Instructions to use wls04/gemma_web with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use wls04/gemma_web with Transformers:

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

pipe = pipeline("image-text-to-text", model="wls04/gemma_web")
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("wls04/gemma_web")
model = AutoModelForMultimodalLM.from_pretrained("wls04/gemma_web")
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 wls04/gemma_web with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "wls04/gemma_web"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wls04/gemma_web",
		"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/wls04/gemma_web

SGLang

How to use wls04/gemma_web 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 "wls04/gemma_web" \
    --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": "wls04/gemma_web",
		"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 "wls04/gemma_web" \
        --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": "wls04/gemma_web",
		"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 wls04/gemma_web with Docker Model Runner:
```
docker model run hf.co/wls04/gemma_web
```

gemma_web

File size: 2,038 Bytes

20acb38

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 1000,
  "global_step": 67,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.15023474178403756,
      "grad_norm": 31.56661033630371,
      "learning_rate": 9.972609476841368e-06,
      "loss": 1.7659,
      "step": 10
    },
    {
      "epoch": 0.3004694835680751,
      "grad_norm": 35.23997497558594,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.5183,
      "step": 20
    },
    {
      "epoch": 0.4507042253521127,
      "grad_norm": 5.03916072845459,
      "learning_rate": 7.033683215379002e-06,
      "loss": 0.278,
      "step": 30
    },
    {
      "epoch": 0.6009389671361502,
      "grad_norm": 3.3426177501678467,
      "learning_rate": 4.477357683661734e-06,
      "loss": 0.2194,
      "step": 40
    },
    {
      "epoch": 0.7511737089201878,
      "grad_norm": 3.752589702606201,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.1969,
      "step": 50
    },
    {
      "epoch": 0.9014084507042254,
      "grad_norm": 5.254862308502197,
      "learning_rate": 4.322727117869951e-07,
      "loss": 0.1813,
      "step": 60
    },
    {
      "epoch": 1.0,
      "step": 67,
      "total_flos": 265909239283712.0,
      "train_loss": 0.48691165269310793,
      "train_runtime": 25225.3606,
      "train_samples_per_second": 0.338,
      "train_steps_per_second": 0.003
    }
  ],
  "logging_steps": 10,
  "max_steps": 67,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 265909239283712.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}