Training in progress, epoch 3
Browse files- model.safetensors +1 -1
- tmp-checkpoint-510/config.json +38 -0
- tmp-checkpoint-510/model.safetensors +3 -0
- tmp-checkpoint-510/optimizer.pt +3 -0
- tmp-checkpoint-510/preprocessor_config.json +22 -0
- tmp-checkpoint-510/rng_state.pth +3 -0
- tmp-checkpoint-510/scheduler.pt +3 -0
- tmp-checkpoint-510/trainer_state.json +367 -0
- tmp-checkpoint-510/training_args.bin +3 -0
- tmp-checkpoint-765/config.json +38 -0
- tmp-checkpoint-765/model.safetensors +3 -0
- tmp-checkpoint-765/optimizer.pt +3 -0
- tmp-checkpoint-765/preprocessor_config.json +22 -0
- tmp-checkpoint-765/rng_state.pth +3 -0
- tmp-checkpoint-765/scheduler.pt +3 -0
- tmp-checkpoint-765/trainer_state.json +537 -0
- tmp-checkpoint-765/training_args.bin +3 -0
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343233204
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc8840a55470dd2e6a6b93487d6ef8d9bc38f276aa0c1c59052c184623f26a4a
|
| 3 |
size 343233204
|
tmp-checkpoint-510/config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "crack",
|
| 13 |
+
"1": "for review",
|
| 14 |
+
"2": "material deterioration",
|
| 15 |
+
"3": "repair",
|
| 16 |
+
"4": "water repair"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"intermediate_size": 3072,
|
| 21 |
+
"label2id": {
|
| 22 |
+
"crack": 0,
|
| 23 |
+
"for review": 1,
|
| 24 |
+
"material deterioration": 2,
|
| 25 |
+
"repair": 3,
|
| 26 |
+
"water repair": 4
|
| 27 |
+
},
|
| 28 |
+
"layer_norm_eps": 1e-12,
|
| 29 |
+
"model_type": "vit",
|
| 30 |
+
"num_attention_heads": 12,
|
| 31 |
+
"num_channels": 3,
|
| 32 |
+
"num_hidden_layers": 12,
|
| 33 |
+
"patch_size": 16,
|
| 34 |
+
"problem_type": "single_label_classification",
|
| 35 |
+
"qkv_bias": true,
|
| 36 |
+
"torch_dtype": "float32",
|
| 37 |
+
"transformers_version": "4.37.2"
|
| 38 |
+
}
|
tmp-checkpoint-510/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aea28430ec5253b8513f16dede6fd1f740b432a3423e22e2570629bf8f5419aa
|
| 3 |
+
size 343233204
|
tmp-checkpoint-510/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a444df4b7ad735786f9bae0afcf0cc3fb6ba7a06b028770e837eec10508db87
|
| 3 |
+
size 686587322
|
tmp-checkpoint-510/preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": true,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTImageProcessor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 224,
|
| 20 |
+
"width": 224
|
| 21 |
+
}
|
| 22 |
+
}
|
tmp-checkpoint-510/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dc6b1f31409e20f3c5b16815345c420ad68241f79a206a022e40c06a0be76e8d
|
| 3 |
+
size 14244
|
tmp-checkpoint-510/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91b857db3da0dbc3e203855b4a23b58444df6ed18f66f40c5f63616dfe236d61
|
| 3 |
+
size 1064
|
tmp-checkpoint-510/trainer_state.json
ADDED
|
@@ -0,0 +1,367 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.498457670211792,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-510",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 510,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04,
|
| 13 |
+
"learning_rate": 6.535947712418301e-07,
|
| 14 |
+
"loss": 1.6179,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.08,
|
| 19 |
+
"learning_rate": 1.3071895424836602e-06,
|
| 20 |
+
"loss": 1.6223,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.12,
|
| 25 |
+
"learning_rate": 1.96078431372549e-06,
|
| 26 |
+
"loss": 1.6263,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.16,
|
| 31 |
+
"learning_rate": 2.6143790849673204e-06,
|
| 32 |
+
"loss": 1.601,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.2,
|
| 37 |
+
"learning_rate": 3.2679738562091506e-06,
|
| 38 |
+
"loss": 1.6074,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.24,
|
| 43 |
+
"learning_rate": 3.92156862745098e-06,
|
| 44 |
+
"loss": 1.5955,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.27,
|
| 49 |
+
"learning_rate": 4.5751633986928105e-06,
|
| 50 |
+
"loss": 1.5698,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.31,
|
| 55 |
+
"learning_rate": 5.228758169934641e-06,
|
| 56 |
+
"loss": 1.5797,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.35,
|
| 61 |
+
"learning_rate": 5.882352941176471e-06,
|
| 62 |
+
"loss": 1.5531,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.39,
|
| 67 |
+
"learning_rate": 6.535947712418301e-06,
|
| 68 |
+
"loss": 1.5434,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.43,
|
| 73 |
+
"learning_rate": 7.1895424836601305e-06,
|
| 74 |
+
"loss": 1.5433,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.47,
|
| 79 |
+
"learning_rate": 7.84313725490196e-06,
|
| 80 |
+
"loss": 1.5329,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.51,
|
| 85 |
+
"learning_rate": 8.496732026143791e-06,
|
| 86 |
+
"loss": 1.4995,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.55,
|
| 91 |
+
"learning_rate": 9.150326797385621e-06,
|
| 92 |
+
"loss": 1.4875,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.59,
|
| 97 |
+
"learning_rate": 9.803921568627451e-06,
|
| 98 |
+
"loss": 1.4863,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.63,
|
| 103 |
+
"learning_rate": 1.0457516339869281e-05,
|
| 104 |
+
"loss": 1.4209,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.67,
|
| 109 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 110 |
+
"loss": 1.4354,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.71,
|
| 115 |
+
"learning_rate": 1.1764705882352942e-05,
|
| 116 |
+
"loss": 1.3572,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.75,
|
| 121 |
+
"learning_rate": 1.2418300653594772e-05,
|
| 122 |
+
"loss": 1.3714,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.78,
|
| 127 |
+
"learning_rate": 1.3071895424836602e-05,
|
| 128 |
+
"loss": 1.3406,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.82,
|
| 133 |
+
"learning_rate": 1.3725490196078432e-05,
|
| 134 |
+
"loss": 1.2574,
|
| 135 |
+
"step": 210
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.86,
|
| 139 |
+
"learning_rate": 1.4379084967320261e-05,
|
| 140 |
+
"loss": 1.2965,
|
| 141 |
+
"step": 220
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.9,
|
| 145 |
+
"learning_rate": 1.5032679738562091e-05,
|
| 146 |
+
"loss": 1.2021,
|
| 147 |
+
"step": 230
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.94,
|
| 151 |
+
"learning_rate": 1.568627450980392e-05,
|
| 152 |
+
"loss": 1.2072,
|
| 153 |
+
"step": 240
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.98,
|
| 157 |
+
"learning_rate": 1.6339869281045753e-05,
|
| 158 |
+
"loss": 1.1457,
|
| 159 |
+
"step": 250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0,
|
| 163 |
+
"eval_accuracy": {
|
| 164 |
+
"accuracy": 0.6941176470588235
|
| 165 |
+
},
|
| 166 |
+
"eval_f1": {
|
| 167 |
+
"f1": 0.683792091280419
|
| 168 |
+
},
|
| 169 |
+
"eval_loss": 1.158827304840088,
|
| 170 |
+
"eval_precision": {
|
| 171 |
+
"precision": 0.7377494408375539
|
| 172 |
+
},
|
| 173 |
+
"eval_recall": {
|
| 174 |
+
"recall": 0.6941484490686618
|
| 175 |
+
},
|
| 176 |
+
"eval_runtime": 25.3262,
|
| 177 |
+
"eval_samples_per_second": 20.137,
|
| 178 |
+
"eval_steps_per_second": 10.069,
|
| 179 |
+
"step": 255
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.02,
|
| 183 |
+
"learning_rate": 1.6993464052287582e-05,
|
| 184 |
+
"loss": 1.0924,
|
| 185 |
+
"step": 260
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.06,
|
| 189 |
+
"learning_rate": 1.7647058823529414e-05,
|
| 190 |
+
"loss": 1.0007,
|
| 191 |
+
"step": 270
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.1,
|
| 195 |
+
"learning_rate": 1.8300653594771242e-05,
|
| 196 |
+
"loss": 1.0054,
|
| 197 |
+
"step": 280
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.14,
|
| 201 |
+
"learning_rate": 1.895424836601307e-05,
|
| 202 |
+
"loss": 0.9365,
|
| 203 |
+
"step": 290
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.18,
|
| 207 |
+
"learning_rate": 1.9607843137254903e-05,
|
| 208 |
+
"loss": 0.8932,
|
| 209 |
+
"step": 300
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.22,
|
| 213 |
+
"learning_rate": 2.0261437908496734e-05,
|
| 214 |
+
"loss": 0.916,
|
| 215 |
+
"step": 310
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.25,
|
| 219 |
+
"learning_rate": 2.0915032679738563e-05,
|
| 220 |
+
"loss": 0.8099,
|
| 221 |
+
"step": 320
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.29,
|
| 225 |
+
"learning_rate": 2.1568627450980395e-05,
|
| 226 |
+
"loss": 0.8211,
|
| 227 |
+
"step": 330
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.33,
|
| 231 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 232 |
+
"loss": 0.6861,
|
| 233 |
+
"step": 340
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.37,
|
| 237 |
+
"learning_rate": 2.2875816993464052e-05,
|
| 238 |
+
"loss": 0.7379,
|
| 239 |
+
"step": 350
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.41,
|
| 243 |
+
"learning_rate": 2.3529411764705884e-05,
|
| 244 |
+
"loss": 0.7349,
|
| 245 |
+
"step": 360
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.45,
|
| 249 |
+
"learning_rate": 2.4183006535947712e-05,
|
| 250 |
+
"loss": 0.6907,
|
| 251 |
+
"step": 370
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.49,
|
| 255 |
+
"learning_rate": 2.4836601307189544e-05,
|
| 256 |
+
"loss": 0.7376,
|
| 257 |
+
"step": 380
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.53,
|
| 261 |
+
"learning_rate": 2.5490196078431373e-05,
|
| 262 |
+
"loss": 0.6827,
|
| 263 |
+
"step": 390
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.57,
|
| 267 |
+
"learning_rate": 2.6143790849673204e-05,
|
| 268 |
+
"loss": 0.6969,
|
| 269 |
+
"step": 400
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.61,
|
| 273 |
+
"learning_rate": 2.6797385620915033e-05,
|
| 274 |
+
"loss": 0.6247,
|
| 275 |
+
"step": 410
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.65,
|
| 279 |
+
"learning_rate": 2.7450980392156865e-05,
|
| 280 |
+
"loss": 0.5244,
|
| 281 |
+
"step": 420
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.69,
|
| 285 |
+
"learning_rate": 2.8104575163398693e-05,
|
| 286 |
+
"loss": 0.7068,
|
| 287 |
+
"step": 430
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.73,
|
| 291 |
+
"learning_rate": 2.8758169934640522e-05,
|
| 292 |
+
"loss": 0.5623,
|
| 293 |
+
"step": 440
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.76,
|
| 297 |
+
"learning_rate": 2.9411764705882354e-05,
|
| 298 |
+
"loss": 0.5931,
|
| 299 |
+
"step": 450
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.8,
|
| 303 |
+
"learning_rate": 3.0065359477124182e-05,
|
| 304 |
+
"loss": 0.6641,
|
| 305 |
+
"step": 460
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.84,
|
| 309 |
+
"learning_rate": 3.0718954248366014e-05,
|
| 310 |
+
"loss": 0.5438,
|
| 311 |
+
"step": 470
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.88,
|
| 315 |
+
"learning_rate": 3.137254901960784e-05,
|
| 316 |
+
"loss": 0.5134,
|
| 317 |
+
"step": 480
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.92,
|
| 321 |
+
"learning_rate": 3.202614379084967e-05,
|
| 322 |
+
"loss": 0.4682,
|
| 323 |
+
"step": 490
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.96,
|
| 327 |
+
"learning_rate": 3.2679738562091506e-05,
|
| 328 |
+
"loss": 0.4762,
|
| 329 |
+
"step": 500
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 2.0,
|
| 333 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 334 |
+
"loss": 0.4558,
|
| 335 |
+
"step": 510
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.0,
|
| 339 |
+
"eval_accuracy": {
|
| 340 |
+
"accuracy": 0.8470588235294118
|
| 341 |
+
},
|
| 342 |
+
"eval_f1": {
|
| 343 |
+
"f1": 0.8510200857474889
|
| 344 |
+
},
|
| 345 |
+
"eval_loss": 0.498457670211792,
|
| 346 |
+
"eval_precision": {
|
| 347 |
+
"precision": 0.8766865984788879
|
| 348 |
+
},
|
| 349 |
+
"eval_recall": {
|
| 350 |
+
"recall": 0.8454231938274491
|
| 351 |
+
},
|
| 352 |
+
"eval_runtime": 10.0517,
|
| 353 |
+
"eval_samples_per_second": 50.738,
|
| 354 |
+
"eval_steps_per_second": 25.369,
|
| 355 |
+
"step": 510
|
| 356 |
+
}
|
| 357 |
+
],
|
| 358 |
+
"logging_steps": 10,
|
| 359 |
+
"max_steps": 7650,
|
| 360 |
+
"num_input_tokens_seen": 0,
|
| 361 |
+
"num_train_epochs": 30,
|
| 362 |
+
"save_steps": 500,
|
| 363 |
+
"total_flos": 3.1617581875052544e+17,
|
| 364 |
+
"train_batch_size": 2,
|
| 365 |
+
"trial_name": null,
|
| 366 |
+
"trial_params": null
|
| 367 |
+
}
|
tmp-checkpoint-510/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:418b85f35316fefe4bc777808f9dff13bb59b607bf93a8504dc8b5709313507f
|
| 3 |
+
size 4728
|
tmp-checkpoint-765/config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "crack",
|
| 13 |
+
"1": "for review",
|
| 14 |
+
"2": "material deterioration",
|
| 15 |
+
"3": "repair",
|
| 16 |
+
"4": "water repair"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"intermediate_size": 3072,
|
| 21 |
+
"label2id": {
|
| 22 |
+
"crack": 0,
|
| 23 |
+
"for review": 1,
|
| 24 |
+
"material deterioration": 2,
|
| 25 |
+
"repair": 3,
|
| 26 |
+
"water repair": 4
|
| 27 |
+
},
|
| 28 |
+
"layer_norm_eps": 1e-12,
|
| 29 |
+
"model_type": "vit",
|
| 30 |
+
"num_attention_heads": 12,
|
| 31 |
+
"num_channels": 3,
|
| 32 |
+
"num_hidden_layers": 12,
|
| 33 |
+
"patch_size": 16,
|
| 34 |
+
"problem_type": "single_label_classification",
|
| 35 |
+
"qkv_bias": true,
|
| 36 |
+
"torch_dtype": "float32",
|
| 37 |
+
"transformers_version": "4.37.2"
|
| 38 |
+
}
|
tmp-checkpoint-765/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc8840a55470dd2e6a6b93487d6ef8d9bc38f276aa0c1c59052c184623f26a4a
|
| 3 |
+
size 343233204
|
tmp-checkpoint-765/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd8d5bfcdbb0ff0423008f642fe704a7ef3181b3cafd9b248ab9d85e23bef1dd
|
| 3 |
+
size 686587322
|
tmp-checkpoint-765/preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": true,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTImageProcessor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 224,
|
| 20 |
+
"width": 224
|
| 21 |
+
}
|
| 22 |
+
}
|
tmp-checkpoint-765/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee19f60ac08292321f1179976bd166d1ed45785f4a7244c14283335e6e126be4
|
| 3 |
+
size 14244
|
tmp-checkpoint-765/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c8984544f991b305fdaa2bdff4442a80ac8b0e93f791609243532fbf334a238
|
| 3 |
+
size 1064
|
tmp-checkpoint-765/trainer_state.json
ADDED
|
@@ -0,0 +1,537 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.44290485978126526,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-765",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 765,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04,
|
| 13 |
+
"learning_rate": 6.535947712418301e-07,
|
| 14 |
+
"loss": 1.6179,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.08,
|
| 19 |
+
"learning_rate": 1.3071895424836602e-06,
|
| 20 |
+
"loss": 1.6223,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.12,
|
| 25 |
+
"learning_rate": 1.96078431372549e-06,
|
| 26 |
+
"loss": 1.6263,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.16,
|
| 31 |
+
"learning_rate": 2.6143790849673204e-06,
|
| 32 |
+
"loss": 1.601,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.2,
|
| 37 |
+
"learning_rate": 3.2679738562091506e-06,
|
| 38 |
+
"loss": 1.6074,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.24,
|
| 43 |
+
"learning_rate": 3.92156862745098e-06,
|
| 44 |
+
"loss": 1.5955,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.27,
|
| 49 |
+
"learning_rate": 4.5751633986928105e-06,
|
| 50 |
+
"loss": 1.5698,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.31,
|
| 55 |
+
"learning_rate": 5.228758169934641e-06,
|
| 56 |
+
"loss": 1.5797,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.35,
|
| 61 |
+
"learning_rate": 5.882352941176471e-06,
|
| 62 |
+
"loss": 1.5531,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.39,
|
| 67 |
+
"learning_rate": 6.535947712418301e-06,
|
| 68 |
+
"loss": 1.5434,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.43,
|
| 73 |
+
"learning_rate": 7.1895424836601305e-06,
|
| 74 |
+
"loss": 1.5433,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.47,
|
| 79 |
+
"learning_rate": 7.84313725490196e-06,
|
| 80 |
+
"loss": 1.5329,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.51,
|
| 85 |
+
"learning_rate": 8.496732026143791e-06,
|
| 86 |
+
"loss": 1.4995,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.55,
|
| 91 |
+
"learning_rate": 9.150326797385621e-06,
|
| 92 |
+
"loss": 1.4875,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.59,
|
| 97 |
+
"learning_rate": 9.803921568627451e-06,
|
| 98 |
+
"loss": 1.4863,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.63,
|
| 103 |
+
"learning_rate": 1.0457516339869281e-05,
|
| 104 |
+
"loss": 1.4209,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.67,
|
| 109 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 110 |
+
"loss": 1.4354,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.71,
|
| 115 |
+
"learning_rate": 1.1764705882352942e-05,
|
| 116 |
+
"loss": 1.3572,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.75,
|
| 121 |
+
"learning_rate": 1.2418300653594772e-05,
|
| 122 |
+
"loss": 1.3714,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.78,
|
| 127 |
+
"learning_rate": 1.3071895424836602e-05,
|
| 128 |
+
"loss": 1.3406,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.82,
|
| 133 |
+
"learning_rate": 1.3725490196078432e-05,
|
| 134 |
+
"loss": 1.2574,
|
| 135 |
+
"step": 210
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.86,
|
| 139 |
+
"learning_rate": 1.4379084967320261e-05,
|
| 140 |
+
"loss": 1.2965,
|
| 141 |
+
"step": 220
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.9,
|
| 145 |
+
"learning_rate": 1.5032679738562091e-05,
|
| 146 |
+
"loss": 1.2021,
|
| 147 |
+
"step": 230
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.94,
|
| 151 |
+
"learning_rate": 1.568627450980392e-05,
|
| 152 |
+
"loss": 1.2072,
|
| 153 |
+
"step": 240
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.98,
|
| 157 |
+
"learning_rate": 1.6339869281045753e-05,
|
| 158 |
+
"loss": 1.1457,
|
| 159 |
+
"step": 250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0,
|
| 163 |
+
"eval_accuracy": {
|
| 164 |
+
"accuracy": 0.6941176470588235
|
| 165 |
+
},
|
| 166 |
+
"eval_f1": {
|
| 167 |
+
"f1": 0.683792091280419
|
| 168 |
+
},
|
| 169 |
+
"eval_loss": 1.158827304840088,
|
| 170 |
+
"eval_precision": {
|
| 171 |
+
"precision": 0.7377494408375539
|
| 172 |
+
},
|
| 173 |
+
"eval_recall": {
|
| 174 |
+
"recall": 0.6941484490686618
|
| 175 |
+
},
|
| 176 |
+
"eval_runtime": 25.3262,
|
| 177 |
+
"eval_samples_per_second": 20.137,
|
| 178 |
+
"eval_steps_per_second": 10.069,
|
| 179 |
+
"step": 255
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.02,
|
| 183 |
+
"learning_rate": 1.6993464052287582e-05,
|
| 184 |
+
"loss": 1.0924,
|
| 185 |
+
"step": 260
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.06,
|
| 189 |
+
"learning_rate": 1.7647058823529414e-05,
|
| 190 |
+
"loss": 1.0007,
|
| 191 |
+
"step": 270
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.1,
|
| 195 |
+
"learning_rate": 1.8300653594771242e-05,
|
| 196 |
+
"loss": 1.0054,
|
| 197 |
+
"step": 280
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.14,
|
| 201 |
+
"learning_rate": 1.895424836601307e-05,
|
| 202 |
+
"loss": 0.9365,
|
| 203 |
+
"step": 290
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.18,
|
| 207 |
+
"learning_rate": 1.9607843137254903e-05,
|
| 208 |
+
"loss": 0.8932,
|
| 209 |
+
"step": 300
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.22,
|
| 213 |
+
"learning_rate": 2.0261437908496734e-05,
|
| 214 |
+
"loss": 0.916,
|
| 215 |
+
"step": 310
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.25,
|
| 219 |
+
"learning_rate": 2.0915032679738563e-05,
|
| 220 |
+
"loss": 0.8099,
|
| 221 |
+
"step": 320
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.29,
|
| 225 |
+
"learning_rate": 2.1568627450980395e-05,
|
| 226 |
+
"loss": 0.8211,
|
| 227 |
+
"step": 330
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.33,
|
| 231 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 232 |
+
"loss": 0.6861,
|
| 233 |
+
"step": 340
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.37,
|
| 237 |
+
"learning_rate": 2.2875816993464052e-05,
|
| 238 |
+
"loss": 0.7379,
|
| 239 |
+
"step": 350
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.41,
|
| 243 |
+
"learning_rate": 2.3529411764705884e-05,
|
| 244 |
+
"loss": 0.7349,
|
| 245 |
+
"step": 360
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.45,
|
| 249 |
+
"learning_rate": 2.4183006535947712e-05,
|
| 250 |
+
"loss": 0.6907,
|
| 251 |
+
"step": 370
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.49,
|
| 255 |
+
"learning_rate": 2.4836601307189544e-05,
|
| 256 |
+
"loss": 0.7376,
|
| 257 |
+
"step": 380
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.53,
|
| 261 |
+
"learning_rate": 2.5490196078431373e-05,
|
| 262 |
+
"loss": 0.6827,
|
| 263 |
+
"step": 390
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.57,
|
| 267 |
+
"learning_rate": 2.6143790849673204e-05,
|
| 268 |
+
"loss": 0.6969,
|
| 269 |
+
"step": 400
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.61,
|
| 273 |
+
"learning_rate": 2.6797385620915033e-05,
|
| 274 |
+
"loss": 0.6247,
|
| 275 |
+
"step": 410
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.65,
|
| 279 |
+
"learning_rate": 2.7450980392156865e-05,
|
| 280 |
+
"loss": 0.5244,
|
| 281 |
+
"step": 420
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.69,
|
| 285 |
+
"learning_rate": 2.8104575163398693e-05,
|
| 286 |
+
"loss": 0.7068,
|
| 287 |
+
"step": 430
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.73,
|
| 291 |
+
"learning_rate": 2.8758169934640522e-05,
|
| 292 |
+
"loss": 0.5623,
|
| 293 |
+
"step": 440
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.76,
|
| 297 |
+
"learning_rate": 2.9411764705882354e-05,
|
| 298 |
+
"loss": 0.5931,
|
| 299 |
+
"step": 450
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.8,
|
| 303 |
+
"learning_rate": 3.0065359477124182e-05,
|
| 304 |
+
"loss": 0.6641,
|
| 305 |
+
"step": 460
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.84,
|
| 309 |
+
"learning_rate": 3.0718954248366014e-05,
|
| 310 |
+
"loss": 0.5438,
|
| 311 |
+
"step": 470
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.88,
|
| 315 |
+
"learning_rate": 3.137254901960784e-05,
|
| 316 |
+
"loss": 0.5134,
|
| 317 |
+
"step": 480
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.92,
|
| 321 |
+
"learning_rate": 3.202614379084967e-05,
|
| 322 |
+
"loss": 0.4682,
|
| 323 |
+
"step": 490
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.96,
|
| 327 |
+
"learning_rate": 3.2679738562091506e-05,
|
| 328 |
+
"loss": 0.4762,
|
| 329 |
+
"step": 500
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 2.0,
|
| 333 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 334 |
+
"loss": 0.4558,
|
| 335 |
+
"step": 510
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.0,
|
| 339 |
+
"eval_accuracy": {
|
| 340 |
+
"accuracy": 0.8470588235294118
|
| 341 |
+
},
|
| 342 |
+
"eval_f1": {
|
| 343 |
+
"f1": 0.8510200857474889
|
| 344 |
+
},
|
| 345 |
+
"eval_loss": 0.498457670211792,
|
| 346 |
+
"eval_precision": {
|
| 347 |
+
"precision": 0.8766865984788879
|
| 348 |
+
},
|
| 349 |
+
"eval_recall": {
|
| 350 |
+
"recall": 0.8454231938274491
|
| 351 |
+
},
|
| 352 |
+
"eval_runtime": 10.0517,
|
| 353 |
+
"eval_samples_per_second": 50.738,
|
| 354 |
+
"eval_steps_per_second": 25.369,
|
| 355 |
+
"step": 510
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 2.04,
|
| 359 |
+
"learning_rate": 3.3986928104575163e-05,
|
| 360 |
+
"loss": 0.4716,
|
| 361 |
+
"step": 520
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 2.08,
|
| 365 |
+
"learning_rate": 3.464052287581699e-05,
|
| 366 |
+
"loss": 0.5375,
|
| 367 |
+
"step": 530
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 2.12,
|
| 371 |
+
"learning_rate": 3.529411764705883e-05,
|
| 372 |
+
"loss": 0.4725,
|
| 373 |
+
"step": 540
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 2.16,
|
| 377 |
+
"learning_rate": 3.5947712418300656e-05,
|
| 378 |
+
"loss": 0.4804,
|
| 379 |
+
"step": 550
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 2.2,
|
| 383 |
+
"learning_rate": 3.6601307189542484e-05,
|
| 384 |
+
"loss": 0.3504,
|
| 385 |
+
"step": 560
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 2.24,
|
| 389 |
+
"learning_rate": 3.725490196078432e-05,
|
| 390 |
+
"loss": 0.4049,
|
| 391 |
+
"step": 570
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.27,
|
| 395 |
+
"learning_rate": 3.790849673202614e-05,
|
| 396 |
+
"loss": 0.4309,
|
| 397 |
+
"step": 580
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 2.31,
|
| 401 |
+
"learning_rate": 3.8562091503267977e-05,
|
| 402 |
+
"loss": 0.4938,
|
| 403 |
+
"step": 590
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 2.35,
|
| 407 |
+
"learning_rate": 3.9215686274509805e-05,
|
| 408 |
+
"loss": 0.5427,
|
| 409 |
+
"step": 600
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 2.39,
|
| 413 |
+
"learning_rate": 3.986928104575164e-05,
|
| 414 |
+
"loss": 0.4719,
|
| 415 |
+
"step": 610
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 2.43,
|
| 419 |
+
"learning_rate": 4.052287581699347e-05,
|
| 420 |
+
"loss": 0.4627,
|
| 421 |
+
"step": 620
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 2.47,
|
| 425 |
+
"learning_rate": 4.11764705882353e-05,
|
| 426 |
+
"loss": 0.5277,
|
| 427 |
+
"step": 630
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 2.51,
|
| 431 |
+
"learning_rate": 4.1830065359477126e-05,
|
| 432 |
+
"loss": 0.3109,
|
| 433 |
+
"step": 640
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 2.55,
|
| 437 |
+
"learning_rate": 4.2483660130718954e-05,
|
| 438 |
+
"loss": 0.3799,
|
| 439 |
+
"step": 650
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 2.59,
|
| 443 |
+
"learning_rate": 4.313725490196079e-05,
|
| 444 |
+
"loss": 0.3515,
|
| 445 |
+
"step": 660
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 2.63,
|
| 449 |
+
"learning_rate": 4.379084967320262e-05,
|
| 450 |
+
"loss": 0.3722,
|
| 451 |
+
"step": 670
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 2.67,
|
| 455 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 456 |
+
"loss": 0.5337,
|
| 457 |
+
"step": 680
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 2.71,
|
| 461 |
+
"learning_rate": 4.5098039215686275e-05,
|
| 462 |
+
"loss": 0.3263,
|
| 463 |
+
"step": 690
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 2.75,
|
| 467 |
+
"learning_rate": 4.5751633986928104e-05,
|
| 468 |
+
"loss": 0.3703,
|
| 469 |
+
"step": 700
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 2.78,
|
| 473 |
+
"learning_rate": 4.640522875816994e-05,
|
| 474 |
+
"loss": 0.4498,
|
| 475 |
+
"step": 710
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 2.82,
|
| 479 |
+
"learning_rate": 4.705882352941177e-05,
|
| 480 |
+
"loss": 0.4109,
|
| 481 |
+
"step": 720
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 2.86,
|
| 485 |
+
"learning_rate": 4.77124183006536e-05,
|
| 486 |
+
"loss": 0.2953,
|
| 487 |
+
"step": 730
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 2.9,
|
| 491 |
+
"learning_rate": 4.8366013071895424e-05,
|
| 492 |
+
"loss": 0.4033,
|
| 493 |
+
"step": 740
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 2.94,
|
| 497 |
+
"learning_rate": 4.901960784313725e-05,
|
| 498 |
+
"loss": 0.4186,
|
| 499 |
+
"step": 750
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 2.98,
|
| 503 |
+
"learning_rate": 4.967320261437909e-05,
|
| 504 |
+
"loss": 0.3368,
|
| 505 |
+
"step": 760
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.0,
|
| 509 |
+
"eval_accuracy": {
|
| 510 |
+
"accuracy": 0.8705882352941177
|
| 511 |
+
},
|
| 512 |
+
"eval_f1": {
|
| 513 |
+
"f1": 0.8713991689505157
|
| 514 |
+
},
|
| 515 |
+
"eval_loss": 0.44290485978126526,
|
| 516 |
+
"eval_precision": {
|
| 517 |
+
"precision": 0.8884985994397759
|
| 518 |
+
},
|
| 519 |
+
"eval_recall": {
|
| 520 |
+
"recall": 0.8711626139817629
|
| 521 |
+
},
|
| 522 |
+
"eval_runtime": 9.9506,
|
| 523 |
+
"eval_samples_per_second": 51.253,
|
| 524 |
+
"eval_steps_per_second": 25.626,
|
| 525 |
+
"step": 765
|
| 526 |
+
}
|
| 527 |
+
],
|
| 528 |
+
"logging_steps": 10,
|
| 529 |
+
"max_steps": 7650,
|
| 530 |
+
"num_input_tokens_seen": 0,
|
| 531 |
+
"num_train_epochs": 30,
|
| 532 |
+
"save_steps": 500,
|
| 533 |
+
"total_flos": 4.7426372812578816e+17,
|
| 534 |
+
"train_batch_size": 2,
|
| 535 |
+
"trial_name": null,
|
| 536 |
+
"trial_params": null
|
| 537 |
+
}
|
tmp-checkpoint-765/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:418b85f35316fefe4bc777808f9dff13bb59b607bf93a8504dc8b5709313507f
|
| 3 |
+
size 4728
|