Profiles2021-2025
Seijin Kobayashi
ETH Zürich Google, Paradigms of Intelligence Team
6Papers
0Last author
7Unique co-authors
1First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers6
- Attention as a HypernetworkAuthor 2 of 5ICLR 2025
- Discovering modular solutions that generalize compositionallyAuthor 2 of 11ICLR 2024 · 24 citations
- Would I have gotten that reward? Long-term credit assignment by counterfactual contribution analysisAuthor 3 of 5NeurIPS 2023 · 16 citations
- The least-control principle for local learning at equilibriumAuthor 3 of 5NeurIPS 2022 · 32 citations
- Disentangling the Predictive Variance of Deep Ensembles through the Neural Tangent KernelAuthor 1 of 4NeurIPS 2022 · 4 citations
- Neural networks with late-phase weightsAuthor 2 of 6ICLR 2021 · 38 citations
