End of training

all_results.json

@@ -1,17 +1,17 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 0.9384550084889643,
-    "eval_f1": 0.9269016697588126,
-    "eval_loss": 0.30563604831695557,
-    "eval_precision": 0.9190581309786607,
-    "eval_recall": 0.9348802395209581,
-    "eval_runtime": 8.103,
+    "eval_accuracy": 0.9681663837011885,
+    "eval_f1": 0.9637654090399701,
+    "eval_loss": 0.18448814749717712,
+    "eval_precision": 0.9619686800894854,
+    "eval_recall": 0.9655688622754491,
+    "eval_runtime": 7.9644,
     "eval_samples": 100,
-    "eval_samples_per_second": 12.341,
-    "eval_steps_per_second": 0.864,
-    "train_loss": 0.7290411071777344,
-    "train_runtime": 1777.9741,
+    "eval_samples_per_second": 12.556,
+    "eval_steps_per_second": 0.879,
+    "train_loss": 0.2323043212890625,
+    "train_runtime": 1807.2991,
     "train_samples": 800,
-    "train_samples_per_second": 8.999,
-    "train_steps_per_second": 0.562
+    "train_samples_per_second": 8.853,
+    "train_steps_per_second": 0.553
 }

eval_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 0.9384550084889643,
-    "eval_f1": 0.9269016697588126,
-    "eval_loss": 0.30563604831695557,
-    "eval_precision": 0.9190581309786607,
-    "eval_recall": 0.9348802395209581,
-    "eval_runtime": 8.103,
+    "eval_accuracy": 0.9681663837011885,
+    "eval_f1": 0.9637654090399701,
+    "eval_loss": 0.18448814749717712,
+    "eval_precision": 0.9619686800894854,
+    "eval_recall": 0.9655688622754491,
+    "eval_runtime": 7.9644,
     "eval_samples": 100,
-    "eval_samples_per_second": 12.341,
-    "eval_steps_per_second": 0.864
+    "eval_samples_per_second": 12.556,
+    "eval_steps_per_second": 0.879
 }

runs/May02_18-53-56_brutasse/events.out.tfevents.1651519687.brutasse.31018.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70da8dadbb44c976605002f3c4d73a5e5efae944bbd380049857b4aae33c6598
+size 512

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "train_loss": 0.7290411071777344,
-    "train_runtime": 1777.9741,
+    "train_loss": 0.2323043212890625,
+    "train_runtime": 1807.2991,
     "train_samples": 800,
-    "train_samples_per_second": 8.999,
-    "train_steps_per_second": 0.562
+    "train_samples_per_second": 8.853,
+    "train_steps_per_second": 0.553
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.9269016697588126,
+  "best_metric": 0.9637654090399701,
   "best_model_checkpoint": "layoutlmv3-test-cord/checkpoint-1000",
   "epoch": 20.0,
   "global_step": 1000,
@@ -9,144 +9,144 @@
   "log_history": [
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.6366723259762309,
-      "eval_f1": 0.5626932325661285,
-      "eval_loss": 1.6054191589355469,
-      "eval_precision": 0.52,
-      "eval_recall": 0.6130239520958084,
-      "eval_runtime": 8.1419,
-      "eval_samples_per_second": 12.282,
-      "eval_steps_per_second": 0.86,
+      "eval_accuracy": 0.8709677419354839,
+      "eval_f1": 0.838591342626559,
+      "eval_loss": 0.5257240533828735,
+      "eval_precision": 0.8223021582733813,
+      "eval_recall": 0.8555389221556886,
+      "eval_runtime": 8.1118,
+      "eval_samples_per_second": 12.328,
+      "eval_steps_per_second": 0.863,
       "step": 100
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8314940577249575,
-      "eval_f1": 0.8096632503660322,
-      "eval_loss": 0.9171701669692993,
-      "eval_precision": 0.7922636103151862,
-      "eval_recall": 0.8278443113772455,
-      "eval_runtime": 8.1716,
-      "eval_samples_per_second": 12.238,
-      "eval_steps_per_second": 0.857,
+      "eval_accuracy": 0.9316638370118846,
+      "eval_f1": 0.9198813056379822,
+      "eval_loss": 0.3200061023235321,
+      "eval_precision": 0.9117647058823529,
+      "eval_recall": 0.9281437125748503,
+      "eval_runtime": 7.9991,
+      "eval_samples_per_second": 12.501,
+      "eval_steps_per_second": 0.875,
       "step": 200
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.866723259762309,
-      "eval_f1": 0.8496683861459101,
-      "eval_loss": 0.6381942629814148,
-      "eval_precision": 0.8367198838896952,
-      "eval_recall": 0.8630239520958084,
-      "eval_runtime": 8.0786,
-      "eval_samples_per_second": 12.378,
-      "eval_steps_per_second": 0.866,
+      "eval_accuracy": 0.9465195246179966,
+      "eval_f1": 0.9360594795539035,
+      "eval_loss": 0.2448921650648117,
+      "eval_precision": 0.9298375184638109,
+      "eval_recall": 0.9423652694610778,
+      "eval_runtime": 7.8919,
+      "eval_samples_per_second": 12.671,
+      "eval_steps_per_second": 0.887,
       "step": 300
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.8960101867572157,
-      "eval_f1": 0.8775811209439529,
-      "eval_loss": 0.4974280595779419,
-      "eval_precision": 0.8648255813953488,
-      "eval_recall": 0.8907185628742516,
-      "eval_runtime": 8.0768,
-      "eval_samples_per_second": 12.381,
-      "eval_steps_per_second": 0.867,
+      "eval_accuracy": 0.9596774193548387,
+      "eval_f1": 0.9503916449086163,
+      "eval_loss": 0.19232001900672913,
+      "eval_precision": 0.9472118959107807,
+      "eval_recall": 0.9535928143712575,
+      "eval_runtime": 8.0551,
+      "eval_samples_per_second": 12.415,
+      "eval_steps_per_second": 0.869,
       "step": 400
     },
     {
       "epoch": 10.0,
-      "learning_rate": 5e-06,
-      "loss": 1.1589,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4328,
       "step": 500
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.916383701188455,
-      "eval_f1": 0.8914243651085757,
-      "eval_loss": 0.41238266229629517,
-      "eval_precision": 0.8769007965242578,
-      "eval_recall": 0.906437125748503,
-      "eval_runtime": 8.1158,
-      "eval_samples_per_second": 12.322,
-      "eval_steps_per_second": 0.863,
+      "eval_accuracy": 0.9681663837011885,
+      "eval_f1": 0.9623274897426333,
+      "eval_loss": 0.18568018078804016,
+      "eval_precision": 0.9591078066914498,
+      "eval_recall": 0.9655688622754491,
+      "eval_runtime": 8.1741,
+      "eval_samples_per_second": 12.234,
+      "eval_steps_per_second": 0.856,
       "step": 500
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.9235993208828522,
-      "eval_f1": 0.9064002959674436,
-      "eval_loss": 0.37666961550712585,
-      "eval_precision": 0.8961228968544257,
-      "eval_recall": 0.9169161676646707,
-      "eval_runtime": 8.1123,
-      "eval_samples_per_second": 12.327,
-      "eval_steps_per_second": 0.863,
+      "eval_accuracy": 0.9656196943972836,
+      "eval_f1": 0.9607476635514018,
+      "eval_loss": 0.20726947486400604,
+      "eval_precision": 0.9596713965646004,
+      "eval_recall": 0.9618263473053892,
+      "eval_runtime": 8.2339,
+      "eval_samples_per_second": 12.145,
+      "eval_steps_per_second": 0.85,
       "step": 600
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.933786078098472,
-      "eval_f1": 0.9210818821785848,
-      "eval_loss": 0.3387647271156311,
-      "eval_precision": 0.9119589141599413,
-      "eval_recall": 0.9303892215568862,
-      "eval_runtime": 8.0833,
-      "eval_samples_per_second": 12.371,
-      "eval_steps_per_second": 0.866,
+      "eval_accuracy": 0.9702886247877759,
+      "eval_f1": 0.9648729446935724,
+      "eval_loss": 0.1803944706916809,
+      "eval_precision": 0.9634328358208956,
+      "eval_recall": 0.9663173652694611,
+      "eval_runtime": 7.9827,
+      "eval_samples_per_second": 12.527,
+      "eval_steps_per_second": 0.877,
       "step": 700
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.9393039049235993,
-      "eval_f1": 0.927643784786642,
-      "eval_loss": 0.3137519955635071,
-      "eval_precision": 0.919793966151582,
-      "eval_recall": 0.9356287425149701,
-      "eval_runtime": 7.8384,
-      "eval_samples_per_second": 12.758,
-      "eval_steps_per_second": 0.893,
+      "eval_accuracy": 0.9664685908319185,
+      "eval_f1": 0.9640987284966342,
+      "eval_loss": 0.18818138539791107,
+      "eval_precision": 0.9633781763826607,
+      "eval_recall": 0.9648203592814372,
+      "eval_runtime": 8.0654,
+      "eval_samples_per_second": 12.399,
+      "eval_steps_per_second": 0.868,
       "step": 800
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.9376061120543294,
-      "eval_f1": 0.9254174397031539,
-      "eval_loss": 0.30727389454841614,
-      "eval_precision": 0.9175864606328182,
-      "eval_recall": 0.9333832335329342,
-      "eval_runtime": 7.8694,
-      "eval_samples_per_second": 12.708,
-      "eval_steps_per_second": 0.89,
+      "eval_accuracy": 0.967741935483871,
+      "eval_f1": 0.9633781763826607,
+      "eval_loss": 0.1799575388431549,
+      "eval_precision": 0.9619402985074627,
+      "eval_recall": 0.9648203592814372,
+      "eval_runtime": 7.9449,
+      "eval_samples_per_second": 12.587,
+      "eval_steps_per_second": 0.881,
       "step": 900
     },
     {
       "epoch": 20.0,
       "learning_rate": 0.0,
-      "loss": 0.2992,
+      "loss": 0.0318,
       "step": 1000
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.9384550084889643,
-      "eval_f1": 0.9269016697588126,
-      "eval_loss": 0.30563604831695557,
-      "eval_precision": 0.9190581309786607,
-      "eval_recall": 0.9348802395209581,
-      "eval_runtime": 7.9092,
-      "eval_samples_per_second": 12.644,
-      "eval_steps_per_second": 0.885,
+      "eval_accuracy": 0.9681663837011885,
+      "eval_f1": 0.9637654090399701,
+      "eval_loss": 0.18448814749717712,
+      "eval_precision": 0.9619686800894854,
+      "eval_recall": 0.9655688622754491,
+      "eval_runtime": 8.0126,
+      "eval_samples_per_second": 12.48,
+      "eval_steps_per_second": 0.874,
       "step": 1000
     },
     {
       "epoch": 20.0,
       "step": 1000,
       "total_flos": 4248755355648000.0,
-      "train_loss": 0.7290411071777344,
-      "train_runtime": 1777.9741,
-      "train_samples_per_second": 8.999,
-      "train_steps_per_second": 0.562
+      "train_loss": 0.2323043212890625,
+      "train_runtime": 1807.2991,
+      "train_samples_per_second": 8.853,
+      "train_steps_per_second": 0.553
     }
   ],
   "max_steps": 1000,