{ "historical_results": true, "new_evaluation_for_release": false, "checkpoint_sha256": "bd82d3c508a7b591b7c1578614231b142689982c842c6864c80baaa213e5dde5", "osworld_selected_suite": { "official_tasks": 369, "selected_tasks": 361, "scored": 356, "successes": 58, "success_rate_scored": 0.16292134831460675, "unscored": 5, "mean_reward_scored": 0.17120773098371703 }, "development_validation": { "step": 3368, "n": 1967, "loss": 5.176659643919462, "raw_action_selector_valid": 0.9415353329944077, "constrained_schema_valid": 0.9994916115912558, "action_accuracy": 0.7559735638027453, "canonical_exact_match": 0.24046771733604475, "coordinate_bin_accuracy": 0.19882246376811594, "action_and_point_within_20_of_1000": 0.41304347826086957, "coordinate_rows": 1104, "single_forward_per_example": true, "by_action": { "double_click": { "n": 199, "exact": 0.005025125628140704, "action_accuracy": 0.5326633165829145 }, "key": { "n": 301, "exact": 0.6744186046511628, "action_accuracy": 0.7740863787375415 }, "left_click": { "n": 584, "exact": 0.0017123287671232876, "action_accuracy": 0.7465753424657534 }, "left_click_drag": { "n": 24, "exact": 0.0, "action_accuracy": 0.7916666666666666 }, "middle_click": { "n": 3, "exact": 0.0, "action_accuracy": 0.3333333333333333 }, "mouse_move": { "n": 36, "exact": 0.0, "action_accuracy": 0.16666666666666666 }, "right_click": { "n": 160, "exact": 0.00625, "action_accuracy": 0.6625 }, "scroll": { "n": 83, "exact": 0.04819277108433735, "action_accuracy": 0.8192771084337349 }, "terminate": { "n": 163, "exact": 0.8773006134969326, "action_accuracy": 0.8773006134969326 }, "triple_click": { "n": 15, "exact": 0.0, "action_accuracy": 0.6 }, "type": { "n": 399, "exact": 0.3007518796992481, "action_accuracy": 0.9022556390977443 } } }, "scope": "Five OSWorld tasks unscored; development validation repeatedly observed; not an official leaderboard submission." }