PhD at Tübingen. Working on post-training diffusion and multimodal models. Previous research interns at Snapchat and Naver Labs. https://sgk98.github.io/https://bsky.app/profile/shyamgopal.bsky.socialhttps://bsky.app/profile/shyamgopal.bsky.social/post/3mnekb3ow7226There's nothing more satisfying than watching the right noise do its magic with diffusion models! A few interesting takeaways I had from this work 🧵 [contains quote post or other embedded content]03 Jun 2026 07:05 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3mnekb3ow7226https://bsky.app/profile/shyamgopal.bsky.social/post/3m6gu2ad24s22Was a very fun (and quick) investigation into biases of multimodal benchmarks, this time on tasks designed for "Spatial Supersensing" introduced by Cambrian-S with some great folks! [contains quote post or other embedded content]25 Nov 2025 07:48 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3m6gu2ad24s22https://bsky.app/profile/shyamgopal.bsky.social/post/3lsl7ol3vg22xWonderful story behind some very nice SSL work! [contains quote post or other embedded content]27 Jun 2025 08:24 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3lsl7ol3vg22xhttps://bsky.app/profile/shyamgopal.bsky.social/post/3lrcgd2472c2tI'm in Nashville this week attending #CVPR2025. Excited to discuss post-training VLMs and diffusion models!11 Jun 2025 03:04 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3lrcgd2472c2thttps://bsky.app/profile/shyamgopal.bsky.social/post/3ljhrjqpgd22gThese are some ridiculously good results from training tiny T2I models purely on ImageNet! It's almost too good to be true. Do check it out! [contains quote post or other embedded content]03 Mar 2025 10:46 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3ljhrjqpgd22ghttps://bsky.app/profile/shyamgopal.bsky.social/post/3lcvo6dxxqs22After a break of over 2 years, I'm attending a conference again! Excited to attend NeurIPS, even more so to be presenting ReNO, getting inference-time scaling and preference optimization to work for text-to-image generation. Do reach out if you'd like to chat!09 Dec 2024 21:27 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3lcvo6dxxqs22https://bsky.app/profile/shyamgopal.bsky.social/post/3lbzkilxgtc2qCheck out this nice work by @confusezius.bsky.social on designing VLMs for few-shot adaptation! [contains quote post or other embedded content]28 Nov 2024 17:07 +0000at://did:plc:n27fjmp4fuiko5cdlfmo7hoe/app.bsky.feed.post/3lbzkilxgtc2q