| { | |
| "best_metric": 0.8819584171696848, | |
| "best_model_checkpoint": "PhoWhisper-medium-vispeech-classifier-v1/checkpoint-1469", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1469, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010211027910142955, | |
| "grad_norm": 79976.4921875, | |
| "learning_rate": 6.122448979591837e-08, | |
| "loss": 2.0788, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02042205582028591, | |
| "grad_norm": 49252.46484375, | |
| "learning_rate": 1.2244897959183673e-07, | |
| "loss": 2.0787, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03063308373042886, | |
| "grad_norm": 66237.7890625, | |
| "learning_rate": 1.836734693877551e-07, | |
| "loss": 2.0734, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.04084411164057182, | |
| "grad_norm": 66530.140625, | |
| "learning_rate": 2.4489795918367347e-07, | |
| "loss": 2.0745, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05105513955071477, | |
| "grad_norm": 75123.3515625, | |
| "learning_rate": 3.0612244897959183e-07, | |
| "loss": 2.0715, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06126616746085772, | |
| "grad_norm": 91761.9921875, | |
| "learning_rate": 3.673469387755102e-07, | |
| "loss": 2.0702, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07147719537100068, | |
| "grad_norm": 76544.7421875, | |
| "learning_rate": 4.2857142857142857e-07, | |
| "loss": 2.0659, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.08168822328114364, | |
| "grad_norm": 85211.5078125, | |
| "learning_rate": 4.897959183673469e-07, | |
| "loss": 2.0632, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09189925119128659, | |
| "grad_norm": 86731.6171875, | |
| "learning_rate": 5.510204081632653e-07, | |
| "loss": 2.0579, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.10211027910142954, | |
| "grad_norm": 73202.421875, | |
| "learning_rate": 6.122448979591837e-07, | |
| "loss": 2.0466, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1123213070115725, | |
| "grad_norm": 99795.828125, | |
| "learning_rate": 6.73469387755102e-07, | |
| "loss": 2.0266, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.12253233492171545, | |
| "grad_norm": 78665.6015625, | |
| "learning_rate": 7.346938775510204e-07, | |
| "loss": 2.0133, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.13274336283185842, | |
| "grad_norm": 109804.5546875, | |
| "learning_rate": 7.959183673469389e-07, | |
| "loss": 2.0033, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.14295439074200136, | |
| "grad_norm": 88988.0234375, | |
| "learning_rate": 8.571428571428571e-07, | |
| "loss": 1.9912, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1531654186521443, | |
| "grad_norm": 163108.234375, | |
| "learning_rate": 9.183673469387755e-07, | |
| "loss": 1.9714, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.16337644656228728, | |
| "grad_norm": 147248.25, | |
| "learning_rate": 9.795918367346939e-07, | |
| "loss": 1.9368, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.17358747447243023, | |
| "grad_norm": 161750.640625, | |
| "learning_rate": 1.0408163265306123e-06, | |
| "loss": 1.8955, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.18379850238257317, | |
| "grad_norm": 189184.296875, | |
| "learning_rate": 1.1020408163265306e-06, | |
| "loss": 1.8849, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.19400953029271614, | |
| "grad_norm": 160164.859375, | |
| "learning_rate": 1.163265306122449e-06, | |
| "loss": 1.8496, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.2042205582028591, | |
| "grad_norm": 147315.296875, | |
| "learning_rate": 1.2244897959183673e-06, | |
| "loss": 1.8118, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.21443158611300203, | |
| "grad_norm": 164890.859375, | |
| "learning_rate": 1.2857142857142856e-06, | |
| "loss": 1.7829, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.224642614023145, | |
| "grad_norm": 195027.578125, | |
| "learning_rate": 1.346938775510204e-06, | |
| "loss": 1.7938, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.23485364193328795, | |
| "grad_norm": 206460.03125, | |
| "learning_rate": 1.4081632653061225e-06, | |
| "loss": 1.7258, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.2450646698434309, | |
| "grad_norm": 247498.296875, | |
| "learning_rate": 1.4693877551020408e-06, | |
| "loss": 1.6498, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.25527569775357384, | |
| "grad_norm": 282144.96875, | |
| "learning_rate": 1.5306122448979593e-06, | |
| "loss": 1.6515, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.26548672566371684, | |
| "grad_norm": 244598.515625, | |
| "learning_rate": 1.5918367346938777e-06, | |
| "loss": 1.6327, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2756977535738598, | |
| "grad_norm": 169616.5, | |
| "learning_rate": 1.6530612244897958e-06, | |
| "loss": 1.576, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.2859087814840027, | |
| "grad_norm": 186143.078125, | |
| "learning_rate": 1.7142857142857143e-06, | |
| "loss": 1.6067, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.29611980939414567, | |
| "grad_norm": 239726.171875, | |
| "learning_rate": 1.7755102040816327e-06, | |
| "loss": 1.5676, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.3063308373042886, | |
| "grad_norm": 224399.15625, | |
| "learning_rate": 1.836734693877551e-06, | |
| "loss": 1.4658, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.31654186521443156, | |
| "grad_norm": 585805.375, | |
| "learning_rate": 1.8979591836734695e-06, | |
| "loss": 1.4288, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.32675289312457456, | |
| "grad_norm": 204082.328125, | |
| "learning_rate": 1.9591836734693877e-06, | |
| "loss": 1.4079, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3369639210347175, | |
| "grad_norm": 253897.078125, | |
| "learning_rate": 2.020408163265306e-06, | |
| "loss": 1.4658, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.34717494894486045, | |
| "grad_norm": 322156.15625, | |
| "learning_rate": 2.0816326530612247e-06, | |
| "loss": 1.3193, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3573859768550034, | |
| "grad_norm": 209002.03125, | |
| "learning_rate": 2.142857142857143e-06, | |
| "loss": 1.3258, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.36759700476514634, | |
| "grad_norm": 239622.390625, | |
| "learning_rate": 2.204081632653061e-06, | |
| "loss": 1.3396, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.3778080326752893, | |
| "grad_norm": 302531.125, | |
| "learning_rate": 2.2653061224489797e-06, | |
| "loss": 1.2724, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.3880190605854323, | |
| "grad_norm": 277807.9375, | |
| "learning_rate": 2.326530612244898e-06, | |
| "loss": 1.2506, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.39823008849557523, | |
| "grad_norm": 543051.25, | |
| "learning_rate": 2.387755102040816e-06, | |
| "loss": 1.3114, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.4084411164057182, | |
| "grad_norm": 278458.59375, | |
| "learning_rate": 2.4489795918367347e-06, | |
| "loss": 1.2505, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4186521443158611, | |
| "grad_norm": 227854.078125, | |
| "learning_rate": 2.510204081632653e-06, | |
| "loss": 1.2032, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.42886317222600406, | |
| "grad_norm": 669112.0625, | |
| "learning_rate": 2.571428571428571e-06, | |
| "loss": 1.2144, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.43907420013614706, | |
| "grad_norm": 291158.375, | |
| "learning_rate": 2.6326530612244897e-06, | |
| "loss": 1.1663, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.44928522804629, | |
| "grad_norm": 322840.65625, | |
| "learning_rate": 2.693877551020408e-06, | |
| "loss": 1.1802, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.45949625595643295, | |
| "grad_norm": 399617.4375, | |
| "learning_rate": 2.7551020408163266e-06, | |
| "loss": 1.0181, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4697072838665759, | |
| "grad_norm": 399545.5, | |
| "learning_rate": 2.816326530612245e-06, | |
| "loss": 1.2468, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.47991831177671884, | |
| "grad_norm": 1116027.875, | |
| "learning_rate": 2.877551020408163e-06, | |
| "loss": 1.2023, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.4901293396868618, | |
| "grad_norm": 300487.4375, | |
| "learning_rate": 2.9387755102040816e-06, | |
| "loss": 1.147, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5003403675970047, | |
| "grad_norm": 413331.65625, | |
| "learning_rate": 3e-06, | |
| "loss": 0.9871, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.5105513955071477, | |
| "grad_norm": 715814.125, | |
| "learning_rate": 2.993192133131619e-06, | |
| "loss": 1.1643, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5207624234172906, | |
| "grad_norm": 1614147.0, | |
| "learning_rate": 2.9863842662632373e-06, | |
| "loss": 0.9421, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.5309734513274337, | |
| "grad_norm": 639895.0625, | |
| "learning_rate": 2.9795763993948565e-06, | |
| "loss": 1.0573, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5411844792375766, | |
| "grad_norm": 345273.84375, | |
| "learning_rate": 2.9727685325264754e-06, | |
| "loss": 1.0337, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.5513955071477196, | |
| "grad_norm": 611370.375, | |
| "learning_rate": 2.9659606656580938e-06, | |
| "loss": 1.1131, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5616065350578625, | |
| "grad_norm": 239903.25, | |
| "learning_rate": 2.9591527987897126e-06, | |
| "loss": 0.9076, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5718175629680055, | |
| "grad_norm": 257958.703125, | |
| "learning_rate": 2.9523449319213314e-06, | |
| "loss": 0.9152, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.5820285908781484, | |
| "grad_norm": 590921.375, | |
| "learning_rate": 2.9455370650529502e-06, | |
| "loss": 0.8795, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.5922396187882913, | |
| "grad_norm": 363986.1875, | |
| "learning_rate": 2.938729198184569e-06, | |
| "loss": 1.0641, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6024506466984343, | |
| "grad_norm": 1878926.875, | |
| "learning_rate": 2.9319213313161875e-06, | |
| "loss": 0.9414, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.6126616746085772, | |
| "grad_norm": 2941328.0, | |
| "learning_rate": 2.9251134644478063e-06, | |
| "loss": 0.9541, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6228727025187202, | |
| "grad_norm": 1327942.75, | |
| "learning_rate": 2.918305597579425e-06, | |
| "loss": 0.8319, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.6330837304288631, | |
| "grad_norm": 161807.125, | |
| "learning_rate": 2.911497730711044e-06, | |
| "loss": 0.9268, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6432947583390062, | |
| "grad_norm": 899555.875, | |
| "learning_rate": 2.9046898638426628e-06, | |
| "loss": 0.8278, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.6535057862491491, | |
| "grad_norm": 268360.0625, | |
| "learning_rate": 2.8978819969742816e-06, | |
| "loss": 0.9021, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6637168141592921, | |
| "grad_norm": 555703.0625, | |
| "learning_rate": 2.8910741301059e-06, | |
| "loss": 1.0146, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.673927842069435, | |
| "grad_norm": 222425.40625, | |
| "learning_rate": 2.884266263237519e-06, | |
| "loss": 1.0005, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.684138869979578, | |
| "grad_norm": 681254.4375, | |
| "learning_rate": 2.877458396369138e-06, | |
| "loss": 0.842, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.6943498978897209, | |
| "grad_norm": 1348221.125, | |
| "learning_rate": 2.8706505295007565e-06, | |
| "loss": 0.9394, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7045609257998638, | |
| "grad_norm": 225728.046875, | |
| "learning_rate": 2.8638426626323753e-06, | |
| "loss": 0.8289, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.7147719537100068, | |
| "grad_norm": 413089.65625, | |
| "learning_rate": 2.857034795763994e-06, | |
| "loss": 1.0175, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7249829816201497, | |
| "grad_norm": 279751.71875, | |
| "learning_rate": 2.8502269288956125e-06, | |
| "loss": 0.9563, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.7351940095302927, | |
| "grad_norm": 422167.125, | |
| "learning_rate": 2.8434190620272318e-06, | |
| "loss": 0.8566, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.7454050374404356, | |
| "grad_norm": 2015877.125, | |
| "learning_rate": 2.83661119515885e-06, | |
| "loss": 0.7012, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.7556160653505786, | |
| "grad_norm": 502859.5, | |
| "learning_rate": 2.829803328290469e-06, | |
| "loss": 0.7563, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.7658270932607216, | |
| "grad_norm": 537785.125, | |
| "learning_rate": 2.822995461422088e-06, | |
| "loss": 0.9854, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.7760381211708646, | |
| "grad_norm": 390231.125, | |
| "learning_rate": 2.8161875945537066e-06, | |
| "loss": 1.0036, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7862491490810075, | |
| "grad_norm": 370246.0625, | |
| "learning_rate": 2.8093797276853255e-06, | |
| "loss": 0.8186, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.7964601769911505, | |
| "grad_norm": 1279901.375, | |
| "learning_rate": 2.8025718608169443e-06, | |
| "loss": 0.9289, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8066712049012934, | |
| "grad_norm": 316319.25, | |
| "learning_rate": 2.7957639939485627e-06, | |
| "loss": 0.7542, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.8168822328114363, | |
| "grad_norm": 577071.5, | |
| "learning_rate": 2.7889561270801815e-06, | |
| "loss": 0.8579, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8270932607215793, | |
| "grad_norm": 412667.71875, | |
| "learning_rate": 2.7821482602118003e-06, | |
| "loss": 0.8424, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.8373042886317222, | |
| "grad_norm": 555197.125, | |
| "learning_rate": 2.775340393343419e-06, | |
| "loss": 0.8225, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.8475153165418652, | |
| "grad_norm": 429715.0, | |
| "learning_rate": 2.768532526475038e-06, | |
| "loss": 1.1042, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.8577263444520081, | |
| "grad_norm": 1407773.25, | |
| "learning_rate": 2.761724659606657e-06, | |
| "loss": 0.9892, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.8679373723621511, | |
| "grad_norm": 328839.6875, | |
| "learning_rate": 2.754916792738275e-06, | |
| "loss": 0.7921, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.8781484002722941, | |
| "grad_norm": 196251.03125, | |
| "learning_rate": 2.748108925869894e-06, | |
| "loss": 0.6631, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.8883594281824371, | |
| "grad_norm": 761081.6875, | |
| "learning_rate": 2.7413010590015133e-06, | |
| "loss": 0.8396, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.89857045609258, | |
| "grad_norm": 905756.375, | |
| "learning_rate": 2.7344931921331317e-06, | |
| "loss": 0.8238, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.908781484002723, | |
| "grad_norm": 1221640.875, | |
| "learning_rate": 2.7276853252647505e-06, | |
| "loss": 0.7228, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.9189925119128659, | |
| "grad_norm": 2633817.0, | |
| "learning_rate": 2.7208774583963693e-06, | |
| "loss": 0.8766, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9292035398230089, | |
| "grad_norm": 454206.53125, | |
| "learning_rate": 2.7140695915279877e-06, | |
| "loss": 0.922, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.9394145677331518, | |
| "grad_norm": 715028.8125, | |
| "learning_rate": 2.707261724659607e-06, | |
| "loss": 0.8212, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.9496255956432947, | |
| "grad_norm": 384052.40625, | |
| "learning_rate": 2.7004538577912254e-06, | |
| "loss": 0.6728, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.9598366235534377, | |
| "grad_norm": 268327.0, | |
| "learning_rate": 2.693645990922844e-06, | |
| "loss": 0.7077, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.9700476514635806, | |
| "grad_norm": 430912.0625, | |
| "learning_rate": 2.686838124054463e-06, | |
| "loss": 0.8267, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.9802586793737236, | |
| "grad_norm": 448801.65625, | |
| "learning_rate": 2.680030257186082e-06, | |
| "loss": 0.6444, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9904697072838666, | |
| "grad_norm": 625466.3125, | |
| "learning_rate": 2.6732223903177007e-06, | |
| "loss": 0.8255, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8819584171696848, | |
| "eval_loss": 0.5119155645370483, | |
| "eval_runtime": 579.0751, | |
| "eval_samples_per_second": 5.15, | |
| "eval_steps_per_second": 0.644, | |
| "step": 1469 | |
| } | |
| ], | |
| "logging_steps": 15, | |
| "max_steps": 7345, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.20189308048384e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |