Model save

README.md

@@ -18,9 +18,9 @@ This model is a fine-tuned version of [MCG-NJU/videomae-base](https://huggingfac
 It achieves the following results on the evaluation set:
 - eval_loss: 0.9552
 - eval_accuracy: 0.7059
-- eval_runtime: 7.648
-- eval_samples_per_second: 13.337
-- eval_steps_per_second: 3.4
+- eval_runtime: 7.5849
+- eval_samples_per_second: 13.448
+- eval_steps_per_second: 3.428
 - epoch: 2.2541
 - step: 186
 

all_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 3.2438524590163933,
-    "eval_accuracy": 0.7647058823529411,
-    "eval_loss": 0.8334438800811768,
+    "eval_accuracy": 0.7058823529411765,
+    "eval_loss": 0.9551951289176941,
     "eval_runtime": 7.7031,
     "eval_samples_per_second": 13.241,
     "eval_steps_per_second": 6.621

test_results.json

@@ -1,8 +1,4 @@
 {
-    "epoch": 3.2438524590163933,
-    "eval_accuracy": 0.7647058823529411,
-    "eval_loss": 0.8334438800811768,
-    "eval_runtime": 7.7031,
-    "eval_samples_per_second": 13.241,
-    "eval_steps_per_second": 6.621
+    "eval_accuracy": 0.7058823529411765,
+    "eval_loss": 0.9551951289176941
 }

trainer_state.json

@@ -1,415 +1,187 @@
 {
-  "best_metric": 0.7021276595744681,
-  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-488",
-  "epoch": 3.2438524590163933,
+  "best_metric": 0.5957446808510638,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-124",
+  "epoch": 2.2540983606557377,
   "eval_steps": 500,
-  "global_step": 488,
+  "global_step": 186,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.020491803278688523,
-      "grad_norm": 17.668354034423828,
-      "learning_rate": 1.0204081632653061e-05,
-      "loss": 0.5899,
+      "epoch": 0.040983606557377046,
+      "grad_norm": 14.630450248718262,
+      "learning_rate": 2e-05,
+      "loss": 1.9953,
       "step": 10
     },
     {
-      "epoch": 0.040983606557377046,
-      "grad_norm": 0.9543402194976807,
-      "learning_rate": 2.0408163265306123e-05,
-      "loss": 0.8724,
+      "epoch": 0.08196721311475409,
+      "grad_norm": 12.040234565734863,
+      "learning_rate": 4e-05,
+      "loss": 2.0122,
       "step": 20
     },
     {
-      "epoch": 0.06147540983606557,
-      "grad_norm": 32.570133209228516,
-      "learning_rate": 3.061224489795919e-05,
-      "loss": 0.8359,
+      "epoch": 0.12295081967213115,
+      "grad_norm": 10.06123161315918,
+      "learning_rate": 4.8858447488584476e-05,
+      "loss": 1.9064,
       "step": 30
     },
     {
-      "epoch": 0.08196721311475409,
-      "grad_norm": 18.297029495239258,
-      "learning_rate": 4.0816326530612245e-05,
-      "loss": 1.0188,
+      "epoch": 0.16393442622950818,
+      "grad_norm": 9.154711723327637,
+      "learning_rate": 4.657534246575342e-05,
+      "loss": 1.955,
       "step": 40
     },
     {
-      "epoch": 0.10245901639344263,
-      "grad_norm": 1.2347818613052368,
-      "learning_rate": 4.988610478359909e-05,
-      "loss": 1.1604,
+      "epoch": 0.20491803278688525,
+      "grad_norm": 17.53380012512207,
+      "learning_rate": 4.4292237442922375e-05,
+      "loss": 1.8111,
       "step": 50
     },
     {
-      "epoch": 0.12295081967213115,
-      "grad_norm": 14.211080551147461,
-      "learning_rate": 4.874715261958998e-05,
-      "loss": 0.8299,
+      "epoch": 0.2459016393442623,
+      "grad_norm": 11.089496612548828,
+      "learning_rate": 4.200913242009132e-05,
+      "loss": 1.8106,
       "step": 60
     },
     {
-      "epoch": 0.14344262295081966,
-      "grad_norm": 74.23954772949219,
-      "learning_rate": 4.760820045558087e-05,
-      "loss": 1.1473,
+      "epoch": 0.2540983606557377,
+      "eval_accuracy": 0.19148936170212766,
+      "eval_loss": 1.8704100847244263,
+      "eval_runtime": 3.5011,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 3.428,
+      "step": 62
+    },
+    {
+      "epoch": 1.0327868852459017,
+      "grad_norm": 13.336834907531738,
+      "learning_rate": 3.9726027397260274e-05,
+      "loss": 1.8751,
       "step": 70
     },
     {
-      "epoch": 0.16393442622950818,
-      "grad_norm": 14.923948287963867,
-      "learning_rate": 4.6469248291571756e-05,
-      "loss": 1.3967,
+      "epoch": 1.0737704918032787,
+      "grad_norm": 11.175029754638672,
+      "learning_rate": 3.744292237442922e-05,
+      "loss": 1.6123,
       "step": 80
     },
     {
-      "epoch": 0.18442622950819673,
-      "grad_norm": 26.641878128051758,
-      "learning_rate": 4.5330296127562645e-05,
-      "loss": 0.9765,
+      "epoch": 1.1147540983606556,
+      "grad_norm": 10.622461318969727,
+      "learning_rate": 3.5159817351598174e-05,
+      "loss": 1.8106,
       "step": 90
     },
     {
-      "epoch": 0.20491803278688525,
-      "grad_norm": 1.4160146713256836,
-      "learning_rate": 4.419134396355353e-05,
-      "loss": 0.8614,
+      "epoch": 1.1557377049180328,
+      "grad_norm": 9.679147720336914,
+      "learning_rate": 3.287671232876712e-05,
+      "loss": 1.567,
       "step": 100
     },
     {
-      "epoch": 0.22540983606557377,
-      "grad_norm": 0.6423073410987854,
-      "learning_rate": 4.305239179954442e-05,
-      "loss": 1.0474,
+      "epoch": 1.1967213114754098,
+      "grad_norm": 14.767767906188965,
+      "learning_rate": 3.059360730593607e-05,
+      "loss": 1.5578,
       "step": 110
     },
     {
-      "epoch": 0.2459016393442623,
-      "grad_norm": 12.656720161437988,
-      "learning_rate": 4.191343963553531e-05,
-      "loss": 1.4877,
+      "epoch": 1.2377049180327868,
+      "grad_norm": 18.987808227539062,
+      "learning_rate": 2.8310502283105023e-05,
+      "loss": 1.4825,
       "step": 120
     },
     {
-      "epoch": 0.2520491803278688,
-      "eval_accuracy": 0.48936170212765956,
-      "eval_loss": 1.489047884941101,
-      "eval_runtime": 3.5398,
-      "eval_samples_per_second": 13.278,
-      "eval_steps_per_second": 6.78,
-      "step": 123
+      "epoch": 1.2540983606557377,
+      "eval_accuracy": 0.5957446808510638,
+      "eval_loss": 1.370785117149353,
+      "eval_runtime": 3.3745,
+      "eval_samples_per_second": 13.928,
+      "eval_steps_per_second": 3.556,
+      "step": 124
     },
     {
-      "epoch": 1.014344262295082,
-      "grad_norm": 151.98411560058594,
-      "learning_rate": 4.07744874715262e-05,
-      "loss": 1.5679,
+      "epoch": 2.0245901639344264,
+      "grad_norm": 19.232662200927734,
+      "learning_rate": 2.6027397260273973e-05,
+      "loss": 1.196,
       "step": 130
     },
     {
-      "epoch": 1.0348360655737705,
-      "grad_norm": 104.5379638671875,
-      "learning_rate": 3.9635535307517086e-05,
-      "loss": 0.7741,
+      "epoch": 2.0655737704918034,
+      "grad_norm": 23.459028244018555,
+      "learning_rate": 2.3744292237442922e-05,
+      "loss": 1.0483,
       "step": 140
     },
     {
-      "epoch": 1.055327868852459,
-      "grad_norm": 11.863714218139648,
-      "learning_rate": 3.8496583143507975e-05,
-      "loss": 0.9627,
+      "epoch": 2.1065573770491803,
+      "grad_norm": 15.685453414916992,
+      "learning_rate": 2.1461187214611872e-05,
+      "loss": 1.0462,
       "step": 150
     },
     {
-      "epoch": 1.0758196721311475,
-      "grad_norm": 2.847776412963867,
-      "learning_rate": 3.735763097949886e-05,
-      "loss": 0.6829,
+      "epoch": 2.1475409836065573,
+      "grad_norm": 16.980884552001953,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 1.1794,
       "step": 160
     },
     {
-      "epoch": 1.096311475409836,
-      "grad_norm": 7.588573455810547,
-      "learning_rate": 3.621867881548975e-05,
-      "loss": 0.8791,
+      "epoch": 2.1885245901639343,
+      "grad_norm": 20.012128829956055,
+      "learning_rate": 1.689497716894977e-05,
+      "loss": 0.8242,
       "step": 170
     },
     {
-      "epoch": 1.1168032786885247,
-      "grad_norm": 39.9682502746582,
-      "learning_rate": 3.507972665148064e-05,
-      "loss": 1.5947,
+      "epoch": 2.2295081967213113,
+      "grad_norm": 12.054617881774902,
+      "learning_rate": 1.4611872146118721e-05,
+      "loss": 0.872,
       "step": 180
     },
     {
-      "epoch": 1.1372950819672132,
-      "grad_norm": 36.84199142456055,
-      "learning_rate": 3.394077448747153e-05,
-      "loss": 0.6469,
-      "step": 190
-    },
-    {
-      "epoch": 1.1577868852459017,
-      "grad_norm": 3.502577304840088,
-      "learning_rate": 3.2801822323462416e-05,
-      "loss": 0.6792,
-      "step": 200
-    },
-    {
-      "epoch": 1.1782786885245902,
-      "grad_norm": 48.51677703857422,
-      "learning_rate": 3.1662870159453304e-05,
-      "loss": 0.8278,
-      "step": 210
-    },
-    {
-      "epoch": 1.1987704918032787,
-      "grad_norm": 18.526588439941406,
-      "learning_rate": 3.052391799544419e-05,
-      "loss": 1.0883,
-      "step": 220
-    },
-    {
-      "epoch": 1.2192622950819672,
-      "grad_norm": 0.22596795856952667,
-      "learning_rate": 2.938496583143508e-05,
-      "loss": 1.5731,
-      "step": 230
-    },
-    {
-      "epoch": 1.2397540983606556,
-      "grad_norm": 43.027931213378906,
-      "learning_rate": 2.824601366742597e-05,
-      "loss": 1.0259,
-      "step": 240
-    },
-    {
-      "epoch": 1.2520491803278688,
-      "eval_accuracy": 0.6595744680851063,
-      "eval_loss": 1.1473325490951538,
-      "eval_runtime": 3.5233,
-      "eval_samples_per_second": 13.34,
-      "eval_steps_per_second": 6.812,
-      "step": 246
-    },
-    {
-      "epoch": 2.0081967213114753,
-      "grad_norm": 16.573293685913086,
-      "learning_rate": 2.7107061503416858e-05,
-      "loss": 0.8229,
-      "step": 250
-    },
-    {
-      "epoch": 2.028688524590164,
-      "grad_norm": 34.50025177001953,
-      "learning_rate": 2.596810933940775e-05,
-      "loss": 1.2741,
-      "step": 260
-    },
-    {
-      "epoch": 2.0491803278688523,
-      "grad_norm": 16.234542846679688,
-      "learning_rate": 2.4829157175398634e-05,
-      "loss": 0.8029,
-      "step": 270
-    },
-    {
-      "epoch": 2.069672131147541,
-      "grad_norm": 36.11598587036133,
-      "learning_rate": 2.3690205011389523e-05,
-      "loss": 0.4739,
-      "step": 280
-    },
-    {
-      "epoch": 2.0901639344262297,
-      "grad_norm": 53.0654182434082,
-      "learning_rate": 2.255125284738041e-05,
-      "loss": 0.9375,
-      "step": 290
-    },
-    {
-      "epoch": 2.110655737704918,
-      "grad_norm": 27.09005355834961,
-      "learning_rate": 2.14123006833713e-05,
-      "loss": 0.5147,
-      "step": 300
-    },
-    {
-      "epoch": 2.1311475409836067,
-      "grad_norm": 1.1381382942199707,
-      "learning_rate": 2.0273348519362188e-05,
-      "loss": 0.3669,
-      "step": 310
-    },
-    {
-      "epoch": 2.151639344262295,
-      "grad_norm": 33.40491485595703,
-      "learning_rate": 1.9134396355353076e-05,
-      "loss": 0.7359,
-      "step": 320
-    },
-    {
-      "epoch": 2.1721311475409837,
-      "grad_norm": 39.87897491455078,
-      "learning_rate": 1.7995444191343964e-05,
-      "loss": 0.8056,
-      "step": 330
-    },
-    {
-      "epoch": 2.192622950819672,
-      "grad_norm": 0.3967530131340027,
-      "learning_rate": 1.6856492027334853e-05,
-      "loss": 0.6106,
-      "step": 340
-    },
-    {
-      "epoch": 2.2131147540983607,
-      "grad_norm": 15.328410148620605,
-      "learning_rate": 1.571753986332574e-05,
-      "loss": 0.5864,
-      "step": 350
-    },
-    {
-      "epoch": 2.2336065573770494,
-      "grad_norm": 0.5930748581886292,
-      "learning_rate": 1.457858769931663e-05,
-      "loss": 0.3236,
-      "step": 360
-    },
-    {
-      "epoch": 2.2520491803278686,
-      "eval_accuracy": 0.6808510638297872,
-      "eval_loss": 1.138521432876587,
-      "eval_runtime": 3.5551,
-      "eval_samples_per_second": 13.221,
-      "eval_steps_per_second": 6.751,
-      "step": 369
-    },
-    {
-      "epoch": 3.002049180327869,
-      "grad_norm": 0.42666512727737427,
-      "learning_rate": 1.3439635535307518e-05,
-      "loss": 0.6331,
-      "step": 370
-    },
-    {
-      "epoch": 3.0225409836065573,
-      "grad_norm": 0.0817425474524498,
-      "learning_rate": 1.2300683371298406e-05,
-      "loss": 0.5621,
-      "step": 380
-    },
-    {
-      "epoch": 3.043032786885246,
-      "grad_norm": 0.2686825692653656,
-      "learning_rate": 1.1161731207289294e-05,
-      "loss": 0.2522,
-      "step": 390
-    },
-    {
-      "epoch": 3.0635245901639343,
-      "grad_norm": 0.08851628750562668,
-      "learning_rate": 1.0022779043280183e-05,
-      "loss": 0.7662,
-      "step": 400
-    },
-    {
-      "epoch": 3.084016393442623,
-      "grad_norm": 0.9150404334068298,
-      "learning_rate": 8.883826879271071e-06,
-      "loss": 1.204,
-      "step": 410
-    },
-    {
-      "epoch": 3.1045081967213113,
-      "grad_norm": 1.546443223953247,
-      "learning_rate": 7.74487471526196e-06,
-      "loss": 0.8539,
-      "step": 420
-    },
-    {
-      "epoch": 3.125,
-      "grad_norm": 5.276226997375488,
-      "learning_rate": 6.605922551252848e-06,
-      "loss": 0.2554,
-      "step": 430
-    },
-    {
-      "epoch": 3.1454918032786887,
-      "grad_norm": 0.19730453193187714,
-      "learning_rate": 5.466970387243736e-06,
-      "loss": 0.2582,
-      "step": 440
-    },
-    {
-      "epoch": 3.165983606557377,
-      "grad_norm": 2.347675085067749,
-      "learning_rate": 4.328018223234624e-06,
-      "loss": 0.4179,
-      "step": 450
-    },
-    {
-      "epoch": 3.1864754098360657,
-      "grad_norm": 0.2406870424747467,
-      "learning_rate": 3.189066059225513e-06,
-      "loss": 0.2018,
-      "step": 460
-    },
-    {
-      "epoch": 3.206967213114754,
-      "grad_norm": 0.8780645728111267,
-      "learning_rate": 2.050113895216401e-06,
-      "loss": 0.2272,
-      "step": 470
-    },
-    {
-      "epoch": 3.2274590163934427,
-      "grad_norm": 0.222975954413414,
-      "learning_rate": 9.111617312072893e-07,
-      "loss": 0.5371,
-      "step": 480
-    },
-    {
-      "epoch": 3.2438524590163933,
-      "eval_accuracy": 0.7021276595744681,
-      "eval_loss": 1.0504409074783325,
-      "eval_runtime": 4.132,
-      "eval_samples_per_second": 11.375,
-      "eval_steps_per_second": 5.808,
-      "step": 488
-    },
-    {
-      "epoch": 3.2438524590163933,
-      "step": 488,
-      "total_flos": 1.2124749082531922e+18,
-      "train_loss": 0.8016270340466108,
-      "train_runtime": 367.025,
-      "train_samples_per_second": 2.659,
-      "train_steps_per_second": 1.33
+      "epoch": 2.2540983606557377,
+      "eval_accuracy": 0.5319148936170213,
+      "eval_loss": 1.0101068019866943,
+      "eval_runtime": 3.379,
+      "eval_samples_per_second": 13.91,
+      "eval_steps_per_second": 3.551,
+      "step": 186
     },
     {
-      "epoch": 3.2438524590163933,
-      "eval_accuracy": 0.7647058823529411,
-      "eval_loss": 0.8334440588951111,
-      "eval_runtime": 7.8397,
-      "eval_samples_per_second": 13.011,
-      "eval_steps_per_second": 6.505,
-      "step": 488
+      "epoch": 2.2540983606557377,
+      "eval_accuracy": 0.7058823529411765,
+      "eval_loss": 0.9551948308944702,
+      "eval_runtime": 7.648,
+      "eval_samples_per_second": 13.337,
+      "eval_steps_per_second": 3.4,
+      "step": 186
     },
     {
-      "epoch": 3.2438524590163933,
-      "eval_accuracy": 0.7647058823529411,
-      "eval_loss": 0.8334438800811768,
-      "eval_runtime": 7.7031,
-      "eval_samples_per_second": 13.241,
-      "eval_steps_per_second": 6.621,
-      "step": 488
+      "epoch": 2.2540983606557377,
+      "eval_accuracy": 0.7058823529411765,
+      "eval_loss": 0.9551951289176941,
+      "eval_runtime": 7.5849,
+      "eval_samples_per_second": 13.448,
+      "eval_steps_per_second": 3.428,
+      "step": 186
     }
   ],
   "logging_steps": 10,
-  "max_steps": 488,
+  "max_steps": 244,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 9223372036854775807,
   "save_steps": 500,
@@ -420,13 +192,13 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": true
+        "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 1.2124749082531922e+18,
-  "train_batch_size": 2,
+  "total_flos": 9.158983119898214e+17,
+  "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null
 }