End of training

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.22972972972973,
-    "eval_accuracy": 0.8387096774193549,
-    "eval_loss": 0.409178227186203,
-    "eval_runtime": 47.0636,
-    "eval_samples_per_second": 3.293,
-    "eval_steps_per_second": 0.425
+    "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
 }

test_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.22972972972973,
-    "eval_accuracy": 0.8387096774193549,
-    "eval_loss": 0.409178227186203,
-    "eval_runtime": 47.0636,
-    "eval_samples_per_second": 3.293,
-    "eval_steps_per_second": 0.425
+    "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
 }

trainer_state.json

@@ -1,177 +1,338 @@
 {
-  "best_metric": 0.8857142857142857,
-  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-148",
-  "epoch": 3.22972972972973,
+  "best_metric": 0.7875,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-95",
+  "epoch": 3.242021276595745,
   "eval_steps": 500,
-  "global_step": 148,
+  "global_step": 376,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.06756756756756757,
-      "grad_norm": 6.447524070739746,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 2.3807,
+      "epoch": 0.026595744680851064,
+      "grad_norm": 12.626161575317383,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 1.6196,
       "step": 10
     },
     {
-      "epoch": 0.13513513513513514,
-      "grad_norm": 7.273045539855957,
-      "learning_rate": 4.81203007518797e-05,
-      "loss": 2.2314,
+      "epoch": 0.05319148936170213,
+      "grad_norm": 5.767611980438232,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.9827,
       "step": 20
     },
     {
-      "epoch": 0.20270270270270271,
-      "grad_norm": 8.625303268432617,
-      "learning_rate": 4.43609022556391e-05,
-      "loss": 2.1403,
+      "epoch": 0.0797872340425532,
+      "grad_norm": 4.556581497192383,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.9375,
       "step": 30
     },
     {
-      "epoch": 0.25675675675675674,
-      "eval_accuracy": 0.5142857142857142,
-      "eval_loss": 1.731579065322876,
-      "eval_runtime": 20.0887,
-      "eval_samples_per_second": 3.485,
-      "eval_steps_per_second": 0.448,
-      "step": 38
-    },
-    {
-      "epoch": 1.0135135135135136,
-      "grad_norm": 8.90733528137207,
-      "learning_rate": 4.0601503759398494e-05,
-      "loss": 1.968,
+      "epoch": 0.10638297872340426,
+      "grad_norm": 5.054460525512695,
+      "learning_rate": 4.970414201183432e-05,
+      "loss": 0.8402,
       "step": 40
     },
     {
-      "epoch": 1.0810810810810811,
-      "grad_norm": 10.461448669433594,
-      "learning_rate": 3.6842105263157895e-05,
-      "loss": 1.5209,
+      "epoch": 0.13297872340425532,
+      "grad_norm": 4.505584239959717,
+      "learning_rate": 4.822485207100592e-05,
+      "loss": 0.582,
       "step": 50
     },
     {
-      "epoch": 1.1486486486486487,
-      "grad_norm": 68.34122467041016,
-      "learning_rate": 3.3082706766917295e-05,
-      "loss": 1.2307,
+      "epoch": 0.1595744680851064,
+      "grad_norm": 3.063162088394165,
+      "learning_rate": 4.674556213017752e-05,
+      "loss": 1.1131,
       "step": 60
     },
     {
-      "epoch": 1.2162162162162162,
-      "grad_norm": 10.887399673461914,
-      "learning_rate": 2.9323308270676693e-05,
-      "loss": 0.8431,
+      "epoch": 0.18617021276595744,
+      "grad_norm": 2.564850330352783,
+      "learning_rate": 4.5266272189349114e-05,
+      "loss": 0.6761,
       "step": 70
     },
     {
-      "epoch": 1.2567567567567568,
-      "eval_accuracy": 0.7285714285714285,
-      "eval_loss": 0.8928426504135132,
-      "eval_runtime": 20.3734,
-      "eval_samples_per_second": 3.436,
-      "eval_steps_per_second": 0.442,
-      "step": 76
-    },
-    {
-      "epoch": 2.027027027027027,
-      "grad_norm": 5.018065452575684,
-      "learning_rate": 2.556390977443609e-05,
-      "loss": 0.8045,
+      "epoch": 0.2127659574468085,
+      "grad_norm": 2.2109172344207764,
+      "learning_rate": 4.378698224852072e-05,
+      "loss": 0.5614,
       "step": 80
     },
     {
-      "epoch": 2.0945945945945947,
-      "grad_norm": 4.78250789642334,
-      "learning_rate": 2.1804511278195487e-05,
-      "loss": 0.5899,
+      "epoch": 0.2393617021276596,
+      "grad_norm": 1.6165765523910522,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.8026,
       "step": 90
     },
     {
-      "epoch": 2.1621621621621623,
-      "grad_norm": 10.711965560913086,
-      "learning_rate": 1.8045112781954888e-05,
-      "loss": 0.5134,
-      "step": 100
+      "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,
+      "step": 95
     },
     {
-      "epoch": 2.22972972972973,
-      "grad_norm": 7.2737650871276855,
-      "learning_rate": 1.4285714285714285e-05,
-      "loss": 0.3876,
-      "step": 110
+      "epoch": 1.0132978723404256,
+      "grad_norm": 1.8220001459121704,
+      "learning_rate": 4.0828402366863904e-05,
+      "loss": 0.6855,
+      "step": 100
     },
     {
-      "epoch": 2.2567567567567566,
-      "eval_accuracy": 0.8285714285714286,
-      "eval_loss": 0.40636110305786133,
-      "eval_runtime": 19.4773,
-      "eval_samples_per_second": 3.594,
-      "eval_steps_per_second": 0.462,
-      "step": 114
+      "epoch": 1.0398936170212767,
+      "grad_norm": 2.690967321395874,
+      "learning_rate": 3.934911242603551e-05,
+      "loss": 0.8036,
+      "step": 110
     },
     {
-      "epoch": 3.0405405405405403,
-      "grad_norm": 2.563310146331787,
-      "learning_rate": 1.0526315789473684e-05,
-      "loss": 0.3354,
+      "epoch": 1.0664893617021276,
+      "grad_norm": 2.7666921615600586,
+      "learning_rate": 3.7869822485207104e-05,
+      "loss": 0.7408,
       "step": 120
     },
     {
-      "epoch": 3.108108108108108,
-      "grad_norm": 5.6177215576171875,
-      "learning_rate": 6.766917293233083e-06,
-      "loss": 0.2464,
+      "epoch": 1.0930851063829787,
+      "grad_norm": 2.5995383262634277,
+      "learning_rate": 3.63905325443787e-05,
+      "loss": 0.7017,
       "step": 130
     },
     {
-      "epoch": 3.175675675675676,
-      "grad_norm": 9.905891418457031,
-      "learning_rate": 3.007518796992481e-06,
-      "loss": 0.2913,
+      "epoch": 1.1196808510638299,
+      "grad_norm": 7.389265537261963,
+      "learning_rate": 3.49112426035503e-05,
+      "loss": 0.6631,
       "step": 140
     },
     {
-      "epoch": 3.22972972972973,
-      "eval_accuracy": 0.8857142857142857,
-      "eval_loss": 0.3383012115955353,
-      "eval_runtime": 20.5339,
-      "eval_samples_per_second": 3.409,
-      "eval_steps_per_second": 0.438,
-      "step": 148
-    },
-    {
-      "epoch": 3.22972972972973,
-      "step": 148,
-      "total_flos": 1.460491890402263e+18,
-      "train_loss": 1.0594123150851276,
-      "train_runtime": 875.9405,
-      "train_samples_per_second": 1.352,
-      "train_steps_per_second": 0.169
-    },
-    {
-      "epoch": 3.22972972972973,
-      "eval_accuracy": 0.8387096774193549,
-      "eval_loss": 0.409178227186203,
-      "eval_runtime": 47.1559,
-      "eval_samples_per_second": 3.287,
-      "eval_steps_per_second": 0.424,
-      "step": 148
-    },
-    {
-      "epoch": 3.22972972972973,
-      "eval_accuracy": 0.8387096774193549,
-      "eval_loss": 0.409178227186203,
-      "eval_runtime": 47.0636,
-      "eval_samples_per_second": 3.293,
-      "eval_steps_per_second": 0.425,
-      "step": 148
+      "epoch": 1.1462765957446808,
+      "grad_norm": 1.8834972381591797,
+      "learning_rate": 3.3431952662721895e-05,
+      "loss": 0.8175,
+      "step": 150
+    },
+    {
+      "epoch": 1.172872340425532,
+      "grad_norm": 1.4516174793243408,
+      "learning_rate": 3.195266272189349e-05,
+      "loss": 0.6888,
+      "step": 160
+    },
+    {
+      "epoch": 1.199468085106383,
+      "grad_norm": 3.942476272583008,
+      "learning_rate": 3.047337278106509e-05,
+      "loss": 0.7102,
+      "step": 170
+    },
+    {
+      "epoch": 1.226063829787234,
+      "grad_norm": 1.646263837814331,
+      "learning_rate": 2.8994082840236685e-05,
+      "loss": 0.7908,
+      "step": 180
+    },
+    {
+      "epoch": 1.252659574468085,
+      "grad_norm": 5.147498607635498,
+      "learning_rate": 2.751479289940829e-05,
+      "loss": 0.8728,
+      "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,
+      "step": 190
+    },
+    {
+      "epoch": 2.026595744680851,
+      "grad_norm": 1.3614023923873901,
+      "learning_rate": 2.6035502958579882e-05,
+      "loss": 0.7153,
+      "step": 200
+    },
+    {
+      "epoch": 2.0531914893617023,
+      "grad_norm": 1.6865636110305786,
+      "learning_rate": 2.4556213017751482e-05,
+      "loss": 0.7114,
+      "step": 210
+    },
+    {
+      "epoch": 2.0797872340425534,
+      "grad_norm": 2.2771339416503906,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.8898,
+      "step": 220
+    },
+    {
+      "epoch": 2.106382978723404,
+      "grad_norm": 2.0663363933563232,
+      "learning_rate": 2.1597633136094676e-05,
+      "loss": 0.8524,
+      "step": 230
+    },
+    {
+      "epoch": 2.132978723404255,
+      "grad_norm": 4.288294792175293,
+      "learning_rate": 2.0118343195266273e-05,
+      "loss": 0.6944,
+      "step": 240
+    },
+    {
+      "epoch": 2.1595744680851063,
+      "grad_norm": 3.9807941913604736,
+      "learning_rate": 1.8639053254437873e-05,
+      "loss": 0.7094,
+      "step": 250
+    },
+    {
+      "epoch": 2.1861702127659575,
+      "grad_norm": 3.6395652294158936,
+      "learning_rate": 1.7159763313609466e-05,
+      "loss": 0.7492,
+      "step": 260
+    },
+    {
+      "epoch": 2.2127659574468086,
+      "grad_norm": 4.759531021118164,
+      "learning_rate": 1.5680473372781066e-05,
+      "loss": 0.5475,
+      "step": 270
+    },
+    {
+      "epoch": 2.2393617021276597,
+      "grad_norm": 2.811213970184326,
+      "learning_rate": 1.4201183431952663e-05,
+      "loss": 0.952,
+      "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,
+      "step": 285
+    },
+    {
+      "epoch": 3.0132978723404253,
+      "grad_norm": 2.360816240310669,
+      "learning_rate": 1.2721893491124262e-05,
+      "loss": 0.6142,
+      "step": 290
+    },
+    {
+      "epoch": 3.0398936170212765,
+      "grad_norm": 3.2934021949768066,
+      "learning_rate": 1.1242603550295859e-05,
+      "loss": 0.6263,
+      "step": 300
+    },
+    {
+      "epoch": 3.0664893617021276,
+      "grad_norm": 2.3915889263153076,
+      "learning_rate": 9.763313609467455e-06,
+      "loss": 0.6514,
+      "step": 310
+    },
+    {
+      "epoch": 3.0930851063829787,
+      "grad_norm": 3.9150514602661133,
+      "learning_rate": 8.284023668639054e-06,
+      "loss": 0.8365,
+      "step": 320
+    },
+    {
+      "epoch": 3.11968085106383,
+      "grad_norm": 3.9853973388671875,
+      "learning_rate": 6.8047337278106515e-06,
+      "loss": 0.7772,
+      "step": 330
+    },
+    {
+      "epoch": 3.146276595744681,
+      "grad_norm": 1.9402360916137695,
+      "learning_rate": 5.325443786982249e-06,
+      "loss": 0.8747,
+      "step": 340
+    },
+    {
+      "epoch": 3.172872340425532,
+      "grad_norm": 3.0497424602508545,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.5998,
+      "step": 350
+    },
+    {
+      "epoch": 3.199468085106383,
+      "grad_norm": 4.310546875,
+      "learning_rate": 2.366863905325444e-06,
+      "loss": 0.8328,
+      "step": 360
+    },
+    {
+      "epoch": 3.226063829787234,
+      "grad_norm": 3.943880796432495,
+      "learning_rate": 8.875739644970415e-07,
+      "loss": 0.8287,
+      "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,
+      "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
+    },
+    {
+      "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,
+      "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,
+      "step": 376
     }
   ],
   "logging_steps": 10,
-  "max_steps": 148,
+  "max_steps": 376,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 9223372036854775807,
   "save_steps": 500,
@@ -187,7 +348,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.460491890402263e+18,
+  "total_flos": 3.7220613152994755e+18,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null