G20 · GOOG · Market coverage: all 20 trading days, 2–30 Jan 2026.
Loading G20 records…
Mean WS21: lower is better. Evaluation settings differ. Source dates describe run or source records, not evaluation completion. Original CSV · Original sources (JSON) · Mode results (JSON)

Loading results…

G20 · GOOG

Score evolution

Generation 250 Generation 500 Other lengths Frontier · lowest recorded value

Three evaluation modes

Leaderboard

Scroll horizontally to view all columns, including Model Zoo, Weights & Biases links and evaluation mode.

Messages Messages Score data path Checkpoint Checkpoint created (UTC) Model Zoo Leaderboard Private · Hugging Face access required Weights & Biases Evaluation mode