Instructions to use vonjack/Phi-3-mini-4k-instruct-LLaMAfied with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use vonjack/Phi-3-mini-4k-instruct-LLaMAfied with Transformers:

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

pipe = pipeline("text-generation", model="vonjack/Phi-3-mini-4k-instruct-LLaMAfied")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("vonjack/Phi-3-mini-4k-instruct-LLaMAfied")
model = AutoModelForCausalLM.from_pretrained("vonjack/Phi-3-mini-4k-instruct-LLaMAfied", 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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use vonjack/Phi-3-mini-4k-instruct-LLaMAfied with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "vonjack/Phi-3-mini-4k-instruct-LLaMAfied"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "vonjack/Phi-3-mini-4k-instruct-LLaMAfied",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/vonjack/Phi-3-mini-4k-instruct-LLaMAfied

SGLang

How to use vonjack/Phi-3-mini-4k-instruct-LLaMAfied 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 "vonjack/Phi-3-mini-4k-instruct-LLaMAfied" \
    --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": "vonjack/Phi-3-mini-4k-instruct-LLaMAfied",
		"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 "vonjack/Phi-3-mini-4k-instruct-LLaMAfied" \
        --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": "vonjack/Phi-3-mini-4k-instruct-LLaMAfied",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use vonjack/Phi-3-mini-4k-instruct-LLaMAfied with Docker Model Runner:
```
docker model run hf.co/vonjack/Phi-3-mini-4k-instruct-LLaMAfied
```

Pipeline explanation

by Mikee283 - opened Apr 24, 2024

Discussion

Mikee283

Apr 24, 2024

Hi, first of all thank you for your work on llamafied model, It saved a me a lot of time and effort and it works perfectly out of the box.

I wish to replicate your results so if you could spare some time to explain the process that would be awsome.

Btw, do you also plan to do the same for 128k model?

Once again thank you for the time you put into this.

vonjack

Owner Apr 25, 2024

Here's the convert code: https://huggingface.co/vonjack/phi-3-mini-4k-instruct-llamafied/blob/main/convert.py
I don't think we can convert 128k model because it uses longrope which not existed in the original llama.

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