Instructions to use BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw with Transformers:

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

pipe = pipeline("image-text-to-text", model="BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw")
model = AutoModelForMultimodalLM.from_pretrained("BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw

SGLang

How to use BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw 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 "BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw" \
    --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": "BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw" \
        --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": "BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw with Docker Model Runner:
```
docker model run hf.co/BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw
```

Qwen3.5-0.8B-EXL3-3.0bpw

Qwen3.5-0.8B quantized to 3.0 bits-per-weight with the EXL3 (bitshift trellis) format, using the 3inst codebook.

Quantized with exllamav3.

Model Description

This is an EXL3-format quantization of Qwen/Qwen3.5-0.8B, the lightweight vision-language variant of the Qwen3.5 family. See the base model card for architecture details, intended use, and limitations.

Property	Value
Base model	Qwen/Qwen3.5-0.8B
Quantization format	EXL3 (QTIP-style bitshift trellis)
Bits per weight	3.0 (head: 6.0)
Codebook	3inst
Parameters	~0.8B
License	Apache 2.0

Quantization Details

Format: EXL3 — bitshift trellis coding with procedural codebook decode.
Calibration: 256 rows x 2048 columns from a general-domain text corpus mix (C4 / Wikipedia / code / technical).
Applied output-channel scales: always.

Evaluation

Metric	fp16	3.0bpw EXL3	Delta
Perplexity (Wiki)	11.21	12.24	+9.2%
KL div (fp16 \|\| 3bit)	—	0.101	—
Top-1 accuracy	0.525	0.512	−2.5pp

Usage

This repository is intended for use with EXL3-capable inference runtimes (ExLlamaV3). It is not a standalone Transformers-format model.

# ExLlamaV3
python chat.py -m BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw

Credits

Base model: Qwen/Qwen3.5-0.8B by Alibaba Group (licensed under Apache 2.0).
Quantizer: ExLlamaV3 by Turboderp.
ROCm/HIP port used for quantization: CarouselAether/rocm_exl3.

This is a community quantization; the base model's capabilities and limitations apply unchanged.

Downloads last month: 18

Safetensors

Model size

0.4B params

Tensor type

F32

BF16

F16

I16

Model tree for BlivionIaG/Qwen3.5-0.8B-exl3-3.0bpw

Base model

Qwen/Qwen3.5-0.8B-Base

Finetuned

Qwen/Qwen3.5-0.8B

Quantized

(234)

this model