Fin de l'entraînement

all_results.json

@@ -1,14 +1,14 @@
 {
     "epoch": 4.0,
-    "eval_loss": 6.874520301818848,
-    "eval_mae": 2.7379310131073,
-    "eval_mse": 7.4963178634643555,
-    "eval_runtime": 1.2013,
-    "eval_samples_per_second": 8.324,
-    "eval_steps_per_second": 1.665,
+    "eval_loss": 6.522369384765625,
+    "eval_mae": 2.7217423915863037,
+    "eval_mse": 7.407970428466797,
+    "eval_runtime": 2.0059,
+    "eval_samples_per_second": 4.985,
+    "eval_steps_per_second": 0.997,
     "total_flos": 0.0,
-    "train_loss": 17.268508911132812,
-    "train_runtime": 30.6646,
-    "train_samples_per_second": 10.175,
-    "train_steps_per_second": 0.261
+    "train_loss": 15.188075065612793,
+    "train_runtime": 31.017,
+    "train_samples_per_second": 10.059,
+    "train_steps_per_second": 0.258
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 4.0,
-    "eval_loss": 6.874520301818848,
-    "eval_mae": 2.7379310131073,
-    "eval_mse": 7.4963178634643555,
-    "eval_runtime": 1.2013,
-    "eval_samples_per_second": 8.324,
-    "eval_steps_per_second": 1.665
+    "eval_loss": 6.522369384765625,
+    "eval_mae": 2.7217423915863037,
+    "eval_mse": 7.407970428466797,
+    "eval_runtime": 2.0059,
+    "eval_samples_per_second": 4.985,
+    "eval_steps_per_second": 0.997
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
     "total_flos": 0.0,
-    "train_loss": 17.268508911132812,
-    "train_runtime": 30.6646,
-    "train_samples_per_second": 10.175,
-    "train_steps_per_second": 0.261
+    "train_loss": 15.188075065612793,
+    "train_runtime": 31.017,
+    "train_samples_per_second": 10.059,
+    "train_steps_per_second": 0.258
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_global_step": 8,
-  "best_metric": 7.4963178634643555,
+  "best_metric": 7.407970428466797,
   "best_model_checkpoint": "./vit_focus/checkpoint-8",
   "epoch": 4.0,
   "eval_steps": 500,
@@ -11,52 +11,52 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "eval_loss": 6.970195770263672,
-      "eval_mae": 2.7422118186950684,
-      "eval_mse": 7.51975154876709,
-      "eval_runtime": 1.2733,
-      "eval_samples_per_second": 7.854,
-      "eval_steps_per_second": 1.571,
+      "eval_loss": 6.890933036804199,
+      "eval_mae": 2.7386960983276367,
+      "eval_mse": 7.500483512878418,
+      "eval_runtime": 1.2828,
+      "eval_samples_per_second": 7.795,
+      "eval_steps_per_second": 1.559,
       "step": 2
     },
     {
       "epoch": 2.0,
-      "eval_loss": 6.951096534729004,
-      "eval_mae": 2.741365671157837,
-      "eval_mse": 7.515117645263672,
-      "eval_runtime": 1.1796,
-      "eval_samples_per_second": 8.478,
-      "eval_steps_per_second": 1.696,
+      "eval_loss": 6.799283504486084,
+      "eval_mae": 2.7345499992370605,
+      "eval_mse": 7.4778008460998535,
+      "eval_runtime": 1.1777,
+      "eval_samples_per_second": 8.491,
+      "eval_steps_per_second": 1.698,
       "step": 4
     },
     {
       "epoch": 3.0,
-      "eval_loss": 6.9106879234313965,
-      "eval_mae": 2.739567279815674,
-      "eval_mse": 7.5052666664123535,
-      "eval_runtime": 1.1711,
-      "eval_samples_per_second": 8.539,
-      "eval_steps_per_second": 1.708,
+      "eval_loss": 6.6634087562561035,
+      "eval_mae": 2.728336811065674,
+      "eval_mse": 7.443875789642334,
+      "eval_runtime": 1.1717,
+      "eval_samples_per_second": 8.535,
+      "eval_steps_per_second": 1.707,
       "step": 6
     },
     {
       "epoch": 4.0,
-      "eval_loss": 6.874520301818848,
-      "eval_mae": 2.7379310131073,
-      "eval_mse": 7.4963178634643555,
-      "eval_runtime": 1.1848,
-      "eval_samples_per_second": 8.44,
-      "eval_steps_per_second": 1.688,
+      "eval_loss": 6.522369384765625,
+      "eval_mae": 2.7217423915863037,
+      "eval_mse": 7.407970428466797,
+      "eval_runtime": 1.2243,
+      "eval_samples_per_second": 8.168,
+      "eval_steps_per_second": 1.634,
       "step": 8
     },
     {
       "epoch": 4.0,
       "step": 8,
       "total_flos": 0.0,
-      "train_loss": 17.268508911132812,
-      "train_runtime": 30.6646,
-      "train_samples_per_second": 10.175,
-      "train_steps_per_second": 0.261
+      "train_loss": 15.188075065612793,
+      "train_runtime": 31.017,
+      "train_samples_per_second": 10.059,
+      "train_steps_per_second": 0.258
     }
   ],
   "logging_steps": 50,