End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.25,
-    "eval_accuracy": 0.7677419354838709,
-    "eval_loss": 0.5838326811790466,
-    "eval_runtime": 30.1878,
-    "eval_samples_per_second": 5.135,
-    "eval_steps_per_second": 1.292
+    "eval_accuracy": 0.9225806451612903,
+    "eval_loss": 0.30445486307144165,
+    "eval_runtime": 29.6038,
+    "eval_samples_per_second": 5.236,
+    "eval_steps_per_second": 1.317
 }

runs/May12_18-11-57_gcp/events.out.tfevents.1715538234.gcp.1903.1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a08856973faf170e6a22240d6c0b2f1ab41b6e78dd87851ec5f789df1298d44e
-size 411
+oid sha256:bff98acb5efd169c6c71d9516f4f608d6c67d67935d9504cea2f5b4ea976306d
+size 734

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.25,
-    "eval_accuracy": 0.7677419354838709,
-    "eval_loss": 0.5838326811790466,
-    "eval_runtime": 30.1878,
-    "eval_samples_per_second": 5.135,
-    "eval_steps_per_second": 1.292
+    "eval_accuracy": 0.9225806451612903,
+    "eval_loss": 0.30445486307144165,
+    "eval_runtime": 29.6038,
+    "eval_samples_per_second": 5.236,
+    "eval_steps_per_second": 1.317
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.8428571428571429,
-  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-150",
+  "best_metric": 0.9,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-300",
   "epoch": 3.25,
   "eval_steps": 500,
   "global_step": 300,
@@ -10,275 +10,275 @@
   "log_history": [
     {
       "epoch": 0.03333333333333333,
-      "grad_norm": 9.030272483825684,
+      "grad_norm": 9.796407699584961,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 2.3471,
+      "loss": 2.2957,
       "step": 10
     },
     {
       "epoch": 0.06666666666666667,
-      "grad_norm": 9.653508186340332,
+      "grad_norm": 9.56534194946289,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 2.2728,
+      "loss": 2.2844,
       "step": 20
     },
     {
       "epoch": 0.1,
-      "grad_norm": 9.514510154724121,
+      "grad_norm": 9.162632942199707,
       "learning_rate": 5e-05,
-      "loss": 2.2423,
+      "loss": 2.1956,
       "step": 30
     },
     {
       "epoch": 0.13333333333333333,
-      "grad_norm": 9.425768852233887,
+      "grad_norm": 9.857256889343262,
       "learning_rate": 4.814814814814815e-05,
-      "loss": 2.0797,
+      "loss": 2.0529,
       "step": 40
     },
     {
       "epoch": 0.16666666666666666,
-      "grad_norm": 12.643138885498047,
+      "grad_norm": 14.574447631835938,
       "learning_rate": 4.62962962962963e-05,
-      "loss": 2.1012,
+      "loss": 2.0628,
       "step": 50
     },
     {
       "epoch": 0.2,
-      "grad_norm": 9.729035377502441,
+      "grad_norm": 9.55310344696045,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 1.7497,
+      "loss": 1.769,
       "step": 60
     },
     {
       "epoch": 0.23333333333333334,
-      "grad_norm": 9.487822532653809,
+      "grad_norm": 12.72282600402832,
       "learning_rate": 4.259259259259259e-05,
-      "loss": 1.7455,
+      "loss": 1.7867,
       "step": 70
     },
     {
       "epoch": 0.25,
-      "eval_accuracy": 0.5857142857142857,
-      "eval_loss": 1.3021408319473267,
-      "eval_runtime": 16.6801,
-      "eval_samples_per_second": 4.197,
-      "eval_steps_per_second": 1.079,
+      "eval_accuracy": 0.5428571428571428,
+      "eval_loss": 1.3080873489379883,
+      "eval_runtime": 16.1891,
+      "eval_samples_per_second": 4.324,
+      "eval_steps_per_second": 1.112,
       "step": 75
     },
     {
       "epoch": 1.0166666666666666,
-      "grad_norm": 9.826342582702637,
+      "grad_norm": 9.65126895904541,
       "learning_rate": 4.074074074074074e-05,
-      "loss": 1.3037,
+      "loss": 1.2955,
       "step": 80
     },
     {
       "epoch": 1.05,
-      "grad_norm": 10.802420616149902,
+      "grad_norm": 9.708106994628906,
       "learning_rate": 3.888888888888889e-05,
-      "loss": 1.128,
+      "loss": 0.9881,
       "step": 90
     },
     {
       "epoch": 1.0833333333333333,
-      "grad_norm": 15.328178405761719,
+      "grad_norm": 13.812542915344238,
       "learning_rate": 3.7037037037037037e-05,
-      "loss": 1.1334,
+      "loss": 1.0424,
       "step": 100
     },
     {
       "epoch": 1.1166666666666667,
-      "grad_norm": 9.453879356384277,
+      "grad_norm": 10.680852890014648,
       "learning_rate": 3.518518518518519e-05,
-      "loss": 0.8507,
+      "loss": 0.8059,
       "step": 110
     },
     {
       "epoch": 1.15,
-      "grad_norm": 10.099981307983398,
+      "grad_norm": 8.754110336303711,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.8617,
+      "loss": 0.7942,
       "step": 120
     },
     {
       "epoch": 1.1833333333333333,
-      "grad_norm": 21.905996322631836,
+      "grad_norm": 12.191349029541016,
       "learning_rate": 3.148148148148148e-05,
-      "loss": 0.6073,
+      "loss": 0.5023,
       "step": 130
     },
     {
       "epoch": 1.2166666666666668,
-      "grad_norm": 8.017053604125977,
+      "grad_norm": 5.203872203826904,
       "learning_rate": 2.962962962962963e-05,
-      "loss": 0.4951,
+      "loss": 0.4333,
       "step": 140
     },
     {
       "epoch": 1.25,
-      "grad_norm": 3.5494937896728516,
+      "grad_norm": 6.432222366333008,
       "learning_rate": 2.777777777777778e-05,
-      "loss": 0.9476,
+      "loss": 0.7871,
       "step": 150
     },
     {
       "epoch": 1.25,
-      "eval_accuracy": 0.8428571428571429,
-      "eval_loss": 0.5081102252006531,
-      "eval_runtime": 16.5404,
-      "eval_samples_per_second": 4.232,
-      "eval_steps_per_second": 1.088,
+      "eval_accuracy": 0.7857142857142857,
+      "eval_loss": 0.5381054282188416,
+      "eval_runtime": 16.23,
+      "eval_samples_per_second": 4.313,
+      "eval_steps_per_second": 1.109,
       "step": 150
     },
     {
       "epoch": 2.033333333333333,
-      "grad_norm": 11.868717193603516,
+      "grad_norm": 5.675662517547607,
       "learning_rate": 2.5925925925925925e-05,
-      "loss": 0.2692,
+      "loss": 0.3323,
       "step": 160
     },
     {
       "epoch": 2.066666666666667,
-      "grad_norm": 11.16838550567627,
+      "grad_norm": 29.48726463317871,
       "learning_rate": 2.4074074074074074e-05,
-      "loss": 0.4435,
+      "loss": 0.4221,
       "step": 170
     },
     {
       "epoch": 2.1,
-      "grad_norm": 4.307368755340576,
+      "grad_norm": 5.2884907722473145,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.3437,
+      "loss": 0.3494,
       "step": 180
     },
     {
       "epoch": 2.1333333333333333,
-      "grad_norm": 14.703262329101562,
+      "grad_norm": 6.56082820892334,
       "learning_rate": 2.037037037037037e-05,
-      "loss": 0.4501,
+      "loss": 0.3741,
       "step": 190
     },
     {
       "epoch": 2.1666666666666665,
-      "grad_norm": 3.9073095321655273,
+      "grad_norm": 3.4989917278289795,
       "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.3173,
+      "loss": 0.3709,
       "step": 200
     },
     {
       "epoch": 2.2,
-      "grad_norm": 7.474828720092773,
+      "grad_norm": 2.7712182998657227,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.1711,
+      "loss": 0.3153,
       "step": 210
     },
     {
       "epoch": 2.2333333333333334,
-      "grad_norm": 1.4244682788848877,
+      "grad_norm": 0.7842445373535156,
       "learning_rate": 1.4814814814814815e-05,
-      "loss": 0.3926,
+      "loss": 0.2923,
       "step": 220
     },
     {
       "epoch": 2.25,
-      "eval_accuracy": 0.7714285714285715,
-      "eval_loss": 0.38518500328063965,
-      "eval_runtime": 16.889,
-      "eval_samples_per_second": 4.145,
-      "eval_steps_per_second": 1.066,
+      "eval_accuracy": 0.7571428571428571,
+      "eval_loss": 0.46805742383003235,
+      "eval_runtime": 16.1856,
+      "eval_samples_per_second": 4.325,
+      "eval_steps_per_second": 1.112,
       "step": 225
     },
     {
       "epoch": 3.0166666666666666,
-      "grad_norm": 1.0246777534484863,
+      "grad_norm": 1.3947151899337769,
       "learning_rate": 1.2962962962962962e-05,
-      "loss": 0.1841,
+      "loss": 0.1865,
       "step": 230
     },
     {
       "epoch": 3.05,
-      "grad_norm": 0.22415301203727722,
+      "grad_norm": 0.3738820552825928,
       "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.12,
+      "loss": 0.1203,
       "step": 240
     },
     {
       "epoch": 3.0833333333333335,
-      "grad_norm": 7.395457744598389,
+      "grad_norm": 11.37885570526123,
       "learning_rate": 9.259259259259259e-06,
-      "loss": 0.1242,
+      "loss": 0.1452,
       "step": 250
     },
     {
       "epoch": 3.1166666666666667,
-      "grad_norm": 0.42695850133895874,
+      "grad_norm": 0.6158033013343811,
       "learning_rate": 7.4074074074074075e-06,
-      "loss": 0.1291,
+      "loss": 0.168,
       "step": 260
     },
     {
       "epoch": 3.15,
-      "grad_norm": 2.703252077102661,
+      "grad_norm": 2.308946371078491,
       "learning_rate": 5.555555555555556e-06,
-      "loss": 0.1147,
+      "loss": 0.1501,
       "step": 270
     },
     {
       "epoch": 3.183333333333333,
-      "grad_norm": 0.6641141772270203,
+      "grad_norm": 1.3887075185775757,
       "learning_rate": 3.7037037037037037e-06,
-      "loss": 0.2027,
+      "loss": 0.1935,
       "step": 280
     },
     {
       "epoch": 3.216666666666667,
-      "grad_norm": 1.9637229442596436,
+      "grad_norm": 1.384949803352356,
       "learning_rate": 1.8518518518518519e-06,
-      "loss": 0.2594,
+      "loss": 0.1793,
       "step": 290
     },
     {
       "epoch": 3.25,
-      "grad_norm": 0.8269719481468201,
+      "grad_norm": 0.7878948450088501,
       "learning_rate": 0.0,
-      "loss": 0.2059,
+      "loss": 0.2033,
       "step": 300
     },
     {
       "epoch": 3.25,
-      "eval_accuracy": 0.8428571428571429,
-      "eval_loss": 0.39890938997268677,
-      "eval_runtime": 16.6405,
-      "eval_samples_per_second": 4.207,
-      "eval_steps_per_second": 1.082,
+      "eval_accuracy": 0.9,
+      "eval_loss": 0.2786923944950104,
+      "eval_runtime": 16.2159,
+      "eval_samples_per_second": 4.317,
+      "eval_steps_per_second": 1.11,
       "step": 300
     },
     {
       "epoch": 3.25,
       "step": 300,
       "total_flos": 1.495384188125184e+18,
-      "train_loss": 0.8531251009305318,
-      "train_runtime": 689.5393,
-      "train_samples_per_second": 1.74,
-      "train_steps_per_second": 0.435
+      "train_loss": 0.8299423436323802,
+      "train_runtime": 668.5558,
+      "train_samples_per_second": 1.795,
+      "train_steps_per_second": 0.449
     },
     {
       "epoch": 3.25,
-      "eval_accuracy": 0.7677419354838709,
-      "eval_loss": 0.5838326811790466,
-      "eval_runtime": 31.2901,
-      "eval_samples_per_second": 4.954,
-      "eval_steps_per_second": 1.246,
+      "eval_accuracy": 0.9225806451612903,
+      "eval_loss": 0.30445483326911926,
+      "eval_runtime": 30.85,
+      "eval_samples_per_second": 5.024,
+      "eval_steps_per_second": 1.264,
       "step": 300
     },
     {
       "epoch": 3.25,
-      "eval_accuracy": 0.7677419354838709,
-      "eval_loss": 0.5838326811790466,
-      "eval_runtime": 30.1878,
-      "eval_samples_per_second": 5.135,
-      "eval_steps_per_second": 1.292,
+      "eval_accuracy": 0.9225806451612903,
+      "eval_loss": 0.30445486307144165,
+      "eval_runtime": 29.6038,
+      "eval_samples_per_second": 5.236,
+      "eval_steps_per_second": 1.317,
       "step": 300
     }
   ],