Instructions to use meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 with Transformers:

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

pipe = pipeline("image-text-to-text", model="meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8")
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("meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8")
model = AutoModelForMultimodalLM.from_pretrained("meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8", 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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
		"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/meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8

SGLang

How to use meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 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 "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8" \
    --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": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
		"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 "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8" \
        --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": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
		"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 meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 with Docker Model Runner:
```
docker model run hf.co/meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8
```

Quantizer: Running into an error with quantization "TypeError: 'dict' object is not callable"

#24

by AaronVogler - opened Apr 16, 2025

Discussion

AaronVogler

Apr 16, 2025

I get the following error when trying to load the model on CPU...

Anyone have any idea as to what's going on?

 File "/usr/local/lib/python3.10/dist-packages/transformers/pipelines/__init__.py", line 942, in pipeline
    framework, model = infer_framework_load_model(
  File "/usr/local/lib/python3.10/dist-packages/transformers/pipelines/base.py", line 291, in infer_framework_load_model
    model = model_class.from_pretrained(model, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/transformers/models/auto/auto_factory.py", line 571, in from_pretrained
    return model_class.from_pretrained(
  File "/usr/local/lib/python3.10/dist-packages/transformers/modeling_utils.py", line 279, in _wrapper
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/transformers/modeling_utils.py", line 4369, in from_pretrained
    hf_quantizer.preprocess_model(
  File "/usr/local/lib/python3.10/dist-packages/transformers/quantizers/base.py", line 224, in preprocess_model
    self._convert_model_for_quantization(model)
  File "/usr/local/lib/python3.10/dist-packages/transformers/quantizers/base.py", line 313, in _convert_model_for_quantization
    parent_module._modules[name] = MODULES_TO_PATCH_FOR_QUANTIZATION[module_class_name](
TypeError: 'dict' object is not callable

    at ChildProcess.<anonymous> (/home/inf3rnus/api/jobRunner/library/container/createTestImage/index.js:61:18)
    at ChildProcess.emit (/home/inf3rnus/api/lib/events.js:519:28)
    at maybeClose (/home/inf3rnus/api/lib/internal/child_process.js:1105:16)
    at ChildProcess._handle.onexit (/home/inf3rnus/api/lib/internal/child_process.js:305:5)
    at Process.callbackTrampoline (node:internal/async_hooks:130:17) {stack: "Error: ERROR: pip's dependency resolver does …Trampoline (node:internal/async_hooks:130:17)", message: "ERROR: pip's dependency resolver does not cu…](
TypeError: 'dict' object is not callable
"}

Thanks,
Aaron

zhuokai

Apr 17, 2025

I got the same error. Were you able to fix it?

AaronVogler

Apr 17, 2025

@zhuokai No, I have been unable to fix it, the non quant versions of the model work on CPU, so I don't know what the deal is, presumably some conflict between compressed-tensors and transformers==4.51.3

wukaixingxp

Apr 17, 2025

Can you try pip install from source? This is a fix in this commit that may help.

BarathwajAnandan

May 3, 2025

yeah, install from source is the fix.

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