Instructions to use FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw with Transformers:

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

pipe = pipeline("text-generation", model="FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw")
model = AutoModelForCausalLM.from_pretrained("FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw

SGLang

How to use FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw 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 "FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw" \
    --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": "FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw",
		"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 "FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw" \
        --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": "FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw with Docker Model Runner:
```
docker model run hf.co/FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw
```

Configuration Parsing Warning:In config.json: "quantization_config.bits" must be an integer

Twilight Miqu is a Story writing model and is composed from sophosympatheia/Midnight-Miqu-70B-v1.5, jukofyork/Dawn-Miqu-70B and jukofyork/Dark-Miqu-70B

It is an experiment to see if large models are more coherent on story writing tasks. Twilight = Midnight + Dawn + Dark

Please see this model card for details and usage instructions. https://huggingface.co/sophosympatheia/Midnight-Miqu-70B-v1.5

This model is based on Miqu so it's capable of 32K context.

All miqu-derived models, including this merge, are only suitable for personal use. Mistral has been cool about it so far, but you should be aware that by downloading this merge you are assuming whatever legal risk is inherent in acquiring and using a model based on leaked weights. This merge comes with no warranties or guarantees of any kind, but you probably already knew that.

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

A big thank you to Mistral, sophosympatheia and jukofyork for the original models!

Downloads last month: 9

Model tree for FluffyKaeloky/Twilight-Miqu-146B-exl2-3.25bpw

jukofyork/Dark-Miqu-70B

sophosympatheia/Midnight-Miqu-70B-v1.5

Merge model

this model