Instructions to use huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated with Transformers:

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

pipe = pipeline("text-generation", model="huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated")
model = AutoModelForMultimodalLM.from_pretrained("huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated

SGLang

How to use huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated 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/DeepSeek-R1-Distill-Llama-70B-abliterated" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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/DeepSeek-R1-Distill-Llama-70B-abliterated" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated with Docker Model Runner:
```
docker model run hf.co/huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated
```

Is this meant to be uncensored?

by BillS1 - opened Apr 24, 2025

Discussion

BillS1

Apr 24, 2025

•

edited Apr 24, 2025

While the 32B consistently complied with NSFW content such as descriptive violence, this 70B variant has (more often than not) refused to generate NSFW content "due to the AI's guidelines", and even directly asking it suggests this unlike the 32B variant. The times when it does allow NSFW is still with its restrictions and even asks if I'm over 18. I also get similar problems with the QWQ abliterated model. Were the NSFW restrictions not properly removed during the abliterating unlike the 32B variant?

MikaSouthworth

Sep 22, 2025

I wonder the same thing

huihui-ai

Owner Sep 22, 2025

The first layer of this model is not ablated, which may still impose restrictions. If all layers are ablated, it could lead to garbled output—this is also a kind of balancing strategy?

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment