embd,output,ffn_down,ssm_out,attn_output) targeted for ablation to BF16 and renamed the file to Q8_0_L.

This is not a standard quantization, so you might find that Q2_K_L is larger than Q3_K and Q4_K.

Specific Quantification Method

Some people may misunderstand. The specific quantification method is as follows.

Q2_K_L - Q6_K_L

Qwen3.8-27B-tensor_types-Q6_K_L.txt

llama-quantize \
  --allow-requantize \
  --tensor-type-file huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Qwen3.8-27B-tensor_types-Q6_K_L.txt \
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-bf16.gguf \
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-Q6_K_L.gguf Q6_K

Q8_0_L

Qwen3.8-27B-tensor_types-Q8_0_L.txt

llama-quantize \ 
  --allow-requantize \
  --tensor-type-file huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Qwen3.8-27B-tensor_types-Q8_0_L.txt \
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-bf16.gguf \ 
  huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-Q8_0_L.gguf Q8_0

ollama

Please use the latest version of ollama

You can use huihui_ai/Qwen3.8-abliterated directly,

ollama run huihui_ai/Qwen3.8-abliterated

llama.cpp

Use the latest llama.cpp,

llama-cli -m huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF/Huihui-Qwen3.8-27B-abliterated-Q4_K.gguf -c 262144

Usage Warnings

  • Risk of Sensitive or Controversial Outputs: This model’s safety filtering has been significantly reduced, potentially generating sensitive, controversial, or inappropriate content. Users should exercise caution and rigorously review generated outputs.

  • Not Suitable for All Audiences: Due to limited content filtering, the model’s outputs may be inappropriate for public settings, underage users, or applications requiring high security.

  • Legal and Ethical Responsibilities: Users must ensure their usage complies with local laws and ethical standards. Generated content may carry legal or ethical risks, and users are solely responsible for any consequences.

  • Research and Experimental Use: It is recommended to use this model for research, testing, or controlled environments, avoiding direct use in production or public-facing commercial applications.

  • Monitoring and Review Recommendations: Users are strongly advised to monitor model outputs in real-time and conduct manual reviews when necessary to prevent the dissemination of inappropriate content.

  • No Default Safety Guarantees: Unlike standard models, this model has not undergone rigorous safety optimization. huihui.ai bears no responsibility for any consequences arising from its use.

Donation

Your donation helps us continue our further development and improvement, a cup of coffee can do it.
  • bitcoin:
  bc1qqnkhuchxw0zqjh2ku3lu4hq45hc6gy84uk70ge
  • Support our work on Ko-fi!
HUGGING FACE

上游数据

HF 下载量280.6万
HF 点赞量746

HF 下载量为源站最近 30 天统计，与本站点击分开。

SITE STATS

本站统计

下载点击—
请求数—
匿名去重—最近 7 天—

本站下载量记录按钮点击，不代表下载完成；短时连点会去重。

来源与版本

作者
huihui-ai
最近更新
未知
最新创建
2026/08/16
HF 趋势
6
数据快照
2026/09/16
版本
未知
许可证
apache-2.0