diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.12946240735346473,
+  "epoch": 0.1941936110301971,
   "eval_steps": 5,
-  "global_step": 1000,
+  "global_step": 1500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10223,6 +10223,5106 @@
       "eval_samples_per_second": 3.125,
       "eval_steps_per_second": 0.213,
       "step": 1000
+    },
+    {
+      "epoch": 0.1295918697608182,
+      "grad_norm": 2.8120885811909413,
+      "learning_rate": 1.2466845648053966e-06,
+      "loss": 3.8772,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12972133216817167,
+      "grad_norm": 5.756038565932446,
+      "learning_rate": 1.2466554442855443e-06,
+      "loss": 4.4307,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12985079457552512,
+      "grad_norm": 1.8124096125196645,
+      "learning_rate": 1.2466261967799672e-06,
+      "loss": 2.8479,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1299802569828786,
+      "grad_norm": 4.381773001776485,
+      "learning_rate": 1.2465968222946397e-06,
+      "loss": 5.085,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13010971939023205,
+      "grad_norm": 5.382581508115201,
+      "learning_rate": 1.2465673208355619e-06,
+      "loss": 4.7542,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13010971939023205,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0799893140792847,
+      "eval_runtime": 14.2326,
+      "eval_samples_per_second": 3.092,
+      "eval_steps_per_second": 0.211,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13023918179758553,
+      "grad_norm": 4.17922696281196,
+      "learning_rate": 1.2465376924087603e-06,
+      "loss": 4.1602,
+      "step": 1006
+    },
+    {
+      "epoch": 0.130368644204939,
+      "grad_norm": 1.3444658394429028,
+      "learning_rate": 1.246507937020287e-06,
+      "loss": 2.7651,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13049810661229247,
+      "grad_norm": 2.557842837131199,
+      "learning_rate": 1.2464780546762201e-06,
+      "loss": 3.1748,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13062756901964592,
+      "grad_norm": 4.350603012131716,
+      "learning_rate": 1.2464480453826638e-06,
+      "loss": 4.8105,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13075703142699938,
+      "grad_norm": 2.3737512312825664,
+      "learning_rate": 1.2464179091457482e-06,
+      "loss": 3.3989,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13075703142699938,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0841618776321411,
+      "eval_runtime": 13.5446,
+      "eval_samples_per_second": 3.249,
+      "eval_steps_per_second": 0.221,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13088649383435286,
+      "grad_norm": 5.674478451377292,
+      "learning_rate": 1.246387645971629e-06,
+      "loss": 5.355,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1310159562417063,
+      "grad_norm": 4.088708377835137,
+      "learning_rate": 1.246357255866488e-06,
+      "loss": 3.7275,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1311454186490598,
+      "grad_norm": 6.34277906172523,
+      "learning_rate": 1.2463267388365332e-06,
+      "loss": 5.2432,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13127488105641325,
+      "grad_norm": 3.3532623510005397,
+      "learning_rate": 1.2462960948879985e-06,
+      "loss": 3.5669,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1314043434637667,
+      "grad_norm": 4.065983366275227,
+      "learning_rate": 1.246265324027143e-06,
+      "loss": 3.603,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1314043434637667,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.08984375,
+      "eval_runtime": 14.51,
+      "eval_samples_per_second": 3.032,
+      "eval_steps_per_second": 0.207,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13153380587112018,
+      "grad_norm": 4.871446791271936,
+      "learning_rate": 1.2462344262602528e-06,
+      "loss": 4.6147,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13166326827847363,
+      "grad_norm": 6.884465471392841,
+      "learning_rate": 1.2462034015936392e-06,
+      "loss": 5.7163,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13179273068582711,
+      "grad_norm": 3.540900098909635,
+      "learning_rate": 1.2461722500336397e-06,
+      "loss": 3.4534,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13192219309318057,
+      "grad_norm": 6.846871483011824,
+      "learning_rate": 1.2461409715866172e-06,
+      "loss": 5.3828,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13205165550053402,
+      "grad_norm": 2.226206677374046,
+      "learning_rate": 1.2461095662589615e-06,
+      "loss": 3.1299,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13205165550053402,
+      "eval_PRM Accuracy": 0.3113207547169811,
+      "eval_PRM F1": 0.21505376344086022,
+      "eval_PRM F1 AUC": 0.5602409638554217,
+      "eval_PRM F1 Neg": 0.3865546218487395,
+      "eval_PRM NPV": 0.23958333333333334,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.12048192771084337,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0929509401321411,
+      "eval_runtime": 14.5103,
+      "eval_samples_per_second": 3.032,
+      "eval_steps_per_second": 0.207,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1321811179078875,
+      "grad_norm": 4.176354096852999,
+      "learning_rate": 1.2460780340570877e-06,
+      "loss": 4.3921,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13231058031524096,
+      "grad_norm": 4.958554188670066,
+      "learning_rate": 1.2460463749874366e-06,
+      "loss": 5.5342,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13244004272259444,
+      "grad_norm": 3.9563417452721175,
+      "learning_rate": 1.2460145890564754e-06,
+      "loss": 4.0376,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1325695051299479,
+      "grad_norm": 3.0904427146375215,
+      "learning_rate": 1.245982676270697e-06,
+      "loss": 3.6658,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13269896753730134,
+      "grad_norm": 4.385039892457533,
+      "learning_rate": 1.24595063663662e-06,
+      "loss": 4.2363,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13269896753730134,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0927734375,
+      "eval_runtime": 14.4587,
+      "eval_samples_per_second": 3.043,
+      "eval_steps_per_second": 0.207,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13282842994465482,
+      "grad_norm": 1.3935763226393512,
+      "learning_rate": 1.2459184701607896e-06,
+      "loss": 2.5908,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13295789235200828,
+      "grad_norm": 4.042920496148491,
+      "learning_rate": 1.2458861768497763e-06,
+      "loss": 3.9229,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13308735475936176,
+      "grad_norm": 5.486748298031092,
+      "learning_rate": 1.2458537567101765e-06,
+      "loss": 4.395,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1332168171667152,
+      "grad_norm": 6.810921713396431,
+      "learning_rate": 1.2458212097486128e-06,
+      "loss": 5.709,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1333462795740687,
+      "grad_norm": 2.9961899744639338,
+      "learning_rate": 1.2457885359717334e-06,
+      "loss": 3.9429,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1333462795740687,
+      "eval_PRM Accuracy": 0.3113207547169811,
+      "eval_PRM F1": 0.21505376344086022,
+      "eval_PRM F1 AUC": 0.5602409638554217,
+      "eval_PRM F1 Neg": 0.3865546218487395,
+      "eval_PRM NPV": 0.23958333333333334,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.12048192771084337,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.096235752105713,
+      "eval_runtime": 14.9658,
+      "eval_samples_per_second": 2.94,
+      "eval_steps_per_second": 0.2,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13347574198142215,
+      "grad_norm": 3.9360839882171677,
+      "learning_rate": 1.245755735386213e-06,
+      "loss": 4.5273,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1336052043887756,
+      "grad_norm": 3.5663962125428466,
+      "learning_rate": 1.2457228079987512e-06,
+      "loss": 4.2832,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13373466679612908,
+      "grad_norm": 2.482586747992713,
+      "learning_rate": 1.2456897538160745e-06,
+      "loss": 2.9658,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13386412920348253,
+      "grad_norm": 3.8268853657919926,
+      "learning_rate": 1.2456565728449349e-06,
+      "loss": 3.5974,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13399359161083602,
+      "grad_norm": 8.482190043082205,
+      "learning_rate": 1.24562326509211e-06,
+      "loss": 5.9277,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13399359161083602,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0987216234207153,
+      "eval_runtime": 14.1289,
+      "eval_samples_per_second": 3.114,
+      "eval_steps_per_second": 0.212,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13412305401818947,
+      "grad_norm": 2.0206646424945482,
+      "learning_rate": 1.2455898305644038e-06,
+      "loss": 3.0186,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13425251642554292,
+      "grad_norm": 5.020851343704947,
+      "learning_rate": 1.2455562692686459e-06,
+      "loss": 3.8423,
+      "step": 1037
+    },
+    {
+      "epoch": 0.1343819788328964,
+      "grad_norm": 5.132360994585813,
+      "learning_rate": 1.245522581211692e-06,
+      "loss": 4.5366,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13451144124024986,
+      "grad_norm": 1.477783236080554,
+      "learning_rate": 1.2454887664004233e-06,
+      "loss": 3.0269,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13464090364760334,
+      "grad_norm": 2.237540431285223,
+      "learning_rate": 1.2454548248417475e-06,
+      "loss": 3.4336,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13464090364760334,
+      "eval_PRM Accuracy": 0.3113207547169811,
+      "eval_PRM F1": 0.21505376344086022,
+      "eval_PRM F1 AUC": 0.5602409638554217,
+      "eval_PRM F1 Neg": 0.3865546218487395,
+      "eval_PRM NPV": 0.23958333333333334,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.12048192771084337,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1044921875,
+      "eval_runtime": 14.3873,
+      "eval_samples_per_second": 3.058,
+      "eval_steps_per_second": 0.209,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1347703660549568,
+      "grad_norm": 3.8002126953119744,
+      "learning_rate": 1.2454207565425973e-06,
+      "loss": 4.0537,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13489982846231025,
+      "grad_norm": 2.7480266287135913,
+      "learning_rate": 1.2453865615099327e-06,
+      "loss": 3.1523,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13502929086966373,
+      "grad_norm": 2.9818602767874163,
+      "learning_rate": 1.245352239750738e-06,
+      "loss": 3.1567,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13515875327701718,
+      "grad_norm": 1.7889731630388341,
+      "learning_rate": 1.2453177912720244e-06,
+      "loss": 3.0554,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13528821568437066,
+      "grad_norm": 2.137420505975883,
+      "learning_rate": 1.2452832160808286e-06,
+      "loss": 2.8738,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13528821568437066,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1065341234207153,
+      "eval_runtime": 14.6277,
+      "eval_samples_per_second": 3.008,
+      "eval_steps_per_second": 0.205,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13541767809172411,
+      "grad_norm": 1.9691915406197775,
+      "learning_rate": 1.2452485141842135e-06,
+      "loss": 2.9021,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13554714049907757,
+      "grad_norm": 2.7479659357203112,
+      "learning_rate": 1.2452136855892674e-06,
+      "loss": 3.9976,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13567660290643105,
+      "grad_norm": 5.13953469653412,
+      "learning_rate": 1.2451787303031051e-06,
+      "loss": 4.9111,
+      "step": 1048
+    },
+    {
+      "epoch": 0.1358060653137845,
+      "grad_norm": 6.077565305237441,
+      "learning_rate": 1.2451436483328663e-06,
+      "loss": 5.792,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13593552772113798,
+      "grad_norm": 5.968102615606787,
+      "learning_rate": 1.2451084396857178e-06,
+      "loss": 4.2515,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13593552772113798,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.111328125,
+      "eval_runtime": 14.4539,
+      "eval_samples_per_second": 3.044,
+      "eval_steps_per_second": 0.208,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13606499012849144,
+      "grad_norm": 4.935480110101971,
+      "learning_rate": 1.2450731043688515e-06,
+      "loss": 3.7729,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13619445253584492,
+      "grad_norm": 4.213728724481223,
+      "learning_rate": 1.2450376423894853e-06,
+      "loss": 3.9072,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13632391494319837,
+      "grad_norm": 4.412709520527811,
+      "learning_rate": 1.245002053754863e-06,
+      "loss": 4.5903,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13645337735055182,
+      "grad_norm": 4.761015015207559,
+      "learning_rate": 1.2449663384722543e-06,
+      "loss": 5.1465,
+      "step": 1054
+    },
+    {
+      "epoch": 0.1365828397579053,
+      "grad_norm": 3.025138872548357,
+      "learning_rate": 1.244930496548955e-06,
+      "loss": 3.2041,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1365828397579053,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1125710010528564,
+      "eval_runtime": 14.0121,
+      "eval_samples_per_second": 3.14,
+      "eval_steps_per_second": 0.214,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13671230216525876,
+      "grad_norm": 4.3390895925318995,
+      "learning_rate": 1.2448945279922863e-06,
+      "loss": 4.3188,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13684176457261224,
+      "grad_norm": 4.890149310380324,
+      "learning_rate": 1.2448584328095955e-06,
+      "loss": 4.2554,
+      "step": 1057
+    },
+    {
+      "epoch": 0.1369712269799657,
+      "grad_norm": 4.870306366542534,
+      "learning_rate": 1.2448222110082557e-06,
+      "loss": 4.3779,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13710068938731915,
+      "grad_norm": 4.237106036947216,
+      "learning_rate": 1.2447858625956663e-06,
+      "loss": 3.938,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13723015179467263,
+      "grad_norm": 2.137305045729939,
+      "learning_rate": 1.244749387579252e-06,
+      "loss": 3.2622,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13723015179467263,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1148792505264282,
+      "eval_runtime": 13.4891,
+      "eval_samples_per_second": 3.262,
+      "eval_steps_per_second": 0.222,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13735961420202608,
+      "grad_norm": 6.874154844975598,
+      "learning_rate": 1.2447127859664633e-06,
+      "loss": 5.7681,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13748907660937956,
+      "grad_norm": 5.518812046908119,
+      "learning_rate": 1.2446760577647774e-06,
+      "loss": 4.4961,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13761853901673302,
+      "grad_norm": 5.635553577183266,
+      "learning_rate": 1.2446392029816962e-06,
+      "loss": 4.4473,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13774800142408647,
+      "grad_norm": 5.895651136352229,
+      "learning_rate": 1.2446022216247482e-06,
+      "loss": 5.8628,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13787746383143995,
+      "grad_norm": 9.150235423220652,
+      "learning_rate": 1.244565113701488e-06,
+      "loss": 6.271,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13787746383143995,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1170099973678589,
+      "eval_runtime": 14.3522,
+      "eval_samples_per_second": 3.066,
+      "eval_steps_per_second": 0.209,
+      "step": 1065
+    },
+    {
+      "epoch": 0.1380069262387934,
+      "grad_norm": 2.3926213237978335,
+      "learning_rate": 1.2445278792194952e-06,
+      "loss": 3.1616,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13813638864614688,
+      "grad_norm": 3.0041194779372837,
+      "learning_rate": 1.2444905181863756e-06,
+      "loss": 3.6079,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13826585105350034,
+      "grad_norm": 7.102381942612581,
+      "learning_rate": 1.2444530306097612e-06,
+      "loss": 5.105,
+      "step": 1068
+    },
+    {
+      "epoch": 0.1383953134608538,
+      "grad_norm": 3.051103199781432,
+      "learning_rate": 1.2444154164973097e-06,
+      "loss": 3.9746,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13852477586820727,
+      "grad_norm": 4.155816663827047,
+      "learning_rate": 1.2443776758567045e-06,
+      "loss": 4.3833,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13852477586820727,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1204723119735718,
+      "eval_runtime": 16.336,
+      "eval_samples_per_second": 2.693,
+      "eval_steps_per_second": 0.184,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13865423827556073,
+      "grad_norm": 2.909044633792644,
+      "learning_rate": 1.2443398086956545e-06,
+      "loss": 3.7295,
+      "step": 1071
+    },
+    {
+      "epoch": 0.1387837006829142,
+      "grad_norm": 6.810916498117859,
+      "learning_rate": 1.2443018150218954e-06,
+      "loss": 5.7261,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13891316309026766,
+      "grad_norm": 2.5246667893864694,
+      "learning_rate": 1.2442636948431877e-06,
+      "loss": 2.7729,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13904262549762114,
+      "grad_norm": 6.535081404466834,
+      "learning_rate": 1.2442254481673185e-06,
+      "loss": 5.9629,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1391720879049746,
+      "grad_norm": 4.1115567819025065,
+      "learning_rate": 1.2441870750021003e-06,
+      "loss": 4.0261,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1391720879049746,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1260653734207153,
+      "eval_runtime": 15.2691,
+      "eval_samples_per_second": 2.882,
+      "eval_steps_per_second": 0.196,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13930155031232805,
+      "grad_norm": 6.11749785808605,
+      "learning_rate": 1.244148575355372e-06,
+      "loss": 5.7808,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13943101271968153,
+      "grad_norm": 3.844043882823037,
+      "learning_rate": 1.2441099492349973e-06,
+      "loss": 4.0398,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13956047512703498,
+      "grad_norm": 3.999562530348418,
+      "learning_rate": 1.2440711966488666e-06,
+      "loss": 4.2388,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13968993753438846,
+      "grad_norm": 1.525558009055224,
+      "learning_rate": 1.2440323176048964e-06,
+      "loss": 2.6057,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13981939994174192,
+      "grad_norm": 3.6527491474817726,
+      "learning_rate": 1.2439933121110278e-06,
+      "loss": 3.2852,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13981939994174192,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1299716234207153,
+      "eval_runtime": 15.1106,
+      "eval_samples_per_second": 2.912,
+      "eval_steps_per_second": 0.199,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13994886234909537,
+      "grad_norm": 2.7800560748337824,
+      "learning_rate": 1.2439541801752288e-06,
+      "loss": 3.6885,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14007832475644885,
+      "grad_norm": 5.713285844726542,
+      "learning_rate": 1.2439149218054928e-06,
+      "loss": 5.0088,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1402077871638023,
+      "grad_norm": 4.513255233030094,
+      "learning_rate": 1.2438755370098393e-06,
+      "loss": 3.6943,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1403372495711558,
+      "grad_norm": 3.9719742935338926,
+      "learning_rate": 1.2438360257963134e-06,
+      "loss": 3.8955,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14046671197850924,
+      "grad_norm": 4.401220468575003,
+      "learning_rate": 1.2437963881729858e-06,
+      "loss": 4.0269,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14046671197850924,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1299716234207153,
+      "eval_runtime": 15.277,
+      "eval_samples_per_second": 2.88,
+      "eval_steps_per_second": 0.196,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1405961743858627,
+      "grad_norm": 2.7065091243488153,
+      "learning_rate": 1.2437566241479536e-06,
+      "loss": 3.5391,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14072563679321617,
+      "grad_norm": 5.291122797587988,
+      "learning_rate": 1.2437167337293393e-06,
+      "loss": 4.2891,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14085509920056963,
+      "grad_norm": 4.4398178392708285,
+      "learning_rate": 1.2436767169252915e-06,
+      "loss": 4.2461,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1409845616079231,
+      "grad_norm": 4.159299303691462,
+      "learning_rate": 1.243636573743984e-06,
+      "loss": 4.019,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14111402401527656,
+      "grad_norm": 5.255831610541715,
+      "learning_rate": 1.2435963041936173e-06,
+      "loss": 3.8638,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14111402401527656,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1344993114471436,
+      "eval_runtime": 20.5228,
+      "eval_samples_per_second": 2.144,
+      "eval_steps_per_second": 0.146,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14124348642263002,
+      "grad_norm": 4.271540775688631,
+      "learning_rate": 1.2435559082824169e-06,
+      "loss": 3.6372,
+      "step": 1091
+    },
+    {
+      "epoch": 0.1413729488299835,
+      "grad_norm": 3.4427712967584525,
+      "learning_rate": 1.2435153860186347e-06,
+      "loss": 3.7021,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14150241123733695,
+      "grad_norm": 5.287523391353101,
+      "learning_rate": 1.2434747374105482e-06,
+      "loss": 3.95,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14163187364469043,
+      "grad_norm": 4.709416318166009,
+      "learning_rate": 1.2434339624664608e-06,
+      "loss": 4.4761,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14176133605204388,
+      "grad_norm": 3.3367113935601274,
+      "learning_rate": 1.2433930611947015e-06,
+      "loss": 4.1387,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14176133605204388,
+      "eval_PRM Accuracy": 0.29245283018867924,
+      "eval_PRM F1": 0.17582417582417584,
+      "eval_PRM F1 AUC": 0.5481927710843374,
+      "eval_PRM F1 Neg": 0.38016528925619836,
+      "eval_PRM NPV": 0.23469387755102042,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.0963855421686747,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1363636255264282,
+      "eval_runtime": 14.3226,
+      "eval_samples_per_second": 3.072,
+      "eval_steps_per_second": 0.209,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14189079845939737,
+      "grad_norm": 2.6404084900379385,
+      "learning_rate": 1.243352033603625e-06,
+      "loss": 3.2983,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14202026086675082,
+      "grad_norm": 3.192939910029371,
+      "learning_rate": 1.2433108797016124e-06,
+      "loss": 3.3809,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14214972327410427,
+      "grad_norm": 3.7648727238369535,
+      "learning_rate": 1.2432695994970698e-06,
+      "loss": 3.9678,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14227918568145775,
+      "grad_norm": 6.6314968524971585,
+      "learning_rate": 1.24322819299843e-06,
+      "loss": 5.1104,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1424086480888112,
+      "grad_norm": 4.524153810847838,
+      "learning_rate": 1.2431866602141506e-06,
+      "loss": 4.5664,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1424086480888112,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1424893140792847,
+      "eval_runtime": 16.1796,
+      "eval_samples_per_second": 2.719,
+      "eval_steps_per_second": 0.185,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1425381104961647,
+      "grad_norm": 2.771733681929019,
+      "learning_rate": 1.243145001152716e-06,
+      "loss": 3.4568,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14266757290351814,
+      "grad_norm": 1.8409856025083404,
+      "learning_rate": 1.2431032158226356e-06,
+      "loss": 2.9897,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1427970353108716,
+      "grad_norm": 6.330061608212643,
+      "learning_rate": 1.243061304232445e-06,
+      "loss": 5.5767,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14292649771822508,
+      "grad_norm": 5.633492517164207,
+      "learning_rate": 1.2430192663907054e-06,
+      "loss": 4.2026,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14305596012557853,
+      "grad_norm": 5.2708392710741885,
+      "learning_rate": 1.2429771023060042e-06,
+      "loss": 4.5283,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14305596012557853,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1439098119735718,
+      "eval_runtime": 14.3628,
+      "eval_samples_per_second": 3.063,
+      "eval_steps_per_second": 0.209,
+      "step": 1105
+    },
+    {
+      "epoch": 0.143185422532932,
+      "grad_norm": 2.700956332923536,
+      "learning_rate": 1.2429348119869536e-06,
+      "loss": 3.0215,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14331488494028546,
+      "grad_norm": 4.310952952859002,
+      "learning_rate": 1.2428923954421932e-06,
+      "loss": 3.9307,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14344434734763892,
+      "grad_norm": 3.311799919455459,
+      "learning_rate": 1.2428498526803866e-06,
+      "loss": 3.7446,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1435738097549924,
+      "grad_norm": 8.659301414845961,
+      "learning_rate": 1.2428071837102243e-06,
+      "loss": 5.3613,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14370327216234585,
+      "grad_norm": 3.2744913060764107,
+      "learning_rate": 1.2427643885404225e-06,
+      "loss": 3.5542,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14370327216234585,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1500355005264282,
+      "eval_runtime": 15.3362,
+      "eval_samples_per_second": 2.869,
+      "eval_steps_per_second": 0.196,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14383273456969933,
+      "grad_norm": 5.068685889283908,
+      "learning_rate": 1.2427214671797228e-06,
+      "loss": 4.2529,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1439621969770528,
+      "grad_norm": 3.4333095693992433,
+      "learning_rate": 1.2426784196368927e-06,
+      "loss": 3.9536,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14409165938440624,
+      "grad_norm": 3.9324156577274665,
+      "learning_rate": 1.242635245920726e-06,
+      "loss": 3.8291,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14422112179175972,
+      "grad_norm": 2.5062848368018926,
+      "learning_rate": 1.242591946040041e-06,
+      "loss": 3.4534,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14435058419911317,
+      "grad_norm": 2.33993888046791,
+      "learning_rate": 1.242548520003683e-06,
+      "loss": 3.0461,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14435058419911317,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.150923252105713,
+      "eval_runtime": 13.6039,
+      "eval_samples_per_second": 3.234,
+      "eval_steps_per_second": 0.221,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14448004660646666,
+      "grad_norm": 5.0594952402452495,
+      "learning_rate": 1.2425049678205228e-06,
+      "loss": 4.5825,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1446095090138201,
+      "grad_norm": 5.093242852258919,
+      "learning_rate": 1.2424612894994566e-06,
+      "loss": 4.644,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1447389714211736,
+      "grad_norm": 5.840839193932989,
+      "learning_rate": 1.2424174850494068e-06,
+      "loss": 5.0425,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14486843382852704,
+      "grad_norm": 6.281788255156286,
+      "learning_rate": 1.2423735544793209e-06,
+      "loss": 5.2183,
+      "step": 1119
+    },
+    {
+      "epoch": 0.1449978962358805,
+      "grad_norm": 2.1010621793288986,
+      "learning_rate": 1.2423294977981732e-06,
+      "loss": 3.2271,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1449978962358805,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.154296875,
+      "eval_runtime": 14.9627,
+      "eval_samples_per_second": 2.941,
+      "eval_steps_per_second": 0.2,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14512735864323398,
+      "grad_norm": 3.7053175155266773,
+      "learning_rate": 1.2422853150149628e-06,
+      "loss": 3.752,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14525682105058743,
+      "grad_norm": 2.5972642614167105,
+      "learning_rate": 1.2422410061387148e-06,
+      "loss": 2.7944,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1453862834579409,
+      "grad_norm": 3.436084182132075,
+      "learning_rate": 1.2421965711784805e-06,
+      "loss": 4.0332,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14551574586529437,
+      "grad_norm": 2.643748949239247,
+      "learning_rate": 1.2421520101433367e-06,
+      "loss": 3.2324,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14564520827264782,
+      "grad_norm": 5.8756578305350935,
+      "learning_rate": 1.2421073230423854e-06,
+      "loss": 5.3511,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14564520827264782,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1598899364471436,
+      "eval_runtime": 14.9721,
+      "eval_samples_per_second": 2.939,
+      "eval_steps_per_second": 0.2,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1457746706800013,
+      "grad_norm": 4.300233658047289,
+      "learning_rate": 1.2420625098847555e-06,
+      "loss": 4.0046,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14590413308735475,
+      "grad_norm": 3.0812323358378713,
+      "learning_rate": 1.2420175706796005e-06,
+      "loss": 3.9351,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14603359549470823,
+      "grad_norm": 3.290230686059332,
+      "learning_rate": 1.2419725054361004e-06,
+      "loss": 4.2256,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1461630579020617,
+      "grad_norm": 3.01406253574201,
+      "learning_rate": 1.2419273141634607e-06,
+      "loss": 3.865,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14629252030941514,
+      "grad_norm": 4.141674327338006,
+      "learning_rate": 1.2418819968709123e-06,
+      "loss": 3.7329,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14629252030941514,
+      "eval_PRM Accuracy": 0.29245283018867924,
+      "eval_PRM F1": 0.17582417582417584,
+      "eval_PRM F1 AUC": 0.5481927710843374,
+      "eval_PRM F1 Neg": 0.38016528925619836,
+      "eval_PRM NPV": 0.23469387755102042,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.0963855421686747,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1622425317764282,
+      "eval_runtime": 15.0433,
+      "eval_samples_per_second": 2.925,
+      "eval_steps_per_second": 0.199,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14642198271676862,
+      "grad_norm": 4.4611758828057555,
+      "learning_rate": 1.2418365535677127e-06,
+      "loss": 4.4844,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14655144512412208,
+      "grad_norm": 3.880257294864339,
+      "learning_rate": 1.241790984263144e-06,
+      "loss": 3.6812,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14668090753147556,
+      "grad_norm": 4.197403417992651,
+      "learning_rate": 1.2417452889665153e-06,
+      "loss": 3.4883,
+      "step": 1133
+    },
+    {
+      "epoch": 0.146810369938829,
+      "grad_norm": 7.548336600024495,
+      "learning_rate": 1.2416994676871601e-06,
+      "loss": 5.3296,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14693983234618246,
+      "grad_norm": 5.314070244950245,
+      "learning_rate": 1.241653520434439e-06,
+      "loss": 4.2246,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14693983234618246,
+      "eval_PRM Accuracy": 0.29245283018867924,
+      "eval_PRM F1": 0.17582417582417584,
+      "eval_PRM F1 AUC": 0.5481927710843374,
+      "eval_PRM F1 Neg": 0.38016528925619836,
+      "eval_PRM NPV": 0.23469387755102042,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.0963855421686747,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.165482997894287,
+      "eval_runtime": 14.2005,
+      "eval_samples_per_second": 3.098,
+      "eval_steps_per_second": 0.211,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14706929475353595,
+      "grad_norm": 4.010158133139364,
+      "learning_rate": 1.2416074472177371e-06,
+      "loss": 3.9932,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1471987571608894,
+      "grad_norm": 3.1664116446901214,
+      "learning_rate": 1.2415612480464663e-06,
+      "loss": 3.7764,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14732821956824288,
+      "grad_norm": 3.7452098581152593,
+      "learning_rate": 1.2415149229300632e-06,
+      "loss": 3.7412,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14745768197559633,
+      "grad_norm": 4.178176679330257,
+      "learning_rate": 1.2414684718779913e-06,
+      "loss": 4.8052,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14758714438294981,
+      "grad_norm": 3.736484115514027,
+      "learning_rate": 1.2414218948997384e-06,
+      "loss": 4.0894,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14758714438294981,
+      "eval_PRM Accuracy": 0.3018867924528302,
+      "eval_PRM F1": 0.1956521739130435,
+      "eval_PRM F1 AUC": 0.5542168674698795,
+      "eval_PRM F1 Neg": 0.38333333333333336,
+      "eval_PRM NPV": 0.23711340206185566,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.10843373493975904,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1712535619735718,
+      "eval_runtime": 15.3806,
+      "eval_samples_per_second": 2.861,
+      "eval_steps_per_second": 0.195,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14771660679030327,
+      "grad_norm": 5.545612288497347,
+      "learning_rate": 1.2413751920048193e-06,
+      "loss": 4.3167,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14784606919765672,
+      "grad_norm": 5.885094342266873,
+      "learning_rate": 1.241328363202774e-06,
+      "loss": 5.2651,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1479755316050102,
+      "grad_norm": 9.48747168940096,
+      "learning_rate": 1.2412814085031678e-06,
+      "loss": 5.1099,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14810499401236366,
+      "grad_norm": 2.7175101031015734,
+      "learning_rate": 1.2412343279155926e-06,
+      "loss": 3.8564,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14823445641971714,
+      "grad_norm": 6.246399926253007,
+      "learning_rate": 1.2411871214496655e-06,
+      "loss": 5.4409,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14823445641971714,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1707209348678589,
+      "eval_runtime": 15.0676,
+      "eval_samples_per_second": 2.92,
+      "eval_steps_per_second": 0.199,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1483639188270706,
+      "grad_norm": 6.462318803632735,
+      "learning_rate": 1.2411397891150293e-06,
+      "loss": 4.7729,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14849338123442404,
+      "grad_norm": 3.3547904268249154,
+      "learning_rate": 1.2410923309213525e-06,
+      "loss": 3.3516,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14862284364177752,
+      "grad_norm": 6.0018818990826865,
+      "learning_rate": 1.2410447468783295e-06,
+      "loss": 4.3984,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14875230604913098,
+      "grad_norm": 3.098226401723943,
+      "learning_rate": 1.2409970369956804e-06,
+      "loss": 3.4839,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14888176845648446,
+      "grad_norm": 6.805595417611291,
+      "learning_rate": 1.2409492012831505e-06,
+      "loss": 5.1567,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14888176845648446,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.17431640625,
+      "eval_runtime": 15.4352,
+      "eval_samples_per_second": 2.851,
+      "eval_steps_per_second": 0.194,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1490112308638379,
+      "grad_norm": 3.0085780082282,
+      "learning_rate": 1.2409012397505118e-06,
+      "loss": 3.1726,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14914069327119137,
+      "grad_norm": 4.219214951166914,
+      "learning_rate": 1.2408531524075613e-06,
+      "loss": 3.6909,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14927015567854485,
+      "grad_norm": 3.131571945617612,
+      "learning_rate": 1.2408049392641215e-06,
+      "loss": 3.3364,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1493996180858983,
+      "grad_norm": 4.558008815484859,
+      "learning_rate": 1.2407566003300408e-06,
+      "loss": 3.6157,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14952908049325178,
+      "grad_norm": 4.057693804613789,
+      "learning_rate": 1.2407081356151937e-06,
+      "loss": 4.2231,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14952908049325178,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1780894994735718,
+      "eval_runtime": 14.8844,
+      "eval_samples_per_second": 2.956,
+      "eval_steps_per_second": 0.202,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14965854290060523,
+      "grad_norm": 9.395021903942308,
+      "learning_rate": 1.2406595451294804e-06,
+      "loss": 5.3145,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1497880053079587,
+      "grad_norm": 4.739429747767751,
+      "learning_rate": 1.240610828882826e-06,
+      "loss": 4.2935,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14991746771531217,
+      "grad_norm": 6.013953177550472,
+      "learning_rate": 1.2405619868851818e-06,
+      "loss": 4.6714,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15004693012266562,
+      "grad_norm": 3.3936756358658933,
+      "learning_rate": 1.240513019146525e-06,
+      "loss": 4.0088,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1501763925300191,
+      "grad_norm": 4.790235402221722,
+      "learning_rate": 1.2404639256768582e-06,
+      "loss": 4.2891,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1501763925300191,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.182661533355713,
+      "eval_runtime": 15.9452,
+      "eval_samples_per_second": 2.759,
+      "eval_steps_per_second": 0.188,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15030585493737256,
+      "grad_norm": 2.7456538713104166,
+      "learning_rate": 1.2404147064862096e-06,
+      "loss": 3.2068,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15043531734472604,
+      "grad_norm": 2.5993777167552086,
+      "learning_rate": 1.2403653615846334e-06,
+      "loss": 3.2368,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1505647797520795,
+      "grad_norm": 2.075738190996234,
+      "learning_rate": 1.240315890982209e-06,
+      "loss": 2.9714,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15069424215943295,
+      "grad_norm": 3.7640632118845887,
+      "learning_rate": 1.2402662946890424e-06,
+      "loss": 4.0061,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15082370456678643,
+      "grad_norm": 4.867948556551821,
+      "learning_rate": 1.2402165727152638e-06,
+      "loss": 4.1699,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15082370456678643,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1837269067764282,
+      "eval_runtime": 14.2197,
+      "eval_samples_per_second": 3.094,
+      "eval_steps_per_second": 0.211,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15095316697413988,
+      "grad_norm": 7.848922824337514,
+      "learning_rate": 1.2401667250710306e-06,
+      "loss": 5.6499,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15108262938149336,
+      "grad_norm": 1.97181000796218,
+      "learning_rate": 1.2401167517665249e-06,
+      "loss": 3.3882,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15121209178884681,
+      "grad_norm": 6.725735942527819,
+      "learning_rate": 1.240066652811955e-06,
+      "loss": 5.21,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15134155419620027,
+      "grad_norm": 3.911916910319721,
+      "learning_rate": 1.2400164282175541e-06,
+      "loss": 3.739,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15147101660355375,
+      "grad_norm": 5.603532945962072,
+      "learning_rate": 1.239966077993582e-06,
+      "loss": 4.5073,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15147101660355375,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1931818723678589,
+      "eval_runtime": 15.3289,
+      "eval_samples_per_second": 2.87,
+      "eval_steps_per_second": 0.196,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1516004790109072,
+      "grad_norm": 3.8454810196359834,
+      "learning_rate": 1.2399156021503236e-06,
+      "loss": 3.4502,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15172994141826068,
+      "grad_norm": 4.184152439986204,
+      "learning_rate": 1.23986500069809e-06,
+      "loss": 4.5081,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15185940382561414,
+      "grad_norm": 5.974191003909128,
+      "learning_rate": 1.2398142736472168e-06,
+      "loss": 5.334,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1519888662329676,
+      "grad_norm": 1.7684190414875718,
+      "learning_rate": 1.2397634210080669e-06,
+      "loss": 3.0549,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15211832864032107,
+      "grad_norm": 3.354370515992573,
+      "learning_rate": 1.2397124427910275e-06,
+      "loss": 3.8457,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15211832864032107,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1973544359207153,
+      "eval_runtime": 14.2302,
+      "eval_samples_per_second": 3.092,
+      "eval_steps_per_second": 0.211,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15224779104767452,
+      "grad_norm": 9.703247303413816,
+      "learning_rate": 1.2396613390065118e-06,
+      "loss": 5.6182,
+      "step": 1176
+    },
+    {
+      "epoch": 0.152377253455028,
+      "grad_norm": 4.2978327684094175,
+      "learning_rate": 1.239610109664959e-06,
+      "loss": 3.9922,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15250671586238146,
+      "grad_norm": 3.3861931761892725,
+      "learning_rate": 1.2395587547768339e-06,
+      "loss": 3.3247,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1526361782697349,
+      "grad_norm": 2.194702128769876,
+      "learning_rate": 1.2395072743526267e-06,
+      "loss": 3.1234,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1527656406770884,
+      "grad_norm": 3.6377293597590983,
+      "learning_rate": 1.239455668402853e-06,
+      "loss": 3.77,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1527656406770884,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.1990412473678589,
+      "eval_runtime": 13.9685,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.215,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15289510308444185,
+      "grad_norm": 2.371487956136997,
+      "learning_rate": 1.2394039369380547e-06,
+      "loss": 3.1914,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15302456549179533,
+      "grad_norm": 2.1579526763975125,
+      "learning_rate": 1.2393520799687989e-06,
+      "loss": 3.189,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15315402789914878,
+      "grad_norm": 5.358489453115939,
+      "learning_rate": 1.2393000975056782e-06,
+      "loss": 4.2988,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15328349030650226,
+      "grad_norm": 6.584407517694455,
+      "learning_rate": 1.2392479895593116e-06,
+      "loss": 4.229,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15341295271385572,
+      "grad_norm": 8.309404189634096,
+      "learning_rate": 1.2391957561403425e-06,
+      "loss": 6.2046,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15341295271385572,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2014826536178589,
+      "eval_runtime": 14.1547,
+      "eval_samples_per_second": 3.109,
+      "eval_steps_per_second": 0.212,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15354241512120917,
+      "grad_norm": 4.74114632109422,
+      "learning_rate": 1.2391433972594414e-06,
+      "loss": 4.1753,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15367187752856265,
+      "grad_norm": 4.390193692733157,
+      "learning_rate": 1.239090912927303e-06,
+      "loss": 4.4556,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1538013399359161,
+      "grad_norm": 2.052791091352851,
+      "learning_rate": 1.2390383031546488e-06,
+      "loss": 2.9971,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15393080234326958,
+      "grad_norm": 2.4078215431835437,
+      "learning_rate": 1.238985567952225e-06,
+      "loss": 3.1335,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15406026475062304,
+      "grad_norm": 6.639587871619227,
+      "learning_rate": 1.2389327073308043e-06,
+      "loss": 5.4448,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15406026475062304,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2037907838821411,
+      "eval_runtime": 14.5114,
+      "eval_samples_per_second": 3.032,
+      "eval_steps_per_second": 0.207,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1541897271579765,
+      "grad_norm": 4.836779897554295,
+      "learning_rate": 1.238879721301184e-06,
+      "loss": 4.4336,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15431918956532997,
+      "grad_norm": 7.268946133144218,
+      "learning_rate": 1.238826609874188e-06,
+      "loss": 6.5786,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15444865197268343,
+      "grad_norm": 2.8229544082602063,
+      "learning_rate": 1.2387733730606651e-06,
+      "loss": 3.4456,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1545781143800369,
+      "grad_norm": 2.8855991056152677,
+      "learning_rate": 1.2387200108714902e-06,
+      "loss": 3.8381,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15470757678739036,
+      "grad_norm": 4.997642536024528,
+      "learning_rate": 1.2386665233175638e-06,
+      "loss": 4.0288,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15470757678739036,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2100497484207153,
+      "eval_runtime": 13.8741,
+      "eval_samples_per_second": 3.171,
+      "eval_steps_per_second": 0.216,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15483703919474381,
+      "grad_norm": 3.9264123046005794,
+      "learning_rate": 1.2386129104098114e-06,
+      "loss": 3.6814,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1549665016020973,
+      "grad_norm": 5.068382546763015,
+      "learning_rate": 1.2385591721591844e-06,
+      "loss": 4.186,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15509596400945075,
+      "grad_norm": 1.4889991711389337,
+      "learning_rate": 1.2385053085766605e-06,
+      "loss": 2.7527,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15522542641680423,
+      "grad_norm": 4.767650595764256,
+      "learning_rate": 1.238451319673242e-06,
+      "loss": 4.3545,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15535488882415768,
+      "grad_norm": 5.081639699506577,
+      "learning_rate": 1.2383972054599575e-06,
+      "loss": 4.8423,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15535488882415768,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2083629369735718,
+      "eval_runtime": 14.4094,
+      "eval_samples_per_second": 3.054,
+      "eval_steps_per_second": 0.208,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15548435123151114,
+      "grad_norm": 3.5174291881211217,
+      "learning_rate": 1.2383429659478608e-06,
+      "loss": 3.3845,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15561381363886462,
+      "grad_norm": 4.005290830675439,
+      "learning_rate": 1.2382886011480312e-06,
+      "loss": 3.6633,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15574327604621807,
+      "grad_norm": 6.014955308875428,
+      "learning_rate": 1.238234111071574e-06,
+      "loss": 4.2661,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15587273845357155,
+      "grad_norm": 1.9366912742329165,
+      "learning_rate": 1.2381794957296201e-06,
+      "loss": 2.7686,
+      "step": 1204
+    },
+    {
+      "epoch": 0.156002200860925,
+      "grad_norm": 8.049063350431043,
+      "learning_rate": 1.2381247551333255e-06,
+      "loss": 5.6768,
+      "step": 1205
+    },
+    {
+      "epoch": 0.156002200860925,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2156871557235718,
+      "eval_runtime": 15.5954,
+      "eval_samples_per_second": 2.821,
+      "eval_steps_per_second": 0.192,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1561316632682785,
+      "grad_norm": 1.8554917158094817,
+      "learning_rate": 1.2380698892938726e-06,
+      "loss": 3.0439,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15626112567563194,
+      "grad_norm": 4.908249250985227,
+      "learning_rate": 1.238014898222468e-06,
+      "loss": 4.4438,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1563905880829854,
+      "grad_norm": 2.756006043001854,
+      "learning_rate": 1.2379597819303454e-06,
+      "loss": 3.3386,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15652005049033887,
+      "grad_norm": 3.6296518464344203,
+      "learning_rate": 1.237904540428763e-06,
+      "loss": 3.5903,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15664951289769233,
+      "grad_norm": 1.3385748688924646,
+      "learning_rate": 1.2378491737290053e-06,
+      "loss": 2.8218,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15664951289769233,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2168856859207153,
+      "eval_runtime": 14.6688,
+      "eval_samples_per_second": 3.0,
+      "eval_steps_per_second": 0.205,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1567789753050458,
+      "grad_norm": 3.23412280352715,
+      "learning_rate": 1.237793681842382e-06,
+      "loss": 3.2202,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15690843771239926,
+      "grad_norm": 2.590075478031581,
+      "learning_rate": 1.2377380647802283e-06,
+      "loss": 3.4526,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15703790011975272,
+      "grad_norm": 3.7604559398392414,
+      "learning_rate": 1.2376823225539056e-06,
+      "loss": 3.9907,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1571673625271062,
+      "grad_norm": 3.145859799107146,
+      "learning_rate": 1.2376264551747998e-06,
+      "loss": 3.7087,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15729682493445965,
+      "grad_norm": 5.218020569732328,
+      "learning_rate": 1.237570462654323e-06,
+      "loss": 3.866,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15729682493445965,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2225230932235718,
+      "eval_runtime": 15.0032,
+      "eval_samples_per_second": 2.933,
+      "eval_steps_per_second": 0.2,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15742628734181313,
+      "grad_norm": 3.7469742939137856,
+      "learning_rate": 1.2375143450039131e-06,
+      "loss": 4.0188,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15755574974916658,
+      "grad_norm": 2.6400573546658612,
+      "learning_rate": 1.2374581022350333e-06,
+      "loss": 3.0256,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15768521215652004,
+      "grad_norm": 4.263157878094292,
+      "learning_rate": 1.237401734359172e-06,
+      "loss": 4.6477,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15781467456387352,
+      "grad_norm": 6.05031999725171,
+      "learning_rate": 1.2373452413878438e-06,
+      "loss": 4.3477,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15794413697122697,
+      "grad_norm": 5.309455621391533,
+      "learning_rate": 1.237288623332588e-06,
+      "loss": 4.7163,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15794413697122697,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2266956567764282,
+      "eval_runtime": 14.5441,
+      "eval_samples_per_second": 3.025,
+      "eval_steps_per_second": 0.206,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15807359937858045,
+      "grad_norm": 3.2215044538193234,
+      "learning_rate": 1.2372318802049706e-06,
+      "loss": 3.3384,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1582030617859339,
+      "grad_norm": 6.635073780420488,
+      "learning_rate": 1.2371750120165821e-06,
+      "loss": 4.4185,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15833252419328736,
+      "grad_norm": 5.25567473429258,
+      "learning_rate": 1.2371180187790394e-06,
+      "loss": 3.8003,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15846198660064084,
+      "grad_norm": 2.0632757968737163,
+      "learning_rate": 1.2370609005039844e-06,
+      "loss": 3.0242,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1585914490079943,
+      "grad_norm": 1.5882276585349184,
+      "learning_rate": 1.2370036572030844e-06,
+      "loss": 2.7036,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1585914490079943,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2283381223678589,
+      "eval_runtime": 14.3375,
+      "eval_samples_per_second": 3.069,
+      "eval_steps_per_second": 0.209,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15872091141534778,
+      "grad_norm": 4.8521324761303894,
+      "learning_rate": 1.2369462888880327e-06,
+      "loss": 3.5801,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15885037382270123,
+      "grad_norm": 3.207378128337535,
+      "learning_rate": 1.2368887955705477e-06,
+      "loss": 3.7197,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1589798362300547,
+      "grad_norm": 6.306439542762054,
+      "learning_rate": 1.236831177262374e-06,
+      "loss": 4.5679,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15910929863740816,
+      "grad_norm": 2.937161580711878,
+      "learning_rate": 1.236773433975281e-06,
+      "loss": 3.0605,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15923876104476162,
+      "grad_norm": 2.516742529686541,
+      "learning_rate": 1.236715565721064e-06,
+      "loss": 3.4512,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15923876104476162,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2337535619735718,
+      "eval_runtime": 14.1828,
+      "eval_samples_per_second": 3.102,
+      "eval_steps_per_second": 0.212,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1593682234521151,
+      "grad_norm": 4.335310410524557,
+      "learning_rate": 1.236657572511544e-06,
+      "loss": 3.855,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15949768585946855,
+      "grad_norm": 3.165587452637354,
+      "learning_rate": 1.236599454358567e-06,
+      "loss": 3.1357,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15962714826682203,
+      "grad_norm": 5.283993449053063,
+      "learning_rate": 1.2365412112740048e-06,
+      "loss": 4.6313,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1597566106741755,
+      "grad_norm": 4.918346096387316,
+      "learning_rate": 1.2364828432697549e-06,
+      "loss": 3.8223,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15988607308152894,
+      "grad_norm": 5.295910560956783,
+      "learning_rate": 1.23642435035774e-06,
+      "loss": 3.6594,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15988607308152894,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2322443723678589,
+      "eval_runtime": 13.8672,
+      "eval_samples_per_second": 3.173,
+      "eval_steps_per_second": 0.216,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16001553548888242,
+      "grad_norm": 5.161114186182553,
+      "learning_rate": 1.2363657325499085e-06,
+      "loss": 4.7429,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16014499789623587,
+      "grad_norm": 2.189717833652239,
+      "learning_rate": 1.2363069898582345e-06,
+      "loss": 3.3198,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16027446030358936,
+      "grad_norm": 5.000926250189146,
+      "learning_rate": 1.236248122294717e-06,
+      "loss": 4.1372,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1604039227109428,
+      "grad_norm": 4.22457709137243,
+      "learning_rate": 1.236189129871381e-06,
+      "loss": 3.9172,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16053338511829626,
+      "grad_norm": 4.48977911398757,
+      "learning_rate": 1.2361300126002774e-06,
+      "loss": 4.179,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16053338511829626,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2400124073028564,
+      "eval_runtime": 14.3539,
+      "eval_samples_per_second": 3.065,
+      "eval_steps_per_second": 0.209,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16066284752564974,
+      "grad_norm": 3.5557076852399754,
+      "learning_rate": 1.2360707704934814e-06,
+      "loss": 3.5537,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1607923099330032,
+      "grad_norm": 4.545641388873724,
+      "learning_rate": 1.2360114035630947e-06,
+      "loss": 3.4026,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16092177234035668,
+      "grad_norm": 2.9613659702626616,
+      "learning_rate": 1.235951911821244e-06,
+      "loss": 3.6074,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16105123474771013,
+      "grad_norm": 4.689840802864478,
+      "learning_rate": 1.2358922952800822e-06,
+      "loss": 4.6143,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16118069715506358,
+      "grad_norm": 3.6422182462756725,
+      "learning_rate": 1.2358325539517867e-06,
+      "loss": 3.7112,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16118069715506358,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2424982786178589,
+      "eval_runtime": 14.6766,
+      "eval_samples_per_second": 2.998,
+      "eval_steps_per_second": 0.204,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16131015956241707,
+      "grad_norm": 5.322698677692095,
+      "learning_rate": 1.235772687848561e-06,
+      "loss": 4.4839,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16143962196977052,
+      "grad_norm": 4.083340946445587,
+      "learning_rate": 1.2357126969826342e-06,
+      "loss": 3.7927,
+      "step": 1247
+    },
+    {
+      "epoch": 0.161569084377124,
+      "grad_norm": 3.7809745288248524,
+      "learning_rate": 1.2356525813662604e-06,
+      "loss": 3.9868,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16169854678447745,
+      "grad_norm": 4.409665189402354,
+      "learning_rate": 1.2355923410117193e-06,
+      "loss": 3.783,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16182800919183093,
+      "grad_norm": 5.564918708338664,
+      "learning_rate": 1.2355319759313167e-06,
+      "loss": 4.75,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16182800919183093,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2487126588821411,
+      "eval_runtime": 14.722,
+      "eval_samples_per_second": 2.989,
+      "eval_steps_per_second": 0.204,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1619574715991844,
+      "grad_norm": 5.442702575649247,
+      "learning_rate": 1.2354714861373828e-06,
+      "loss": 4.0142,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16208693400653784,
+      "grad_norm": 2.6369781872365317,
+      "learning_rate": 1.2354108716422744e-06,
+      "loss": 3.54,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16221639641389132,
+      "grad_norm": 2.9925332747938342,
+      "learning_rate": 1.235350132458373e-06,
+      "loss": 3.5942,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16234585882124478,
+      "grad_norm": 5.268174377931261,
+      "learning_rate": 1.2352892685980856e-06,
+      "loss": 4.5454,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16247532122859826,
+      "grad_norm": 2.2686559422744974,
+      "learning_rate": 1.2352282800738451e-06,
+      "loss": 2.9609,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16247532122859826,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2497336864471436,
+      "eval_runtime": 13.888,
+      "eval_samples_per_second": 3.168,
+      "eval_steps_per_second": 0.216,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1626047836359517,
+      "grad_norm": 5.607461141405671,
+      "learning_rate": 1.2351671668981098e-06,
+      "loss": 4.8423,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16273424604330516,
+      "grad_norm": 5.850364440689246,
+      "learning_rate": 1.2351059290833631e-06,
+      "loss": 3.8564,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16286370845065865,
+      "grad_norm": 5.8228065719225235,
+      "learning_rate": 1.235044566642114e-06,
+      "loss": 4.0498,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1629931708580121,
+      "grad_norm": 3.526841380655274,
+      "learning_rate": 1.2349830795868972e-06,
+      "loss": 3.4312,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16312263326536558,
+      "grad_norm": 2.100369748774491,
+      "learning_rate": 1.2349214679302726e-06,
+      "loss": 3.0347,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16312263326536558,
+      "eval_PRM Accuracy": 0.2830188679245283,
+      "eval_PRM F1": 0.15555555555555556,
+      "eval_PRM F1 AUC": 0.5421686746987951,
+      "eval_PRM F1 Neg": 0.3770491803278688,
+      "eval_PRM NPV": 0.23232323232323232,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.08433734939759036,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.253373622894287,
+      "eval_runtime": 14.4406,
+      "eval_samples_per_second": 3.047,
+      "eval_steps_per_second": 0.208,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16325209567271903,
+      "grad_norm": 2.7409557649454532,
+      "learning_rate": 1.2348597316848257e-06,
+      "loss": 3.1353,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1633815580800725,
+      "grad_norm": 5.024298758837595,
+      "learning_rate": 1.2347978708631671e-06,
+      "loss": 4.2914,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16351102048742597,
+      "grad_norm": 2.2232604215036864,
+      "learning_rate": 1.2347358854779337e-06,
+      "loss": 3.2371,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16364048289477942,
+      "grad_norm": 4.994951838441288,
+      "learning_rate": 1.2346737755417867e-06,
+      "loss": 5.2429,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1637699453021329,
+      "grad_norm": 5.528739361458071,
+      "learning_rate": 1.2346115410674133e-06,
+      "loss": 4.6021,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1637699453021329,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2551935911178589,
+      "eval_runtime": 13.7653,
+      "eval_samples_per_second": 3.196,
+      "eval_steps_per_second": 0.218,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16389940770948636,
+      "grad_norm": 4.479468314724481,
+      "learning_rate": 1.2345491820675268e-06,
+      "loss": 3.6948,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1640288701168398,
+      "grad_norm": 2.2989754140783627,
+      "learning_rate": 1.2344866985548646e-06,
+      "loss": 3.1912,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1641583325241933,
+      "grad_norm": 3.4286256750104673,
+      "learning_rate": 1.2344240905421906e-06,
+      "loss": 3.6475,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16428779493154674,
+      "grad_norm": 4.436999930092937,
+      "learning_rate": 1.2343613580422934e-06,
+      "loss": 3.7156,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16441725733890022,
+      "grad_norm": 3.6373496122515894,
+      "learning_rate": 1.2342985010679879e-06,
+      "loss": 4.1511,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16441725733890022,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.261186122894287,
+      "eval_runtime": 13.9727,
+      "eval_samples_per_second": 3.149,
+      "eval_steps_per_second": 0.215,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16454671974625368,
+      "grad_norm": 6.167071431690786,
+      "learning_rate": 1.2342355196321135e-06,
+      "loss": 4.5303,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16467618215360716,
+      "grad_norm": 5.847337076567583,
+      "learning_rate": 1.2341724137475355e-06,
+      "loss": 4.4424,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1648056445609606,
+      "grad_norm": 6.031013766434898,
+      "learning_rate": 1.2341091834271449e-06,
+      "loss": 4.2832,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16493510696831407,
+      "grad_norm": 3.126763617576407,
+      "learning_rate": 1.2340458286838573e-06,
+      "loss": 3.2061,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16506456937566755,
+      "grad_norm": 3.6081058992460684,
+      "learning_rate": 1.2339823495306145e-06,
+      "loss": 3.8472,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16506456937566755,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.26416015625,
+      "eval_runtime": 15.217,
+      "eval_samples_per_second": 2.892,
+      "eval_steps_per_second": 0.197,
+      "step": 1275
+    },
+    {
+      "epoch": 0.165194031783021,
+      "grad_norm": 5.146830772092357,
+      "learning_rate": 1.2339187459803831e-06,
+      "loss": 4.2148,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16532349419037448,
+      "grad_norm": 3.9509885467502297,
+      "learning_rate": 1.233855018046156e-06,
+      "loss": 3.8145,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16545295659772793,
+      "grad_norm": 3.1281148078265506,
+      "learning_rate": 1.23379116574095e-06,
+      "loss": 3.6353,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1655824190050814,
+      "grad_norm": 2.3341125663017976,
+      "learning_rate": 1.2337271890778093e-06,
+      "loss": 2.8467,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16571188141243487,
+      "grad_norm": 6.860199796848311,
+      "learning_rate": 1.2336630880698015e-06,
+      "loss": 4.2173,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16571188141243487,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2681552171707153,
+      "eval_runtime": 15.3122,
+      "eval_samples_per_second": 2.874,
+      "eval_steps_per_second": 0.196,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16584134381978832,
+      "grad_norm": 2.1309762049815233,
+      "learning_rate": 1.233598862730021e-06,
+      "loss": 2.9844,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1659708062271418,
+      "grad_norm": 1.8571640811545418,
+      "learning_rate": 1.2335345130715873e-06,
+      "loss": 3.1099,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16610026863449526,
+      "grad_norm": 2.737781773506582,
+      "learning_rate": 1.2334700391076445e-06,
+      "loss": 3.0718,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1662297310418487,
+      "grad_norm": 3.7674622944115628,
+      "learning_rate": 1.2334054408513636e-06,
+      "loss": 3.7505,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1663591934492022,
+      "grad_norm": 1.9738861955669518,
+      "learning_rate": 1.233340718315939e-06,
+      "loss": 3.2979,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1663591934492022,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2753018140792847,
+      "eval_runtime": 13.6962,
+      "eval_samples_per_second": 3.213,
+      "eval_steps_per_second": 0.219,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16648865585655565,
+      "grad_norm": 3.836604791934664,
+      "learning_rate": 1.2332758715145926e-06,
+      "loss": 3.3862,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16661811826390913,
+      "grad_norm": 8.380308107723765,
+      "learning_rate": 1.23321090046057e-06,
+      "loss": 5.3901,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16674758067126258,
+      "grad_norm": 4.966178759450729,
+      "learning_rate": 1.2331458051671432e-06,
+      "loss": 4.5332,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16687704307861603,
+      "grad_norm": 1.543012905635976,
+      "learning_rate": 1.233080585647609e-06,
+      "loss": 2.7429,
+      "step": 1289
+    },
+    {
+      "epoch": 0.16700650548596951,
+      "grad_norm": 3.495369542423535,
+      "learning_rate": 1.2330152419152903e-06,
+      "loss": 3.7144,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16700650548596951,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2757457494735718,
+      "eval_runtime": 13.8008,
+      "eval_samples_per_second": 3.188,
+      "eval_steps_per_second": 0.217,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16713596789332297,
+      "grad_norm": 4.883376146174935,
+      "learning_rate": 1.2329497739835343e-06,
+      "loss": 4.0574,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16726543030067645,
+      "grad_norm": 4.421732827938649,
+      "learning_rate": 1.2328841818657145e-06,
+      "loss": 4.2456,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1673948927080299,
+      "grad_norm": 2.4483818735743843,
+      "learning_rate": 1.2328184655752292e-06,
+      "loss": 3.2094,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16752435511538338,
+      "grad_norm": 2.029621505907904,
+      "learning_rate": 1.2327526251255024e-06,
+      "loss": 2.8154,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16765381752273684,
+      "grad_norm": 1.5805262062388568,
+      "learning_rate": 1.2326866605299834e-06,
+      "loss": 2.7754,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16765381752273684,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2783203125,
+      "eval_runtime": 14.3194,
+      "eval_samples_per_second": 3.073,
+      "eval_steps_per_second": 0.21,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1677832799300903,
+      "grad_norm": 3.2089745667494283,
+      "learning_rate": 1.2326205718021466e-06,
+      "loss": 3.4331,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16791274233744377,
+      "grad_norm": 4.429837592989631,
+      "learning_rate": 1.232554358955492e-06,
+      "loss": 3.4827,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16804220474479722,
+      "grad_norm": 5.786722189374458,
+      "learning_rate": 1.2324880220035455e-06,
+      "loss": 3.9502,
+      "step": 1298
+    },
+    {
+      "epoch": 0.1681716671521507,
+      "grad_norm": 4.642693871719533,
+      "learning_rate": 1.2324215609598568e-06,
+      "loss": 4.0615,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16830112955950416,
+      "grad_norm": 4.978951094202258,
+      "learning_rate": 1.2323549758380026e-06,
+      "loss": 4.196,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16830112955950416,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2800959348678589,
+      "eval_runtime": 14.1293,
+      "eval_samples_per_second": 3.114,
+      "eval_steps_per_second": 0.212,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1684305919668576,
+      "grad_norm": 3.0785794770733723,
+      "learning_rate": 1.232288266651584e-06,
+      "loss": 3.0371,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1685600543742111,
+      "grad_norm": 3.8312546894832047,
+      "learning_rate": 1.2322214334142279e-06,
+      "loss": 3.6953,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16868951678156455,
+      "grad_norm": 4.189659697476781,
+      "learning_rate": 1.232154476139586e-06,
+      "loss": 3.6094,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16881897918891803,
+      "grad_norm": 2.557533080396666,
+      "learning_rate": 1.232087394841336e-06,
+      "loss": 3.3228,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16894844159627148,
+      "grad_norm": 2.045420861296134,
+      "learning_rate": 1.2320201895331807e-06,
+      "loss": 2.9675,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16894844159627148,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2811611890792847,
+      "eval_runtime": 14.8844,
+      "eval_samples_per_second": 2.956,
+      "eval_steps_per_second": 0.202,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16907790400362493,
+      "grad_norm": 5.105850138085465,
+      "learning_rate": 1.2319528602288477e-06,
+      "loss": 4.3081,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16920736641097842,
+      "grad_norm": 6.682389823064708,
+      "learning_rate": 1.2318854069420906e-06,
+      "loss": 4.4058,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16933682881833187,
+      "grad_norm": 5.176659050729813,
+      "learning_rate": 1.2318178296866881e-06,
+      "loss": 3.3955,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16946629122568535,
+      "grad_norm": 2.808313562781416,
+      "learning_rate": 1.2317501284764443e-06,
+      "loss": 2.9983,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1695957536330388,
+      "grad_norm": 4.748615457273775,
+      "learning_rate": 1.2316823033251887e-06,
+      "loss": 4.0342,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1695957536330388,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2867542505264282,
+      "eval_runtime": 14.0369,
+      "eval_samples_per_second": 3.135,
+      "eval_steps_per_second": 0.214,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16972521604039228,
+      "grad_norm": 2.7328742941141178,
+      "learning_rate": 1.2316143542467756e-06,
+      "loss": 3.1914,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16985467844774574,
+      "grad_norm": 2.375136785115866,
+      "learning_rate": 1.2315462812550853e-06,
+      "loss": 2.936,
+      "step": 1312
+    },
+    {
+      "epoch": 0.1699841408550992,
+      "grad_norm": 5.309048653227103,
+      "learning_rate": 1.2314780843640225e-06,
+      "loss": 4.2681,
+      "step": 1313
+    },
+    {
+      "epoch": 0.17011360326245267,
+      "grad_norm": 2.631556473990196,
+      "learning_rate": 1.2314097635875185e-06,
+      "loss": 3.1074,
+      "step": 1314
+    },
+    {
+      "epoch": 0.17024306566980613,
+      "grad_norm": 6.010473160130362,
+      "learning_rate": 1.2313413189395289e-06,
+      "loss": 3.7925,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17024306566980613,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2906605005264282,
+      "eval_runtime": 14.1661,
+      "eval_samples_per_second": 3.106,
+      "eval_steps_per_second": 0.212,
+      "step": 1315
+    },
+    {
+      "epoch": 0.1703725280771596,
+      "grad_norm": 1.2910117682548108,
+      "learning_rate": 1.231272750434035e-06,
+      "loss": 2.4407,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17050199048451306,
+      "grad_norm": 4.9717678405248,
+      "learning_rate": 1.2312040580850432e-06,
+      "loss": 4.2549,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17063145289186651,
+      "grad_norm": 3.8740745628307462,
+      "learning_rate": 1.231135241906585e-06,
+      "loss": 4.0288,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17076091529922,
+      "grad_norm": 8.644608909763942,
+      "learning_rate": 1.2310663019127184e-06,
+      "loss": 5.8799,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17089037770657345,
+      "grad_norm": 5.570092690640336,
+      "learning_rate": 1.2309972381175249e-06,
+      "loss": 4.2168,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17089037770657345,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2937678098678589,
+      "eval_runtime": 13.9362,
+      "eval_samples_per_second": 3.157,
+      "eval_steps_per_second": 0.215,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17101984011392693,
+      "grad_norm": 5.064122521810584,
+      "learning_rate": 1.2309280505351127e-06,
+      "loss": 4.7144,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17114930252128038,
+      "grad_norm": 3.9690626796665454,
+      "learning_rate": 1.2308587391796143e-06,
+      "loss": 3.6675,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17127876492863384,
+      "grad_norm": 3.4272165270525634,
+      "learning_rate": 1.2307893040651884e-06,
+      "loss": 3.5312,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17140822733598732,
+      "grad_norm": 1.3454091952728904,
+      "learning_rate": 1.2307197452060182e-06,
+      "loss": 2.7812,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17153768974334077,
+      "grad_norm": 2.851743842237999,
+      "learning_rate": 1.2306500626163129e-06,
+      "loss": 3.2869,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17153768974334077,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.2962092161178589,
+      "eval_runtime": 15.2331,
+      "eval_samples_per_second": 2.888,
+      "eval_steps_per_second": 0.197,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17166715215069425,
+      "grad_norm": 2.327670258714756,
+      "learning_rate": 1.2305802563103062e-06,
+      "loss": 3.3921,
+      "step": 1326
+    },
+    {
+      "epoch": 0.1717966145580477,
+      "grad_norm": 3.668682339811261,
+      "learning_rate": 1.2305103263022577e-06,
+      "loss": 3.8506,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17192607696540116,
+      "grad_norm": 1.4938868264866891,
+      "learning_rate": 1.2304402726064517e-06,
+      "loss": 2.7803,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17205553937275464,
+      "grad_norm": 2.9262951989712733,
+      "learning_rate": 1.2303700952371986e-06,
+      "loss": 3.563,
+      "step": 1329
+    },
+    {
+      "epoch": 0.1721850017801081,
+      "grad_norm": 3.3946300834250827,
+      "learning_rate": 1.230299794208833e-06,
+      "loss": 3.4497,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1721850017801081,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.299272060394287,
+      "eval_runtime": 15.5694,
+      "eval_samples_per_second": 2.826,
+      "eval_steps_per_second": 0.193,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17231446418746157,
+      "grad_norm": 2.20692189896672,
+      "learning_rate": 1.2302293695357157e-06,
+      "loss": 2.844,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17244392659481503,
+      "grad_norm": 4.710959935707902,
+      "learning_rate": 1.2301588212322322e-06,
+      "loss": 4.3613,
+      "step": 1332
+    },
+    {
+      "epoch": 0.1725733890021685,
+      "grad_norm": 4.846085251219363,
+      "learning_rate": 1.2300881493127936e-06,
+      "loss": 3.3613,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17270285140952196,
+      "grad_norm": 2.1744362773088453,
+      "learning_rate": 1.2300173537918357e-06,
+      "loss": 2.8889,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17283231381687542,
+      "grad_norm": 4.429681532794174,
+      "learning_rate": 1.2299464346838202e-06,
+      "loss": 3.9434,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17283231381687542,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3029118776321411,
+      "eval_runtime": 14.9721,
+      "eval_samples_per_second": 2.939,
+      "eval_steps_per_second": 0.2,
+      "step": 1335
+    },
+    {
+      "epoch": 0.1729617762242289,
+      "grad_norm": 3.2691116304344794,
+      "learning_rate": 1.2298753920032337e-06,
+      "loss": 3.6699,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17309123863158235,
+      "grad_norm": 5.346699742618801,
+      "learning_rate": 1.229804225764588e-06,
+      "loss": 3.5596,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17322070103893583,
+      "grad_norm": 4.555852549866342,
+      "learning_rate": 1.2297329359824206e-06,
+      "loss": 3.4106,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17335016344628928,
+      "grad_norm": 4.213042149473917,
+      "learning_rate": 1.2296615226712938e-06,
+      "loss": 4.061,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17347962585364274,
+      "grad_norm": 3.0846887771916194,
+      "learning_rate": 1.2295899858457946e-06,
+      "loss": 3.6709,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17347962585364274,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.302778720855713,
+      "eval_runtime": 15.1779,
+      "eval_samples_per_second": 2.899,
+      "eval_steps_per_second": 0.198,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17360908826099622,
+      "grad_norm": 5.731358875225708,
+      "learning_rate": 1.2295183255205368e-06,
+      "loss": 4.2327,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17373855066834967,
+      "grad_norm": 5.4889352136054494,
+      "learning_rate": 1.2294465417101575e-06,
+      "loss": 4.585,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17386801307570315,
+      "grad_norm": 4.53134594373362,
+      "learning_rate": 1.229374634429321e-06,
+      "loss": 3.7031,
+      "step": 1343
+    },
+    {
+      "epoch": 0.1739974754830566,
+      "grad_norm": 4.257610992869868,
+      "learning_rate": 1.2293026036927146e-06,
+      "loss": 3.3718,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17412693789041006,
+      "grad_norm": 4.8198308965127366,
+      "learning_rate": 1.2292304495150535e-06,
+      "loss": 4.1067,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17412693789041006,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3063298463821411,
+      "eval_runtime": 14.4331,
+      "eval_samples_per_second": 3.049,
+      "eval_steps_per_second": 0.208,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17425640029776354,
+      "grad_norm": 3.5659979881622537,
+      "learning_rate": 1.2291581719110754e-06,
+      "loss": 3.6414,
+      "step": 1346
+    },
+    {
+      "epoch": 0.174385862705117,
+      "grad_norm": 4.515344822694987,
+      "learning_rate": 1.2290857708955451e-06,
+      "loss": 3.8911,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17451532511247048,
+      "grad_norm": 2.1358031247437914,
+      "learning_rate": 1.229013246483252e-06,
+      "loss": 2.9033,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17464478751982393,
+      "grad_norm": 6.95039018439194,
+      "learning_rate": 1.2289405986890102e-06,
+      "loss": 4.8418,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17477424992717738,
+      "grad_norm": 2.5381980477001598,
+      "learning_rate": 1.2288678275276599e-06,
+      "loss": 3.2893,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17477424992717738,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.313432216644287,
+      "eval_runtime": 14.6309,
+      "eval_samples_per_second": 3.007,
+      "eval_steps_per_second": 0.205,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17490371233453086,
+      "grad_norm": 3.5215403014602464,
+      "learning_rate": 1.2287949330140662e-06,
+      "loss": 3.416,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17503317474188432,
+      "grad_norm": 2.2585148190601405,
+      "learning_rate": 1.2287219151631186e-06,
+      "loss": 3.1743,
+      "step": 1352
+    },
+    {
+      "epoch": 0.1751626371492378,
+      "grad_norm": 4.376247385684798,
+      "learning_rate": 1.2286487739897334e-06,
+      "loss": 4.624,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17529209955659125,
+      "grad_norm": 1.882267898193611,
+      "learning_rate": 1.2285755095088506e-06,
+      "loss": 2.9172,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17542156196394473,
+      "grad_norm": 7.536306206174726,
+      "learning_rate": 1.2285021217354359e-06,
+      "loss": 5.127,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17542156196394473,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3166725635528564,
+      "eval_runtime": 14.631,
+      "eval_samples_per_second": 3.007,
+      "eval_steps_per_second": 0.205,
+      "step": 1355
+    },
+    {
+      "epoch": 0.1755510243712982,
+      "grad_norm": 6.011467643089391,
+      "learning_rate": 1.2284286106844805e-06,
+      "loss": 4.1162,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17568048677865164,
+      "grad_norm": 5.668595721896454,
+      "learning_rate": 1.2283549763710005e-06,
+      "loss": 3.8235,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17580994918600512,
+      "grad_norm": 7.105469287931377,
+      "learning_rate": 1.228281218810037e-06,
+      "loss": 4.7415,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17593941159335857,
+      "grad_norm": 3.684870119131621,
+      "learning_rate": 1.2282073380166568e-06,
+      "loss": 3.4263,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17606887400071206,
+      "grad_norm": 6.516145571408506,
+      "learning_rate": 1.2281333340059512e-06,
+      "loss": 4.542,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17606887400071206,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.320268154144287,
+      "eval_runtime": 15.377,
+      "eval_samples_per_second": 2.861,
+      "eval_steps_per_second": 0.195,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1761983364080655,
+      "grad_norm": 4.591769682088052,
+      "learning_rate": 1.2280592067930371e-06,
+      "loss": 3.9644,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17632779881541896,
+      "grad_norm": 3.502751007308499,
+      "learning_rate": 1.2279849563930567e-06,
+      "loss": 3.6277,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17645726122277244,
+      "grad_norm": 3.586476933477245,
+      "learning_rate": 1.227910582821177e-06,
+      "loss": 3.2766,
+      "step": 1363
+    },
+    {
+      "epoch": 0.1765867236301259,
+      "grad_norm": 4.671775247889118,
+      "learning_rate": 1.2278360860925904e-06,
+      "loss": 3.6133,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17671618603747938,
+      "grad_norm": 2.7099668128122816,
+      "learning_rate": 1.2277614662225142e-06,
+      "loss": 3.2119,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17671618603747938,
+      "eval_PRM Accuracy": 0.27358490566037735,
+      "eval_PRM F1": 0.1348314606741573,
+      "eval_PRM F1 AUC": 0.536144578313253,
+      "eval_PRM F1 Neg": 0.37398373983739835,
+      "eval_PRM NPV": 0.23,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.07228915662650602,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.326615810394287,
+      "eval_runtime": 14.0466,
+      "eval_samples_per_second": 3.132,
+      "eval_steps_per_second": 0.214,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17684564844483283,
+      "grad_norm": 1.986955188132743,
+      "learning_rate": 1.227686723226191e-06,
+      "loss": 3.3013,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17697511085218628,
+      "grad_norm": 3.1356637150410545,
+      "learning_rate": 1.227611857118889e-06,
+      "loss": 3.6294,
+      "step": 1367
+    },
+    {
+      "epoch": 0.17710457325953977,
+      "grad_norm": 4.861762151968143,
+      "learning_rate": 1.2275368679159002e-06,
+      "loss": 3.8896,
+      "step": 1368
+    },
+    {
+      "epoch": 0.17723403566689322,
+      "grad_norm": 4.958798865951923,
+      "learning_rate": 1.2274617556325438e-06,
+      "loss": 4.8574,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1773634980742467,
+      "grad_norm": 2.7922007386505765,
+      "learning_rate": 1.2273865202841623e-06,
+      "loss": 2.9448,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1773634980742467,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3270152807235718,
+      "eval_runtime": 13.8583,
+      "eval_samples_per_second": 3.175,
+      "eval_steps_per_second": 0.216,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17749296048160015,
+      "grad_norm": 3.004859369357232,
+      "learning_rate": 1.2273111618861242e-06,
+      "loss": 3.6426,
+      "step": 1371
+    },
+    {
+      "epoch": 0.1776224228889536,
+      "grad_norm": 5.478934915785173,
+      "learning_rate": 1.227235680453823e-06,
+      "loss": 3.8804,
+      "step": 1372
+    },
+    {
+      "epoch": 0.1777518852963071,
+      "grad_norm": 6.952092575527702,
+      "learning_rate": 1.2271600760026771e-06,
+      "loss": 4.3699,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17788134770366054,
+      "grad_norm": 3.567671522731947,
+      "learning_rate": 1.2270843485481307e-06,
+      "loss": 3.3667,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17801081011101402,
+      "grad_norm": 3.4333439439210265,
+      "learning_rate": 1.2270084981056523e-06,
+      "loss": 3.2964,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17801081011101402,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3323420286178589,
+      "eval_runtime": 15.235,
+      "eval_samples_per_second": 2.888,
+      "eval_steps_per_second": 0.197,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17814027251836748,
+      "grad_norm": 2.3634794909945134,
+      "learning_rate": 1.2269325246907362e-06,
+      "loss": 2.8354,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17826973492572096,
+      "grad_norm": 9.23330369497804,
+      "learning_rate": 1.226856428318901e-06,
+      "loss": 6.0547,
+      "step": 1377
+    },
+    {
+      "epoch": 0.1783991973330744,
+      "grad_norm": 7.978881425745575,
+      "learning_rate": 1.2267802090056916e-06,
+      "loss": 4.6367,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17852865974042786,
+      "grad_norm": 6.601964940830656,
+      "learning_rate": 1.226703866766677e-06,
+      "loss": 4.6245,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17865812214778135,
+      "grad_norm": 2.8174317770600537,
+      "learning_rate": 1.2266274016174512e-06,
+      "loss": 3.7234,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17865812214778135,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3362038135528564,
+      "eval_runtime": 14.7693,
+      "eval_samples_per_second": 2.979,
+      "eval_steps_per_second": 0.203,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1787875845551348,
+      "grad_norm": 2.342349869371614,
+      "learning_rate": 1.2265508135736344e-06,
+      "loss": 3.1567,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17891704696248828,
+      "grad_norm": 2.59091194915207,
+      "learning_rate": 1.2264741026508712e-06,
+      "loss": 3.0042,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17904650936984173,
+      "grad_norm": 4.304851786405775,
+      "learning_rate": 1.226397268864831e-06,
+      "loss": 3.7949,
+      "step": 1383
+    },
+    {
+      "epoch": 0.1791759717771952,
+      "grad_norm": 2.7736195632209615,
+      "learning_rate": 1.2263203122312089e-06,
+      "loss": 2.9407,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17930543418454867,
+      "grad_norm": 5.7694747836934495,
+      "learning_rate": 1.2262432327657251e-06,
+      "loss": 4.4531,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17930543418454867,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.341264247894287,
+      "eval_runtime": 14.5408,
+      "eval_samples_per_second": 3.026,
+      "eval_steps_per_second": 0.206,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17943489659190212,
+      "grad_norm": 2.4345289511602974,
+      "learning_rate": 1.226166030484124e-06,
+      "loss": 3.4282,
+      "step": 1386
+    },
+    {
+      "epoch": 0.1795643589992556,
+      "grad_norm": 5.992520859307226,
+      "learning_rate": 1.2260887054021763e-06,
+      "loss": 3.6562,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17969382140660906,
+      "grad_norm": 1.2621959680004389,
+      "learning_rate": 1.226011257535677e-06,
+      "loss": 2.6984,
+      "step": 1388
+    },
+    {
+      "epoch": 0.1798232838139625,
+      "grad_norm": 3.1977805600564166,
+      "learning_rate": 1.2259336869004462e-06,
+      "loss": 3.3037,
+      "step": 1389
+    },
+    {
+      "epoch": 0.179952746221316,
+      "grad_norm": 4.075018442422728,
+      "learning_rate": 1.2258559935123297e-06,
+      "loss": 3.2825,
+      "step": 1390
+    },
+    {
+      "epoch": 0.179952746221316,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.341752529144287,
+      "eval_runtime": 14.6531,
+      "eval_samples_per_second": 3.003,
+      "eval_steps_per_second": 0.205,
+      "step": 1390
+    },
+    {
+      "epoch": 0.18008220862866944,
+      "grad_norm": 3.903661339265306,
+      "learning_rate": 1.2257781773871977e-06,
+      "loss": 3.5542,
+      "step": 1391
+    },
+    {
+      "epoch": 0.18021167103602292,
+      "grad_norm": 3.790227953159591,
+      "learning_rate": 1.2257002385409455e-06,
+      "loss": 3.5815,
+      "step": 1392
+    },
+    {
+      "epoch": 0.18034113344337638,
+      "grad_norm": 5.398416413517164,
+      "learning_rate": 1.2256221769894943e-06,
+      "loss": 3.5898,
+      "step": 1393
+    },
+    {
+      "epoch": 0.18047059585072983,
+      "grad_norm": 4.4279447629335005,
+      "learning_rate": 1.225543992748789e-06,
+      "loss": 3.9104,
+      "step": 1394
+    },
+    {
+      "epoch": 0.1806000582580833,
+      "grad_norm": 1.9659534356002728,
+      "learning_rate": 1.2254656858348013e-06,
+      "loss": 2.8894,
+      "step": 1395
+    },
+    {
+      "epoch": 0.1806000582580833,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3474787473678589,
+      "eval_runtime": 14.6952,
+      "eval_samples_per_second": 2.994,
+      "eval_steps_per_second": 0.204,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18072952066543677,
+      "grad_norm": 3.3695500855419356,
+      "learning_rate": 1.2253872562635258e-06,
+      "loss": 3.0664,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18085898307279025,
+      "grad_norm": 2.923048469363371,
+      "learning_rate": 1.2253087040509841e-06,
+      "loss": 3.0024,
+      "step": 1397
+    },
+    {
+      "epoch": 0.1809884454801437,
+      "grad_norm": 2.630209307642065,
+      "learning_rate": 1.2252300292132222e-06,
+      "loss": 3.0532,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18111790788749718,
+      "grad_norm": 5.333966673548329,
+      "learning_rate": 1.2251512317663102e-06,
+      "loss": 4.4951,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18124737029485063,
+      "grad_norm": 4.768577762469869,
+      "learning_rate": 1.225072311726345e-06,
+      "loss": 4.1709,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18124737029485063,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.351518154144287,
+      "eval_runtime": 14.5067,
+      "eval_samples_per_second": 3.033,
+      "eval_steps_per_second": 0.207,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1813768327022041,
+      "grad_norm": 2.873356644696701,
+      "learning_rate": 1.2249932691094472e-06,
+      "loss": 3.0024,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18150629510955757,
+      "grad_norm": 1.4921759905025873,
+      "learning_rate": 1.2249141039317626e-06,
+      "loss": 2.5623,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18163575751691102,
+      "grad_norm": 6.404901401988275,
+      "learning_rate": 1.2248348162094628e-06,
+      "loss": 4.8521,
+      "step": 1403
+    },
+    {
+      "epoch": 0.1817652199242645,
+      "grad_norm": 5.340069187711158,
+      "learning_rate": 1.2247554059587435e-06,
+      "loss": 3.8372,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18189468233161796,
+      "grad_norm": 2.818863663253772,
+      "learning_rate": 1.2246758731958261e-06,
+      "loss": 3.5308,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18189468233161796,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3575993776321411,
+      "eval_runtime": 14.1089,
+      "eval_samples_per_second": 3.119,
+      "eval_steps_per_second": 0.213,
+      "step": 1405
+    },
+    {
+      "epoch": 0.1820241447389714,
+      "grad_norm": 6.504035357386434,
+      "learning_rate": 1.224596217936957e-06,
+      "loss": 3.6816,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1821536071463249,
+      "grad_norm": 4.226594780373103,
+      "learning_rate": 1.2245164401984067e-06,
+      "loss": 3.3032,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18228306955367835,
+      "grad_norm": 6.523869931802508,
+      "learning_rate": 1.2244365399964722e-06,
+      "loss": 4.1499,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18241253196103183,
+      "grad_norm": 2.2729241748769664,
+      "learning_rate": 1.224356517347474e-06,
+      "loss": 3.054,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18254199436838528,
+      "grad_norm": 2.232002427245416,
+      "learning_rate": 1.224276372267759e-06,
+      "loss": 2.9849,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18254199436838528,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3580876588821411,
+      "eval_runtime": 14.3326,
+      "eval_samples_per_second": 3.07,
+      "eval_steps_per_second": 0.209,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18267145677573873,
+      "grad_norm": 1.8013043044712627,
+      "learning_rate": 1.2241961047736982e-06,
+      "loss": 3.1763,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18280091918309221,
+      "grad_norm": 5.0167684023817225,
+      "learning_rate": 1.224115714881688e-06,
+      "loss": 4.2461,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18293038159044567,
+      "grad_norm": 5.604938238406695,
+      "learning_rate": 1.2240352026081492e-06,
+      "loss": 4.6792,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18305984399779915,
+      "grad_norm": 2.3286796253397615,
+      "learning_rate": 1.2239545679695284e-06,
+      "loss": 3.5737,
+      "step": 1414
+    },
+    {
+      "epoch": 0.1831893064051526,
+      "grad_norm": 3.3881593882231855,
+      "learning_rate": 1.2238738109822968e-06,
+      "loss": 3.2437,
+      "step": 1415
+    },
+    {
+      "epoch": 0.1831893064051526,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.357865810394287,
+      "eval_runtime": 15.073,
+      "eval_samples_per_second": 2.919,
+      "eval_steps_per_second": 0.199,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18331876881250606,
+      "grad_norm": 2.8725253858671307,
+      "learning_rate": 1.223792931662951e-06,
+      "loss": 3.8125,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18344823121985954,
+      "grad_norm": 4.2082521312796075,
+      "learning_rate": 1.2237119300280117e-06,
+      "loss": 3.8577,
+      "step": 1417
+    },
+    {
+      "epoch": 0.183577693627213,
+      "grad_norm": 3.664366756612207,
+      "learning_rate": 1.2236308060940253e-06,
+      "loss": 3.8291,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18370715603456647,
+      "grad_norm": 2.8286714352794804,
+      "learning_rate": 1.223549559877563e-06,
+      "loss": 3.3262,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18383661844191992,
+      "grad_norm": 1.8582336316662238,
+      "learning_rate": 1.2234681913952213e-06,
+      "loss": 2.6143,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18383661844191992,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3638583421707153,
+      "eval_runtime": 14.2613,
+      "eval_samples_per_second": 3.085,
+      "eval_steps_per_second": 0.21,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1839660808492734,
+      "grad_norm": 4.6996909869227075,
+      "learning_rate": 1.223386700663621e-06,
+      "loss": 3.7173,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18409554325662686,
+      "grad_norm": 4.496564894942089,
+      "learning_rate": 1.2233050876994086e-06,
+      "loss": 3.564,
+      "step": 1422
+    },
+    {
+      "epoch": 0.1842250056639803,
+      "grad_norm": 1.581812736719712,
+      "learning_rate": 1.2232233525192545e-06,
+      "loss": 2.5459,
+      "step": 1423
+    },
+    {
+      "epoch": 0.1843544680713338,
+      "grad_norm": 6.236373517235444,
+      "learning_rate": 1.2231414951398554e-06,
+      "loss": 3.665,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18448393047868725,
+      "grad_norm": 6.203911927277913,
+      "learning_rate": 1.223059515577932e-06,
+      "loss": 4.0923,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18448393047868725,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3698064088821411,
+      "eval_runtime": 14.1682,
+      "eval_samples_per_second": 3.106,
+      "eval_steps_per_second": 0.212,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18461339288604073,
+      "grad_norm": 3.4833382453688655,
+      "learning_rate": 1.2229774138502306e-06,
+      "loss": 3.0735,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18474285529339418,
+      "grad_norm": 4.347187809491616,
+      "learning_rate": 1.222895189973522e-06,
+      "loss": 3.8179,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18487231770074763,
+      "grad_norm": 3.476425995636057,
+      "learning_rate": 1.222812843964602e-06,
+      "loss": 3.5542,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18500178010810112,
+      "grad_norm": 4.092272962578211,
+      "learning_rate": 1.2227303758402918e-06,
+      "loss": 4.2007,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18513124251545457,
+      "grad_norm": 6.370381941248504,
+      "learning_rate": 1.2226477856174366e-06,
+      "loss": 4.5801,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18513124251545457,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.369140625,
+      "eval_runtime": 14.0312,
+      "eval_samples_per_second": 3.136,
+      "eval_steps_per_second": 0.214,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18526070492280805,
+      "grad_norm": 3.5251596519985045,
+      "learning_rate": 1.2225650733129073e-06,
+      "loss": 3.3674,
+      "step": 1431
+    },
+    {
+      "epoch": 0.1853901673301615,
+      "grad_norm": 1.2361749130373705,
+      "learning_rate": 1.2224822389435997e-06,
+      "loss": 2.4565,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18551962973751496,
+      "grad_norm": 2.81013731681298,
+      "learning_rate": 1.2223992825264347e-06,
+      "loss": 3.1934,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18564909214486844,
+      "grad_norm": 3.483636615304478,
+      "learning_rate": 1.2223162040783572e-06,
+      "loss": 3.1353,
+      "step": 1434
+    },
+    {
+      "epoch": 0.1857785545522219,
+      "grad_norm": 6.99999596571565,
+      "learning_rate": 1.2222330036163381e-06,
+      "loss": 4.8738,
+      "step": 1435
+    },
+    {
+      "epoch": 0.1857785545522219,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3704723119735718,
+      "eval_runtime": 14.1881,
+      "eval_samples_per_second": 3.101,
+      "eval_steps_per_second": 0.211,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18590801695957537,
+      "grad_norm": 2.3749684960020843,
+      "learning_rate": 1.2221496811573726e-06,
+      "loss": 3.2864,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18603747936692883,
+      "grad_norm": 4.7779905519446935,
+      "learning_rate": 1.222066236718481e-06,
+      "loss": 3.647,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18616694177428228,
+      "grad_norm": 7.29843151157389,
+      "learning_rate": 1.2219826703167087e-06,
+      "loss": 3.9585,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18629640418163576,
+      "grad_norm": 2.57725108373773,
+      "learning_rate": 1.2218989819691257e-06,
+      "loss": 3.1404,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18642586658898921,
+      "grad_norm": 4.434566243366233,
+      "learning_rate": 1.221815171692827e-06,
+      "loss": 4.1277,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18642586658898921,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3776633739471436,
+      "eval_runtime": 14.7808,
+      "eval_samples_per_second": 2.977,
+      "eval_steps_per_second": 0.203,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1865553289963427,
+      "grad_norm": 3.5594124522576163,
+      "learning_rate": 1.2217312395049324e-06,
+      "loss": 3.4541,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18668479140369615,
+      "grad_norm": 6.615842610368984,
+      "learning_rate": 1.2216471854225873e-06,
+      "loss": 4.4895,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18681425381104963,
+      "grad_norm": 3.457320657983824,
+      "learning_rate": 1.221563009462961e-06,
+      "loss": 3.5063,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18694371621840308,
+      "grad_norm": 4.806707468572244,
+      "learning_rate": 1.221478711643248e-06,
+      "loss": 3.5095,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18707317862575654,
+      "grad_norm": 4.667653187215275,
+      "learning_rate": 1.221394291980668e-06,
+      "loss": 3.3647,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18707317862575654,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3777521848678589,
+      "eval_runtime": 14.2176,
+      "eval_samples_per_second": 3.095,
+      "eval_steps_per_second": 0.211,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18720264103311002,
+      "grad_norm": 2.3540701328498144,
+      "learning_rate": 1.2213097504924659e-06,
+      "loss": 3.0569,
+      "step": 1446
+    },
+    {
+      "epoch": 0.18733210344046347,
+      "grad_norm": 3.9234725048015746,
+      "learning_rate": 1.2212250871959106e-06,
+      "loss": 3.4436,
+      "step": 1447
+    },
+    {
+      "epoch": 0.18746156584781695,
+      "grad_norm": 6.268158860585557,
+      "learning_rate": 1.221140302108296e-06,
+      "loss": 4.4556,
+      "step": 1448
+    },
+    {
+      "epoch": 0.1875910282551704,
+      "grad_norm": 5.735086994523188,
+      "learning_rate": 1.2210553952469418e-06,
+      "loss": 3.9031,
+      "step": 1449
+    },
+    {
+      "epoch": 0.18772049066252386,
+      "grad_norm": 3.566209530027245,
+      "learning_rate": 1.2209703666291912e-06,
+      "loss": 3.6167,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18772049066252386,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3832119703292847,
+      "eval_runtime": 14.1238,
+      "eval_samples_per_second": 3.115,
+      "eval_steps_per_second": 0.212,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18784995306987734,
+      "grad_norm": 2.7759442862299575,
+      "learning_rate": 1.2208852162724138e-06,
+      "loss": 3.4282,
+      "step": 1451
+    },
+    {
+      "epoch": 0.1879794154772308,
+      "grad_norm": 6.03578138417162,
+      "learning_rate": 1.2207999441940028e-06,
+      "loss": 5.1118,
+      "step": 1452
+    },
+    {
+      "epoch": 0.18810887788458427,
+      "grad_norm": 5.045494044161035,
+      "learning_rate": 1.2207145504113769e-06,
+      "loss": 4.0557,
+      "step": 1453
+    },
+    {
+      "epoch": 0.18823834029193773,
+      "grad_norm": 4.701816933182676,
+      "learning_rate": 1.2206290349419795e-06,
+      "loss": 3.6936,
+      "step": 1454
+    },
+    {
+      "epoch": 0.18836780269929118,
+      "grad_norm": 3.737544974401831,
+      "learning_rate": 1.220543397803279e-06,
+      "loss": 3.4189,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18836780269929118,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.388716220855713,
+      "eval_runtime": 15.1962,
+      "eval_samples_per_second": 2.895,
+      "eval_steps_per_second": 0.197,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18849726510664466,
+      "grad_norm": 2.98202953594671,
+      "learning_rate": 1.2204576390127681e-06,
+      "loss": 3.0361,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18862672751399812,
+      "grad_norm": 4.190718268486176,
+      "learning_rate": 1.2203717585879652e-06,
+      "loss": 3.7644,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1887561899213516,
+      "grad_norm": 4.759119772432647,
+      "learning_rate": 1.220285756546413e-06,
+      "loss": 3.416,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18888565232870505,
+      "grad_norm": 9.405091538903855,
+      "learning_rate": 1.2201996329056793e-06,
+      "loss": 5.927,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1890151147360585,
+      "grad_norm": 4.871610230477146,
+      "learning_rate": 1.220113387683356e-06,
+      "loss": 4.231,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1890151147360585,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.391157627105713,
+      "eval_runtime": 14.3757,
+      "eval_samples_per_second": 3.061,
+      "eval_steps_per_second": 0.209,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18914457714341198,
+      "grad_norm": 3.093044372326429,
+      "learning_rate": 1.2200270208970614e-06,
+      "loss": 2.948,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18927403955076544,
+      "grad_norm": 9.632522186253102,
+      "learning_rate": 1.2199405325644367e-06,
+      "loss": 6.4829,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18940350195811892,
+      "grad_norm": 2.4259934682234907,
+      "learning_rate": 1.2198539227031495e-06,
+      "loss": 3.144,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18953296436547237,
+      "grad_norm": 2.0793635957219996,
+      "learning_rate": 1.2197671913308914e-06,
+      "loss": 2.9604,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18966242677282585,
+      "grad_norm": 3.162156193891948,
+      "learning_rate": 1.2196803384653792e-06,
+      "loss": 3.2866,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18966242677282585,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.3975497484207153,
+      "eval_runtime": 15.0403,
+      "eval_samples_per_second": 2.925,
+      "eval_steps_per_second": 0.199,
+      "step": 1465
+    },
+    {
+      "epoch": 0.1897918891801793,
+      "grad_norm": 3.3374046567024465,
+      "learning_rate": 1.219593364124354e-06,
+      "loss": 3.667,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18992135158753276,
+      "grad_norm": 1.8849838522909952,
+      "learning_rate": 1.2195062683255825e-06,
+      "loss": 2.9072,
+      "step": 1467
+    },
+    {
+      "epoch": 0.19005081399488624,
+      "grad_norm": 1.5145599328612633,
+      "learning_rate": 1.2194190510868555e-06,
+      "loss": 2.8418,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1901802764022397,
+      "grad_norm": 4.268043726959668,
+      "learning_rate": 1.2193317124259894e-06,
+      "loss": 3.79,
+      "step": 1469
+    },
+    {
+      "epoch": 0.19030973880959318,
+      "grad_norm": 5.7488014538831225,
+      "learning_rate": 1.2192442523608239e-06,
+      "loss": 4.5181,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19030973880959318,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.400390625,
+      "eval_runtime": 14.9333,
+      "eval_samples_per_second": 2.946,
+      "eval_steps_per_second": 0.201,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19043920121694663,
+      "grad_norm": 3.2764681327677074,
+      "learning_rate": 1.2191566709092253e-06,
+      "loss": 3.1045,
+      "step": 1471
+    },
+    {
+      "epoch": 0.19056866362430008,
+      "grad_norm": 4.4796397902105625,
+      "learning_rate": 1.2190689680890837e-06,
+      "loss": 3.4404,
+      "step": 1472
+    },
+    {
+      "epoch": 0.19069812603165356,
+      "grad_norm": 4.800960715818861,
+      "learning_rate": 1.2189811439183141e-06,
+      "loss": 3.9375,
+      "step": 1473
+    },
+    {
+      "epoch": 0.19082758843900702,
+      "grad_norm": 3.6856454316834553,
+      "learning_rate": 1.2188931984148566e-06,
+      "loss": 2.9868,
+      "step": 1474
+    },
+    {
+      "epoch": 0.1909570508463605,
+      "grad_norm": 4.624893420071821,
+      "learning_rate": 1.2188051315966754e-06,
+      "loss": 3.5354,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1909570508463605,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4031428098678589,
+      "eval_runtime": 16.5659,
+      "eval_samples_per_second": 2.656,
+      "eval_steps_per_second": 0.181,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19108651325371395,
+      "grad_norm": 2.4215697306628003,
+      "learning_rate": 1.2187169434817604e-06,
+      "loss": 3.3665,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1912159756610674,
+      "grad_norm": 3.09264095899073,
+      "learning_rate": 1.2186286340881253e-06,
+      "loss": 3.5483,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1913454380684209,
+      "grad_norm": 4.027535046427222,
+      "learning_rate": 1.2185402034338095e-06,
+      "loss": 3.4644,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19147490047577434,
+      "grad_norm": 3.351719821789138,
+      "learning_rate": 1.218451651536877e-06,
+      "loss": 3.1255,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19160436288312782,
+      "grad_norm": 5.150281941211811,
+      "learning_rate": 1.2183629784154155e-06,
+      "loss": 3.9524,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19160436288312782,
+      "eval_PRM Accuracy": 0.25471698113207547,
+      "eval_PRM F1": 0.09195402298850575,
+      "eval_PRM F1 AUC": 0.5240963855421686,
+      "eval_PRM F1 Neg": 0.368,
+      "eval_PRM NPV": 0.22549019607843138,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.04819277108433735,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4035866260528564,
+      "eval_runtime": 18.7342,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 0.16,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19173382529048127,
+      "grad_norm": 2.517345404719067,
+      "learning_rate": 1.218274184087539e-06,
+      "loss": 3.2493,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19186328769783473,
+      "grad_norm": 3.244307149627311,
+      "learning_rate": 1.218185268571385e-06,
+      "loss": 3.4783,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1919927501051882,
+      "grad_norm": 2.058493891977264,
+      "learning_rate": 1.2180962318851168e-06,
+      "loss": 2.584,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19212221251254166,
+      "grad_norm": 3.7502324472681434,
+      "learning_rate": 1.2180070740469217e-06,
+      "loss": 3.4126,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19225167491989514,
+      "grad_norm": 3.784012039523587,
+      "learning_rate": 1.2179177950750118e-06,
+      "loss": 3.4277,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19225167491989514,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4086025953292847,
+      "eval_runtime": 14.3657,
+      "eval_samples_per_second": 3.063,
+      "eval_steps_per_second": 0.209,
+      "step": 1485
+    },
+    {
+      "epoch": 0.1923811373272486,
+      "grad_norm": 3.2291766620010343,
+      "learning_rate": 1.2178283949876246e-06,
+      "loss": 3.4443,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19251059973460208,
+      "grad_norm": 4.61459480144291,
+      "learning_rate": 1.2177388738030216e-06,
+      "loss": 3.6772,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19264006214195553,
+      "grad_norm": 3.5463107007493044,
+      "learning_rate": 1.2176492315394894e-06,
+      "loss": 3.6357,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19276952454930898,
+      "grad_norm": 4.270266157919265,
+      "learning_rate": 1.217559468215339e-06,
+      "loss": 3.9058,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19289898695666247,
+      "grad_norm": 7.575439693616707,
+      "learning_rate": 1.217469583848907e-06,
+      "loss": 4.4536,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19289898695666247,
+      "eval_PRM Accuracy": 0.2641509433962264,
+      "eval_PRM F1": 0.11363636363636363,
+      "eval_PRM F1 AUC": 0.5301204819277109,
+      "eval_PRM F1 Neg": 0.3709677419354839,
+      "eval_PRM NPV": 0.22772277227722773,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.060240963855421686,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4122425317764282,
+      "eval_runtime": 16.104,
+      "eval_samples_per_second": 2.732,
+      "eval_steps_per_second": 0.186,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19302844936401592,
+      "grad_norm": 2.972792525981474,
+      "learning_rate": 1.2173795784585534e-06,
+      "loss": 3.1672,
+      "step": 1491
+    },
+    {
+      "epoch": 0.1931579117713694,
+      "grad_norm": 2.3418243128705636,
+      "learning_rate": 1.217289452062664e-06,
+      "loss": 3.063,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19328737417872285,
+      "grad_norm": 5.567828310167252,
+      "learning_rate": 1.217199204679649e-06,
+      "loss": 4.2148,
+      "step": 1493
+    },
+    {
+      "epoch": 0.1934168365860763,
+      "grad_norm": 4.261413175583276,
+      "learning_rate": 1.2171088363279426e-06,
+      "loss": 3.416,
+      "step": 1494
+    },
+    {
+      "epoch": 0.1935462989934298,
+      "grad_norm": 3.3535930303205053,
+      "learning_rate": 1.2170183470260053e-06,
+      "loss": 3.0527,
+      "step": 1495
+    },
+    {
+      "epoch": 0.1935462989934298,
+      "eval_PRM Accuracy": 0.25471698113207547,
+      "eval_PRM F1": 0.09195402298850575,
+      "eval_PRM F1 AUC": 0.5240963855421686,
+      "eval_PRM F1 Neg": 0.368,
+      "eval_PRM NPV": 0.22549019607843138,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.04819277108433735,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.415971279144287,
+      "eval_runtime": 15.1716,
+      "eval_samples_per_second": 2.9,
+      "eval_steps_per_second": 0.198,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19367576140078324,
+      "grad_norm": 1.5289685576856569,
+      "learning_rate": 1.2169277367923207e-06,
+      "loss": 2.6704,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19380522380813672,
+      "grad_norm": 7.267548086395574,
+      "learning_rate": 1.2168370056453983e-06,
+      "loss": 4.4214,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19393468621549018,
+      "grad_norm": 4.30219798549593,
+      "learning_rate": 1.2167461536037712e-06,
+      "loss": 3.2427,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19406414862284363,
+      "grad_norm": 2.676643384513489,
+      "learning_rate": 1.2166551806859984e-06,
+      "loss": 3.0469,
+      "step": 1499
+    },
+    {
+      "epoch": 0.1941936110301971,
+      "grad_norm": 7.036672441155513,
+      "learning_rate": 1.2165640869106624e-06,
+      "loss": 5.48,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1941936110301971,
+      "eval_PRM Accuracy": 0.25471698113207547,
+      "eval_PRM F1": 0.09195402298850575,
+      "eval_PRM F1 AUC": 0.5240963855421686,
+      "eval_PRM F1 Neg": 0.368,
+      "eval_PRM NPV": 0.22549019607843138,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.04819277108433735,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4214755296707153,
+      "eval_runtime": 13.9305,
+      "eval_samples_per_second": 3.159,
+      "eval_steps_per_second": 0.215,
+      "step": 1500
     }
   ],
   "logging_steps": 1,
@@ -10242,7 +15342,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 838862698414080.0,
+  "total_flos": 1259448129847296.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null