End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.242021276595745,
     "eval_accuracy": 0.7633136094674556,
-    "eval_loss": 0.8817277550697327,
-    "eval_runtime": 14.8211,
-    "eval_samples_per_second": 11.403,
-    "eval_steps_per_second": 1.484
+    "eval_loss": 0.8992812633514404,
+    "eval_runtime": 12.8676,
+    "eval_samples_per_second": 13.134,
+    "eval_steps_per_second": 1.71
 }

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.242021276595745,
     "eval_accuracy": 0.7633136094674556,
-    "eval_loss": 0.8817277550697327,
-    "eval_runtime": 14.8211,
-    "eval_samples_per_second": 11.403,
-    "eval_steps_per_second": 1.484
+    "eval_loss": 0.8992812633514404,
+    "eval_runtime": 12.8676,
+    "eval_samples_per_second": 13.134,
+    "eval_steps_per_second": 1.71
 }

trainer_state.json

@@ -10,324 +10,324 @@
   "log_history": [
     {
       "epoch": 0.026595744680851064,
-      "grad_norm": 12.65776252746582,
+      "grad_norm": 11.770811080932617,
       "learning_rate": 1.3157894736842106e-05,
-      "loss": 1.7008,
+      "loss": 1.5077,
       "step": 10
     },
     {
       "epoch": 0.05319148936170213,
-      "grad_norm": 6.046972274780273,
+      "grad_norm": 6.303102493286133,
       "learning_rate": 2.6315789473684212e-05,
-      "loss": 0.9981,
+      "loss": 0.9267,
       "step": 20
     },
     {
       "epoch": 0.0797872340425532,
-      "grad_norm": 9.531058311462402,
+      "grad_norm": 5.121222019195557,
       "learning_rate": 3.9473684210526316e-05,
-      "loss": 0.9371,
+      "loss": 0.9311,
       "step": 30
     },
     {
       "epoch": 0.10638297872340426,
-      "grad_norm": 6.416304111480713,
+      "grad_norm": 5.0847344398498535,
       "learning_rate": 4.970414201183432e-05,
-      "loss": 0.8702,
+      "loss": 0.8635,
       "step": 40
     },
     {
       "epoch": 0.13297872340425532,
-      "grad_norm": 4.882875442504883,
+      "grad_norm": 4.5861735343933105,
       "learning_rate": 4.822485207100592e-05,
-      "loss": 0.5812,
+      "loss": 0.5788,
       "step": 50
     },
     {
       "epoch": 0.1595744680851064,
-      "grad_norm": 3.9771392345428467,
+      "grad_norm": 3.4006526470184326,
       "learning_rate": 4.674556213017752e-05,
-      "loss": 1.1064,
+      "loss": 1.0821,
       "step": 60
     },
     {
       "epoch": 0.18617021276595744,
-      "grad_norm": 2.8228275775909424,
+      "grad_norm": 2.8479793071746826,
       "learning_rate": 4.5266272189349114e-05,
-      "loss": 0.6824,
+      "loss": 0.6913,
       "step": 70
     },
     {
       "epoch": 0.2127659574468085,
-      "grad_norm": 2.11715030670166,
+      "grad_norm": 2.3751566410064697,
       "learning_rate": 4.378698224852072e-05,
-      "loss": 0.5582,
+      "loss": 0.5482,
       "step": 80
     },
     {
       "epoch": 0.2393617021276596,
-      "grad_norm": 1.6437604427337646,
+      "grad_norm": 1.6777262687683105,
       "learning_rate": 4.230769230769231e-05,
-      "loss": 0.7968,
+      "loss": 0.8086,
       "step": 90
     },
     {
       "epoch": 0.2526595744680851,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7984516620635986,
-      "eval_runtime": 15.2987,
-      "eval_samples_per_second": 10.458,
-      "eval_steps_per_second": 1.307,
+      "eval_loss": 0.8058681488037109,
+      "eval_runtime": 13.8118,
+      "eval_samples_per_second": 11.584,
+      "eval_steps_per_second": 1.448,
       "step": 95
     },
     {
       "epoch": 1.0132978723404256,
-      "grad_norm": 1.7461698055267334,
+      "grad_norm": 2.0221545696258545,
       "learning_rate": 4.0828402366863904e-05,
-      "loss": 0.6967,
+      "loss": 0.6867,
       "step": 100
     },
     {
       "epoch": 1.0398936170212767,
-      "grad_norm": 2.8989648818969727,
+      "grad_norm": 2.89918851852417,
       "learning_rate": 3.934911242603551e-05,
-      "loss": 0.8106,
+      "loss": 0.7971,
       "step": 110
     },
     {
       "epoch": 1.0664893617021276,
-      "grad_norm": 2.848541021347046,
+      "grad_norm": 2.958357810974121,
       "learning_rate": 3.7869822485207104e-05,
-      "loss": 0.741,
+      "loss": 0.7237,
       "step": 120
     },
     {
       "epoch": 1.0930851063829787,
-      "grad_norm": 2.6322543621063232,
+      "grad_norm": 2.737426996231079,
       "learning_rate": 3.63905325443787e-05,
-      "loss": 0.7027,
+      "loss": 0.7045,
       "step": 130
     },
     {
       "epoch": 1.1196808510638299,
-      "grad_norm": 7.683568954467773,
+      "grad_norm": 8.279816627502441,
       "learning_rate": 3.49112426035503e-05,
-      "loss": 0.6579,
+      "loss": 0.6584,
       "step": 140
     },
     {
       "epoch": 1.1462765957446808,
-      "grad_norm": 1.7948863506317139,
+      "grad_norm": 1.7154802083969116,
       "learning_rate": 3.3431952662721895e-05,
-      "loss": 0.8164,
+      "loss": 0.8076,
       "step": 150
     },
     {
       "epoch": 1.172872340425532,
-      "grad_norm": 1.682706594467163,
+      "grad_norm": 1.5087482929229736,
       "learning_rate": 3.195266272189349e-05,
-      "loss": 0.6967,
+      "loss": 0.7007,
       "step": 160
     },
     {
       "epoch": 1.199468085106383,
-      "grad_norm": 3.870861530303955,
+      "grad_norm": 4.256617069244385,
       "learning_rate": 3.047337278106509e-05,
-      "loss": 0.7218,
+      "loss": 0.7228,
       "step": 170
     },
     {
       "epoch": 1.226063829787234,
-      "grad_norm": 1.6848540306091309,
+      "grad_norm": 1.8001201152801514,
       "learning_rate": 2.8994082840236685e-05,
-      "loss": 0.7805,
+      "loss": 0.7797,
       "step": 180
     },
     {
       "epoch": 1.252659574468085,
-      "grad_norm": 5.371254920959473,
+      "grad_norm": 5.422004222869873,
       "learning_rate": 2.751479289940829e-05,
-      "loss": 0.8744,
+      "loss": 0.8755,
       "step": 190
     },
     {
       "epoch": 1.252659574468085,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7836956977844238,
-      "eval_runtime": 11.0004,
-      "eval_samples_per_second": 14.545,
-      "eval_steps_per_second": 1.818,
+      "eval_loss": 0.7764860987663269,
+      "eval_runtime": 10.4189,
+      "eval_samples_per_second": 15.357,
+      "eval_steps_per_second": 1.92,
       "step": 190
     },
     {
       "epoch": 2.026595744680851,
-      "grad_norm": 1.314697027206421,
+      "grad_norm": 1.4132887125015259,
       "learning_rate": 2.6035502958579882e-05,
-      "loss": 0.7122,
+      "loss": 0.7022,
       "step": 200
     },
     {
       "epoch": 2.0531914893617023,
-      "grad_norm": 1.7407877445220947,
+      "grad_norm": 1.8760045766830444,
       "learning_rate": 2.4556213017751482e-05,
-      "loss": 0.7189,
+      "loss": 0.7098,
       "step": 210
     },
     {
       "epoch": 2.0797872340425534,
-      "grad_norm": 2.266850471496582,
+      "grad_norm": 2.369459629058838,
       "learning_rate": 2.307692307692308e-05,
-      "loss": 0.8913,
+      "loss": 0.8856,
       "step": 220
     },
     {
       "epoch": 2.106382978723404,
-      "grad_norm": 2.1507058143615723,
+      "grad_norm": 2.2688181400299072,
       "learning_rate": 2.1597633136094676e-05,
-      "loss": 0.8554,
+      "loss": 0.8472,
       "step": 230
     },
     {
       "epoch": 2.132978723404255,
-      "grad_norm": 4.038496971130371,
+      "grad_norm": 4.264322280883789,
       "learning_rate": 2.0118343195266273e-05,
-      "loss": 0.6919,
+      "loss": 0.6886,
       "step": 240
     },
     {
       "epoch": 2.1595744680851063,
-      "grad_norm": 4.005491256713867,
+      "grad_norm": 3.999258518218994,
       "learning_rate": 1.8639053254437873e-05,
-      "loss": 0.7086,
+      "loss": 0.7047,
       "step": 250
     },
     {
       "epoch": 2.1861702127659575,
-      "grad_norm": 3.721208095550537,
+      "grad_norm": 3.8878672122955322,
       "learning_rate": 1.7159763313609466e-05,
-      "loss": 0.7478,
+      "loss": 0.7449,
       "step": 260
     },
     {
       "epoch": 2.2127659574468086,
-      "grad_norm": 4.95754337310791,
+      "grad_norm": 4.918434143066406,
       "learning_rate": 1.5680473372781066e-05,
-      "loss": 0.5457,
+      "loss": 0.5438,
       "step": 270
     },
     {
       "epoch": 2.2393617021276597,
-      "grad_norm": 2.7324411869049072,
+      "grad_norm": 2.8296828269958496,
       "learning_rate": 1.4201183431952663e-05,
-      "loss": 0.9479,
+      "loss": 0.9334,
       "step": 280
     },
     {
       "epoch": 2.252659574468085,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7812246084213257,
-      "eval_runtime": 11.2394,
-      "eval_samples_per_second": 14.236,
-      "eval_steps_per_second": 1.779,
+      "eval_loss": 0.7846399545669556,
+      "eval_runtime": 10.9493,
+      "eval_samples_per_second": 14.613,
+      "eval_steps_per_second": 1.827,
       "step": 285
     },
     {
       "epoch": 3.0132978723404253,
-      "grad_norm": 2.182974338531494,
+      "grad_norm": 2.3935749530792236,
       "learning_rate": 1.2721893491124262e-05,
-      "loss": 0.6173,
+      "loss": 0.6163,
       "step": 290
     },
     {
       "epoch": 3.0398936170212765,
-      "grad_norm": 3.512352228164673,
+      "grad_norm": 3.57572340965271,
       "learning_rate": 1.1242603550295859e-05,
-      "loss": 0.6313,
+      "loss": 0.6265,
       "step": 300
     },
     {
       "epoch": 3.0664893617021276,
-      "grad_norm": 2.436189651489258,
+      "grad_norm": 2.4465370178222656,
       "learning_rate": 9.763313609467455e-06,
-      "loss": 0.6465,
+      "loss": 0.64,
       "step": 310
     },
     {
       "epoch": 3.0930851063829787,
-      "grad_norm": 4.008547782897949,
+      "grad_norm": 4.222550868988037,
       "learning_rate": 8.284023668639054e-06,
-      "loss": 0.8407,
+      "loss": 0.8452,
       "step": 320
     },
     {
       "epoch": 3.11968085106383,
-      "grad_norm": 4.314539909362793,
+      "grad_norm": 3.980487823486328,
       "learning_rate": 6.8047337278106515e-06,
-      "loss": 0.7826,
+      "loss": 0.7681,
       "step": 330
     },
     {
       "epoch": 3.146276595744681,
-      "grad_norm": 2.1060965061187744,
+      "grad_norm": 1.943975806236267,
       "learning_rate": 5.325443786982249e-06,
-      "loss": 0.8714,
+      "loss": 0.8814,
       "step": 340
     },
     {
       "epoch": 3.172872340425532,
-      "grad_norm": 3.250807046890259,
+      "grad_norm": 3.3926122188568115,
       "learning_rate": 3.846153846153847e-06,
-      "loss": 0.6124,
+      "loss": 0.5995,
       "step": 350
     },
     {
       "epoch": 3.199468085106383,
-      "grad_norm": 4.283337593078613,
+      "grad_norm": 4.575331211090088,
       "learning_rate": 2.366863905325444e-06,
-      "loss": 0.8395,
+      "loss": 0.8319,
       "step": 360
     },
     {
       "epoch": 3.226063829787234,
-      "grad_norm": 4.142538070678711,
+      "grad_norm": 4.1369309425354,
       "learning_rate": 8.875739644970415e-07,
-      "loss": 0.8282,
+      "loss": 0.8263,
       "step": 370
     },
     {
       "epoch": 3.242021276595745,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7884188294410706,
-      "eval_runtime": 11.8341,
-      "eval_samples_per_second": 13.52,
-      "eval_steps_per_second": 1.69,
+      "eval_loss": 0.7844525575637817,
+      "eval_runtime": 11.1167,
+      "eval_samples_per_second": 14.393,
+      "eval_steps_per_second": 1.799,
       "step": 376
     },
     {
       "epoch": 3.242021276595745,
       "step": 376,
       "total_flos": 3.7220613152994755e+18,
-      "train_loss": 0.785655234088289,
-      "train_runtime": 616.1212,
-      "train_samples_per_second": 4.882,
-      "train_steps_per_second": 0.61
+      "train_loss": 0.7742417086946204,
+      "train_runtime": 578.534,
+      "train_samples_per_second": 5.199,
+      "train_steps_per_second": 0.65
     },
     {
       "epoch": 3.242021276595745,
       "eval_accuracy": 0.7633136094674556,
-      "eval_loss": 0.8817278742790222,
-      "eval_runtime": 19.9254,
-      "eval_samples_per_second": 8.482,
-      "eval_steps_per_second": 1.104,
+      "eval_loss": 0.8992813229560852,
+      "eval_runtime": 20.2942,
+      "eval_samples_per_second": 8.327,
+      "eval_steps_per_second": 1.084,
       "step": 376
     },
     {
       "epoch": 3.242021276595745,
       "eval_accuracy": 0.7633136094674556,
-      "eval_loss": 0.8817277550697327,
-      "eval_runtime": 14.8211,
-      "eval_samples_per_second": 11.403,
-      "eval_steps_per_second": 1.484,
+      "eval_loss": 0.8992812633514404,
+      "eval_runtime": 12.8676,
+      "eval_samples_per_second": 13.134,
+      "eval_steps_per_second": 1.71,
       "step": 376
     }
   ],