Instructions to use mlx-community/Meta-Llama-3.1-8B-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mlx-community/Meta-Llama-3.1-8B-4bit with Transformers:

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

pipe = pipeline("text-generation", model="mlx-community/Meta-Llama-3.1-8B-4bit")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("mlx-community/Meta-Llama-3.1-8B-4bit")
model = AutoModelForCausalLM.from_pretrained("mlx-community/Meta-Llama-3.1-8B-4bit", device_map="auto")

MLX

How to use mlx-community/Meta-Llama-3.1-8B-4bit with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm
# if on a CUDA device, also pip install mlx[cuda]

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Meta-Llama-3.1-8B-4bit")

prompt = "Once upon a time in"
text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

vLLM

How to use mlx-community/Meta-Llama-3.1-8B-4bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mlx-community/Meta-Llama-3.1-8B-4bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/Meta-Llama-3.1-8B-4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/mlx-community/Meta-Llama-3.1-8B-4bit

SGLang

How to use mlx-community/Meta-Llama-3.1-8B-4bit 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 "mlx-community/Meta-Llama-3.1-8B-4bit" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/Meta-Llama-3.1-8B-4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "mlx-community/Meta-Llama-3.1-8B-4bit" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/Meta-Llama-3.1-8B-4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

MLX LM

How to use mlx-community/Meta-Llama-3.1-8B-4bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Generate some text
mlx_lm.generate --model "mlx-community/Meta-Llama-3.1-8B-4bit" --prompt "Once upon a time"

Docker Model Runner
How to use mlx-community/Meta-Llama-3.1-8B-4bit with Docker Model Runner:
```
docker model run hf.co/mlx-community/Meta-Llama-3.1-8B-4bit
```
Atomic Chat

Error when trying to load the model according to instructions

by mattany - opened Jul 28, 2024

Discussion

mattany

Jul 28, 2024

I used the command

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Meta-Llama-3.1-8B-4bit")

and got the error

ValueError Traceback (most recent call last)
in <cell line: 6>()
4 from mlx_lm import load, generate
5
----> 6 model, tokenizer = load("mlx-community/Meta-Llama-3.1-8B-4bit")
7 # response = generate(model, tokenizer, prompt="hello", verbose=True)
8

2 frames
/usr/local/lib/python3.10/dist-packages/mlx/nn/layers/base.py in load_weights(self, file_or_weights, strict)
162 if extras := (new_weights.keys() - curr_weights.keys()):
163 extras = " ".join(extras)
--> 164 raise ValueError(f"Received parameters not in model: {extras}.")
165 if missing := (curr_weights.keys() - new_weights.keys()):
166 missing = " ".join(missing)

ValueError: Received parameters not in model: model.embed_tokens.scales model.embed_tokens.biases.

Running in colab by the way with a T4. Do I have to be running on apple silicon for this model to work?

mattany

Jul 28, 2024

okay it's because the version installed of mlx was outdated. I didn't manage to fix it since pip couldn't get an up to date version of mlx. I guess the model is meant to run on apple silicone only

mattany changed discussion status to closed Jul 28, 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