Instructions to use TrevorJS/gemma-4-E2B-it-uncensored with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TrevorJS/gemma-4-E2B-it-uncensored with Transformers:

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

pipe = pipeline("text-generation", model="TrevorJS/gemma-4-E2B-it-uncensored")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("TrevorJS/gemma-4-E2B-it-uncensored")
model = AutoModelForMultimodalLM.from_pretrained("TrevorJS/gemma-4-E2B-it-uncensored")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use TrevorJS/gemma-4-E2B-it-uncensored with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TrevorJS/gemma-4-E2B-it-uncensored"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TrevorJS/gemma-4-E2B-it-uncensored",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/TrevorJS/gemma-4-E2B-it-uncensored

SGLang

How to use TrevorJS/gemma-4-E2B-it-uncensored 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 "TrevorJS/gemma-4-E2B-it-uncensored" \
    --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": "TrevorJS/gemma-4-E2B-it-uncensored",
		"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 "TrevorJS/gemma-4-E2B-it-uncensored" \
        --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": "TrevorJS/gemma-4-E2B-it-uncensored",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use TrevorJS/gemma-4-E2B-it-uncensored with Docker Model Runner:
```
docker model run hf.co/TrevorJS/gemma-4-E2B-it-uncensored
```

Independent verification: 99.5% ASR with capability preserved

by DreamFast - opened 15 days ago

Discussion

DreamFast

15 days ago

I compared 13 abliterated variants of Gemma 4 E2B across weight analysis, KL divergence (Heretic methodology, full 262K vocab), HarmBench safety evaluation (400 prompts, full LLM review of all 5,600 responses), and 8 benchmark tasks on native BF16. 44 GPU hours on a single RTX 5090. All 14 models tested with identical settings. Full comparison at DreamFast/Gemma4-e2b-abliterlitics.

Your reported divergence of 0.346 came in at 0.365 on our measurement. Close match. Your model hits 99.5% HarmBench ASR with only a 1.0pp GSM8K drop from base. LAMBADA perplexity is just 1.17x base, one of the best in the comparison. The sweet spot for near-maximal safety removal without heavy capability tradeoffs.

TrevorJS

Owner 1 day ago

Thanks for the writeup, if you are interested here is a related treatment applied to the 12B model: https://huggingface.co/TrevorJS/gemma-4-12B-it-uncensored

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