3 papers
cs.LG2026
The risk of KV cache compression
Lukas Haverbeck, Carmen Amo Alonso, Andres Felipe Posada-Moreno +2
Transformer inference on long sequences is expensive because softmax attention repeatedly reads from a large KV cache. The prevalent approach to this bottleneck is KV cache compres…
cs.LG2025
Kernel conditional tests from learning-theoretic bounds
Pierre-François Massiani, Christian Fiedler, Lukas Haverbeck +2
We propose a framework for hypothesis testing on conditional probability distributions, which we then use to construct statistical tests of functionals of conditional distributions…
cs.LG2025
Viability of Future Actions: Robust Safety in Reinforcement Learning via Entropy Regularization
Pierre-François Massiani, Alexander von Rohr, Lukas Haverbeck +1
Despite the many recent advances in reinforcement learning (RL), the question of learning policies that robustly satisfy state constraints under unknown disturbances remains open.…