| { | |
| "best_metric": 0.2655230164527893, | |
| "best_model_checkpoint": "./convnext-tiny-upgrade-384-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": 19.57126808166504, | |
| "learning_rate": 2.9975536558892034e-05, | |
| "loss": 2.7921, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 32.66088104248047, | |
| "learning_rate": 2.9902226030228252e-05, | |
| "loss": 1.698, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 20.374221801757812, | |
| "learning_rate": 2.9780307537715396e-05, | |
| "loss": 1.2474, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 25.04788589477539, | |
| "learning_rate": 2.9610178754135005e-05, | |
| "loss": 1.0608, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 14.614296913146973, | |
| "learning_rate": 2.939239460421746e-05, | |
| "loss": 0.9343, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8409542743538767, | |
| "eval_loss": 0.5731540322303772, | |
| "eval_runtime": 70.5042, | |
| "eval_samples_per_second": 35.672, | |
| "eval_steps_per_second": 1.121, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 26.049169540405273, | |
| "learning_rate": 2.9127665454592872e-05, | |
| "loss": 0.8642, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 22.884756088256836, | |
| "learning_rate": 2.8816854796722754e-05, | |
| "loss": 0.7679, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 21.125539779663086, | |
| "learning_rate": 2.8460976430370375e-05, | |
| "loss": 0.7144, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "grad_norm": 32.00399398803711, | |
| "learning_rate": 2.8061191156796658e-05, | |
| "loss": 0.7373, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 25.68100357055664, | |
| "learning_rate": 2.7618802992467718e-05, | |
| "loss": 0.6422, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 30.877243041992188, | |
| "learning_rate": 2.7135254915624213e-05, | |
| "loss": 0.6456, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8842942345924454, | |
| "eval_loss": 0.4129641652107239, | |
| "eval_runtime": 70.4926, | |
| "eval_samples_per_second": 35.678, | |
| "eval_steps_per_second": 1.121, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 19.94837760925293, | |
| "learning_rate": 2.661212415958624e-05, | |
| "loss": 0.5854, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 24.665042877197266, | |
| "learning_rate": 2.6051117068146073e-05, | |
| "loss": 0.5862, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 17.881755828857422, | |
| "learning_rate": 2.5454063529829405e-05, | |
| "loss": 0.5626, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 25.268552780151367, | |
| "learning_rate": 2.482291100917928e-05, | |
| "loss": 0.5462, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 19.982158660888672, | |
| "learning_rate": 2.4159718194531573e-05, | |
| "loss": 0.5478, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9025844930417495, | |
| "eval_loss": 0.3537212908267975, | |
| "eval_runtime": 69.6425, | |
| "eval_samples_per_second": 36.113, | |
| "eval_steps_per_second": 1.134, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 28.108890533447266, | |
| "learning_rate": 2.3466648283001542e-05, | |
| "loss": 0.5215, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "grad_norm": 22.890396118164062, | |
| "learning_rate": 2.274596192458443e-05, | |
| "loss": 0.5037, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "grad_norm": 24.4963321685791, | |
| "learning_rate": 2.2000009848385107e-05, | |
| "loss": 0.4873, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 23.050090789794922, | |
| "learning_rate": 2.12312251950283e-05, | |
| "loss": 0.5107, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "grad_norm": 14.223852157592773, | |
| "learning_rate": 2.0442115580259615e-05, | |
| "loss": 0.5071, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 28.922595977783203, | |
| "learning_rate": 1.963525491562421e-05, | |
| "loss": 0.466, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9180914512922466, | |
| "eval_loss": 0.3011625111103058, | |
| "eval_runtime": 70.4042, | |
| "eval_samples_per_second": 35.722, | |
| "eval_steps_per_second": 1.122, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "grad_norm": 15.419177055358887, | |
| "learning_rate": 1.8813275012902307e-05, | |
| "loss": 0.4539, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "grad_norm": 18.684144973754883, | |
| "learning_rate": 1.7978856999686182e-05, | |
| "loss": 0.434, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "grad_norm": 17.165298461914062, | |
| "learning_rate": 1.713472257409928e-05, | |
| "loss": 0.426, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "grad_norm": 13.944131851196289, | |
| "learning_rate": 1.6283625127182596e-05, | |
| "loss": 0.449, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "grad_norm": 19.61562728881836, | |
| "learning_rate": 1.5428340761905444e-05, | |
| "loss": 0.4619, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9141153081510934, | |
| "eval_loss": 0.303144633769989, | |
| "eval_runtime": 70.2862, | |
| "eval_samples_per_second": 35.782, | |
| "eval_steps_per_second": 1.124, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "grad_norm": 15.433624267578125, | |
| "learning_rate": 1.4571659238094557e-05, | |
| "loss": 0.4093, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "grad_norm": 24.75423812866211, | |
| "learning_rate": 1.3716374872817408e-05, | |
| "loss": 0.4117, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "grad_norm": 19.66401481628418, | |
| "learning_rate": 1.2865277425900725e-05, | |
| "loss": 0.3987, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 21.91676902770996, | |
| "learning_rate": 1.2021143000313822e-05, | |
| "loss": 0.3868, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "grad_norm": 13.552069664001465, | |
| "learning_rate": 1.11867249870977e-05, | |
| "loss": 0.4236, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 12.01144790649414, | |
| "learning_rate": 1.036474508437579e-05, | |
| "loss": 0.4046, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9157057654075547, | |
| "eval_loss": 0.29709887504577637, | |
| "eval_runtime": 70.5426, | |
| "eval_samples_per_second": 35.652, | |
| "eval_steps_per_second": 1.12, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "grad_norm": 5.744084358215332, | |
| "learning_rate": 9.557884419740387e-06, | |
| "loss": 0.3854, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "grad_norm": 16.507299423217773, | |
| "learning_rate": 8.768774804971705e-06, | |
| "loss": 0.3689, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "grad_norm": 17.598424911499023, | |
| "learning_rate": 7.999990151614895e-06, | |
| "loss": 0.3805, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "grad_norm": 20.120948791503906, | |
| "learning_rate": 7.254038075415573e-06, | |
| "loss": 0.352, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "grad_norm": 12.273432731628418, | |
| "learning_rate": 6.533351716998466e-06, | |
| "loss": 0.3852, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.9204771371769384, | |
| "eval_loss": 0.27634668350219727, | |
| "eval_runtime": 69.9648, | |
| "eval_samples_per_second": 35.947, | |
| "eval_steps_per_second": 1.129, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "grad_norm": 16.243404388427734, | |
| "learning_rate": 5.840281805468427e-06, | |
| "loss": 0.3342, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "grad_norm": 16.89383316040039, | |
| "learning_rate": 5.177088990820725e-06, | |
| "loss": 0.3651, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "grad_norm": 24.284339904785156, | |
| "learning_rate": 4.5459364701706e-06, | |
| "loss": 0.3668, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "grad_norm": 17.376432418823242, | |
| "learning_rate": 3.948882931853924e-06, | |
| "loss": 0.3521, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "grad_norm": 9.699359893798828, | |
| "learning_rate": 3.3878758404137627e-06, | |
| "loss": 0.36, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 61.89017105102539, | |
| "learning_rate": 2.86474508437579e-06, | |
| "loss": 0.3346, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9224652087475149, | |
| "eval_loss": 0.27124735713005066, | |
| "eval_runtime": 70.1639, | |
| "eval_samples_per_second": 35.845, | |
| "eval_steps_per_second": 1.126, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "grad_norm": 20.770275115966797, | |
| "learning_rate": 2.38119700753228e-06, | |
| "loss": 0.327, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "grad_norm": 16.55147361755371, | |
| "learning_rate": 1.9388088432033446e-06, | |
| "loss": 0.3215, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "grad_norm": 31.432947158813477, | |
| "learning_rate": 1.5390235696296268e-06, | |
| "loss": 0.3265, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "grad_norm": 23.547870635986328, | |
| "learning_rate": 1.1831452032772499e-06, | |
| "loss": 0.3433, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "grad_norm": 7.18527889251709, | |
| "learning_rate": 8.723345454071308e-07, | |
| "loss": 0.3386, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.9220675944333996, | |
| "eval_loss": 0.2671603262424469, | |
| "eval_runtime": 70.5325, | |
| "eval_samples_per_second": 35.657, | |
| "eval_steps_per_second": 1.12, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "grad_norm": 18.18093490600586, | |
| "learning_rate": 6.076053957825411e-07, | |
| "loss": 0.3248, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "grad_norm": 20.491416931152344, | |
| "learning_rate": 3.8982124586499804e-07, | |
| "loss": 0.3149, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "grad_norm": 26.094270706176758, | |
| "learning_rate": 2.1969246228460526e-07, | |
| "loss": 0.3316, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "grad_norm": 23.82662582397461, | |
| "learning_rate": 9.777396977174668e-08, | |
| "loss": 0.3519, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "grad_norm": 23.638530731201172, | |
| "learning_rate": 2.4463441107965278e-08, | |
| "loss": 0.34, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 30.666988372802734, | |
| "learning_rate": 0.0, | |
| "loss": 0.3462, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.9244532803180915, | |
| "eval_loss": 0.2655230164527893, | |
| "eval_runtime": 70.9285, | |
| "eval_samples_per_second": 35.458, | |
| "eval_steps_per_second": 1.114, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 5500, | |
| "total_flos": 1.301428412334932e+19, | |
| "train_loss": 0.5498046708540483, | |
| "train_runtime": 10510.7539, | |
| "train_samples_per_second": 16.727, | |
| "train_steps_per_second": 0.523 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 5500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 1.301428412334932e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |