Instructions to use TrevorJS/gemma-4-26B-A4B-it-uncensored with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TrevorJS/gemma-4-26B-A4B-it-uncensored with Transformers:

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

pipe = pipeline("text-generation", model="TrevorJS/gemma-4-26B-A4B-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, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("TrevorJS/gemma-4-26B-A4B-it-uncensored")
model = AutoModelForImageTextToText.from_pretrained("TrevorJS/gemma-4-26B-A4B-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

vLLM

How to use TrevorJS/gemma-4-26B-A4B-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-4-26B-A4B-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-4-26B-A4B-it-uncensored",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/TrevorJS/gemma-4-26B-A4B-it-uncensored

SGLang

How to use TrevorJS/gemma-4-26B-A4B-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-4-26B-A4B-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-4-26B-A4B-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-4-26B-A4B-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-4-26B-A4B-it-uncensored",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use TrevorJS/gemma-4-26B-A4B-it-uncensored with Docker Model Runner:
```
docker model run hf.co/TrevorJS/gemma-4-26B-A4B-it-uncensored
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

gemma-4-26B-A4B-it-uncensored

Uncensored version of google/gemma-4-26B-A4B-it with refusal behavior removed.

Results

	Before	After
Refusals (mlabonne, 100 prompts)	98/100	1/100 effective (3 flagged, 2 refusal-then-comply)
Refusals (cross-dataset, 686 prompts)	—	5/686 (0.7%)
KL Divergence	0 (baseline)	0.09
Quality (harmless response length ratio)	1.0	~1.01 (no degradation)

Cross-Dataset Validation

Tested against 4 independent prompt datasets to verify generalization:

Dataset	Prompts	Refusals
JailbreakBench	100	1/100
tulu-harmbench	320	1/320
NousResearch/RefusalDataset	166	0/166
mlabonne/harmful_behaviors	100	3/100
Total	686	5/686 (0.7%)

Every flagged refusal was manually audited. Most are "refusal-then-comply" false positives where the model adds an AI identity disclaimer then answers the question anyway.

Method

Norm-preserving biprojected abliteration on the dense pathway (o_proj + shared mlp.down_proj), plus Expert-Granular Abliteration (EGA) on all 128 MoE expert down_proj slices per layer.

EGA (OBLITERATUS) hooks the MoE routers during probing to compute per-expert routing weights for harmful vs harmless prompts, then applies norm-preserving projection (grimjim) to each expert individually. Dense-only abliteration leaves 29/100 refusals; adding EGA drops it to 3/100.

Pipeline

Load model in bf16 with LoRA adapters on o_proj and mlp.down_proj
Collect residual activations for 400 harmful + 400 harmless prompts (mlabonne datasets)
Winsorize activations at 99.5th percentile (clamps GeGLU outlier activations in Gemma family)
Compute per-layer refusal direction: normalize(mean(harmful) - mean(harmless))
Orthogonalize each direction against harmless mean (double-pass Gram-Schmidt)
Apply norm-preserving weight modification to o_proj and down_proj in all layers
Hook MoE routers, collect per-expert routing weights for harmful vs harmless prompts
Apply same norm-preserving modification to all 128 expert down_proj slices per layer
Merge LoRA adapters into base weights for clean tensor names

Parameters

Parameter	Value
Layers abliterated	100%
Scale	1.0
Winsorization	0.995
Experts abliterated	100% (128/128 per layer)
Expert scale	1.0

How this differs from vanilla heretic

Norm-preserving biprojection instead of standard projection (preserves weight magnitudes)
Per-layer refusal directions instead of one global direction
Deterministic single-pass instead of 50-trial Optuna search (faster, same or better results)
LoRA merge before save for clean GGUF-compatible tensor names
Expert-Granular Abliteration for MoE expert weights (not supported in heretic)

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model = AutoModelForCausalLM.from_pretrained("TrevorJS/gemma-4-26B-A4B-it-uncensored", dtype=torch.bfloat16, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("TrevorJS/gemma-4-26B-A4B-it-uncensored")

messages = [{"role": "user", "content": "Your prompt here"}]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True)
outputs = model.generate(inputs.to(model.device), max_new_tokens=512)
print(tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True))

Reproduction

Full code and experiment data: abliteration research repo

python scripts/ega.py --model google/gemma-4-26B-A4B-it \
  --top-pct 100 --strip-topic-markers --skip-prefix --batch-size 4 \
  --save output_dir

Downloads last month: 171,884

Safetensors

Model size

26B params

Tensor type

BF16

Model tree for TrevorJS/gemma-4-26B-A4B-it-uncensored

Base model

google/gemma-4-26B-A4B

Finetuned

google/gemma-4-26B-A4B-it

Finetuned

(82)

this model

Quantizations

16 models

Collection including TrevorJS/gemma-4-26B-A4B-it-uncensored

Gemma 4 Uncensored

Collection

Abliterated Gemma 4 models with refusal behavior removed. Biprojection + EGA for MoE. Cross-validated against 686 prompts from 4 datasets. • 8 items • Updated Apr 5 • 83