| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.999281732051852, | |
| "global_step": 4565, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.5e-06, | |
| "loss": 2.2671, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3e-06, | |
| "loss": 1.8863, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.8591, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 6e-06, | |
| "loss": 1.7121, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.5e-06, | |
| "loss": 1.6481, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9e-06, | |
| "loss": 1.606, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.05e-05, | |
| "loss": 1.6154, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.5695, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 1.4993, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.4618, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.65e-05, | |
| "loss": 1.4893, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.4824, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.95e-05, | |
| "loss": 1.4306, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.4697, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.25e-05, | |
| "loss": 1.4275, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.4353, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.55e-05, | |
| "loss": 1.4811, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.4272, | |
| "step": 900 | |
| }, | |
| { | |
| "HasAns_exact": 58.11939375193319, | |
| "HasAns_f1": 72.62489483900757, | |
| "HasAns_total": 3233, | |
| "NoAns_exact": 76.07361963190183, | |
| "NoAns_f1": 76.07361963190183, | |
| "NoAns_total": 163, | |
| "best_exact": 60.129564193168434, | |
| "best_exact_thresh": 0.0, | |
| "best_f1": 73.93883539885493, | |
| "best_f1_thresh": 0.0, | |
| "epoch": 1.0, | |
| "exact": 58.9811542991755, | |
| "f1": 72.79042550486203, | |
| "step": 913, | |
| "total": 3396 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 1.168, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3e-05, | |
| "loss": 1.1045, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.9579242636746143e-05, | |
| "loss": 1.1772, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.9158485273492286e-05, | |
| "loss": 1.1478, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.873772791023843e-05, | |
| "loss": 1.2113, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.831697054698457e-05, | |
| "loss": 1.2509, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7896213183730714e-05, | |
| "loss": 1.1653, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.747545582047686e-05, | |
| "loss": 1.1098, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.7054698457223e-05, | |
| "loss": 1.1876, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.6633941093969145e-05, | |
| "loss": 1.1803, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.6213183730715288e-05, | |
| "loss": 1.1392, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.579242636746143e-05, | |
| "loss": 1.158, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.5371669004207573e-05, | |
| "loss": 1.0864, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.4950911640953716e-05, | |
| "loss": 1.2314, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.453015427769986e-05, | |
| "loss": 1.174, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.4109396914446005e-05, | |
| "loss": 1.1833, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.3688639551192144e-05, | |
| "loss": 1.1636, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.326788218793829e-05, | |
| "loss": 1.1598, | |
| "step": 1800 | |
| }, | |
| { | |
| "HasAns_exact": 60.19177234766471, | |
| "HasAns_f1": 75.40873260282125, | |
| "HasAns_total": 3233, | |
| "NoAns_exact": 52.147239263803684, | |
| "NoAns_f1": 52.147239263803684, | |
| "NoAns_total": 163, | |
| "best_exact": 62.102473498233216, | |
| "best_exact_thresh": 0.0, | |
| "best_f1": 76.58905550792727, | |
| "best_f1_thresh": 0.0, | |
| "epoch": 2.0, | |
| "exact": 59.80565371024735, | |
| "f1": 74.29223571994143, | |
| "step": 1826, | |
| "total": 3396 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.2847124824684433e-05, | |
| "loss": 0.962, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.2426367461430575e-05, | |
| "loss": 0.7194, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.2005610098176718e-05, | |
| "loss": 0.7277, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.158485273492286e-05, | |
| "loss": 0.6882, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.1164095371669003e-05, | |
| "loss": 0.6856, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.074333800841515e-05, | |
| "loss": 0.7305, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.032258064516129e-05, | |
| "loss": 0.7234, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1.9901823281907435e-05, | |
| "loss": 0.6732, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1.9481065918653574e-05, | |
| "loss": 0.7105, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1.906030855539972e-05, | |
| "loss": 0.7333, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1.8639551192145863e-05, | |
| "loss": 0.6969, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1.8218793828892005e-05, | |
| "loss": 0.7382, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1.7798036465638148e-05, | |
| "loss": 0.7801, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.7377279102384294e-05, | |
| "loss": 0.7708, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.6956521739130433e-05, | |
| "loss": 0.7032, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.653576437587658e-05, | |
| "loss": 0.7041, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.611500701262272e-05, | |
| "loss": 0.7627, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.5694249649368865e-05, | |
| "loss": 0.7428, | |
| "step": 2700 | |
| }, | |
| { | |
| "HasAns_exact": 59.44942777605939, | |
| "HasAns_f1": 75.24381473873625, | |
| "HasAns_total": 3233, | |
| "NoAns_exact": 56.441717791411044, | |
| "NoAns_f1": 56.441717791411044, | |
| "NoAns_total": 163, | |
| "best_exact": 61.39575971731449, | |
| "best_exact_thresh": 0.0, | |
| "best_f1": 76.43205331281922, | |
| "best_f1_thresh": 0.0, | |
| "epoch": 3.0, | |
| "exact": 59.305064782096586, | |
| "f1": 74.34135837760138, | |
| "step": 2739, | |
| "total": 3396 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.5273492286115007e-05, | |
| "loss": 0.6496, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.4852734922861152e-05, | |
| "loss": 0.4199, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.4431977559607294e-05, | |
| "loss": 0.4245, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.4011220196353437e-05, | |
| "loss": 0.4192, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.359046283309958e-05, | |
| "loss": 0.4115, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.3169705469845722e-05, | |
| "loss": 0.4392, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.2748948106591867e-05, | |
| "loss": 0.3749, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.232819074333801e-05, | |
| "loss": 0.4264, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.1907433380084152e-05, | |
| "loss": 0.4127, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 1.1486676016830295e-05, | |
| "loss": 0.4517, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 1.1065918653576439e-05, | |
| "loss": 0.4202, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 1.0645161290322582e-05, | |
| "loss": 0.413, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.0224403927068724e-05, | |
| "loss": 0.4274, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 9.803646563814867e-06, | |
| "loss": 0.3953, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 9.382889200561011e-06, | |
| "loss": 0.4115, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 8.962131837307154e-06, | |
| "loss": 0.4382, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 8.541374474053297e-06, | |
| "loss": 0.3753, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 8.12061711079944e-06, | |
| "loss": 0.4231, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 7.699859747545584e-06, | |
| "loss": 0.4801, | |
| "step": 3650 | |
| }, | |
| { | |
| "HasAns_exact": 59.48035879987628, | |
| "HasAns_f1": 75.10338801562317, | |
| "HasAns_total": 3233, | |
| "NoAns_exact": 65.03067484662577, | |
| "NoAns_f1": 65.03067484662577, | |
| "NoAns_total": 163, | |
| "best_exact": 61.42520612485277, | |
| "best_exact_thresh": 0.0, | |
| "best_f1": 76.29836674161058, | |
| "best_f1_thresh": 0.0, | |
| "epoch": 4.0, | |
| "exact": 59.74676089517079, | |
| "f1": 74.61992151192865, | |
| "step": 3652, | |
| "total": 3396 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 7.2791023842917254e-06, | |
| "loss": 0.2907, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 6.858345021037868e-06, | |
| "loss": 0.2268, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 6.437587657784012e-06, | |
| "loss": 0.2669, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 6.016830294530154e-06, | |
| "loss": 0.2185, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 5.596072931276297e-06, | |
| "loss": 0.2734, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 5.17531556802244e-06, | |
| "loss": 0.2407, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 4.754558204768583e-06, | |
| "loss": 0.2524, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 4.3338008415147266e-06, | |
| "loss": 0.2545, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 3.913043478260869e-06, | |
| "loss": 0.2268, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 3.4922861150070127e-06, | |
| "loss": 0.2433, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.071528751753156e-06, | |
| "loss": 0.2661, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.650771388499299e-06, | |
| "loss": 0.2574, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.230014025245442e-06, | |
| "loss": 0.212, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 1.8092566619915848e-06, | |
| "loss": 0.2434, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.388499298737728e-06, | |
| "loss": 0.28, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 9.67741935483871e-07, | |
| "loss": 0.2294, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 5.46984572230014e-07, | |
| "loss": 0.2103, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.2622720897615708e-07, | |
| "loss": 0.2526, | |
| "step": 4550 | |
| }, | |
| { | |
| "HasAns_exact": 57.96473863284875, | |
| "HasAns_f1": 74.35908528391936, | |
| "HasAns_total": 3233, | |
| "NoAns_exact": 65.03067484662577, | |
| "NoAns_f1": 65.03067484662577, | |
| "NoAns_total": 163, | |
| "best_exact": 59.98233215547703, | |
| "best_exact_thresh": 0.0, | |
| "best_f1": 75.58978878766521, | |
| "best_f1_thresh": 0.0, | |
| "epoch": 5.0, | |
| "exact": 58.303886925795055, | |
| "f1": 73.91134355798329, | |
| "step": 4565, | |
| "total": 3396 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 4565, | |
| "total_flos": 1.5001363873057997e+17, | |
| "train_runtime": 22638.32, | |
| "train_samples_per_second": 0.202 | |
| } | |
| ], | |
| "max_steps": 4565, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.5001363873057997e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |