End of training

all_results.json

@@ -1,17 +1,17 @@
 {
     "epoch": 21.28,
-    "eval_accuracy": 0.9908389585342333,
-    "eval_f1": 0.9868823000898472,
-    "eval_loss": 0.05581057444214821,
-    "eval_precision": 0.9845822875582646,
+    "eval_accuracy": 0.9905978784956606,
+    "eval_f1": 0.9879474725670084,
+    "eval_loss": 0.058533914387226105,
+    "eval_precision": 0.986704994610133,
     "eval_recall": 0.989193083573487,
-    "eval_runtime": 15.2739,
+    "eval_runtime": 13.9248,
     "eval_samples": 189,
-    "eval_samples_per_second": 12.374,
-    "eval_steps_per_second": 0.786,
-    "train_loss": 0.1608761215209961,
-    "train_runtime": 1880.5995,
+    "eval_samples_per_second": 13.573,
+    "eval_steps_per_second": 0.862,
+    "train_loss": 0.16068801975250244,
+    "train_runtime": 1757.2853,
     "train_samples": 752,
-    "train_samples_per_second": 8.508,
-    "train_steps_per_second": 0.532
+    "train_samples_per_second": 9.105,
+    "train_steps_per_second": 0.569
 }

eval_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 21.28,
-    "eval_accuracy": 0.9908389585342333,
-    "eval_f1": 0.9868823000898472,
-    "eval_loss": 0.05581057444214821,
-    "eval_precision": 0.9845822875582646,
+    "eval_accuracy": 0.9905978784956606,
+    "eval_f1": 0.9879474725670084,
+    "eval_loss": 0.058533914387226105,
+    "eval_precision": 0.986704994610133,
     "eval_recall": 0.989193083573487,
-    "eval_runtime": 15.2739,
+    "eval_runtime": 13.9248,
     "eval_samples": 189,
-    "eval_samples_per_second": 12.374,
-    "eval_steps_per_second": 0.786
+    "eval_samples_per_second": 13.573,
+    "eval_steps_per_second": 0.862
 }

runs/Dec20_22-54-58_pod-pasha/events.out.tfevents.1671578896.pod-pasha.546411.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f1f54c44016fdcb07e7ab4264c1d91bc832540bcb358903a19844908e93caed
+size 512

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 21.28,
-    "train_loss": 0.1608761215209961,
-    "train_runtime": 1880.5995,
+    "train_loss": 0.16068801975250244,
+    "train_runtime": 1757.2853,
     "train_samples": 752,
-    "train_samples_per_second": 8.508,
-    "train_steps_per_second": 0.532
+    "train_samples_per_second": 9.105,
+    "train_steps_per_second": 0.569
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.9868823000898472,
-  "best_model_checkpoint": "pasha/checkpoint-1000",
+  "best_metric": 0.9879474725670084,
+  "best_model_checkpoint": "pasha/checkpoint-500",
   "epoch": 21.27659574468085,
   "global_step": 1000,
   "is_hyper_param_search": false,
@@ -9,144 +9,144 @@
   "log_history": [
     {
       "epoch": 2.13,
-      "eval_accuracy": 0.9566055930568949,
-      "eval_f1": 0.9482633863965269,
-      "eval_loss": 0.2661653161048889,
-      "eval_precision": 0.9523982558139535,
-      "eval_recall": 0.944164265129683,
-      "eval_runtime": 15.4131,
-      "eval_samples_per_second": 12.262,
-      "eval_steps_per_second": 0.779,
+      "eval_accuracy": 0.9570877531340405,
+      "eval_f1": 0.9485879797248371,
+      "eval_loss": 0.2664182484149933,
+      "eval_precision": 0.9534206695778749,
+      "eval_recall": 0.9438040345821326,
+      "eval_runtime": 13.9899,
+      "eval_samples_per_second": 13.51,
+      "eval_steps_per_second": 0.858,
       "step": 100
     },
     {
       "epoch": 4.26,
-      "eval_accuracy": 0.9850530376084861,
-      "eval_f1": 0.9795185052102047,
-      "eval_loss": 0.1026068776845932,
-      "eval_precision": 0.9770609318996416,
-      "eval_recall": 0.9819884726224783,
-      "eval_runtime": 15.204,
-      "eval_samples_per_second": 12.431,
-      "eval_steps_per_second": 0.789,
+      "eval_accuracy": 0.983847637415622,
+      "eval_f1": 0.9778975741239893,
+      "eval_loss": 0.10435084253549576,
+      "eval_precision": 0.9756185012549301,
+      "eval_recall": 0.9801873198847262,
+      "eval_runtime": 13.823,
+      "eval_samples_per_second": 13.673,
+      "eval_steps_per_second": 0.868,
       "step": 200
     },
     {
       "epoch": 6.38,
-      "eval_accuracy": 0.9884281581485053,
-      "eval_f1": 0.9849137931034483,
-      "eval_loss": 0.07217290997505188,
-      "eval_precision": 0.9820916905444126,
-      "eval_recall": 0.9877521613832853,
-      "eval_runtime": 15.2143,
-      "eval_samples_per_second": 12.423,
-      "eval_steps_per_second": 0.789,
+      "eval_accuracy": 0.9903567984570878,
+      "eval_f1": 0.987601078167116,
+      "eval_loss": 0.06718672811985016,
+      "eval_precision": 0.9852993904625313,
+      "eval_recall": 0.9899135446685879,
+      "eval_runtime": 13.9839,
+      "eval_samples_per_second": 13.516,
+      "eval_steps_per_second": 0.858,
       "step": 300
     },
     {
       "epoch": 8.51,
-      "eval_accuracy": 0.9891513982642237,
-      "eval_f1": 0.9857785778577858,
-      "eval_loss": 0.060767240822315216,
-      "eval_precision": 0.9852464915437208,
-      "eval_recall": 0.9863112391930836,
-      "eval_runtime": 15.3275,
-      "eval_samples_per_second": 12.331,
-      "eval_steps_per_second": 0.783,
+      "eval_accuracy": 0.9884281581485053,
+      "eval_f1": 0.9841783531103919,
+      "eval_loss": 0.06342343986034393,
+      "eval_precision": 0.9824120603015075,
+      "eval_recall": 0.9859510086455331,
+      "eval_runtime": 13.8826,
+      "eval_samples_per_second": 13.614,
+      "eval_steps_per_second": 0.864,
       "step": 400
     },
     {
       "epoch": 10.64,
       "learning_rate": 5e-06,
-      "loss": 0.2962,
+      "loss": 0.2958,
       "step": 500
     },
     {
       "epoch": 10.64,
-      "eval_accuracy": 0.9889103182256509,
-      "eval_f1": 0.9854185418541853,
-      "eval_loss": 0.060581281781196594,
-      "eval_precision": 0.9848866498740554,
-      "eval_recall": 0.9859510086455331,
-      "eval_runtime": 14.8027,
-      "eval_samples_per_second": 12.768,
-      "eval_steps_per_second": 0.811,
+      "eval_accuracy": 0.9905978784956606,
+      "eval_f1": 0.9879474725670084,
+      "eval_loss": 0.058533914387226105,
+      "eval_precision": 0.986704994610133,
+      "eval_recall": 0.989193083573487,
+      "eval_runtime": 13.739,
+      "eval_samples_per_second": 13.756,
+      "eval_steps_per_second": 0.873,
       "step": 500
     },
     {
       "epoch": 12.77,
-      "eval_accuracy": 0.9920443587270974,
-      "eval_f1": 0.988501616960115,
-      "eval_loss": 0.0517994724214077,
-      "eval_precision": 0.986021505376344,
-      "eval_recall": 0.9909942363112392,
-      "eval_runtime": 14.7335,
-      "eval_samples_per_second": 12.828,
-      "eval_steps_per_second": 0.814,
+      "eval_accuracy": 0.9927675988428158,
+      "eval_f1": 0.9908322847384505,
+      "eval_loss": 0.051136456429958344,
+      "eval_precision": 0.9888769285970578,
+      "eval_recall": 0.9927953890489913,
+      "eval_runtime": 14.0393,
+      "eval_samples_per_second": 13.462,
+      "eval_steps_per_second": 0.855,
       "step": 600
     },
     {
       "epoch": 14.89,
-      "eval_accuracy": 0.9922854387656702,
-      "eval_f1": 0.988679245283019,
-      "eval_loss": 0.052589546889066696,
-      "eval_precision": 0.9863750448189316,
-      "eval_recall": 0.9909942363112392,
-      "eval_runtime": 14.9339,
-      "eval_samples_per_second": 12.656,
-      "eval_steps_per_second": 0.804,
+      "eval_accuracy": 0.992526518804243,
+      "eval_f1": 0.9895795903701042,
+      "eval_loss": 0.05025022476911545,
+      "eval_precision": 0.9870967741935484,
+      "eval_recall": 0.9920749279538905,
+      "eval_runtime": 14.0132,
+      "eval_samples_per_second": 13.487,
+      "eval_steps_per_second": 0.856,
       "step": 700
     },
     {
       "epoch": 17.02,
       "eval_accuracy": 0.991321118611379,
-      "eval_f1": 0.9872416891284815,
-      "eval_loss": 0.05428989231586456,
-      "eval_precision": 0.984940839010398,
-      "eval_recall": 0.9895533141210374,
-      "eval_runtime": 14.8022,
-      "eval_samples_per_second": 12.768,
-      "eval_steps_per_second": 0.811,
+      "eval_f1": 0.9881380301941048,
+      "eval_loss": 0.05291323363780975,
+      "eval_precision": 0.9860114777618364,
+      "eval_recall": 0.9902737752161384,
+      "eval_runtime": 13.9493,
+      "eval_samples_per_second": 13.549,
+      "eval_steps_per_second": 0.86,
       "step": 800
     },
     {
       "epoch": 19.15,
-      "eval_accuracy": 0.9910800385728061,
-      "eval_f1": 0.9867002156721782,
-      "eval_loss": 0.05573796480894089,
-      "eval_precision": 0.9845767575322812,
-      "eval_recall": 0.9888328530259366,
-      "eval_runtime": 14.7741,
-      "eval_samples_per_second": 12.793,
-      "eval_steps_per_second": 0.812,
+      "eval_accuracy": 0.9903567984570878,
+      "eval_f1": 0.986704994610133,
+      "eval_loss": 0.0581122450530529,
+      "eval_precision": 0.9842293906810036,
+      "eval_recall": 0.989193083573487,
+      "eval_runtime": 13.9545,
+      "eval_samples_per_second": 13.544,
+      "eval_steps_per_second": 0.86,
       "step": 900
     },
     {
       "epoch": 21.28,
       "learning_rate": 0.0,
-      "loss": 0.0255,
+      "loss": 0.0256,
       "step": 1000
     },
     {
       "epoch": 21.28,
-      "eval_accuracy": 0.9908389585342333,
-      "eval_f1": 0.9868823000898472,
-      "eval_loss": 0.05581057444214821,
-      "eval_precision": 0.9845822875582646,
-      "eval_recall": 0.989193083573487,
-      "eval_runtime": 15.103,
-      "eval_samples_per_second": 12.514,
-      "eval_steps_per_second": 0.795,
+      "eval_accuracy": 0.990115718418515,
+      "eval_f1": 0.9868775840373899,
+      "eval_loss": 0.057054802775382996,
+      "eval_precision": 0.984930032292788,
+      "eval_recall": 0.9888328530259366,
+      "eval_runtime": 14.2093,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 0.845,
       "step": 1000
     },
     {
       "epoch": 21.28,
       "step": 1000,
       "total_flos": 4247054450688000.0,
-      "train_loss": 0.1608761215209961,
-      "train_runtime": 1880.5995,
-      "train_samples_per_second": 8.508,
-      "train_steps_per_second": 0.532
+      "train_loss": 0.16068801975250244,
+      "train_runtime": 1757.2853,
+      "train_samples_per_second": 9.105,
+      "train_steps_per_second": 0.569
     }
   ],
   "max_steps": 1000,