🍻 cheers

README.md

@@ -2,6 +2,7 @@
 license: apache-2.0
 base_model: google/vit-base-patch16-224
 tags:
+- image-classification
 - generated_from_trainer
 metrics:
 - accuracy
@@ -18,13 +19,13 @@ should probably proofread and complete it, then remove this comment. -->
 
 # vit-epsilon-5e-9
 
-This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on an unknown dataset.
+This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the skin-cancer dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6199
-- Accuracy: 0.8582
-- Precision: 0.8561
-- Recall: 0.8582
-- F1: 0.8541
+- Loss: 0.4961
+- Accuracy: 0.8252
+- Precision: 0.8358
+- Recall: 0.8252
+- F1: 0.8286
 
 ## Model description
 

all_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 21.0,
+    "eval_accuracy": 0.8252427184466019,
+    "eval_f1": 0.8286042614270734,
+    "eval_loss": 0.49607226252555847,
+    "eval_precision": 0.8357667938878295,
+    "eval_recall": 0.8252427184466019,
+    "eval_runtime": 23.1247,
+    "eval_samples_per_second": 124.715,
+    "eval_steps_per_second": 15.611,
+    "total_flos": 8.345331343683404e+18,
+    "train_loss": 0.6726648568081371,
+    "train_runtime": 2517.1784,
+    "train_samples_per_second": 203.72,
+    "train_steps_per_second": 12.752
+}

eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 21.0,
+    "eval_accuracy": 0.8252427184466019,
+    "eval_f1": 0.8286042614270734,
+    "eval_loss": 0.49607226252555847,
+    "eval_precision": 0.8357667938878295,
+    "eval_recall": 0.8252427184466019,
+    "eval_runtime": 23.1247,
+    "eval_samples_per_second": 124.715,
+    "eval_steps_per_second": 15.611
+}

runs/Mar22_10-26-56_80a90583e6d6/events.out.tfevents.1711105788.80a90583e6d6.2864.9

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a6a6317da93a0b7bb55beebbae214d136c93b795cf7e8c98b49ad74af34bc93
+size 560

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 21.0,
+    "total_flos": 8.345331343683404e+18,
+    "train_loss": 0.6726648568081371,
+    "train_runtime": 2517.1784,
+    "train_samples_per_second": 203.72,
+    "train_steps_per_second": 12.752
+}

trainer_state.json

@@ -0,0 +1,429 @@
+{
+  "best_metric": 0.49607226252555847,
+  "best_model_checkpoint": "./vit-epsilon-5e-9/checkpoint-3531",
+  "epoch": 21.0,
+  "eval_steps": 500,
+  "global_step": 6741,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "grad_norm": 26.712631225585938,
+      "learning_rate": 1.8291979226774382e-05,
+      "loss": 1.7673,
+      "step": 321
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6889736477115118,
+      "eval_f1": 0.6301904633567856,
+      "eval_loss": 0.9545517563819885,
+      "eval_precision": 0.6434923013126813,
+      "eval_recall": 0.6889736477115118,
+      "eval_runtime": 23.0294,
+      "eval_samples_per_second": 125.231,
+      "eval_steps_per_second": 15.676,
+      "step": 321
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 24.46586036682129,
+      "learning_rate": 3.681477207155222e-05,
+      "loss": 1.1951,
+      "step": 642
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7437586685159501,
+      "eval_f1": 0.7198853448190381,
+      "eval_loss": 0.7243568301200867,
+      "eval_precision": 0.7325414111372327,
+      "eval_recall": 0.7437586685159501,
+      "eval_runtime": 22.9972,
+      "eval_samples_per_second": 125.406,
+      "eval_steps_per_second": 15.698,
+      "step": 642
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 3.9905879497528076,
+      "learning_rate": 5.5337564916330066e-05,
+      "loss": 1.0711,
+      "step": 963
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7552011095700416,
+      "eval_f1": 0.7223710830172695,
+      "eval_loss": 0.6499444246292114,
+      "eval_precision": 0.7393727177504272,
+      "eval_recall": 0.7552011095700416,
+      "eval_runtime": 22.9477,
+      "eval_samples_per_second": 125.677,
+      "eval_steps_per_second": 15.731,
+      "step": 963
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 27.717905044555664,
+      "learning_rate": 7.386035776110792e-05,
+      "loss": 0.9944,
+      "step": 1284
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7590152565880721,
+      "eval_f1": 0.7682688045967295,
+      "eval_loss": 0.590674638748169,
+      "eval_precision": 0.7962204668682246,
+      "eval_recall": 0.7590152565880721,
+      "eval_runtime": 22.8277,
+      "eval_samples_per_second": 126.338,
+      "eval_steps_per_second": 15.814,
+      "step": 1284
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 7.2339677810668945,
+      "learning_rate": 9.238315060588575e-05,
+      "loss": 0.9231,
+      "step": 1605
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7083911234396671,
+      "eval_f1": 0.7305944258102485,
+      "eval_loss": 0.6987636685371399,
+      "eval_precision": 0.8053894786564816,
+      "eval_recall": 0.7083911234396671,
+      "eval_runtime": 23.2435,
+      "eval_samples_per_second": 124.078,
+      "eval_steps_per_second": 15.531,
+      "step": 1605
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 6.594783306121826,
+      "learning_rate": 9.984715255878176e-05,
+      "loss": 0.9889,
+      "step": 1926
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.7746185852981969,
+      "eval_f1": 0.7808322063026315,
+      "eval_loss": 0.5912254452705383,
+      "eval_precision": 0.7919129111414196,
+      "eval_recall": 0.7746185852981969,
+      "eval_runtime": 22.6534,
+      "eval_samples_per_second": 127.31,
+      "eval_steps_per_second": 15.936,
+      "step": 1926
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 17.52560043334961,
+      "learning_rate": 9.889061131437471e-05,
+      "loss": 0.8818,
+      "step": 2247
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.7569348127600555,
+      "eval_f1": 0.7696667824220995,
+      "eval_loss": 0.6373791098594666,
+      "eval_precision": 0.8000628884069328,
+      "eval_recall": 0.7569348127600555,
+      "eval_runtime": 22.8375,
+      "eval_samples_per_second": 126.283,
+      "eval_steps_per_second": 15.807,
+      "step": 2247
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 3.2242636680603027,
+      "learning_rate": 9.707265436104638e-05,
+      "loss": 0.7973,
+      "step": 2568
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.7579750346740638,
+      "eval_f1": 0.7684127710230544,
+      "eval_loss": 0.6658095717430115,
+      "eval_precision": 0.7925317692672306,
+      "eval_recall": 0.7579750346740638,
+      "eval_runtime": 23.0108,
+      "eval_samples_per_second": 125.332,
+      "eval_steps_per_second": 15.688,
+      "step": 2568
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 5.601225852966309,
+      "learning_rate": 9.44253127296151e-05,
+      "loss": 0.7525,
+      "step": 2889
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.8044382801664355,
+      "eval_f1": 0.8068084072735682,
+      "eval_loss": 0.5219517350196838,
+      "eval_precision": 0.8124082874354884,
+      "eval_recall": 0.8044382801664355,
+      "eval_runtime": 22.9081,
+      "eval_samples_per_second": 125.895,
+      "eval_steps_per_second": 15.759,
+      "step": 2889
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 8.540093421936035,
+      "learning_rate": 9.099523058358976e-05,
+      "loss": 0.6938,
+      "step": 3210
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.789875173370319,
+      "eval_f1": 0.7980316905301443,
+      "eval_loss": 0.5633859038352966,
+      "eval_precision": 0.8334903184113343,
+      "eval_recall": 0.789875173370319,
+      "eval_runtime": 22.9512,
+      "eval_samples_per_second": 125.658,
+      "eval_steps_per_second": 15.729,
+      "step": 3210
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 24.208213806152344,
+      "learning_rate": 8.685682824178951e-05,
+      "loss": 0.6354,
+      "step": 3531
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.8252427184466019,
+      "eval_f1": 0.8286042614270734,
+      "eval_loss": 0.49607226252555847,
+      "eval_precision": 0.8357667938878295,
+      "eval_recall": 0.8252427184466019,
+      "eval_runtime": 22.5858,
+      "eval_samples_per_second": 127.691,
+      "eval_steps_per_second": 15.983,
+      "step": 3531
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 43.178890228271484,
+      "learning_rate": 8.205719438083829e-05,
+      "loss": 0.5602,
+      "step": 3852
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.8141470180305131,
+      "eval_f1": 0.818467222844118,
+      "eval_loss": 0.5485954284667969,
+      "eval_precision": 0.8275933063656342,
+      "eval_recall": 0.8141470180305131,
+      "eval_runtime": 22.5135,
+      "eval_samples_per_second": 128.101,
+      "eval_steps_per_second": 16.035,
+      "step": 3852
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 32.53014373779297,
+      "learning_rate": 7.669273692531118e-05,
+      "loss": 0.44,
+      "step": 4173
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.8141470180305131,
+      "eval_f1": 0.8155366968674153,
+      "eval_loss": 0.6554011702537537,
+      "eval_precision": 0.8442196675228097,
+      "eval_recall": 0.8141470180305131,
+      "eval_runtime": 22.8299,
+      "eval_samples_per_second": 126.326,
+      "eval_steps_per_second": 15.813,
+      "step": 4173
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 2.291996955871582,
+      "learning_rate": 7.085797357089247e-05,
+      "loss": 0.4704,
+      "step": 4494
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.823509015256588,
+      "eval_f1": 0.8287228429152137,
+      "eval_loss": 0.5703807473182678,
+      "eval_precision": 0.8431399994729947,
+      "eval_recall": 0.823509015256588,
+      "eval_runtime": 22.7856,
+      "eval_samples_per_second": 126.571,
+      "eval_steps_per_second": 15.843,
+      "step": 4494
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 1.2885832786560059,
+      "learning_rate": 6.46557084486047e-05,
+      "loss": 0.4275,
+      "step": 4815
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.8141470180305131,
+      "eval_f1": 0.8230327037475541,
+      "eval_loss": 0.5562968254089355,
+      "eval_precision": 0.845939062780455,
+      "eval_recall": 0.8141470180305131,
+      "eval_runtime": 22.7036,
+      "eval_samples_per_second": 127.028,
+      "eval_steps_per_second": 15.901,
+      "step": 4815
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 7.533414363861084,
+      "learning_rate": 5.8195220793532045e-05,
+      "loss": 0.3511,
+      "step": 5136
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.8072122052704577,
+      "eval_f1": 0.816559487569828,
+      "eval_loss": 0.5933049917221069,
+      "eval_precision": 0.8401813483694368,
+      "eval_recall": 0.8072122052704577,
+      "eval_runtime": 22.8717,
+      "eval_samples_per_second": 126.095,
+      "eval_steps_per_second": 15.784,
+      "step": 5136
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 3.8387444019317627,
+      "learning_rate": 5.15903395270923e-05,
+      "loss": 0.2853,
+      "step": 5457
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.8436199722607489,
+      "eval_f1": 0.8470414632738617,
+      "eval_loss": 0.5246312618255615,
+      "eval_precision": 0.8541559260812904,
+      "eval_recall": 0.8436199722607489,
+      "eval_runtime": 22.7545,
+      "eval_samples_per_second": 126.744,
+      "eval_steps_per_second": 15.865,
+      "step": 5457
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 14.802148818969727,
+      "learning_rate": 4.495743767726598e-05,
+      "loss": 0.2691,
+      "step": 5778
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.8509015256588072,
+      "eval_f1": 0.8519257296049367,
+      "eval_loss": 0.5257188677787781,
+      "eval_precision": 0.8550518836398507,
+      "eval_recall": 0.8509015256588072,
+      "eval_runtime": 22.5177,
+      "eval_samples_per_second": 128.077,
+      "eval_steps_per_second": 16.032,
+      "step": 5778
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 0.6039425730705261,
+      "learning_rate": 3.843351058509752e-05,
+      "loss": 0.2134,
+      "step": 6099
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.8332177531206657,
+      "eval_f1": 0.8404222882108485,
+      "eval_loss": 0.6391225457191467,
+      "eval_precision": 0.8553108175252582,
+      "eval_recall": 0.8332177531206657,
+      "eval_runtime": 22.8047,
+      "eval_samples_per_second": 126.465,
+      "eval_steps_per_second": 15.83,
+      "step": 6099
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.08735406398773193,
+      "learning_rate": 3.209279054309716e-05,
+      "loss": 0.224,
+      "step": 6420
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.8488210818307905,
+      "eval_f1": 0.8496776636057133,
+      "eval_loss": 0.6296781897544861,
+      "eval_precision": 0.8537272864205,
+      "eval_recall": 0.8488210818307905,
+      "eval_runtime": 22.8485,
+      "eval_samples_per_second": 126.223,
+      "eval_steps_per_second": 15.8,
+      "step": 6420
+    },
+    {
+      "epoch": 21.0,
+      "grad_norm": 11.383289337158203,
+      "learning_rate": 2.6067582030088143e-05,
+      "loss": 0.1843,
+      "step": 6741
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.8581830790568654,
+      "eval_f1": 0.8541053823981534,
+      "eval_loss": 0.6199486255645752,
+      "eval_precision": 0.8560897758057323,
+      "eval_recall": 0.8581830790568654,
+      "eval_runtime": 22.9703,
+      "eval_samples_per_second": 125.553,
+      "eval_steps_per_second": 15.716,
+      "step": 6741
+    },
+    {
+      "epoch": 21.0,
+      "step": 6741,
+      "total_flos": 8.345331343683404e+18,
+      "train_loss": 0.6726648568081371,
+      "train_runtime": 2517.1784,
+      "train_samples_per_second": 203.72,
+      "train_steps_per_second": 12.752
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 32100,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 100,
+  "save_steps": 500,
+  "total_flos": 8.345331343683404e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}