| { | |
| "best_metric": 0.9315417408943176, | |
| "best_model_checkpoint": "./albert_sentiment_model\\checkpoint-2000", | |
| "epoch": 2.3094688221709005, | |
| "eval_steps": 500, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.007698229407236336, | |
| "grad_norm": 25.626720428466797, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.9667, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015396458814472672, | |
| "grad_norm": 35.37693405151367, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.0464, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023094688221709007, | |
| "grad_norm": 22.576478958129883, | |
| "learning_rate": 3e-06, | |
| "loss": 1.9565, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.030792917628945343, | |
| "grad_norm": 21.228620529174805, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.9974, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03849114703618168, | |
| "grad_norm": 27.480052947998047, | |
| "learning_rate": 5e-06, | |
| "loss": 1.9431, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.046189376443418015, | |
| "grad_norm": 27.75535774230957, | |
| "learning_rate": 6e-06, | |
| "loss": 1.8815, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05388760585065435, | |
| "grad_norm": 17.684391021728516, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 1.8753, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.061585835257890686, | |
| "grad_norm": 21.62338638305664, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.8629, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06928406466512702, | |
| "grad_norm": 43.68171691894531, | |
| "learning_rate": 9e-06, | |
| "loss": 1.8003, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07698229407236336, | |
| "grad_norm": 22.630041122436523, | |
| "learning_rate": 1e-05, | |
| "loss": 1.7636, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08468052347959969, | |
| "grad_norm": 30.946426391601562, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 1.7074, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09237875288683603, | |
| "grad_norm": 30.420377731323242, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.6716, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10007698229407236, | |
| "grad_norm": 48.35504150390625, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 1.4465, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1077752117013087, | |
| "grad_norm": 51.120361328125, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 1.4337, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11547344110854503, | |
| "grad_norm": 29.794086456298828, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.3535, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12317167051578137, | |
| "grad_norm": 34.01353454589844, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.3233, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1308698999230177, | |
| "grad_norm": 47.1700439453125, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 1.342, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13856812933025403, | |
| "grad_norm": 34.01679611206055, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.4129, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1462663587374904, | |
| "grad_norm": 27.47703742980957, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.3769, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15396458814472672, | |
| "grad_norm": 60.44585037231445, | |
| "learning_rate": 2e-05, | |
| "loss": 1.3133, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16166281755196305, | |
| "grad_norm": 23.065670013427734, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.2592, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.16936104695919937, | |
| "grad_norm": 70.744873046875, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 1.3765, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17705927636643573, | |
| "grad_norm": 47.14573287963867, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 1.3629, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18475750577367206, | |
| "grad_norm": 26.078250885009766, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.1865, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.1924557351809084, | |
| "grad_norm": 63.99885940551758, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.3399, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20015396458814472, | |
| "grad_norm": 24.421600341796875, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 1.1854, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.20785219399538107, | |
| "grad_norm": 16.544498443603516, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.2511, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.2155504234026174, | |
| "grad_norm": 52.77935791015625, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 1.2485, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22324865280985373, | |
| "grad_norm": 68.4757080078125, | |
| "learning_rate": 2.9e-05, | |
| "loss": 1.0838, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23094688221709006, | |
| "grad_norm": 26.108631134033203, | |
| "learning_rate": 3e-05, | |
| "loss": 1.3167, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23864511162432642, | |
| "grad_norm": 16.388147354125977, | |
| "learning_rate": 3.1e-05, | |
| "loss": 1.0942, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.24634334103156275, | |
| "grad_norm": 27.521499633789062, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 1.3823, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2540415704387991, | |
| "grad_norm": 18.420778274536133, | |
| "learning_rate": 3.3e-05, | |
| "loss": 1.1258, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2617397998460354, | |
| "grad_norm": 16.579042434692383, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.159, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.26943802925327176, | |
| "grad_norm": 48.20399856567383, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.1556, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.27713625866050806, | |
| "grad_norm": 12.147098541259766, | |
| "learning_rate": 3.6e-05, | |
| "loss": 1.1309, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2848344880677444, | |
| "grad_norm": 52.27555847167969, | |
| "learning_rate": 3.7e-05, | |
| "loss": 1.3251, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2925327174749808, | |
| "grad_norm": 42.3845100402832, | |
| "learning_rate": 3.8e-05, | |
| "loss": 1.3005, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3002309468822171, | |
| "grad_norm": 29.475187301635742, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 1.2515, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.30792917628945343, | |
| "grad_norm": 60.33949279785156, | |
| "learning_rate": 4e-05, | |
| "loss": 1.2975, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3156274056966898, | |
| "grad_norm": 47.650211334228516, | |
| "learning_rate": 4.1e-05, | |
| "loss": 1.2688, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.3233256351039261, | |
| "grad_norm": 20.859683990478516, | |
| "learning_rate": 4.2e-05, | |
| "loss": 1.2035, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33102386451116245, | |
| "grad_norm": 16.41029930114746, | |
| "learning_rate": 4.3e-05, | |
| "loss": 1.1203, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.33872209391839875, | |
| "grad_norm": 12.632010459899902, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.1197, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3464203233256351, | |
| "grad_norm": 36.34510803222656, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.2378, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.35411855273287146, | |
| "grad_norm": 52.647926330566406, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.2022, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36181678214010776, | |
| "grad_norm": 88.51087188720703, | |
| "learning_rate": 4.7e-05, | |
| "loss": 1.0755, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3695150115473441, | |
| "grad_norm": 27.021543502807617, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.2264, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3772132409545804, | |
| "grad_norm": 13.925777435302734, | |
| "learning_rate": 4.9e-05, | |
| "loss": 1.0287, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3849114703618168, | |
| "grad_norm": 17.583614349365234, | |
| "learning_rate": 5e-05, | |
| "loss": 0.979, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3849114703618168, | |
| "eval_accuracy": 0.5254041570438799, | |
| "eval_f1": 0.4643762968149376, | |
| "eval_loss": 1.1208206415176392, | |
| "eval_precision": 0.47090986837941123, | |
| "eval_recall": 0.5254041570438799, | |
| "eval_runtime": 7.9109, | |
| "eval_samples_per_second": 328.406, | |
| "eval_steps_per_second": 41.082, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.39260969976905313, | |
| "grad_norm": 29.161548614501953, | |
| "learning_rate": 4.998700285937094e-05, | |
| "loss": 1.1457, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40030792917628943, | |
| "grad_norm": 45.45001220703125, | |
| "learning_rate": 4.997400571874188e-05, | |
| "loss": 1.4028, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4080061585835258, | |
| "grad_norm": 20.393016815185547, | |
| "learning_rate": 4.9961008578112814e-05, | |
| "loss": 1.3193, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.41570438799076215, | |
| "grad_norm": 9.973402976989746, | |
| "learning_rate": 4.9948011437483756e-05, | |
| "loss": 1.088, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.42340261739799845, | |
| "grad_norm": 41.007110595703125, | |
| "learning_rate": 4.993501429685469e-05, | |
| "loss": 1.3676, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.4311008468052348, | |
| "grad_norm": 13.304949760437012, | |
| "learning_rate": 4.992201715622563e-05, | |
| "loss": 1.3313, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4387990762124711, | |
| "grad_norm": 10.015753746032715, | |
| "learning_rate": 4.990902001559657e-05, | |
| "loss": 1.0298, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.44649730561970746, | |
| "grad_norm": 67.66326904296875, | |
| "learning_rate": 4.989602287496751e-05, | |
| "loss": 1.0524, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4541955350269438, | |
| "grad_norm": 28.423460006713867, | |
| "learning_rate": 4.9883025734338444e-05, | |
| "loss": 1.1959, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4618937644341801, | |
| "grad_norm": 18.55489158630371, | |
| "learning_rate": 4.9870028593709386e-05, | |
| "loss": 1.0586, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4695919938414165, | |
| "grad_norm": 36.054500579833984, | |
| "learning_rate": 4.985703145308033e-05, | |
| "loss": 1.2009, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.47729022324865283, | |
| "grad_norm": 18.236879348754883, | |
| "learning_rate": 4.984403431245126e-05, | |
| "loss": 1.1012, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.48498845265588914, | |
| "grad_norm": 20.71109962463379, | |
| "learning_rate": 4.9831037171822204e-05, | |
| "loss": 1.1486, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4926866820631255, | |
| "grad_norm": 28.097261428833008, | |
| "learning_rate": 4.981804003119314e-05, | |
| "loss": 1.1645, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5003849114703618, | |
| "grad_norm": 20.31618309020996, | |
| "learning_rate": 4.980504289056408e-05, | |
| "loss": 1.1256, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5080831408775982, | |
| "grad_norm": 126.53520202636719, | |
| "learning_rate": 4.9792045749935016e-05, | |
| "loss": 1.4823, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5157813702848345, | |
| "grad_norm": 14.448310852050781, | |
| "learning_rate": 4.977904860930596e-05, | |
| "loss": 1.2872, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5234795996920708, | |
| "grad_norm": 16.949066162109375, | |
| "learning_rate": 4.976605146867689e-05, | |
| "loss": 1.2012, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5311778290993071, | |
| "grad_norm": 14.372076034545898, | |
| "learning_rate": 4.9753054328047835e-05, | |
| "loss": 1.1928, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5388760585065435, | |
| "grad_norm": 76.28020477294922, | |
| "learning_rate": 4.974005718741877e-05, | |
| "loss": 1.1082, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5465742879137798, | |
| "grad_norm": 41.16048049926758, | |
| "learning_rate": 4.972706004678971e-05, | |
| "loss": 1.0006, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5542725173210161, | |
| "grad_norm": 64.6050033569336, | |
| "learning_rate": 4.9714062906160646e-05, | |
| "loss": 1.2638, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5619707467282525, | |
| "grad_norm": 35.458457946777344, | |
| "learning_rate": 4.970106576553159e-05, | |
| "loss": 1.0841, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5696689761354888, | |
| "grad_norm": 33.32878494262695, | |
| "learning_rate": 4.968806862490252e-05, | |
| "loss": 1.2451, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5773672055427251, | |
| "grad_norm": 26.29853630065918, | |
| "learning_rate": 4.967507148427346e-05, | |
| "loss": 1.0487, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5850654349499615, | |
| "grad_norm": 6.7931084632873535, | |
| "learning_rate": 4.96620743436444e-05, | |
| "loss": 1.0836, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5927636643571979, | |
| "grad_norm": 11.772964477539062, | |
| "learning_rate": 4.9649077203015335e-05, | |
| "loss": 0.9459, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6004618937644342, | |
| "grad_norm": 83.65714263916016, | |
| "learning_rate": 4.9636080062386276e-05, | |
| "loss": 1.0921, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6081601231716706, | |
| "grad_norm": 67.42829132080078, | |
| "learning_rate": 4.962308292175721e-05, | |
| "loss": 1.0981, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6158583525789069, | |
| "grad_norm": 40.03876876831055, | |
| "learning_rate": 4.961008578112815e-05, | |
| "loss": 1.1155, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6235565819861432, | |
| "grad_norm": 14.459259033203125, | |
| "learning_rate": 4.9597088640499095e-05, | |
| "loss": 1.2369, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6312548113933796, | |
| "grad_norm": 43.21743392944336, | |
| "learning_rate": 4.9584091499870037e-05, | |
| "loss": 1.1154, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6389530408006159, | |
| "grad_norm": 29.028535842895508, | |
| "learning_rate": 4.957109435924097e-05, | |
| "loss": 1.0684, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6466512702078522, | |
| "grad_norm": 3421.910400390625, | |
| "learning_rate": 4.9558097218611907e-05, | |
| "loss": 1.1415, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6543494996150885, | |
| "grad_norm": 50.08725357055664, | |
| "learning_rate": 4.954510007798285e-05, | |
| "loss": 1.5334, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6620477290223249, | |
| "grad_norm": 31.24684715270996, | |
| "learning_rate": 4.953210293735378e-05, | |
| "loss": 1.1275, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6697459584295612, | |
| "grad_norm": 56.685611724853516, | |
| "learning_rate": 4.9519105796724725e-05, | |
| "loss": 1.1105, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6774441878367975, | |
| "grad_norm": 59.59196853637695, | |
| "learning_rate": 4.950610865609566e-05, | |
| "loss": 1.296, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6851424172440339, | |
| "grad_norm": 59.091697692871094, | |
| "learning_rate": 4.94931115154666e-05, | |
| "loss": 1.0462, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6928406466512702, | |
| "grad_norm": 25.288618087768555, | |
| "learning_rate": 4.948011437483754e-05, | |
| "loss": 0.9684, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7005388760585065, | |
| "grad_norm": 34.281044006347656, | |
| "learning_rate": 4.946711723420848e-05, | |
| "loss": 1.0789, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7082371054657429, | |
| "grad_norm": 12.016758918762207, | |
| "learning_rate": 4.945412009357941e-05, | |
| "loss": 1.1008, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7159353348729792, | |
| "grad_norm": 23.659439086914062, | |
| "learning_rate": 4.9441122952950355e-05, | |
| "loss": 1.3602, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7236335642802155, | |
| "grad_norm": 12.893980026245117, | |
| "learning_rate": 4.942812581232129e-05, | |
| "loss": 1.3067, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7313317936874519, | |
| "grad_norm": 36.1422004699707, | |
| "learning_rate": 4.941512867169223e-05, | |
| "loss": 1.2795, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7390300230946882, | |
| "grad_norm": 24.229007720947266, | |
| "learning_rate": 4.940213153106317e-05, | |
| "loss": 0.8846, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7467282525019245, | |
| "grad_norm": 12.185916900634766, | |
| "learning_rate": 4.938913439043411e-05, | |
| "loss": 1.0618, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7544264819091608, | |
| "grad_norm": 58.11173629760742, | |
| "learning_rate": 4.9376137249805044e-05, | |
| "loss": 1.0567, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7621247113163973, | |
| "grad_norm": 17.494789123535156, | |
| "learning_rate": 4.936314010917598e-05, | |
| "loss": 0.9427, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7698229407236336, | |
| "grad_norm": 21.812406539916992, | |
| "learning_rate": 4.935014296854692e-05, | |
| "loss": 1.0871, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7698229407236336, | |
| "eval_accuracy": 0.5400307929176289, | |
| "eval_f1": 0.5212757553318326, | |
| "eval_loss": 1.0452667474746704, | |
| "eval_precision": 0.5391802513141176, | |
| "eval_recall": 0.5400307929176289, | |
| "eval_runtime": 7.9557, | |
| "eval_samples_per_second": 326.559, | |
| "eval_steps_per_second": 40.851, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7775211701308699, | |
| "grad_norm": 68.46436309814453, | |
| "learning_rate": 4.9337145827917855e-05, | |
| "loss": 1.3711, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.7852193995381063, | |
| "grad_norm": 25.05400848388672, | |
| "learning_rate": 4.9324148687288804e-05, | |
| "loss": 1.1931, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7929176289453426, | |
| "grad_norm": 24.60456085205078, | |
| "learning_rate": 4.931115154665974e-05, | |
| "loss": 1.1109, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8006158583525789, | |
| "grad_norm": 24.26364517211914, | |
| "learning_rate": 4.929815440603068e-05, | |
| "loss": 1.0804, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8083140877598153, | |
| "grad_norm": 22.994844436645508, | |
| "learning_rate": 4.9285157265401615e-05, | |
| "loss": 0.8045, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8160123171670516, | |
| "grad_norm": 16.76064682006836, | |
| "learning_rate": 4.927216012477256e-05, | |
| "loss": 1.1683, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8237105465742879, | |
| "grad_norm": 46.755393981933594, | |
| "learning_rate": 4.925916298414349e-05, | |
| "loss": 1.1455, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.8314087759815243, | |
| "grad_norm": 46.636077880859375, | |
| "learning_rate": 4.924616584351443e-05, | |
| "loss": 0.9044, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8391070053887606, | |
| "grad_norm": 25.199445724487305, | |
| "learning_rate": 4.923316870288537e-05, | |
| "loss": 1.0345, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8468052347959969, | |
| "grad_norm": 37.11675262451172, | |
| "learning_rate": 4.9220171562256304e-05, | |
| "loss": 1.2491, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8545034642032333, | |
| "grad_norm": 19.960737228393555, | |
| "learning_rate": 4.9207174421627246e-05, | |
| "loss": 0.9029, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8622016936104696, | |
| "grad_norm": 81.35137939453125, | |
| "learning_rate": 4.919417728099818e-05, | |
| "loss": 1.2947, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8698999230177059, | |
| "grad_norm": 32.889427185058594, | |
| "learning_rate": 4.918118014036912e-05, | |
| "loss": 1.1014, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8775981524249422, | |
| "grad_norm": 35.65376281738281, | |
| "learning_rate": 4.916818299974006e-05, | |
| "loss": 1.1406, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.8852963818321786, | |
| "grad_norm": 309.11114501953125, | |
| "learning_rate": 4.9155185859111e-05, | |
| "loss": 0.9684, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8929946112394149, | |
| "grad_norm": 19.624103546142578, | |
| "learning_rate": 4.9142188718481934e-05, | |
| "loss": 0.9999, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9006928406466512, | |
| "grad_norm": 35.4536018371582, | |
| "learning_rate": 4.9129191577852876e-05, | |
| "loss": 1.3579, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9083910700538876, | |
| "grad_norm": 94.22552490234375, | |
| "learning_rate": 4.911619443722381e-05, | |
| "loss": 1.0926, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9160892994611239, | |
| "grad_norm": 23.519554138183594, | |
| "learning_rate": 4.910319729659475e-05, | |
| "loss": 1.1782, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9237875288683602, | |
| "grad_norm": 25.848222732543945, | |
| "learning_rate": 4.909020015596569e-05, | |
| "loss": 1.0085, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9314857582755967, | |
| "grad_norm": 40.45154571533203, | |
| "learning_rate": 4.907720301533663e-05, | |
| "loss": 1.0384, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.939183987682833, | |
| "grad_norm": 26.536237716674805, | |
| "learning_rate": 4.9064205874707564e-05, | |
| "loss": 1.006, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9468822170900693, | |
| "grad_norm": 17.983915328979492, | |
| "learning_rate": 4.9051208734078506e-05, | |
| "loss": 1.1031, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9545804464973057, | |
| "grad_norm": 46.20643615722656, | |
| "learning_rate": 4.903821159344945e-05, | |
| "loss": 0.9322, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.962278675904542, | |
| "grad_norm": 59.26303482055664, | |
| "learning_rate": 4.902521445282038e-05, | |
| "loss": 1.1146, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9699769053117783, | |
| "grad_norm": 42.50138473510742, | |
| "learning_rate": 4.9012217312191324e-05, | |
| "loss": 1.1769, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9776751347190146, | |
| "grad_norm": 44.46087646484375, | |
| "learning_rate": 4.899922017156226e-05, | |
| "loss": 1.113, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.985373364126251, | |
| "grad_norm": 24.701412200927734, | |
| "learning_rate": 4.89862230309332e-05, | |
| "loss": 1.0445, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.9930715935334873, | |
| "grad_norm": 19.296676635742188, | |
| "learning_rate": 4.8973225890304136e-05, | |
| "loss": 1.075, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.0007698229407236, | |
| "grad_norm": 24.70547866821289, | |
| "learning_rate": 4.896022874967508e-05, | |
| "loss": 0.8709, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.0084680523479599, | |
| "grad_norm": 35.57961654663086, | |
| "learning_rate": 4.894723160904601e-05, | |
| "loss": 1.0378, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.0161662817551964, | |
| "grad_norm": 30.505125045776367, | |
| "learning_rate": 4.893423446841695e-05, | |
| "loss": 0.992, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.0238645111624327, | |
| "grad_norm": 58.703643798828125, | |
| "learning_rate": 4.892123732778789e-05, | |
| "loss": 1.1885, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.031562740569669, | |
| "grad_norm": 28.3648681640625, | |
| "learning_rate": 4.8908240187158824e-05, | |
| "loss": 1.2337, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.0392609699769053, | |
| "grad_norm": 24.309017181396484, | |
| "learning_rate": 4.8895243046529766e-05, | |
| "loss": 1.0578, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.0469591993841416, | |
| "grad_norm": 26.426084518432617, | |
| "learning_rate": 4.88822459059007e-05, | |
| "loss": 0.9974, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.054657428791378, | |
| "grad_norm": 27.277679443359375, | |
| "learning_rate": 4.886924876527164e-05, | |
| "loss": 1.0032, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.0623556581986142, | |
| "grad_norm": 22.838211059570312, | |
| "learning_rate": 4.885625162464258e-05, | |
| "loss": 1.1148, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.0700538876058507, | |
| "grad_norm": 41.72176742553711, | |
| "learning_rate": 4.884325448401352e-05, | |
| "loss": 0.9433, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.077752117013087, | |
| "grad_norm": 72.71926879882812, | |
| "learning_rate": 4.8830257343384455e-05, | |
| "loss": 0.8138, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.0854503464203233, | |
| "grad_norm": 20.042091369628906, | |
| "learning_rate": 4.8817260202755396e-05, | |
| "loss": 1.0195, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.0931485758275596, | |
| "grad_norm": 20.96778678894043, | |
| "learning_rate": 4.880426306212633e-05, | |
| "loss": 1.0269, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.100846805234796, | |
| "grad_norm": 32.662322998046875, | |
| "learning_rate": 4.879126592149727e-05, | |
| "loss": 0.891, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.1085450346420322, | |
| "grad_norm": 19.693790435791016, | |
| "learning_rate": 4.8778268780868215e-05, | |
| "loss": 1.054, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.1162432640492688, | |
| "grad_norm": 20.91014289855957, | |
| "learning_rate": 4.876527164023915e-05, | |
| "loss": 0.9135, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.123941493456505, | |
| "grad_norm": 52.75801086425781, | |
| "learning_rate": 4.875227449961009e-05, | |
| "loss": 1.0464, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.1316397228637414, | |
| "grad_norm": 42.71292495727539, | |
| "learning_rate": 4.8739277358981026e-05, | |
| "loss": 1.0075, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.1393379522709777, | |
| "grad_norm": 21.153303146362305, | |
| "learning_rate": 4.872628021835197e-05, | |
| "loss": 1.2111, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.147036181678214, | |
| "grad_norm": 44.54867172241211, | |
| "learning_rate": 4.87132830777229e-05, | |
| "loss": 0.8313, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.1547344110854503, | |
| "grad_norm": 48.92774963378906, | |
| "learning_rate": 4.8700285937093845e-05, | |
| "loss": 1.0748, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.1547344110854503, | |
| "eval_accuracy": 0.5658198614318707, | |
| "eval_f1": 0.5450550292077301, | |
| "eval_loss": 0.9736452102661133, | |
| "eval_precision": 0.5647189867888521, | |
| "eval_recall": 0.5658198614318707, | |
| "eval_runtime": 8.0094, | |
| "eval_samples_per_second": 324.367, | |
| "eval_steps_per_second": 40.577, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.1624326404926868, | |
| "grad_norm": 39.90788650512695, | |
| "learning_rate": 4.868728879646478e-05, | |
| "loss": 1.0338, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.170130869899923, | |
| "grad_norm": 35.67189025878906, | |
| "learning_rate": 4.867429165583572e-05, | |
| "loss": 1.0185, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.1778290993071594, | |
| "grad_norm": 76.6269302368164, | |
| "learning_rate": 4.8661294515206657e-05, | |
| "loss": 1.0013, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.1855273287143957, | |
| "grad_norm": 166.01284790039062, | |
| "learning_rate": 4.86482973745776e-05, | |
| "loss": 1.0846, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.193225558121632, | |
| "grad_norm": 26.522552490234375, | |
| "learning_rate": 4.863530023394853e-05, | |
| "loss": 1.012, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.2009237875288683, | |
| "grad_norm": 23.47018051147461, | |
| "learning_rate": 4.862230309331947e-05, | |
| "loss": 0.9853, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.2086220169361046, | |
| "grad_norm": 16.074838638305664, | |
| "learning_rate": 4.860930595269041e-05, | |
| "loss": 0.9196, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.2163202463433411, | |
| "grad_norm": 21.358341217041016, | |
| "learning_rate": 4.8596308812061345e-05, | |
| "loss": 1.0509, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.2240184757505774, | |
| "grad_norm": 28.035846710205078, | |
| "learning_rate": 4.858331167143229e-05, | |
| "loss": 0.9299, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.2317167051578137, | |
| "grad_norm": 23.313568115234375, | |
| "learning_rate": 4.857031453080322e-05, | |
| "loss": 0.9581, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.23941493456505, | |
| "grad_norm": 26.382869720458984, | |
| "learning_rate": 4.8557317390174163e-05, | |
| "loss": 0.8374, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.2471131639722863, | |
| "grad_norm": 62.66410446166992, | |
| "learning_rate": 4.85443202495451e-05, | |
| "loss": 1.1268, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.2548113933795226, | |
| "grad_norm": 32.7457160949707, | |
| "learning_rate": 4.853132310891604e-05, | |
| "loss": 0.7796, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.2625096227867592, | |
| "grad_norm": 45.75529861450195, | |
| "learning_rate": 4.851832596828698e-05, | |
| "loss": 1.0265, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.2702078521939955, | |
| "grad_norm": 73.32510375976562, | |
| "learning_rate": 4.850532882765792e-05, | |
| "loss": 1.3302, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.2779060816012318, | |
| "grad_norm": 51.92557144165039, | |
| "learning_rate": 4.849233168702886e-05, | |
| "loss": 1.2056, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.285604311008468, | |
| "grad_norm": 9.508386611938477, | |
| "learning_rate": 4.8479334546399794e-05, | |
| "loss": 0.8817, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.2933025404157044, | |
| "grad_norm": 19.91646957397461, | |
| "learning_rate": 4.8466337405770735e-05, | |
| "loss": 1.1081, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.3010007698229407, | |
| "grad_norm": 22.991626739501953, | |
| "learning_rate": 4.845334026514167e-05, | |
| "loss": 0.9953, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.308698999230177, | |
| "grad_norm": 36.05852127075195, | |
| "learning_rate": 4.844034312451261e-05, | |
| "loss": 0.9753, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.3163972286374133, | |
| "grad_norm": 30.5816593170166, | |
| "learning_rate": 4.842734598388355e-05, | |
| "loss": 1.0043, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.3240954580446498, | |
| "grad_norm": 32.22935485839844, | |
| "learning_rate": 4.841434884325449e-05, | |
| "loss": 0.977, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.331793687451886, | |
| "grad_norm": 93.55306243896484, | |
| "learning_rate": 4.8401351702625424e-05, | |
| "loss": 0.959, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.3394919168591224, | |
| "grad_norm": 20.172109603881836, | |
| "learning_rate": 4.8388354561996365e-05, | |
| "loss": 0.8522, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.3471901462663587, | |
| "grad_norm": 34.432254791259766, | |
| "learning_rate": 4.83753574213673e-05, | |
| "loss": 0.9968, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.354888375673595, | |
| "grad_norm": 9.962430000305176, | |
| "learning_rate": 4.836236028073824e-05, | |
| "loss": 0.8846, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.3625866050808315, | |
| "grad_norm": 16.258909225463867, | |
| "learning_rate": 4.834936314010918e-05, | |
| "loss": 0.9685, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.3702848344880678, | |
| "grad_norm": 600.8819580078125, | |
| "learning_rate": 4.833636599948012e-05, | |
| "loss": 1.2891, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.3779830638953041, | |
| "grad_norm": 19.667699813842773, | |
| "learning_rate": 4.8323368858851054e-05, | |
| "loss": 0.9624, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.3856812933025404, | |
| "grad_norm": 28.33806037902832, | |
| "learning_rate": 4.831037171822199e-05, | |
| "loss": 0.9677, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.3933795227097767, | |
| "grad_norm": 17.19879722595215, | |
| "learning_rate": 4.829737457759293e-05, | |
| "loss": 0.7864, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.401077752117013, | |
| "grad_norm": 68.86786651611328, | |
| "learning_rate": 4.8284377436963865e-05, | |
| "loss": 1.0756, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.4087759815242493, | |
| "grad_norm": 19.253087997436523, | |
| "learning_rate": 4.827138029633481e-05, | |
| "loss": 0.902, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.4164742109314856, | |
| "grad_norm": 22.85369300842285, | |
| "learning_rate": 4.825838315570575e-05, | |
| "loss": 0.9561, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.4241724403387221, | |
| "grad_norm": 34.35043716430664, | |
| "learning_rate": 4.824538601507669e-05, | |
| "loss": 0.938, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.4318706697459584, | |
| "grad_norm": 29.028274536132812, | |
| "learning_rate": 4.8232388874447626e-05, | |
| "loss": 0.9659, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.4395688991531947, | |
| "grad_norm": 15.733872413635254, | |
| "learning_rate": 4.821939173381857e-05, | |
| "loss": 0.955, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.447267128560431, | |
| "grad_norm": 45.1998176574707, | |
| "learning_rate": 4.82063945931895e-05, | |
| "loss": 1.0997, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.4549653579676676, | |
| "grad_norm": 62.95772171020508, | |
| "learning_rate": 4.819339745256044e-05, | |
| "loss": 0.9975, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.4626635873749039, | |
| "grad_norm": 13.520893096923828, | |
| "learning_rate": 4.818040031193138e-05, | |
| "loss": 0.9224, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.4703618167821402, | |
| "grad_norm": 44.204837799072266, | |
| "learning_rate": 4.8167403171302314e-05, | |
| "loss": 1.0823, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.4780600461893765, | |
| "grad_norm": 18.848365783691406, | |
| "learning_rate": 4.8154406030673256e-05, | |
| "loss": 0.8767, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.4857582755966128, | |
| "grad_norm": 48.03689193725586, | |
| "learning_rate": 4.814140889004419e-05, | |
| "loss": 1.1274, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.493456505003849, | |
| "grad_norm": 29.64519691467285, | |
| "learning_rate": 4.812841174941513e-05, | |
| "loss": 0.8109, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.5011547344110854, | |
| "grad_norm": 56.1988639831543, | |
| "learning_rate": 4.811541460878607e-05, | |
| "loss": 1.1295, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.5088529638183217, | |
| "grad_norm": 39.60947036743164, | |
| "learning_rate": 4.810241746815701e-05, | |
| "loss": 1.1057, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.516551193225558, | |
| "grad_norm": 14.454174041748047, | |
| "learning_rate": 4.8089420327527944e-05, | |
| "loss": 1.098, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.5242494226327945, | |
| "grad_norm": 18.127788543701172, | |
| "learning_rate": 4.8076423186898886e-05, | |
| "loss": 0.947, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.5319476520400308, | |
| "grad_norm": 30.042423248291016, | |
| "learning_rate": 4.806342604626982e-05, | |
| "loss": 0.8846, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.539645881447267, | |
| "grad_norm": 31.532861709594727, | |
| "learning_rate": 4.805042890564076e-05, | |
| "loss": 0.908, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.539645881447267, | |
| "eval_accuracy": 0.5873749037721324, | |
| "eval_f1": 0.5698601671652778, | |
| "eval_loss": 0.9315417408943176, | |
| "eval_precision": 0.5808478011880224, | |
| "eval_recall": 0.5873749037721324, | |
| "eval_runtime": 8.0056, | |
| "eval_samples_per_second": 324.524, | |
| "eval_steps_per_second": 40.597, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.5473441108545036, | |
| "grad_norm": 87.31982421875, | |
| "learning_rate": 4.80374317650117e-05, | |
| "loss": 1.2135, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.55504234026174, | |
| "grad_norm": 36.29372787475586, | |
| "learning_rate": 4.802443462438263e-05, | |
| "loss": 0.9446, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.5627405696689762, | |
| "grad_norm": 65.0682144165039, | |
| "learning_rate": 4.8011437483753574e-05, | |
| "loss": 0.942, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.5704387990762125, | |
| "grad_norm": 7.867269992828369, | |
| "learning_rate": 4.799844034312451e-05, | |
| "loss": 0.9343, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.5781370284834488, | |
| "grad_norm": 18.922204971313477, | |
| "learning_rate": 4.798544320249546e-05, | |
| "loss": 1.0242, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.5858352578906851, | |
| "grad_norm": 22.182636260986328, | |
| "learning_rate": 4.797244606186639e-05, | |
| "loss": 0.8997, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.5935334872979214, | |
| "grad_norm": 13.871538162231445, | |
| "learning_rate": 4.7959448921237335e-05, | |
| "loss": 1.0238, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.6012317167051577, | |
| "grad_norm": 24.956518173217773, | |
| "learning_rate": 4.794645178060827e-05, | |
| "loss": 0.8555, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.608929946112394, | |
| "grad_norm": 27.833717346191406, | |
| "learning_rate": 4.793345463997921e-05, | |
| "loss": 1.1057, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.6166281755196303, | |
| "grad_norm": 15.595669746398926, | |
| "learning_rate": 4.7920457499350146e-05, | |
| "loss": 0.8761, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.6243264049268669, | |
| "grad_norm": 33.80525207519531, | |
| "learning_rate": 4.790746035872109e-05, | |
| "loss": 0.9889, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.6320246343341032, | |
| "grad_norm": 21.781145095825195, | |
| "learning_rate": 4.789446321809202e-05, | |
| "loss": 1.055, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.6397228637413395, | |
| "grad_norm": 29.12350082397461, | |
| "learning_rate": 4.788146607746296e-05, | |
| "loss": 1.0033, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.647421093148576, | |
| "grad_norm": 20.39841079711914, | |
| "learning_rate": 4.78684689368339e-05, | |
| "loss": 1.0698, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.6551193225558123, | |
| "grad_norm": 9.309553146362305, | |
| "learning_rate": 4.7855471796204835e-05, | |
| "loss": 0.9092, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.6628175519630486, | |
| "grad_norm": 28.19939613342285, | |
| "learning_rate": 4.7842474655575776e-05, | |
| "loss": 0.9727, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.670515781370285, | |
| "grad_norm": 18.627004623413086, | |
| "learning_rate": 4.782947751494671e-05, | |
| "loss": 0.8224, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.6782140107775212, | |
| "grad_norm": 15.454500198364258, | |
| "learning_rate": 4.781648037431765e-05, | |
| "loss": 0.9459, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.6859122401847575, | |
| "grad_norm": 25.361587524414062, | |
| "learning_rate": 4.780348323368859e-05, | |
| "loss": 0.8389, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.6936104695919938, | |
| "grad_norm": 48.25446701049805, | |
| "learning_rate": 4.779048609305953e-05, | |
| "loss": 1.1498, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.70130869899923, | |
| "grad_norm": 45.576908111572266, | |
| "learning_rate": 4.7777488952430465e-05, | |
| "loss": 0.7871, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.7090069284064664, | |
| "grad_norm": 28.02678871154785, | |
| "learning_rate": 4.7764491811801407e-05, | |
| "loss": 0.8891, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.7167051578137027, | |
| "grad_norm": 17.72388458251953, | |
| "learning_rate": 4.775149467117234e-05, | |
| "loss": 1.0627, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 1.7244033872209392, | |
| "grad_norm": 14.796937942504883, | |
| "learning_rate": 4.773849753054328e-05, | |
| "loss": 0.8403, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.7321016166281755, | |
| "grad_norm": 18.26534652709961, | |
| "learning_rate": 4.7725500389914225e-05, | |
| "loss": 1.0399, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.7397998460354118, | |
| "grad_norm": 12.606345176696777, | |
| "learning_rate": 4.771250324928516e-05, | |
| "loss": 0.7362, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.7474980754426483, | |
| "grad_norm": 10.439709663391113, | |
| "learning_rate": 4.76995061086561e-05, | |
| "loss": 1.0418, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 1.7551963048498846, | |
| "grad_norm": 24.162647247314453, | |
| "learning_rate": 4.768650896802704e-05, | |
| "loss": 0.9489, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.762894534257121, | |
| "grad_norm": 34.678993225097656, | |
| "learning_rate": 4.767351182739798e-05, | |
| "loss": 0.8907, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 1.7705927636643572, | |
| "grad_norm": 22.55309295654297, | |
| "learning_rate": 4.7660514686768913e-05, | |
| "loss": 0.8239, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.7782909930715936, | |
| "grad_norm": 22.526445388793945, | |
| "learning_rate": 4.7647517546139855e-05, | |
| "loss": 1.0999, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 1.7859892224788299, | |
| "grad_norm": 19.45115089416504, | |
| "learning_rate": 4.763452040551079e-05, | |
| "loss": 0.8858, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.7936874518860662, | |
| "grad_norm": 29.1611270904541, | |
| "learning_rate": 4.762152326488173e-05, | |
| "loss": 1.0418, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 1.8013856812933025, | |
| "grad_norm": 22.248546600341797, | |
| "learning_rate": 4.760852612425267e-05, | |
| "loss": 0.8722, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.8090839107005388, | |
| "grad_norm": 78.22930908203125, | |
| "learning_rate": 4.75955289836236e-05, | |
| "loss": 0.8606, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.816782140107775, | |
| "grad_norm": 16.559967041015625, | |
| "learning_rate": 4.7582531842994544e-05, | |
| "loss": 0.9108, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.8244803695150116, | |
| "grad_norm": 20.415922164916992, | |
| "learning_rate": 4.756953470236548e-05, | |
| "loss": 1.0512, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 1.8321785989222479, | |
| "grad_norm": 34.354740142822266, | |
| "learning_rate": 4.755653756173642e-05, | |
| "loss": 0.9482, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.8398768283294842, | |
| "grad_norm": 26.14980697631836, | |
| "learning_rate": 4.7543540421107355e-05, | |
| "loss": 0.9213, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 1.8475750577367207, | |
| "grad_norm": 137.22198486328125, | |
| "learning_rate": 4.75305432804783e-05, | |
| "loss": 1.0509, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.855273287143957, | |
| "grad_norm": 82.37857818603516, | |
| "learning_rate": 4.751754613984923e-05, | |
| "loss": 0.8466, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 1.8629715165511933, | |
| "grad_norm": 31.272960662841797, | |
| "learning_rate": 4.7504548999220174e-05, | |
| "loss": 0.9717, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.8706697459584296, | |
| "grad_norm": 33.75693893432617, | |
| "learning_rate": 4.749155185859111e-05, | |
| "loss": 1.0565, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 1.878367975365666, | |
| "grad_norm": 21.18593406677246, | |
| "learning_rate": 4.747855471796205e-05, | |
| "loss": 0.8595, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.8860662047729022, | |
| "grad_norm": 23.71518898010254, | |
| "learning_rate": 4.7465557577332985e-05, | |
| "loss": 0.8621, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.8937644341801385, | |
| "grad_norm": 51.083431243896484, | |
| "learning_rate": 4.745256043670393e-05, | |
| "loss": 0.9675, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.9014626635873748, | |
| "grad_norm": 59.243743896484375, | |
| "learning_rate": 4.743956329607487e-05, | |
| "loss": 0.7884, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 1.9091608929946111, | |
| "grad_norm": 25.169973373413086, | |
| "learning_rate": 4.7426566155445804e-05, | |
| "loss": 1.0393, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.9168591224018474, | |
| "grad_norm": 19.93684196472168, | |
| "learning_rate": 4.7413569014816746e-05, | |
| "loss": 1.0029, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 1.924557351809084, | |
| "grad_norm": 14.145678520202637, | |
| "learning_rate": 4.740057187418768e-05, | |
| "loss": 1.0533, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.924557351809084, | |
| "eval_accuracy": 0.5569668976135489, | |
| "eval_f1": 0.5318009748153606, | |
| "eval_loss": 0.9681846499443054, | |
| "eval_precision": 0.5749243005518604, | |
| "eval_recall": 0.5569668976135489, | |
| "eval_runtime": 8.0276, | |
| "eval_samples_per_second": 323.634, | |
| "eval_steps_per_second": 40.485, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.9322555812163202, | |
| "grad_norm": 15.758938789367676, | |
| "learning_rate": 4.738757473355862e-05, | |
| "loss": 0.8101, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 1.9399538106235565, | |
| "grad_norm": 13.67740535736084, | |
| "learning_rate": 4.737457759292956e-05, | |
| "loss": 0.7461, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.947652040030793, | |
| "grad_norm": 15.8422269821167, | |
| "learning_rate": 4.73615804523005e-05, | |
| "loss": 1.0919, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 1.9553502694380294, | |
| "grad_norm": 150.54856872558594, | |
| "learning_rate": 4.7348583311671434e-05, | |
| "loss": 0.8516, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.9630484988452657, | |
| "grad_norm": 73.50398254394531, | |
| "learning_rate": 4.7335586171042376e-05, | |
| "loss": 0.9784, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.970746728252502, | |
| "grad_norm": 16.76059913635254, | |
| "learning_rate": 4.732258903041331e-05, | |
| "loss": 0.8167, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.9784449576597383, | |
| "grad_norm": 29.237688064575195, | |
| "learning_rate": 4.730959188978425e-05, | |
| "loss": 1.0536, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 1.9861431870669746, | |
| "grad_norm": 21.804590225219727, | |
| "learning_rate": 4.729659474915519e-05, | |
| "loss": 0.9558, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.9938414164742109, | |
| "grad_norm": 21.874589920043945, | |
| "learning_rate": 4.728359760852612e-05, | |
| "loss": 0.9748, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 2.001539645881447, | |
| "grad_norm": 32.83283996582031, | |
| "learning_rate": 4.7270600467897064e-05, | |
| "loss": 1.0102, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.0092378752886835, | |
| "grad_norm": 16.153419494628906, | |
| "learning_rate": 4.7257603327268e-05, | |
| "loss": 0.8345, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 2.0169361046959198, | |
| "grad_norm": 54.94171142578125, | |
| "learning_rate": 4.724460618663894e-05, | |
| "loss": 0.8617, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 2.024634334103156, | |
| "grad_norm": 39.80078887939453, | |
| "learning_rate": 4.7231609046009876e-05, | |
| "loss": 0.8463, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 2.032332563510393, | |
| "grad_norm": 21.685455322265625, | |
| "learning_rate": 4.721861190538082e-05, | |
| "loss": 0.8706, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 2.040030792917629, | |
| "grad_norm": 26.26354217529297, | |
| "learning_rate": 4.720561476475175e-05, | |
| "loss": 0.8641, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.0477290223248654, | |
| "grad_norm": 139.40414428710938, | |
| "learning_rate": 4.7192617624122694e-05, | |
| "loss": 0.7537, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 2.0554272517321017, | |
| "grad_norm": 49.84522247314453, | |
| "learning_rate": 4.7179620483493636e-05, | |
| "loss": 0.7604, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 2.063125481139338, | |
| "grad_norm": 17.73415756225586, | |
| "learning_rate": 4.716662334286457e-05, | |
| "loss": 0.7963, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 2.0708237105465743, | |
| "grad_norm": 16.671457290649414, | |
| "learning_rate": 4.715362620223551e-05, | |
| "loss": 0.8546, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 2.0785219399538106, | |
| "grad_norm": 15.332406044006348, | |
| "learning_rate": 4.714062906160645e-05, | |
| "loss": 0.784, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.086220169361047, | |
| "grad_norm": 76.87459564208984, | |
| "learning_rate": 4.712763192097739e-05, | |
| "loss": 0.9221, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 2.0939183987682832, | |
| "grad_norm": 21.800127029418945, | |
| "learning_rate": 4.7114634780348324e-05, | |
| "loss": 1.0555, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 2.1016166281755195, | |
| "grad_norm": 31.496492385864258, | |
| "learning_rate": 4.7101637639719266e-05, | |
| "loss": 0.8673, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 2.109314857582756, | |
| "grad_norm": 9.187371253967285, | |
| "learning_rate": 4.70886404990902e-05, | |
| "loss": 0.9217, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 2.117013086989992, | |
| "grad_norm": 19.336244583129883, | |
| "learning_rate": 4.707564335846114e-05, | |
| "loss": 0.8739, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.1247113163972284, | |
| "grad_norm": 28.658281326293945, | |
| "learning_rate": 4.706264621783208e-05, | |
| "loss": 0.8937, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 2.132409545804465, | |
| "grad_norm": 32.35395431518555, | |
| "learning_rate": 4.704964907720302e-05, | |
| "loss": 0.9527, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 2.1401077752117015, | |
| "grad_norm": 39.60884094238281, | |
| "learning_rate": 4.7036651936573955e-05, | |
| "loss": 0.9266, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 2.147806004618938, | |
| "grad_norm": 16.985599517822266, | |
| "learning_rate": 4.7023654795944896e-05, | |
| "loss": 0.9875, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 2.155504234026174, | |
| "grad_norm": 14.943406105041504, | |
| "learning_rate": 4.701065765531583e-05, | |
| "loss": 0.8371, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.1632024634334104, | |
| "grad_norm": 18.414188385009766, | |
| "learning_rate": 4.699766051468677e-05, | |
| "loss": 0.8526, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 2.1709006928406467, | |
| "grad_norm": 12.915871620178223, | |
| "learning_rate": 4.698466337405771e-05, | |
| "loss": 0.9252, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 2.178598922247883, | |
| "grad_norm": 38.03502655029297, | |
| "learning_rate": 4.697166623342864e-05, | |
| "loss": 0.8437, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 2.1862971516551193, | |
| "grad_norm": 163.22010803222656, | |
| "learning_rate": 4.6958669092799585e-05, | |
| "loss": 0.8906, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 2.1939953810623556, | |
| "grad_norm": 15.631921768188477, | |
| "learning_rate": 4.694567195217052e-05, | |
| "loss": 0.9374, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.201693610469592, | |
| "grad_norm": 91.86471557617188, | |
| "learning_rate": 4.693267481154146e-05, | |
| "loss": 0.895, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 2.209391839876828, | |
| "grad_norm": 43.340171813964844, | |
| "learning_rate": 4.69196776709124e-05, | |
| "loss": 0.9779, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 2.2170900692840645, | |
| "grad_norm": 17.621532440185547, | |
| "learning_rate": 4.6906680530283345e-05, | |
| "loss": 0.9628, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 2.2247882986913012, | |
| "grad_norm": 22.588891983032227, | |
| "learning_rate": 4.689368338965428e-05, | |
| "loss": 0.8364, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 2.2324865280985375, | |
| "grad_norm": 17.839075088500977, | |
| "learning_rate": 4.688068624902522e-05, | |
| "loss": 0.8933, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.240184757505774, | |
| "grad_norm": 63.97893524169922, | |
| "learning_rate": 4.6867689108396157e-05, | |
| "loss": 0.8879, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 2.24788298691301, | |
| "grad_norm": 25.88020133972168, | |
| "learning_rate": 4.685469196776709e-05, | |
| "loss": 1.1249, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 2.2555812163202464, | |
| "grad_norm": 61.59385299682617, | |
| "learning_rate": 4.684169482713803e-05, | |
| "loss": 0.9538, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 2.2632794457274827, | |
| "grad_norm": 313.2900390625, | |
| "learning_rate": 4.682869768650897e-05, | |
| "loss": 0.875, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 2.270977675134719, | |
| "grad_norm": 24.295677185058594, | |
| "learning_rate": 4.681570054587991e-05, | |
| "loss": 1.0275, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.2786759045419553, | |
| "grad_norm": 20.744089126586914, | |
| "learning_rate": 4.6802703405250845e-05, | |
| "loss": 0.9133, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 2.2863741339491916, | |
| "grad_norm": 52.287078857421875, | |
| "learning_rate": 4.678970626462179e-05, | |
| "loss": 0.7897, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 2.294072363356428, | |
| "grad_norm": 21.226224899291992, | |
| "learning_rate": 4.677670912399272e-05, | |
| "loss": 0.9604, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 2.3017705927636642, | |
| "grad_norm": 12.787373542785645, | |
| "learning_rate": 4.6763711983363663e-05, | |
| "loss": 0.8511, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 2.3094688221709005, | |
| "grad_norm": 22.66034507751465, | |
| "learning_rate": 4.67507148427346e-05, | |
| "loss": 1.0062, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.3094688221709005, | |
| "eval_accuracy": 0.588144726712856, | |
| "eval_f1": 0.5671167090208166, | |
| "eval_loss": 0.9636054039001465, | |
| "eval_precision": 0.5876103570700653, | |
| "eval_recall": 0.588144726712856, | |
| "eval_runtime": 7.9352, | |
| "eval_samples_per_second": 327.401, | |
| "eval_steps_per_second": 40.957, | |
| "step": 3000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 38970, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 30, | |
| "save_steps": 500, | |
| "total_flos": 277929582871200.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |