Instructions to use SatouLilly/Lumiere-120b-3.0bpw-h6-exl2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SatouLilly/Lumiere-120b-3.0bpw-h6-exl2 with Transformers:

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

pipe = pipeline("text-generation", model="SatouLilly/Lumiere-120b-3.0bpw-h6-exl2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SatouLilly/Lumiere-120b-3.0bpw-h6-exl2")
model = AutoModelForCausalLM.from_pretrained("SatouLilly/Lumiere-120b-3.0bpw-h6-exl2", 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 SatouLilly/Lumiere-120b-3.0bpw-h6-exl2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SatouLilly/Lumiere-120b-3.0bpw-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": "SatouLilly/Lumiere-120b-3.0bpw-h6-exl2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SatouLilly/Lumiere-120b-3.0bpw-h6-exl2

SGLang

How to use SatouLilly/Lumiere-120b-3.0bpw-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 "SatouLilly/Lumiere-120b-3.0bpw-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": "SatouLilly/Lumiere-120b-3.0bpw-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 "SatouLilly/Lumiere-120b-3.0bpw-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": "SatouLilly/Lumiere-120b-3.0bpw-h6-exl2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SatouLilly/Lumiere-120b-3.0bpw-h6-exl2 with Docker Model Runner:
```
docker model run hf.co/SatouLilly/Lumiere-120b-3.0bpw-h6-exl2
```

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

EXL2 quants of bcse/Lumiere-120b

base_model:

Undi95/Miqu-70B-Alpaca-DPO
Sao10K/Euryale-1.3-L2-70B library_name: transformers tags:
mergekit
merge

Lumiere-120b

This is a merge of pre-trained language models created using mergekit.

Quants

Merge Details

Merge Method

This model was merged using the linear merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

merge_method: linear
parameters:
  weight: 1.0
slices:
  - sources:
      - model: Undi95/Miqu-70B-Alpaca-DPO
        layer_range: [0, 1]
      - model: Sao10K/Euryale-1.3-L2-70B
        layer_range: [0, 1]
        parameters:
          weight: 0
  - sources:
      - model: Undi95/Miqu-70B-Alpaca-DPO
        layer_range: [1, 20]
  - sources:
      - model: Sao10K/Euryale-1.3-L2-70B
        layer_range: [10, 30]
  - sources:
      - model: Undi95/Miqu-70B-Alpaca-DPO
        layer_range: [20, 40]
  - sources:
      - model: Sao10K/Euryale-1.3-L2-70B
        layer_range: [30, 50]
  - sources:
      - model: Undi95/Miqu-70B-Alpaca-DPO
        layer_range: [40, 60]
  - sources:
      - model: Sao10K/Euryale-1.3-L2-70B
        layer_range: [50, 70]
  - sources:
      - model: Undi95/Miqu-70B-Alpaca-DPO
        layer_range: [60, 79]
  - sources:
      - model: Undi95/Miqu-70B-Alpaca-DPO
        layer_range: [79, 80]
      - model: Sao10K/Euryale-1.3-L2-70B
        layer_range: [79, 80]
        parameters:
          weight: 0
dtype: float16
tokenizer_source: model:Undi95/Miqu-70B-Alpaca-DPO

Downloads last month: 19

Paper for SatouLilly/Lumiere-120b-3.0bpw-h6-exl2

Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time

Paper • 2203.05482 • Published Mar 10, 2022 • 9