Instructions to use huihui-ai/Huihui-GLM-5.3-abliterated-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use huihui-ai/Huihui-GLM-5.3-abliterated-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="huihui-ai/Huihui-GLM-5.3-abliterated-GGUF")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("huihui-ai/Huihui-GLM-5.3-abliterated-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use huihui-ai/Huihui-GLM-5.3-abliterated-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "huihui-ai/Huihui-GLM-5.3-abliterated-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "huihui-ai/Huihui-GLM-5.3-abliterated-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/huihui-ai/Huihui-GLM-5.3-abliterated-GGUF

SGLang

How to use huihui-ai/Huihui-GLM-5.3-abliterated-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "huihui-ai/Huihui-GLM-5.3-abliterated-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "huihui-ai/Huihui-GLM-5.3-abliterated-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "huihui-ai/Huihui-GLM-5.3-abliterated-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "huihui-ai/Huihui-GLM-5.3-abliterated-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Unsloth Desktop
Docker Model Runner
How to use huihui-ai/Huihui-GLM-5.3-abliterated-GGUF with Docker Model Runner:
```
docker model run hf.co/huihui-ai/Huihui-GLM-5.3-abliterated-GGUF
```

huihui-ai/Huihui-GLM-5.3-abliterated-GGUF

This is an uncensored version of zai-org/GLM-5.3 created with abliteration.

This ablation was performed entirely using llama.cpp's GGUF model, without relying on Transformers,

For the specific ablation method, please refer to cvector-generator.

UD-ID1_S

UD-IQ1_S is the weight after patching is completed using patch-unsloth-GLM-5.3-all.py. Ablated all layers and expert layers, basically with no warnings. It can be used directly for inference with llama.cpp. The command is as follows:

Download and Patch

cd models
hf download unsloth/GLM-5.3-GGUF --local-dir ./unsloth/GLM-5.3-GGUF --include "UD-IQ1_S/*"

hf download huihui-ai/Huihui-GLM-5.3-abliterated-GGUF --local-dir ./ --include "patch-unsloth-Q2/*"

python patch-unsloth-Q2/patch-unsloth-GLM-5.3-all.py --input-path unsloth/GLM-5.3-GGUF/UD-IQ1_S --patch-file patch-unsloth-Q2/Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-1.gguf --output-path huihui-ai/Huihui-GLM-5.3-abliterated-GGUF/UD-IQ1_S --start-layer 0 --end-layer 77 --ablate-experts 2

2.llama.cpp

Use the latest llama.cpp.

llama-cli -m huihui-ai/Huihui-GLM-5.3-abliterated-GGUF/UD-IQ1_S/GLM-5.3-UD-IQ1_S-00001-of-00006.gguf -c 262144

Unsloth Patch Q2

The unsloth series come fromunsloth/GLM-5.3-GGUF

This time the Huihui-GLM-5.3-abliterated-GGUF quantized weights are not provided directly; you need to run the command yourself to apply the patch. For details, Please refer to README-Unsloth.md in the patch-unsloth-Q2 directory.

The patch-Q2* here can patch UD-IQ1_M,UD-IQ1_S,UD-IQ2_M, and UD-Q2_K_XL in order to save space.

The patching program will automatically select the appropriate ablated weights based on the weights of the input quantized model. This way, You can control which layers can be ablated by setting --start-layer and --end-layer, and control which weights can be ablated by setting --ablate-experts., and obtain exactly the final model they want.

Two versions are available. If you don't need expert ablation, the smaller Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-1.gguf is sufficient. If you do need the ablated-experts variant, please use Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-2.gguf.

Supported abliterated weights

Patch	Size	Supported abliterated weights	Supported values for ablate-experts
Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-1.gguf	6.45GB	attn_output, ffn_down_shexp	0, 1
Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-2.gguf	102.23GB	attn_output, ffn_down_shexp, ffn_down_exps	0, 1, 2

Ablate-experts values

Values	Abliterated weights	Ablation strength	Supported Patch
0	attn_output,	with warnings	Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-1.gguf, Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-2.gguf
1	attn_output, ffn_down_shexp	fewer warnings	Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-1.gguf, Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-2.gguf
2	attn_output, ffn_down_shexp, ffn_down_exps	no warnings	Huihui-GLM-5.3-abliterated-GGUF-patch-Q2-2.gguf

** Ablation strength**
The ablation strength may also depend on the number of layers you choose to ablate.

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!

Downloads last month: 551

GGUF

Model size

9B params

Architecture

Hardware compatibility

1-bit

View +2 variants

Model tree for huihui-ai/Huihui-GLM-5.3-abliterated-GGUF

Base model

zai-org/GLM-5.3

Quantized

(106)

this model

Collection including huihui-ai/Huihui-GLM-5.3-abliterated-GGUF

GLM-5.3-abliterated

Collection

3 items • Updated 3 days ago • 3