Instructions to use LoneStriker/miqu-1-103b-2.4bpw-h6-exl2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LoneStriker/miqu-1-103b-2.4bpw-h6-exl2 with Transformers:

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

pipe = pipeline("text-generation", model="LoneStriker/miqu-1-103b-2.4bpw-h6-exl2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("LoneStriker/miqu-1-103b-2.4bpw-h6-exl2")
model = AutoModelForCausalLM.from_pretrained("LoneStriker/miqu-1-103b-2.4bpw-h6-exl2")
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 LoneStriker/miqu-1-103b-2.4bpw-h6-exl2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LoneStriker/miqu-1-103b-2.4bpw-h6-exl2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LoneStriker/miqu-1-103b-2.4bpw-h6-exl2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/LoneStriker/miqu-1-103b-2.4bpw-h6-exl2

SGLang

How to use LoneStriker/miqu-1-103b-2.4bpw-h6-exl2 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 "LoneStriker/miqu-1-103b-2.4bpw-h6-exl2" \
    --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": "LoneStriker/miqu-1-103b-2.4bpw-h6-exl2",
		"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 "LoneStriker/miqu-1-103b-2.4bpw-h6-exl2" \
        --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": "LoneStriker/miqu-1-103b-2.4bpw-h6-exl2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use LoneStriker/miqu-1-103b-2.4bpw-h6-exl2 with Docker Model Runner:
```
docker model run hf.co/LoneStriker/miqu-1-103b-2.4bpw-h6-exl2
```

Any way to speed up generation on a Windows 11 PC, using a single 24GB card (4090), with Text-Generation-WebUI

by clevnumb - opened Mar 10, 2024

Discussion

clevnumb

Mar 10, 2024

•

edited Mar 10, 2024

Primary goal, SPEED, secondary, MORE CONTEXT

Right now I'm @ 4096 context , 8-bit cache, alpha and compress are both set to 1 (should I raise those??)

...and getting a very slow less than a half token/sec (0.24 & 0.35 with last two tests....unusable really)

Is anything available to speed up this generation even if I must stay at this context? Any other way? Thank you.

LoneStriker

Owner Mar 13, 2024

This model won't fit into 24 GB VRAM, so you are swapping to system RAM. There's no way to speed things up other than running a smaller model. Try a 2.4 bpw 70B model or a Mixtral 8x7B model at less than 4.0bpw.

mjzdev

Mar 13, 2024

This comment has been hidden

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