Instructions to use PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds with Transformers:

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

pipe = pipeline("image-text-to-text", model="PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds")
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("PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds")
model = AutoModelForImageTextToText.from_pretrained("PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds")
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 PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds",
		"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/PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds

SGLang

How to use PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds 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 "PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds" \
    --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": "PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds",
		"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 "PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds" \
        --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": "PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds",
		"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 PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds with Docker Model Runner:
```
docker model run hf.co/PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds
```

gemma-3-4b-pt-InitializedEmbeds

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the Linear merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

merge_method: linear
dtype: bfloat16
models:
  - model: google/gemma-3-4b-pt
    parameters:
      weight: 1.0
  - model: google/gemma-3-4b-it
    parameters:
      weight: 0.0
tokenizer:
  source: google/gemma-3-4b-it
  tokens:
    <start_of_turn>:
      source: google/gemma-3-4b-it
      force: true
    <end_of_turn>:
      source: google/gemma-3-4b-it
      force: true

Downloads last month: 78

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds

google/gemma-3-4b-it

google/gemma-3-4b-pt

Merge model

this model

Paper for PJMixers-Dev/gemma-3-4b-pt-InitializedEmbeds

Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time

Paper • 2203.05482 • Published Mar 10, 2022 • 8