Alibaba Cloud has launched two open-source large vision language models (LVLM), Qwen-VL and Qwen-VL-Chat, capable of comprehending images and texts, answering questions, and more in both English and Chinese.
Facts
- Qwen-VL is a multimodal LVLM that can handle image inputs and text prompts in English and Chinese, performing tasks like answering questions related to images and generating image captions.
- Qwen-VL-Chat is designed for complex interactions, including comparing multiple images and engaging in multi-round question answering, with creative capabilities like writing poetry and summarizing image content.
- Alibaba Cloud has shared the model’s code, weights, and documentation with the open-source community via ModelScope and Hugging Face. Companies with over 100 million monthly users can request a license for commercial use.
- These models could potentially assist visually impaired individuals during online shopping by providing information based on image comprehension.
- Qwen-VL outperforms other large vision language models in various visual language tasks, including captioning, question answering, and object detection.
- Qwen-VL-Chat achieves leading results in both Chinese and English text-image dialogue and alignment tests, according to Alibaba Cloud’s benchmark.
- Earlier, Alibaba Cloud open-sourced its 7-billion-parameter LLMs, Qwen-7B and Qwen-7B-Chat, contributing to the open-source community with over 400,000 downloads within a month of launch.
