| { | |
| "best_metric": 0.26184722781181335, | |
| "best_model_checkpoint": "./convnext-nano-new-1e-4/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": 19.579008102416992, | |
| "learning_rate": 9.991845519630678e-05, | |
| "loss": 2.3491, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 31.21416664123535, | |
| "learning_rate": 9.967408676742751e-05, | |
| "loss": 1.2966, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 15.68994140625, | |
| "learning_rate": 9.926769179238466e-05, | |
| "loss": 1.0475, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 17.57521629333496, | |
| "learning_rate": 9.870059584711668e-05, | |
| "loss": 0.9611, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 19.856332778930664, | |
| "learning_rate": 9.797464868072488e-05, | |
| "loss": 0.8703, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8473161033797216, | |
| "eval_loss": 0.5358882546424866, | |
| "eval_runtime": 60.8074, | |
| "eval_samples_per_second": 41.36, | |
| "eval_steps_per_second": 1.299, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 17.288860321044922, | |
| "learning_rate": 9.709221818197624e-05, | |
| "loss": 0.8327, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 15.169794082641602, | |
| "learning_rate": 9.60561826557425e-05, | |
| "loss": 0.7284, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 14.436149597167969, | |
| "learning_rate": 9.486992143456792e-05, | |
| "loss": 0.6652, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "grad_norm": 21.993471145629883, | |
| "learning_rate": 9.353730385598887e-05, | |
| "loss": 0.7387, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 17.582576751708984, | |
| "learning_rate": 9.206267664155907e-05, | |
| "loss": 0.678, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 23.1961727142334, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.6685, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8854870775347913, | |
| "eval_loss": 0.41691645979881287, | |
| "eval_runtime": 60.6738, | |
| "eval_samples_per_second": 41.451, | |
| "eval_steps_per_second": 1.302, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 13.541474342346191, | |
| "learning_rate": 8.870708053195413e-05, | |
| "loss": 0.5784, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 14.434982299804688, | |
| "learning_rate": 8.683705689382024e-05, | |
| "loss": 0.6053, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 9.445209503173828, | |
| "learning_rate": 8.484687843276469e-05, | |
| "loss": 0.5885, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 18.613405227661133, | |
| "learning_rate": 8.274303669726426e-05, | |
| "loss": 0.5734, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 16.23317527770996, | |
| "learning_rate": 8.053239398177191e-05, | |
| "loss": 0.5564, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9041749502982107, | |
| "eval_loss": 0.3498544692993164, | |
| "eval_runtime": 61.3833, | |
| "eval_samples_per_second": 40.972, | |
| "eval_steps_per_second": 1.287, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 17.68219566345215, | |
| "learning_rate": 7.822216094333847e-05, | |
| "loss": 0.5309, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "grad_norm": 13.154298782348633, | |
| "learning_rate": 7.58198730819481e-05, | |
| "loss": 0.5068, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "grad_norm": 12.262152671813965, | |
| "learning_rate": 7.333336616128369e-05, | |
| "loss": 0.4878, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 13.857909202575684, | |
| "learning_rate": 7.077075065009433e-05, | |
| "loss": 0.4992, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "grad_norm": 11.467909812927246, | |
| "learning_rate": 6.814038526753205e-05, | |
| "loss": 0.4905, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 28.545787811279297, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.4515, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9141153081510934, | |
| "eval_loss": 0.31653931736946106, | |
| "eval_runtime": 60.4281, | |
| "eval_samples_per_second": 41.62, | |
| "eval_steps_per_second": 1.307, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "grad_norm": 15.704607963562012, | |
| "learning_rate": 6.271091670967436e-05, | |
| "loss": 0.4485, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "grad_norm": 15.504982948303223, | |
| "learning_rate": 5.992952333228728e-05, | |
| "loss": 0.4265, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "grad_norm": 10.627553939819336, | |
| "learning_rate": 5.7115741913664264e-05, | |
| "loss": 0.4132, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "grad_norm": 14.083510398864746, | |
| "learning_rate": 5.427875042394199e-05, | |
| "loss": 0.4428, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "grad_norm": 7.471627235412598, | |
| "learning_rate": 5.142780253968481e-05, | |
| "loss": 0.442, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9081510934393638, | |
| "eval_loss": 0.3228101134300232, | |
| "eval_runtime": 60.6673, | |
| "eval_samples_per_second": 41.456, | |
| "eval_steps_per_second": 1.302, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "grad_norm": 14.514873504638672, | |
| "learning_rate": 4.85721974603152e-05, | |
| "loss": 0.383, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "grad_norm": 10.791834831237793, | |
| "learning_rate": 4.5721249576058027e-05, | |
| "loss": 0.39, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "grad_norm": 11.45054817199707, | |
| "learning_rate": 4.288425808633575e-05, | |
| "loss": 0.3562, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 12.328774452209473, | |
| "learning_rate": 4.007047666771274e-05, | |
| "loss": 0.3457, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "grad_norm": 14.327699661254883, | |
| "learning_rate": 3.728908329032567e-05, | |
| "loss": 0.3893, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 16.291210174560547, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.3799, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9157057654075547, | |
| "eval_loss": 0.3088889420032501, | |
| "eval_runtime": 61.3966, | |
| "eval_samples_per_second": 40.963, | |
| "eval_steps_per_second": 1.287, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "grad_norm": 2.8535711765289307, | |
| "learning_rate": 3.1859614732467954e-05, | |
| "loss": 0.3477, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "grad_norm": 13.33967399597168, | |
| "learning_rate": 2.9229249349905684e-05, | |
| "loss": 0.3299, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "grad_norm": 14.240943908691406, | |
| "learning_rate": 2.6666633838716314e-05, | |
| "loss": 0.3453, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "grad_norm": 13.373929023742676, | |
| "learning_rate": 2.418012691805191e-05, | |
| "loss": 0.313, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "grad_norm": 8.548375129699707, | |
| "learning_rate": 2.1777839056661554e-05, | |
| "loss": 0.3311, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.925248508946322, | |
| "eval_loss": 0.2745630145072937, | |
| "eval_runtime": 60.3297, | |
| "eval_samples_per_second": 41.688, | |
| "eval_steps_per_second": 1.309, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "grad_norm": 18.050865173339844, | |
| "learning_rate": 1.946760601822809e-05, | |
| "loss": 0.2854, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "grad_norm": 13.02624225616455, | |
| "learning_rate": 1.725696330273575e-05, | |
| "loss": 0.2972, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "grad_norm": 13.168692588806152, | |
| "learning_rate": 1.5153121567235335e-05, | |
| "loss": 0.3127, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "grad_norm": 12.672035217285156, | |
| "learning_rate": 1.3162943106179749e-05, | |
| "loss": 0.2909, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "grad_norm": 14.670820236206055, | |
| "learning_rate": 1.1292919468045877e-05, | |
| "loss": 0.2881, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 25.280366897583008, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.2726, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9248508946322067, | |
| "eval_loss": 0.26891395449638367, | |
| "eval_runtime": 60.3039, | |
| "eval_samples_per_second": 41.705, | |
| "eval_steps_per_second": 1.31, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "grad_norm": 16.034042358398438, | |
| "learning_rate": 7.937323358440935e-06, | |
| "loss": 0.2687, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "grad_norm": 11.272032737731934, | |
| "learning_rate": 6.462696144011149e-06, | |
| "loss": 0.2633, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "grad_norm": 12.020566940307617, | |
| "learning_rate": 5.13007856543209e-06, | |
| "loss": 0.2478, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "grad_norm": 15.785858154296875, | |
| "learning_rate": 3.9438173442575e-06, | |
| "loss": 0.2792, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "grad_norm": 8.982597351074219, | |
| "learning_rate": 2.9077818180237693e-06, | |
| "loss": 0.2711, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.927634194831014, | |
| "eval_loss": 0.2651021480560303, | |
| "eval_runtime": 61.2169, | |
| "eval_samples_per_second": 41.083, | |
| "eval_steps_per_second": 1.29, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "grad_norm": 11.63708209991455, | |
| "learning_rate": 2.0253513192751373e-06, | |
| "loss": 0.246, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "grad_norm": 9.835166931152344, | |
| "learning_rate": 1.2994041528833266e-06, | |
| "loss": 0.2542, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "grad_norm": 4.863436222076416, | |
| "learning_rate": 7.323082076153509e-07, | |
| "loss": 0.264, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "grad_norm": 15.75218677520752, | |
| "learning_rate": 3.2591323257248893e-07, | |
| "loss": 0.2809, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "grad_norm": 15.889384269714355, | |
| "learning_rate": 8.15448036932176e-08, | |
| "loss": 0.2674, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 48.06371307373047, | |
| "learning_rate": 0.0, | |
| "loss": 0.2758, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.9308151093439364, | |
| "eval_loss": 0.26184722781181335, | |
| "eval_runtime": 60.5892, | |
| "eval_samples_per_second": 41.509, | |
| "eval_steps_per_second": 1.304, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 5500, | |
| "total_flos": 7.000491898906214e+18, | |
| "train_loss": 0.49915996170043947, | |
| "train_runtime": 8370.5251, | |
| "train_samples_per_second": 21.003, | |
| "train_steps_per_second": 0.657 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 5500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 7.000491898906214e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |