Report
Language-model learning paper wins COLM Outstanding Paper; diagnostic-error abstract takes DEX26 People’s Choice
The poster describes ordered skill emergence and says top models caught many physician diagnostic errors, though big gaps remain.
TLDR
A post announces that “What do Language Models Learn and When?” received COLM’s Outstanding Paper award and says language-model skills emerge in a consistent, compositional order. In a follow-up, the poster says Ahmed Hassoon’s benchmark of AI for diagnostic error correction won DEX26’s People’s Choice Best Abstract. The poster says top large language models caught many physician diagnostic errors, but big gaps remain.
Combined views
26
1 Source, first seen ago
