Reaction
Mark Tenenholtz calls a small-class model’s 13.5% SWE-sweep result “insane”
Ofir Press praises Gemini’s SWE-sweep results, as Mark Tenenholtz highlights 13.5% for a small-class model.
TLDR
Mark Tenenholtz says a small-class model scored 13.5% on SWE-sweep, calling it an “insane result” and congratulating Google DeepMind. Separately, Ofir Press congratulated the Gemini team on its “strong” SWE-sweep results.
Combined views
—
2 Sources, first seen ago
Comments