Upload folder using huggingface_hub
Browse files- checkpoint-3450/config.json +72 -0
- checkpoint-3450/model.safetensors +3 -0
- checkpoint-3450/optimizer.pt +3 -0
- checkpoint-3450/preprocessor_config.json +36 -0
- checkpoint-3450/rng_state.pth +3 -0
- checkpoint-3450/scheduler.pt +3 -0
- checkpoint-3450/trainer_state.json +345 -0
- checkpoint-3450/training_args.bin +3 -0
- config.json +10 -12
- model.safetensors +2 -2
- training_args.bin +1 -1
checkpoint-3450/config.json
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-224-in21k",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ViTForImageClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"encoder_stride": 16,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.0,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "01",
|
| 13 |
+
"1": "02",
|
| 14 |
+
"2": "03",
|
| 15 |
+
"3": "04",
|
| 16 |
+
"4": "05",
|
| 17 |
+
"5": "06-07",
|
| 18 |
+
"6": "08-09",
|
| 19 |
+
"7": "10-12",
|
| 20 |
+
"8": "13-15",
|
| 21 |
+
"9": "16-20",
|
| 22 |
+
"10": "21-25",
|
| 23 |
+
"11": "26-30",
|
| 24 |
+
"12": "31-35",
|
| 25 |
+
"13": "36-40",
|
| 26 |
+
"14": "41-45",
|
| 27 |
+
"15": "46-50",
|
| 28 |
+
"16": "51-55",
|
| 29 |
+
"17": "56-60",
|
| 30 |
+
"18": "61-70",
|
| 31 |
+
"19": "71-80",
|
| 32 |
+
"20": "81-90",
|
| 33 |
+
"21": "90+"
|
| 34 |
+
},
|
| 35 |
+
"image_size": 224,
|
| 36 |
+
"initializer_range": 0.02,
|
| 37 |
+
"intermediate_size": 3072,
|
| 38 |
+
"label2id": {
|
| 39 |
+
"01": 0,
|
| 40 |
+
"02": 1,
|
| 41 |
+
"03": 2,
|
| 42 |
+
"04": 3,
|
| 43 |
+
"05": 4,
|
| 44 |
+
"06-07": 5,
|
| 45 |
+
"08-09": 6,
|
| 46 |
+
"10-12": 7,
|
| 47 |
+
"13-15": 8,
|
| 48 |
+
"16-20": 9,
|
| 49 |
+
"21-25": 10,
|
| 50 |
+
"26-30": 11,
|
| 51 |
+
"31-35": 12,
|
| 52 |
+
"36-40": 13,
|
| 53 |
+
"41-45": 14,
|
| 54 |
+
"46-50": 15,
|
| 55 |
+
"51-55": 16,
|
| 56 |
+
"56-60": 17,
|
| 57 |
+
"61-70": 18,
|
| 58 |
+
"71-80": 19,
|
| 59 |
+
"81-90": 20,
|
| 60 |
+
"90+": 21
|
| 61 |
+
},
|
| 62 |
+
"layer_norm_eps": 1e-12,
|
| 63 |
+
"model_type": "vit",
|
| 64 |
+
"num_attention_heads": 12,
|
| 65 |
+
"num_channels": 3,
|
| 66 |
+
"num_hidden_layers": 12,
|
| 67 |
+
"patch_size": 16,
|
| 68 |
+
"problem_type": "single_label_classification",
|
| 69 |
+
"qkv_bias": true,
|
| 70 |
+
"torch_dtype": "float32",
|
| 71 |
+
"transformers_version": "4.41.1"
|
| 72 |
+
}
|
checkpoint-3450/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:794ac67b4aaf79e01e2addd21b15f9e8ca8fb6adf09679466daed316d3ca94ce
|
| 3 |
+
size 343285496
|
checkpoint-3450/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:495ab4370a5dddfaba44b9795c0e7d069ab3c83803c6792a3d85994c9c343aa5
|
| 3 |
+
size 686691898
|
checkpoint-3450/preprocessor_config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_valid_processor_keys": [
|
| 3 |
+
"images",
|
| 4 |
+
"do_resize",
|
| 5 |
+
"size",
|
| 6 |
+
"resample",
|
| 7 |
+
"do_rescale",
|
| 8 |
+
"rescale_factor",
|
| 9 |
+
"do_normalize",
|
| 10 |
+
"image_mean",
|
| 11 |
+
"image_std",
|
| 12 |
+
"return_tensors",
|
| 13 |
+
"data_format",
|
| 14 |
+
"input_data_format"
|
| 15 |
+
],
|
| 16 |
+
"do_normalize": true,
|
| 17 |
+
"do_rescale": true,
|
| 18 |
+
"do_resize": true,
|
| 19 |
+
"image_mean": [
|
| 20 |
+
0.5,
|
| 21 |
+
0.5,
|
| 22 |
+
0.5
|
| 23 |
+
],
|
| 24 |
+
"image_processor_type": "ViTImageProcessor",
|
| 25 |
+
"image_std": [
|
| 26 |
+
0.5,
|
| 27 |
+
0.5,
|
| 28 |
+
0.5
|
| 29 |
+
],
|
| 30 |
+
"resample": 2,
|
| 31 |
+
"rescale_factor": 0.00392156862745098,
|
| 32 |
+
"size": {
|
| 33 |
+
"height": 224,
|
| 34 |
+
"width": 224
|
| 35 |
+
}
|
| 36 |
+
}
|
checkpoint-3450/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d830bc0bdad9a43c538efa339ac7fa041ed81e0f4a462da9bc03c5d546028f7
|
| 3 |
+
size 14244
|
checkpoint-3450/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:750efa1a992d3127b077e2c4466c2d711d73d777c3732a6272e57941d326ecd2
|
| 3 |
+
size 1064
|
checkpoint-3450/trainer_state.json
ADDED
|
@@ -0,0 +1,345 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.5474669933319092,
|
| 3 |
+
"best_model_checkpoint": "facial_age_estimator/checkpoint-3450",
|
| 4 |
+
"epoch": 30.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 3450,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_accuracy": 0.26057633353770693,
|
| 14 |
+
"eval_loss": 2.853342294692993,
|
| 15 |
+
"eval_runtime": 95.848,
|
| 16 |
+
"eval_samples_per_second": 102.099,
|
| 17 |
+
"eval_steps_per_second": 1.596,
|
| 18 |
+
"step": 115
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 2.0,
|
| 22 |
+
"eval_accuracy": 0.31943592887798894,
|
| 23 |
+
"eval_loss": 2.498396635055542,
|
| 24 |
+
"eval_runtime": 86.0515,
|
| 25 |
+
"eval_samples_per_second": 113.723,
|
| 26 |
+
"eval_steps_per_second": 1.778,
|
| 27 |
+
"step": 230
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 3.0,
|
| 31 |
+
"eval_accuracy": 0.35080727569997955,
|
| 32 |
+
"eval_loss": 2.3049697875976562,
|
| 33 |
+
"eval_runtime": 85.5701,
|
| 34 |
+
"eval_samples_per_second": 114.362,
|
| 35 |
+
"eval_steps_per_second": 1.788,
|
| 36 |
+
"step": 345
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 4.0,
|
| 40 |
+
"eval_accuracy": 0.3734927447373799,
|
| 41 |
+
"eval_loss": 2.1921744346618652,
|
| 42 |
+
"eval_runtime": 84.3843,
|
| 43 |
+
"eval_samples_per_second": 115.969,
|
| 44 |
+
"eval_steps_per_second": 1.813,
|
| 45 |
+
"step": 460
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 4.3478260869565215,
|
| 49 |
+
"grad_norm": 1.9413998126983643,
|
| 50 |
+
"learning_rate": 8.676470588235295e-06,
|
| 51 |
+
"loss": 2.4591,
|
| 52 |
+
"step": 500
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 5.0,
|
| 56 |
+
"eval_accuracy": 0.38565297363580625,
|
| 57 |
+
"eval_loss": 2.10524320602417,
|
| 58 |
+
"eval_runtime": 83.7671,
|
| 59 |
+
"eval_samples_per_second": 116.824,
|
| 60 |
+
"eval_steps_per_second": 1.826,
|
| 61 |
+
"step": 575
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 6.0,
|
| 65 |
+
"eval_accuracy": 0.39505415900265684,
|
| 66 |
+
"eval_loss": 2.0314905643463135,
|
| 67 |
+
"eval_runtime": 84.7557,
|
| 68 |
+
"eval_samples_per_second": 115.461,
|
| 69 |
+
"eval_steps_per_second": 1.805,
|
| 70 |
+
"step": 690
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 7.0,
|
| 74 |
+
"eval_accuracy": 0.4055793991416309,
|
| 75 |
+
"eval_loss": 1.9862908124923706,
|
| 76 |
+
"eval_runtime": 82.3593,
|
| 77 |
+
"eval_samples_per_second": 118.821,
|
| 78 |
+
"eval_steps_per_second": 1.858,
|
| 79 |
+
"step": 805
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 8.0,
|
| 83 |
+
"eval_accuracy": 0.41702432045779686,
|
| 84 |
+
"eval_loss": 1.925173282623291,
|
| 85 |
+
"eval_runtime": 85.8756,
|
| 86 |
+
"eval_samples_per_second": 113.956,
|
| 87 |
+
"eval_steps_per_second": 1.782,
|
| 88 |
+
"step": 920
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 8.695652173913043,
|
| 92 |
+
"grad_norm": 3.8010852336883545,
|
| 93 |
+
"learning_rate": 7.205882352941177e-06,
|
| 94 |
+
"loss": 1.8939,
|
| 95 |
+
"step": 1000
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 9.0,
|
| 99 |
+
"eval_accuracy": 0.41998773758430413,
|
| 100 |
+
"eval_loss": 1.8860591650009155,
|
| 101 |
+
"eval_runtime": 85.2312,
|
| 102 |
+
"eval_samples_per_second": 114.817,
|
| 103 |
+
"eval_steps_per_second": 1.795,
|
| 104 |
+
"step": 1035
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 10.0,
|
| 108 |
+
"eval_accuracy": 0.4385857347230738,
|
| 109 |
+
"eval_loss": 1.8425257205963135,
|
| 110 |
+
"eval_runtime": 84.7668,
|
| 111 |
+
"eval_samples_per_second": 115.446,
|
| 112 |
+
"eval_steps_per_second": 1.805,
|
| 113 |
+
"step": 1150
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 11.0,
|
| 117 |
+
"eval_accuracy": 0.44338851420396486,
|
| 118 |
+
"eval_loss": 1.8084428310394287,
|
| 119 |
+
"eval_runtime": 84.3931,
|
| 120 |
+
"eval_samples_per_second": 115.957,
|
| 121 |
+
"eval_steps_per_second": 1.813,
|
| 122 |
+
"step": 1265
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 12.0,
|
| 126 |
+
"eval_accuracy": 0.44430819538115673,
|
| 127 |
+
"eval_loss": 1.775467872619629,
|
| 128 |
+
"eval_runtime": 83.7762,
|
| 129 |
+
"eval_samples_per_second": 116.811,
|
| 130 |
+
"eval_steps_per_second": 1.826,
|
| 131 |
+
"step": 1380
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 13.0,
|
| 135 |
+
"eval_accuracy": 0.4494175352544451,
|
| 136 |
+
"eval_loss": 1.7456355094909668,
|
| 137 |
+
"eval_runtime": 84.1752,
|
| 138 |
+
"eval_samples_per_second": 116.258,
|
| 139 |
+
"eval_steps_per_second": 1.818,
|
| 140 |
+
"step": 1495
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 13.043478260869565,
|
| 144 |
+
"grad_norm": 2.846041202545166,
|
| 145 |
+
"learning_rate": 5.735294117647059e-06,
|
| 146 |
+
"loss": 1.6732,
|
| 147 |
+
"step": 1500
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 14.0,
|
| 151 |
+
"eval_accuracy": 0.45299407316574697,
|
| 152 |
+
"eval_loss": 1.7259840965270996,
|
| 153 |
+
"eval_runtime": 84.9249,
|
| 154 |
+
"eval_samples_per_second": 115.231,
|
| 155 |
+
"eval_steps_per_second": 1.802,
|
| 156 |
+
"step": 1610
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 15.0,
|
| 160 |
+
"eval_accuracy": 0.4582055998365011,
|
| 161 |
+
"eval_loss": 1.7003978490829468,
|
| 162 |
+
"eval_runtime": 83.6465,
|
| 163 |
+
"eval_samples_per_second": 116.992,
|
| 164 |
+
"eval_steps_per_second": 1.829,
|
| 165 |
+
"step": 1725
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 16.0,
|
| 169 |
+
"eval_accuracy": 0.46607398324136523,
|
| 170 |
+
"eval_loss": 1.6746762990951538,
|
| 171 |
+
"eval_runtime": 83.5102,
|
| 172 |
+
"eval_samples_per_second": 117.183,
|
| 173 |
+
"eval_steps_per_second": 1.832,
|
| 174 |
+
"step": 1840
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 17.0,
|
| 178 |
+
"eval_accuracy": 0.4717964438994482,
|
| 179 |
+
"eval_loss": 1.6561590433120728,
|
| 180 |
+
"eval_runtime": 84.4244,
|
| 181 |
+
"eval_samples_per_second": 115.914,
|
| 182 |
+
"eval_steps_per_second": 1.812,
|
| 183 |
+
"step": 1955
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 17.391304347826086,
|
| 187 |
+
"grad_norm": 3.774789333343506,
|
| 188 |
+
"learning_rate": 4.264705882352942e-06,
|
| 189 |
+
"loss": 1.5216,
|
| 190 |
+
"step": 2000
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 18.0,
|
| 194 |
+
"eval_accuracy": 0.4754751686082158,
|
| 195 |
+
"eval_loss": 1.641586422920227,
|
| 196 |
+
"eval_runtime": 85.8369,
|
| 197 |
+
"eval_samples_per_second": 114.007,
|
| 198 |
+
"eval_steps_per_second": 1.782,
|
| 199 |
+
"step": 2070
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 19.0,
|
| 203 |
+
"eval_accuracy": 0.47588391579807887,
|
| 204 |
+
"eval_loss": 1.629101037979126,
|
| 205 |
+
"eval_runtime": 85.0024,
|
| 206 |
+
"eval_samples_per_second": 115.126,
|
| 207 |
+
"eval_steps_per_second": 1.8,
|
| 208 |
+
"step": 2185
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 20.0,
|
| 212 |
+
"eval_accuracy": 0.47946045370938073,
|
| 213 |
+
"eval_loss": 1.6132134199142456,
|
| 214 |
+
"eval_runtime": 83.8043,
|
| 215 |
+
"eval_samples_per_second": 116.772,
|
| 216 |
+
"eval_steps_per_second": 1.826,
|
| 217 |
+
"step": 2300
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 21.0,
|
| 221 |
+
"eval_accuracy": 0.48732883711424485,
|
| 222 |
+
"eval_loss": 1.5983059406280518,
|
| 223 |
+
"eval_runtime": 82.4226,
|
| 224 |
+
"eval_samples_per_second": 118.73,
|
| 225 |
+
"eval_steps_per_second": 1.856,
|
| 226 |
+
"step": 2415
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 21.73913043478261,
|
| 230 |
+
"grad_norm": 4.8250579833984375,
|
| 231 |
+
"learning_rate": 2.7941176470588237e-06,
|
| 232 |
+
"loss": 1.4196,
|
| 233 |
+
"step": 2500
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 22.0,
|
| 237 |
+
"eval_accuracy": 0.48681790312691603,
|
| 238 |
+
"eval_loss": 1.588654637336731,
|
| 239 |
+
"eval_runtime": 83.9042,
|
| 240 |
+
"eval_samples_per_second": 116.633,
|
| 241 |
+
"eval_steps_per_second": 1.824,
|
| 242 |
+
"step": 2530
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 23.0,
|
| 246 |
+
"eval_accuracy": 0.4901900674432863,
|
| 247 |
+
"eval_loss": 1.5788466930389404,
|
| 248 |
+
"eval_runtime": 85.2579,
|
| 249 |
+
"eval_samples_per_second": 114.781,
|
| 250 |
+
"eval_steps_per_second": 1.795,
|
| 251 |
+
"step": 2645
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 24.0,
|
| 255 |
+
"eval_accuracy": 0.4943797261393828,
|
| 256 |
+
"eval_loss": 1.572027325630188,
|
| 257 |
+
"eval_runtime": 85.4444,
|
| 258 |
+
"eval_samples_per_second": 114.531,
|
| 259 |
+
"eval_steps_per_second": 1.791,
|
| 260 |
+
"step": 2760
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 25.0,
|
| 264 |
+
"eval_accuracy": 0.4872266503167791,
|
| 265 |
+
"eval_loss": 1.572544813156128,
|
| 266 |
+
"eval_runtime": 85.0642,
|
| 267 |
+
"eval_samples_per_second": 115.043,
|
| 268 |
+
"eval_steps_per_second": 1.799,
|
| 269 |
+
"step": 2875
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 26.0,
|
| 273 |
+
"eval_accuracy": 0.4948906601267116,
|
| 274 |
+
"eval_loss": 1.5591200590133667,
|
| 275 |
+
"eval_runtime": 83.9021,
|
| 276 |
+
"eval_samples_per_second": 116.636,
|
| 277 |
+
"eval_steps_per_second": 1.824,
|
| 278 |
+
"step": 2990
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 26.08695652173913,
|
| 282 |
+
"grad_norm": 3.582409381866455,
|
| 283 |
+
"learning_rate": 1.323529411764706e-06,
|
| 284 |
+
"loss": 1.3513,
|
| 285 |
+
"step": 3000
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 27.0,
|
| 289 |
+
"eval_accuracy": 0.49080318822808094,
|
| 290 |
+
"eval_loss": 1.5544283390045166,
|
| 291 |
+
"eval_runtime": 85.2923,
|
| 292 |
+
"eval_samples_per_second": 114.735,
|
| 293 |
+
"eval_steps_per_second": 1.794,
|
| 294 |
+
"step": 3105
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 28.0,
|
| 298 |
+
"eval_accuracy": 0.4963212752912324,
|
| 299 |
+
"eval_loss": 1.55410635471344,
|
| 300 |
+
"eval_runtime": 83.7558,
|
| 301 |
+
"eval_samples_per_second": 116.84,
|
| 302 |
+
"eval_steps_per_second": 1.827,
|
| 303 |
+
"step": 3220
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 29.0,
|
| 307 |
+
"eval_accuracy": 0.49509503372164315,
|
| 308 |
+
"eval_loss": 1.5487123727798462,
|
| 309 |
+
"eval_runtime": 85.4702,
|
| 310 |
+
"eval_samples_per_second": 114.496,
|
| 311 |
+
"eval_steps_per_second": 1.79,
|
| 312 |
+
"step": 3335
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 30.0,
|
| 316 |
+
"eval_accuracy": 0.4971387696709585,
|
| 317 |
+
"eval_loss": 1.5474669933319092,
|
| 318 |
+
"eval_runtime": 82.7937,
|
| 319 |
+
"eval_samples_per_second": 118.197,
|
| 320 |
+
"eval_steps_per_second": 1.848,
|
| 321 |
+
"step": 3450
|
| 322 |
+
}
|
| 323 |
+
],
|
| 324 |
+
"logging_steps": 500,
|
| 325 |
+
"max_steps": 3450,
|
| 326 |
+
"num_input_tokens_seen": 0,
|
| 327 |
+
"num_train_epochs": 30,
|
| 328 |
+
"save_steps": 500,
|
| 329 |
+
"stateful_callbacks": {
|
| 330 |
+
"TrainerControl": {
|
| 331 |
+
"args": {
|
| 332 |
+
"should_epoch_stop": false,
|
| 333 |
+
"should_evaluate": false,
|
| 334 |
+
"should_log": false,
|
| 335 |
+
"should_save": true,
|
| 336 |
+
"should_training_stop": true
|
| 337 |
+
},
|
| 338 |
+
"attributes": {}
|
| 339 |
+
}
|
| 340 |
+
},
|
| 341 |
+
"total_flos": 3.4128939348225147e+19,
|
| 342 |
+
"train_batch_size": 128,
|
| 343 |
+
"trial_name": null,
|
| 344 |
+
"trial_params": null
|
| 345 |
+
}
|
checkpoint-3450/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76ee3f055c95efecd4fcb9240a1dde4af41411a36d523b3a39b9672edbfd9ef4
|
| 3 |
+
size 5048
|
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"architectures": [
|
| 4 |
"ViTForImageClassification"
|
| 5 |
],
|
|
@@ -27,11 +27,10 @@
|
|
| 27 |
"15": "46-50",
|
| 28 |
"16": "51-55",
|
| 29 |
"17": "56-60",
|
| 30 |
-
"18": "61-
|
| 31 |
-
"19": "
|
| 32 |
-
"20": "
|
| 33 |
-
"21": "
|
| 34 |
-
"22": "90+"
|
| 35 |
},
|
| 36 |
"image_size": 224,
|
| 37 |
"initializer_range": 0.02,
|
|
@@ -55,11 +54,10 @@
|
|
| 55 |
"46-50": 15,
|
| 56 |
"51-55": 16,
|
| 57 |
"56-60": 17,
|
| 58 |
-
"61-
|
| 59 |
-
"
|
| 60 |
-
"
|
| 61 |
-
"
|
| 62 |
-
"90+": 22
|
| 63 |
},
|
| 64 |
"layer_norm_eps": 1e-12,
|
| 65 |
"model_type": "vit",
|
|
@@ -70,5 +68,5 @@
|
|
| 70 |
"problem_type": "single_label_classification",
|
| 71 |
"qkv_bias": true,
|
| 72 |
"torch_dtype": "float32",
|
| 73 |
-
"transformers_version": "4.41.
|
| 74 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-224-in21k",
|
| 3 |
"architectures": [
|
| 4 |
"ViTForImageClassification"
|
| 5 |
],
|
|
|
|
| 27 |
"15": "46-50",
|
| 28 |
"16": "51-55",
|
| 29 |
"17": "56-60",
|
| 30 |
+
"18": "61-70",
|
| 31 |
+
"19": "71-80",
|
| 32 |
+
"20": "81-90",
|
| 33 |
+
"21": "90+"
|
|
|
|
| 34 |
},
|
| 35 |
"image_size": 224,
|
| 36 |
"initializer_range": 0.02,
|
|
|
|
| 54 |
"46-50": 15,
|
| 55 |
"51-55": 16,
|
| 56 |
"56-60": 17,
|
| 57 |
+
"61-70": 18,
|
| 58 |
+
"71-80": 19,
|
| 59 |
+
"81-90": 20,
|
| 60 |
+
"90+": 21
|
|
|
|
| 61 |
},
|
| 62 |
"layer_norm_eps": 1e-12,
|
| 63 |
"model_type": "vit",
|
|
|
|
| 68 |
"problem_type": "single_label_classification",
|
| 69 |
"qkv_bias": true,
|
| 70 |
"torch_dtype": "float32",
|
| 71 |
+
"transformers_version": "4.41.1"
|
| 72 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:794ac67b4aaf79e01e2addd21b15f9e8ca8fb6adf09679466daed316d3ca94ce
|
| 3 |
+
size 343285496
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5048
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76ee3f055c95efecd4fcb9240a1dde4af41411a36d523b3a39b9672edbfd9ef4
|
| 3 |
size 5048
|