Add evaluation results and output files to model card

README.md

@@ -5,16 +5,19 @@ tags:
   - model
   - evaluation
 metrics:
-  - average_precision: 0.98
-  - roc_auc: 0.98
+  - average_precision: 0.69
+  - roc_auc: 0.80
+  - best threshold according to F1: 0.49
 ---
 
 # Binary Classification Model
 
 ## Evaluation Results
 
-**Average Precision:** 0.98  
-**ROC AUC:** 0.98
+**Average Precision:** 0.69  
+**ROC AUC:** 0.80
+**best threshold according to F1: 0.49
+
 
 ## Visualizations
 

checkpoint-60/config.json

@@ -0,0 +1,23 @@
+{
+  "_name_or_path": "ZivKassnerNK/binary_classification_model_v3.1.3_spines",
+  "architectures": [
+    "ViTForImageClassification"
+  ],
+  "attention_probs_dropout_prob": 0.0,
+  "encoder_stride": 16,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.0,
+  "hidden_size": 256,
+  "image_size": 32,
+  "initializer_range": 0.02,
+  "intermediate_size": 512,
+  "layer_norm_eps": 1e-12,
+  "model_type": "vit",
+  "num_attention_heads": 8,
+  "num_channels": 3,
+  "num_hidden_layers": 4,
+  "patch_size": 4,
+  "qkv_bias": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.46.2"
+}

checkpoint-60/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15c3e4eadac387d483c826c1b623a2d3556ddc29b917fb4af6ac56d20f6bc19a
+size 8563512

checkpoint-60/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:199a160bf7c2384f9978abcd0839d2e41a02de869543a5cf1b9e930265ba8cba
+size 17153530

checkpoint-60/preprocessor_config.json

@@ -0,0 +1,22 @@
+{
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": false,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "ViTFeatureExtractor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "height": 32,
+    "width": 32
+  }
+}

checkpoint-60/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7083fc19c48bbb45c07abf88c54fd55da4575c8de4143ad56fb803aa19e03e1c
+size 14244

checkpoint-60/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe62f9b4684e8fb0c29de8f793bdd1b8fa855b987b0ead0080529f92de69f770
+size 1000

checkpoint-60/trainer_state.json

@@ -0,0 +1,633 @@
+{
+  "best_metric": 0.8018098205014093,
+  "best_model_checkpoint": "/tmp/logs/binary_classification_model_v3.1.3_spines/checkpoint-60",
+  "epoch": 15.0,
+  "eval_steps": 500,
+  "global_step": 60,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.25,
+      "grad_norm": 274159.65625,
+      "learning_rate": 1.605445671718469e-07,
+      "loss": 0.4196,
+      "step": 1
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 123727.359375,
+      "learning_rate": 3.210891343436938e-07,
+      "loss": 0.4092,
+      "step": 2
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 222667.140625,
+      "learning_rate": 4.816337015155407e-07,
+      "loss": 0.4168,
+      "step": 3
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 209992.453125,
+      "learning_rate": 6.421782686873876e-07,
+      "loss": 0.4225,
+      "step": 4
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7976561341047322,
+      "eval_f1": 0.7792165749433474,
+      "eval_loss": 0.41963040828704834,
+      "eval_precision": 0.795702479338843,
+      "eval_recall": 0.7633999365683476,
+      "eval_runtime": 3.391,
+      "eval_samples_per_second": 1987.931,
+      "eval_steps_per_second": 1.18,
+      "step": 4
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 109456.3203125,
+      "learning_rate": 8.027228358592346e-07,
+      "loss": 0.4175,
+      "step": 5
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 263527.1875,
+      "learning_rate": 9.632674030310814e-07,
+      "loss": 0.4227,
+      "step": 6
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 201531.734375,
+      "learning_rate": 1.1238119702029283e-06,
+      "loss": 0.4214,
+      "step": 7
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 185288.390625,
+      "learning_rate": 1.2843565373747752e-06,
+      "loss": 0.4262,
+      "step": 8
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7937991395935321,
+      "eval_f1": 0.7719067935674434,
+      "eval_loss": 0.4199751019477844,
+      "eval_precision": 0.7997279836790208,
+      "eval_recall": 0.7459562321598477,
+      "eval_runtime": 3.5857,
+      "eval_samples_per_second": 1879.992,
+      "eval_steps_per_second": 1.116,
+      "step": 8
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 155418.6875,
+      "learning_rate": 1.444901104546622e-06,
+      "loss": 0.4239,
+      "step": 9
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 57884.13671875,
+      "learning_rate": 1.6054456717184691e-06,
+      "loss": 0.4192,
+      "step": 10
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 67545.6796875,
+      "learning_rate": 1.765990238890316e-06,
+      "loss": 0.4241,
+      "step": 11
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 205635.140625,
+      "learning_rate": 1.926534806062163e-06,
+      "loss": 0.4239,
+      "step": 12
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.79884290164664,
+      "eval_f1": 0.7818532818532818,
+      "eval_loss": 0.42012614011764526,
+      "eval_precision": 0.7933398628795298,
+      "eval_recall": 0.7706945765937203,
+      "eval_runtime": 3.6476,
+      "eval_samples_per_second": 1848.072,
+      "eval_steps_per_second": 1.097,
+      "step": 12
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 330362.96875,
+      "learning_rate": 2.0870793732340097e-06,
+      "loss": 0.4369,
+      "step": 13
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 340636.6875,
+      "learning_rate": 2.2476239404058566e-06,
+      "loss": 0.4186,
+      "step": 14
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 224574.828125,
+      "learning_rate": 2.408168507577704e-06,
+      "loss": 0.4217,
+      "step": 15
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 234618.53125,
+      "learning_rate": 2.5687130747495503e-06,
+      "loss": 0.397,
+      "step": 16
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7969144043910399,
+      "eval_f1": 0.7780119993513864,
+      "eval_loss": 0.4202006161212921,
+      "eval_precision": 0.7959522229595223,
+      "eval_recall": 0.7608626704725658,
+      "eval_runtime": 3.7223,
+      "eval_samples_per_second": 1810.979,
+      "eval_steps_per_second": 1.075,
+      "step": 16
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 214805.421875,
+      "learning_rate": 2.729257641921397e-06,
+      "loss": 0.4257,
+      "step": 17
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 193107.890625,
+      "learning_rate": 2.889802209093244e-06,
+      "loss": 0.421,
+      "step": 18
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 185903.40625,
+      "learning_rate": 3.0503467762650913e-06,
+      "loss": 0.4226,
+      "step": 19
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 146775.96875,
+      "learning_rate": 3.2108913434369382e-06,
+      "loss": 0.4049,
+      "step": 20
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7930574098798397,
+      "eval_f1": 0.7698399604025739,
+      "eval_loss": 0.4199390411376953,
+      "eval_precision": 0.8022696011004127,
+      "eval_recall": 0.739930225182366,
+      "eval_runtime": 3.7358,
+      "eval_samples_per_second": 1804.441,
+      "eval_steps_per_second": 1.071,
+      "step": 20
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 193645.421875,
+      "learning_rate": 3.371435910608785e-06,
+      "loss": 0.4233,
+      "step": 21
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 626666.875,
+      "learning_rate": 3.531980477780632e-06,
+      "loss": 0.421,
+      "step": 22
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 317973.40625,
+      "learning_rate": 3.692525044952479e-06,
+      "loss": 0.4353,
+      "step": 23
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 185018.9375,
+      "learning_rate": 3.853069612124326e-06,
+      "loss": 0.413,
+      "step": 24
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.7991395935321169,
+      "eval_f1": 0.7809061488673139,
+      "eval_loss": 0.4188691973686218,
+      "eval_precision": 0.7971589032044929,
+      "eval_recall": 0.7653028861401839,
+      "eval_runtime": 3.7978,
+      "eval_samples_per_second": 1774.966,
+      "eval_steps_per_second": 1.053,
+      "step": 24
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 113566.4609375,
+      "learning_rate": 4.0136141792961726e-06,
+      "loss": 0.412,
+      "step": 25
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 261597.5625,
+      "learning_rate": 4.1741587464680194e-06,
+      "loss": 0.4271,
+      "step": 26
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 322553.15625,
+      "learning_rate": 4.334703313639866e-06,
+      "loss": 0.4138,
+      "step": 27
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 299670.59375,
+      "learning_rate": 4.495247880811713e-06,
+      "loss": 0.4336,
+      "step": 28
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.7982495178756861,
+      "eval_f1": 0.7792924375202857,
+      "eval_loss": 0.4183463454246521,
+      "eval_precision": 0.7979395147889664,
+      "eval_recall": 0.7614969869965112,
+      "eval_runtime": 3.822,
+      "eval_samples_per_second": 1763.716,
+      "eval_steps_per_second": 1.047,
+      "step": 28
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 129755.515625,
+      "learning_rate": 4.655792447983561e-06,
+      "loss": 0.4236,
+      "step": 29
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 149775.578125,
+      "learning_rate": 4.816337015155408e-06,
+      "loss": 0.4221,
+      "step": 30
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 452987.9375,
+      "learning_rate": 4.976881582327254e-06,
+      "loss": 0.4276,
+      "step": 31
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 406737.96875,
+      "learning_rate": 5.137426149499101e-06,
+      "loss": 0.408,
+      "step": 32
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.7954309449636553,
+      "eval_f1": 0.7733026467203682,
+      "eval_loss": 0.4189365804195404,
+      "eval_precision": 0.8027303754266212,
+      "eval_recall": 0.7459562321598477,
+      "eval_runtime": 4.1779,
+      "eval_samples_per_second": 1613.482,
+      "eval_steps_per_second": 0.957,
+      "step": 32
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 173327.0625,
+      "learning_rate": 5.297970716670948e-06,
+      "loss": 0.4021,
+      "step": 33
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 132129.640625,
+      "learning_rate": 5.458515283842794e-06,
+      "loss": 0.4216,
+      "step": 34
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 328514.625,
+      "learning_rate": 5.619059851014642e-06,
+      "loss": 0.4214,
+      "step": 35
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 296242.96875,
+      "learning_rate": 5.779604418186488e-06,
+      "loss": 0.4328,
+      "step": 36
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.8003263610740247,
+      "eval_f1": 0.7828331719909648,
+      "eval_loss": 0.4176911413669586,
+      "eval_precision": 0.7967159277504106,
+      "eval_recall": 0.7694259435458294,
+      "eval_runtime": 3.9359,
+      "eval_samples_per_second": 1712.676,
+      "eval_steps_per_second": 1.016,
+      "step": 36
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 164642.25,
+      "learning_rate": 5.940148985358336e-06,
+      "loss": 0.4296,
+      "step": 37
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 386216.90625,
+      "learning_rate": 6.100693552530183e-06,
+      "loss": 0.4082,
+      "step": 38
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 341470.5625,
+      "learning_rate": 6.2612381197020296e-06,
+      "loss": 0.4235,
+      "step": 39
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 161297.625,
+      "learning_rate": 6.4217826868738764e-06,
+      "loss": 0.4109,
+      "step": 40
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.7982495178756861,
+      "eval_f1": 0.7789336801040312,
+      "eval_loss": 0.41719672083854675,
+      "eval_precision": 0.7989329776592198,
+      "eval_recall": 0.7599111956866477,
+      "eval_runtime": 3.9721,
+      "eval_samples_per_second": 1697.089,
+      "eval_steps_per_second": 1.007,
+      "step": 40
+    },
+    {
+      "epoch": 10.25,
+      "grad_norm": 336130.65625,
+      "learning_rate": 6.5823272540457225e-06,
+      "loss": 0.4119,
+      "step": 41
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 538575.9375,
+      "learning_rate": 6.74287182121757e-06,
+      "loss": 0.4189,
+      "step": 42
+    },
+    {
+      "epoch": 10.75,
+      "grad_norm": 566222.0625,
+      "learning_rate": 6.903416388389417e-06,
+      "loss": 0.4167,
+      "step": 43
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 502448.53125,
+      "learning_rate": 7.063960955561264e-06,
+      "loss": 0.4242,
+      "step": 44
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.8006230529595015,
+      "eval_f1": 0.785234899328859,
+      "eval_loss": 0.41683289408683777,
+      "eval_precision": 0.7913043478260869,
+      "eval_recall": 0.7792578496669839,
+      "eval_runtime": 4.0247,
+      "eval_samples_per_second": 1674.916,
+      "eval_steps_per_second": 0.994,
+      "step": 44
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 461244.03125,
+      "learning_rate": 7.224505522733111e-06,
+      "loss": 0.4241,
+      "step": 45
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 354418.75,
+      "learning_rate": 7.385050089904958e-06,
+      "loss": 0.4113,
+      "step": 46
+    },
+    {
+      "epoch": 11.75,
+      "grad_norm": 367365.15625,
+      "learning_rate": 7.5455946570768045e-06,
+      "loss": 0.4066,
+      "step": 47
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 241178.4375,
+      "learning_rate": 7.706139224248651e-06,
+      "loss": 0.4237,
+      "step": 48
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.7964693665628245,
+      "eval_f1": 0.7763286599282687,
+      "eval_loss": 0.41600093245506287,
+      "eval_precision": 0.7987252599798725,
+      "eval_recall": 0.7551538217570568,
+      "eval_runtime": 4.068,
+      "eval_samples_per_second": 1657.071,
+      "eval_steps_per_second": 0.983,
+      "step": 48
+    },
+    {
+      "epoch": 12.25,
+      "grad_norm": 236949.75,
+      "learning_rate": 7.866683791420498e-06,
+      "loss": 0.4187,
+      "step": 49
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 367486.0,
+      "learning_rate": 8.027228358592345e-06,
+      "loss": 0.4213,
+      "step": 50
+    },
+    {
+      "epoch": 12.75,
+      "grad_norm": 594396.3125,
+      "learning_rate": 8.187772925764192e-06,
+      "loss": 0.4117,
+      "step": 51
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 535547.0,
+      "learning_rate": 8.348317492936039e-06,
+      "loss": 0.4294,
+      "step": 52
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.8001780151312862,
+      "eval_f1": 0.7838921867479545,
+      "eval_loss": 0.4154762029647827,
+      "eval_precision": 0.7931818181818182,
+      "eval_recall": 0.7748176339993657,
+      "eval_runtime": 4.1194,
+      "eval_samples_per_second": 1636.391,
+      "eval_steps_per_second": 0.971,
+      "step": 52
+    },
+    {
+      "epoch": 13.25,
+      "grad_norm": 192222.078125,
+      "learning_rate": 8.508862060107887e-06,
+      "loss": 0.4259,
+      "step": 53
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 169889.9375,
+      "learning_rate": 8.669406627279733e-06,
+      "loss": 0.4129,
+      "step": 54
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 802483.4375,
+      "learning_rate": 8.82995119445158e-06,
+      "loss": 0.4143,
+      "step": 55
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 1152123.0,
+      "learning_rate": 8.990495761623426e-06,
+      "loss": 0.4141,
+      "step": 56
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.7955792909063937,
+      "eval_f1": 0.7706391478029294,
+      "eval_loss": 0.41698139905929565,
+      "eval_precision": 0.8108581436077058,
+      "eval_recall": 0.7342213764668569,
+      "eval_runtime": 4.1658,
+      "eval_samples_per_second": 1618.183,
+      "eval_steps_per_second": 0.96,
+      "step": 56
+    },
+    {
+      "epoch": 14.25,
+      "grad_norm": 431923.5,
+      "learning_rate": 9.151040328795273e-06,
+      "loss": 0.4048,
+      "step": 57
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 107792.703125,
+      "learning_rate": 9.311584895967122e-06,
+      "loss": 0.4099,
+      "step": 58
+    },
+    {
+      "epoch": 14.75,
+      "grad_norm": 361818.65625,
+      "learning_rate": 9.472129463138967e-06,
+      "loss": 0.4235,
+      "step": 59
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 672806.375,
+      "learning_rate": 9.632674030310815e-06,
+      "loss": 0.4233,
+      "step": 60
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.8018098205014093,
+      "eval_f1": 0.7872611464968153,
+      "eval_loss": 0.41520848870277405,
+      "eval_precision": 0.7905340582027502,
+      "eval_recall": 0.7840152235965747,
+      "eval_runtime": 4.2184,
+      "eval_samples_per_second": 1598.016,
+      "eval_steps_per_second": 0.948,
+      "step": 60
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 80,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9208628260945920.0,
+  "train_batch_size": 2048,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-60/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a962d081d3ebaf098ba4aae1ae99c3f381953544a4002d54b37e070680980a81
+size 5368

config.json

@@ -1,4 +1,5 @@
 {
+  "_name_or_path": "ZivKassnerNK/binary_classification_model_v3.1.3_spines",
   "architectures": [
     "ViTForImageClassification"
   ],
@@ -18,5 +19,5 @@
   "patch_size": 4,
   "qkv_bias": true,
   "torch_dtype": "float32",
-  "transformers_version": "4.46.1"
+  "transformers_version": "4.46.2"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d3763285e8dd082d9840123bbe4a55f1faaefe6dabaa87bbe48938bee65866d2
+oid sha256:92713d053262518988d9a71ee99c08290f4284d2816caca747d6ffa93ff0b342
 size 8563512

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df1535d56f69b10dec594d0116a6d77d9220713be79d301e5d938e5c3d7f1940
+oid sha256:a962d081d3ebaf098ba4aae1ae99c3f381953544a4002d54b37e070680980a81
 size 5368