←
Hakki C. Karaimer30 posts
Hakki C. Karaimer avatar

Hakki C. Karaimer

@karaimer

Staff Research Scientist at Samsung Research America (SRA) — Previously Qualcomm & AMD · postdoc EPFL · PhD York University · TEV-SINGA PhD scholar NUS.

◷ Joined March 2011
Babak Taati avatar
⟳ Hakki C. Karaimer retweeted
Babak Taati@Babak_Taati·Oct 30

Kalman filter for the win 💪

📢🚀Happy to share our recent work. SpecKCast: Spectral Kalman Forecasting for Diffusion Acceleration 🌐Webpage: rajabi2001.github.io/speckca… 📄Paper: arxiv.org/abs/2610.26837 A huge thanks to @karaimer, @DaveLindell, @Babak_Taati, and Michael S. Brown. @Samsung_RA @VectorInst @UofTCompSci

1:34
Javad Rajabi avatar
⟳ Hakki C. Karaimer retweeted
Javad Rajabi@rajabi2001·Oct 23

✨SpecKCast is a training-free method that speeds up diffusion transformers by predicting the model's output at most denoising steps instead of computing it. A lightweight Kalman filter makes these predictions in the frequency domain and estimates how wrong they might be. The transformer runs only when the forecast becomes too uncertain, achieving high speedup while preserving generation quality. Feel free to check out our paper!

Javad Rajabi avatar
⟳ Hakki C. Karaimer retweeted
Javad Rajabi@rajabi2001·Oct 19

🔥Check out our recent work; it turned into a really fun and interesting paper! Duplication is usually considered a flaw in diffusion models when generating high-resolution images. We looked at it from a different perspective and exploited it for a very interesting use case: photomosaic images. 🧩Make sure to check out our cool demo: huggingface.co/spaces/kooros…

Happy to share that PhotoQuilt got accepted to #WACV2027 🎉 Some bugs are mistakes. Some are ideas in disguise. Training-free photomosaics at any resolution. Zoom in and every tile is its own image. 🔗 kooroshrh.github.io/photo-qu… 🤗 Check out our demo: huggingface.co/spaces/kooros…

1:06
Amirhossein Kazerouni avatar
⟳ Hakki C. Karaimer retweeted
Amirhossein Kazerouni@amirkz99·Sep 24

🎉 Super excited to share that AVIS has been accepted to #NeurIPS2026! 🚀🔥 Huge congratulations to the amazing team! See you at NeurIPS! 🥳 📄 Paper: arxiv.org/abs/2606.11576 💻 Code: github.com/SamsungLabs/avis 🌐 Website: avis-vlm.github.io/

🚀 New paper: AVIS: Adaptive Visual Inference Scaling for Vision-Language Models Most test-time scaling methods for vision-language models optimize a single axis of compute: either scale visual context or perform more reasoning rollouts. AVIS jointly adapts both. It allocates compute across visual context scaling and visual reasoning scaling on a per-query basis, combining efficient visual token pruning with adaptive reasoning search. 📊 Across diverse image and video reasoning benchmarks, AVIS achieves a better accuracy–compute trade-off than methods that scale only visual context or only reasoning, while remaining deployment-friendly through shared-prefill inference and KV-cache reuse. Vision-language inference is not just about seeing more or thinking longer. AVIS learns when to do each. 📄 Paper: arxiv.org/abs/2606.11576 🌐 Website: avis-vlm.github.io Amazing teamwork with @amirkz99 @karaimer @CSProfKGD @Babak_Taati

Armen Jeddi avatar
⟳ Hakki C. Karaimer retweeted
Armen Jeddi@armen_jeddi·Jun 12

🚀 New paper: AVIS: Adaptive Visual Inference Scaling for Vision-Language Models Most test-time scaling methods for vision-language models optimize a single axis of compute: either scale visual context or perform more reasoning rollouts. AVIS jointly adapts both. It allocates compute across visual context scaling and visual reasoning scaling on a per-query basis, combining efficient visual token pruning with adaptive reasoning search. 📊 Across diverse image and video reasoning benchmarks, AVIS achieves a better accuracy–compute trade-off than methods that scale only visual context or only reasoning, while remaining deployment-friendly through shared-prefill inference and KV-cache reuse. Vision-language inference is not just about seeing more or thinking longer. AVIS learns when to do each. 📄 Paper: arxiv.org/abs/2606.11576 🌐 Website: avis-vlm.github.io Amazing teamwork with @amirkz99 @karaimer @CSProfKGD @Babak_Taati

Armen Jeddi avatar
⟳ Hakki C. Karaimer retweeted
Armen Jeddi@armen_jeddi·May 15

🚀 New paper: GEAR: Genetic AutoResearch for Agentic Code Evolution Most autonomous research agents search like greedy hill climbers: edit one program, run one experiment, and keep only the best result. GEAR replaces this with a population-based frontier of research states, using mutation, crossover, and an evolvable search policy to preserve and recombine promising ideas. 📊 Under the same AutoResearch setup and 100-experiment budget, all GEAR variants outperform the baseline. Research is not one path. GEAR gives agents a frontier. 📄 Paper: arxiv.org/abs/2605.13874 🌐 Website: genetic-autoresearch.github.… #autoresearch #llm #agentic_ai #agentic_coding @karaimer @CSProfKGD @Babak_Taati

Post attachmentPost attachment
Armen Jeddi avatar
⟳ Hakki C. Karaimer retweeted
Armen Jeddi@armen_jeddi·Apr 21

I’ll be in Rio 🇧🇷 presenting LoopFormer at #ICLR2026! Stop by our poster in Pavilion 3 this Thursday, 10:30 AM–1 PM — or DM me if you’d like to chat.

Post attachment
Armen Jeddi avatar
⟳ Hakki C. Karaimer retweeted
Armen Jeddi@armen_jeddi·Apr 17

Loops, loops, loops… LoopFormer. Looped Language Models are becoming one of the defining themes of 2026, and next week I’ll be presenting our paper, LoopFormer, at #ICLR2026. Lmk if you want to chat about looped LMs in Rio.

Post attachment
Armen Jeddi avatar
⟳ Hakki C. Karaimer retweeted
Armen Jeddi@armen_jeddi·Feb 13

LoopFormer is the next step in Recursive Language Modeling (RLMs): it treats iterative refinement in looped Transformers as a trajectory in representation space, conditioning each loop on time and step size so the same model is both parameter-efficient and compute-adaptive.

Armen Jeddi avatar
⟳ Hakki C. Karaimer retweeted
Armen Jeddi@armen_jeddi·Feb 13

New Paper: (ICLR 2026): LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation arxiv.org/abs/2602.11451