Instructions to use meta-llama/Llama-3.2-3B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use meta-llama/Llama-3.2-3B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="meta-llama/Llama-3.2-3B-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-3B-Instruct")
model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.2-3B-Instruct", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use meta-llama/Llama-3.2-3B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "meta-llama/Llama-3.2-3B-Instruct"
# 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-3.2-3B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/meta-llama/Llama-3.2-3B-Instruct

SGLang

How to use meta-llama/Llama-3.2-3B-Instruct 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-3.2-3B-Instruct" \
    --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-3.2-3B-Instruct",
		"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 "meta-llama/Llama-3.2-3B-Instruct" \
        --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-3.2-3B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use meta-llama/Llama-3.2-3B-Instruct with Docker Model Runner:
```
docker model run hf.co/meta-llama/Llama-3.2-3B-Instruct
```

Token indices sequence length is longer than the specified maximum sequence length for this model (269923 > 131072)

#23

by wasimsafdar - opened Oct 15, 2024

Discussion

wasimsafdar

Oct 15, 2024

I am getting the above-mentioned error while using llama3.2 for vision. If I use an image URL, it works fine. However, if I upload the same image from a local directory, I get an error. I am using transformed from hugging face. My device is a Mac M1 pro, with 16 GB RAM.

There is also a runtime error "RuntimeError: MPS backend out of memory (MPS allocated: 17.08 GB, other allocations: 672.00 KB, max allowed: 18.13 GB). Tried to allocate 1.54 GB on private pool. Use PYTORCH_MPS_HIGH_WATERMARK_RATIO=0.0 to disable upper limit for memory allocations (may cause system failure)."

I searched Google and people said we need to set "PYTORCH_MPS_HIGH_WATERMARK_RATIO" to 0.7. However, I do not know how to set it in PyTorch. Most answers are related to Stable diffusion webui.

This is my code.

import warnings
from transformers import pipeline
import torch
access_token = "hf_...."

warnings.filterwarnings('ignore')

model_id = "meta-llama/Llama-3.2-3B-Instruct"
pipe = pipeline(
"text-generation",
model=model_id,
token=access_token,
torch_dtype=torch.bfloat16,
device_map="auto",
)

import base64

def encode_image(image_path):
with open(image_path, "rb") as image_file:
return base64.b64encode(image_file.read()).decode('utf-8')
base64_image = encode_image("/Users/mac/Downloads/Llama_Repo.jpeg")

messages = [
{"role": "user",
"content": [
{"type": "text",
"text": "describe the image in one sentence"
},
{"type": "image_url",
"image_url": {"url": f"data:image/jpeg;base64,{base64_image}"}
}
]
},
]

outputs = pipe(
messages,
max_new_tokens=256,
)
response = outputs[0]["generated_text"][-1]["content"]
print(response)

Please guide me on how to solve it in Pytorch. Thanks

jdc4429

Oct 27, 2024

I have nothing to do with this project, just thought I would mention that your posting a problem about a vision model. This is not a vision model. This is 3B instruct. Vision model is 11B.

wasimsafdar

Oct 27, 2024

@jdc4429 You are right. This is not a vision model but it is fined-tuned to reason images and lighter weight than heavier vision model. I understood the issue. It was related to Python configuration. Setting PYTORCH_MPS_HIGH_WATERMARK_RATIO=0.0 resolves the issue.

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