| { | |
| "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": 0.57, | |
| "eval_loss": 0.627579391002655, | |
| "eval_runtime": 2.9468, | |
| "eval_samples_per_second": 33.935, | |
| "eval_steps_per_second": 4.412, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.57, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6135770678520203, | |
| "eval_runtime": 3.0105, | |
| "eval_samples_per_second": 33.217, | |
| "eval_steps_per_second": 4.318, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6773912906646729, | |
| "eval_runtime": 3.0612, | |
| "eval_samples_per_second": 32.667, | |
| "eval_steps_per_second": 4.247, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5963658690452576, | |
| "eval_runtime": 3.096, | |
| "eval_samples_per_second": 32.3, | |
| "eval_steps_per_second": 4.199, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5315800309181213, | |
| "eval_runtime": 3.1161, | |
| "eval_samples_per_second": 32.091, | |
| "eval_steps_per_second": 4.172, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5231013298034668, | |
| "eval_runtime": 3.1314, | |
| "eval_samples_per_second": 31.935, | |
| "eval_steps_per_second": 4.151, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5155523419380188, | |
| "eval_runtime": 3.1398, | |
| "eval_samples_per_second": 31.849, | |
| "eval_steps_per_second": 4.14, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6215571761131287, | |
| "eval_runtime": 3.1438, | |
| "eval_samples_per_second": 31.809, | |
| "eval_steps_per_second": 4.135, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5012997388839722, | |
| "eval_runtime": 3.1482, | |
| "eval_samples_per_second": 31.764, | |
| "eval_steps_per_second": 4.129, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5733979940414429, | |
| "eval_runtime": 3.1518, | |
| "eval_samples_per_second": 31.728, | |
| "eval_steps_per_second": 4.125, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4682706594467163, | |
| "eval_runtime": 3.1601, | |
| "eval_samples_per_second": 31.645, | |
| "eval_steps_per_second": 4.114, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5332651734352112, | |
| "eval_runtime": 3.1621, | |
| "eval_samples_per_second": 31.624, | |
| "eval_steps_per_second": 4.111, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6740408539772034, | |
| "eval_runtime": 3.1473, | |
| "eval_samples_per_second": 31.773, | |
| "eval_steps_per_second": 4.131, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5184910893440247, | |
| "eval_runtime": 3.1461, | |
| "eval_samples_per_second": 31.786, | |
| "eval_steps_per_second": 4.132, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5030907392501831, | |
| "eval_runtime": 3.1466, | |
| "eval_samples_per_second": 31.78, | |
| "eval_steps_per_second": 4.131, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5397804379463196, | |
| "eval_runtime": 3.1405, | |
| "eval_samples_per_second": 31.843, | |
| "eval_steps_per_second": 4.14, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5245677828788757, | |
| "eval_runtime": 3.1417, | |
| "eval_samples_per_second": 31.83, | |
| "eval_steps_per_second": 4.138, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.7413635849952698, | |
| "eval_runtime": 3.149, | |
| "eval_samples_per_second": 31.756, | |
| "eval_steps_per_second": 4.128, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.6816568970680237, | |
| "eval_runtime": 3.1399, | |
| "eval_samples_per_second": 31.849, | |
| "eval_steps_per_second": 4.14, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.7352, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6655824184417725, | |
| "eval_runtime": 3.1414, | |
| "eval_samples_per_second": 31.833, | |
| "eval_steps_per_second": 4.138, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5838807225227356, | |
| "eval_runtime": 3.1408, | |
| "eval_samples_per_second": 31.839, | |
| "eval_steps_per_second": 4.139, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 20, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6626308560371399, | |
| "eval_runtime": 3.1413, | |
| "eval_samples_per_second": 31.834, | |
| "eval_steps_per_second": 4.138, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 20, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5017469525337219, | |
| "eval_runtime": 3.1427, | |
| "eval_samples_per_second": 31.82, | |
| "eval_steps_per_second": 4.137, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 20, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5168166160583496, | |
| "eval_runtime": 3.1452, | |
| "eval_samples_per_second": 31.795, | |
| "eval_steps_per_second": 4.133, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 20, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.78, | |
| "eval_loss": 0.5911619067192078, | |
| "eval_runtime": 3.1431, | |
| "eval_samples_per_second": 31.816, | |
| "eval_steps_per_second": 4.136, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.77, | |
| "eval_loss": 0.5595811605453491, | |
| "eval_runtime": 3.1433, | |
| "eval_samples_per_second": 31.813, | |
| "eval_steps_per_second": 4.136, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.77, | |
| "eval_loss": 0.48836490511894226, | |
| "eval_runtime": 3.1431, | |
| "eval_samples_per_second": 31.816, | |
| "eval_steps_per_second": 4.136, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.47384893894195557, | |
| "eval_runtime": 3.142, | |
| "eval_samples_per_second": 31.826, | |
| "eval_steps_per_second": 4.137, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5052289366722107, | |
| "eval_runtime": 3.1453, | |
| "eval_samples_per_second": 31.794, | |
| "eval_steps_per_second": 4.133, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.6162938475608826, | |
| "eval_runtime": 3.1453, | |
| "eval_samples_per_second": 31.793, | |
| "eval_steps_per_second": 4.133, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5823907256126404, | |
| "eval_runtime": 3.143, | |
| "eval_samples_per_second": 31.817, | |
| "eval_steps_per_second": 4.136, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4995167553424835, | |
| "eval_runtime": 3.1421, | |
| "eval_samples_per_second": 31.826, | |
| "eval_steps_per_second": 4.137, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4935573935508728, | |
| "eval_runtime": 3.1423, | |
| "eval_samples_per_second": 31.824, | |
| "eval_steps_per_second": 4.137, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5463616847991943, | |
| "eval_runtime": 3.1442, | |
| "eval_samples_per_second": 31.804, | |
| "eval_steps_per_second": 4.135, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5164341926574707, | |
| "eval_runtime": 3.1446, | |
| "eval_samples_per_second": 31.801, | |
| "eval_steps_per_second": 4.134, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5088481307029724, | |
| "eval_runtime": 3.1452, | |
| "eval_samples_per_second": 31.795, | |
| "eval_steps_per_second": 4.133, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.599052906036377, | |
| "eval_runtime": 3.1458, | |
| "eval_samples_per_second": 31.788, | |
| "eval_steps_per_second": 4.132, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.49629107117652893, | |
| "eval_runtime": 3.1523, | |
| "eval_samples_per_second": 31.723, | |
| "eval_steps_per_second": 4.124, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5085676312446594, | |
| "eval_runtime": 3.1502, | |
| "eval_samples_per_second": 31.744, | |
| "eval_steps_per_second": 4.127, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.411, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5202763080596924, | |
| "eval_runtime": 3.1439, | |
| "eval_samples_per_second": 31.808, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5843695998191833, | |
| "eval_runtime": 3.1411, | |
| "eval_samples_per_second": 31.836, | |
| "eval_steps_per_second": 4.139, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5285233855247498, | |
| "eval_runtime": 3.1423, | |
| "eval_samples_per_second": 31.824, | |
| "eval_steps_per_second": 4.137, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5552850961685181, | |
| "eval_runtime": 3.1448, | |
| "eval_samples_per_second": 31.799, | |
| "eval_steps_per_second": 4.134, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5588386058807373, | |
| "eval_runtime": 3.1454, | |
| "eval_samples_per_second": 31.793, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5391697883605957, | |
| "eval_runtime": 3.1435, | |
| "eval_samples_per_second": 31.812, | |
| "eval_steps_per_second": 4.136, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5494285225868225, | |
| "eval_runtime": 3.1433, | |
| "eval_samples_per_second": 31.814, | |
| "eval_steps_per_second": 4.136, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.4982169270515442, | |
| "eval_runtime": 3.1427, | |
| "eval_samples_per_second": 31.819, | |
| "eval_steps_per_second": 4.137, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5374172329902649, | |
| "eval_runtime": 3.1407, | |
| "eval_samples_per_second": 31.84, | |
| "eval_steps_per_second": 4.139, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5730433464050293, | |
| "eval_runtime": 3.142, | |
| "eval_samples_per_second": 31.827, | |
| "eval_steps_per_second": 4.137, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.51490718126297, | |
| "eval_runtime": 3.1446, | |
| "eval_samples_per_second": 31.8, | |
| "eval_steps_per_second": 4.134, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.49485138058662415, | |
| "eval_runtime": 3.1403, | |
| "eval_samples_per_second": 31.844, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5295071601867676, | |
| "eval_runtime": 3.1397, | |
| "eval_samples_per_second": 31.851, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5223097801208496, | |
| "eval_runtime": 3.1423, | |
| "eval_samples_per_second": 31.824, | |
| "eval_steps_per_second": 4.137, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5616713762283325, | |
| "eval_runtime": 3.141, | |
| "eval_samples_per_second": 31.837, | |
| "eval_steps_per_second": 4.139, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5373037457466125, | |
| "eval_runtime": 3.1394, | |
| "eval_samples_per_second": 31.853, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.48570069670677185, | |
| "eval_runtime": 3.1398, | |
| "eval_samples_per_second": 31.849, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.49541863799095154, | |
| "eval_runtime": 3.1394, | |
| "eval_samples_per_second": 31.853, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5024493932723999, | |
| "eval_runtime": 3.1403, | |
| "eval_samples_per_second": 31.844, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.497109591960907, | |
| "eval_runtime": 3.1489, | |
| "eval_samples_per_second": 31.757, | |
| "eval_steps_per_second": 4.128, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.318, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5264896154403687, | |
| "eval_runtime": 3.1479, | |
| "eval_samples_per_second": 31.767, | |
| "eval_steps_per_second": 4.13, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4966976046562195, | |
| "eval_runtime": 3.1464, | |
| "eval_samples_per_second": 31.783, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.497234046459198, | |
| "eval_runtime": 3.1465, | |
| "eval_samples_per_second": 31.782, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4908214509487152, | |
| "eval_runtime": 3.1342, | |
| "eval_samples_per_second": 31.906, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5055844783782959, | |
| "eval_runtime": 3.1392, | |
| "eval_samples_per_second": 31.855, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5230618119239807, | |
| "eval_runtime": 3.1268, | |
| "eval_samples_per_second": 31.981, | |
| "eval_steps_per_second": 4.158, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.47373998165130615, | |
| "eval_runtime": 3.1252, | |
| "eval_samples_per_second": 31.998, | |
| "eval_steps_per_second": 4.16, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5015537142753601, | |
| "eval_runtime": 3.1238, | |
| "eval_samples_per_second": 32.012, | |
| "eval_steps_per_second": 4.162, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.49883437156677246, | |
| "eval_runtime": 3.1204, | |
| "eval_samples_per_second": 32.047, | |
| "eval_steps_per_second": 4.166, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5276287794113159, | |
| "eval_runtime": 3.1254, | |
| "eval_samples_per_second": 31.996, | |
| "eval_steps_per_second": 4.16, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.4912046790122986, | |
| "eval_runtime": 3.1233, | |
| "eval_samples_per_second": 32.018, | |
| "eval_steps_per_second": 4.162, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.48646289110183716, | |
| "eval_runtime": 3.1257, | |
| "eval_samples_per_second": 31.993, | |
| "eval_steps_per_second": 4.159, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.475396066904068, | |
| "eval_runtime": 3.1256, | |
| "eval_samples_per_second": 31.994, | |
| "eval_steps_per_second": 4.159, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.49221473932266235, | |
| "eval_runtime": 3.1249, | |
| "eval_samples_per_second": 32.001, | |
| "eval_steps_per_second": 4.16, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4884068965911865, | |
| "eval_runtime": 3.1256, | |
| "eval_samples_per_second": 31.994, | |
| "eval_steps_per_second": 4.159, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.48677849769592285, | |
| "eval_runtime": 3.1234, | |
| "eval_samples_per_second": 32.017, | |
| "eval_steps_per_second": 4.162, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.48715740442276, | |
| "eval_runtime": 3.1231, | |
| "eval_samples_per_second": 32.02, | |
| "eval_steps_per_second": 4.163, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4847962260246277, | |
| "eval_runtime": 3.1233, | |
| "eval_samples_per_second": 32.017, | |
| "eval_steps_per_second": 4.162, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4922550320625305, | |
| "eval_runtime": 3.1248, | |
| "eval_samples_per_second": 32.002, | |
| "eval_steps_per_second": 4.16, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.4888269007205963, | |
| "eval_runtime": 3.1211, | |
| "eval_samples_per_second": 32.04, | |
| "eval_steps_per_second": 4.165, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.287, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4912576377391815, | |
| "eval_runtime": 3.1223, | |
| "eval_samples_per_second": 32.028, | |
| "eval_steps_per_second": 4.164, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.78, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.43781012725830076, | |
| "train_runtime": 1665.2419, | |
| "train_samples_per_second": 19.216, | |
| "train_steps_per_second": 1.201 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |