Tag
This paper studies online adaptation strategies for discrete diffusion models in molecular optimization, identifying complementary components like acquisition, reward shaping, debiasing, replay, and validity control that improve feedback efficiency on small-molecule and protein tasks.
TxBench-PP is a benchmark for evaluating AI agents on small-molecule preclinical pharmacology tasks. Across 16 model-harness configurations, the best system achieved only 59.3% accuracy, indicating significant room for improvement.