| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 80.0, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.49578723311424255, | |
| "eval_runtime": 2.9005, | |
| "eval_samples_per_second": 34.477, | |
| "eval_steps_per_second": 4.482, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.46, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.5955537557601929, | |
| "eval_runtime": 2.9308, | |
| "eval_samples_per_second": 34.12, | |
| "eval_steps_per_second": 4.436, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.54, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.5376533269882202, | |
| "eval_runtime": 2.9692, | |
| "eval_samples_per_second": 33.679, | |
| "eval_steps_per_second": 4.378, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.54, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4201565980911255, | |
| "eval_runtime": 2.9986, | |
| "eval_samples_per_second": 33.349, | |
| "eval_steps_per_second": 4.335, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.43667927384376526, | |
| "eval_runtime": 3.0016, | |
| "eval_samples_per_second": 33.316, | |
| "eval_steps_per_second": 4.331, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.43700677156448364, | |
| "eval_runtime": 3.0099, | |
| "eval_samples_per_second": 33.223, | |
| "eval_steps_per_second": 4.319, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.42071959376335144, | |
| "eval_runtime": 3.0221, | |
| "eval_samples_per_second": 33.09, | |
| "eval_steps_per_second": 4.302, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.4423440098762512, | |
| "eval_runtime": 3.0262, | |
| "eval_samples_per_second": 33.044, | |
| "eval_steps_per_second": 4.296, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.41069385409355164, | |
| "eval_runtime": 3.0271, | |
| "eval_samples_per_second": 33.035, | |
| "eval_steps_per_second": 4.294, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.4332168996334076, | |
| "eval_runtime": 3.0297, | |
| "eval_samples_per_second": 33.007, | |
| "eval_steps_per_second": 4.291, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.4055112898349762, | |
| "eval_runtime": 3.031, | |
| "eval_samples_per_second": 32.992, | |
| "eval_steps_per_second": 4.289, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.4375588893890381, | |
| "eval_runtime": 3.0325, | |
| "eval_samples_per_second": 32.977, | |
| "eval_steps_per_second": 4.287, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.40624508261680603, | |
| "eval_runtime": 3.0305, | |
| "eval_samples_per_second": 32.998, | |
| "eval_steps_per_second": 4.29, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4000048339366913, | |
| "eval_runtime": 3.0315, | |
| "eval_samples_per_second": 32.986, | |
| "eval_steps_per_second": 4.288, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.40517398715019226, | |
| "eval_runtime": 3.0336, | |
| "eval_samples_per_second": 32.965, | |
| "eval_steps_per_second": 4.285, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.3960806727409363, | |
| "eval_runtime": 3.0345, | |
| "eval_samples_per_second": 32.955, | |
| "eval_steps_per_second": 4.284, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 15, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.3975551724433899, | |
| "eval_runtime": 3.0333, | |
| "eval_samples_per_second": 32.968, | |
| "eval_steps_per_second": 4.286, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 15, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.41863301396369934, | |
| "eval_runtime": 3.0326, | |
| "eval_samples_per_second": 32.975, | |
| "eval_steps_per_second": 4.287, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 15, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.4303971230983734, | |
| "eval_runtime": 3.0325, | |
| "eval_samples_per_second": 32.976, | |
| "eval_steps_per_second": 4.287, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 15, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.731, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4357996881008148, | |
| "eval_runtime": 3.0345, | |
| "eval_samples_per_second": 32.955, | |
| "eval_steps_per_second": 4.284, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.41348427534103394, | |
| "eval_runtime": 3.0336, | |
| "eval_samples_per_second": 32.964, | |
| "eval_steps_per_second": 4.285, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4180167019367218, | |
| "eval_runtime": 3.0311, | |
| "eval_samples_per_second": 32.991, | |
| "eval_steps_per_second": 4.289, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.46266496181488037, | |
| "eval_runtime": 3.0317, | |
| "eval_samples_per_second": 32.985, | |
| "eval_steps_per_second": 4.288, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.4150283932685852, | |
| "eval_runtime": 3.0391, | |
| "eval_samples_per_second": 32.904, | |
| "eval_steps_per_second": 4.278, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4004598557949066, | |
| "eval_runtime": 3.032, | |
| "eval_samples_per_second": 32.982, | |
| "eval_steps_per_second": 4.288, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4123203158378601, | |
| "eval_runtime": 3.0302, | |
| "eval_samples_per_second": 33.001, | |
| "eval_steps_per_second": 4.29, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4341800808906555, | |
| "eval_runtime": 3.0344, | |
| "eval_samples_per_second": 32.955, | |
| "eval_steps_per_second": 4.284, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.45509403944015503, | |
| "eval_runtime": 3.0328, | |
| "eval_samples_per_second": 32.973, | |
| "eval_steps_per_second": 4.287, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.42222651839256287, | |
| "eval_runtime": 3.0333, | |
| "eval_samples_per_second": 32.967, | |
| "eval_steps_per_second": 4.286, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4225788116455078, | |
| "eval_runtime": 3.0312, | |
| "eval_samples_per_second": 32.99, | |
| "eval_steps_per_second": 4.289, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.47024014592170715, | |
| "eval_runtime": 3.0363, | |
| "eval_samples_per_second": 32.934, | |
| "eval_steps_per_second": 4.281, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.41000667214393616, | |
| "eval_runtime": 3.0339, | |
| "eval_samples_per_second": 32.961, | |
| "eval_steps_per_second": 4.285, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4317716658115387, | |
| "eval_runtime": 3.0312, | |
| "eval_samples_per_second": 32.99, | |
| "eval_steps_per_second": 4.289, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4446772634983063, | |
| "eval_runtime": 3.0338, | |
| "eval_samples_per_second": 32.962, | |
| "eval_steps_per_second": 4.285, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.3881460130214691, | |
| "eval_runtime": 3.0331, | |
| "eval_samples_per_second": 32.969, | |
| "eval_steps_per_second": 4.286, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.42335742712020874, | |
| "eval_runtime": 3.0313, | |
| "eval_samples_per_second": 32.989, | |
| "eval_steps_per_second": 4.289, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4869128465652466, | |
| "eval_runtime": 3.0313, | |
| "eval_samples_per_second": 32.989, | |
| "eval_steps_per_second": 4.289, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4352262020111084, | |
| "eval_runtime": 3.0335, | |
| "eval_samples_per_second": 32.965, | |
| "eval_steps_per_second": 4.285, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4464748501777649, | |
| "eval_runtime": 3.0309, | |
| "eval_samples_per_second": 32.993, | |
| "eval_steps_per_second": 4.289, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.5086, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.413508802652359, | |
| "eval_runtime": 3.0326, | |
| "eval_samples_per_second": 32.974, | |
| "eval_steps_per_second": 4.287, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4060712456703186, | |
| "eval_runtime": 3.0312, | |
| "eval_samples_per_second": 32.99, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.44368666410446167, | |
| "eval_runtime": 3.0306, | |
| "eval_samples_per_second": 32.997, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4461361765861511, | |
| "eval_runtime": 3.0306, | |
| "eval_samples_per_second": 32.997, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.41435202956199646, | |
| "eval_runtime": 3.0303, | |
| "eval_samples_per_second": 33.0, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.3973015248775482, | |
| "eval_runtime": 3.0301, | |
| "eval_samples_per_second": 33.002, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 34, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.45110881328582764, | |
| "eval_runtime": 3.0313, | |
| "eval_samples_per_second": 32.989, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4273070991039276, | |
| "eval_runtime": 3.0301, | |
| "eval_samples_per_second": 33.002, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4100435972213745, | |
| "eval_runtime": 3.0299, | |
| "eval_samples_per_second": 33.004, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4208601415157318, | |
| "eval_runtime": 3.0294, | |
| "eval_samples_per_second": 33.01, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.41906824707984924, | |
| "eval_runtime": 3.0315, | |
| "eval_samples_per_second": 32.987, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.40232476592063904, | |
| "eval_runtime": 3.0299, | |
| "eval_samples_per_second": 33.004, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.40375789999961853, | |
| "eval_runtime": 3.0318, | |
| "eval_samples_per_second": 32.984, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.41481122374534607, | |
| "eval_runtime": 3.0329, | |
| "eval_samples_per_second": 32.972, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.42626023292541504, | |
| "eval_runtime": 3.0313, | |
| "eval_samples_per_second": 32.989, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.43311816453933716, | |
| "eval_runtime": 3.035, | |
| "eval_samples_per_second": 32.949, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4372536838054657, | |
| "eval_runtime": 3.0311, | |
| "eval_samples_per_second": 32.991, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.40811866521835327, | |
| "eval_runtime": 3.0315, | |
| "eval_samples_per_second": 32.987, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.40780752897262573, | |
| "eval_runtime": 3.0299, | |
| "eval_samples_per_second": 33.004, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.42498713731765747, | |
| "eval_runtime": 3.0363, | |
| "eval_samples_per_second": 32.935, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.4268, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4223678708076477, | |
| "eval_runtime": 3.0288, | |
| "eval_samples_per_second": 33.016, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4255259335041046, | |
| "eval_runtime": 3.0305, | |
| "eval_samples_per_second": 32.997, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.41144871711730957, | |
| "eval_runtime": 3.0322, | |
| "eval_samples_per_second": 32.98, | |
| "eval_steps_per_second": 4.287, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.42656296491622925, | |
| "eval_runtime": 3.0304, | |
| "eval_samples_per_second": 32.999, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4097367227077484, | |
| "eval_runtime": 3.03, | |
| "eval_samples_per_second": 33.004, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.40528982877731323, | |
| "eval_runtime": 3.0318, | |
| "eval_samples_per_second": 32.983, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4050586223602295, | |
| "eval_runtime": 3.0289, | |
| "eval_samples_per_second": 33.015, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.41345199942588806, | |
| "eval_runtime": 3.031, | |
| "eval_samples_per_second": 32.993, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.3958780765533447, | |
| "eval_runtime": 3.0359, | |
| "eval_samples_per_second": 32.939, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4162384867668152, | |
| "eval_runtime": 3.0305, | |
| "eval_samples_per_second": 32.998, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.4060596227645874, | |
| "eval_runtime": 3.0315, | |
| "eval_samples_per_second": 32.986, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4015595614910126, | |
| "eval_runtime": 3.03, | |
| "eval_samples_per_second": 33.003, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4193567931652069, | |
| "eval_runtime": 3.0293, | |
| "eval_samples_per_second": 33.011, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4097726345062256, | |
| "eval_runtime": 3.0293, | |
| "eval_samples_per_second": 33.011, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.41785815358161926, | |
| "eval_runtime": 3.0313, | |
| "eval_samples_per_second": 32.989, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.41046372056007385, | |
| "eval_runtime": 3.0316, | |
| "eval_samples_per_second": 32.986, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.413968950510025, | |
| "eval_runtime": 3.0295, | |
| "eval_samples_per_second": 33.008, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.4080766439437866, | |
| "eval_runtime": 3.0307, | |
| "eval_samples_per_second": 32.996, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.40437057614326477, | |
| "eval_runtime": 3.0312, | |
| "eval_samples_per_second": 32.99, | |
| "eval_steps_per_second": 4.289, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.3996027410030365, | |
| "eval_runtime": 3.033, | |
| "eval_samples_per_second": 32.971, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.3915, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.40144088864326477, | |
| "eval_runtime": 3.0353, | |
| "eval_samples_per_second": 32.946, | |
| "eval_steps_per_second": 4.283, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.5144821853637696, | |
| "train_runtime": 1625.7247, | |
| "train_samples_per_second": 19.684, | |
| "train_steps_per_second": 1.23 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |