Arrived in San Diego for #NeurIPS2025! Reach out if you are around too! Happy to chat! We'll present AttentionChains on Thursday afternoon! neurips.cc/virtual/2025...
NeurIPS Poster Attention (as Discrete-Time Markov) ChainsNeurIPS 2025
neurips.cc
Olaf Dünkel
@oduenkel.bsky.social
ELLIS PhD @ MPI & Oxford - Generative Models for Vision https://odunkel.github.io/
Currently travelling to #ICCV2025 and looking forward to presenting DIY-SC and CNS-Bench there! DIY-SC: #538 at poster session 2 CNS-Bench: #1839 at poster session 5 Happy to chat during the poster sessions or at some other time if you are around!
Super excited to introduce ✨ AnyUp: Universal Feature Upsampling 🔎 Upsample any feature - really any feature - with the same upsampler, no need for cumbersome retraining. SOTA feature upsampling results while being feature-agnostic at inference time. 🌐 wimmerth.github.io/anyup/
You read this? You’ll likely read the linked post too. You like it? Your followers might see it too. In other words: Attention here → attention to Yotam’s post. We explore how transformer attention can be propagated—like PageRank, but for attention. Fun work with @yotamerel.bsky.social
Attention (as Discrete-Time Markov) Chains
Attention (as Discrete-Time Markov) Chains
yoterel.github.io
Yotam Erel@yotamerel.bsky.social · last yr.{8/8} This framework offers a new way to probe and reason about attention ! 📄 Paper: “Attention (as Discrete-Time Markov) Chains” 🔗 yoterel.github.io/attention_ch... 👥 Yotam Erel*, @oduenkel.bsky.social*, Rishabh Dabral, Vlad Golyanik, Christian Theobalt, Amit Bermano *denotes equal contribution
Are you using DINOv2 for tasks that require semantic features? DIY-SC might be the alternative! It refines DINOv2 or SD+DINOv2 features and achieves a new SOTA on the semantic correspondence dataset SPair-71k when not relying on annotated keypoints! [1/6] genintel.github.io/DIY-SC