| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.0, | |
| "eval_steps": 50, | |
| "global_step": 90, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 4.725662708282471, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.2963, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.30786412954330444, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.30786412954330444, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 172.38189697265625, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 172.38189697265625, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.785310745239258, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.785310745239258, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 612.4951171875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.6612959956709956, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 612.4951171875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 612.4951171875, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 612.4951171875, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.3111322224140167, | |
| "eval_runtime": 6.0296, | |
| "eval_samples_per_second": 1.824, | |
| "eval_steps_per_second": 0.332, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 4.223657608032227, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2833, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3102341890335083, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3102341890335083, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 173.5974578857422, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 173.5974578857422, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.73798179626465, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.73798179626465, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 610.9443359375, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 610.9443359375, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 610.9443359375, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 610.9443359375, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.309558629989624, | |
| "eval_runtime": 5.9488, | |
| "eval_samples_per_second": 1.849, | |
| "eval_steps_per_second": 0.336, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 3.3170320987701416, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2568, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3211533725261688, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3211533725261688, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 179.75201416015625, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 179.75201416015625, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.536266326904297, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.536266326904297, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 605.5822143554688, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 605.5822143554688, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 605.5822143554688, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 605.5822143554688, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.304984450340271, | |
| "eval_runtime": 7.2979, | |
| "eval_samples_per_second": 1.507, | |
| "eval_steps_per_second": 0.274, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 2.2900643348693848, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.2177, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3480110168457031, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3480110168457031, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 195.35403442382812, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 195.35403442382812, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.041318893432617, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.041318893432617, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 593.150634765625, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 593.150634765625, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 593.150634765625, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.7237959956709956, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 593.150634765625, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.29575565457344055, | |
| "eval_runtime": 6.0903, | |
| "eval_samples_per_second": 1.806, | |
| "eval_steps_per_second": 0.328, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 1.5022671222686768, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.1797, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3965001106262207, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3965001106262207, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 223.7809295654297, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 223.7809295654297, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 26.097108840942383, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 26.097108840942383, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 570.329345703125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.682129329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 570.329345703125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 570.329345703125, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 570.329345703125, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.28263404965400696, | |
| "eval_runtime": 6.0031, | |
| "eval_samples_per_second": 1.832, | |
| "eval_steps_per_second": 0.333, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 0.6781304478645325, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1419, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.43997177481651306, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.43997177481651306, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 249.57528686523438, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 249.57528686523438, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 25.200597763061523, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 25.200597763061523, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 550.5452880859375, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 550.5452880859375, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 550.5452880859375, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 550.5452880859375, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.276452898979187, | |
| "eval_runtime": 5.4943, | |
| "eval_samples_per_second": 2.002, | |
| "eval_steps_per_second": 0.364, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.4487136900424957, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 0.1057, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.37805190682411194, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.37805190682411194, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 211.4693603515625, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 211.4693603515625, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 26.396636962890625, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 26.396636962890625, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 582.5875244140625, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 582.5875244140625, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 582.5875244140625, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 582.5875244140625, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.295420378446579, | |
| "eval_runtime": 6.1313, | |
| "eval_samples_per_second": 1.794, | |
| "eval_steps_per_second": 0.326, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 0.16805221140384674, | |
| "learning_rate": 1.555555555555556e-05, | |
| "loss": 0.0815, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.31827351450920105, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.31827351450920105, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 178.09083557128906, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 178.09083557128906, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.63851547241211, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.63851547241211, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 612.4237060546875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 612.4237060546875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 612.4237060546875, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 612.4237060546875, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.3164561986923218, | |
| "eval_runtime": 6.5067, | |
| "eval_samples_per_second": 1.691, | |
| "eval_steps_per_second": 0.307, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 0.16789764165878296, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.0664, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.30298155546188354, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.30298155546188354, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 169.24606323242188, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 169.24606323242188, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.91897964477539, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.91897964477539, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 618.7899169921875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 618.7899169921875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 618.7899169921875, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 618.7899169921875, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.31994009017944336, | |
| "eval_runtime": 6.8319, | |
| "eval_samples_per_second": 1.61, | |
| "eval_steps_per_second": 0.293, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 0.1519097536802292, | |
| "learning_rate": 1.1111111111111113e-05, | |
| "loss": 0.0497, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.31682828068733215, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.31682828068733215, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 176.45004272460938, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 176.45004272460938, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.598316192626953, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.598316192626953, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 611.2236328125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 611.2236328125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 611.2236328125, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.650879329004329, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 611.2236328125, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.31398141384124756, | |
| "eval_runtime": 6.4168, | |
| "eval_samples_per_second": 1.714, | |
| "eval_steps_per_second": 0.312, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "grad_norm": 0.17287826538085938, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0402, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.33553600311279297, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.33553600311279297, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 186.30569458007812, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 186.30569458007812, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.178071975708008, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.178071975708008, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 601.0965576171875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 601.0965576171875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 601.0965576171875, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.6321293290043289, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 601.0965576171875, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.3081442415714264, | |
| "eval_runtime": 6.4435, | |
| "eval_samples_per_second": 1.707, | |
| "eval_steps_per_second": 0.31, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "grad_norm": 0.1389617770910263, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0346, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3391532003879547, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3391532003879547, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 187.9766082763672, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 187.9766082763672, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.080123901367188, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.080123901367188, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 598.4072265625, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 598.4072265625, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 598.4072265625, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 598.4072265625, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.30722811818122864, | |
| "eval_runtime": 6.555, | |
| "eval_samples_per_second": 1.678, | |
| "eval_steps_per_second": 0.305, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "grad_norm": 0.10099858045578003, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0293, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.34109100699424744, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.34109100699424744, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 189.05935668945312, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 189.05935668945312, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.041587829589844, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.041587829589844, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 597.2989501953125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 597.2989501953125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 597.2989501953125, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 597.2989501953125, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.30660775303840637, | |
| "eval_runtime": 5.9875, | |
| "eval_samples_per_second": 1.837, | |
| "eval_steps_per_second": 0.334, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "grad_norm": 0.12931989133358002, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0302, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3385917544364929, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3385917544364929, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 187.77444458007812, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 187.77444458007812, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.100971221923828, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.100971221923828, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 598.5726318359375, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 598.5726318359375, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 598.5726318359375, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 598.5726318359375, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.30758312344551086, | |
| "eval_runtime": 5.1895, | |
| "eval_samples_per_second": 2.12, | |
| "eval_steps_per_second": 0.385, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 0.16915012896060944, | |
| "learning_rate": 0.0, | |
| "loss": 0.0287, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.3379952907562256, | |
| "eval_custom-arc-semantics-data-jp_cosine_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.3379952907562256, | |
| "eval_custom-arc-semantics-data-jp_cosine_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_cosine_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 187.5118865966797, | |
| "eval_custom-arc-semantics-data-jp_dot_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_dot_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_dot_f1_threshold": 187.5118865966797, | |
| "eval_custom-arc-semantics-data-jp_dot_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_dot_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 27.118305206298828, | |
| "eval_custom-arc-semantics-data-jp_euclidean_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 27.118305206298828, | |
| "eval_custom-arc-semantics-data-jp_euclidean_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_euclidean_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 598.9317626953125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_ap": 0.619629329004329, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 598.9317626953125, | |
| "eval_custom-arc-semantics-data-jp_manhattan_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_manhattan_recall": 0.875, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy": 0.6363636363636364, | |
| "eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 598.9317626953125, | |
| "eval_custom-arc-semantics-data-jp_max_ap": 0.6946293290043289, | |
| "eval_custom-arc-semantics-data-jp_max_f1": 0.7777777777777777, | |
| "eval_custom-arc-semantics-data-jp_max_f1_threshold": 598.9317626953125, | |
| "eval_custom-arc-semantics-data-jp_max_precision": 0.7, | |
| "eval_custom-arc-semantics-data-jp_max_recall": 0.875, | |
| "eval_loss": 0.3078480362892151, | |
| "eval_runtime": 5.9867, | |
| "eval_samples_per_second": 1.837, | |
| "eval_steps_per_second": 0.334, | |
| "step": 90 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 90, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |