Upload folder using huggingface_hub

training_checkpoints/checkpoint-9645/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3fce45b115b232a03755bc389bf96ae449717fb4e34dd6f0815441f86e005e0
+oid sha256:d5a43044f3883b5937338842d04d582eea99655defaa4ff1e77553e429df37f9
 size 498652812

training_checkpoints/checkpoint-9645/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aeadf0536441f75b4bab51564512611c8b0488ba60b41cc25c020d21e38ca40b
+oid sha256:ed8dd2655c8a3227c36f2923c26e803748cf4f8ce86e33ff2419fb6a235f566f
 size 997425402

training_checkpoints/checkpoint-9645/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2fe148baf5ae4abec5ba0f1cd141ef03f380ade8d0f9fa21aa75fa9ef31a6675
+oid sha256:2623804f91f689ff1e336231975ac841f18f91a752aa8dc11c4b026dff4a80b8
 size 14244

training_checkpoints/checkpoint-9645/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:05ce34d26c60a61f9275f5b7997ecbd615a6dc14ec6c37d3e624bdcc1c4a6f10
+oid sha256:bb87ef656eef2e06eb1a123d6b14bdd40712b87ff8494549a22ba029d8fac413
 size 1064

training_checkpoints/checkpoint-9645/trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.3416002470875682,
-  "best_model_checkpoint": "roberta-base/checkpoint-9645",
+  "best_metric": 0.7850116201409046,
+  "best_model_checkpoint": "roberta-base/checkpoint-7073",
   "epoch": 15.0,
   "eval_steps": 500,
   "global_step": 9645,
@@ -10,465 +10,465 @@
   "log_history": [
     {
       "epoch": 0.78,
-      "grad_norm": 6.534026145935059,
+      "grad_norm": 16.478424072265625,
       "learning_rate": 4.742871954380509e-05,
-      "loss": 2.2214,
+      "loss": 1.4921,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.24554608830364058,
-      "eval_f1_macro": 0.02628524046434494,
-      "eval_f1_micro": 0.24554608830364058,
-      "eval_f1_weighted": 0.09681356964210704,
-      "eval_loss": 2.1982107162475586,
-      "eval_macro_fpr": 0.06666666666666667,
-      "eval_macro_sensitivity": 0.06666666666666667,
-      "eval_macro_specificity": 0.9333333333333333,
-      "eval_precision": 0.06029288148121926,
-      "eval_precision_macro": 0.016369739220242704,
-      "eval_recall": 0.24554608830364058,
-      "eval_recall_macro": 0.06666666666666667,
-      "eval_runtime": 29.4439,
-      "eval_samples_per_second": 43.846,
-      "eval_steps_per_second": 5.502,
-      "eval_weighted_fpr": 0.17997043606799704,
-      "eval_weighted_sensitivity": 0.24554608830364058,
-      "eval_weighted_specificity": 0.7544539116963594,
+      "eval_accuracy": 0.6762199845081333,
+      "eval_f1_macro": 0.36762215914350266,
+      "eval_f1_micro": 0.6762199845081333,
+      "eval_f1_weighted": 0.6381553142782734,
+      "eval_loss": 1.0307424068450928,
+      "eval_macro_fpr": 0.03508857639652692,
+      "eval_macro_sensitivity": 0.4027183655220987,
+      "eval_macro_specificity": 0.9746655126680946,
+      "eval_precision": 0.6361951106730956,
+      "eval_precision_macro": 0.3745780279088578,
+      "eval_recall": 0.6762199845081333,
+      "eval_recall_macro": 0.4027183655220987,
+      "eval_runtime": 65.0146,
+      "eval_samples_per_second": 19.857,
+      "eval_steps_per_second": 2.492,
+      "eval_weighted_fpr": 0.03306962025316456,
+      "eval_weighted_sensitivity": 0.6762199845081333,
+      "eval_weighted_specificity": 0.9437627055132857,
       "step": 643
     },
     {
       "epoch": 1.56,
-      "grad_norm": 6.595059871673584,
+      "grad_norm": 11.052178382873535,
       "learning_rate": 4.483670295489891e-05,
-      "loss": 2.1462,
+      "loss": 0.9019,
       "step": 1000
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.3361735089078234,
-      "eval_f1_macro": 0.06552880630375763,
-      "eval_f1_micro": 0.3361735089078234,
-      "eval_f1_weighted": 0.2010480796875688,
-      "eval_loss": 1.9634978771209717,
-      "eval_macro_fpr": 0.09092506634165738,
-      "eval_macro_sensitivity": 0.11547020741602194,
-      "eval_macro_specificity": 0.9457066851291582,
-      "eval_precision": 0.1591168191834871,
-      "eval_precision_macro": 0.05308920125936356,
-      "eval_recall": 0.3361735089078234,
-      "eval_recall_macro": 0.11547020741602194,
-      "eval_runtime": 57.2389,
-      "eval_samples_per_second": 22.555,
-      "eval_steps_per_second": 2.83,
-      "eval_weighted_fpr": 0.12361171210154334,
-      "eval_weighted_sensitivity": 0.3361735089078234,
-      "eval_weighted_specificity": 0.84942676802955,
+      "eval_accuracy": 0.737412858249419,
+      "eval_f1_macro": 0.4767493392755858,
+      "eval_f1_micro": 0.737412858249419,
+      "eval_f1_weighted": 0.7197263095885438,
+      "eval_loss": 0.8135831952095032,
+      "eval_macro_fpr": 0.025093207469991694,
+      "eval_macro_sensitivity": 0.5148741493607558,
+      "eval_macro_specificity": 0.9804978566683986,
+      "eval_precision": 0.7158327481302543,
+      "eval_precision_macro": 0.4625056436667857,
+      "eval_recall": 0.737412858249419,
+      "eval_recall_macro": 0.5148741493607558,
+      "eval_runtime": 78.3833,
+      "eval_samples_per_second": 16.47,
+      "eval_steps_per_second": 2.067,
+      "eval_weighted_fpr": 0.024804273066510573,
+      "eval_weighted_sensitivity": 0.737412858249419,
+      "eval_weighted_specificity": 0.9700549917765612,
       "step": 1286
     },
     {
       "epoch": 2.33,
-      "grad_norm": 41.05898666381836,
-      "learning_rate": 4.224987039917055e-05,
-      "loss": 1.995,
+      "grad_norm": 16.875791549682617,
+      "learning_rate": 4.224468636599274e-05,
+      "loss": 0.7714,
       "step": 1500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.3555383423702556,
-      "eval_f1_macro": 0.09312335075016234,
-      "eval_f1_micro": 0.3555383423702556,
-      "eval_f1_weighted": 0.2541975590745503,
-      "eval_loss": 1.932247281074524,
-      "eval_macro_fpr": 0.09537141294157879,
-      "eval_macro_sensitivity": 0.14351969127084646,
-      "eval_macro_specificity": 0.9497795632717652,
-      "eval_precision": 0.23412920314698393,
-      "eval_precision_macro": 0.09260367628400626,
-      "eval_recall": 0.3555383423702556,
-      "eval_recall_macro": 0.14351969127084646,
-      "eval_runtime": 59.8887,
-      "eval_samples_per_second": 21.557,
-      "eval_steps_per_second": 2.705,
-      "eval_weighted_fpr": 0.11463213006337834,
-      "eval_weighted_sensitivity": 0.3555383423702556,
-      "eval_weighted_specificity": 0.8911551067062222,
+      "eval_accuracy": 0.7621998450813323,
+      "eval_f1_macro": 0.5064301619528891,
+      "eval_f1_micro": 0.7621998450813323,
+      "eval_f1_weighted": 0.7440787580787277,
+      "eval_loss": 0.8493772149085999,
+      "eval_macro_fpr": 0.022439395825223803,
+      "eval_macro_sensitivity": 0.5353612181578932,
+      "eval_macro_specificity": 0.9822327546765126,
+      "eval_precision": 0.7385639759374297,
+      "eval_precision_macro": 0.505572707108006,
+      "eval_recall": 0.7621998450813323,
+      "eval_recall_macro": 0.5353612181578932,
+      "eval_runtime": 62.8149,
+      "eval_samples_per_second": 20.552,
+      "eval_steps_per_second": 2.579,
+      "eval_weighted_fpr": 0.02179933252858056,
+      "eval_weighted_sensitivity": 0.7621998450813323,
+      "eval_weighted_specificity": 0.9712914750663563,
       "step": 1929
     },
     {
       "epoch": 3.11,
-      "grad_norm": 4.287881851196289,
-      "learning_rate": 3.96630378434422e-05,
-      "loss": 2.0064,
+      "grad_norm": 59.48842239379883,
+      "learning_rate": 3.9652669777086575e-05,
+      "loss": 0.7018,
       "step": 2000
     },
     {
       "epoch": 3.89,
-      "grad_norm": 5.713494777679443,
-      "learning_rate": 3.7076205287713846e-05,
-      "loss": 1.8909,
+      "grad_norm": 1.6855363845825195,
+      "learning_rate": 3.706583722135822e-05,
+      "loss": 0.5351,
       "step": 2500
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.3787761425251743,
-      "eval_f1_macro": 0.14496386130680605,
-      "eval_f1_micro": 0.3787761425251743,
-      "eval_f1_weighted": 0.29773714484044667,
-      "eval_loss": 1.8874640464782715,
-      "eval_macro_fpr": 0.08706956494211866,
-      "eval_macro_sensitivity": 0.2191470222337379,
-      "eval_macro_specificity": 0.9525422186332462,
-      "eval_precision": 0.28749388015966415,
-      "eval_precision_macro": 0.12703293957809556,
-      "eval_recall": 0.3787761425251743,
-      "eval_recall_macro": 0.2191470222337379,
-      "eval_runtime": 69.4305,
-      "eval_samples_per_second": 18.594,
-      "eval_steps_per_second": 2.333,
-      "eval_weighted_fpr": 0.10486401673640168,
-      "eval_weighted_sensitivity": 0.3787761425251743,
-      "eval_weighted_specificity": 0.9093571369735185,
+      "eval_accuracy": 0.7645236250968241,
+      "eval_f1_macro": 0.544445143585482,
+      "eval_f1_micro": 0.7645236250968241,
+      "eval_f1_weighted": 0.7545581478665218,
+      "eval_loss": 0.9904704093933105,
+      "eval_macro_fpr": 0.022368675533492623,
+      "eval_macro_sensitivity": 0.5653277909563001,
+      "eval_macro_specificity": 0.9821237739107975,
+      "eval_precision": 0.7656722240785093,
+      "eval_precision_macro": 0.5495765557734748,
+      "eval_recall": 0.7645236250968241,
+      "eval_recall_macro": 0.5653277909563001,
+      "eval_runtime": 80.7211,
+      "eval_samples_per_second": 15.993,
+      "eval_steps_per_second": 2.007,
+      "eval_weighted_fpr": 0.021526695935419914,
+      "eval_weighted_sensitivity": 0.7645236250968241,
+      "eval_weighted_specificity": 0.9673329835651372,
       "step": 2572
     },
     {
       "epoch": 4.67,
-      "grad_norm": 18.91859245300293,
-      "learning_rate": 3.4484188698807675e-05,
-      "loss": 1.7621,
+      "grad_norm": 41.18514633178711,
+      "learning_rate": 3.447382063245205e-05,
+      "loss": 0.4521,
       "step": 3000
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.42835011618900076,
-      "eval_f1_macro": 0.1745447316852942,
-      "eval_f1_micro": 0.4283501161890007,
-      "eval_f1_weighted": 0.3593798293323463,
-      "eval_loss": 1.7168936729431152,
-      "eval_macro_fpr": 0.07431772683182448,
-      "eval_macro_sensitivity": 0.2270221309090987,
-      "eval_macro_specificity": 0.95583829077833,
-      "eval_precision": 0.374728097525222,
-      "eval_precision_macro": 0.18202607308331148,
-      "eval_recall": 0.42835011618900076,
-      "eval_recall_macro": 0.2270221309090987,
-      "eval_runtime": 58.7135,
-      "eval_samples_per_second": 21.988,
-      "eval_steps_per_second": 2.759,
-      "eval_weighted_fpr": 0.08702830188679245,
-      "eval_weighted_sensitivity": 0.42835011618900076,
-      "eval_weighted_specificity": 0.9092242454859507,
+      "eval_accuracy": 0.7900852052672347,
+      "eval_f1_macro": 0.6116525431724917,
+      "eval_f1_micro": 0.7900852052672347,
+      "eval_f1_weighted": 0.7794301250682801,
+      "eval_loss": 1.0259578227996826,
+      "eval_macro_fpr": 0.019043445988564375,
+      "eval_macro_sensitivity": 0.6315170743424566,
+      "eval_macro_specificity": 0.984305159354804,
+      "eval_precision": 0.7882093327742286,
+      "eval_precision_macro": 0.6420940275483907,
+      "eval_recall": 0.7900852052672347,
+      "eval_recall_macro": 0.6315170743424566,
+      "eval_runtime": 56.9964,
+      "eval_samples_per_second": 22.651,
+      "eval_steps_per_second": 2.842,
+      "eval_weighted_fpr": 0.018624149542986735,
+      "eval_weighted_sensitivity": 0.7900852052672347,
+      "eval_weighted_specificity": 0.9744921850548264,
       "step": 3215
     },
     {
       "epoch": 5.44,
-      "grad_norm": 7.408999919891357,
-      "learning_rate": 3.1892172109901504e-05,
-      "loss": 1.6254,
+      "grad_norm": 92.02596282958984,
+      "learning_rate": 3.1881804043545884e-05,
+      "loss": 0.3466,
       "step": 3500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.4500387296669249,
-      "eval_f1_macro": 0.18588331301714964,
-      "eval_f1_micro": 0.4500387296669249,
-      "eval_f1_weighted": 0.39697994452590485,
-      "eval_loss": 1.8831233978271484,
-      "eval_macro_fpr": 0.07944022523045574,
-      "eval_macro_sensitivity": 0.2544695464493438,
-      "eval_macro_specificity": 0.9582711519830998,
-      "eval_precision": 0.3947542941359455,
-      "eval_precision_macro": 0.1877780146030796,
-      "eval_recall": 0.4500387296669249,
-      "eval_recall_macro": 0.2544695464493438,
-      "eval_runtime": 65.7966,
-      "eval_samples_per_second": 19.621,
-      "eval_steps_per_second": 2.462,
-      "eval_weighted_fpr": 0.08028041610131162,
-      "eval_weighted_sensitivity": 0.4500387296669249,
-      "eval_weighted_specificity": 0.92402855007957,
+      "eval_accuracy": 0.7869868319132456,
+      "eval_f1_macro": 0.6118575336738351,
+      "eval_f1_micro": 0.7869868319132455,
+      "eval_f1_weighted": 0.7889880056842663,
+      "eval_loss": 1.0384949445724487,
+      "eval_macro_fpr": 0.019220806323432427,
+      "eval_macro_sensitivity": 0.633518783056355,
+      "eval_macro_specificity": 0.984312476194111,
+      "eval_precision": 0.8134105657387156,
+      "eval_precision_macro": 0.6722188161248251,
+      "eval_recall": 0.7869868319132456,
+      "eval_recall_macro": 0.633518783056355,
+      "eval_runtime": 58.1615,
+      "eval_samples_per_second": 22.197,
+      "eval_steps_per_second": 2.785,
+      "eval_weighted_fpr": 0.018966825298296436,
+      "eval_weighted_sensitivity": 0.7869868319132456,
+      "eval_weighted_specificity": 0.9777003109984174,
       "step": 3858
     },
     {
       "epoch": 6.22,
-      "grad_norm": 8.139199256896973,
-      "learning_rate": 2.9300155520995337e-05,
-      "loss": 1.4996,
+      "grad_norm": 80.3629379272461,
+      "learning_rate": 2.9294971487817523e-05,
+      "loss": 0.3321,
       "step": 4000
     },
     {
       "epoch": 7.0,
-      "grad_norm": 40.22551727294922,
-      "learning_rate": 2.671332296526698e-05,
-      "loss": 1.4139,
+      "grad_norm": 0.3377046585083008,
+      "learning_rate": 2.6702954898911353e-05,
+      "loss": 0.2333,
       "step": 4500
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.46398140975987606,
-      "eval_f1_macro": 0.22845278849292336,
-      "eval_f1_micro": 0.4639814097598761,
-      "eval_f1_weighted": 0.4297793996222424,
-      "eval_loss": 1.662539005279541,
-      "eval_macro_fpr": 0.07171513504664777,
-      "eval_macro_sensitivity": 0.2949397340388095,
-      "eval_macro_specificity": 0.9601261188979237,
-      "eval_precision": 0.451602882474783,
-      "eval_precision_macro": 0.232765042661752,
-      "eval_recall": 0.46398140975987606,
-      "eval_recall_macro": 0.2949397340388095,
-      "eval_runtime": 60.1854,
-      "eval_samples_per_second": 21.45,
-      "eval_steps_per_second": 2.692,
-      "eval_weighted_fpr": 0.07622824410663141,
-      "eval_weighted_sensitivity": 0.46398140975987606,
-      "eval_weighted_specificity": 0.9379103737089811,
+      "eval_accuracy": 0.8117738187451587,
+      "eval_f1_macro": 0.665659747836216,
+      "eval_f1_micro": 0.8117738187451587,
+      "eval_f1_weighted": 0.8057624075524522,
+      "eval_loss": 1.1464879512786865,
+      "eval_macro_fpr": 0.01693769060060091,
+      "eval_macro_sensitivity": 0.6784424177464591,
+      "eval_macro_specificity": 0.9857572234742,
+      "eval_precision": 0.8064435087918365,
+      "eval_precision_macro": 0.6794791733053616,
+      "eval_recall": 0.8117738187451587,
+      "eval_recall_macro": 0.6784424177464591,
+      "eval_runtime": 54.6308,
+      "eval_samples_per_second": 23.631,
+      "eval_steps_per_second": 2.965,
+      "eval_weighted_fpr": 0.016292323164599398,
+      "eval_weighted_sensitivity": 0.8117738187451587,
+      "eval_weighted_specificity": 0.9745845333678407,
       "step": 4501
     },
     {
       "epoch": 7.78,
-      "grad_norm": 5.4409565925598145,
-      "learning_rate": 2.412130637636081e-05,
-      "loss": 1.2681,
+      "grad_norm": 0.19802448153495789,
+      "learning_rate": 2.4110938310005186e-05,
+      "loss": 0.1658,
       "step": 5000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5383423702556158,
-      "eval_f1_macro": 0.2703391292878622,
-      "eval_f1_micro": 0.5383423702556158,
-      "eval_f1_weighted": 0.48876940771692845,
-      "eval_loss": 1.4461586475372314,
-      "eval_macro_fpr": 0.056020043825060734,
-      "eval_macro_sensitivity": 0.3381295733874737,
-      "eval_macro_specificity": 0.9655641856187219,
-      "eval_precision": 0.5269280210096271,
-      "eval_precision_macro": 0.2765942082966614,
-      "eval_recall": 0.5383423702556158,
-      "eval_recall_macro": 0.3381295733874737,
-      "eval_runtime": 64.8556,
-      "eval_samples_per_second": 19.906,
-      "eval_steps_per_second": 2.498,
-      "eval_weighted_fpr": 0.057718380786364515,
-      "eval_weighted_sensitivity": 0.5383423702556158,
-      "eval_weighted_specificity": 0.9451204140252133,
+      "eval_accuracy": 0.814872192099148,
+      "eval_f1_macro": 0.7328612504365998,
+      "eval_f1_micro": 0.814872192099148,
+      "eval_f1_weighted": 0.8120704640541792,
+      "eval_loss": 1.241939902305603,
+      "eval_macro_fpr": 0.016489901852411035,
+      "eval_macro_sensitivity": 0.7440371076362254,
+      "eval_macro_specificity": 0.9861286253461167,
+      "eval_precision": 0.8148683401292235,
+      "eval_precision_macro": 0.7318970148814324,
+      "eval_recall": 0.814872192099148,
+      "eval_recall_macro": 0.7440371076362254,
+      "eval_runtime": 57.719,
+      "eval_samples_per_second": 22.367,
+      "eval_steps_per_second": 2.807,
+      "eval_weighted_fpr": 0.015968463954032203,
+      "eval_weighted_sensitivity": 0.814872192099148,
+      "eval_weighted_specificity": 0.9770571880926026,
       "step": 5144
     },
     {
       "epoch": 8.55,
-      "grad_norm": 6.948422908782959,
-      "learning_rate": 2.152928978745464e-05,
-      "loss": 1.2351,
+      "grad_norm": 0.0870741754770279,
+      "learning_rate": 2.151892172109902e-05,
+      "loss": 0.1597,
       "step": 5500
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.5515104570100697,
-      "eval_f1_macro": 0.29219268534520737,
-      "eval_f1_micro": 0.5515104570100697,
-      "eval_f1_weighted": 0.5183378322557094,
-      "eval_loss": 1.3802897930145264,
-      "eval_macro_fpr": 0.05395959885799773,
-      "eval_macro_sensitivity": 0.33994734275998406,
-      "eval_macro_specificity": 0.966222282286659,
-      "eval_precision": 0.5182882506526222,
-      "eval_precision_macro": 0.2800150936557356,
-      "eval_recall": 0.5515104570100697,
-      "eval_recall_macro": 0.33994734275998406,
-      "eval_runtime": 95.1023,
-      "eval_samples_per_second": 13.575,
-      "eval_steps_per_second": 1.703,
-      "eval_weighted_fpr": 0.05489712714515976,
-      "eval_weighted_sensitivity": 0.5515104570100697,
-      "eval_weighted_specificity": 0.9418237772898147,
+      "eval_accuracy": 0.8179705654531371,
+      "eval_f1_macro": 0.7804377723463752,
+      "eval_f1_micro": 0.817970565453137,
+      "eval_f1_weighted": 0.819003836369197,
+      "eval_loss": 1.3440543413162231,
+      "eval_macro_fpr": 0.015977205363795557,
+      "eval_macro_sensitivity": 0.770217713596195,
+      "eval_macro_specificity": 0.9863181364411756,
+      "eval_precision": 0.8259393911103454,
+      "eval_precision_macro": 0.8314287989687705,
+      "eval_recall": 0.8179705654531371,
+      "eval_recall_macro": 0.770217713596195,
+      "eval_runtime": 90.353,
+      "eval_samples_per_second": 14.288,
+      "eval_steps_per_second": 1.793,
+      "eval_weighted_fpr": 0.01564684732671949,
+      "eval_weighted_sensitivity": 0.8179705654531371,
+      "eval_weighted_specificity": 0.9768014811644973,
       "step": 5787
     },
     {
       "epoch": 9.33,
-      "grad_norm": 17.84939956665039,
-      "learning_rate": 1.8937273198548472e-05,
-      "loss": 1.1161,
+      "grad_norm": 0.013821585103869438,
+      "learning_rate": 1.8926905132192844e-05,
+      "loss": 0.11,
       "step": 6000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5747482571649883,
-      "eval_f1_macro": 0.3016038026372148,
-      "eval_f1_micro": 0.5747482571649883,
-      "eval_f1_weighted": 0.5350722531862059,
-      "eval_loss": 1.3405604362487793,
-      "eval_macro_fpr": 0.050184494231821634,
-      "eval_macro_sensitivity": 0.35749011781425877,
-      "eval_macro_specificity": 0.968005386818716,
-      "eval_precision": 0.5342322193126818,
-      "eval_precision_macro": 0.2899738892872161,
-      "eval_recall": 0.5747482571649883,
-      "eval_recall_macro": 0.35749011781425877,
-      "eval_runtime": 59.1957,
-      "eval_samples_per_second": 21.809,
-      "eval_steps_per_second": 2.737,
-      "eval_weighted_fpr": 0.05019658041510469,
-      "eval_weighted_sensitivity": 0.5747482571649883,
-      "eval_weighted_specificity": 0.945332545115751,
+      "eval_accuracy": 0.8024786986831913,
+      "eval_f1_macro": 0.7500128517100584,
+      "eval_f1_micro": 0.8024786986831914,
+      "eval_f1_weighted": 0.8064576225913597,
+      "eval_loss": 1.3504691123962402,
+      "eval_macro_fpr": 0.01779575845463305,
+      "eval_macro_sensitivity": 0.765413440395755,
+      "eval_macro_specificity": 0.9852391037313044,
+      "eval_precision": 0.8152310151046661,
+      "eval_precision_macro": 0.7519799672719661,
+      "eval_recall": 0.8024786986831913,
+      "eval_recall_macro": 0.765413440395755,
+      "eval_runtime": 55.2658,
+      "eval_samples_per_second": 23.36,
+      "eval_steps_per_second": 2.931,
+      "eval_weighted_fpr": 0.017277593332881633,
+      "eval_weighted_sensitivity": 0.8024786986831913,
+      "eval_weighted_specificity": 0.9761078572863756,
       "step": 6430
     },
     {
       "epoch": 10.11,
-      "grad_norm": 34.46049499511719,
-      "learning_rate": 1.6345256609642305e-05,
-      "loss": 1.0384,
+      "grad_norm": 0.009372069500386715,
+      "learning_rate": 1.6334888543286677e-05,
+      "loss": 0.0741,
       "step": 6500
     },
     {
       "epoch": 10.89,
-      "grad_norm": 18.093345642089844,
-      "learning_rate": 1.3753240020736132e-05,
-      "loss": 0.9747,
+      "grad_norm": 0.017231620848178864,
+      "learning_rate": 1.374805598755832e-05,
+      "loss": 0.0747,
       "step": 7000
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.6150271107668474,
-      "eval_f1_macro": 0.3261856936782915,
-      "eval_f1_micro": 0.6150271107668474,
-      "eval_f1_weighted": 0.5766051994823087,
-      "eval_loss": 1.380952000617981,
-      "eval_macro_fpr": 0.0444013730647484,
-      "eval_macro_sensitivity": 0.3747372380690592,
-      "eval_macro_specificity": 0.9705808122777527,
-      "eval_precision": 0.5606186818865333,
-      "eval_precision_macro": 0.3083540755636765,
-      "eval_recall": 0.6150271107668474,
-      "eval_recall_macro": 0.3747372380690592,
-      "eval_runtime": 54.8103,
-      "eval_samples_per_second": 23.554,
-      "eval_steps_per_second": 2.956,
-      "eval_weighted_fpr": 0.04279686558167571,
-      "eval_weighted_sensitivity": 0.6150271107668474,
-      "eval_weighted_specificity": 0.9436850733994432,
+      "eval_accuracy": 0.8280402788536019,
+      "eval_f1_macro": 0.7850116201409046,
+      "eval_f1_micro": 0.8280402788536019,
+      "eval_f1_weighted": 0.8252613057748646,
+      "eval_loss": 1.3682292699813843,
+      "eval_macro_fpr": 0.015248972295162136,
+      "eval_macro_sensitivity": 0.7819796586438489,
+      "eval_macro_specificity": 0.9869417363825405,
+      "eval_precision": 0.8277410447478016,
+      "eval_precision_macro": 0.8096252898700513,
+      "eval_recall": 0.8280402788536019,
+      "eval_recall_macro": 0.7819796586438489,
+      "eval_runtime": 58.5549,
+      "eval_samples_per_second": 22.048,
+      "eval_steps_per_second": 2.767,
+      "eval_weighted_fpr": 0.014616802739004477,
+      "eval_weighted_sensitivity": 0.8280402788536019,
+      "eval_weighted_specificity": 0.976085766884506,
       "step": 7073
     },
     {
       "epoch": 11.66,
-      "grad_norm": 18.54058265686035,
-      "learning_rate": 1.1161223431829965e-05,
-      "loss": 0.9145,
+      "grad_norm": 0.039045702666044235,
+      "learning_rate": 1.1156039398652151e-05,
+      "loss": 0.0519,
       "step": 7500
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.6196746707978311,
-      "eval_f1_macro": 0.33300079962435747,
-      "eval_f1_micro": 0.6196746707978311,
-      "eval_f1_weighted": 0.5913426830324416,
-      "eval_loss": 1.273578405380249,
-      "eval_macro_fpr": 0.04205592092867989,
-      "eval_macro_sensitivity": 0.38075990989198766,
-      "eval_macro_specificity": 0.9713642071716353,
-      "eval_precision": 0.5857440172055212,
-      "eval_precision_macro": 0.3198823734674615,
-      "eval_recall": 0.6196746707978311,
-      "eval_recall_macro": 0.38075990989198766,
-      "eval_runtime": 53.8432,
-      "eval_samples_per_second": 23.977,
-      "eval_steps_per_second": 3.009,
-      "eval_weighted_fpr": 0.041998118210589346,
-      "eval_weighted_sensitivity": 0.6196746707978311,
-      "eval_weighted_specificity": 0.9507884367766963,
+      "eval_accuracy": 0.8164213787761425,
+      "eval_f1_macro": 0.7698779986816631,
+      "eval_f1_micro": 0.8164213787761425,
+      "eval_f1_weighted": 0.8170206182613167,
+      "eval_loss": 1.4436842203140259,
+      "eval_macro_fpr": 0.016423979995821146,
+      "eval_macro_sensitivity": 0.7750707038634387,
+      "eval_macro_specificity": 0.9861294812620465,
+      "eval_precision": 0.8188455079677682,
+      "eval_precision_macro": 0.7730875104850559,
+      "eval_recall": 0.8164213787761425,
+      "eval_recall_macro": 0.7750707038634387,
+      "eval_runtime": 56.1974,
+      "eval_samples_per_second": 22.973,
+      "eval_steps_per_second": 2.883,
+      "eval_weighted_fpr": 0.01580737677582872,
+      "eval_weighted_sensitivity": 0.8164213787761425,
+      "eval_weighted_specificity": 0.9755208401545548,
       "step": 7716
     },
     {
       "epoch": 12.44,
-      "grad_norm": 2.550917863845825,
-      "learning_rate": 8.574390876101608e-06,
-      "loss": 0.8879,
+      "grad_norm": 220.98574829101562,
+      "learning_rate": 8.564022809745984e-06,
+      "loss": 0.0324,
       "step": 8000
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.635166537567777,
-      "eval_f1_macro": 0.3368012141860189,
-      "eval_f1_micro": 0.635166537567777,
-      "eval_f1_weighted": 0.6018918369596469,
-      "eval_loss": 1.3463146686553955,
-      "eval_macro_fpr": 0.039790684997017586,
-      "eval_macro_sensitivity": 0.3876584042144534,
-      "eval_macro_specificity": 0.972524656182703,
-      "eval_precision": 0.591491914014134,
-      "eval_precision_macro": 0.3194926993208432,
-      "eval_recall": 0.635166537567777,
-      "eval_recall_macro": 0.3876584042144534,
-      "eval_runtime": 53.0494,
-      "eval_samples_per_second": 24.336,
-      "eval_steps_per_second": 3.054,
-      "eval_weighted_fpr": 0.039410927955819595,
-      "eval_weighted_sensitivity": 0.635166537567777,
-      "eval_weighted_specificity": 0.9527033051727678,
+      "eval_accuracy": 0.8086754453911696,
+      "eval_f1_macro": 0.7637730945937946,
+      "eval_f1_micro": 0.8086754453911695,
+      "eval_f1_weighted": 0.8096616104204358,
+      "eval_loss": 1.4511244297027588,
+      "eval_macro_fpr": 0.017099071788154886,
+      "eval_macro_sensitivity": 0.7801835333077355,
+      "eval_macro_specificity": 0.9857069816183608,
+      "eval_precision": 0.8126700320980054,
+      "eval_precision_macro": 0.755164725157429,
+      "eval_recall": 0.8086754453911696,
+      "eval_recall_macro": 0.7801835333077355,
+      "eval_runtime": 62.3649,
+      "eval_samples_per_second": 20.701,
+      "eval_steps_per_second": 2.598,
+      "eval_weighted_fpr": 0.016618448496265894,
+      "eval_weighted_sensitivity": 0.8086754453911696,
+      "eval_weighted_specificity": 0.9769292788842445,
       "step": 8359
     },
     {
       "epoch": 13.22,
-      "grad_norm": 20.480079650878906,
-      "learning_rate": 5.9823742871954386e-06,
-      "loss": 0.818,
+      "grad_norm": 0.02690161019563675,
+      "learning_rate": 5.9720062208398135e-06,
+      "loss": 0.0349,
       "step": 8500
     },
     {
       "epoch": 14.0,
-      "grad_norm": 23.086130142211914,
-      "learning_rate": 3.390357698289269e-06,
-      "loss": 0.8034,
+      "grad_norm": 0.014679946005344391,
+      "learning_rate": 3.379989631933644e-06,
+      "loss": 0.0184,
       "step": 9000
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.6312935708752905,
-      "eval_f1_macro": 0.34083702910591834,
-      "eval_f1_micro": 0.6312935708752905,
-      "eval_f1_weighted": 0.6028724695599472,
-      "eval_loss": 1.3203202486038208,
-      "eval_macro_fpr": 0.04021000066286082,
-      "eval_macro_sensitivity": 0.3887264350085824,
-      "eval_macro_specificity": 0.9722365595844839,
-      "eval_precision": 0.5912742157720344,
-      "eval_precision_macro": 0.3223323142886684,
-      "eval_recall": 0.6312935708752905,
-      "eval_recall_macro": 0.3887264350085824,
-      "eval_runtime": 53.2083,
-      "eval_samples_per_second": 24.263,
-      "eval_steps_per_second": 3.045,
-      "eval_weighted_fpr": 0.04004711425206125,
-      "eval_weighted_sensitivity": 0.6312935708752905,
-      "eval_weighted_specificity": 0.9522548228919672,
+      "eval_accuracy": 0.8140975987606507,
+      "eval_f1_macro": 0.7681382733031227,
+      "eval_f1_micro": 0.8140975987606507,
+      "eval_f1_weighted": 0.8158709627059858,
+      "eval_loss": 1.5005292892456055,
+      "eval_macro_fpr": 0.016486730283890944,
+      "eval_macro_sensitivity": 0.7847943200142441,
+      "eval_macro_specificity": 0.9861151984488189,
+      "eval_precision": 0.8196264811835474,
+      "eval_precision_macro": 0.7613127212329543,
+      "eval_recall": 0.8140975987606507,
+      "eval_recall_macro": 0.7847943200142441,
+      "eval_runtime": 55.9241,
+      "eval_samples_per_second": 23.085,
+      "eval_steps_per_second": 2.897,
+      "eval_weighted_fpr": 0.016049217600641968,
+      "eval_weighted_sensitivity": 0.8140975987606507,
+      "eval_weighted_specificity": 0.9776303779716344,
       "step": 9002
     },
     {
       "epoch": 14.77,
-      "grad_norm": 33.89963912963867,
-      "learning_rate": 7.983411093831002e-07,
-      "loss": 0.7438,
+      "grad_norm": 0.007979592308402061,
+      "learning_rate": 7.931570762052877e-07,
+      "loss": 0.0137,
       "step": 9500
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.6343919442292796,
-      "eval_f1_macro": 0.3416002470875682,
-      "eval_f1_micro": 0.6343919442292796,
-      "eval_f1_weighted": 0.605100970801756,
-      "eval_loss": 1.3312515020370483,
-      "eval_macro_fpr": 0.03982433244262638,
-      "eval_macro_sensitivity": 0.38980632312034796,
-      "eval_macro_specificity": 0.9724569035536236,
-      "eval_precision": 0.5926286717233002,
-      "eval_precision_macro": 0.32309677130337533,
-      "eval_recall": 0.6343919442292796,
-      "eval_recall_macro": 0.38980632312034796,
-      "eval_runtime": 59.3158,
-      "eval_samples_per_second": 21.765,
-      "eval_steps_per_second": 2.731,
-      "eval_weighted_fpr": 0.039537610990115594,
-      "eval_weighted_sensitivity": 0.6343919442292796,
-      "eval_weighted_specificity": 0.9524616090750742,
+      "eval_accuracy": 0.8164213787761425,
+      "eval_f1_macro": 0.7664691580433438,
+      "eval_f1_micro": 0.8164213787761425,
+      "eval_f1_weighted": 0.8173112279420088,
+      "eval_loss": 1.4994738101959229,
+      "eval_macro_fpr": 0.01630779929417336,
+      "eval_macro_sensitivity": 0.777963226880043,
+      "eval_macro_specificity": 0.9862332623102597,
+      "eval_precision": 0.8193184364806528,
+      "eval_precision_macro": 0.7619970590670772,
+      "eval_recall": 0.8164213787761425,
+      "eval_recall_macro": 0.777963226880043,
+      "eval_runtime": 61.5289,
+      "eval_samples_per_second": 20.982,
+      "eval_steps_per_second": 2.633,
+      "eval_weighted_fpr": 0.01580737677582872,
+      "eval_weighted_sensitivity": 0.8164213787761425,
+      "eval_weighted_specificity": 0.9770775558777535,
       "step": 9645
     }
   ],

training_checkpoints/checkpoint-9645/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6da235c4e1e89b590bb6c4dd543edc36c67b04a23baad1858c6c64a331ed4bd1
+oid sha256:beb8f128167a5d05587c32e6e3bff806f690d3ab0996792b1d08193bcb3299f2
 size 4920