| { | |
| "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": 0.38, | |
| "eval_loss": 0.5778459906578064, | |
| "eval_runtime": 2.9164, | |
| "eval_samples_per_second": 34.289, | |
| "eval_steps_per_second": 4.458, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.38, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5810088515281677, | |
| "eval_runtime": 2.9673, | |
| "eval_samples_per_second": 33.701, | |
| "eval_steps_per_second": 4.381, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6271185874938965, | |
| "eval_runtime": 3.0283, | |
| "eval_samples_per_second": 33.022, | |
| "eval_steps_per_second": 4.293, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5771670341491699, | |
| "eval_runtime": 3.068, | |
| "eval_samples_per_second": 32.595, | |
| "eval_steps_per_second": 4.237, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5290189981460571, | |
| "eval_runtime": 3.0958, | |
| "eval_samples_per_second": 32.301, | |
| "eval_steps_per_second": 4.199, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5351535677909851, | |
| "eval_runtime": 3.1094, | |
| "eval_samples_per_second": 32.16, | |
| "eval_steps_per_second": 4.181, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5321515798568726, | |
| "eval_runtime": 3.1264, | |
| "eval_samples_per_second": 31.985, | |
| "eval_steps_per_second": 4.158, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5976082682609558, | |
| "eval_runtime": 3.1371, | |
| "eval_samples_per_second": 31.876, | |
| "eval_steps_per_second": 4.144, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5289967656135559, | |
| "eval_runtime": 3.1422, | |
| "eval_samples_per_second": 31.825, | |
| "eval_steps_per_second": 4.137, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.569958508014679, | |
| "eval_runtime": 3.1468, | |
| "eval_samples_per_second": 31.778, | |
| "eval_steps_per_second": 4.131, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5132102966308594, | |
| "eval_runtime": 3.1512, | |
| "eval_samples_per_second": 31.734, | |
| "eval_steps_per_second": 4.125, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5155261754989624, | |
| "eval_runtime": 3.163, | |
| "eval_samples_per_second": 31.616, | |
| "eval_steps_per_second": 4.11, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5048606395721436, | |
| "eval_runtime": 3.1576, | |
| "eval_samples_per_second": 31.67, | |
| "eval_steps_per_second": 4.117, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.507844865322113, | |
| "eval_runtime": 3.1514, | |
| "eval_samples_per_second": 31.732, | |
| "eval_steps_per_second": 4.125, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.48210325837135315, | |
| "eval_runtime": 3.149, | |
| "eval_samples_per_second": 31.756, | |
| "eval_steps_per_second": 4.128, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5371115207672119, | |
| "eval_runtime": 3.1482, | |
| "eval_samples_per_second": 31.765, | |
| "eval_steps_per_second": 4.129, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 15, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5406803488731384, | |
| "eval_runtime": 3.147, | |
| "eval_samples_per_second": 31.776, | |
| "eval_steps_per_second": 4.131, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 15, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6440941095352173, | |
| "eval_runtime": 3.1467, | |
| "eval_samples_per_second": 31.779, | |
| "eval_steps_per_second": 4.131, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 17, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5787262916564941, | |
| "eval_runtime": 3.1434, | |
| "eval_samples_per_second": 31.812, | |
| "eval_steps_per_second": 4.136, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 17, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.0075, | |
| "loss": 0.6402, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5646016597747803, | |
| "eval_runtime": 3.1464, | |
| "eval_samples_per_second": 31.782, | |
| "eval_steps_per_second": 4.132, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 17, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5552840828895569, | |
| "eval_runtime": 3.1468, | |
| "eval_samples_per_second": 31.779, | |
| "eval_steps_per_second": 4.131, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 17, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.613720178604126, | |
| "eval_runtime": 3.1448, | |
| "eval_samples_per_second": 31.799, | |
| "eval_steps_per_second": 4.134, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.49484309554100037, | |
| "eval_runtime": 3.146, | |
| "eval_samples_per_second": 31.786, | |
| "eval_steps_per_second": 4.132, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.550967812538147, | |
| "eval_runtime": 3.1467, | |
| "eval_samples_per_second": 31.78, | |
| "eval_steps_per_second": 4.131, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5984699130058289, | |
| "eval_runtime": 3.146, | |
| "eval_samples_per_second": 31.786, | |
| "eval_steps_per_second": 4.132, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.566030740737915, | |
| "eval_runtime": 3.1483, | |
| "eval_samples_per_second": 31.763, | |
| "eval_steps_per_second": 4.129, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.523222029209137, | |
| "eval_runtime": 3.1485, | |
| "eval_samples_per_second": 31.762, | |
| "eval_steps_per_second": 4.129, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5381032824516296, | |
| "eval_runtime": 3.1492, | |
| "eval_samples_per_second": 31.754, | |
| "eval_steps_per_second": 4.128, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5234293937683105, | |
| "eval_runtime": 3.147, | |
| "eval_samples_per_second": 31.776, | |
| "eval_steps_per_second": 4.131, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6145434975624084, | |
| "eval_runtime": 3.1487, | |
| "eval_samples_per_second": 31.759, | |
| "eval_steps_per_second": 4.129, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 21, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5482304096221924, | |
| "eval_runtime": 3.1469, | |
| "eval_samples_per_second": 31.777, | |
| "eval_steps_per_second": 4.131, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5246371626853943, | |
| "eval_runtime": 3.1482, | |
| "eval_samples_per_second": 31.764, | |
| "eval_steps_per_second": 4.129, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5257811546325684, | |
| "eval_runtime": 3.1597, | |
| "eval_samples_per_second": 31.649, | |
| "eval_steps_per_second": 4.114, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5278013348579407, | |
| "eval_runtime": 3.1476, | |
| "eval_samples_per_second": 31.771, | |
| "eval_steps_per_second": 4.13, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5244772434234619, | |
| "eval_runtime": 3.1478, | |
| "eval_samples_per_second": 31.769, | |
| "eval_steps_per_second": 4.13, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5072507262229919, | |
| "eval_runtime": 3.1456, | |
| "eval_samples_per_second": 31.79, | |
| "eval_steps_per_second": 4.133, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4982739984989166, | |
| "eval_runtime": 3.1457, | |
| "eval_samples_per_second": 31.79, | |
| "eval_steps_per_second": 4.133, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5076562166213989, | |
| "eval_runtime": 3.147, | |
| "eval_samples_per_second": 31.776, | |
| "eval_steps_per_second": 4.131, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5262712240219116, | |
| "eval_runtime": 3.1494, | |
| "eval_samples_per_second": 31.752, | |
| "eval_steps_per_second": 4.128, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.3719, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5096136331558228, | |
| "eval_runtime": 3.1456, | |
| "eval_samples_per_second": 31.79, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5338824391365051, | |
| "eval_runtime": 3.1496, | |
| "eval_samples_per_second": 31.75, | |
| "eval_steps_per_second": 4.128, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.4963531792163849, | |
| "eval_runtime": 3.1494, | |
| "eval_samples_per_second": 31.752, | |
| "eval_steps_per_second": 4.128, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.4831782877445221, | |
| "eval_runtime": 3.1482, | |
| "eval_samples_per_second": 31.764, | |
| "eval_steps_per_second": 4.129, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.49397972226142883, | |
| "eval_runtime": 3.1476, | |
| "eval_samples_per_second": 31.77, | |
| "eval_steps_per_second": 4.13, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.49817508459091187, | |
| "eval_runtime": 3.1507, | |
| "eval_samples_per_second": 31.739, | |
| "eval_steps_per_second": 4.126, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5449243187904358, | |
| "eval_runtime": 3.1528, | |
| "eval_samples_per_second": 31.718, | |
| "eval_steps_per_second": 4.123, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5174975395202637, | |
| "eval_runtime": 3.1533, | |
| "eval_samples_per_second": 31.713, | |
| "eval_steps_per_second": 4.123, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5207518339157104, | |
| "eval_runtime": 3.1444, | |
| "eval_samples_per_second": 31.803, | |
| "eval_steps_per_second": 4.134, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5281075835227966, | |
| "eval_runtime": 3.1471, | |
| "eval_samples_per_second": 31.775, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 41, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.4939603805541992, | |
| "eval_runtime": 3.1477, | |
| "eval_samples_per_second": 31.77, | |
| "eval_steps_per_second": 4.13, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5019869208335876, | |
| "eval_runtime": 3.1472, | |
| "eval_samples_per_second": 31.774, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5010063052177429, | |
| "eval_runtime": 3.1485, | |
| "eval_samples_per_second": 31.761, | |
| "eval_steps_per_second": 4.129, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.4799458384513855, | |
| "eval_runtime": 3.1487, | |
| "eval_samples_per_second": 31.759, | |
| "eval_steps_per_second": 4.129, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5205541253089905, | |
| "eval_runtime": 3.1466, | |
| "eval_samples_per_second": 31.78, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5148220062255859, | |
| "eval_runtime": 3.1506, | |
| "eval_samples_per_second": 31.74, | |
| "eval_steps_per_second": 4.126, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.481507271528244, | |
| "eval_runtime": 3.1452, | |
| "eval_samples_per_second": 31.795, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4951218068599701, | |
| "eval_runtime": 3.1457, | |
| "eval_samples_per_second": 31.789, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.507718563079834, | |
| "eval_runtime": 3.145, | |
| "eval_samples_per_second": 31.797, | |
| "eval_steps_per_second": 4.134, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5000278353691101, | |
| "eval_runtime": 3.146, | |
| "eval_samples_per_second": 31.786, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0025, | |
| "loss": 0.3121, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5124149918556213, | |
| "eval_runtime": 3.1487, | |
| "eval_samples_per_second": 31.759, | |
| "eval_steps_per_second": 4.129, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.4891398549079895, | |
| "eval_runtime": 3.1463, | |
| "eval_samples_per_second": 31.784, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.4993504583835602, | |
| "eval_runtime": 3.1451, | |
| "eval_samples_per_second": 31.795, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.49470365047454834, | |
| "eval_runtime": 3.1452, | |
| "eval_samples_per_second": 31.794, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4832766652107239, | |
| "eval_runtime": 3.1469, | |
| "eval_samples_per_second": 31.777, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5135172605514526, | |
| "eval_runtime": 3.1462, | |
| "eval_samples_per_second": 31.785, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.4803304374217987, | |
| "eval_runtime": 3.1472, | |
| "eval_samples_per_second": 31.775, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5058041214942932, | |
| "eval_runtime": 3.1437, | |
| "eval_samples_per_second": 31.81, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.48404258489608765, | |
| "eval_runtime": 3.1474, | |
| "eval_samples_per_second": 31.772, | |
| "eval_steps_per_second": 4.13, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5051174759864807, | |
| "eval_runtime": 3.1517, | |
| "eval_samples_per_second": 31.728, | |
| "eval_steps_per_second": 4.125, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4883221387863159, | |
| "eval_runtime": 3.1453, | |
| "eval_samples_per_second": 31.794, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.497234046459198, | |
| "eval_runtime": 3.146, | |
| "eval_samples_per_second": 31.786, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4789247214794159, | |
| "eval_runtime": 3.1455, | |
| "eval_samples_per_second": 31.791, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.49844783544540405, | |
| "eval_runtime": 3.1428, | |
| "eval_samples_per_second": 31.818, | |
| "eval_steps_per_second": 4.136, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4913429021835327, | |
| "eval_runtime": 3.1433, | |
| "eval_samples_per_second": 31.813, | |
| "eval_steps_per_second": 4.136, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4879249632358551, | |
| "eval_runtime": 3.1452, | |
| "eval_samples_per_second": 31.794, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.4901810884475708, | |
| "eval_runtime": 3.1415, | |
| "eval_samples_per_second": 31.832, | |
| "eval_steps_per_second": 4.138, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.48560553789138794, | |
| "eval_runtime": 3.1442, | |
| "eval_samples_per_second": 31.804, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.48929107189178467, | |
| "eval_runtime": 3.1449, | |
| "eval_samples_per_second": 31.797, | |
| "eval_steps_per_second": 4.134, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.4907485842704773, | |
| "eval_runtime": 3.1529, | |
| "eval_samples_per_second": 31.717, | |
| "eval_steps_per_second": 4.123, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2906, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.4898410737514496, | |
| "eval_runtime": 3.1539, | |
| "eval_samples_per_second": 31.707, | |
| "eval_steps_per_second": 4.122, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.4037043533325195, | |
| "train_runtime": 1668.0695, | |
| "train_samples_per_second": 19.184, | |
| "train_steps_per_second": 1.199 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |