re-trained model after eliminating annotation errors

README.md

@@ -1,4 +0,0 @@
----
-license: afl-3.0
-inference: false
----

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fa17b658ff6291efbf0919babd01f5eec92da9643b39758cdaa3b6db9476e1db
+oid sha256:5276bb8be599a0e1a57ad1a71f3975ae5e7e6b7cabcbcbde26fe5fd41cbd0853
 size 997361565

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41adba97b7f678cbf4f5086d500f1346c7fa47d372988e0f0b6d5a52361a1d50
+oid sha256:f5a7aee6788246a2848ebff5ac3618a37d758a581da5a7ead3c2bf124d51f6d4
 size 498694189

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a9d4831f96a909c7446a40c7bdd8e8f202f41f3fc50a85bbf9b303cfdfcd011
+oid sha256:573ba4da91fda3dbf4c67c67e7334810f4f5c1c52e4466d2c4c1a4fe2dc8d0d4
 size 14567

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2c73139dd76e847347cf52b339302ca357b5775c05b771f462b21d2da0fdbee2
+oid sha256:b71b07341c8d2987279f87286ec850204a8e1765279b2b57ded211a066f7e1b3
 size 623

trainer_state.json

@@ -1,382 +1,416 @@
 {
-  "best_metric": 0.6491296227815271,
-  "best_model_checkpoint": "./CARES/checkpoints/roberta-stratified/run-0/checkpoint-3266",
-  "epoch": 23.163120567375888,
-  "global_step": 3266,
+  "best_metric": 0.7672541719320296,
+  "best_model_checkpoint": "./CARES/checkpoints/roberta-stratified/run-6/checkpoint-3550",
+  "epoch": 50.0,
+  "global_step": 3550,
   "is_hyper_param_search": true,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.01,
-      "eval_loss": 0.3462072014808655,
-      "eval_macro_f1": 0.0007381889763779527,
-      "eval_macro_precision": 0.0625,
-      "eval_macro_recall": 0.0003712871287128713,
-      "eval_micro_f1": 0.0039447731755424065,
-      "eval_micro_precision": 1.0,
-      "eval_micro_recall": 0.001976284584980237,
-      "eval_runtime": 2.8803,
-      "eval_samples_per_second": 337.116,
-      "eval_steps_per_second": 10.763,
+      "epoch": 2.0,
+      "eval_loss": 0.24808131158351898,
+      "eval_macro_f1": 0.05481647030175482,
+      "eval_macro_precision": 0.17800453514739228,
+      "eval_macro_recall": 0.03330847588023425,
+      "eval_micro_f1": 0.13594611145131658,
+      "eval_micro_precision": 0.9568965517241379,
+      "eval_micro_recall": 0.07317073170731707,
+      "eval_runtime": 2.8186,
+      "eval_samples_per_second": 342.724,
+      "eval_steps_per_second": 21.642,
       "step": 142
     },
     {
-      "epoch": 2.01,
-      "eval_loss": 0.26299169659614563,
-      "eval_macro_f1": 0.03133608815426997,
-      "eval_macro_precision": 0.051470588235294115,
-      "eval_macro_recall": 0.022524752475247524,
-      "eval_micro_f1": 0.20931569867740082,
-      "eval_micro_precision": 0.8235294117647058,
-      "eval_micro_recall": 0.11989459815546773,
-      "eval_runtime": 2.7712,
-      "eval_samples_per_second": 350.396,
-      "eval_steps_per_second": 11.187,
+      "epoch": 4.0,
+      "eval_loss": 0.16814221441745758,
+      "eval_macro_f1": 0.37346208328620767,
+      "eval_macro_precision": 0.4376472360907343,
+      "eval_macro_recall": 0.3342883491414155,
+      "eval_micro_f1": 0.7248062015503876,
+      "eval_micro_precision": 0.8795860771401693,
+      "eval_micro_recall": 0.6163480553724456,
+      "eval_runtime": 2.8192,
+      "eval_samples_per_second": 342.653,
+      "eval_steps_per_second": 21.638,
       "step": 284
     },
     {
-      "epoch": 3.02,
-      "eval_loss": 0.2253277599811554,
-      "eval_macro_f1": 0.06167745523885891,
-      "eval_macro_precision": 0.16397973284200829,
-      "eval_macro_recall": 0.06444494429328504,
-      "eval_micro_f1": 0.43727272727272726,
-      "eval_micro_precision": 0.7052785923753666,
-      "eval_micro_recall": 0.3168642951251647,
-      "eval_runtime": 2.7754,
-      "eval_samples_per_second": 349.86,
-      "eval_steps_per_second": 11.17,
+      "epoch": 6.0,
+      "eval_loss": 0.13150478899478912,
+      "eval_macro_f1": 0.4319986220599845,
+      "eval_macro_precision": 0.6389953058630822,
+      "eval_macro_recall": 0.37598442828818834,
+      "eval_micro_f1": 0.7641723356009071,
+      "eval_micro_precision": 0.895482728077945,
+      "eval_micro_recall": 0.6664469347396177,
+      "eval_runtime": 2.8191,
+      "eval_samples_per_second": 342.658,
+      "eval_steps_per_second": 21.638,
       "step": 426
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 1.858287403029302e-05,
-      "loss": 0.3344,
+      "epoch": 7.04,
+      "learning_rate": 2.8662355855330125e-05,
+      "loss": 0.2172,
       "step": 500
     },
     {
-      "epoch": 4.03,
-      "eval_loss": 0.19568884372711182,
-      "eval_macro_f1": 0.19533405609363325,
-      "eval_macro_precision": 0.3105566636324695,
-      "eval_macro_recall": 0.1838257881936085,
-      "eval_micro_f1": 0.6221161495624502,
-      "eval_micro_precision": 0.785140562248996,
-      "eval_micro_recall": 0.5151515151515151,
-      "eval_runtime": 2.775,
-      "eval_samples_per_second": 349.909,
-      "eval_steps_per_second": 11.171,
+      "epoch": 8.0,
+      "eval_loss": 0.1121998205780983,
+      "eval_macro_f1": 0.5386615940425785,
+      "eval_macro_precision": 0.7631201629660762,
+      "eval_macro_recall": 0.46714612883136086,
+      "eval_micro_f1": 0.8077485380116959,
+      "eval_micro_precision": 0.9064807219031994,
+      "eval_micro_recall": 0.7284113381674358,
+      "eval_runtime": 2.8196,
+      "eval_samples_per_second": 342.598,
+      "eval_steps_per_second": 21.634,
       "step": 568
     },
     {
-      "epoch": 5.04,
-      "eval_loss": 0.16411657631397247,
-      "eval_macro_f1": 0.2724936969098707,
-      "eval_macro_precision": 0.3798299682715689,
-      "eval_macro_recall": 0.24456296636870606,
-      "eval_micro_f1": 0.6832491255343957,
-      "eval_micro_precision": 0.833175355450237,
-      "eval_micro_recall": 0.5790513833992095,
-      "eval_runtime": 2.7735,
-      "eval_samples_per_second": 350.097,
-      "eval_steps_per_second": 11.177,
+      "epoch": 10.0,
+      "eval_loss": 0.10608664900064468,
+      "eval_macro_f1": 0.5851323545796894,
+      "eval_macro_precision": 0.7572045278273729,
+      "eval_macro_recall": 0.5132727928856573,
+      "eval_micro_f1": 0.8146802325581396,
+      "eval_micro_precision": 0.9076923076923077,
+      "eval_micro_recall": 0.7389584706657878,
+      "eval_runtime": 2.8204,
+      "eval_samples_per_second": 342.507,
+      "eval_steps_per_second": 21.628,
       "step": 710
     },
     {
-      "epoch": 6.04,
-      "eval_loss": 0.15434952080249786,
-      "eval_macro_f1": 0.33512731711181853,
-      "eval_macro_precision": 0.42023284164245034,
-      "eval_macro_recall": 0.31527094611085327,
-      "eval_micro_f1": 0.7207672445592034,
-      "eval_micro_precision": 0.8189438390611903,
-      "eval_micro_recall": 0.6436100131752306,
-      "eval_runtime": 2.7765,
-      "eval_samples_per_second": 349.72,
-      "eval_steps_per_second": 11.165,
+      "epoch": 12.0,
+      "eval_loss": 0.100788913667202,
+      "eval_macro_f1": 0.628825318779795,
+      "eval_macro_precision": 0.750817062947988,
+      "eval_macro_recall": 0.5578116563013593,
+      "eval_micro_f1": 0.8245363766048501,
+      "eval_micro_precision": 0.8982128982128982,
+      "eval_micro_recall": 0.7620303230059328,
+      "eval_runtime": 2.8211,
+      "eval_samples_per_second": 342.425,
+      "eval_steps_per_second": 21.623,
       "step": 852
     },
     {
-      "epoch": 7.05,
-      "eval_loss": 0.14012500643730164,
-      "eval_macro_f1": 0.3798067714294707,
-      "eval_macro_precision": 0.5469114307081249,
-      "eval_macro_recall": 0.3317304014614234,
-      "eval_micro_f1": 0.7399624765478424,
-      "eval_micro_precision": 0.8596338273757629,
-      "eval_micro_recall": 0.6495388669301713,
-      "eval_runtime": 2.774,
-      "eval_samples_per_second": 350.033,
-      "eval_steps_per_second": 11.175,
+      "epoch": 14.0,
+      "eval_loss": 0.09952697157859802,
+      "eval_macro_f1": 0.648367052483795,
+      "eval_macro_precision": 0.7408904755242313,
+      "eval_macro_recall": 0.5923039028415692,
+      "eval_micro_f1": 0.8379418970948547,
+      "eval_micro_precision": 0.8932835820895523,
+      "eval_micro_recall": 0.7890573500329597,
+      "eval_runtime": 2.8203,
+      "eval_samples_per_second": 342.519,
+      "eval_steps_per_second": 21.629,
       "step": 994
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 3.0452387239092086e-05,
-      "loss": 0.1403,
+      "epoch": 14.08,
+      "learning_rate": 2.6490965260229357e-05,
+      "loss": 0.0388,
       "step": 1000
     },
     {
-      "epoch": 8.06,
-      "eval_loss": 0.1425222009420395,
-      "eval_macro_f1": 0.4279380705267567,
-      "eval_macro_precision": 0.603624913309443,
-      "eval_macro_recall": 0.3703454980587147,
-      "eval_micro_f1": 0.7544954128440367,
-      "eval_micro_precision": 0.851698425849213,
-      "eval_micro_recall": 0.6772068511198946,
-      "eval_runtime": 2.7729,
-      "eval_samples_per_second": 350.179,
-      "eval_steps_per_second": 11.18,
+      "epoch": 16.0,
+      "eval_loss": 0.10003374516963959,
+      "eval_macro_f1": 0.7100297932545989,
+      "eval_macro_precision": 0.8731926397728601,
+      "eval_macro_recall": 0.6387457691830838,
+      "eval_micro_f1": 0.837847344354555,
+      "eval_micro_precision": 0.8981900452488688,
+      "eval_micro_recall": 0.7851021753460777,
+      "eval_runtime": 2.8204,
+      "eval_samples_per_second": 342.505,
+      "eval_steps_per_second": 21.628,
       "step": 1136
     },
     {
-      "epoch": 9.06,
-      "eval_loss": 0.12965930998325348,
-      "eval_macro_f1": 0.516908938978679,
-      "eval_macro_precision": 0.6952135806026781,
-      "eval_macro_recall": 0.4441065897269686,
-      "eval_micro_f1": 0.7703488372093023,
-      "eval_micro_precision": 0.8589951377633711,
-      "eval_micro_recall": 0.6982872200263505,
-      "eval_runtime": 2.7725,
-      "eval_samples_per_second": 350.226,
-      "eval_steps_per_second": 11.181,
+      "epoch": 18.0,
+      "eval_loss": 0.10116878896951675,
+      "eval_macro_f1": 0.7102542897490418,
+      "eval_macro_precision": 0.8708979085725551,
+      "eval_macro_recall": 0.6278706438272531,
+      "eval_micro_f1": 0.8369795342272407,
+      "eval_micro_precision": 0.9005315110098709,
+      "eval_micro_recall": 0.7818061964403428,
+      "eval_runtime": 2.821,
+      "eval_samples_per_second": 342.431,
+      "eval_steps_per_second": 21.624,
       "step": 1278
     },
     {
-      "epoch": 10.07,
-      "eval_loss": 0.12852537631988525,
-      "eval_macro_f1": 0.5388984906238257,
-      "eval_macro_precision": 0.7130990451123878,
-      "eval_macro_recall": 0.4711434333008039,
-      "eval_micro_f1": 0.7730547550432276,
-      "eval_micro_precision": 0.8529411764705882,
-      "eval_micro_recall": 0.7068511198945981,
-      "eval_runtime": 2.7737,
-      "eval_samples_per_second": 350.077,
-      "eval_steps_per_second": 11.176,
+      "epoch": 20.0,
+      "eval_loss": 0.10277832299470901,
+      "eval_macro_f1": 0.7155479835258449,
+      "eval_macro_precision": 0.8815899776220457,
+      "eval_macro_recall": 0.635461059766739,
+      "eval_micro_f1": 0.8361702127659574,
+      "eval_micro_precision": 0.9048349961627015,
+      "eval_micro_recall": 0.7771918259723137,
+      "eval_runtime": 2.8201,
+      "eval_samples_per_second": 342.543,
+      "eval_steps_per_second": 21.631,
       "step": 1420
     },
     {
-      "epoch": 10.64,
-      "learning_rate": 2.929008238263819e-05,
-      "loss": 0.0558,
+      "epoch": 21.13,
+      "learning_rate": 2.431957466512859e-05,
+      "loss": 0.0147,
       "step": 1500
     },
     {
-      "epoch": 11.08,
-      "eval_loss": 0.13507980108261108,
-      "eval_macro_f1": 0.5653547754610186,
-      "eval_macro_precision": 0.697025101599949,
-      "eval_macro_recall": 0.49809191604893205,
-      "eval_micro_f1": 0.7797808412866738,
-      "eval_micro_precision": 0.8413424866514111,
-      "eval_micro_recall": 0.7266139657444005,
-      "eval_runtime": 2.7736,
-      "eval_samples_per_second": 350.083,
-      "eval_steps_per_second": 11.177,
+      "epoch": 22.0,
+      "eval_loss": 0.10351855307817459,
+      "eval_macro_f1": 0.7203873361325647,
+      "eval_macro_precision": 0.87817690487536,
+      "eval_macro_recall": 0.6443209986334517,
+      "eval_micro_f1": 0.837389770723104,
+      "eval_micro_precision": 0.9006069802731411,
+      "eval_micro_recall": 0.7824653922214898,
+      "eval_runtime": 2.8203,
+      "eval_samples_per_second": 342.52,
+      "eval_steps_per_second": 21.629,
       "step": 1562
     },
     {
-      "epoch": 12.09,
-      "eval_loss": 0.1355280876159668,
-      "eval_macro_f1": 0.5794821565833095,
-      "eval_macro_precision": 0.6818697605893614,
-      "eval_macro_recall": 0.5236900724860433,
-      "eval_micro_f1": 0.7849877236057524,
-      "eval_micro_precision": 0.8394598649662416,
-      "eval_micro_recall": 0.7371541501976284,
-      "eval_runtime": 2.7759,
-      "eval_samples_per_second": 349.801,
-      "eval_steps_per_second": 11.168,
+      "epoch": 24.0,
+      "eval_loss": 0.10557578504085541,
+      "eval_macro_f1": 0.7298035823714943,
+      "eval_macro_precision": 0.8804692132123697,
+      "eval_macro_recall": 0.6544575680820387,
+      "eval_micro_f1": 0.8399153737658673,
+      "eval_micro_precision": 0.9029567854435178,
+      "eval_micro_recall": 0.7851021753460777,
+      "eval_runtime": 2.8192,
+      "eval_samples_per_second": 342.647,
+      "eval_steps_per_second": 21.637,
       "step": 1704
     },
     {
-      "epoch": 13.09,
-      "eval_loss": 0.13659390807151794,
-      "eval_macro_f1": 0.5891915194817472,
-      "eval_macro_precision": 0.69343533514632,
-      "eval_macro_recall": 0.5302292848265341,
-      "eval_micro_f1": 0.7858642407277815,
-      "eval_micro_precision": 0.8380597014925373,
-      "eval_micro_recall": 0.7397891963109354,
-      "eval_runtime": 2.7739,
-      "eval_samples_per_second": 350.043,
-      "eval_steps_per_second": 11.175,
+      "epoch": 26.0,
+      "eval_loss": 0.10698197782039642,
+      "eval_macro_f1": 0.719671126038702,
+      "eval_macro_precision": 0.8799359996888543,
+      "eval_macro_recall": 0.6422398711062759,
+      "eval_micro_f1": 0.8375706214689265,
+      "eval_micro_precision": 0.9019011406844106,
+      "eval_micro_recall": 0.7818061964403428,
+      "eval_runtime": 2.8198,
+      "eval_samples_per_second": 342.58,
+      "eval_steps_per_second": 21.633,
       "step": 1846
     },
     {
-      "epoch": 14.1,
-      "eval_loss": 0.13488434255123138,
-      "eval_macro_f1": 0.5900914578255574,
-      "eval_macro_precision": 0.6881746937300715,
-      "eval_macro_recall": 0.5327336547593478,
-      "eval_micro_f1": 0.7906326459279972,
-      "eval_micro_precision": 0.842144452717796,
-      "eval_micro_recall": 0.7450592885375494,
-      "eval_runtime": 2.772,
-      "eval_samples_per_second": 350.292,
-      "eval_steps_per_second": 11.183,
+      "epoch": 28.0,
+      "eval_loss": 0.10863872617483139,
+      "eval_macro_f1": 0.7318024399797272,
+      "eval_macro_precision": 0.8781101314061108,
+      "eval_macro_recall": 0.6548756753680312,
+      "eval_micro_f1": 0.8389143461402891,
+      "eval_micro_precision": 0.9015151515151515,
+      "eval_micro_recall": 0.7844429795649308,
+      "eval_runtime": 2.8198,
+      "eval_samples_per_second": 342.583,
+      "eval_steps_per_second": 21.633,
       "step": 1988
     },
     {
-      "epoch": 14.18,
-      "learning_rate": 2.8127777526184293e-05,
-      "loss": 0.0229,
+      "epoch": 28.17,
+      "learning_rate": 2.214818407002782e-05,
+      "loss": 0.0082,
       "step": 2000
     },
     {
-      "epoch": 15.11,
-      "eval_loss": 0.14160068333148956,
-      "eval_macro_f1": 0.5920064296505824,
-      "eval_macro_precision": 0.6848089552512611,
-      "eval_macro_recall": 0.5382146732964487,
-      "eval_micro_f1": 0.7904066736183525,
-      "eval_micro_precision": 0.8366445916114791,
-      "eval_micro_recall": 0.7490118577075099,
-      "eval_runtime": 2.7738,
-      "eval_samples_per_second": 350.061,
-      "eval_steps_per_second": 11.176,
+      "epoch": 30.0,
+      "eval_loss": 0.1103406548500061,
+      "eval_macro_f1": 0.728656455732704,
+      "eval_macro_precision": 0.8784457053881349,
+      "eval_macro_recall": 0.6552278977528909,
+      "eval_micro_f1": 0.8400702987697715,
+      "eval_micro_precision": 0.8998493975903614,
+      "eval_micro_recall": 0.7877389584706658,
+      "eval_runtime": 2.8227,
+      "eval_samples_per_second": 342.231,
+      "eval_steps_per_second": 21.611,
       "step": 2130
     },
     {
-      "epoch": 16.11,
-      "eval_loss": 0.14674818515777588,
-      "eval_macro_f1": 0.5928173166365653,
-      "eval_macro_precision": 0.6894481117094009,
-      "eval_macro_recall": 0.5355030217646541,
-      "eval_micro_f1": 0.7838118298166724,
-      "eval_micro_precision": 0.8252002913328478,
-      "eval_micro_recall": 0.7463768115942029,
-      "eval_runtime": 2.7753,
-      "eval_samples_per_second": 349.871,
-      "eval_steps_per_second": 11.17,
+      "epoch": 32.0,
+      "eval_loss": 0.11085448414087296,
+      "eval_macro_f1": 0.7344651790603305,
+      "eval_macro_precision": 0.8647882525531505,
+      "eval_macro_recall": 0.6579821769650391,
+      "eval_micro_f1": 0.8402116402116402,
+      "eval_micro_precision": 0.9036418816388467,
+      "eval_micro_recall": 0.7851021753460777,
+      "eval_runtime": 2.8204,
+      "eval_samples_per_second": 342.508,
+      "eval_steps_per_second": 21.628,
       "step": 2272
     },
     {
-      "epoch": 17.12,
-      "eval_loss": 0.15975715219974518,
-      "eval_macro_f1": 0.6260279452973108,
-      "eval_macro_precision": 0.7038398537082291,
-      "eval_macro_recall": 0.5866152515992036,
-      "eval_micro_f1": 0.7704974271012006,
-      "eval_micro_precision": 0.80386542591267,
-      "eval_micro_recall": 0.7397891963109354,
-      "eval_runtime": 2.773,
-      "eval_samples_per_second": 350.168,
-      "eval_steps_per_second": 11.179,
+      "epoch": 34.0,
+      "eval_loss": 0.11268670856952667,
+      "eval_macro_f1": 0.7371526225457369,
+      "eval_macro_precision": 0.8672638823695995,
+      "eval_macro_recall": 0.6591834104386446,
+      "eval_micro_f1": 0.8397323001056711,
+      "eval_micro_precision": 0.9016641452344932,
+      "eval_micro_recall": 0.7857613711272248,
+      "eval_runtime": 2.8203,
+      "eval_samples_per_second": 342.515,
+      "eval_steps_per_second": 21.629,
       "step": 2414
     },
     {
-      "epoch": 17.73,
-      "learning_rate": 2.6965472669730396e-05,
-      "loss": 0.0137,
+      "epoch": 35.21,
+      "learning_rate": 1.9976793474927056e-05,
+      "loss": 0.0056,
       "step": 2500
     },
     {
-      "epoch": 18.13,
-      "eval_loss": 0.15081025660037994,
-      "eval_macro_f1": 0.6271302596876451,
-      "eval_macro_precision": 0.7314709578369518,
-      "eval_macro_recall": 0.5704921787075566,
-      "eval_micro_f1": 0.7809989521480964,
-      "eval_micro_precision": 0.8312267657992565,
-      "eval_micro_recall": 0.7364953886693018,
-      "eval_runtime": 2.7721,
-      "eval_samples_per_second": 350.282,
-      "eval_steps_per_second": 11.183,
+      "epoch": 36.0,
+      "eval_loss": 0.11449939757585526,
+      "eval_macro_f1": 0.7435877211427984,
+      "eval_macro_precision": 0.8803719443456752,
+      "eval_macro_recall": 0.6671207777218493,
+      "eval_micro_f1": 0.8435852372583479,
+      "eval_micro_precision": 0.9036144578313253,
+      "eval_micro_recall": 0.7910349373764007,
+      "eval_runtime": 2.8219,
+      "eval_samples_per_second": 342.324,
+      "eval_steps_per_second": 21.617,
       "step": 2556
     },
     {
-      "epoch": 19.13,
-      "eval_loss": 0.14972703158855438,
-      "eval_macro_f1": 0.5953077373311189,
-      "eval_macro_precision": 0.6771131830689183,
-      "eval_macro_recall": 0.5419002355296083,
-      "eval_micro_f1": 0.7900552486187846,
-      "eval_micro_precision": 0.8301886792452831,
-      "eval_micro_recall": 0.7536231884057971,
-      "eval_runtime": 2.7725,
-      "eval_samples_per_second": 350.229,
-      "eval_steps_per_second": 11.181,
+      "epoch": 38.0,
+      "eval_loss": 0.1153542622923851,
+      "eval_macro_f1": 0.7450728657264144,
+      "eval_macro_precision": 0.879277762192774,
+      "eval_macro_recall": 0.6681699293595025,
+      "eval_micro_f1": 0.8419570573741639,
+      "eval_micro_precision": 0.9033232628398792,
+      "eval_micro_recall": 0.7883981542518128,
+      "eval_runtime": 2.8191,
+      "eval_samples_per_second": 342.661,
+      "eval_steps_per_second": 21.638,
       "step": 2698
     },
     {
-      "epoch": 20.14,
-      "eval_loss": 0.16402311623096466,
-      "eval_macro_f1": 0.5892530505072842,
-      "eval_macro_precision": 0.6727972959139897,
-      "eval_macro_recall": 0.5367187180351005,
-      "eval_micro_f1": 0.779319916724497,
-      "eval_micro_precision": 0.8233137829912024,
-      "eval_micro_recall": 0.7397891963109354,
-      "eval_runtime": 2.7746,
-      "eval_samples_per_second": 349.963,
-      "eval_steps_per_second": 11.173,
+      "epoch": 40.0,
+      "eval_loss": 0.11756419390439987,
+      "eval_macro_f1": 0.744466331732378,
+      "eval_macro_precision": 0.8796020530253235,
+      "eval_macro_recall": 0.6677481100215579,
+      "eval_micro_f1": 0.8419570573741639,
+      "eval_micro_precision": 0.9033232628398792,
+      "eval_micro_recall": 0.7883981542518128,
+      "eval_runtime": 2.8181,
+      "eval_samples_per_second": 342.785,
+      "eval_steps_per_second": 21.646,
       "step": 2840
     },
     {
-      "epoch": 21.15,
-      "eval_loss": 0.16335928440093994,
-      "eval_macro_f1": 0.6332863501223459,
-      "eval_macro_precision": 0.7578187629295322,
-      "eval_macro_recall": 0.5685989504723161,
-      "eval_micro_f1": 0.7924791086350974,
-      "eval_micro_precision": 0.8404726735598228,
-      "eval_micro_recall": 0.7496706192358367,
-      "eval_runtime": 2.7725,
-      "eval_samples_per_second": 350.229,
-      "eval_steps_per_second": 11.181,
+      "epoch": 42.0,
+      "eval_loss": 0.11846820265054703,
+      "eval_macro_f1": 0.7434279482403103,
+      "eval_macro_precision": 0.8813168271357439,
+      "eval_macro_recall": 0.6651456053129234,
+      "eval_micro_f1": 0.8427717200140696,
+      "eval_micro_precision": 0.9034690799396682,
+      "eval_micro_recall": 0.7897165458141068,
+      "eval_runtime": 2.8203,
+      "eval_samples_per_second": 342.52,
+      "eval_steps_per_second": 21.629,
       "step": 2982
     },
     {
-      "epoch": 21.28,
-      "learning_rate": 2.58031678132765e-05,
-      "loss": 0.011,
+      "epoch": 42.25,
+      "learning_rate": 1.7805402879826288e-05,
+      "loss": 0.004,
       "step": 3000
     },
     {
-      "epoch": 22.16,
-      "eval_loss": 0.16375195980072021,
-      "eval_macro_f1": 0.6343242676719218,
-      "eval_macro_precision": 0.7473720060268463,
-      "eval_macro_recall": 0.5714537925969617,
-      "eval_micro_f1": 0.7909878682842287,
-      "eval_micro_precision": 0.8346744696415508,
-      "eval_micro_recall": 0.7516469038208169,
-      "eval_runtime": 2.7735,
-      "eval_samples_per_second": 350.093,
-      "eval_steps_per_second": 11.177,
+      "epoch": 44.0,
+      "eval_loss": 0.11996057629585266,
+      "eval_macro_f1": 0.7394753977857482,
+      "eval_macro_precision": 0.8783969916238008,
+      "eval_macro_recall": 0.6650662737686901,
+      "eval_micro_f1": 0.8427717200140696,
+      "eval_micro_precision": 0.9034690799396682,
+      "eval_micro_recall": 0.7897165458141068,
+      "eval_runtime": 2.82,
+      "eval_samples_per_second": 342.552,
+      "eval_steps_per_second": 21.631,
       "step": 3124
     },
     {
-      "epoch": 23.16,
-      "eval_loss": 0.16090567409992218,
-      "eval_macro_f1": 0.6491296227815271,
-      "eval_macro_precision": 0.7487080551943797,
-      "eval_macro_recall": 0.5927079340096836,
-      "eval_micro_f1": 0.7961299239806496,
-      "eval_micro_precision": 0.8372093023255814,
-      "eval_micro_recall": 0.758893280632411,
-      "eval_runtime": 2.7735,
-      "eval_samples_per_second": 350.094,
-      "eval_steps_per_second": 11.177,
+      "epoch": 46.0,
+      "eval_loss": 0.12046220153570175,
+      "eval_macro_f1": 0.7423287819151947,
+      "eval_macro_precision": 0.8810998402784964,
+      "eval_macro_recall": 0.6639465302563645,
+      "eval_micro_f1": 0.8432546671363156,
+      "eval_micro_precision": 0.905446293494705,
+      "eval_micro_recall": 0.7890573500329597,
+      "eval_runtime": 2.819,
+      "eval_samples_per_second": 342.677,
+      "eval_steps_per_second": 21.639,
       "step": 3266
+    },
+    {
+      "epoch": 48.0,
+      "eval_loss": 0.12299305200576782,
+      "eval_macro_f1": 0.7584256567760465,
+      "eval_macro_precision": 0.9421857420436691,
+      "eval_macro_recall": 0.6745195368122752,
+      "eval_micro_f1": 0.8419570573741639,
+      "eval_micro_precision": 0.9033232628398792,
+      "eval_micro_recall": 0.7883981542518128,
+      "eval_runtime": 2.8184,
+      "eval_samples_per_second": 342.744,
+      "eval_steps_per_second": 21.643,
+      "step": 3408
+    },
+    {
+      "epoch": 49.3,
+      "learning_rate": 1.563401228472552e-05,
+      "loss": 0.0031,
+      "step": 3500
+    },
+    {
+      "epoch": 50.0,
+      "eval_loss": 0.12287621200084686,
+      "eval_macro_f1": 0.7672541719320296,
+      "eval_macro_precision": 0.9303028445349943,
+      "eval_macro_recall": 0.6844658294965831,
+      "eval_micro_f1": 0.8449122807017543,
+      "eval_micro_precision": 0.9032258064516129,
+      "eval_micro_recall": 0.7936717205009888,
+      "eval_runtime": 2.8187,
+      "eval_samples_per_second": 342.711,
+      "eval_steps_per_second": 21.641,
+      "step": 3550
     }
   ],
-  "max_steps": 14100,
+  "max_steps": 7100,
   "num_train_epochs": 100,
-  "total_flos": 1.2446273222900736e+16,
+  "total_flos": 3.007255197834797e+16,
   "trial_name": null,
   "trial_params": {
-    "adam_epsilon": 6.356433345691159e-10,
-    "learning_rate": 3.084757089028641e-05,
-    "per_device_eval_batch_size": 32,
-    "per_device_train_batch_size": 16,
-    "seed": 326,
-    "warmup_steps": 830,
-    "weight_decay": 0.0009910374448883887
+    "adam_epsilon": 6.447418463180699e-08,
+    "learning_rate": 3.0290898801655698e-05,
+    "per_device_eval_batch_size": 16,
+    "per_device_train_batch_size": 32,
+    "seed": 320,
+    "warmup_steps": 125,
+    "weight_decay": 4.5126980713116176e-08
   }
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8b7f323c1b5dbfa731b4c4816647448376c6ed3b12e43aeda40bcbd1093c542
+oid sha256:f8c9e36bee277bf832a3d25b8f22fe9e82c88fe3d841e639970624efd5b1bd10
 size 3439