Instructions to use LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2 with Transformers:

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

pipe = pipeline("text-generation", model="LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2")
model = AutoModelForCausalLM.from_pretrained("LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2

SGLang

How to use LoneStriker/DaringLotus-10.7B-8.0bpw-h8-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/DaringLotus-10.7B-8.0bpw-h8-exl2" \
    --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": "LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2",
		"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 "LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2" \
        --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": "LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2 with Docker Model Runner:
```
docker model run hf.co/LoneStriker/DaringLotus-10.7B-8.0bpw-h8-exl2
```

I managed to do a heavy density DARE TIES merge of SnowLotus and it's parent models (unusual strategy I know) that seems okay (prose not too bad, not incoherent). Early impressions are that this has slightly different prose - maybe a touch more GPT in there, as it talks of connections, but not at all to the degree that many more synthetically based models do. You probably will find that unobtrusive. Like it's sister model it can and does take lore, character cards and in context chat at times and creates with it, and is very descriptive. I cannot tell which is more coherent - occasionally they both get confused (as is typical with smaller models particularly onces with better prose). I did notice that when in particular contexts, SnowLotus' tendancy for exagerated excalation seemed stronger with this model. So there are differences (some prose and tone differences at least), and testin will probably tell which you prefer.

They share more in common that they do differences - descriptive, fairly creative, occassionally confused but also sometimes surprisingly bright. And the prose has lots of similarities too, it's not generally your 'light, lyrical and poetic' affair.

Summary at least so far, is this one is slightly more gptish in prose and more inclined to escalate scenarios and descriptions in a sort of enthusiastic manner. Both do feed a lot off context, so if you give them stuff they should not be mild or timid.

Downloads last month: 17