About Me

Shot by my wonderful friend Liqi Xu in Shenzhen, Jan 2024.

I am Jinghong Chen (Eric), a PhD student at the University of Cambridge. I write concise posts for a technical audience interested in Natural Language Generation (NLG), Vision-Language Models (VLM), and Diffusion Models. You can find my list of publications below. Reach me at jc2124 (at) cam (dot) ac (dot) uk.

I am author of the research framework Runway for ML. It provides off-the-shelf experiment configuration system and data pipelines capable of large-scale data processing. Runway for ML is used by top vision-language researchers and AIGC start-ups. Find out more here.

I am also the Principal AI Scientist at To0 Space, an AIGC start-up that builds Stable Diffusion models to generate high-quality architecture renderings that follow architects' designs and choice of style. Our service is hosted here.

I started Kendo in 2023. I was the winner (beginner) of the at Varsity 2023 and runner-up (Kyu) at Varisty 2024. I am on the Cambridge Second Team in 2024.

Publications

Schema-Guided Semantic Accuracy: Faithfulness in Task-Oriented Dialogue Response Generation. Jinghong Chen, Weizhe Lin, Bill Byrne. https://arxiv.org/abs/2301.12568. 2022.
Fine-grained Late-interaction Multi-modal Retrieval for Retrieval Augmented Visual Question Answering. Weizhe Lin, Jinghong Chen, Jingbiao Mei, Alexandru Coca, Bill Byrne. The Thirty-seventh Conference on Neural Information Processing Systems (NeurIPS). https://arxiv.org/abs/2309.17133. 2023. 3-minute walkthrough: https://jinghong-chen.ghost.io/fined-grained-late-interaction-multimodal-retrieval-flmr/.
Direct Preference Optimization for Neural Machine Translation with Minimum Bayes Risk Decoding. Guangyu Yang, Jinghong Chen, Weizhe Lin, Bill Byrne. https://arxiv.org/abs/2311.08380. 2023. 3-minute walkthrough: https://jinghong-chen.ghost.io/learning-from-mbr-decoding-using-direct-preference-optimization/
Improving hateful memes detection via learning hatefulness-aware embedding space through retrieval-guided contrastive learning. Jingbiao Mei, Jinghong Chen, Weizhe Lin, Bill Byrne, Marcus Tomalin. https://arxiv.org/abs/2311.08110. 2023. 3-minute-walkthrough: https://www.jinghong-chen.net/3-minute-pitch-retrieval-guided-contrastive-learning-for-hateful-memes-detection/
Grounding Description-Driven Dialogue State Trackers with Knowledge-Seeking Turns. Alexandra Coca, Bo-Hsiang Tseng, Jinghong Chen, Jingbiao Mei, Bill Byrne. The 24th Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL). https://arxiv.org/abs/2301.12568. 2023.