| { | |
| "best_metric": 0.42871516942977905, | |
| "best_model_checkpoint": "./convnext-tiny-upgrade-1k-224-batch-32/checkpoint-5500", | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 5500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 9.459319114685059, | |
| "learning_rate": 2.9975536558892034e-05, | |
| "loss": 3.1737, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 15.453875541687012, | |
| "learning_rate": 2.9902226030228252e-05, | |
| "loss": 2.5523, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 14.928990364074707, | |
| "learning_rate": 2.9780307537715396e-05, | |
| "loss": 2.0438, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 14.474929809570312, | |
| "learning_rate": 2.9610178754135005e-05, | |
| "loss": 1.7769, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 16.29086685180664, | |
| "learning_rate": 2.939239460421746e-05, | |
| "loss": 1.5523, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7009940357852883, | |
| "eval_loss": 1.2082535028457642, | |
| "eval_runtime": 49.3679, | |
| "eval_samples_per_second": 50.944, | |
| "eval_steps_per_second": 1.6, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 23.075525283813477, | |
| "learning_rate": 2.9127665454592872e-05, | |
| "loss": 1.448, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 22.86272430419922, | |
| "learning_rate": 2.8816854796722754e-05, | |
| "loss": 1.2812, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 18.48836898803711, | |
| "learning_rate": 2.8460976430370375e-05, | |
| "loss": 1.2282, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "grad_norm": 32.918296813964844, | |
| "learning_rate": 2.8061191156796658e-05, | |
| "loss": 1.2056, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 19.546428680419922, | |
| "learning_rate": 2.7618802992467718e-05, | |
| "loss": 1.1179, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 39.9352912902832, | |
| "learning_rate": 2.7135254915624213e-05, | |
| "loss": 1.0852, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7960238568588469, | |
| "eval_loss": 0.7955119013786316, | |
| "eval_runtime": 49.699, | |
| "eval_samples_per_second": 50.605, | |
| "eval_steps_per_second": 1.59, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 31.628759384155273, | |
| "learning_rate": 2.661212415958624e-05, | |
| "loss": 1.0018, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 20.639881134033203, | |
| "learning_rate": 2.6051117068146073e-05, | |
| "loss": 1.0004, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 34.07547378540039, | |
| "learning_rate": 2.5454063529829405e-05, | |
| "loss": 0.9302, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 21.45457649230957, | |
| "learning_rate": 2.482291100917928e-05, | |
| "loss": 0.9218, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 16.903533935546875, | |
| "learning_rate": 2.4159718194531573e-05, | |
| "loss": 0.9179, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8258449304174951, | |
| "eval_loss": 0.6425434947013855, | |
| "eval_runtime": 50.6126, | |
| "eval_samples_per_second": 49.691, | |
| "eval_steps_per_second": 1.561, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 20.296306610107422, | |
| "learning_rate": 2.3466648283001542e-05, | |
| "loss": 0.8552, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "grad_norm": 25.9653263092041, | |
| "learning_rate": 2.274596192458443e-05, | |
| "loss": 0.8283, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "grad_norm": 14.738792419433594, | |
| "learning_rate": 2.2000009848385107e-05, | |
| "loss": 0.8234, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 17.272687911987305, | |
| "learning_rate": 2.12312251950283e-05, | |
| "loss": 0.8101, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "grad_norm": 19.54085922241211, | |
| "learning_rate": 2.0442115580259615e-05, | |
| "loss": 0.8235, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 30.610021591186523, | |
| "learning_rate": 1.963525491562421e-05, | |
| "loss": 0.7621, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8548707753479126, | |
| "eval_loss": 0.5425807237625122, | |
| "eval_runtime": 49.9135, | |
| "eval_samples_per_second": 50.387, | |
| "eval_steps_per_second": 1.583, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "grad_norm": 17.56275749206543, | |
| "learning_rate": 1.8813275012902307e-05, | |
| "loss": 0.743, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "grad_norm": 21.964000701904297, | |
| "learning_rate": 1.7978856999686182e-05, | |
| "loss": 0.7278, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "grad_norm": 19.241918563842773, | |
| "learning_rate": 1.713472257409928e-05, | |
| "loss": 0.7047, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "grad_norm": 30.8186092376709, | |
| "learning_rate": 1.6283625127182596e-05, | |
| "loss": 0.7558, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "grad_norm": 16.47164535522461, | |
| "learning_rate": 1.5428340761905444e-05, | |
| "loss": 0.7506, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8624254473161034, | |
| "eval_loss": 0.5018457770347595, | |
| "eval_runtime": 49.7681, | |
| "eval_samples_per_second": 50.534, | |
| "eval_steps_per_second": 1.587, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "grad_norm": 21.695837020874023, | |
| "learning_rate": 1.4571659238094557e-05, | |
| "loss": 0.693, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "grad_norm": 19.108768463134766, | |
| "learning_rate": 1.3716374872817408e-05, | |
| "loss": 0.6944, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "grad_norm": 19.964969635009766, | |
| "learning_rate": 1.2865277425900725e-05, | |
| "loss": 0.656, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 17.055917739868164, | |
| "learning_rate": 1.2021143000313822e-05, | |
| "loss": 0.6643, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "grad_norm": 15.315804481506348, | |
| "learning_rate": 1.11867249870977e-05, | |
| "loss": 0.6955, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 33.08659744262695, | |
| "learning_rate": 1.036474508437579e-05, | |
| "loss": 0.6774, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.868389662027833, | |
| "eval_loss": 0.4792053699493408, | |
| "eval_runtime": 49.8325, | |
| "eval_samples_per_second": 50.469, | |
| "eval_steps_per_second": 1.585, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "grad_norm": 20.445402145385742, | |
| "learning_rate": 9.557884419740387e-06, | |
| "loss": 0.6568, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "grad_norm": 15.636629104614258, | |
| "learning_rate": 8.768774804971705e-06, | |
| "loss": 0.606, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "grad_norm": 18.987926483154297, | |
| "learning_rate": 7.999990151614895e-06, | |
| "loss": 0.663, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "grad_norm": 14.477697372436523, | |
| "learning_rate": 7.254038075415573e-06, | |
| "loss": 0.6101, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "grad_norm": 24.2838191986084, | |
| "learning_rate": 6.533351716998466e-06, | |
| "loss": 0.6364, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8743538767395627, | |
| "eval_loss": 0.4525967836380005, | |
| "eval_runtime": 49.992, | |
| "eval_samples_per_second": 50.308, | |
| "eval_steps_per_second": 1.58, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "grad_norm": 23.1962947845459, | |
| "learning_rate": 5.840281805468427e-06, | |
| "loss": 0.5968, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "grad_norm": 21.290063858032227, | |
| "learning_rate": 5.177088990820725e-06, | |
| "loss": 0.6077, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "grad_norm": 25.02341079711914, | |
| "learning_rate": 4.5459364701706e-06, | |
| "loss": 0.6175, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "grad_norm": 21.104934692382812, | |
| "learning_rate": 3.948882931853924e-06, | |
| "loss": 0.5993, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "grad_norm": 17.36994171142578, | |
| "learning_rate": 3.3878758404137627e-06, | |
| "loss": 0.607, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 35.08428955078125, | |
| "learning_rate": 2.86474508437579e-06, | |
| "loss": 0.5961, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.8799204771371769, | |
| "eval_loss": 0.4362061619758606, | |
| "eval_runtime": 49.5582, | |
| "eval_samples_per_second": 50.748, | |
| "eval_steps_per_second": 1.594, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "grad_norm": 20.05402183532715, | |
| "learning_rate": 2.38119700753228e-06, | |
| "loss": 0.5695, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "grad_norm": 30.20684242248535, | |
| "learning_rate": 1.9388088432033446e-06, | |
| "loss": 0.5647, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "grad_norm": 21.600929260253906, | |
| "learning_rate": 1.5390235696296268e-06, | |
| "loss": 0.5637, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "grad_norm": 32.336669921875, | |
| "learning_rate": 1.1831452032772499e-06, | |
| "loss": 0.5916, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "grad_norm": 14.266773223876953, | |
| "learning_rate": 8.723345454071308e-07, | |
| "loss": 0.602, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.882703777335984, | |
| "eval_loss": 0.43156278133392334, | |
| "eval_runtime": 50.0221, | |
| "eval_samples_per_second": 50.278, | |
| "eval_steps_per_second": 1.579, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "grad_norm": 27.497398376464844, | |
| "learning_rate": 6.076053957825411e-07, | |
| "loss": 0.5848, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "grad_norm": 17.4595890045166, | |
| "learning_rate": 3.8982124586499804e-07, | |
| "loss": 0.5724, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "grad_norm": 30.596479415893555, | |
| "learning_rate": 2.1969246228460526e-07, | |
| "loss": 0.5712, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "grad_norm": 17.515758514404297, | |
| "learning_rate": 9.777396977174668e-08, | |
| "loss": 0.5994, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "grad_norm": 25.090110778808594, | |
| "learning_rate": 2.4463441107965278e-08, | |
| "loss": 0.5685, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 22.852954864501953, | |
| "learning_rate": 0.0, | |
| "loss": 0.5896, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8850894632206759, | |
| "eval_loss": 0.42871516942977905, | |
| "eval_runtime": 49.9143, | |
| "eval_samples_per_second": 50.386, | |
| "eval_steps_per_second": 1.583, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 5500, | |
| "total_flos": 4.4284716808619213e+18, | |
| "train_loss": 0.8922947484796697, | |
| "train_runtime": 5842.588, | |
| "train_samples_per_second": 30.091, | |
| "train_steps_per_second": 0.941 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 5500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 4.4284716808619213e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |