Instructions to use Avuja/Void-Citrus-L3.3-70B-4.25-EXL3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Avuja/Void-Citrus-L3.3-70B-4.25-EXL3 with Transformers:

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

pipe = pipeline("text-generation", model="Avuja/Void-Citrus-L3.3-70B-4.25-EXL3")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Avuja/Void-Citrus-L3.3-70B-4.25-EXL3")
model = AutoModelForCausalLM.from_pretrained("Avuja/Void-Citrus-L3.3-70B-4.25-EXL3", 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 Avuja/Void-Citrus-L3.3-70B-4.25-EXL3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Avuja/Void-Citrus-L3.3-70B-4.25-EXL3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Avuja/Void-Citrus-L3.3-70B-4.25-EXL3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Avuja/Void-Citrus-L3.3-70B-4.25-EXL3

SGLang

How to use Avuja/Void-Citrus-L3.3-70B-4.25-EXL3 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 "Avuja/Void-Citrus-L3.3-70B-4.25-EXL3" \
    --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": "Avuja/Void-Citrus-L3.3-70B-4.25-EXL3",
		"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 "Avuja/Void-Citrus-L3.3-70B-4.25-EXL3" \
        --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": "Avuja/Void-Citrus-L3.3-70B-4.25-EXL3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Avuja/Void-Citrus-L3.3-70B-4.25-EXL3 with Docker Model Runner:
```
docker model run hf.co/Avuja/Void-Citrus-L3.3-70B-4.25-EXL3
```

Void-Citrus-L3.3-70B-4.25-EXL3 / README.md

Avuja

Upload folder using huggingface_hub

4c9aeed verified 4 months ago

preview code

Raw

History Blame Contribute Delete

5.62 kB

metadata

base_model:
  - Steelskull/L3.3-MS-Nevoria-70b
  - zerofata/L3.3-GeneticLemonade-Final-v2-70B
  - Steelskull/L3.3-San-Mai-R1-70b
  - CrucibleLab/L3.3-70B-Loki-V2.0
  - TheDrummer/Anubis-70B-v1.1
  - sophosympatheia/Strawberrylemonade-L3-70B-v1.2
tags:
  - merge
  - mergekit
  - lazymergekit
  - Void-Citrus
  - L3.3
  - 70B
  - exl3
  - exllamav3
language:
  - en
library_name: transformers
quantized_by: Avuja

Original model by: Darkknight535 →
Quantized by: Avuja →

Quantization Info

This is an ExLlamaV3 (EXL3) quantization of Void-Citrus-L3.3-70B.

Format: EXL34.25 bpwHead bits: 670B

Requires ExLlamaV3 or a compatible inference engine such as TabbyAPI.

Original Model Information

Void-Citrus is built around Nevoria stability, boosted by Lemonade logic, Loki chaos, and Anubis boldness. StrawberryLemonade adds warmth to keep responses alive and expressive.

70B MergeVoid-Citrus ThemePure Black CoreL3.3 Series

Model Composition

Nevoria Anchor — Core stability + RP foundation
Genetic Lemonade — Fire + sharp reasoning boost
San-Mai R1 — Dialogue refinement + balance
Loki V2 — Chaos creativity + uniqueness
Anubis — Removes quietness, adds bold personality
Strawberry Lemonade — Sweetness + expressive warmth

Credits & Support

Original Model Support SteelSkull Support Zerofata Support TheDrummer Support Sophosympatheia

Avuja
/

Void-Citrus-L3.3-70B-4.25-EXL3

Void-Citrus-L3.3-70B — EXL3 4.25bpw

Quantization Info

Original Model Information

Model Composition

Credits & Support