New better model

sonic/config.json

@@ -20,7 +20,7 @@
   "id2label": {
     "0": "Tails the fox",
     "1": "Amy Rose",
-    "10": "Shadow the hedgehog",
+    "10": "Sonic the hedgehog",
     "11": "Sally Acorn",
     "12": "Tangle the lemur",
     "13": "mobian",
@@ -30,7 +30,7 @@
     "17": "Tikal the echidna",
     "18": "Wave the swallow",
     "19": "Jet the hawk",
-    "2": "Rogue the bat",
+    "2": "Knuckles the echidna",
     "20": "Honey the cat",
     "21": "Surge the tenrec",
     "22": "Starline the platypus",
@@ -44,35 +44,35 @@
     "3": "Whisper the wolf",
     "30": "Storm the albatross",
     "31": "Maria Robotnik",
-    "4": "Knuckles the echidna",
-    "5": "Blaze the cat",
-    "6": "Silver the hedgehog",
-    "7": "Cream the rabbit",
-    "8": "Vanilla the rabbit",
-    "9": "Sonic the hedgehog"
+    "4": "Rogue the bat",
+    "5": "Silver the hedgehog",
+    "6": "Blaze the cat",
+    "7": "Vanilla the rabbit",
+    "8": "Cream the rabbit",
+    "9": "Shadow the hedgehog"
   },
   "image_size": 384,
   "initializer_range": 0.02,
   "label2id": {
     "Amy Rose": "1",
     "Big the cat": "15",
-    "Blaze the cat": "5",
+    "Blaze the cat": "6",
     "Bunnie Rabbot": "28",
     "Charmy the bee": "25",
     "Cosmo the seedrian": "14",
-    "Cream the rabbit": "7",
+    "Cream the rabbit": "8",
     "Espio the chameleon": "26",
     "Honey the cat": "20",
     "Jet the hawk": "19",
-    "Knuckles the echidna": "4",
+    "Knuckles the echidna": "2",
     "Maria Robotnik": "31",
     "Marine the Raccoon": "29",
     "Mighty the armadillo": "23",
-    "Rogue the bat": "2",
+    "Rogue the bat": "4",
     "Sally Acorn": "11",
-    "Shadow the hedgehog": "10",
-    "Silver the hedgehog": "6",
-    "Sonic the hedgehog": "9",
+    "Shadow the hedgehog": "9",
+    "Silver the hedgehog": "5",
+    "Sonic the hedgehog": "10",
     "Starline the platypus": "22",
     "Sticks the badger": "16",
     "Storm the albatross": "30",
@@ -80,7 +80,7 @@
     "Tails the fox": "0",
     "Tangle the lemur": "12",
     "Tikal the echidna": "17",
-    "Vanilla the rabbit": "8",
+    "Vanilla the rabbit": "7",
     "Vector the crocodile": "24",
     "Wave the swallow": "18",
     "Whisper the wolf": "3",

sonic/pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ec5e3d6a60bdf3f519adb879379016d6bd97bb046a8f5a18afaac0cbae52c570
+oid sha256:1f52490f57b1e76d2aa86890ff44ec63cde9b7581f02c8fffac5292887bf7565
 size 350515757

sonic/trainer_state.json

@@ -1,8 +1,8 @@
 {
-  "best_metric": 0.08468733727931976,
-  "best_model_checkpoint": "./SonicCharacterClassifier\\checkpoint-1600",
-  "epoch": 6.042296072507553,
-  "global_step": 2000,
+  "best_metric": 0.0789622887969017,
+  "best_model_checkpoint": "./SonicCharacterClassifier\\checkpoint-1800",
+  "epoch": 9.06344410876133,
+  "global_step": 3000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,177 +10,262 @@
     {
       "epoch": 0.6,
       "learning_rate": 4.6978851963746225e-05,
-      "loss": 0.2038,
+      "loss": 0.2032,
       "step": 200
     },
     {
       "epoch": 0.6,
-      "eval_accuracy": 0.9509220795568983,
-      "eval_loss": 0.1493668258190155,
-      "eval_precision": 0.4276334957981484,
-      "eval_recall": 0.34057603409646653,
-      "eval_runtime": 245.2223,
-      "eval_samples_per_second": 8.099,
-      "eval_steps_per_second": 1.015,
+      "eval_accuracy": 0.9495591939546599,
+      "eval_loss": 0.1497950255870819,
+      "eval_precision": 0.38202921623576536,
+      "eval_recall": 0.2866420054831203,
+      "eval_runtime": 264.6722,
+      "eval_samples_per_second": 7.5,
+      "eval_steps_per_second": 0.941,
       "step": 200
     },
     {
       "epoch": 1.21,
       "learning_rate": 4.395770392749245e-05,
-      "loss": 0.1299,
+      "loss": 0.1244,
       "step": 400
     },
     {
       "epoch": 1.21,
-      "eval_accuracy": 0.9626132930513596,
-      "eval_loss": 0.11390183866024017,
-      "eval_precision": 0.6821628829936983,
-      "eval_recall": 0.6035811496119172,
-      "eval_runtime": 245.5675,
-      "eval_samples_per_second": 8.087,
-      "eval_steps_per_second": 1.014,
+      "eval_accuracy": 0.9624212846347607,
+      "eval_loss": 0.1109040230512619,
+      "eval_precision": 0.6335885608933468,
+      "eval_recall": 0.5364579884489264,
+      "eval_runtime": 261.6322,
+      "eval_samples_per_second": 7.587,
+      "eval_steps_per_second": 0.952,
       "step": 400
     },
     {
       "epoch": 1.81,
       "learning_rate": 4.093655589123867e-05,
-      "loss": 0.0952,
+      "loss": 0.0902,
       "step": 600
     },
     {
       "epoch": 1.81,
-      "eval_accuracy": 0.9677901560926485,
-      "eval_loss": 0.10017536580562592,
-      "eval_precision": 0.7746363586620384,
-      "eval_recall": 0.71393109788423,
-      "eval_runtime": 246.2016,
-      "eval_samples_per_second": 8.067,
-      "eval_steps_per_second": 1.011,
+      "eval_accuracy": 0.9697890428211587,
+      "eval_loss": 0.09568421542644501,
+      "eval_precision": 0.7973267857987819,
+      "eval_recall": 0.7373828567737021,
+      "eval_runtime": 263.8635,
+      "eval_samples_per_second": 7.523,
+      "eval_steps_per_second": 0.944,
       "step": 600
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.79154078549849e-05,
-      "loss": 0.0681,
+      "loss": 0.0692,
       "step": 800
     },
     {
       "epoch": 2.42,
-      "eval_accuracy": 0.9708113041289023,
-      "eval_loss": 0.09041494876146317,
-      "eval_precision": 0.8065139068160218,
-      "eval_recall": 0.7231326368063123,
-      "eval_runtime": 245.4942,
-      "eval_samples_per_second": 8.09,
-      "eval_steps_per_second": 1.014,
+      "eval_accuracy": 0.9711901763224181,
+      "eval_loss": 0.08928817510604858,
+      "eval_precision": 0.8029630032715678,
+      "eval_recall": 0.7031733867707225,
+      "eval_runtime": 256.6147,
+      "eval_samples_per_second": 7.735,
+      "eval_steps_per_second": 0.97,
       "step": 800
     },
     {
       "epoch": 3.02,
       "learning_rate": 3.489425981873112e-05,
-      "loss": 0.0608,
+      "loss": 0.0536,
       "step": 1000
     },
     {
       "epoch": 3.02,
-      "eval_accuracy": 0.972904078549849,
-      "eval_loss": 0.08679082244634628,
-      "eval_precision": 0.8209949551458537,
-      "eval_recall": 0.7289373926706963,
-      "eval_runtime": 244.3239,
-      "eval_samples_per_second": 8.129,
-      "eval_steps_per_second": 1.019,
+      "eval_accuracy": 0.9734571788413098,
+      "eval_loss": 0.08212397247552872,
+      "eval_precision": 0.8281340329136301,
+      "eval_recall": 0.7454393318947714,
+      "eval_runtime": 262.8056,
+      "eval_samples_per_second": 7.553,
+      "eval_steps_per_second": 0.947,
       "step": 1000
     },
     {
       "epoch": 3.63,
       "learning_rate": 3.187311178247734e-05,
-      "loss": 0.0393,
+      "loss": 0.0383,
       "step": 1200
     },
     {
       "epoch": 3.63,
-      "eval_accuracy": 0.9730771651560927,
-      "eval_loss": 0.08753738552331924,
-      "eval_precision": 0.8360383490292852,
-      "eval_recall": 0.7486598211097333,
-      "eval_runtime": 248.2529,
-      "eval_samples_per_second": 8.0,
-      "eval_steps_per_second": 1.003,
+      "eval_accuracy": 0.9749685138539043,
+      "eval_loss": 0.07910189032554626,
+      "eval_precision": 0.8499502404462407,
+      "eval_recall": 0.7928208095506756,
+      "eval_runtime": 269.2715,
+      "eval_samples_per_second": 7.372,
+      "eval_steps_per_second": 0.925,
       "step": 1200
     },
     {
       "epoch": 4.23,
       "learning_rate": 2.8851963746223565e-05,
-      "loss": 0.0347,
+      "loss": 0.0335,
       "step": 1400
     },
     {
       "epoch": 4.23,
-      "eval_accuracy": 0.9733761329305136,
-      "eval_loss": 0.08492986112833023,
-      "eval_precision": 0.8388275873170129,
-      "eval_recall": 0.7801557843198007,
-      "eval_runtime": 248.4693,
-      "eval_samples_per_second": 7.993,
-      "eval_steps_per_second": 1.002,
+      "eval_accuracy": 0.9746851385390428,
+      "eval_loss": 0.08017772436141968,
+      "eval_precision": 0.8415176727834163,
+      "eval_recall": 0.7714390648131707,
+      "eval_runtime": 262.8468,
+      "eval_samples_per_second": 7.552,
+      "eval_steps_per_second": 0.947,
       "step": 1400
     },
     {
       "epoch": 4.83,
       "learning_rate": 2.583081570996979e-05,
-      "loss": 0.0256,
+      "loss": 0.0245,
       "step": 1600
     },
     {
       "epoch": 4.83,
-      "eval_accuracy": 0.9739583333333334,
-      "eval_loss": 0.08468733727931976,
-      "eval_precision": 0.8475558611132625,
-      "eval_recall": 0.788949676665331,
-      "eval_runtime": 248.7261,
-      "eval_samples_per_second": 7.985,
-      "eval_steps_per_second": 1.001,
+      "eval_accuracy": 0.9751259445843828,
+      "eval_loss": 0.08046023547649384,
+      "eval_precision": 0.8452890204657136,
+      "eval_recall": 0.7795725037362784,
+      "eval_runtime": 292.5617,
+      "eval_samples_per_second": 6.785,
+      "eval_steps_per_second": 0.851,
       "step": 1600
     },
     {
       "epoch": 5.44,
       "learning_rate": 2.2809667673716012e-05,
-      "loss": 0.0191,
+      "loss": 0.0199,
       "step": 1800
     },
     {
       "epoch": 5.44,
-      "eval_accuracy": 0.9742573011077543,
-      "eval_loss": 0.0882001668214798,
-      "eval_precision": 0.845859505500522,
-      "eval_recall": 0.7649921950675245,
-      "eval_runtime": 251.833,
-      "eval_samples_per_second": 7.886,
-      "eval_steps_per_second": 0.989,
+      "eval_accuracy": 0.975787153652393,
+      "eval_loss": 0.0789622887969017,
+      "eval_precision": 0.8522677350637055,
+      "eval_recall": 0.7916681676884358,
+      "eval_runtime": 269.8991,
+      "eval_samples_per_second": 7.355,
+      "eval_steps_per_second": 0.923,
       "step": 1800
     },
     {
       "epoch": 6.04,
       "learning_rate": 1.9788519637462235e-05,
-      "loss": 0.0154,
+      "loss": 0.0169,
       "step": 2000
     },
     {
       "epoch": 6.04,
-      "eval_accuracy": 0.9743674471299094,
-      "eval_loss": 0.08864131569862366,
-      "eval_precision": 0.8535254440088273,
-      "eval_recall": 0.7735381205368478,
-      "eval_runtime": 245.1302,
-      "eval_samples_per_second": 8.102,
-      "eval_steps_per_second": 1.016,
+      "eval_accuracy": 0.9749842569269521,
+      "eval_loss": 0.08285234868526459,
+      "eval_precision": 0.8439850500744712,
+      "eval_recall": 0.7729562503247884,
+      "eval_runtime": 265.7593,
+      "eval_samples_per_second": 7.469,
+      "eval_steps_per_second": 0.937,
       "step": 2000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.676737160120846e-05,
+      "loss": 0.0132,
+      "step": 2200
+    },
+    {
+      "epoch": 6.65,
+      "eval_accuracy": 0.9755982367758187,
+      "eval_loss": 0.08093218505382538,
+      "eval_precision": 0.8554658866593972,
+      "eval_recall": 0.7986969530652608,
+      "eval_runtime": 261.442,
+      "eval_samples_per_second": 7.593,
+      "eval_steps_per_second": 0.952,
+      "step": 2200
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3746223564954682e-05,
+      "loss": 0.0113,
+      "step": 2400
+    },
+    {
+      "epoch": 7.25,
+      "eval_accuracy": 0.9752676322418136,
+      "eval_loss": 0.08354648947715759,
+      "eval_precision": 0.8497939903151792,
+      "eval_recall": 0.7866075777488188,
+      "eval_runtime": 259.2176,
+      "eval_samples_per_second": 7.658,
+      "eval_steps_per_second": 0.961,
+      "step": 2400
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0725075528700906e-05,
+      "loss": 0.0102,
+      "step": 2600
+    },
+    {
+      "epoch": 7.85,
+      "eval_accuracy": 0.9757241813602016,
+      "eval_loss": 0.08276553452014923,
+      "eval_precision": 0.8552711059293521,
+      "eval_recall": 0.7972149485255522,
+      "eval_runtime": 259.024,
+      "eval_samples_per_second": 7.663,
+      "eval_steps_per_second": 0.961,
+      "step": 2600
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.70392749244713e-06,
+      "loss": 0.0089,
+      "step": 2800
+    },
+    {
+      "epoch": 8.46,
+      "eval_accuracy": 0.976227959697733,
+      "eval_loss": 0.08310715854167938,
+      "eval_precision": 0.8535510581843376,
+      "eval_recall": 0.7942174286631556,
+      "eval_runtime": 266.8436,
+      "eval_samples_per_second": 7.439,
+      "eval_steps_per_second": 0.933,
+      "step": 2800
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.682779456193353e-06,
+      "loss": 0.0077,
+      "step": 3000
+    },
+    {
+      "epoch": 9.06,
+      "eval_accuracy": 0.9758658690176323,
+      "eval_loss": 0.08518018573522568,
+      "eval_precision": 0.8593849643219924,
+      "eval_recall": 0.7900506494585933,
+      "eval_runtime": 260.5553,
+      "eval_samples_per_second": 7.618,
+      "eval_steps_per_second": 0.956,
+      "step": 3000
     }
   ],
   "max_steps": 3310,
   "num_train_epochs": 10,
-  "total_flos": 1.1154739870524506e+19,
+  "total_flos": 1.6730017245781033e+19,
   "trial_name": null,
   "trial_params": null
 }