Instructions to use JLKang/ViSpec-llava-1.5-7b-hf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use JLKang/ViSpec-llava-1.5-7b-hf with Transformers:

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

pipe = pipeline("image-text-to-text", model="JLKang/ViSpec-llava-1.5-7b-hf")

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("JLKang/ViSpec-llava-1.5-7b-hf")
model = AutoModelForImageTextToText.from_pretrained("JLKang/ViSpec-llava-1.5-7b-hf")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use JLKang/ViSpec-llava-1.5-7b-hf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "JLKang/ViSpec-llava-1.5-7b-hf"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "JLKang/ViSpec-llava-1.5-7b-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/JLKang/ViSpec-llava-1.5-7b-hf

SGLang

How to use JLKang/ViSpec-llava-1.5-7b-hf 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 "JLKang/ViSpec-llava-1.5-7b-hf" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "JLKang/ViSpec-llava-1.5-7b-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "JLKang/ViSpec-llava-1.5-7b-hf" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "JLKang/ViSpec-llava-1.5-7b-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use JLKang/ViSpec-llava-1.5-7b-hf with Docker Model Runner:
```
docker model run hf.co/JLKang/ViSpec-llava-1.5-7b-hf
```

ViSpec: Accelerating Vision-Language Models with Vision-Aware Speculative Decoding

Jialiang Kang, Han Shu, Wenshuo Li, Yingjie Zhai, Xinghao Chen

Overview

Speculative decoding is a widely adopted technique for accelerating inference in large language models (LLMs), yet its application to vision-language models (VLMs) remains underexplored, with existing methods achieving only modest speedups ($<1.5\times$). This gap is increasingly significant as multimodal capabilities become central to large-scale models. We hypothesize that large VLMs can effectively filter redundant image information layer by layer without compromising textual comprehension, whereas smaller draft models struggle to do so. To address this, we introduce Vision-Aware Speculative Decoding (ViSpec), a novel framework tailored for VLMs. ViSpec employs a lightweight vision adaptor module to compress image tokens into a compact representation, which is seamlessly integrated into the draft model's attention mechanism while preserving original image positional information. Additionally, we extract a global feature vector for each input image and augment all subsequent text tokens with this feature to enhance multimodal coherence. To overcome the scarcity of multimodal datasets with long assistant responses, we curate a specialized training dataset by repurposing existing datasets and generating extended outputs using the target VLM with modified prompts. Our training strategy mitigates the risk of the draft model exploiting direct access to the target model's hidden states, which could otherwise lead to shortcut learning when training solely on target model outputs. Extensive experiments validate ViSpec, achieving, to our knowledge, the first substantial speedup in VLM speculative decoding.

Citation

If you find our work useful, please consider citing:

@inproceedings{vispec,
  title={ViSpec: Accelerating Vision-Language Models with Vision-Aware Speculative Decoding},
  author={Kang, Jialiang and Shu, Han and Li, Wenshuo and Zhai, Yingjie and Chen, Xinghao},
  booktitle={Annual Conference on Neural Information Processing Systems},
  year={2025}
}