| { | |
| "best_metric": 87.64485843737634, | |
| "best_model_checkpoint": "/home/ccasimiro/ccasimiro/berta/src/qa/roberta-base-ca-v2-qa/catalanqa/checkpoint-1134", | |
| "epoch": 10.0, | |
| "global_step": 5670, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.955908289241622e-05, | |
| "loss": 5.04, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9118165784832456e-05, | |
| "loss": 3.614, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.8677248677248676e-05, | |
| "loss": 3.1231, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.82363315696649e-05, | |
| "loss": 2.7369, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.779541446208113e-05, | |
| "loss": 2.3199, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.7354497354497356e-05, | |
| "loss": 2.0626, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.691358024691358e-05, | |
| "loss": 1.8434, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.647266313932981e-05, | |
| "loss": 1.5806, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.603174603174603e-05, | |
| "loss": 1.4498, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.559082892416226e-05, | |
| "loss": 1.2956, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.5149911816578484e-05, | |
| "loss": 1.1497, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 69.11247803163445, | |
| "eval_f1": 85.23057847114657, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.470899470899471e-05, | |
| "loss": 0.9974, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.426807760141094e-05, | |
| "loss": 0.9077, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.3827160493827164e-05, | |
| "loss": 0.8397, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.3386243386243384e-05, | |
| "loss": 0.7758, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.294532627865962e-05, | |
| "loss": 0.7537, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.250440917107584e-05, | |
| "loss": 0.7319, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.2063492063492065e-05, | |
| "loss": 0.7263, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.162257495590829e-05, | |
| "loss": 0.7032, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.118165784832452e-05, | |
| "loss": 0.6714, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.074074074074074e-05, | |
| "loss": 0.6854, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.029982363315697e-05, | |
| "loss": 0.6446, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 73.85764499121265, | |
| "eval_f1": 87.64485843737634, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.985890652557319e-05, | |
| "loss": 0.588, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.941798941798942e-05, | |
| "loss": 0.4666, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.8977072310405645e-05, | |
| "loss": 0.4807, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.853615520282187e-05, | |
| "loss": 0.4549, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.809523809523809e-05, | |
| "loss": 0.4482, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.7654320987654326e-05, | |
| "loss": 0.4739, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.7213403880070546e-05, | |
| "loss": 0.4336, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.677248677248677e-05, | |
| "loss": 0.3979, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 3.6331569664903e-05, | |
| "loss": 0.5031, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.5890652557319226e-05, | |
| "loss": 0.4193, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.5449735449735446e-05, | |
| "loss": 0.4714, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.500881834215168e-05, | |
| "loss": 0.416, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_exact_match": 73.94551845342707, | |
| "eval_f1": 87.39678260774708, | |
| "step": 1701 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.45679012345679e-05, | |
| "loss": 0.2783, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.412698412698413e-05, | |
| "loss": 0.271, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.3686067019400353e-05, | |
| "loss": 0.2807, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.324514991181658e-05, | |
| "loss": 0.3413, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.280423280423281e-05, | |
| "loss": 0.3236, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.2363315696649034e-05, | |
| "loss": 0.3034, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 3.1922398589065254e-05, | |
| "loss": 0.3067, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.148148148148148e-05, | |
| "loss": 0.2978, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 3.104056437389771e-05, | |
| "loss": 0.3084, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 3.0599647266313934e-05, | |
| "loss": 0.3331, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.0158730158730158e-05, | |
| "loss": 0.3132, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_exact_match": 73.59402460456943, | |
| "eval_f1": 87.53281609001948, | |
| "step": 2268 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 2.9717813051146388e-05, | |
| "loss": 0.2315, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.927689594356261e-05, | |
| "loss": 0.2073, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 2.8835978835978838e-05, | |
| "loss": 0.1865, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 2.839506172839506e-05, | |
| "loss": 0.1868, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 2.795414462081129e-05, | |
| "loss": 0.2252, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.7513227513227512e-05, | |
| "loss": 0.2252, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.7072310405643742e-05, | |
| "loss": 0.198, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.6631393298059965e-05, | |
| "loss": 0.2137, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.6190476190476192e-05, | |
| "loss": 0.2006, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.5749559082892416e-05, | |
| "loss": 0.2478, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 2.5308641975308646e-05, | |
| "loss": 0.2094, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_exact_match": 72.84710017574693, | |
| "eval_f1": 87.02126420060829, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.4867724867724866e-05, | |
| "loss": 0.1825, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.4426807760141093e-05, | |
| "loss": 0.1354, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.398589065255732e-05, | |
| "loss": 0.1383, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.3544973544973546e-05, | |
| "loss": 0.1417, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 2.310405643738977e-05, | |
| "loss": 0.174, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 2.2663139329805996e-05, | |
| "loss": 0.1297, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 0.1254, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 2.1781305114638447e-05, | |
| "loss": 0.1467, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 2.1340388007054674e-05, | |
| "loss": 0.1429, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.08994708994709e-05, | |
| "loss": 0.1401, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.0458553791887124e-05, | |
| "loss": 0.1309, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.001763668430335e-05, | |
| "loss": 0.1286, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_exact_match": 73.76977152899825, | |
| "eval_f1": 87.38988637413004, | |
| "step": 3402 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.9576719576719577e-05, | |
| "loss": 0.102, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 1.91358024691358e-05, | |
| "loss": 0.1046, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 1.8694885361552028e-05, | |
| "loss": 0.0967, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 1.8253968253968254e-05, | |
| "loss": 0.1127, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 1.781305114638448e-05, | |
| "loss": 0.093, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 1.7372134038800705e-05, | |
| "loss": 0.1085, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 1.693121693121693e-05, | |
| "loss": 0.0913, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.6490299823633158e-05, | |
| "loss": 0.1008, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 1.604938271604938e-05, | |
| "loss": 0.1191, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 1.560846560846561e-05, | |
| "loss": 0.095, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 1.5167548500881834e-05, | |
| "loss": 0.1145, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_exact_match": 73.94551845342707, | |
| "eval_f1": 87.51076671024968, | |
| "step": 3969 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.472663139329806e-05, | |
| "loss": 0.0907, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 1.4285714285714285e-05, | |
| "loss": 0.0702, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 1.384479717813051e-05, | |
| "loss": 0.0683, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 1.3403880070546737e-05, | |
| "loss": 0.0877, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 1.2962962962962962e-05, | |
| "loss": 0.085, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 1.252204585537919e-05, | |
| "loss": 0.0673, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 1.2081128747795414e-05, | |
| "loss": 0.0664, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 1.164021164021164e-05, | |
| "loss": 0.0984, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 1.1199294532627866e-05, | |
| "loss": 0.0635, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 1.0758377425044091e-05, | |
| "loss": 0.0866, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 1.0317460317460318e-05, | |
| "loss": 0.0744, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_exact_match": 73.85764499121265, | |
| "eval_f1": 87.23571814342841, | |
| "step": 4536 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 9.876543209876543e-06, | |
| "loss": 0.0788, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 9.435626102292768e-06, | |
| "loss": 0.0556, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 8.994708994708995e-06, | |
| "loss": 0.0415, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 8.55379188712522e-06, | |
| "loss": 0.052, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 8.112874779541445e-06, | |
| "loss": 0.0636, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 7.671957671957672e-06, | |
| "loss": 0.0481, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 7.231040564373897e-06, | |
| "loss": 0.0545, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 6.790123456790123e-06, | |
| "loss": 0.0597, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 6.349206349206349e-06, | |
| "loss": 0.0437, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 5.908289241622575e-06, | |
| "loss": 0.0722, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 5.4673721340388e-06, | |
| "loss": 0.062, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 5.026455026455026e-06, | |
| "loss": 0.0719, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_exact_match": 74.07732864674868, | |
| "eval_f1": 87.42888206161075, | |
| "step": 5103 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 4.585537918871252e-06, | |
| "loss": 0.0346, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 9.17, | |
| "learning_rate": 4.144620811287478e-06, | |
| "loss": 0.0373, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 3.7037037037037037e-06, | |
| "loss": 0.0333, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 3.2627865961199293e-06, | |
| "loss": 0.0384, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.8218694885361552e-06, | |
| "loss": 0.0413, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 2.3809523809523808e-06, | |
| "loss": 0.0511, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 1.9400352733686067e-06, | |
| "loss": 0.0453, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 1.4991181657848325e-06, | |
| "loss": 0.0422, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 1.0582010582010582e-06, | |
| "loss": 0.0572, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 6.17283950617284e-07, | |
| "loss": 0.0494, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 1.763668430335097e-07, | |
| "loss": 0.0444, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_exact_match": 74.34094903339192, | |
| "eval_f1": 87.50452307749782, | |
| "step": 5670 | |
| } | |
| ], | |
| "max_steps": 5670, | |
| "num_train_epochs": 10, | |
| "total_flos": 6912296207806464.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |