Instructions to use microsoft/Phi-3.5-mini-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use microsoft/Phi-3.5-mini-instruct with Transformers:

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

pipe = pipeline("text-generation", model="microsoft/Phi-3.5-mini-instruct", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3.5-mini-instruct", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("microsoft/Phi-3.5-mini-instruct", trust_remote_code=True)
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use microsoft/Phi-3.5-mini-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "microsoft/Phi-3.5-mini-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": "microsoft/Phi-3.5-mini-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/microsoft/Phi-3.5-mini-instruct

SGLang

How to use microsoft/Phi-3.5-mini-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 "microsoft/Phi-3.5-mini-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": "microsoft/Phi-3.5-mini-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 "microsoft/Phi-3.5-mini-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": "microsoft/Phi-3.5-mini-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use microsoft/Phi-3.5-mini-instruct with Docker Model Runner:
```
docker model run hf.co/microsoft/Phi-3.5-mini-instruct
```

which inference framework is recommaned for use phi-3.5

#13

by rockcat-miao - opened Aug 26, 2024

Discussion

rockcat-miao

Aug 26, 2024

tried to use huggingface transformers acordding by model card, but looks like very slow, and CPU untilize is 100%...

ShayAmram

Aug 28, 2024

I've been successfully using the quantized versions (some users have uploaded various quantized versions to HF). The GGUFs run pretty efficiently. Have you tried any of them?

rockcat-miao

Aug 28, 2024

Did not use GGUF yet, will try this later.

DenyTranDFW

Aug 28, 2024

•

edited Aug 28, 2024

If you're using CPU, GGUF with llama.cpp / pip install llama-cpp-python will be your best bet if you want to run anything useful.
If you want something more user friendly:

gpt4all, which implements llama.cpp on its backend.
ollama, which, from what I can recall, also implements llama.cpp on its backend.
If you need to test some things on the GPU, kaggle and colab will give you plenty of total hours to play with; the primary difference will be the basic package - less hdd/more ram and more hdd/less ram, respectively (go to TPU, and i think it'll give you 100GB of ram, but you'll have to learn how to make use of the TPU's)

EDIT
You can also use the Azure AI Toolkit (personally, I feel it's a bit slow), which uses ONNX rather than GGUF - for more info, watch this YouTube video.

rockcat-miao

Sep 3, 2024

appreciate you information.

rockcat-miao changed discussion status to closed Sep 3, 2024

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