Instructions to use cyboghostginx/Llama3.1-8B-mlx-4Bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cyboghostginx/Llama3.1-8B-mlx-4Bit with Transformers:

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

pipe = pipeline("text-generation", model="cyboghostginx/Llama3.1-8B-mlx-4Bit")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cyboghostginx/Llama3.1-8B-mlx-4Bit")
model = AutoModelForCausalLM.from_pretrained("cyboghostginx/Llama3.1-8B-mlx-4Bit", device_map="auto")

MLX

How to use cyboghostginx/Llama3.1-8B-mlx-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("cyboghostginx/Llama3.1-8B-mlx-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 cyboghostginx/Llama3.1-8B-mlx-4Bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cyboghostginx/Llama3.1-8B-mlx-4Bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cyboghostginx/Llama3.1-8B-mlx-4Bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cyboghostginx/Llama3.1-8B-mlx-4Bit

SGLang

How to use cyboghostginx/Llama3.1-8B-mlx-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 "cyboghostginx/Llama3.1-8B-mlx-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": "cyboghostginx/Llama3.1-8B-mlx-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 "cyboghostginx/Llama3.1-8B-mlx-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": "cyboghostginx/Llama3.1-8B-mlx-4Bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

MLX LM

How to use cyboghostginx/Llama3.1-8B-mlx-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 "cyboghostginx/Llama3.1-8B-mlx-4Bit" --prompt "Once upon a time"

Docker Model Runner
How to use cyboghostginx/Llama3.1-8B-mlx-4Bit with Docker Model Runner:
```
docker model run hf.co/cyboghostginx/Llama3.1-8B-mlx-4Bit
```
Atomic Chat

Llama3.1-8B-mlx-4Bit

Built with Llama. 4-bit MLX conversion of Meta's Llama 3.1 8B base model for Apple silicon. 4.5 GB, single shard, converted with mlx-lm 0.26.4 from cyboghostginx/Llama3.1-8B. Weights only, no fine-tuning.

This is the base model, not Instruct. It ships no chat template, so it completes text rather than answering turns. For chat, quantize an Instruct checkpoint instead.

pip install mlx-lm
mlx_lm.generate --model cyboghostginx/Llama3.1-8B-mlx-4Bit \
                --prompt "The three laws of robotics are" --max-tokens 256

from mlx_lm import load, generate

model, tokenizer = load("cyboghostginx/Llama3.1-8B-mlx-4Bit")
print(generate(model, tokenizer, prompt="The three laws of robotics are", verbose=True))

Higher precision: 8-bit MLX, 8.5 GB.

License

Downloads last month: 9

Safetensors

Model size

1B params

Tensor type

BF16

U32

MLX

Hardware compatibility

4-bit

Model tree for cyboghostginx/Llama3.1-8B-mlx-4Bit

Base model

meta-llama/Llama-3.1-8B

Finetuned

cyboghostginx/Llama3.1-8B

Quantized

(2)

this model