Not-For-All-Audiences

Instructions to use Ionze/Lilitu-L3.3-70b-0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Ionze/Lilitu-L3.3-70b-0.1 with Transformers:

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

pipe = pipeline("text-generation", model="Ionze/Lilitu-L3.3-70b-0.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Ionze/Lilitu-L3.3-70b-0.1")
model = AutoModelForCausalLM.from_pretrained("Ionze/Lilitu-L3.3-70b-0.1", device_map="auto")

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Ionze/Lilitu-L3.3-70b-0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Ionze/Lilitu-L3.3-70b-0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Ionze/Lilitu-L3.3-70b-0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Ionze/Lilitu-L3.3-70b-0.1

SGLang

How to use Ionze/Lilitu-L3.3-70b-0.1 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 "Ionze/Lilitu-L3.3-70b-0.1" \
    --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": "Ionze/Lilitu-L3.3-70b-0.1",
		"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 "Ionze/Lilitu-L3.3-70b-0.1" \
        --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": "Ionze/Lilitu-L3.3-70b-0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Ionze/Lilitu-L3.3-70b-0.1 with Docker Model Runner:
```
docker model run hf.co/Ionze/Lilitu-L3.3-70b-0.1
```

Request for 4.25bpw (Great model!)

by Apple1037 - opened Dec 24, 2025

Discussion

Apple1037

Dec 24, 2025

Hi, thank you for the release! I was using your 4.5 bpw and it seems slightly larger than another model at 4.5 bpw (yours is 41.4 gb while zerofata_L3.3-GeneticLemonade-Unleashed-v3-70B-4.5bpw-hb6-exl3 is 39.3 gb) and the 2 gb difference is unfortunately slightly limiting. Any chance you could do a 4.25 quant? I would do it myself but I don't want to mess with whatever you are doing which is clearly working.

Sorry for the trouble!

Ionze

Owner Dec 27, 2025

Hi, I've generated and uploaded the 4.25bpw version, hope this works for you!
https://huggingface.co/Ionze/Lilitu-L3.3-70b-0.1-exl3-4.25bpw

Apple1037

Dec 28, 2025

Thank you! Yes this fits perfectly at 64K on my 2x 3090 and I'm hardly noticing a difference between the 4.5, should be my general model for some time to come.

Apple1037 changed discussion status to closed Dec 28, 2025

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