Instructions to use Minachist/Qwen3.6-27B-INT8-AutoRound with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Minachist/Qwen3.6-27B-INT8-AutoRound with Transformers:

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

pipe = pipeline("image-text-to-text", model="Minachist/Qwen3.6-27B-INT8-AutoRound")
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, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("Minachist/Qwen3.6-27B-INT8-AutoRound")
model = AutoModelForImageTextToText.from_pretrained("Minachist/Qwen3.6-27B-INT8-AutoRound")
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

vLLM

How to use Minachist/Qwen3.6-27B-INT8-AutoRound with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Minachist/Qwen3.6-27B-INT8-AutoRound"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Minachist/Qwen3.6-27B-INT8-AutoRound",
		"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/Minachist/Qwen3.6-27B-INT8-AutoRound

SGLang

How to use Minachist/Qwen3.6-27B-INT8-AutoRound 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 "Minachist/Qwen3.6-27B-INT8-AutoRound" \
    --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": "Minachist/Qwen3.6-27B-INT8-AutoRound",
		"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 "Minachist/Qwen3.6-27B-INT8-AutoRound" \
        --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": "Minachist/Qwen3.6-27B-INT8-AutoRound",
		"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 Minachist/Qwen3.6-27B-INT8-AutoRound with Docker Model Runner:
```
docker model run hf.co/Minachist/Qwen3.6-27B-INT8-AutoRound
```

Me again

by qenme - opened 3 days ago

Discussion

qenme

3 days ago

•

edited 3 days ago

Hey, me again just happened to be looking at your model card and I noticed this "With these settings, you get around 129k context. You can also add --kv-cache-dtype fp8_e4m3 --calculate-kv-scales args to get about 252k tokens." I wasn't aware of the calculate kv scales so I did some research and it's been deprecated. It silently corrupts on Qwen3.5 (https://github.com/vllm-project/vllm/pull/37565).

Also, as a follow up to my previous discussion, I ended up being able to recreate this quant in llama-cpp and it works very well.

/home/user/llm/mtp/llama.cpp/build/bin/llama-quantize
--tensor-type token_embd=bf16
--tensor-type output=bf16
--tensor-type output_norm=bf16
--tensor-type post_attention_norm=bf16
--tensor-type attn_q_norm=bf16
--tensor-type attn_k_norm=bf16
--tensor-type attn_qkv=bf16
--tensor-type attn_gate=bf16
--tensor-type ssm_a=bf16
--tensor-type ssm_alpha=bf16
--tensor-type ssm_beta=bf16
--tensor-type ssm_conv1d=bf16
--tensor-type ssm_dt.bias=bf16
--tensor-type ssm_norm=bf16
--tensor-type ssm_out=bf16
/home/user/llm/models/Qwen3.6-27B/BF16/Qwen3.6-27B-BF16-MTP.gguf
/home/user/llm/models/Qwen3.6-27B/BF16/Qwen3.6-27B-Q8-BIGBOY.gguf
q8_0

Minachist

Owner 3 days ago

Updated the readme, thank you!

Minachist

Owner 3 days ago

Also, you might want to consider uploading it to huggingface yourself. I am sure others would find it useful. I don't use llama.cpp anymore since it lacked MTP and tensor parallelism support, which doesn't suit my use case, but it's still widely used by many people.

qenme

2 days ago

I may upload it. 30GB will take an overnight session, lol. Llama CPP now supports tensor and MTP. Although the new MTP feature degrades PP a bit. Overall, they are improving well. Thanks again for introducing me to this quant. I'll keep a look out for future quants from you.

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