|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.1204393023553411, |
|
"global_step": 5000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 9.9e-06, |
|
"loss": 3.2945, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 9.800000000000001e-06, |
|
"loss": 3.1789, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 9.7e-06, |
|
"loss": 3.2237, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 9.600000000000001e-06, |
|
"loss": 3.2381, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.5e-06, |
|
"loss": 3.2048, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.4e-06, |
|
"loss": 3.2129, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.3e-06, |
|
"loss": 3.218, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.200000000000002e-06, |
|
"loss": 3.1789, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.100000000000001e-06, |
|
"loss": 3.2067, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9e-06, |
|
"loss": 3.1704, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 8.900000000000001e-06, |
|
"loss": 3.1543, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 8.8e-06, |
|
"loss": 3.1896, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.700000000000001e-06, |
|
"loss": 3.1658, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.6e-06, |
|
"loss": 3.2246, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.5e-06, |
|
"loss": 3.1984, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.400000000000001e-06, |
|
"loss": 3.1704, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.3e-06, |
|
"loss": 3.2175, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.2e-06, |
|
"loss": 3.1687, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.1e-06, |
|
"loss": 3.1709, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 3.1268, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.9e-06, |
|
"loss": 3.1634, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.800000000000002e-06, |
|
"loss": 3.0684, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.7e-06, |
|
"loss": 3.1026, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.600000000000001e-06, |
|
"loss": 3.0981, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 3.1482, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.4e-06, |
|
"loss": 3.0771, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.3e-06, |
|
"loss": 3.1484, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.2000000000000005e-06, |
|
"loss": 3.0779, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 7.100000000000001e-06, |
|
"loss": 3.1395, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 7e-06, |
|
"loss": 3.1112, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.9e-06, |
|
"loss": 3.087, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.800000000000001e-06, |
|
"loss": 3.1027, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.700000000000001e-06, |
|
"loss": 3.0695, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.600000000000001e-06, |
|
"loss": 3.1138, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.5000000000000004e-06, |
|
"loss": 3.1111, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.4000000000000006e-06, |
|
"loss": 3.0658, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.300000000000001e-06, |
|
"loss": 3.0523, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 6.200000000000001e-06, |
|
"loss": 3.0924, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 6.1e-06, |
|
"loss": 3.0793, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 6e-06, |
|
"loss": 3.0692, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 5.9e-06, |
|
"loss": 3.1456, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 5.8e-06, |
|
"loss": 3.0418, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 5.7e-06, |
|
"loss": 3.0855, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 5.600000000000001e-06, |
|
"loss": 3.074, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 5.500000000000001e-06, |
|
"loss": 3.0617, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.400000000000001e-06, |
|
"loss": 3.0639, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.300000000000001e-06, |
|
"loss": 3.0636, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.2e-06, |
|
"loss": 2.9724, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.1e-06, |
|
"loss": 3.0637, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5e-06, |
|
"loss": 3.0511, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.9000000000000005e-06, |
|
"loss": 3.0514, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.800000000000001e-06, |
|
"loss": 3.0162, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.7e-06, |
|
"loss": 3.0779, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.600000000000001e-06, |
|
"loss": 3.031, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.5e-06, |
|
"loss": 3.092, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.4e-06, |
|
"loss": 3.0638, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.3e-06, |
|
"loss": 3.0483, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.2000000000000004e-06, |
|
"loss": 2.992, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.1e-06, |
|
"loss": 3.0777, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 3.0417, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 3.900000000000001e-06, |
|
"loss": 2.976, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 3.8000000000000005e-06, |
|
"loss": 3.0749, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.7e-06, |
|
"loss": 3.085, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.6000000000000003e-06, |
|
"loss": 3.0375, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.5e-06, |
|
"loss": 3.0349, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.4000000000000005e-06, |
|
"loss": 3.0532, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.3000000000000006e-06, |
|
"loss": 3.1154, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.2000000000000003e-06, |
|
"loss": 2.9963, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.1000000000000004e-06, |
|
"loss": 3.0137, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3e-06, |
|
"loss": 3.0581, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.9e-06, |
|
"loss": 2.9854, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.8000000000000003e-06, |
|
"loss": 2.9893, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.7000000000000004e-06, |
|
"loss": 2.9317, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.6e-06, |
|
"loss": 3.0318, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.5e-06, |
|
"loss": 3.0336, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 3.0419, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.3000000000000004e-06, |
|
"loss": 3.0265, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.2e-06, |
|
"loss": 2.9753, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 2.1000000000000002e-06, |
|
"loss": 3.021, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"loss": 3.0237, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9000000000000002e-06, |
|
"loss": 2.9938, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.8000000000000001e-06, |
|
"loss": 3.0129, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.7000000000000002e-06, |
|
"loss": 2.9862, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.6000000000000001e-06, |
|
"loss": 3.041, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.5e-06, |
|
"loss": 3.0256, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.4000000000000001e-06, |
|
"loss": 2.9721, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.3e-06, |
|
"loss": 3.0076, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.2000000000000002e-06, |
|
"loss": 3.0082, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.1e-06, |
|
"loss": 3.0354, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"loss": 2.9866, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.000000000000001e-07, |
|
"loss": 3.0599, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 8.000000000000001e-07, |
|
"loss": 2.9539, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 7.000000000000001e-07, |
|
"loss": 2.96, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 6.000000000000001e-07, |
|
"loss": 3.0198, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 5.000000000000001e-07, |
|
"loss": 3.0299, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 4.0000000000000003e-07, |
|
"loss": 2.9731, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 3.0000000000000004e-07, |
|
"loss": 3.0531, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 2.0000000000000002e-07, |
|
"loss": 2.9388, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.0000000000000001e-07, |
|
"loss": 3.0682, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 0.0, |
|
"loss": 2.9654, |
|
"step": 5000 |
|
} |
|
], |
|
"max_steps": 5000, |
|
"num_train_epochs": 1, |
|
"total_flos": 1.38662884933632e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|