| { | |
| "best_metric": 0.00048321104259230196, | |
| "best_model_checkpoint": "./home/mkapadni/scratch/finetuning_experiments/models/t5_large_finetuned_meetup_d1/checkpoint-22880", | |
| "epoch": 5.0, | |
| "global_step": 22880, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 60.4275, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 57.9583, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.2e-05, | |
| "loss": 54.114, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 46.0159, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2e-05, | |
| "loss": 38.0285, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9910634495084898e-05, | |
| "loss": 34.5486, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9821268990169794e-05, | |
| "loss": 32.3095, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9731903485254694e-05, | |
| "loss": 30.5876, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.964253798033959e-05, | |
| "loss": 28.5188, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9553172475424487e-05, | |
| "loss": 5.5749, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9463806970509383e-05, | |
| "loss": 0.1655, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9374441465594283e-05, | |
| "loss": 0.1157, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.928507596067918e-05, | |
| "loss": 0.1249, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9195710455764076e-05, | |
| "loss": 0.0992, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9106344950848975e-05, | |
| "loss": 0.0703, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.901697944593387e-05, | |
| "loss": 0.0694, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8927613941018768e-05, | |
| "loss": 0.0567, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.8838248436103664e-05, | |
| "loss": 0.0478, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8748882931188564e-05, | |
| "loss": 0.0412, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.865951742627346e-05, | |
| "loss": 0.0377, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.8570151921358357e-05, | |
| "loss": 0.0324, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.8480786416443253e-05, | |
| "loss": 0.0291, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.839142091152815e-05, | |
| "loss": 0.0243, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.830205540661305e-05, | |
| "loss": 0.0206, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.8212689901697946e-05, | |
| "loss": 0.0207, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.8123324396782845e-05, | |
| "loss": 0.0158, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.803395889186774e-05, | |
| "loss": 0.0201, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.7944593386952638e-05, | |
| "loss": 0.0182, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.7855227882037534e-05, | |
| "loss": 0.0131, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.776586237712243e-05, | |
| "loss": 0.0121, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.767649687220733e-05, | |
| "loss": 0.0116, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.7587131367292227e-05, | |
| "loss": 0.0132, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7497765862377123e-05, | |
| "loss": 0.0132, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.740840035746202e-05, | |
| "loss": 0.0105, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.731903485254692e-05, | |
| "loss": 0.0095, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.7229669347631816e-05, | |
| "loss": 0.0132, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7140303842716712e-05, | |
| "loss": 0.0105, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7050938337801612e-05, | |
| "loss": 0.0103, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6961572832886508e-05, | |
| "loss": 0.0075, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.6872207327971404e-05, | |
| "loss": 0.01, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.67828418230563e-05, | |
| "loss": 0.0106, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.66934763181412e-05, | |
| "loss": 0.0068, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.6604110813226097e-05, | |
| "loss": 0.0073, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.6514745308310993e-05, | |
| "loss": 0.007, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.642537980339589e-05, | |
| "loss": 0.0054, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.0016077366890385747, | |
| "eval_runtime": 582.2402, | |
| "eval_samples_per_second": 1.913, | |
| "eval_steps_per_second": 1.913, | |
| "step": 4576 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.6336014298480786e-05, | |
| "loss": 0.0069, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.6246648793565686e-05, | |
| "loss": 0.0035, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.6157283288650582e-05, | |
| "loss": 0.0054, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.6067917783735482e-05, | |
| "loss": 0.0054, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.5978552278820375e-05, | |
| "loss": 0.0027, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.5889186773905275e-05, | |
| "loss": 0.0047, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.579982126899017e-05, | |
| "loss": 0.0049, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.5710455764075067e-05, | |
| "loss": 0.0033, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.5621090259159967e-05, | |
| "loss": 0.0044, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.5531724754244863e-05, | |
| "loss": 0.0037, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.544235924932976e-05, | |
| "loss": 0.0034, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.5352993744414656e-05, | |
| "loss": 0.0042, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.5263628239499556e-05, | |
| "loss": 0.0045, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.517426273458445e-05, | |
| "loss": 0.0036, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.5084897229669349e-05, | |
| "loss": 0.004, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.4995531724754247e-05, | |
| "loss": 0.0045, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.4906166219839145e-05, | |
| "loss": 0.0031, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.481680071492404e-05, | |
| "loss": 0.0028, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.4727435210008937e-05, | |
| "loss": 0.003, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4638069705093835e-05, | |
| "loss": 0.0035, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4548704200178732e-05, | |
| "loss": 0.0028, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.445933869526363e-05, | |
| "loss": 0.0036, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4369973190348528e-05, | |
| "loss": 0.0047, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4280607685433422e-05, | |
| "loss": 0.0016, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.419124218051832e-05, | |
| "loss": 0.0034, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4101876675603219e-05, | |
| "loss": 0.0037, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.4012511170688115e-05, | |
| "loss": 8.1941, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3923145665773013e-05, | |
| "loss": 28.1153, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3833780160857911e-05, | |
| "loss": 34.0805, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3744414655942809e-05, | |
| "loss": 34.9717, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3655049151027704e-05, | |
| "loss": 33.9775, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3565683646112602e-05, | |
| "loss": 30.7487, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.34763181411975e-05, | |
| "loss": 15.3977, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.3386952636282396e-05, | |
| "loss": 1.1498, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.3297587131367293e-05, | |
| "loss": 0.1552, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.320822162645219e-05, | |
| "loss": 0.0114, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.3118856121537087e-05, | |
| "loss": 0.0909, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.3029490616621985e-05, | |
| "loss": 0.0104, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.2940125111706883e-05, | |
| "loss": 0.0026, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.2850759606791778e-05, | |
| "loss": 0.0028, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.2761394101876676e-05, | |
| "loss": 0.0026, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2672028596961574e-05, | |
| "loss": 0.0026, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.2582663092046472e-05, | |
| "loss": 0.0018, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.2493297587131368e-05, | |
| "loss": 0.0023, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.2403932082216266e-05, | |
| "loss": 0.1979, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.2314566577301164e-05, | |
| "loss": 0.0038, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.0005774666788056493, | |
| "eval_runtime": 582.3329, | |
| "eval_samples_per_second": 1.913, | |
| "eval_steps_per_second": 1.913, | |
| "step": 9152 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.2225201072386059e-05, | |
| "loss": 0.0013, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.2135835567470957e-05, | |
| "loss": 0.0623, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.2046470062555855e-05, | |
| "loss": 0.0007, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.1957104557640751e-05, | |
| "loss": 0.0019, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.186773905272565e-05, | |
| "loss": 0.002, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.1778373547810546e-05, | |
| "loss": 0.0012, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.1689008042895442e-05, | |
| "loss": 0.0018, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.159964253798034e-05, | |
| "loss": 0.0016, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.1510277033065238e-05, | |
| "loss": 0.0013, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.1420911528150136e-05, | |
| "loss": 0.0011, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.1331546023235031e-05, | |
| "loss": 0.0021, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.1242180518319929e-05, | |
| "loss": 0.0019, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.1152815013404827e-05, | |
| "loss": 0.0017, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1063449508489723e-05, | |
| "loss": 0.0013, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.0974084003574621e-05, | |
| "loss": 0.0015, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.088471849865952e-05, | |
| "loss": 0.0021, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0795352993744414e-05, | |
| "loss": 0.0012, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0705987488829312e-05, | |
| "loss": 0.001, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.061662198391421e-05, | |
| "loss": 0.0011, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 1.0527256478999108e-05, | |
| "loss": 0.0011, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.0437890974084005e-05, | |
| "loss": 0.0024, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.0348525469168903e-05, | |
| "loss": 0.0012, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.0259159964253799e-05, | |
| "loss": 0.0009, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1.0169794459338695e-05, | |
| "loss": 0.0008, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.0080428954423593e-05, | |
| "loss": 0.0014, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 9.99106344950849e-06, | |
| "loss": 0.0021, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 9.901697944593388e-06, | |
| "loss": 0.001, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 9.812332439678284e-06, | |
| "loss": 0.0014, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 9.722966934763182e-06, | |
| "loss": 0.0014, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 9.63360142984808e-06, | |
| "loss": 0.0012, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 9.544235924932977e-06, | |
| "loss": 0.001, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 9.454870420017875e-06, | |
| "loss": 0.0004, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 9.365504915102771e-06, | |
| "loss": 0.0011, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 9.276139410187667e-06, | |
| "loss": 0.0013, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 9.186773905272565e-06, | |
| "loss": 0.0005, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 9.097408400357462e-06, | |
| "loss": 0.0006, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 9.00804289544236e-06, | |
| "loss": 0.0005, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 8.918677390527258e-06, | |
| "loss": 0.001, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 8.829311885612154e-06, | |
| "loss": 0.0013, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.739946380697052e-06, | |
| "loss": 0.0013, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.650580875781949e-06, | |
| "loss": 0.0007, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 8.561215370866847e-06, | |
| "loss": 0.0012, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 8.471849865951743e-06, | |
| "loss": 0.0008, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.38248436103664e-06, | |
| "loss": 0.0015, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 8.293118856121537e-06, | |
| "loss": 0.0011, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 8.203753351206435e-06, | |
| "loss": 0.001, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.0004990746383555233, | |
| "eval_runtime": 582.349, | |
| "eval_samples_per_second": 1.913, | |
| "eval_steps_per_second": 1.913, | |
| "step": 13728 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 8.114387846291332e-06, | |
| "loss": 0.0006, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.02502234137623e-06, | |
| "loss": 0.0007, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.935656836461126e-06, | |
| "loss": 0.0012, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 7.846291331546024e-06, | |
| "loss": 0.0013, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.75692582663092e-06, | |
| "loss": 0.0004, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.667560321715819e-06, | |
| "loss": 0.0015, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 7.578194816800716e-06, | |
| "loss": 0.0011, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 7.488829311885612e-06, | |
| "loss": 0.0009, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 7.39946380697051e-06, | |
| "loss": 0.001, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 7.3100983020554075e-06, | |
| "loss": 0.0004, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 7.220732797140304e-06, | |
| "loss": 0.0013, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 7.131367292225202e-06, | |
| "loss": 0.0007, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 7.042001787310099e-06, | |
| "loss": 0.0007, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 6.9526362823949954e-06, | |
| "loss": 0.0004, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 6.8632707774798935e-06, | |
| "loss": 0.0007, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 6.77390527256479e-06, | |
| "loss": 0.0003, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.684539767649688e-06, | |
| "loss": 0.0014, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 6.595174262734585e-06, | |
| "loss": 0.0005, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.5058087578194814e-06, | |
| "loss": 0.0008, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 6.4164432529043795e-06, | |
| "loss": 0.0009, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.327077747989277e-06, | |
| "loss": 0.0004, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 6.237712243074175e-06, | |
| "loss": 0.0008, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 6.148346738159071e-06, | |
| "loss": 0.0009, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 6.058981233243968e-06, | |
| "loss": 0.001, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 5.9696157283288655e-06, | |
| "loss": 0.0005, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 5.880250223413763e-06, | |
| "loss": 0.0011, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.790884718498661e-06, | |
| "loss": 0.0003, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 5.701519213583557e-06, | |
| "loss": 0.0003, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 5.612153708668454e-06, | |
| "loss": 0.0004, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 5.522788203753352e-06, | |
| "loss": 0.0006, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 5.433422698838249e-06, | |
| "loss": 0.0004, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 5.344057193923146e-06, | |
| "loss": 0.0005, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 5.254691689008043e-06, | |
| "loss": 0.0005, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 5.16532618409294e-06, | |
| "loss": 0.0006, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 5.075960679177838e-06, | |
| "loss": 0.001, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 4.986595174262735e-06, | |
| "loss": 0.0002, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 4.897229669347633e-06, | |
| "loss": 0.0006, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 4.80786416443253e-06, | |
| "loss": 0.0006, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 4.718498659517426e-06, | |
| "loss": 0.0006, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 4.6291331546023235e-06, | |
| "loss": 0.0005, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.5397676496872215e-06, | |
| "loss": 0.0007, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 4.450402144772119e-06, | |
| "loss": 0.0004, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 4.361036639857015e-06, | |
| "loss": 0.0004, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 4.271671134941912e-06, | |
| "loss": 0.0004, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 4.18230563002681e-06, | |
| "loss": 0.0002, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.0929401251117075e-06, | |
| "loss": 0.0007, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.0004931196453981102, | |
| "eval_runtime": 582.3074, | |
| "eval_samples_per_second": 1.913, | |
| "eval_steps_per_second": 1.913, | |
| "step": 18304 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.003574620196605e-06, | |
| "loss": 0.0007, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.914209115281501e-06, | |
| "loss": 0.0005, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.824843610366399e-06, | |
| "loss": 0.0007, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.7354781054512963e-06, | |
| "loss": 0.0003, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.6461126005361935e-06, | |
| "loss": 0.0004, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.5567470956210903e-06, | |
| "loss": 0.0002, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.4673815907059875e-06, | |
| "loss": 0.0004, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.378016085790885e-06, | |
| "loss": 0.0003, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.2886505808757823e-06, | |
| "loss": 0.0009, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.199285075960679e-06, | |
| "loss": 0.0007, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.1099195710455763e-06, | |
| "loss": 0.0006, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 3.020554066130474e-06, | |
| "loss": 0.0006, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 2.931188561215371e-06, | |
| "loss": 0.0002, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 2.8418230563002683e-06, | |
| "loss": 0.0004, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 2.7524575513851655e-06, | |
| "loss": 0.0004, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.6630920464700628e-06, | |
| "loss": 0.0004, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 2.57372654155496e-06, | |
| "loss": 0.0005, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 2.484361036639857e-06, | |
| "loss": 0.0003, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 2.3949955317247544e-06, | |
| "loss": 0.0004, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.3056300268096516e-06, | |
| "loss": 0.0003, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.2162645218945488e-06, | |
| "loss": 0.0004, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.1268990169794464e-06, | |
| "loss": 0.0005, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.037533512064343e-06, | |
| "loss": 0.0003, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 1.9481680071492408e-06, | |
| "loss": 0.0003, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 1.8588025022341378e-06, | |
| "loss": 0.0003, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 1.769436997319035e-06, | |
| "loss": 0.0003, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 1.6800714924039322e-06, | |
| "loss": 0.001, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 1.5907059874888296e-06, | |
| "loss": 0.0008, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 1.5013404825737266e-06, | |
| "loss": 0.0002, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 1.411974977658624e-06, | |
| "loss": 0.0005, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 1.322609472743521e-06, | |
| "loss": 0.0004, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 1.2332439678284184e-06, | |
| "loss": 0.0003, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 1.1438784629133156e-06, | |
| "loss": 0.0001, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.0545129579982128e-06, | |
| "loss": 0.0006, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 9.6514745308311e-07, | |
| "loss": 0.0004, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 8.757819481680072e-07, | |
| "loss": 0.0002, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 7.864164432529045e-07, | |
| "loss": 0.0004, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 6.970509383378017e-07, | |
| "loss": 0.0001, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 6.076854334226989e-07, | |
| "loss": 0.0003, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 5.183199285075961e-07, | |
| "loss": 0.0004, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 4.2895442359249334e-07, | |
| "loss": 0.0003, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.3958891867739054e-07, | |
| "loss": 0.0003, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 2.502234137622878e-07, | |
| "loss": 0.0004, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 1.60857908847185e-07, | |
| "loss": 0.0005, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 7.149240393208222e-08, | |
| "loss": 0.0002, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.00048321104259230196, | |
| "eval_runtime": 582.3092, | |
| "eval_samples_per_second": 1.913, | |
| "eval_steps_per_second": 1.913, | |
| "step": 22880 | |
| } | |
| ], | |
| "max_steps": 22880, | |
| "num_train_epochs": 5, | |
| "total_flos": 2.1093236286160896e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |