| { | |
| "best_metric": 88.70164904870342, | |
| "best_model_checkpoint": "output/ft_minilm_roberta_base_tokenizer/checkpoint-55000", | |
| "epoch": 4.9679342426158435, | |
| "global_step": 55000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.972902176858459e-05, | |
| "loss": 3.6092, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_exact_match": 38.34437086092715, | |
| "eval_f1": 53.04195171599385, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.9458043537169182e-05, | |
| "loss": 2.3309, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_exact_match": 57.48344370860927, | |
| "eval_f1": 69.86399789823338, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.9187065305753772e-05, | |
| "loss": 1.869, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_exact_match": 60.8325449385052, | |
| "eval_f1": 73.39213373938078, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.8916087074338363e-05, | |
| "loss": 1.714, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_exact_match": 67.55912961210974, | |
| "eval_f1": 78.46892865109527, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.864510884292295e-05, | |
| "loss": 1.596, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_exact_match": 69.00662251655629, | |
| "eval_f1": 79.38516098327496, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.8374130611507544e-05, | |
| "loss": 1.4757, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_exact_match": 70.37842951750237, | |
| "eval_f1": 80.6431787888757, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.8103152380092135e-05, | |
| "loss": 1.4583, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_exact_match": 70.35004730368969, | |
| "eval_f1": 80.79000685737387, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.783217414867672e-05, | |
| "loss": 1.409, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "eval_exact_match": 72.25165562913908, | |
| "eval_f1": 81.90341789407236, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.7561195917261312e-05, | |
| "loss": 1.3424, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_exact_match": 72.48817407757805, | |
| "eval_f1": 82.40953492685638, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7290217685845906e-05, | |
| "loss": 1.35, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_exact_match": 73.56669820245979, | |
| "eval_f1": 83.04642576372542, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.7019239454430493e-05, | |
| "loss": 1.3199, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_exact_match": 73.97350993377484, | |
| "eval_f1": 83.20596198058827, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.6748261223015084e-05, | |
| "loss": 1.2529, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "eval_exact_match": 74.53169347209082, | |
| "eval_f1": 83.5595619348799, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.6477282991599678e-05, | |
| "loss": 1.2637, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "eval_exact_match": 75.10879848628193, | |
| "eval_f1": 84.03364368528639, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.6206304760184265e-05, | |
| "loss": 1.2265, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_exact_match": 74.9763481551561, | |
| "eval_f1": 84.28762156566786, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.5935326528768855e-05, | |
| "loss": 1.2274, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "eval_exact_match": 74.78713339640493, | |
| "eval_f1": 84.24767651424737, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.5664348297353446e-05, | |
| "loss": 1.2305, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "eval_exact_match": 75.47776726584674, | |
| "eval_f1": 84.74489870588677, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.539337006593804e-05, | |
| "loss": 1.2175, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_exact_match": 76.7833491012299, | |
| "eval_f1": 85.271914926584, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.5122391834522627e-05, | |
| "loss": 1.182, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "eval_exact_match": 76.45222327341533, | |
| "eval_f1": 85.33283456122666, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.4851413603107218e-05, | |
| "loss": 1.1853, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "eval_exact_match": 76.96310312204352, | |
| "eval_f1": 85.2924473992702, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.4580435371691808e-05, | |
| "loss": 1.1664, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_exact_match": 76.04541154210028, | |
| "eval_f1": 85.02171266178281, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.43094571402764e-05, | |
| "loss": 1.1932, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "eval_exact_match": 75.78051087984863, | |
| "eval_f1": 85.14984769166524, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.403847890886099e-05, | |
| "loss": 1.1671, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_exact_match": 77.6158940397351, | |
| "eval_f1": 85.8681189259958, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.376750067744558e-05, | |
| "loss": 1.028, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_exact_match": 77.33207190160833, | |
| "eval_f1": 86.13476093529944, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.349652244603017e-05, | |
| "loss": 1.0269, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "eval_exact_match": 77.31315042573321, | |
| "eval_f1": 85.78642346203148, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.322554421461476e-05, | |
| "loss": 1.0276, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "eval_exact_match": 77.62535477767265, | |
| "eval_f1": 86.04542466345086, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 2.295456598319935e-05, | |
| "loss": 1.0251, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "eval_exact_match": 77.96594134342479, | |
| "eval_f1": 86.0039567640003, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.268358775178394e-05, | |
| "loss": 1.0153, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "eval_exact_match": 77.918637653737, | |
| "eval_f1": 85.94724245197462, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.2412609520368532e-05, | |
| "loss": 1.0273, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_exact_match": 77.88079470198676, | |
| "eval_f1": 86.43469515459539, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.2141631288953123e-05, | |
| "loss": 1.0184, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_exact_match": 78.30652790917692, | |
| "eval_f1": 86.49960884571237, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.187065305753771e-05, | |
| "loss": 1.0129, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_exact_match": 77.84295175023652, | |
| "eval_f1": 86.34551242938619, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.15996748261223e-05, | |
| "loss": 1.0298, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "eval_exact_match": 78.36329233680227, | |
| "eval_f1": 86.62513916601708, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.1328696594706894e-05, | |
| "loss": 0.9999, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_exact_match": 78.59981078524125, | |
| "eval_f1": 86.75304072732975, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.105771836329148e-05, | |
| "loss": 0.957, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "eval_exact_match": 78.69441816461683, | |
| "eval_f1": 86.9310767775865, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.0786740131876072e-05, | |
| "loss": 1.0023, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "eval_exact_match": 78.80794701986756, | |
| "eval_f1": 87.1131371682942, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.0515761900460666e-05, | |
| "loss": 0.999, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_exact_match": 78.69441816461683, | |
| "eval_f1": 86.85637778000766, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.0244783669045253e-05, | |
| "loss": 1.0011, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "eval_exact_match": 79.120151371807, | |
| "eval_f1": 87.31859279800312, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.9973805437629844e-05, | |
| "loss": 1.0053, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "eval_exact_match": 79.29990539262063, | |
| "eval_f1": 87.20927813197312, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.9702827206214434e-05, | |
| "loss": 0.9997, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "eval_exact_match": 79.0728476821192, | |
| "eval_f1": 87.14281225266565, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.9431848974799025e-05, | |
| "loss": 0.9993, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "eval_exact_match": 78.77010406811732, | |
| "eval_f1": 86.96515368656466, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9160870743383615e-05, | |
| "loss": 0.9712, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_exact_match": 78.86471144749291, | |
| "eval_f1": 87.139815910334, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.8889892511968206e-05, | |
| "loss": 0.9586, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "eval_exact_match": 79.60264900662251, | |
| "eval_f1": 87.67189897254278, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8618914280552793e-05, | |
| "loss": 0.9855, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "eval_exact_match": 79.29044465468307, | |
| "eval_f1": 87.35142110336871, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.8347936049137387e-05, | |
| "loss": 0.9577, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_exact_match": 79.71617786187322, | |
| "eval_f1": 87.58757029471191, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.8076957817721978e-05, | |
| "loss": 0.9764, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_exact_match": 79.45127719962157, | |
| "eval_f1": 87.47183032338337, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.7805979586306565e-05, | |
| "loss": 0.8665, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_exact_match": 79.66887417218543, | |
| "eval_f1": 87.64021806679038, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.753500135489116e-05, | |
| "loss": 0.8516, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_exact_match": 79.66887417218543, | |
| "eval_f1": 87.7551254398469, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.726402312347575e-05, | |
| "loss": 0.8475, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "eval_exact_match": 79.2336802270577, | |
| "eval_f1": 87.42734800717938, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.699304489206034e-05, | |
| "loss": 0.8682, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "eval_exact_match": 79.85808893093662, | |
| "eval_f1": 87.76673576734072, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.6722066660644927e-05, | |
| "loss": 0.885, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "eval_exact_match": 79.52696310312204, | |
| "eval_f1": 87.60864611417429, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.645108842922952e-05, | |
| "loss": 0.8821, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "eval_exact_match": 79.99053926206244, | |
| "eval_f1": 87.83466920921494, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.618011019781411e-05, | |
| "loss": 0.8379, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "eval_exact_match": 79.68779564806054, | |
| "eval_f1": 87.65155459405366, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.59091319663987e-05, | |
| "loss": 0.8406, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "eval_exact_match": 79.71617786187322, | |
| "eval_f1": 87.80269063903106, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.563815373498329e-05, | |
| "loss": 0.8817, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "eval_exact_match": 79.76348155156103, | |
| "eval_f1": 87.6863825089024, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.5367175503567883e-05, | |
| "loss": 0.8487, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "eval_exact_match": 79.87701040681173, | |
| "eval_f1": 87.85589778459244, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.509619727215247e-05, | |
| "loss": 0.8474, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "eval_exact_match": 80.13245033112582, | |
| "eval_f1": 87.88619033429767, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.482521904073706e-05, | |
| "loss": 0.844, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "eval_exact_match": 80.22705771050141, | |
| "eval_f1": 87.94538843278356, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1.4554240809321651e-05, | |
| "loss": 0.8165, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_exact_match": 80.1135288552507, | |
| "eval_f1": 87.94195235666041, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1.4283262577906243e-05, | |
| "loss": 0.8857, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "eval_exact_match": 80.3027436140019, | |
| "eval_f1": 88.06711014071337, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1.4012284346490832e-05, | |
| "loss": 0.8881, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "eval_exact_match": 79.46073793755913, | |
| "eval_f1": 87.63361243864426, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.3741306115075423e-05, | |
| "loss": 0.8478, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "eval_exact_match": 80.24597918637653, | |
| "eval_f1": 88.16985739641302, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.3470327883660013e-05, | |
| "loss": 0.8322, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "eval_exact_match": 80.13245033112582, | |
| "eval_f1": 87.86660396184342, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.3199349652244604e-05, | |
| "loss": 0.8407, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "eval_exact_match": 80.07568590350047, | |
| "eval_f1": 87.92805047246172, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.2928371420829193e-05, | |
| "loss": 0.8427, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "eval_exact_match": 80.17975402081362, | |
| "eval_f1": 88.0783939864499, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.2657393189413785e-05, | |
| "loss": 0.8334, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "eval_exact_match": 80.19867549668874, | |
| "eval_f1": 88.06310378279342, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.2386414957998374e-05, | |
| "loss": 0.8234, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "eval_exact_match": 80.22705771050141, | |
| "eval_f1": 88.08144574072385, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.2115436726582964e-05, | |
| "loss": 0.8344, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "eval_exact_match": 80.66225165562913, | |
| "eval_f1": 88.25754342761562, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 1.1844458495167555e-05, | |
| "loss": 0.7908, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "eval_exact_match": 80.3027436140019, | |
| "eval_f1": 88.2107177195419, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.1573480263752145e-05, | |
| "loss": 0.7268, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "eval_exact_match": 80.47303689687796, | |
| "eval_f1": 88.33146167313825, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.1302502032336736e-05, | |
| "loss": 0.7416, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "eval_exact_match": 80.33112582781457, | |
| "eval_f1": 88.20770283656012, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 1.1031523800921326e-05, | |
| "loss": 0.7289, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "eval_exact_match": 80.27436140018922, | |
| "eval_f1": 88.08675881507963, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.0760545569505917e-05, | |
| "loss": 0.7338, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "eval_exact_match": 80.32166508987702, | |
| "eval_f1": 88.24483276676646, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.0489567338090507e-05, | |
| "loss": 0.7992, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_exact_match": 80.48249763481552, | |
| "eval_f1": 88.24667332064766, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.0218589106675098e-05, | |
| "loss": 0.7497, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "eval_exact_match": 80.0, | |
| "eval_f1": 87.95371777369881, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 9.947610875259687e-06, | |
| "loss": 0.7389, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "eval_exact_match": 80.45411542100284, | |
| "eval_f1": 88.24729821889156, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 9.676632643844279e-06, | |
| "loss": 0.748, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "eval_exact_match": 80.49195837275307, | |
| "eval_f1": 88.27202955999378, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 9.405654412428868e-06, | |
| "loss": 0.7528, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "eval_exact_match": 80.59602649006622, | |
| "eval_f1": 88.27897205791908, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 9.134676181013458e-06, | |
| "loss": 0.7544, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "eval_exact_match": 80.71901608325449, | |
| "eval_f1": 88.29085294791585, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 8.863697949598049e-06, | |
| "loss": 0.7185, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "eval_exact_match": 80.77578051087986, | |
| "eval_f1": 88.45178084091113, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.59271971818264e-06, | |
| "loss": 0.711, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "eval_exact_match": 80.54872280037843, | |
| "eval_f1": 88.23006827695103, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 8.32174148676723e-06, | |
| "loss": 0.7767, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "eval_exact_match": 80.51087984862819, | |
| "eval_f1": 88.21518807166053, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 8.05076325535182e-06, | |
| "loss": 0.7299, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "eval_exact_match": 80.42573320719016, | |
| "eval_f1": 88.14657248237756, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 7.779785023936411e-06, | |
| "loss": 0.7443, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "eval_exact_match": 80.45411542100284, | |
| "eval_f1": 88.25882487493044, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 7.508806792521e-06, | |
| "loss": 0.7672, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "eval_exact_match": 80.79470198675497, | |
| "eval_f1": 88.35767687765708, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 7.237828561105591e-06, | |
| "loss": 0.7466, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "eval_exact_match": 80.81362346263009, | |
| "eval_f1": 88.49878983679895, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 6.966850329690182e-06, | |
| "loss": 0.7294, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "eval_exact_match": 80.66225165562913, | |
| "eval_f1": 88.4219719582272, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 6.695872098274772e-06, | |
| "loss": 0.7554, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "eval_exact_match": 80.80416272469253, | |
| "eval_f1": 88.36066468276921, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 6.424893866859363e-06, | |
| "loss": 0.761, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "eval_exact_match": 80.63386944181646, | |
| "eval_f1": 88.41315914502825, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.1539156354439526e-06, | |
| "loss": 0.7363, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "eval_exact_match": 80.65279091769158, | |
| "eval_f1": 88.34571302471203, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.882937404028543e-06, | |
| "loss": 0.7057, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "eval_exact_match": 80.91769157994324, | |
| "eval_f1": 88.47587754641178, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 5.611959172613134e-06, | |
| "loss": 0.6499, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "eval_exact_match": 80.9271523178808, | |
| "eval_f1": 88.48670524401668, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 5.340980941197724e-06, | |
| "loss": 0.6538, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "eval_exact_match": 80.89877010406812, | |
| "eval_f1": 88.44196466728391, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.070002709782314e-06, | |
| "loss": 0.6996, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "eval_exact_match": 80.98391674550615, | |
| "eval_f1": 88.49930789298939, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 4.799024478366904e-06, | |
| "loss": 0.6394, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "eval_exact_match": 80.6244087038789, | |
| "eval_f1": 88.34556116700644, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 4.528046246951495e-06, | |
| "loss": 0.6721, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "eval_exact_match": 80.83254493850521, | |
| "eval_f1": 88.44910098996408, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 4.257068015536085e-06, | |
| "loss": 0.6854, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "eval_exact_match": 80.86092715231788, | |
| "eval_f1": 88.53212448905681, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.986089784120676e-06, | |
| "loss": 0.6761, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "eval_exact_match": 80.6717123935667, | |
| "eval_f1": 88.4448096587991, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.715111552705266e-06, | |
| "loss": 0.6943, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "eval_exact_match": 80.52980132450331, | |
| "eval_f1": 88.33280893923866, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.444133321289856e-06, | |
| "loss": 0.6655, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "eval_exact_match": 80.93661305581836, | |
| "eval_f1": 88.56888817902892, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.173155089874447e-06, | |
| "loss": 0.7029, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "eval_exact_match": 80.84200567644277, | |
| "eval_f1": 88.51593698418615, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.9021768584590373e-06, | |
| "loss": 0.6648, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "eval_exact_match": 80.83254493850521, | |
| "eval_f1": 88.61534249704374, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.6311986270436274e-06, | |
| "loss": 0.6901, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "eval_exact_match": 80.94607379375591, | |
| "eval_f1": 88.66269781300855, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 2.360220395628218e-06, | |
| "loss": 0.6784, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "eval_exact_match": 80.879848628193, | |
| "eval_f1": 88.58225782143238, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.089242164212808e-06, | |
| "loss": 0.6543, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "eval_exact_match": 80.99337748344371, | |
| "eval_f1": 88.67206922431193, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 1.8182639327973986e-06, | |
| "loss": 0.6419, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "eval_exact_match": 80.97445600756859, | |
| "eval_f1": 88.6884615459273, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.5472857013819889e-06, | |
| "loss": 0.6617, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "eval_exact_match": 80.85146641438033, | |
| "eval_f1": 88.58266438928703, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 1.2763074699665794e-06, | |
| "loss": 0.7022, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "eval_exact_match": 80.89877010406812, | |
| "eval_f1": 88.62217290793005, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 1.0053292385511697e-06, | |
| "loss": 0.662, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "eval_exact_match": 80.96499526963103, | |
| "eval_f1": 88.5709480325892, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 7.343510071357602e-07, | |
| "loss": 0.6708, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "eval_exact_match": 80.83254493850521, | |
| "eval_f1": 88.61340923000925, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 4.6337277572035044e-07, | |
| "loss": 0.6504, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "eval_exact_match": 81.01229895931883, | |
| "eval_f1": 88.68117331158336, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 1.9239454430494084e-07, | |
| "loss": 0.6681, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "eval_exact_match": 81.03122043519394, | |
| "eval_f1": 88.70164904870342, | |
| "step": 55000 | |
| } | |
| ], | |
| "max_steps": 55355, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.079484291072e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |