Instructions to use jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128 with Transformers:

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

pipe = pipeline("image-text-to-text", model="jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128", trust_remote_code=True)
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, AutoModelForCausalLM

processor = AutoProcessor.from_pretrained("jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128", trust_remote_code=True, 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 jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128",
		"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/jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128

SGLang

How to use jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128 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 "jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128" \
    --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": "jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128",
		"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 "jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128" \
        --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": "jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128",
		"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 jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128 with Docker Model Runner:
```
docker model run hf.co/jkim96/Qwen3.5-35B-A3B-DASHQ-INT3-g128
```

jkim96 commited on 6 days ago

Commit

0cb768d

verified ·

1 Parent(s): 49c9c8e

Add DASH-Q remote-code inference (Triton decode kernel)

Browse files

Files changed (1) hide show

README.md +11 -17

README.md CHANGED Viewed

@@ -16,8 +16,6 @@ tags:
 > **DASH-Q** — Diagonal-Aware Shrinkage for Robust PTQ.
 > `INT3` · group size 128 · **17.4800 GB** (from 71.9039 GB — **4.1x smaller**)
-This checkpoint runs directly with Transformers: the packed quantized layers and a Triton decode kernel are bundled in the repository, so no additional package is needed.
 ## Usage
 ```python
@@ -34,23 +32,19 @@ inputs = tokenizer(text, return_tensors="pt").to(model.device)
 print(tokenizer.decode(model.generate(**inputs, max_new_tokens=256)[0]))
 ```
-No extra package is required: this repository carries its own inference code
-(`modeling_dashq.py`) and a Triton weight-only GEMV decode kernel
-(`dashq_kernel.py`). Quantized layers are rebuilt from `dashq_config.json` at load
-time and converted to the kernel automatically; on CPU or without Triton the model
-falls back to a PyTorch dequantize-and-matmul path.
-Requirements: `transformers`, `torch`, and `triton` (bundled with CUDA builds of
-PyTorch). The [DASH-Q repository](https://github.com/JaeminK/dashq) is only needed
-to quantize your own models.
-### Runtime format
-| Field | Value |
-| --- | --- |
-| Weights | group-wise asymmetric integers (3-bit, group size 128) packed into int32 words |
-| Decode kernel | Triton K-major GEMV (supports 2/3/4/8-bit at any group size) |
-| Prefill | unpack-and-matmul on the same buffers |
 ## Quantization

 > **DASH-Q** — Diagonal-Aware Shrinkage for Robust PTQ.
 > `INT3` · group size 128 · **17.4800 GB** (from 71.9039 GB — **4.1x smaller**)
 ## Usage
 ```python
 print(tokenizer.decode(model.generate(**inputs, max_new_tokens=256)[0]))
 ```
+`trust_remote_code=True` is required: the checkpoint ships the quantized-layer
+implementation (`modeling_dashq.py`) and a Triton weight-only GEMV decode kernel
+(`dashq_kernel.py`). Without Triton or on CPU the model falls back to a PyTorch
+dequantize-and-matmul path.
+### Requirements
+| Package | Minimum | Verified with |
+| --- | --- | --- |
+| `torch` | 2.4 | 2.12.1+cu130 |
+| `transformers` | 5.8 | 5.9.0 |
+| `triton` | 3.0 (Linux; bundled with CUDA builds of PyTorch) | 3.7.1 |
+| `huggingface_hub` | 1.5 (pulled in by transformers) | 1.15.0 |
 ## Quantization