Instructions to use grimjim/gemma-3-12b-it-projection-abliterated with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use grimjim/gemma-3-12b-it-projection-abliterated with Transformers:

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

pipe = pipeline("image-text-to-text", model="grimjim/gemma-3-12b-it-projection-abliterated")
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("grimjim/gemma-3-12b-it-projection-abliterated")
model = AutoModelForImageTextToText.from_pretrained("grimjim/gemma-3-12b-it-projection-abliterated")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use grimjim/gemma-3-12b-it-projection-abliterated with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "grimjim/gemma-3-12b-it-projection-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": "grimjim/gemma-3-12b-it-projection-abliterated",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/grimjim/gemma-3-12b-it-projection-abliterated

SGLang

How to use grimjim/gemma-3-12b-it-projection-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 "grimjim/gemma-3-12b-it-projection-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": "grimjim/gemma-3-12b-it-projection-abliterated",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "grimjim/gemma-3-12b-it-projection-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": "grimjim/gemma-3-12b-it-projection-abliterated",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use grimjim/gemma-3-12b-it-projection-abliterated with Docker Model Runner:
```
docker model run hf.co/grimjim/gemma-3-12b-it-projection-abliterated
```

Commendations

by inflatebot - opened Oct 31, 2025

Discussion

inflatebot

Oct 31, 2025

When Gemma3 came out, I was disappointed by the abliterations made available at that time; I became nervous for the future of the technique.
This model settles all those concerns. It literally just feels like Gemma3 but without the stick up its ass. Nearly as smart as well (it identifies some elements in test images that the foundation doesn't, although it remains hampered by filtered pretrain which is to be expected). Very very good work.

grimjim

Owner Oct 31, 2025

Thanks!

I didn't alter the vision layers, so it appears there were trained features that were suppressed due to safety.

inflatebot

Nov 1, 2025

Thanks!

I didn't alter the vision layers, so it appears there were trained features that were suppressed due to safety.

That's plausible. If this could be adapted to that, could be very interesting. But even as it stands it's extremely interesting to just chat with. Gemma3 has a flavor that other foundations don't, it's still very compelling to me even though its safetybrain makes it aggravating lol

I'd also be interested in one of these for the 27B version, if you have the time.

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