| { | |
| "best_metric": 0.3894089162349701, | |
| "best_model_checkpoint": "prostate-mri-T2w-v05/checkpoint-6223", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 6223, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.025871766029246346, | |
| "grad_norm": 62.908626556396484, | |
| "learning_rate": 2.5871766029246344e-07, | |
| "loss": 0.9082, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05174353205849269, | |
| "grad_norm": 17.856958389282227, | |
| "learning_rate": 5.174353205849269e-07, | |
| "loss": 0.9712, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.07761529808773904, | |
| "grad_norm": 14.738483428955078, | |
| "learning_rate": 7.761529808773904e-07, | |
| "loss": 0.7274, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.10348706411698538, | |
| "grad_norm": 11.495190620422363, | |
| "learning_rate": 1.0348706411698538e-06, | |
| "loss": 0.5668, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12935883014623173, | |
| "grad_norm": 14.365918159484863, | |
| "learning_rate": 1.2935883014623174e-06, | |
| "loss": 0.5766, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15523059617547807, | |
| "grad_norm": 4.656925678253174, | |
| "learning_rate": 1.5523059617547808e-06, | |
| "loss": 0.5059, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.18110236220472442, | |
| "grad_norm": 9.39604663848877, | |
| "learning_rate": 1.8110236220472441e-06, | |
| "loss": 0.4405, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.20697412823397077, | |
| "grad_norm": 9.823530197143555, | |
| "learning_rate": 2.0697412823397075e-06, | |
| "loss": 0.3319, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.23284589426321708, | |
| "grad_norm": 10.687150955200195, | |
| "learning_rate": 2.328458942632171e-06, | |
| "loss": 0.4602, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.25871766029246346, | |
| "grad_norm": 1.1277879476547241, | |
| "learning_rate": 2.5871766029246347e-06, | |
| "loss": 0.3568, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.2845894263217098, | |
| "grad_norm": 0.47847771644592285, | |
| "learning_rate": 2.845894263217098e-06, | |
| "loss": 0.6281, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.31046119235095615, | |
| "grad_norm": 0.5296943187713623, | |
| "learning_rate": 3.1046119235095615e-06, | |
| "loss": 0.3346, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.3363329583802025, | |
| "grad_norm": 11.51565170288086, | |
| "learning_rate": 3.3633295838020245e-06, | |
| "loss": 0.5892, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.36220472440944884, | |
| "grad_norm": 18.972782135009766, | |
| "learning_rate": 3.6220472440944883e-06, | |
| "loss": 0.7544, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.3880764904386952, | |
| "grad_norm": 10.762083053588867, | |
| "learning_rate": 3.880764904386952e-06, | |
| "loss": 0.6232, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.41394825646794153, | |
| "grad_norm": 11.391585350036621, | |
| "learning_rate": 4.139482564679415e-06, | |
| "loss": 0.4856, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.4398200224971879, | |
| "grad_norm": 1.076779842376709, | |
| "learning_rate": 4.3982002249718785e-06, | |
| "loss": 0.3987, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.46569178852643417, | |
| "grad_norm": 11.604365348815918, | |
| "learning_rate": 4.656917885264342e-06, | |
| "loss": 0.5967, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.4915635545556805, | |
| "grad_norm": 10.939277648925781, | |
| "learning_rate": 4.915635545556806e-06, | |
| "loss": 0.3712, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.5174353205849269, | |
| "grad_norm": 10.872827529907227, | |
| "learning_rate": 5.1743532058492695e-06, | |
| "loss": 0.4234, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5433070866141733, | |
| "grad_norm": 13.358455657958984, | |
| "learning_rate": 5.433070866141733e-06, | |
| "loss": 0.6415, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.5691788526434196, | |
| "grad_norm": 0.5069867968559265, | |
| "learning_rate": 5.691788526434196e-06, | |
| "loss": 0.3304, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.595050618672666, | |
| "grad_norm": 11.854174613952637, | |
| "learning_rate": 5.95050618672666e-06, | |
| "loss": 0.2956, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.6209223847019123, | |
| "grad_norm": 0.6237831711769104, | |
| "learning_rate": 6.209223847019123e-06, | |
| "loss": 0.5922, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.6467941507311586, | |
| "grad_norm": 18.399309158325195, | |
| "learning_rate": 6.467941507311586e-06, | |
| "loss": 0.5529, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.672665916760405, | |
| "grad_norm": 1.0723528861999512, | |
| "learning_rate": 6.726659167604049e-06, | |
| "loss": 0.5464, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.6985376827896513, | |
| "grad_norm": 10.857867240905762, | |
| "learning_rate": 6.985376827896513e-06, | |
| "loss": 0.5258, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.7244094488188977, | |
| "grad_norm": 11.417038917541504, | |
| "learning_rate": 7.244094488188977e-06, | |
| "loss": 0.4599, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.750281214848144, | |
| "grad_norm": 12.366191864013672, | |
| "learning_rate": 7.50281214848144e-06, | |
| "loss": 0.5278, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.7761529808773904, | |
| "grad_norm": 59.656211853027344, | |
| "learning_rate": 7.761529808773903e-06, | |
| "loss": 0.4802, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.8020247469066367, | |
| "grad_norm": 1.6525479555130005, | |
| "learning_rate": 8.020247469066367e-06, | |
| "loss": 0.3855, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.8278965129358831, | |
| "grad_norm": 11.62271499633789, | |
| "learning_rate": 8.27896512935883e-06, | |
| "loss": 0.5377, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.8537682789651294, | |
| "grad_norm": 11.242469787597656, | |
| "learning_rate": 8.537682789651294e-06, | |
| "loss": 0.6707, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.8796400449943758, | |
| "grad_norm": 1.673377513885498, | |
| "learning_rate": 8.796400449943757e-06, | |
| "loss": 0.434, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.905511811023622, | |
| "grad_norm": 0.3748597800731659, | |
| "learning_rate": 9.055118110236222e-06, | |
| "loss": 0.4709, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.9313835770528683, | |
| "grad_norm": 10.111214637756348, | |
| "learning_rate": 9.313835770528684e-06, | |
| "loss": 0.5416, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.9572553430821147, | |
| "grad_norm": 11.05431842803955, | |
| "learning_rate": 9.572553430821147e-06, | |
| "loss": 0.4895, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.983127109111361, | |
| "grad_norm": 0.5102259516716003, | |
| "learning_rate": 9.831271091113612e-06, | |
| "loss": 0.6304, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7857142857142857, | |
| "eval_auc": 0.8011167033658384, | |
| "eval_f1": 0.0, | |
| "eval_loss": 0.7771126627922058, | |
| "eval_precision": 0.0, | |
| "eval_recall": 0.0, | |
| "eval_runtime": 2.7441, | |
| "eval_samples_per_second": 173.466, | |
| "eval_steps_per_second": 21.865, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.0089988751406074, | |
| "grad_norm": 13.26599407196045, | |
| "learning_rate": 1.0089988751406076e-05, | |
| "loss": 0.3247, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.0348706411698538, | |
| "grad_norm": 11.029207229614258, | |
| "learning_rate": 1.0348706411698539e-05, | |
| "loss": 0.6095, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.0607424071991, | |
| "grad_norm": 0.07530930638313293, | |
| "learning_rate": 1.0607424071991e-05, | |
| "loss": 0.1472, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 1.0866141732283465, | |
| "grad_norm": 0.6263224482536316, | |
| "learning_rate": 1.0866141732283466e-05, | |
| "loss": 0.6615, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 1.1124859392575928, | |
| "grad_norm": 9.975605964660645, | |
| "learning_rate": 1.1124859392575929e-05, | |
| "loss": 0.4082, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 1.1383577052868392, | |
| "grad_norm": 10.850635528564453, | |
| "learning_rate": 1.1383577052868393e-05, | |
| "loss": 0.4151, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 1.1642294713160855, | |
| "grad_norm": 10.857563972473145, | |
| "learning_rate": 1.1642294713160854e-05, | |
| "loss": 0.4441, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.190101237345332, | |
| "grad_norm": 4.742630958557129, | |
| "learning_rate": 1.190101237345332e-05, | |
| "loss": 0.7853, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 1.2159730033745781, | |
| "grad_norm": 0.09981119632720947, | |
| "learning_rate": 1.2159730033745783e-05, | |
| "loss": 0.2316, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 1.2418447694038246, | |
| "grad_norm": 0.37830597162246704, | |
| "learning_rate": 1.2418447694038246e-05, | |
| "loss": 0.5479, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 1.2677165354330708, | |
| "grad_norm": 1.0691417455673218, | |
| "learning_rate": 1.2677165354330708e-05, | |
| "loss": 0.342, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 1.2935883014623173, | |
| "grad_norm": 10.283402442932129, | |
| "learning_rate": 1.2935883014623173e-05, | |
| "loss": 0.591, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.3194600674915635, | |
| "grad_norm": 2.4355926513671875, | |
| "learning_rate": 1.3194600674915636e-05, | |
| "loss": 0.486, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 1.34533183352081, | |
| "grad_norm": 7.244462966918945, | |
| "learning_rate": 1.3453318335208098e-05, | |
| "loss": 0.5206, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 1.3712035995500562, | |
| "grad_norm": 9.888749122619629, | |
| "learning_rate": 1.3712035995500563e-05, | |
| "loss": 0.4504, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 1.3970753655793025, | |
| "grad_norm": 3.3459055423736572, | |
| "learning_rate": 1.3970753655793026e-05, | |
| "loss": 0.4415, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 1.422947131608549, | |
| "grad_norm": 0.175747811794281, | |
| "learning_rate": 1.4229471316085491e-05, | |
| "loss": 0.2943, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.4488188976377954, | |
| "grad_norm": 4.791985988616943, | |
| "learning_rate": 1.4488188976377953e-05, | |
| "loss": 0.566, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 1.4746906636670416, | |
| "grad_norm": 12.70669174194336, | |
| "learning_rate": 1.4746906636670415e-05, | |
| "loss": 0.4687, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 1.5005624296962878, | |
| "grad_norm": 0.04389326646924019, | |
| "learning_rate": 1.500562429696288e-05, | |
| "loss": 0.2629, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 1.5264341957255343, | |
| "grad_norm": 4.369255542755127, | |
| "learning_rate": 1.5264341957255343e-05, | |
| "loss": 0.472, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 1.5523059617547807, | |
| "grad_norm": 4.454276084899902, | |
| "learning_rate": 1.5523059617547807e-05, | |
| "loss": 0.4307, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.578177727784027, | |
| "grad_norm": 1.1173179149627686, | |
| "learning_rate": 1.578177727784027e-05, | |
| "loss": 0.3431, | |
| "step": 1403 | |
| }, | |
| { | |
| "epoch": 1.6040494938132732, | |
| "grad_norm": 3.0332882404327393, | |
| "learning_rate": 1.6040494938132734e-05, | |
| "loss": 0.2545, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 1.6299212598425197, | |
| "grad_norm": 10.521413803100586, | |
| "learning_rate": 1.6299212598425197e-05, | |
| "loss": 0.6555, | |
| "step": 1449 | |
| }, | |
| { | |
| "epoch": 1.6557930258717661, | |
| "grad_norm": 3.8012301921844482, | |
| "learning_rate": 1.655793025871766e-05, | |
| "loss": 0.37, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 1.6816647919010124, | |
| "grad_norm": 0.05941835045814514, | |
| "learning_rate": 1.6816647919010127e-05, | |
| "loss": 0.3091, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 1.7075365579302586, | |
| "grad_norm": 0.29451972246170044, | |
| "learning_rate": 1.7075365579302587e-05, | |
| "loss": 0.3511, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 1.733408323959505, | |
| "grad_norm": 3.2587544918060303, | |
| "learning_rate": 1.733408323959505e-05, | |
| "loss": 0.6164, | |
| "step": 1541 | |
| }, | |
| { | |
| "epoch": 1.7592800899887515, | |
| "grad_norm": 3.4113192558288574, | |
| "learning_rate": 1.7592800899887514e-05, | |
| "loss": 0.3817, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 1.7851518560179978, | |
| "grad_norm": 11.975400924682617, | |
| "learning_rate": 1.7851518560179977e-05, | |
| "loss": 0.3421, | |
| "step": 1587 | |
| }, | |
| { | |
| "epoch": 1.811023622047244, | |
| "grad_norm": 1.5508025884628296, | |
| "learning_rate": 1.8110236220472444e-05, | |
| "loss": 0.6197, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.8368953880764904, | |
| "grad_norm": 0.24332457780838013, | |
| "learning_rate": 1.8368953880764904e-05, | |
| "loss": 0.4227, | |
| "step": 1633 | |
| }, | |
| { | |
| "epoch": 1.862767154105737, | |
| "grad_norm": 0.10154075175523758, | |
| "learning_rate": 1.8627671541057367e-05, | |
| "loss": 0.4287, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 1.8886389201349831, | |
| "grad_norm": 10.398853302001953, | |
| "learning_rate": 1.8886389201349834e-05, | |
| "loss": 0.3014, | |
| "step": 1679 | |
| }, | |
| { | |
| "epoch": 1.9145106861642294, | |
| "grad_norm": 0.8489835858345032, | |
| "learning_rate": 1.9145106861642294e-05, | |
| "loss": 0.3982, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 1.9403824521934758, | |
| "grad_norm": 0.8622331619262695, | |
| "learning_rate": 1.940382452193476e-05, | |
| "loss": 0.5771, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.9662542182227223, | |
| "grad_norm": 2.847560405731201, | |
| "learning_rate": 1.9662542182227224e-05, | |
| "loss": 0.4758, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 1.9921259842519685, | |
| "grad_norm": 0.5839571952819824, | |
| "learning_rate": 1.9921259842519684e-05, | |
| "loss": 0.2732, | |
| "step": 1771 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7857142857142857, | |
| "eval_auc": 0.7965817342979973, | |
| "eval_f1": 0.0, | |
| "eval_loss": 0.9842509627342224, | |
| "eval_precision": 0.0, | |
| "eval_recall": 0.0, | |
| "eval_runtime": 2.74, | |
| "eval_samples_per_second": 173.72, | |
| "eval_steps_per_second": 21.897, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 2.0179977502812148, | |
| "grad_norm": 0.23481443524360657, | |
| "learning_rate": 2.017997750281215e-05, | |
| "loss": 0.2917, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 2.043869516310461, | |
| "grad_norm": 0.1679057627916336, | |
| "learning_rate": 2.043869516310461e-05, | |
| "loss": 0.5162, | |
| "step": 1817 | |
| }, | |
| { | |
| "epoch": 2.0697412823397077, | |
| "grad_norm": 5.455684185028076, | |
| "learning_rate": 2.0697412823397078e-05, | |
| "loss": 0.5789, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.095613048368954, | |
| "grad_norm": 7.542656421661377, | |
| "learning_rate": 2.095613048368954e-05, | |
| "loss": 0.5314, | |
| "step": 1863 | |
| }, | |
| { | |
| "epoch": 2.1214848143982, | |
| "grad_norm": 2.186872720718384, | |
| "learning_rate": 2.1214848143982e-05, | |
| "loss": 0.3816, | |
| "step": 1886 | |
| }, | |
| { | |
| "epoch": 2.1473565804274464, | |
| "grad_norm": 0.19364489614963531, | |
| "learning_rate": 2.1473565804274468e-05, | |
| "loss": 0.336, | |
| "step": 1909 | |
| }, | |
| { | |
| "epoch": 2.173228346456693, | |
| "grad_norm": 0.20303942263126373, | |
| "learning_rate": 2.173228346456693e-05, | |
| "loss": 0.421, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 2.1991001124859393, | |
| "grad_norm": 0.5988159775733948, | |
| "learning_rate": 2.1991001124859395e-05, | |
| "loss": 0.4587, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 2.2249718785151855, | |
| "grad_norm": 7.407308101654053, | |
| "learning_rate": 2.2249718785151858e-05, | |
| "loss": 0.6277, | |
| "step": 1978 | |
| }, | |
| { | |
| "epoch": 2.2508436445444318, | |
| "grad_norm": 2.7972147464752197, | |
| "learning_rate": 2.2508436445444318e-05, | |
| "loss": 0.4025, | |
| "step": 2001 | |
| }, | |
| { | |
| "epoch": 2.2767154105736784, | |
| "grad_norm": 14.71440601348877, | |
| "learning_rate": 2.2767154105736785e-05, | |
| "loss": 0.4362, | |
| "step": 2024 | |
| }, | |
| { | |
| "epoch": 2.3025871766029247, | |
| "grad_norm": 0.6409655213356018, | |
| "learning_rate": 2.302587176602925e-05, | |
| "loss": 0.4014, | |
| "step": 2047 | |
| }, | |
| { | |
| "epoch": 2.328458942632171, | |
| "grad_norm": 0.0657491534948349, | |
| "learning_rate": 2.328458942632171e-05, | |
| "loss": 0.3942, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.354330708661417, | |
| "grad_norm": 0.9892528653144836, | |
| "learning_rate": 2.3543307086614175e-05, | |
| "loss": 0.5464, | |
| "step": 2093 | |
| }, | |
| { | |
| "epoch": 2.380202474690664, | |
| "grad_norm": 0.6635844111442566, | |
| "learning_rate": 2.380202474690664e-05, | |
| "loss": 0.4767, | |
| "step": 2116 | |
| }, | |
| { | |
| "epoch": 2.40607424071991, | |
| "grad_norm": 7.151478290557861, | |
| "learning_rate": 2.4060742407199102e-05, | |
| "loss": 0.4637, | |
| "step": 2139 | |
| }, | |
| { | |
| "epoch": 2.4319460067491563, | |
| "grad_norm": 0.7382416129112244, | |
| "learning_rate": 2.4319460067491565e-05, | |
| "loss": 0.3082, | |
| "step": 2162 | |
| }, | |
| { | |
| "epoch": 2.4578177727784025, | |
| "grad_norm": 9.286602973937988, | |
| "learning_rate": 2.4578177727784025e-05, | |
| "loss": 0.5282, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 2.483689538807649, | |
| "grad_norm": 4.055200099945068, | |
| "learning_rate": 2.4836895388076492e-05, | |
| "loss": 0.457, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 2.5095613048368954, | |
| "grad_norm": 0.12764699757099152, | |
| "learning_rate": 2.5095613048368956e-05, | |
| "loss": 0.3318, | |
| "step": 2231 | |
| }, | |
| { | |
| "epoch": 2.5354330708661417, | |
| "grad_norm": 2.9867072105407715, | |
| "learning_rate": 2.5354330708661416e-05, | |
| "loss": 0.3542, | |
| "step": 2254 | |
| }, | |
| { | |
| "epoch": 2.5613048368953883, | |
| "grad_norm": 4.703469753265381, | |
| "learning_rate": 2.5613048368953886e-05, | |
| "loss": 0.4748, | |
| "step": 2277 | |
| }, | |
| { | |
| "epoch": 2.5871766029246346, | |
| "grad_norm": 9.484485626220703, | |
| "learning_rate": 2.5871766029246346e-05, | |
| "loss": 0.4135, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.613048368953881, | |
| "grad_norm": 18.093868255615234, | |
| "learning_rate": 2.613048368953881e-05, | |
| "loss": 0.4656, | |
| "step": 2323 | |
| }, | |
| { | |
| "epoch": 2.638920134983127, | |
| "grad_norm": 0.21779067814350128, | |
| "learning_rate": 2.6389201349831272e-05, | |
| "loss": 0.3057, | |
| "step": 2346 | |
| }, | |
| { | |
| "epoch": 2.6647919010123733, | |
| "grad_norm": 0.7615957260131836, | |
| "learning_rate": 2.6647919010123736e-05, | |
| "loss": 0.4652, | |
| "step": 2369 | |
| }, | |
| { | |
| "epoch": 2.69066366704162, | |
| "grad_norm": 0.35096365213394165, | |
| "learning_rate": 2.6906636670416196e-05, | |
| "loss": 0.3751, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 2.716535433070866, | |
| "grad_norm": 0.5412973761558533, | |
| "learning_rate": 2.7165354330708666e-05, | |
| "loss": 0.1823, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 2.7424071991001124, | |
| "grad_norm": 1.9220446348190308, | |
| "learning_rate": 2.7424071991001126e-05, | |
| "loss": 0.6485, | |
| "step": 2438 | |
| }, | |
| { | |
| "epoch": 2.7682789651293587, | |
| "grad_norm": 11.08309268951416, | |
| "learning_rate": 2.768278965129359e-05, | |
| "loss": 0.4777, | |
| "step": 2461 | |
| }, | |
| { | |
| "epoch": 2.794150731158605, | |
| "grad_norm": 0.0344441793859005, | |
| "learning_rate": 2.7941507311586053e-05, | |
| "loss": 0.2201, | |
| "step": 2484 | |
| }, | |
| { | |
| "epoch": 2.8200224971878516, | |
| "grad_norm": 0.9013801217079163, | |
| "learning_rate": 2.8200224971878513e-05, | |
| "loss": 0.471, | |
| "step": 2507 | |
| }, | |
| { | |
| "epoch": 2.845894263217098, | |
| "grad_norm": 1.6073826551437378, | |
| "learning_rate": 2.8458942632170983e-05, | |
| "loss": 0.5535, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 2.871766029246344, | |
| "grad_norm": 3.7488861083984375, | |
| "learning_rate": 2.8717660292463443e-05, | |
| "loss": 0.525, | |
| "step": 2553 | |
| }, | |
| { | |
| "epoch": 2.8976377952755907, | |
| "grad_norm": 0.2803422808647156, | |
| "learning_rate": 2.8976377952755906e-05, | |
| "loss": 0.4241, | |
| "step": 2576 | |
| }, | |
| { | |
| "epoch": 2.923509561304837, | |
| "grad_norm": 2.339632749557495, | |
| "learning_rate": 2.923509561304837e-05, | |
| "loss": 0.3422, | |
| "step": 2599 | |
| }, | |
| { | |
| "epoch": 2.949381327334083, | |
| "grad_norm": 0.10544496774673462, | |
| "learning_rate": 2.949381327334083e-05, | |
| "loss": 0.2698, | |
| "step": 2622 | |
| }, | |
| { | |
| "epoch": 2.9752530933633294, | |
| "grad_norm": 5.151175498962402, | |
| "learning_rate": 2.97525309336333e-05, | |
| "loss": 0.5485, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7815126050420168, | |
| "eval_auc": 0.7921778337003251, | |
| "eval_f1": 0.0, | |
| "eval_loss": 0.7073759436607361, | |
| "eval_precision": 0.0, | |
| "eval_recall": 0.0, | |
| "eval_runtime": 2.7321, | |
| "eval_samples_per_second": 174.227, | |
| "eval_steps_per_second": 21.961, | |
| "step": 2667 | |
| }, | |
| { | |
| "epoch": 3.001124859392576, | |
| "grad_norm": 7.475840091705322, | |
| "learning_rate": 3.001124859392576e-05, | |
| "loss": 0.2998, | |
| "step": 2668 | |
| }, | |
| { | |
| "epoch": 3.0269966254218224, | |
| "grad_norm": 0.36270514130592346, | |
| "learning_rate": 3.0269966254218223e-05, | |
| "loss": 0.3207, | |
| "step": 2691 | |
| }, | |
| { | |
| "epoch": 3.0528683914510686, | |
| "grad_norm": 3.655841112136841, | |
| "learning_rate": 3.052868391451069e-05, | |
| "loss": 0.376, | |
| "step": 2714 | |
| }, | |
| { | |
| "epoch": 3.078740157480315, | |
| "grad_norm": 0.28142863512039185, | |
| "learning_rate": 3.078740157480315e-05, | |
| "loss": 0.3376, | |
| "step": 2737 | |
| }, | |
| { | |
| "epoch": 3.1046119235095615, | |
| "grad_norm": 0.1056133359670639, | |
| "learning_rate": 3.1046119235095613e-05, | |
| "loss": 0.43, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 3.1304836895388077, | |
| "grad_norm": 10.104964256286621, | |
| "learning_rate": 3.130483689538808e-05, | |
| "loss": 0.4603, | |
| "step": 2783 | |
| }, | |
| { | |
| "epoch": 3.156355455568054, | |
| "grad_norm": 8.072303771972656, | |
| "learning_rate": 3.156355455568054e-05, | |
| "loss": 0.3976, | |
| "step": 2806 | |
| }, | |
| { | |
| "epoch": 3.1822272215973, | |
| "grad_norm": 10.715200424194336, | |
| "learning_rate": 3.1822272215973004e-05, | |
| "loss": 0.4361, | |
| "step": 2829 | |
| }, | |
| { | |
| "epoch": 3.208098987626547, | |
| "grad_norm": 1.3668488264083862, | |
| "learning_rate": 3.208098987626547e-05, | |
| "loss": 0.2668, | |
| "step": 2852 | |
| }, | |
| { | |
| "epoch": 3.233970753655793, | |
| "grad_norm": 1.127758502960205, | |
| "learning_rate": 3.233970753655794e-05, | |
| "loss": 0.3974, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.2598425196850394, | |
| "grad_norm": 0.872011125087738, | |
| "learning_rate": 3.2598425196850394e-05, | |
| "loss": 0.3557, | |
| "step": 2898 | |
| }, | |
| { | |
| "epoch": 3.2857142857142856, | |
| "grad_norm": 15.174445152282715, | |
| "learning_rate": 3.285714285714286e-05, | |
| "loss": 0.4542, | |
| "step": 2921 | |
| }, | |
| { | |
| "epoch": 3.3115860517435323, | |
| "grad_norm": 3.535675048828125, | |
| "learning_rate": 3.311586051743532e-05, | |
| "loss": 0.3665, | |
| "step": 2944 | |
| }, | |
| { | |
| "epoch": 3.3374578177727785, | |
| "grad_norm": 0.1963585466146469, | |
| "learning_rate": 3.3374578177727784e-05, | |
| "loss": 0.4275, | |
| "step": 2967 | |
| }, | |
| { | |
| "epoch": 3.3633295838020247, | |
| "grad_norm": 1.3127784729003906, | |
| "learning_rate": 3.3633295838020254e-05, | |
| "loss": 0.4884, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 3.389201349831271, | |
| "grad_norm": 1.3405988216400146, | |
| "learning_rate": 3.389201349831271e-05, | |
| "loss": 0.2996, | |
| "step": 3013 | |
| }, | |
| { | |
| "epoch": 3.415073115860517, | |
| "grad_norm": 9.742510795593262, | |
| "learning_rate": 3.4150731158605174e-05, | |
| "loss": 0.4061, | |
| "step": 3036 | |
| }, | |
| { | |
| "epoch": 3.440944881889764, | |
| "grad_norm": 0.6711751222610474, | |
| "learning_rate": 3.440944881889764e-05, | |
| "loss": 0.3684, | |
| "step": 3059 | |
| }, | |
| { | |
| "epoch": 3.46681664791901, | |
| "grad_norm": 9.26884937286377, | |
| "learning_rate": 3.46681664791901e-05, | |
| "loss": 0.4557, | |
| "step": 3082 | |
| }, | |
| { | |
| "epoch": 3.4926884139482564, | |
| "grad_norm": 11.18705940246582, | |
| "learning_rate": 3.492688413948257e-05, | |
| "loss": 0.3843, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 3.518560179977503, | |
| "grad_norm": 0.6143022775650024, | |
| "learning_rate": 3.518560179977503e-05, | |
| "loss": 0.2909, | |
| "step": 3128 | |
| }, | |
| { | |
| "epoch": 3.5444319460067493, | |
| "grad_norm": 0.13090083003044128, | |
| "learning_rate": 3.544431946006749e-05, | |
| "loss": 0.3688, | |
| "step": 3151 | |
| }, | |
| { | |
| "epoch": 3.5703037120359955, | |
| "grad_norm": 8.838688850402832, | |
| "learning_rate": 3.5703037120359954e-05, | |
| "loss": 0.4144, | |
| "step": 3174 | |
| }, | |
| { | |
| "epoch": 3.5961754780652417, | |
| "grad_norm": 8.904288291931152, | |
| "learning_rate": 3.596175478065242e-05, | |
| "loss": 0.3711, | |
| "step": 3197 | |
| }, | |
| { | |
| "epoch": 3.622047244094488, | |
| "grad_norm": 0.6677899360656738, | |
| "learning_rate": 3.622047244094489e-05, | |
| "loss": 0.5376, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 3.6479190101237347, | |
| "grad_norm": 2.987168073654175, | |
| "learning_rate": 3.647919010123735e-05, | |
| "loss": 0.3642, | |
| "step": 3243 | |
| }, | |
| { | |
| "epoch": 3.673790776152981, | |
| "grad_norm": 9.179577827453613, | |
| "learning_rate": 3.673790776152981e-05, | |
| "loss": 0.6663, | |
| "step": 3266 | |
| }, | |
| { | |
| "epoch": 3.699662542182227, | |
| "grad_norm": 1.0456788539886475, | |
| "learning_rate": 3.699662542182227e-05, | |
| "loss": 0.478, | |
| "step": 3289 | |
| }, | |
| { | |
| "epoch": 3.725534308211474, | |
| "grad_norm": 8.08246898651123, | |
| "learning_rate": 3.7255343082114735e-05, | |
| "loss": 0.4601, | |
| "step": 3312 | |
| }, | |
| { | |
| "epoch": 3.75140607424072, | |
| "grad_norm": 8.33056926727295, | |
| "learning_rate": 3.7514060742407205e-05, | |
| "loss": 0.4343, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 3.7772778402699663, | |
| "grad_norm": 2.994161367416382, | |
| "learning_rate": 3.777277840269967e-05, | |
| "loss": 0.3037, | |
| "step": 3358 | |
| }, | |
| { | |
| "epoch": 3.8031496062992125, | |
| "grad_norm": 0.5124155282974243, | |
| "learning_rate": 3.8031496062992125e-05, | |
| "loss": 0.4441, | |
| "step": 3381 | |
| }, | |
| { | |
| "epoch": 3.8290213723284587, | |
| "grad_norm": 9.554545402526855, | |
| "learning_rate": 3.829021372328459e-05, | |
| "loss": 0.4561, | |
| "step": 3404 | |
| }, | |
| { | |
| "epoch": 3.8548931383577054, | |
| "grad_norm": 2.4750254154205322, | |
| "learning_rate": 3.854893138357705e-05, | |
| "loss": 0.5598, | |
| "step": 3427 | |
| }, | |
| { | |
| "epoch": 3.8807649043869517, | |
| "grad_norm": 0.18071837723255157, | |
| "learning_rate": 3.880764904386952e-05, | |
| "loss": 0.392, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.906636670416198, | |
| "grad_norm": 0.9068869948387146, | |
| "learning_rate": 3.9066366704161985e-05, | |
| "loss": 0.4358, | |
| "step": 3473 | |
| }, | |
| { | |
| "epoch": 3.9325084364454446, | |
| "grad_norm": 2.832122325897217, | |
| "learning_rate": 3.932508436445445e-05, | |
| "loss": 0.3503, | |
| "step": 3496 | |
| }, | |
| { | |
| "epoch": 3.958380202474691, | |
| "grad_norm": 8.372138023376465, | |
| "learning_rate": 3.9583802024746905e-05, | |
| "loss": 0.3665, | |
| "step": 3519 | |
| }, | |
| { | |
| "epoch": 3.984251968503937, | |
| "grad_norm": 0.4025736451148987, | |
| "learning_rate": 3.984251968503937e-05, | |
| "loss": 0.3274, | |
| "step": 3542 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7815126050420168, | |
| "eval_auc": 0.8276973891160743, | |
| "eval_f1": 0.08771929824561403, | |
| "eval_loss": 0.6576350927352905, | |
| "eval_precision": 0.4166666666666667, | |
| "eval_recall": 0.049019607843137254, | |
| "eval_runtime": 2.7248, | |
| "eval_samples_per_second": 174.693, | |
| "eval_steps_per_second": 22.02, | |
| "step": 3556 | |
| }, | |
| { | |
| "epoch": 4.010123734533184, | |
| "grad_norm": 11.520211219787598, | |
| "learning_rate": 4.010123734533184e-05, | |
| "loss": 0.3699, | |
| "step": 3565 | |
| }, | |
| { | |
| "epoch": 4.0359955005624295, | |
| "grad_norm": 3.6669209003448486, | |
| "learning_rate": 4.03599550056243e-05, | |
| "loss": 0.4811, | |
| "step": 3588 | |
| }, | |
| { | |
| "epoch": 4.061867266591676, | |
| "grad_norm": 6.646744251251221, | |
| "learning_rate": 4.0618672665916766e-05, | |
| "loss": 0.3609, | |
| "step": 3611 | |
| }, | |
| { | |
| "epoch": 4.087739032620922, | |
| "grad_norm": 0.03924195468425751, | |
| "learning_rate": 4.087739032620922e-05, | |
| "loss": 0.3344, | |
| "step": 3634 | |
| }, | |
| { | |
| "epoch": 4.113610798650169, | |
| "grad_norm": 3.2102909088134766, | |
| "learning_rate": 4.1136107986501686e-05, | |
| "loss": 0.6133, | |
| "step": 3657 | |
| }, | |
| { | |
| "epoch": 4.139482564679415, | |
| "grad_norm": 8.028974533081055, | |
| "learning_rate": 4.1394825646794156e-05, | |
| "loss": 0.322, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 4.165354330708661, | |
| "grad_norm": 0.7968009114265442, | |
| "learning_rate": 4.165354330708662e-05, | |
| "loss": 0.337, | |
| "step": 3703 | |
| }, | |
| { | |
| "epoch": 4.191226096737908, | |
| "grad_norm": 8.355441093444824, | |
| "learning_rate": 4.191226096737908e-05, | |
| "loss": 0.3879, | |
| "step": 3726 | |
| }, | |
| { | |
| "epoch": 4.2170978627671545, | |
| "grad_norm": 0.24170738458633423, | |
| "learning_rate": 4.217097862767154e-05, | |
| "loss": 0.3252, | |
| "step": 3749 | |
| }, | |
| { | |
| "epoch": 4.2429696287964, | |
| "grad_norm": 5.390313625335693, | |
| "learning_rate": 4.2429696287964e-05, | |
| "loss": 0.507, | |
| "step": 3772 | |
| }, | |
| { | |
| "epoch": 4.268841394825647, | |
| "grad_norm": 5.854102611541748, | |
| "learning_rate": 4.268841394825647e-05, | |
| "loss": 0.3849, | |
| "step": 3795 | |
| }, | |
| { | |
| "epoch": 4.294713160854893, | |
| "grad_norm": 8.469975471496582, | |
| "learning_rate": 4.2947131608548936e-05, | |
| "loss": 0.315, | |
| "step": 3818 | |
| }, | |
| { | |
| "epoch": 4.320584926884139, | |
| "grad_norm": 3.3237178325653076, | |
| "learning_rate": 4.32058492688414e-05, | |
| "loss": 0.7087, | |
| "step": 3841 | |
| }, | |
| { | |
| "epoch": 4.346456692913386, | |
| "grad_norm": 1.0640898942947388, | |
| "learning_rate": 4.346456692913386e-05, | |
| "loss": 0.4298, | |
| "step": 3864 | |
| }, | |
| { | |
| "epoch": 4.372328458942632, | |
| "grad_norm": 0.4618189334869385, | |
| "learning_rate": 4.372328458942632e-05, | |
| "loss": 0.3469, | |
| "step": 3887 | |
| }, | |
| { | |
| "epoch": 4.398200224971879, | |
| "grad_norm": 0.6110893487930298, | |
| "learning_rate": 4.398200224971879e-05, | |
| "loss": 0.4085, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 4.424071991001125, | |
| "grad_norm": 8.723979949951172, | |
| "learning_rate": 4.424071991001125e-05, | |
| "loss": 0.263, | |
| "step": 3933 | |
| }, | |
| { | |
| "epoch": 4.449943757030371, | |
| "grad_norm": 19.94253158569336, | |
| "learning_rate": 4.4499437570303716e-05, | |
| "loss": 0.3318, | |
| "step": 3956 | |
| }, | |
| { | |
| "epoch": 4.475815523059618, | |
| "grad_norm": 0.9224587082862854, | |
| "learning_rate": 4.475815523059618e-05, | |
| "loss": 0.462, | |
| "step": 3979 | |
| }, | |
| { | |
| "epoch": 4.5016872890888635, | |
| "grad_norm": 9.54886531829834, | |
| "learning_rate": 4.5016872890888636e-05, | |
| "loss": 0.4471, | |
| "step": 4002 | |
| }, | |
| { | |
| "epoch": 4.52755905511811, | |
| "grad_norm": 3.35776686668396, | |
| "learning_rate": 4.52755905511811e-05, | |
| "loss": 0.4742, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 4.553430821147357, | |
| "grad_norm": 7.887851715087891, | |
| "learning_rate": 4.553430821147357e-05, | |
| "loss": 0.4714, | |
| "step": 4048 | |
| }, | |
| { | |
| "epoch": 4.579302587176603, | |
| "grad_norm": 3.327909469604492, | |
| "learning_rate": 4.5793025871766033e-05, | |
| "loss": 0.6452, | |
| "step": 4071 | |
| }, | |
| { | |
| "epoch": 4.605174353205849, | |
| "grad_norm": 1.3243390321731567, | |
| "learning_rate": 4.60517435320585e-05, | |
| "loss": 0.3745, | |
| "step": 4094 | |
| }, | |
| { | |
| "epoch": 4.631046119235096, | |
| "grad_norm": 0.7224089503288269, | |
| "learning_rate": 4.631046119235096e-05, | |
| "loss": 0.2628, | |
| "step": 4117 | |
| }, | |
| { | |
| "epoch": 4.656917885264342, | |
| "grad_norm": 0.017284361645579338, | |
| "learning_rate": 4.656917885264342e-05, | |
| "loss": 0.317, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 4.6827896512935885, | |
| "grad_norm": 14.2166166305542, | |
| "learning_rate": 4.682789651293589e-05, | |
| "loss": 0.464, | |
| "step": 4163 | |
| }, | |
| { | |
| "epoch": 4.708661417322834, | |
| "grad_norm": 0.49798861145973206, | |
| "learning_rate": 4.708661417322835e-05, | |
| "loss": 0.3336, | |
| "step": 4186 | |
| }, | |
| { | |
| "epoch": 4.734533183352081, | |
| "grad_norm": 1.5816570520401, | |
| "learning_rate": 4.7345331833520814e-05, | |
| "loss": 0.423, | |
| "step": 4209 | |
| }, | |
| { | |
| "epoch": 4.760404949381328, | |
| "grad_norm": 4.431649208068848, | |
| "learning_rate": 4.760404949381328e-05, | |
| "loss": 0.4474, | |
| "step": 4232 | |
| }, | |
| { | |
| "epoch": 4.786276715410573, | |
| "grad_norm": 7.487512588500977, | |
| "learning_rate": 4.7862767154105734e-05, | |
| "loss": 0.32, | |
| "step": 4255 | |
| }, | |
| { | |
| "epoch": 4.81214848143982, | |
| "grad_norm": 3.283292293548584, | |
| "learning_rate": 4.8121484814398204e-05, | |
| "loss": 0.4767, | |
| "step": 4278 | |
| }, | |
| { | |
| "epoch": 4.838020247469066, | |
| "grad_norm": 7.209409236907959, | |
| "learning_rate": 4.838020247469067e-05, | |
| "loss": 0.2593, | |
| "step": 4301 | |
| }, | |
| { | |
| "epoch": 4.863892013498313, | |
| "grad_norm": 7.526238441467285, | |
| "learning_rate": 4.863892013498313e-05, | |
| "loss": 0.3552, | |
| "step": 4324 | |
| }, | |
| { | |
| "epoch": 4.889763779527559, | |
| "grad_norm": 0.18729792535305023, | |
| "learning_rate": 4.8897637795275594e-05, | |
| "loss": 0.3581, | |
| "step": 4347 | |
| }, | |
| { | |
| "epoch": 4.915635545556805, | |
| "grad_norm": 0.21440748870372772, | |
| "learning_rate": 4.915635545556805e-05, | |
| "loss": 0.2921, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 4.941507311586052, | |
| "grad_norm": 12.171503067016602, | |
| "learning_rate": 4.941507311586052e-05, | |
| "loss": 0.5248, | |
| "step": 4393 | |
| }, | |
| { | |
| "epoch": 4.967379077615298, | |
| "grad_norm": 0.14398285746574402, | |
| "learning_rate": 4.9673790776152984e-05, | |
| "loss": 0.4253, | |
| "step": 4416 | |
| }, | |
| { | |
| "epoch": 4.993250843644544, | |
| "grad_norm": 2.316239595413208, | |
| "learning_rate": 4.993250843644545e-05, | |
| "loss": 0.4106, | |
| "step": 4439 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7815126050420168, | |
| "eval_auc": 0.8033579742057251, | |
| "eval_f1": 0.0, | |
| "eval_loss": 0.50748610496521, | |
| "eval_precision": 0.0, | |
| "eval_recall": 0.0, | |
| "eval_runtime": 2.7592, | |
| "eval_samples_per_second": 172.516, | |
| "eval_steps_per_second": 21.746, | |
| "step": 4445 | |
| }, | |
| { | |
| "epoch": 5.019122609673791, | |
| "grad_norm": 4.349539279937744, | |
| "learning_rate": 4.997875265591801e-05, | |
| "loss": 0.3046, | |
| "step": 4462 | |
| }, | |
| { | |
| "epoch": 5.0449943757030375, | |
| "grad_norm": 0.3448341190814972, | |
| "learning_rate": 4.995000624921885e-05, | |
| "loss": 0.3477, | |
| "step": 4485 | |
| }, | |
| { | |
| "epoch": 5.070866141732283, | |
| "grad_norm": 0.6105608940124512, | |
| "learning_rate": 4.9921259842519685e-05, | |
| "loss": 0.5053, | |
| "step": 4508 | |
| }, | |
| { | |
| "epoch": 5.09673790776153, | |
| "grad_norm": 8.691744804382324, | |
| "learning_rate": 4.989251343582053e-05, | |
| "loss": 0.3679, | |
| "step": 4531 | |
| }, | |
| { | |
| "epoch": 5.122609673790776, | |
| "grad_norm": 2.42826771736145, | |
| "learning_rate": 4.9863767029121364e-05, | |
| "loss": 0.2206, | |
| "step": 4554 | |
| }, | |
| { | |
| "epoch": 5.1484814398200225, | |
| "grad_norm": 2.756155490875244, | |
| "learning_rate": 4.98350206224222e-05, | |
| "loss": 0.6075, | |
| "step": 4577 | |
| }, | |
| { | |
| "epoch": 5.174353205849269, | |
| "grad_norm": 0.6143556833267212, | |
| "learning_rate": 4.9806274215723035e-05, | |
| "loss": 0.2272, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 5.200224971878515, | |
| "grad_norm": 0.17892147600650787, | |
| "learning_rate": 4.977752780902387e-05, | |
| "loss": 0.5256, | |
| "step": 4623 | |
| }, | |
| { | |
| "epoch": 5.226096737907762, | |
| "grad_norm": 0.12368874251842499, | |
| "learning_rate": 4.9748781402324714e-05, | |
| "loss": 0.5569, | |
| "step": 4646 | |
| }, | |
| { | |
| "epoch": 5.251968503937007, | |
| "grad_norm": 0.048263806849718094, | |
| "learning_rate": 4.972003499562555e-05, | |
| "loss": 0.2298, | |
| "step": 4669 | |
| }, | |
| { | |
| "epoch": 5.277840269966254, | |
| "grad_norm": 4.790956020355225, | |
| "learning_rate": 4.9691288588926386e-05, | |
| "loss": 0.5415, | |
| "step": 4692 | |
| }, | |
| { | |
| "epoch": 5.303712035995501, | |
| "grad_norm": 5.684449672698975, | |
| "learning_rate": 4.966254218222722e-05, | |
| "loss": 0.3086, | |
| "step": 4715 | |
| }, | |
| { | |
| "epoch": 5.329583802024747, | |
| "grad_norm": 0.4559657871723175, | |
| "learning_rate": 4.9633795775528064e-05, | |
| "loss": 0.3575, | |
| "step": 4738 | |
| }, | |
| { | |
| "epoch": 5.355455568053993, | |
| "grad_norm": 7.768259048461914, | |
| "learning_rate": 4.96050493688289e-05, | |
| "loss": 0.4601, | |
| "step": 4761 | |
| }, | |
| { | |
| "epoch": 5.38132733408324, | |
| "grad_norm": 0.44012290239334106, | |
| "learning_rate": 4.9576302962129736e-05, | |
| "loss": 0.3265, | |
| "step": 4784 | |
| }, | |
| { | |
| "epoch": 5.407199100112486, | |
| "grad_norm": 0.22265122830867767, | |
| "learning_rate": 4.954755655543057e-05, | |
| "loss": 0.6152, | |
| "step": 4807 | |
| }, | |
| { | |
| "epoch": 5.433070866141732, | |
| "grad_norm": 4.463795185089111, | |
| "learning_rate": 4.951881014873141e-05, | |
| "loss": 0.3046, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 5.458942632170979, | |
| "grad_norm": 5.145118236541748, | |
| "learning_rate": 4.949006374203225e-05, | |
| "loss": 0.4773, | |
| "step": 4853 | |
| }, | |
| { | |
| "epoch": 5.484814398200225, | |
| "grad_norm": 0.24177011847496033, | |
| "learning_rate": 4.9461317335333087e-05, | |
| "loss": 0.2634, | |
| "step": 4876 | |
| }, | |
| { | |
| "epoch": 5.5106861642294716, | |
| "grad_norm": 0.8902063369750977, | |
| "learning_rate": 4.943257092863392e-05, | |
| "loss": 0.3362, | |
| "step": 4899 | |
| }, | |
| { | |
| "epoch": 5.536557930258717, | |
| "grad_norm": 0.39589497447013855, | |
| "learning_rate": 4.940382452193476e-05, | |
| "loss": 0.6168, | |
| "step": 4922 | |
| }, | |
| { | |
| "epoch": 5.562429696287964, | |
| "grad_norm": 1.736325979232788, | |
| "learning_rate": 4.9375078115235594e-05, | |
| "loss": 0.5466, | |
| "step": 4945 | |
| }, | |
| { | |
| "epoch": 5.588301462317211, | |
| "grad_norm": 2.3112049102783203, | |
| "learning_rate": 4.934633170853644e-05, | |
| "loss": 0.3546, | |
| "step": 4968 | |
| }, | |
| { | |
| "epoch": 5.6141732283464565, | |
| "grad_norm": 0.4776885509490967, | |
| "learning_rate": 4.931758530183727e-05, | |
| "loss": 0.3755, | |
| "step": 4991 | |
| }, | |
| { | |
| "epoch": 5.640044994375703, | |
| "grad_norm": 0.09375002235174179, | |
| "learning_rate": 4.928883889513811e-05, | |
| "loss": 0.2044, | |
| "step": 5014 | |
| }, | |
| { | |
| "epoch": 5.665916760404949, | |
| "grad_norm": 11.319817543029785, | |
| "learning_rate": 4.9260092488438945e-05, | |
| "loss": 0.5937, | |
| "step": 5037 | |
| }, | |
| { | |
| "epoch": 5.691788526434196, | |
| "grad_norm": 1.1792347431182861, | |
| "learning_rate": 4.923134608173979e-05, | |
| "loss": 0.3856, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 5.717660292463442, | |
| "grad_norm": 6.615360736846924, | |
| "learning_rate": 4.920259967504062e-05, | |
| "loss": 0.3669, | |
| "step": 5083 | |
| }, | |
| { | |
| "epoch": 5.743532058492688, | |
| "grad_norm": 6.705723762512207, | |
| "learning_rate": 4.917385326834146e-05, | |
| "loss": 0.2221, | |
| "step": 5106 | |
| }, | |
| { | |
| "epoch": 5.769403824521935, | |
| "grad_norm": 2.312495470046997, | |
| "learning_rate": 4.9145106861642295e-05, | |
| "loss": 0.4308, | |
| "step": 5129 | |
| }, | |
| { | |
| "epoch": 5.7952755905511815, | |
| "grad_norm": 11.482613563537598, | |
| "learning_rate": 4.911636045494313e-05, | |
| "loss": 0.3644, | |
| "step": 5152 | |
| }, | |
| { | |
| "epoch": 5.821147356580427, | |
| "grad_norm": 4.5011982917785645, | |
| "learning_rate": 4.9087614048243973e-05, | |
| "loss": 0.5318, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 5.847019122609674, | |
| "grad_norm": 3.1337974071502686, | |
| "learning_rate": 4.905886764154481e-05, | |
| "loss": 0.4274, | |
| "step": 5198 | |
| }, | |
| { | |
| "epoch": 5.872890888638921, | |
| "grad_norm": 0.16391399502754211, | |
| "learning_rate": 4.9030121234845645e-05, | |
| "loss": 0.3304, | |
| "step": 5221 | |
| }, | |
| { | |
| "epoch": 5.898762654668166, | |
| "grad_norm": 0.630847692489624, | |
| "learning_rate": 4.900137482814649e-05, | |
| "loss": 0.4878, | |
| "step": 5244 | |
| }, | |
| { | |
| "epoch": 5.924634420697413, | |
| "grad_norm": 3.9900641441345215, | |
| "learning_rate": 4.897262842144732e-05, | |
| "loss": 0.3846, | |
| "step": 5267 | |
| }, | |
| { | |
| "epoch": 5.950506186726659, | |
| "grad_norm": 5.799670219421387, | |
| "learning_rate": 4.894388201474816e-05, | |
| "loss": 0.3469, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 5.9763779527559056, | |
| "grad_norm": 0.49845901131629944, | |
| "learning_rate": 4.8915135608048996e-05, | |
| "loss": 0.3733, | |
| "step": 5313 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.7857142857142857, | |
| "eval_auc": 0.7628709237705779, | |
| "eval_f1": 0.0, | |
| "eval_loss": 0.6193206310272217, | |
| "eval_precision": 0.0, | |
| "eval_recall": 0.0, | |
| "eval_runtime": 2.7283, | |
| "eval_samples_per_second": 174.467, | |
| "eval_steps_per_second": 21.992, | |
| "step": 5334 | |
| }, | |
| { | |
| "epoch": 6.002249718785152, | |
| "grad_norm": 0.5700758695602417, | |
| "learning_rate": 4.888638920134984e-05, | |
| "loss": 0.3025, | |
| "step": 5336 | |
| }, | |
| { | |
| "epoch": 6.028121484814398, | |
| "grad_norm": 0.3606770932674408, | |
| "learning_rate": 4.8857642794650674e-05, | |
| "loss": 0.383, | |
| "step": 5359 | |
| }, | |
| { | |
| "epoch": 6.053993250843645, | |
| "grad_norm": 0.136641263961792, | |
| "learning_rate": 4.88288963879515e-05, | |
| "loss": 0.2687, | |
| "step": 5382 | |
| }, | |
| { | |
| "epoch": 6.0798650168728905, | |
| "grad_norm": 0.3220977187156677, | |
| "learning_rate": 4.8800149981252346e-05, | |
| "loss": 0.4189, | |
| "step": 5405 | |
| }, | |
| { | |
| "epoch": 6.105736782902137, | |
| "grad_norm": 0.8333466053009033, | |
| "learning_rate": 4.877140357455318e-05, | |
| "loss": 0.3677, | |
| "step": 5428 | |
| }, | |
| { | |
| "epoch": 6.131608548931384, | |
| "grad_norm": 0.08014432340860367, | |
| "learning_rate": 4.8742657167854025e-05, | |
| "loss": 0.3624, | |
| "step": 5451 | |
| }, | |
| { | |
| "epoch": 6.15748031496063, | |
| "grad_norm": 8.192437171936035, | |
| "learning_rate": 4.871391076115486e-05, | |
| "loss": 0.4036, | |
| "step": 5474 | |
| }, | |
| { | |
| "epoch": 6.183352080989876, | |
| "grad_norm": 6.411347389221191, | |
| "learning_rate": 4.8685164354455696e-05, | |
| "loss": 0.3529, | |
| "step": 5497 | |
| }, | |
| { | |
| "epoch": 6.209223847019123, | |
| "grad_norm": 0.006956384517252445, | |
| "learning_rate": 4.865641794775653e-05, | |
| "loss": 0.1559, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 6.235095613048369, | |
| "grad_norm": 0.1893380731344223, | |
| "learning_rate": 4.862767154105737e-05, | |
| "loss": 0.5716, | |
| "step": 5543 | |
| }, | |
| { | |
| "epoch": 6.2609673790776155, | |
| "grad_norm": 0.47687458992004395, | |
| "learning_rate": 4.859892513435821e-05, | |
| "loss": 0.399, | |
| "step": 5566 | |
| }, | |
| { | |
| "epoch": 6.286839145106861, | |
| "grad_norm": 4.109427452087402, | |
| "learning_rate": 4.857017872765905e-05, | |
| "loss": 0.3858, | |
| "step": 5589 | |
| }, | |
| { | |
| "epoch": 6.312710911136108, | |
| "grad_norm": 0.6130828857421875, | |
| "learning_rate": 4.854143232095988e-05, | |
| "loss": 0.3434, | |
| "step": 5612 | |
| }, | |
| { | |
| "epoch": 6.338582677165355, | |
| "grad_norm": 2.936263084411621, | |
| "learning_rate": 4.851268591426072e-05, | |
| "loss": 0.7265, | |
| "step": 5635 | |
| }, | |
| { | |
| "epoch": 6.3644544431946, | |
| "grad_norm": 0.4069950580596924, | |
| "learning_rate": 4.848393950756156e-05, | |
| "loss": 0.2387, | |
| "step": 5658 | |
| }, | |
| { | |
| "epoch": 6.390326209223847, | |
| "grad_norm": 3.4261677265167236, | |
| "learning_rate": 4.84551931008624e-05, | |
| "loss": 0.4388, | |
| "step": 5681 | |
| }, | |
| { | |
| "epoch": 6.416197975253094, | |
| "grad_norm": 4.583181381225586, | |
| "learning_rate": 4.842644669416323e-05, | |
| "loss": 0.321, | |
| "step": 5704 | |
| }, | |
| { | |
| "epoch": 6.44206974128234, | |
| "grad_norm": 4.256218433380127, | |
| "learning_rate": 4.839770028746407e-05, | |
| "loss": 0.3261, | |
| "step": 5727 | |
| }, | |
| { | |
| "epoch": 6.467941507311586, | |
| "grad_norm": 1.0955240726470947, | |
| "learning_rate": 4.8368953880764905e-05, | |
| "loss": 0.3753, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 6.493813273340832, | |
| "grad_norm": 7.2450361251831055, | |
| "learning_rate": 4.834020747406575e-05, | |
| "loss": 0.2428, | |
| "step": 5773 | |
| }, | |
| { | |
| "epoch": 6.519685039370079, | |
| "grad_norm": 5.181588172912598, | |
| "learning_rate": 4.831146106736658e-05, | |
| "loss": 0.3341, | |
| "step": 5796 | |
| }, | |
| { | |
| "epoch": 6.545556805399325, | |
| "grad_norm": 2.217822790145874, | |
| "learning_rate": 4.828271466066742e-05, | |
| "loss": 0.4004, | |
| "step": 5819 | |
| }, | |
| { | |
| "epoch": 6.571428571428571, | |
| "grad_norm": 0.3265409469604492, | |
| "learning_rate": 4.8253968253968255e-05, | |
| "loss": 0.4519, | |
| "step": 5842 | |
| }, | |
| { | |
| "epoch": 6.597300337457818, | |
| "grad_norm": 3.4193084239959717, | |
| "learning_rate": 4.822522184726909e-05, | |
| "loss": 0.4347, | |
| "step": 5865 | |
| }, | |
| { | |
| "epoch": 6.6231721034870645, | |
| "grad_norm": 3.1898536682128906, | |
| "learning_rate": 4.8196475440569934e-05, | |
| "loss": 0.2508, | |
| "step": 5888 | |
| }, | |
| { | |
| "epoch": 6.64904386951631, | |
| "grad_norm": 5.070316314697266, | |
| "learning_rate": 4.816772903387077e-05, | |
| "loss": 0.4112, | |
| "step": 5911 | |
| }, | |
| { | |
| "epoch": 6.674915635545557, | |
| "grad_norm": 0.23132477700710297, | |
| "learning_rate": 4.8138982627171605e-05, | |
| "loss": 0.2206, | |
| "step": 5934 | |
| }, | |
| { | |
| "epoch": 6.700787401574803, | |
| "grad_norm": 4.237873554229736, | |
| "learning_rate": 4.811023622047244e-05, | |
| "loss": 0.5504, | |
| "step": 5957 | |
| }, | |
| { | |
| "epoch": 6.7266591676040495, | |
| "grad_norm": 0.857448160648346, | |
| "learning_rate": 4.808148981377328e-05, | |
| "loss": 0.3815, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 6.752530933633296, | |
| "grad_norm": 5.193442344665527, | |
| "learning_rate": 4.805274340707412e-05, | |
| "loss": 0.3129, | |
| "step": 6003 | |
| }, | |
| { | |
| "epoch": 6.778402699662542, | |
| "grad_norm": 4.156959056854248, | |
| "learning_rate": 4.8023997000374956e-05, | |
| "loss": 0.4625, | |
| "step": 6026 | |
| }, | |
| { | |
| "epoch": 6.804274465691789, | |
| "grad_norm": 0.4680914878845215, | |
| "learning_rate": 4.799525059367579e-05, | |
| "loss": 0.2069, | |
| "step": 6049 | |
| }, | |
| { | |
| "epoch": 6.830146231721034, | |
| "grad_norm": 5.006599426269531, | |
| "learning_rate": 4.796650418697663e-05, | |
| "loss": 0.4084, | |
| "step": 6072 | |
| }, | |
| { | |
| "epoch": 6.856017997750281, | |
| "grad_norm": 6.840053081512451, | |
| "learning_rate": 4.793775778027747e-05, | |
| "loss": 0.2994, | |
| "step": 6095 | |
| }, | |
| { | |
| "epoch": 6.881889763779528, | |
| "grad_norm": 0.11373735964298248, | |
| "learning_rate": 4.7909011373578306e-05, | |
| "loss": 0.2934, | |
| "step": 6118 | |
| }, | |
| { | |
| "epoch": 6.907761529808774, | |
| "grad_norm": 5.152923107147217, | |
| "learning_rate": 4.788026496687914e-05, | |
| "loss": 0.2687, | |
| "step": 6141 | |
| }, | |
| { | |
| "epoch": 6.93363329583802, | |
| "grad_norm": 0.021206054836511612, | |
| "learning_rate": 4.785151856017998e-05, | |
| "loss": 0.3019, | |
| "step": 6164 | |
| }, | |
| { | |
| "epoch": 6.959505061867267, | |
| "grad_norm": 0.06403312087059021, | |
| "learning_rate": 4.7822772153480814e-05, | |
| "loss": 0.2471, | |
| "step": 6187 | |
| }, | |
| { | |
| "epoch": 6.985376827896513, | |
| "grad_norm": 3.1448378562927246, | |
| "learning_rate": 4.7794025746781657e-05, | |
| "loss": 0.6191, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.819327731092437, | |
| "eval_auc": 0.8453785257418476, | |
| "eval_f1": 0.4625, | |
| "eval_loss": 0.3894089162349701, | |
| "eval_precision": 0.6379310344827587, | |
| "eval_recall": 0.3627450980392157, | |
| "eval_runtime": 2.7066, | |
| "eval_samples_per_second": 175.867, | |
| "eval_steps_per_second": 22.168, | |
| "step": 6223 | |
| } | |
| ], | |
| "logging_steps": 23, | |
| "max_steps": 44450, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 20, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.0174156175154545e+19, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |