End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.242021276595745,
     "eval_accuracy": 0.7633136094674556,
-    "eval_loss": 0.893736720085144,
-    "eval_runtime": 11.4267,
-    "eval_samples_per_second": 14.79,
-    "eval_steps_per_second": 1.925
+    "eval_loss": 0.8817277550697327,
+    "eval_runtime": 14.8211,
+    "eval_samples_per_second": 11.403,
+    "eval_steps_per_second": 1.484
 }

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.242021276595745,
     "eval_accuracy": 0.7633136094674556,
-    "eval_loss": 0.893736720085144,
-    "eval_runtime": 11.4267,
-    "eval_samples_per_second": 14.79,
-    "eval_steps_per_second": 1.925
+    "eval_loss": 0.8817277550697327,
+    "eval_runtime": 14.8211,
+    "eval_samples_per_second": 11.403,
+    "eval_steps_per_second": 1.484
 }

trainer_state.json

@@ -10,324 +10,324 @@
   "log_history": [
     {
       "epoch": 0.026595744680851064,
-      "grad_norm": 12.626161575317383,
+      "grad_norm": 12.65776252746582,
       "learning_rate": 1.3157894736842106e-05,
-      "loss": 1.6196,
+      "loss": 1.7008,
       "step": 10
     },
     {
       "epoch": 0.05319148936170213,
-      "grad_norm": 5.767611980438232,
+      "grad_norm": 6.046972274780273,
       "learning_rate": 2.6315789473684212e-05,
-      "loss": 0.9827,
+      "loss": 0.9981,
       "step": 20
     },
     {
       "epoch": 0.0797872340425532,
-      "grad_norm": 4.556581497192383,
+      "grad_norm": 9.531058311462402,
       "learning_rate": 3.9473684210526316e-05,
-      "loss": 0.9375,
+      "loss": 0.9371,
       "step": 30
     },
     {
       "epoch": 0.10638297872340426,
-      "grad_norm": 5.054460525512695,
+      "grad_norm": 6.416304111480713,
       "learning_rate": 4.970414201183432e-05,
-      "loss": 0.8402,
+      "loss": 0.8702,
       "step": 40
     },
     {
       "epoch": 0.13297872340425532,
-      "grad_norm": 4.505584239959717,
+      "grad_norm": 4.882875442504883,
       "learning_rate": 4.822485207100592e-05,
-      "loss": 0.582,
+      "loss": 0.5812,
       "step": 50
     },
     {
       "epoch": 0.1595744680851064,
-      "grad_norm": 3.063162088394165,
+      "grad_norm": 3.9771392345428467,
       "learning_rate": 4.674556213017752e-05,
-      "loss": 1.1131,
+      "loss": 1.1064,
       "step": 60
     },
     {
       "epoch": 0.18617021276595744,
-      "grad_norm": 2.564850330352783,
+      "grad_norm": 2.8228275775909424,
       "learning_rate": 4.5266272189349114e-05,
-      "loss": 0.6761,
+      "loss": 0.6824,
       "step": 70
     },
     {
       "epoch": 0.2127659574468085,
-      "grad_norm": 2.2109172344207764,
+      "grad_norm": 2.11715030670166,
       "learning_rate": 4.378698224852072e-05,
-      "loss": 0.5614,
+      "loss": 0.5582,
       "step": 80
     },
     {
       "epoch": 0.2393617021276596,
-      "grad_norm": 1.6165765523910522,
+      "grad_norm": 1.6437604427337646,
       "learning_rate": 4.230769230769231e-05,
-      "loss": 0.8026,
+      "loss": 0.7968,
       "step": 90
     },
     {
       "epoch": 0.2526595744680851,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.8083112835884094,
-      "eval_runtime": 16.3601,
-      "eval_samples_per_second": 9.78,
-      "eval_steps_per_second": 1.222,
+      "eval_loss": 0.7984516620635986,
+      "eval_runtime": 15.2987,
+      "eval_samples_per_second": 10.458,
+      "eval_steps_per_second": 1.307,
       "step": 95
     },
     {
       "epoch": 1.0132978723404256,
-      "grad_norm": 1.8220001459121704,
+      "grad_norm": 1.7461698055267334,
       "learning_rate": 4.0828402366863904e-05,
-      "loss": 0.6855,
+      "loss": 0.6967,
       "step": 100
     },
     {
       "epoch": 1.0398936170212767,
-      "grad_norm": 2.690967321395874,
+      "grad_norm": 2.8989648818969727,
       "learning_rate": 3.934911242603551e-05,
-      "loss": 0.8036,
+      "loss": 0.8106,
       "step": 110
     },
     {
       "epoch": 1.0664893617021276,
-      "grad_norm": 2.7666921615600586,
+      "grad_norm": 2.848541021347046,
       "learning_rate": 3.7869822485207104e-05,
-      "loss": 0.7408,
+      "loss": 0.741,
       "step": 120
     },
     {
       "epoch": 1.0930851063829787,
-      "grad_norm": 2.5995383262634277,
+      "grad_norm": 2.6322543621063232,
       "learning_rate": 3.63905325443787e-05,
-      "loss": 0.7017,
+      "loss": 0.7027,
       "step": 130
     },
     {
       "epoch": 1.1196808510638299,
-      "grad_norm": 7.389265537261963,
+      "grad_norm": 7.683568954467773,
       "learning_rate": 3.49112426035503e-05,
-      "loss": 0.6631,
+      "loss": 0.6579,
       "step": 140
     },
     {
       "epoch": 1.1462765957446808,
-      "grad_norm": 1.8834972381591797,
+      "grad_norm": 1.7948863506317139,
       "learning_rate": 3.3431952662721895e-05,
-      "loss": 0.8175,
+      "loss": 0.8164,
       "step": 150
     },
     {
       "epoch": 1.172872340425532,
-      "grad_norm": 1.4516174793243408,
+      "grad_norm": 1.682706594467163,
       "learning_rate": 3.195266272189349e-05,
-      "loss": 0.6888,
+      "loss": 0.6967,
       "step": 160
     },
     {
       "epoch": 1.199468085106383,
-      "grad_norm": 3.942476272583008,
+      "grad_norm": 3.870861530303955,
       "learning_rate": 3.047337278106509e-05,
-      "loss": 0.7102,
+      "loss": 0.7218,
       "step": 170
     },
     {
       "epoch": 1.226063829787234,
-      "grad_norm": 1.646263837814331,
+      "grad_norm": 1.6848540306091309,
       "learning_rate": 2.8994082840236685e-05,
-      "loss": 0.7908,
+      "loss": 0.7805,
       "step": 180
     },
     {
       "epoch": 1.252659574468085,
-      "grad_norm": 5.147498607635498,
+      "grad_norm": 5.371254920959473,
       "learning_rate": 2.751479289940829e-05,
-      "loss": 0.8728,
+      "loss": 0.8744,
       "step": 190
     },
     {
       "epoch": 1.252659574468085,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7828324437141418,
-      "eval_runtime": 16.1891,
-      "eval_samples_per_second": 9.883,
-      "eval_steps_per_second": 1.235,
+      "eval_loss": 0.7836956977844238,
+      "eval_runtime": 11.0004,
+      "eval_samples_per_second": 14.545,
+      "eval_steps_per_second": 1.818,
       "step": 190
     },
     {
       "epoch": 2.026595744680851,
-      "grad_norm": 1.3614023923873901,
+      "grad_norm": 1.314697027206421,
       "learning_rate": 2.6035502958579882e-05,
-      "loss": 0.7153,
+      "loss": 0.7122,
       "step": 200
     },
     {
       "epoch": 2.0531914893617023,
-      "grad_norm": 1.6865636110305786,
+      "grad_norm": 1.7407877445220947,
       "learning_rate": 2.4556213017751482e-05,
-      "loss": 0.7114,
+      "loss": 0.7189,
       "step": 210
     },
     {
       "epoch": 2.0797872340425534,
-      "grad_norm": 2.2771339416503906,
+      "grad_norm": 2.266850471496582,
       "learning_rate": 2.307692307692308e-05,
-      "loss": 0.8898,
+      "loss": 0.8913,
       "step": 220
     },
     {
       "epoch": 2.106382978723404,
-      "grad_norm": 2.0663363933563232,
+      "grad_norm": 2.1507058143615723,
       "learning_rate": 2.1597633136094676e-05,
-      "loss": 0.8524,
+      "loss": 0.8554,
       "step": 230
     },
     {
       "epoch": 2.132978723404255,
-      "grad_norm": 4.288294792175293,
+      "grad_norm": 4.038496971130371,
       "learning_rate": 2.0118343195266273e-05,
-      "loss": 0.6944,
+      "loss": 0.6919,
       "step": 240
     },
     {
       "epoch": 2.1595744680851063,
-      "grad_norm": 3.9807941913604736,
+      "grad_norm": 4.005491256713867,
       "learning_rate": 1.8639053254437873e-05,
-      "loss": 0.7094,
+      "loss": 0.7086,
       "step": 250
     },
     {
       "epoch": 2.1861702127659575,
-      "grad_norm": 3.6395652294158936,
+      "grad_norm": 3.721208095550537,
       "learning_rate": 1.7159763313609466e-05,
-      "loss": 0.7492,
+      "loss": 0.7478,
       "step": 260
     },
     {
       "epoch": 2.2127659574468086,
-      "grad_norm": 4.759531021118164,
+      "grad_norm": 4.95754337310791,
       "learning_rate": 1.5680473372781066e-05,
-      "loss": 0.5475,
+      "loss": 0.5457,
       "step": 270
     },
     {
       "epoch": 2.2393617021276597,
-      "grad_norm": 2.811213970184326,
+      "grad_norm": 2.7324411869049072,
       "learning_rate": 1.4201183431952663e-05,
-      "loss": 0.952,
+      "loss": 0.9479,
       "step": 280
     },
     {
       "epoch": 2.252659574468085,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7974634170532227,
-      "eval_runtime": 12.6508,
-      "eval_samples_per_second": 12.647,
-      "eval_steps_per_second": 1.581,
+      "eval_loss": 0.7812246084213257,
+      "eval_runtime": 11.2394,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 1.779,
       "step": 285
     },
     {
       "epoch": 3.0132978723404253,
-      "grad_norm": 2.360816240310669,
+      "grad_norm": 2.182974338531494,
       "learning_rate": 1.2721893491124262e-05,
-      "loss": 0.6142,
+      "loss": 0.6173,
       "step": 290
     },
     {
       "epoch": 3.0398936170212765,
-      "grad_norm": 3.2934021949768066,
+      "grad_norm": 3.512352228164673,
       "learning_rate": 1.1242603550295859e-05,
-      "loss": 0.6263,
+      "loss": 0.6313,
       "step": 300
     },
     {
       "epoch": 3.0664893617021276,
-      "grad_norm": 2.3915889263153076,
+      "grad_norm": 2.436189651489258,
       "learning_rate": 9.763313609467455e-06,
-      "loss": 0.6514,
+      "loss": 0.6465,
       "step": 310
     },
     {
       "epoch": 3.0930851063829787,
-      "grad_norm": 3.9150514602661133,
+      "grad_norm": 4.008547782897949,
       "learning_rate": 8.284023668639054e-06,
-      "loss": 0.8365,
+      "loss": 0.8407,
       "step": 320
     },
     {
       "epoch": 3.11968085106383,
-      "grad_norm": 3.9853973388671875,
+      "grad_norm": 4.314539909362793,
       "learning_rate": 6.8047337278106515e-06,
-      "loss": 0.7772,
+      "loss": 0.7826,
       "step": 330
     },
     {
       "epoch": 3.146276595744681,
-      "grad_norm": 1.9402360916137695,
+      "grad_norm": 2.1060965061187744,
       "learning_rate": 5.325443786982249e-06,
-      "loss": 0.8747,
+      "loss": 0.8714,
       "step": 340
     },
     {
       "epoch": 3.172872340425532,
-      "grad_norm": 3.0497424602508545,
+      "grad_norm": 3.250807046890259,
       "learning_rate": 3.846153846153847e-06,
-      "loss": 0.5998,
+      "loss": 0.6124,
       "step": 350
     },
     {
       "epoch": 3.199468085106383,
-      "grad_norm": 4.310546875,
+      "grad_norm": 4.283337593078613,
       "learning_rate": 2.366863905325444e-06,
-      "loss": 0.8328,
+      "loss": 0.8395,
       "step": 360
     },
     {
       "epoch": 3.226063829787234,
-      "grad_norm": 3.943880796432495,
+      "grad_norm": 4.142538070678711,
       "learning_rate": 8.875739644970415e-07,
-      "loss": 0.8287,
+      "loss": 0.8282,
       "step": 370
     },
     {
       "epoch": 3.242021276595745,
       "eval_accuracy": 0.7875,
-      "eval_loss": 0.7972937822341919,
-      "eval_runtime": 11.2004,
-      "eval_samples_per_second": 14.285,
-      "eval_steps_per_second": 1.786,
+      "eval_loss": 0.7884188294410706,
+      "eval_runtime": 11.8341,
+      "eval_samples_per_second": 13.52,
+      "eval_steps_per_second": 1.69,
       "step": 376
     },
     {
       "epoch": 3.242021276595745,
       "step": 376,
       "total_flos": 3.7220613152994755e+18,
-      "train_loss": 0.7813464815312243,
-      "train_runtime": 603.3994,
-      "train_samples_per_second": 4.985,
-      "train_steps_per_second": 0.623
+      "train_loss": 0.785655234088289,
+      "train_runtime": 616.1212,
+      "train_samples_per_second": 4.882,
+      "train_steps_per_second": 0.61
     },
     {
       "epoch": 3.242021276595745,
       "eval_accuracy": 0.7633136094674556,
-      "eval_loss": 0.8937366008758545,
-      "eval_runtime": 12.7948,
-      "eval_samples_per_second": 13.209,
-      "eval_steps_per_second": 1.719,
+      "eval_loss": 0.8817278742790222,
+      "eval_runtime": 19.9254,
+      "eval_samples_per_second": 8.482,
+      "eval_steps_per_second": 1.104,
       "step": 376
     },
     {
       "epoch": 3.242021276595745,
       "eval_accuracy": 0.7633136094674556,
-      "eval_loss": 0.893736720085144,
-      "eval_runtime": 11.4267,
-      "eval_samples_per_second": 14.79,
-      "eval_steps_per_second": 1.925,
+      "eval_loss": 0.8817277550697327,
+      "eval_runtime": 14.8211,
+      "eval_samples_per_second": 11.403,
+      "eval_steps_per_second": 1.484,
       "step": 376
     }
   ],