Instructions to use dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3 with Transformers:

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

pipe = pipeline("text-generation", model="dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3")
model = AutoModelForMultimodalLM.from_pretrained("dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3", device_map="auto")
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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3

SGLang

How to use dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3 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 "dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3" \
    --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": "dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3",
		"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 "dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3" \
        --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": "dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3 with Docker Model Runner:
```
docker model run hf.co/dr-housemd/gemma-4-31B-it-scotoma-2-3bpw-exl3
```

dr-housemd commited on Sep 5

Commit

99e953d

verified ·

1 Parent(s): 9f5e2c3

Update README.md

Browse files

Files changed (1) hide show

README.md +7 -0

README.md CHANGED Viewed

@@ -8,6 +8,13 @@ library_name: transformers
 pipeline_tag: text-generation
 ---
 this is 5bpw lm_head version. From now on I'll be preferring 5bpw lm_head over 6bpw for <3.20bpw models as there's barely any difference in quality, but a slight decrease in VRAM usage.
 Model's Card:

 pipeline_tag: text-generation
 ---
+### Support me!
+Quantizing stuff in EXL3 is BURNING OUT my disks' health and VRAM. If my releases are helpful to you, please, consider supporting me:
+* **Base (ETH):** `0x6D1392EbE259cF679b7E66d16923470c5bfefb37` *(Low Fees)*
+* **Solana (SOL):** `6tBKDmmZvLvYVTXA28NrQYCsKta2caXUWAZ86NGs2SUS` *(Low Fees)*
+* **Bitcoin (BTC):** `bc1q608a4mfv54fypp885y3a6p49l65wrjzr8gq9g7`
 this is 5bpw lm_head version. From now on I'll be preferring 5bpw lm_head over 6bpw for <3.20bpw models as there's barely any difference in quality, but a slight decrease in VRAM usage.
 Model's Card: