David Seunghyun Yoon

Research Scientist @AdobeResearch #NLProc, #ML

[David Seunghyun Yoon @david_s_yoon]

Excited to share that StreamGaze has been accepted to #CVPR2026!

[Daeun Lee @danadaeun]

🥳 Happy to announce that StreamGaze is accepted to #CVPR2026!

👀 We introduce the first benchmark that evaluates gaze-guided temporal reasoning (past, present, and future) and proactive understanding for streaming video understanding. We find that all MLLMs fall far below human performance, particularly in temporal continuity, gaze grounding, and proactive prediction.

💗 Huge thanks to my last year's AdobeResearch team: Subhojyoti Mukherjee, Branislav Kveton, Ryan A. Rossi, Viet Lai, David Seunghyun Yoon, Trung Bui, Franck Dernoncourt, and my advisor Mohit Bansal 😃

[David Seunghyun Yoon @david_s_yoon]

Key contributions to the captioning model SlimVLM-3B, which takes a kid’s sketch and generates text to send to Firefly for image and 3D Generation. Thanks to amazing collaborator Viet and leader Trung.

[David Seunghyun Yoon @david_s_yoon]

🌊 Thrilled to see the launch of Project Aqua! It’s incredibly rewarding to see a product I contributed to make its way into the world. 🚀

https://t.co/llCNtkDv1t

[Keivan Rezaei @RezaeiKeivan]

Join us today 4:30 at Exhibit Hall C,D,E #4311 to check out our poster "Localizing Knowledge in Diffusion Transformers".

[David Seunghyun Yoon @david_s_yoon]

🚀 Another presentation is happening today! Come stop by our poster session 😊

Program Synthesis via Test-Time Transduction
📄 Paper: https://t.co/T8HIViPSIn

📅 Thu, Dec 4, 2025
⏰ 4:30 PM – 7:30 PM PST
📍 Exhibit Hall C, D, E — #2005

#NeurIPS2025

[David Seunghyun Yoon @david_s_yoon]

Come to our poster session for a discussion! 😊

Offline RL by Reward-Weighted Fine-Tuning for Conversation Optimization
📄 Paper:: https://t.co/OSfCAKr13x

📅 Thu, Dec 4, 2025
⏰ 11:00 AM – 2:00 PM PST
📍 Exhibit Hall C, D, E — #210

#NeurIPS2025

[David Seunghyun Yoon @david_s_yoon]

Come to our session!

[Adobe Research @AdobeResearch]

Adobe researchers are sharing groundbreaking work this week at the premiere AI technology conference NeurIPS2025! We are proud to be presenting 24 papers that connect AI to the future of creativity, design, and data intelligence. We explore generative model innovation, efficient learning, personalized experiences, and multimodal understanding--work that could unlock new creative possibilities for millions of people.

Dive into the details!
https://t.co/1QNYkjWl4v

#NeurIPS2025 #AI

[David Seunghyun Yoon @david_s_yoon]

Adobe Research @ #NeurIPS2025!

Lots of exciting work will be presented this week. If you're interested in a deep discussion, potential collaborations, or internship opportunities, feel free to reach out — I’m around as well 😀
papers: https://t.co/ugflkpj99O

[David Seunghyun Yoon @david_s_yoon]

Excited to share our work on understanding streaming video. Check our paper and dataset!

[Daeun Lee @danadaeun]

🤔 We rely on gaze to guide our actions, but can current MLLMs truly understand it and infer our intentions?

Introducing StreamGaze 👀, the first benchmark that evaluates gaze-guided temporal reasoning (past, present, and future) and proactive understanding in streaming video settings.

➡️ Gaze-Guided Streaming Benchmark: 10 tasks spanning past, present, and proactive reasoning, from gaze-sequence matching to alerting when objects appear within the FOV area.

➡️ Gaze-Guided Streaming Data Construction Pipeline: We align egocentric videos with raw gaze trajectories using fixation extraction, region-specific visual prompting, and scanpath construction to generate spatio-temporally grounded QA pairs. This process is human-verified.

➡️ Comprehensive Evaluation of State-of-the-Art MLLMs: Across all gaze-conditioned streaming tasks, we highlight fundamental limits of current MLLMs. All MLLMs fall far below human performance. Models particularly struggle with temporal continuity, gaze grounding, and proactive prediction.

[SambaNova @SambaNovaAI]

🚀 World record performance: SambaNova is running Llama 3.1 405B at 114 t/s with full precision accuracy, in only one rack. Verified by @ArtificialAnlys! 🦙

This speed unlocks so many use cases for enterprises and developers that we cannot wait to see them built on our platform.

Apply for early access today: https://t.co/CSlbJbTFVj

[Adobe Research @AdobeResearch]

Check out Adobe Research's new work in Computer Vision at #CVPR2024! https://t.co/hXaFfCbhmX

[Anh Totti Nguyen @anh_ng8]

If you are at #NAACL2024 today, come by our poster 4️⃣9️⃣ to check out our explainable, **editable**, part-based image classifier!

Users can intervene the classification process or even modify the classifier by simply editing text descriptors.

[David Seunghyun Yoon @david_s_yoon]

Poster presentation of our exciting work :) Great job! @ArchikiPrasad

[Archiki Prasad @ArchikiPrasad]

Thanks for the shoutout + covering our #ACL2023nlp work on MeetingQA, @JayAlammar @cohere @forai_ml! It was great interacting with you 😃

PS: For those interested, details at 👉 https://t.co/t706r2ZxLL

cc/ Trung, @david_s_yoon, Hanieh, @FranckDernoncou and @mohitban47

[Archiki Prasad @ArchikiPrasad]

New #ACL2023nlp paper: 𝗠𝗲𝗲𝘁𝗶𝗻𝗴𝗤𝗔!

We explore how good LMs are at answering questions in complex conversational meeting settings (with rhetorical + discussion seeking questions & multi-span + multi-speaker answers)

https://t.co/CjrG3CYTpT
@AdobeResearch @uncnlp
🧵👇

[Adobe Research @AdobeResearch]

Adobe researchers are presenting new #ComputerVision work at this week’s #CVPR2023. In addition to the publications, @Adobe authors have also contributed to the conference in many different ways. Check out the blog post to learn more! https://t.co/CarCfJr2QI

[Geoffrey Hinton @geoffreyhinton]

In the NYT today, Cade Metz implies that I left Google so that I could criticize Google. Actually, I left so that I could talk about the dangers of AI without considering how this impacts Google. Google has acted very responsibly.

[Sundar Pichai @sundarpichai]

WebGPU in Chromium 113, excited for the web and great to see this ship to stable!

[Adobe Research @AdobeResearch]

Adobe researchers are presenting new work at #EMNLP2022, one of the top research conferences on #NaturalLanguageProcessing. Check out the blog post to learn more! https://t.co/ZxQebCGtnm

[Adobe Research @AdobeResearch]

Check out the list of @Adobe co-authored papers at this week's #COLING2022, one of the top research conferences on #NaturalLanguageProcessing. https://t.co/vjXqzEThZq

[Adobe Research @AdobeResearch]

Check out the full list of @Adobe co-authored papers and other contributions at this year's @Siggraph, the premier academic conference in computer graphics and interactive techniques! https://t.co/prsqzffyPt