Loading the SOTA2 catalog…
Language Modeling on 1.3B 26B-token pre-training corpus (val) benchmark leaderboard · SOTA2 Research