End of training

README.md

@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [facebook/dinov2-small-imagenet1k-1-layer](https://huggingface.co/facebook/dinov2-small-imagenet1k-1-layer) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4129
-- Accuracy: 0.8604
+- Loss: 0.3366
+- Accuracy: 0.8982
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 9.577777777777778,
-    "eval_accuracy": 0.9080459770114943,
-    "eval_loss": 0.37552881240844727,
-    "eval_runtime": 61.3305,
-    "eval_samples_per_second": 9.93,
-    "eval_steps_per_second": 0.636,
+    "eval_accuracy": 0.8981937602627258,
+    "eval_loss": 0.33657345175743103,
+    "eval_runtime": 68.6419,
+    "eval_samples_per_second": 8.872,
+    "eval_steps_per_second": 0.568,
     "total_flos": 2.7458744422511e+17,
-    "train_loss": 0.4097213279117237,
-    "train_runtime": 2363.2087,
-    "train_samples_per_second": 6.089,
-    "train_steps_per_second": 0.093
+    "train_loss": 0.37115615768866106,
+    "train_runtime": 2504.6096,
+    "train_samples_per_second": 5.745,
+    "train_steps_per_second": 0.088
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 9.577777777777778,
-    "eval_accuracy": 0.9080459770114943,
-    "eval_loss": 0.37552881240844727,
-    "eval_runtime": 61.3305,
-    "eval_samples_per_second": 9.93,
-    "eval_steps_per_second": 0.636
+    "eval_accuracy": 0.8981937602627258,
+    "eval_loss": 0.33657345175743103,
+    "eval_runtime": 68.6419,
+    "eval_samples_per_second": 8.872,
+    "eval_steps_per_second": 0.568
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 9.577777777777778,
     "total_flos": 2.7458744422511e+17,
-    "train_loss": 0.4097213279117237,
-    "train_runtime": 2363.2087,
-    "train_samples_per_second": 6.089,
-    "train_steps_per_second": 0.093
+    "train_loss": 0.37115615768866106,
+    "train_runtime": 2504.6096,
+    "train_samples_per_second": 5.745,
+    "train_steps_per_second": 0.088
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.9080459770114943,
+  "best_metric": 0.8981937602627258,
   "best_model_checkpoint": "dinov2-small-imagenet1k-1-layer-finetuned-noh\\checkpoint-46",
   "epoch": 9.577777777777778,
   "eval_steps": 500,
@@ -10,256 +10,256 @@
   "log_history": [
     {
       "epoch": 0.4444444444444444,
-      "grad_norm": 467.71881103515625,
+      "grad_norm": 349.2164306640625,
       "learning_rate": 2.272727272727273e-05,
-      "loss": 0.7354,
+      "loss": 0.611,
       "step": 10
     },
     {
       "epoch": 0.8888888888888888,
-      "grad_norm": 400.1885986328125,
+      "grad_norm": 174.1605682373047,
       "learning_rate": 4.545454545454546e-05,
-      "loss": 0.6331,
+      "loss": 0.4924,
       "step": 20
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7651888341543513,
-      "eval_loss": 0.5416210889816284,
-      "eval_runtime": 69.3532,
-      "eval_samples_per_second": 8.781,
-      "eval_steps_per_second": 0.562,
+      "eval_accuracy": 0.8325123152709359,
+      "eval_loss": 0.5212485194206238,
+      "eval_runtime": 65.7326,
+      "eval_samples_per_second": 9.265,
+      "eval_steps_per_second": 0.593,
       "step": 23
     },
     {
       "epoch": 1.3111111111111111,
-      "grad_norm": 38.50715255737305,
+      "grad_norm": 74.73553466796875,
       "learning_rate": 4.797979797979798e-05,
-      "loss": 0.5552,
+      "loss": 0.5101,
       "step": 30
     },
     {
       "epoch": 1.7555555555555555,
-      "grad_norm": 288.6549072265625,
+      "grad_norm": 69.5820083618164,
       "learning_rate": 4.545454545454546e-05,
-      "loss": 0.4913,
+      "loss": 0.5732,
       "step": 40
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.9080459770114943,
-      "eval_loss": 0.37552881240844727,
-      "eval_runtime": 62.5531,
-      "eval_samples_per_second": 9.736,
-      "eval_steps_per_second": 0.623,
+      "eval_accuracy": 0.8981937602627258,
+      "eval_loss": 0.33657345175743103,
+      "eval_runtime": 69.7181,
+      "eval_samples_per_second": 8.735,
+      "eval_steps_per_second": 0.559,
       "step": 46
     },
     {
       "epoch": 2.1777777777777776,
-      "grad_norm": 73.34674835205078,
+      "grad_norm": 36.10708999633789,
       "learning_rate": 4.292929292929293e-05,
-      "loss": 0.4369,
+      "loss": 0.3947,
       "step": 50
     },
     {
       "epoch": 2.6222222222222222,
-      "grad_norm": 61.16302490234375,
+      "grad_norm": 220.30311584472656,
       "learning_rate": 4.0404040404040405e-05,
-      "loss": 0.4642,
+      "loss": 0.5639,
       "step": 60
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.6141215106732348,
-      "eval_loss": 0.7140535116195679,
-      "eval_runtime": 68.2847,
-      "eval_samples_per_second": 8.919,
-      "eval_steps_per_second": 0.571,
+      "eval_accuracy": 0.8489326765188834,
+      "eval_loss": 0.3906978666782379,
+      "eval_runtime": 70.3418,
+      "eval_samples_per_second": 8.658,
+      "eval_steps_per_second": 0.554,
       "step": 69
     },
     {
       "epoch": 3.0444444444444443,
-      "grad_norm": 245.7981414794922,
+      "grad_norm": 58.00751495361328,
       "learning_rate": 3.787878787878788e-05,
-      "loss": 0.4971,
+      "loss": 0.4467,
       "step": 70
     },
     {
       "epoch": 3.488888888888889,
-      "grad_norm": 59.85107421875,
+      "grad_norm": 61.74748992919922,
       "learning_rate": 3.535353535353535e-05,
-      "loss": 0.4248,
+      "loss": 0.521,
       "step": 80
     },
     {
       "epoch": 3.9333333333333336,
-      "grad_norm": 26.801841735839844,
+      "grad_norm": 52.76782989501953,
       "learning_rate": 3.282828282828283e-05,
-      "loss": 0.4451,
+      "loss": 0.4759,
       "step": 90
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8045977011494253,
-      "eval_loss": 0.4347776770591736,
-      "eval_runtime": 63.7065,
-      "eval_samples_per_second": 9.559,
-      "eval_steps_per_second": 0.612,
+      "eval_accuracy": 0.8817733990147784,
+      "eval_loss": 0.3481961786746979,
+      "eval_runtime": 69.182,
+      "eval_samples_per_second": 8.803,
+      "eval_steps_per_second": 0.564,
       "step": 92
     },
     {
       "epoch": 4.355555555555555,
-      "grad_norm": 125.90353393554688,
+      "grad_norm": 83.23902893066406,
       "learning_rate": 3.0303030303030306e-05,
-      "loss": 0.3966,
+      "loss": 0.3662,
       "step": 100
     },
     {
       "epoch": 4.8,
-      "grad_norm": 27.96786880493164,
+      "grad_norm": 40.003013610839844,
       "learning_rate": 2.777777777777778e-05,
-      "loss": 0.4095,
+      "loss": 0.3757,
       "step": 110
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.8029556650246306,
-      "eval_loss": 0.5060437917709351,
-      "eval_runtime": 66.7998,
-      "eval_samples_per_second": 9.117,
-      "eval_steps_per_second": 0.584,
+      "eval_accuracy": 0.8275862068965517,
+      "eval_loss": 0.392085462808609,
+      "eval_runtime": 67.7229,
+      "eval_samples_per_second": 8.993,
+      "eval_steps_per_second": 0.576,
       "step": 115
     },
     {
       "epoch": 5.222222222222222,
-      "grad_norm": 28.08486557006836,
+      "grad_norm": 54.831241607666016,
       "learning_rate": 2.5252525252525256e-05,
-      "loss": 0.3595,
+      "loss": 0.3124,
       "step": 120
     },
     {
       "epoch": 5.666666666666667,
-      "grad_norm": 44.42402267456055,
+      "grad_norm": 45.63344192504883,
       "learning_rate": 2.272727272727273e-05,
-      "loss": 0.3399,
+      "loss": 0.3356,
       "step": 130
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.7372742200328407,
-      "eval_loss": 0.546423614025116,
-      "eval_runtime": 63.3896,
-      "eval_samples_per_second": 9.607,
-      "eval_steps_per_second": 0.615,
+      "eval_accuracy": 0.896551724137931,
+      "eval_loss": 0.3183760344982147,
+      "eval_runtime": 67.9276,
+      "eval_samples_per_second": 8.965,
+      "eval_steps_per_second": 0.574,
       "step": 138
     },
     {
       "epoch": 6.088888888888889,
-      "grad_norm": 93.54395294189453,
+      "grad_norm": 52.133689880371094,
       "learning_rate": 2.0202020202020203e-05,
-      "loss": 0.333,
+      "loss": 0.2883,
       "step": 140
     },
     {
       "epoch": 6.533333333333333,
-      "grad_norm": 56.091121673583984,
+      "grad_norm": 35.971134185791016,
       "learning_rate": 1.7676767676767676e-05,
-      "loss": 0.434,
+      "loss": 0.3177,
       "step": 150
     },
     {
       "epoch": 6.977777777777778,
-      "grad_norm": 24.830839157104492,
+      "grad_norm": 57.66413879394531,
       "learning_rate": 1.5151515151515153e-05,
-      "loss": 0.3304,
+      "loss": 0.2521,
       "step": 160
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.8883415435139573,
-      "eval_loss": 0.32744449377059937,
-      "eval_runtime": 63.7561,
-      "eval_samples_per_second": 9.552,
-      "eval_steps_per_second": 0.612,
+      "eval_accuracy": 0.8571428571428571,
+      "eval_loss": 0.39921438694000244,
+      "eval_runtime": 65.0226,
+      "eval_samples_per_second": 9.366,
+      "eval_steps_per_second": 0.6,
       "step": 161
     },
     {
       "epoch": 7.4,
-      "grad_norm": 59.196083068847656,
+      "grad_norm": 74.5230484008789,
       "learning_rate": 1.2626262626262628e-05,
-      "loss": 0.2963,
+      "loss": 0.2193,
       "step": 170
     },
     {
       "epoch": 7.844444444444444,
-      "grad_norm": 22.917722702026367,
+      "grad_norm": 80.1707534790039,
       "learning_rate": 1.0101010101010101e-05,
-      "loss": 0.3539,
+      "loss": 0.2981,
       "step": 180
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.8604269293924466,
-      "eval_loss": 0.38926875591278076,
-      "eval_runtime": 63.7316,
-      "eval_samples_per_second": 9.556,
-      "eval_steps_per_second": 0.612,
+      "eval_accuracy": 0.8702791461412152,
+      "eval_loss": 0.3903738856315613,
+      "eval_runtime": 67.5642,
+      "eval_samples_per_second": 9.014,
+      "eval_steps_per_second": 0.577,
       "step": 184
     },
     {
       "epoch": 8.266666666666667,
-      "grad_norm": 87.52684783935547,
+      "grad_norm": 82.75043487548828,
       "learning_rate": 7.5757575757575764e-06,
-      "loss": 0.2673,
+      "loss": 0.1812,
       "step": 190
     },
     {
       "epoch": 8.71111111111111,
-      "grad_norm": 21.081283569335938,
+      "grad_norm": 54.377445220947266,
       "learning_rate": 5.050505050505051e-06,
-      "loss": 0.2849,
+      "loss": 0.2302,
       "step": 200
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.8637110016420362,
-      "eval_loss": 0.3758099675178528,
-      "eval_runtime": 62.4924,
-      "eval_samples_per_second": 9.745,
-      "eval_steps_per_second": 0.624,
+      "eval_accuracy": 0.8719211822660099,
+      "eval_loss": 0.3987477421760559,
+      "eval_runtime": 67.5557,
+      "eval_samples_per_second": 9.015,
+      "eval_steps_per_second": 0.577,
       "step": 207
     },
     {
       "epoch": 9.133333333333333,
-      "grad_norm": 63.325809478759766,
+      "grad_norm": 65.00289916992188,
       "learning_rate": 2.5252525252525253e-06,
-      "loss": 0.265,
+      "loss": 0.2021,
       "step": 210
     },
     {
       "epoch": 9.577777777777778,
-      "grad_norm": 38.6427116394043,
+      "grad_norm": 52.96971130371094,
       "learning_rate": 0.0,
-      "loss": 0.2605,
+      "loss": 0.1979,
       "step": 220
     },
     {
       "epoch": 9.577777777777778,
-      "eval_accuracy": 0.8489326765188834,
-      "eval_loss": 0.3968973755836487,
-      "eval_runtime": 63.9436,
-      "eval_samples_per_second": 9.524,
-      "eval_steps_per_second": 0.61,
+      "eval_accuracy": 0.8604269293924466,
+      "eval_loss": 0.4129030406475067,
+      "eval_runtime": 66.9132,
+      "eval_samples_per_second": 9.101,
+      "eval_steps_per_second": 0.583,
       "step": 220
     },
     {
       "epoch": 9.577777777777778,
       "step": 220,
       "total_flos": 2.7458744422511e+17,
-      "train_loss": 0.4097213279117237,
-      "train_runtime": 2363.2087,
-      "train_samples_per_second": 6.089,
-      "train_steps_per_second": 0.093
+      "train_loss": 0.37115615768866106,
+      "train_runtime": 2504.6096,
+      "train_samples_per_second": 5.745,
+      "train_steps_per_second": 0.088
     }
   ],
   "logging_steps": 10,