Training in progress, epoch 11
Browse files- model.safetensors +1 -1
- tmp-checkpoint-1785/config.json +38 -0
- tmp-checkpoint-1785/model.safetensors +3 -0
- tmp-checkpoint-1785/optimizer.pt +3 -0
- tmp-checkpoint-1785/preprocessor_config.json +22 -0
- tmp-checkpoint-1785/rng_state.pth +3 -0
- tmp-checkpoint-1785/scheduler.pt +3 -0
- tmp-checkpoint-1785/trainer_state.json +1229 -0
- tmp-checkpoint-1785/training_args.bin +3 -0
- tmp-checkpoint-2040/config.json +38 -0
- tmp-checkpoint-2040/model.safetensors +3 -0
- tmp-checkpoint-2040/optimizer.pt +3 -0
- tmp-checkpoint-2040/preprocessor_config.json +22 -0
- tmp-checkpoint-2040/rng_state.pth +3 -0
- tmp-checkpoint-2040/scheduler.pt +3 -0
- tmp-checkpoint-2040/trainer_state.json +1405 -0
- tmp-checkpoint-2040/training_args.bin +3 -0
- tmp-checkpoint-2295/config.json +38 -0
- tmp-checkpoint-2295/model.safetensors +3 -0
- tmp-checkpoint-2295/optimizer.pt +3 -0
- tmp-checkpoint-2295/preprocessor_config.json +22 -0
- tmp-checkpoint-2295/rng_state.pth +3 -0
- tmp-checkpoint-2295/scheduler.pt +3 -0
- tmp-checkpoint-2295/trainer_state.json +1575 -0
- tmp-checkpoint-2295/training_args.bin +3 -0
- tmp-checkpoint-2550/config.json +38 -0
- tmp-checkpoint-2550/model.safetensors +3 -0
- tmp-checkpoint-2550/optimizer.pt +3 -0
- tmp-checkpoint-2550/preprocessor_config.json +22 -0
- tmp-checkpoint-2550/rng_state.pth +3 -0
- tmp-checkpoint-2550/scheduler.pt +3 -0
- tmp-checkpoint-2550/trainer_state.json +1751 -0
- tmp-checkpoint-2550/training_args.bin +3 -0
- tmp-checkpoint-2805/config.json +38 -0
- tmp-checkpoint-2805/model.safetensors +3 -0
- tmp-checkpoint-2805/optimizer.pt +3 -0
- tmp-checkpoint-2805/preprocessor_config.json +22 -0
- tmp-checkpoint-2805/rng_state.pth +3 -0
- tmp-checkpoint-2805/scheduler.pt +3 -0
- tmp-checkpoint-2805/trainer_state.json +1921 -0
- tmp-checkpoint-2805/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:fd0bdc5483d786e2b3f7c85777a9c669be94de7087b07a6141776eb245d18eb9
|
| 3 |
size 343233204
|
tmp-checkpoint-1785/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-1785/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e3c93d17c523d3fb9fa8e71795519468a1cbb4da362202d60f90781cef73c631
|
| 3 |
+
size 343233204
|
tmp-checkpoint-1785/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:490bc59a437475fe48c1f80c63d351f330f3cf1c5065b3fd84be13d30746a31f
|
| 3 |
+
size 686587322
|
tmp-checkpoint-1785/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-1785/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e622fae879a8d9eb9b2344460e495c3f37f31358f19d3d47cedcadd4ab4d6e14
|
| 3 |
+
size 14244
|
tmp-checkpoint-1785/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1722b165a59712e461f5b9774b72af9f4c592277b59741488608138f7864791b
|
| 3 |
+
size 1064
|
tmp-checkpoint-1785/trainer_state.json
ADDED
|
@@ -0,0 +1,1229 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.36420056223869324,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-1785",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1785,
|
| 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 |
+
"epoch": 3.02,
|
| 529 |
+
"learning_rate": 4.9963689179375456e-05,
|
| 530 |
+
"loss": 0.3347,
|
| 531 |
+
"step": 770
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 3.06,
|
| 535 |
+
"learning_rate": 4.9891067538126364e-05,
|
| 536 |
+
"loss": 0.3339,
|
| 537 |
+
"step": 780
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 3.1,
|
| 541 |
+
"learning_rate": 4.981844589687727e-05,
|
| 542 |
+
"loss": 0.5573,
|
| 543 |
+
"step": 790
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 3.14,
|
| 547 |
+
"learning_rate": 4.974582425562818e-05,
|
| 548 |
+
"loss": 0.5463,
|
| 549 |
+
"step": 800
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.18,
|
| 553 |
+
"learning_rate": 4.967320261437909e-05,
|
| 554 |
+
"loss": 0.3453,
|
| 555 |
+
"step": 810
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.22,
|
| 559 |
+
"learning_rate": 4.9600580973129996e-05,
|
| 560 |
+
"loss": 0.3814,
|
| 561 |
+
"step": 820
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.25,
|
| 565 |
+
"learning_rate": 4.9527959331880904e-05,
|
| 566 |
+
"loss": 0.4434,
|
| 567 |
+
"step": 830
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 3.29,
|
| 571 |
+
"learning_rate": 4.945533769063181e-05,
|
| 572 |
+
"loss": 0.3675,
|
| 573 |
+
"step": 840
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 3.33,
|
| 577 |
+
"learning_rate": 4.938271604938271e-05,
|
| 578 |
+
"loss": 0.5386,
|
| 579 |
+
"step": 850
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 3.37,
|
| 583 |
+
"learning_rate": 4.931009440813363e-05,
|
| 584 |
+
"loss": 0.2692,
|
| 585 |
+
"step": 860
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.41,
|
| 589 |
+
"learning_rate": 4.9237472766884536e-05,
|
| 590 |
+
"loss": 0.4815,
|
| 591 |
+
"step": 870
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.45,
|
| 595 |
+
"learning_rate": 4.9164851125635444e-05,
|
| 596 |
+
"loss": 0.4178,
|
| 597 |
+
"step": 880
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.49,
|
| 601 |
+
"learning_rate": 4.909222948438635e-05,
|
| 602 |
+
"loss": 0.3618,
|
| 603 |
+
"step": 890
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.53,
|
| 607 |
+
"learning_rate": 4.901960784313725e-05,
|
| 608 |
+
"loss": 0.6841,
|
| 609 |
+
"step": 900
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 3.57,
|
| 613 |
+
"learning_rate": 4.894698620188817e-05,
|
| 614 |
+
"loss": 0.471,
|
| 615 |
+
"step": 910
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.61,
|
| 619 |
+
"learning_rate": 4.8874364560639076e-05,
|
| 620 |
+
"loss": 0.313,
|
| 621 |
+
"step": 920
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 3.65,
|
| 625 |
+
"learning_rate": 4.8801742919389983e-05,
|
| 626 |
+
"loss": 0.2437,
|
| 627 |
+
"step": 930
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 3.69,
|
| 631 |
+
"learning_rate": 4.8729121278140885e-05,
|
| 632 |
+
"loss": 0.4748,
|
| 633 |
+
"step": 940
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.73,
|
| 637 |
+
"learning_rate": 4.865649963689179e-05,
|
| 638 |
+
"loss": 0.5519,
|
| 639 |
+
"step": 950
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.76,
|
| 643 |
+
"learning_rate": 4.85838779956427e-05,
|
| 644 |
+
"loss": 0.3051,
|
| 645 |
+
"step": 960
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.8,
|
| 649 |
+
"learning_rate": 4.8511256354393615e-05,
|
| 650 |
+
"loss": 0.3847,
|
| 651 |
+
"step": 970
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 3.84,
|
| 655 |
+
"learning_rate": 4.843863471314452e-05,
|
| 656 |
+
"loss": 0.2454,
|
| 657 |
+
"step": 980
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.88,
|
| 661 |
+
"learning_rate": 4.8366013071895424e-05,
|
| 662 |
+
"loss": 0.379,
|
| 663 |
+
"step": 990
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 3.92,
|
| 667 |
+
"learning_rate": 4.829339143064633e-05,
|
| 668 |
+
"loss": 0.4485,
|
| 669 |
+
"step": 1000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 3.96,
|
| 673 |
+
"learning_rate": 4.822076978939724e-05,
|
| 674 |
+
"loss": 0.2484,
|
| 675 |
+
"step": 1010
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.0,
|
| 679 |
+
"learning_rate": 4.814814814814815e-05,
|
| 680 |
+
"loss": 0.4402,
|
| 681 |
+
"step": 1020
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 4.0,
|
| 685 |
+
"eval_accuracy": {
|
| 686 |
+
"accuracy": 0.8784313725490196
|
| 687 |
+
},
|
| 688 |
+
"eval_f1": {
|
| 689 |
+
"f1": 0.880471094803992
|
| 690 |
+
},
|
| 691 |
+
"eval_loss": 0.3946979343891144,
|
| 692 |
+
"eval_precision": {
|
| 693 |
+
"precision": 0.8952890487430221
|
| 694 |
+
},
|
| 695 |
+
"eval_recall": {
|
| 696 |
+
"recall": 0.8776488582339645
|
| 697 |
+
},
|
| 698 |
+
"eval_runtime": 10.1771,
|
| 699 |
+
"eval_samples_per_second": 50.113,
|
| 700 |
+
"eval_steps_per_second": 25.056,
|
| 701 |
+
"step": 1020
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 4.04,
|
| 705 |
+
"learning_rate": 4.807552650689906e-05,
|
| 706 |
+
"loss": 0.3494,
|
| 707 |
+
"step": 1030
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 4.08,
|
| 711 |
+
"learning_rate": 4.8002904865649964e-05,
|
| 712 |
+
"loss": 0.3327,
|
| 713 |
+
"step": 1040
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 4.12,
|
| 717 |
+
"learning_rate": 4.793028322440087e-05,
|
| 718 |
+
"loss": 0.2898,
|
| 719 |
+
"step": 1050
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 4.16,
|
| 723 |
+
"learning_rate": 4.785766158315178e-05,
|
| 724 |
+
"loss": 0.3965,
|
| 725 |
+
"step": 1060
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 4.2,
|
| 729 |
+
"learning_rate": 4.778503994190269e-05,
|
| 730 |
+
"loss": 0.3607,
|
| 731 |
+
"step": 1070
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.24,
|
| 735 |
+
"learning_rate": 4.77124183006536e-05,
|
| 736 |
+
"loss": 0.3714,
|
| 737 |
+
"step": 1080
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.27,
|
| 741 |
+
"learning_rate": 4.7639796659404504e-05,
|
| 742 |
+
"loss": 0.1796,
|
| 743 |
+
"step": 1090
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 4.31,
|
| 747 |
+
"learning_rate": 4.756717501815541e-05,
|
| 748 |
+
"loss": 0.1874,
|
| 749 |
+
"step": 1100
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 4.35,
|
| 753 |
+
"learning_rate": 4.749455337690632e-05,
|
| 754 |
+
"loss": 0.3462,
|
| 755 |
+
"step": 1110
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 4.39,
|
| 759 |
+
"learning_rate": 4.742193173565723e-05,
|
| 760 |
+
"loss": 0.2082,
|
| 761 |
+
"step": 1120
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 4.43,
|
| 765 |
+
"learning_rate": 4.7349310094408136e-05,
|
| 766 |
+
"loss": 0.4531,
|
| 767 |
+
"step": 1130
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 4.47,
|
| 771 |
+
"learning_rate": 4.7276688453159044e-05,
|
| 772 |
+
"loss": 0.3951,
|
| 773 |
+
"step": 1140
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.51,
|
| 777 |
+
"learning_rate": 4.720406681190995e-05,
|
| 778 |
+
"loss": 0.2337,
|
| 779 |
+
"step": 1150
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.55,
|
| 783 |
+
"learning_rate": 4.713144517066086e-05,
|
| 784 |
+
"loss": 0.2118,
|
| 785 |
+
"step": 1160
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 4.59,
|
| 789 |
+
"learning_rate": 4.705882352941177e-05,
|
| 790 |
+
"loss": 0.373,
|
| 791 |
+
"step": 1170
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 4.63,
|
| 795 |
+
"learning_rate": 4.6986201888162675e-05,
|
| 796 |
+
"loss": 0.1484,
|
| 797 |
+
"step": 1180
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 4.67,
|
| 801 |
+
"learning_rate": 4.691358024691358e-05,
|
| 802 |
+
"loss": 0.2568,
|
| 803 |
+
"step": 1190
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 4.71,
|
| 807 |
+
"learning_rate": 4.684095860566449e-05,
|
| 808 |
+
"loss": 0.3067,
|
| 809 |
+
"step": 1200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 4.75,
|
| 813 |
+
"learning_rate": 4.67683369644154e-05,
|
| 814 |
+
"loss": 0.2007,
|
| 815 |
+
"step": 1210
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.78,
|
| 819 |
+
"learning_rate": 4.669571532316631e-05,
|
| 820 |
+
"loss": 0.2493,
|
| 821 |
+
"step": 1220
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.82,
|
| 825 |
+
"learning_rate": 4.6623093681917215e-05,
|
| 826 |
+
"loss": 0.2717,
|
| 827 |
+
"step": 1230
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 4.86,
|
| 831 |
+
"learning_rate": 4.6550472040668116e-05,
|
| 832 |
+
"loss": 0.4134,
|
| 833 |
+
"step": 1240
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 4.9,
|
| 837 |
+
"learning_rate": 4.647785039941903e-05,
|
| 838 |
+
"loss": 0.2417,
|
| 839 |
+
"step": 1250
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 4.94,
|
| 843 |
+
"learning_rate": 4.640522875816994e-05,
|
| 844 |
+
"loss": 0.2573,
|
| 845 |
+
"step": 1260
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 4.98,
|
| 849 |
+
"learning_rate": 4.633260711692085e-05,
|
| 850 |
+
"loss": 0.3937,
|
| 851 |
+
"step": 1270
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 5.0,
|
| 855 |
+
"eval_accuracy": {
|
| 856 |
+
"accuracy": 0.8725490196078431
|
| 857 |
+
},
|
| 858 |
+
"eval_f1": {
|
| 859 |
+
"f1": 0.8703780158419334
|
| 860 |
+
},
|
| 861 |
+
"eval_loss": 0.4377364218235016,
|
| 862 |
+
"eval_precision": {
|
| 863 |
+
"precision": 0.8900760323128502
|
| 864 |
+
},
|
| 865 |
+
"eval_recall": {
|
| 866 |
+
"recall": 0.8691138648585458
|
| 867 |
+
},
|
| 868 |
+
"eval_runtime": 10.1762,
|
| 869 |
+
"eval_samples_per_second": 50.117,
|
| 870 |
+
"eval_steps_per_second": 25.058,
|
| 871 |
+
"step": 1275
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 5.02,
|
| 875 |
+
"learning_rate": 4.6259985475671755e-05,
|
| 876 |
+
"loss": 0.165,
|
| 877 |
+
"step": 1280
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 5.06,
|
| 881 |
+
"learning_rate": 4.6187363834422656e-05,
|
| 882 |
+
"loss": 0.1452,
|
| 883 |
+
"step": 1290
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 5.1,
|
| 887 |
+
"learning_rate": 4.6114742193173564e-05,
|
| 888 |
+
"loss": 0.1548,
|
| 889 |
+
"step": 1300
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 5.14,
|
| 893 |
+
"learning_rate": 4.604212055192448e-05,
|
| 894 |
+
"loss": 0.4443,
|
| 895 |
+
"step": 1310
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 5.18,
|
| 899 |
+
"learning_rate": 4.5969498910675387e-05,
|
| 900 |
+
"loss": 0.4322,
|
| 901 |
+
"step": 1320
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 5.22,
|
| 905 |
+
"learning_rate": 4.5896877269426294e-05,
|
| 906 |
+
"loss": 0.2355,
|
| 907 |
+
"step": 1330
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 5.25,
|
| 911 |
+
"learning_rate": 4.5824255628177196e-05,
|
| 912 |
+
"loss": 0.1949,
|
| 913 |
+
"step": 1340
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 5.29,
|
| 917 |
+
"learning_rate": 4.5751633986928104e-05,
|
| 918 |
+
"loss": 0.1959,
|
| 919 |
+
"step": 1350
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.33,
|
| 923 |
+
"learning_rate": 4.567901234567901e-05,
|
| 924 |
+
"loss": 0.3841,
|
| 925 |
+
"step": 1360
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 5.37,
|
| 929 |
+
"learning_rate": 4.5606390704429926e-05,
|
| 930 |
+
"loss": 0.1401,
|
| 931 |
+
"step": 1370
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 5.41,
|
| 935 |
+
"learning_rate": 4.5533769063180834e-05,
|
| 936 |
+
"loss": 0.1148,
|
| 937 |
+
"step": 1380
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 5.45,
|
| 941 |
+
"learning_rate": 4.5461147421931735e-05,
|
| 942 |
+
"loss": 0.234,
|
| 943 |
+
"step": 1390
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 5.49,
|
| 947 |
+
"learning_rate": 4.538852578068264e-05,
|
| 948 |
+
"loss": 0.3899,
|
| 949 |
+
"step": 1400
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 5.53,
|
| 953 |
+
"learning_rate": 4.531590413943355e-05,
|
| 954 |
+
"loss": 0.4873,
|
| 955 |
+
"step": 1410
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 5.57,
|
| 959 |
+
"learning_rate": 4.5243282498184466e-05,
|
| 960 |
+
"loss": 0.2483,
|
| 961 |
+
"step": 1420
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.61,
|
| 965 |
+
"learning_rate": 4.5170660856935374e-05,
|
| 966 |
+
"loss": 0.2654,
|
| 967 |
+
"step": 1430
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 5.65,
|
| 971 |
+
"learning_rate": 4.5098039215686275e-05,
|
| 972 |
+
"loss": 0.2035,
|
| 973 |
+
"step": 1440
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 5.69,
|
| 977 |
+
"learning_rate": 4.502541757443718e-05,
|
| 978 |
+
"loss": 0.3744,
|
| 979 |
+
"step": 1450
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 5.73,
|
| 983 |
+
"learning_rate": 4.495279593318809e-05,
|
| 984 |
+
"loss": 0.2139,
|
| 985 |
+
"step": 1460
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 5.76,
|
| 989 |
+
"learning_rate": 4.4880174291939e-05,
|
| 990 |
+
"loss": 0.2212,
|
| 991 |
+
"step": 1470
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 5.8,
|
| 995 |
+
"learning_rate": 4.480755265068991e-05,
|
| 996 |
+
"loss": 0.376,
|
| 997 |
+
"step": 1480
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 5.84,
|
| 1001 |
+
"learning_rate": 4.4734931009440815e-05,
|
| 1002 |
+
"loss": 0.2399,
|
| 1003 |
+
"step": 1490
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.88,
|
| 1007 |
+
"learning_rate": 4.466230936819172e-05,
|
| 1008 |
+
"loss": 0.3534,
|
| 1009 |
+
"step": 1500
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 5.92,
|
| 1013 |
+
"learning_rate": 4.458968772694263e-05,
|
| 1014 |
+
"loss": 0.3667,
|
| 1015 |
+
"step": 1510
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 5.96,
|
| 1019 |
+
"learning_rate": 4.451706608569354e-05,
|
| 1020 |
+
"loss": 0.496,
|
| 1021 |
+
"step": 1520
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 6.0,
|
| 1025 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 1026 |
+
"loss": 0.4042,
|
| 1027 |
+
"step": 1530
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 6.0,
|
| 1031 |
+
"eval_accuracy": {
|
| 1032 |
+
"accuracy": 0.8862745098039215
|
| 1033 |
+
},
|
| 1034 |
+
"eval_f1": {
|
| 1035 |
+
"f1": 0.8859196308355288
|
| 1036 |
+
},
|
| 1037 |
+
"eval_loss": 0.39891302585601807,
|
| 1038 |
+
"eval_precision": {
|
| 1039 |
+
"precision": 0.8972093144987638
|
| 1040 |
+
},
|
| 1041 |
+
"eval_recall": {
|
| 1042 |
+
"recall": 0.8834955381497934
|
| 1043 |
+
},
|
| 1044 |
+
"eval_runtime": 10.1174,
|
| 1045 |
+
"eval_samples_per_second": 50.408,
|
| 1046 |
+
"eval_steps_per_second": 25.204,
|
| 1047 |
+
"step": 1530
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 6.04,
|
| 1051 |
+
"learning_rate": 4.4371822803195355e-05,
|
| 1052 |
+
"loss": 0.2637,
|
| 1053 |
+
"step": 1540
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 6.08,
|
| 1057 |
+
"learning_rate": 4.429920116194626e-05,
|
| 1058 |
+
"loss": 0.2821,
|
| 1059 |
+
"step": 1550
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 6.12,
|
| 1063 |
+
"learning_rate": 4.422657952069717e-05,
|
| 1064 |
+
"loss": 0.1201,
|
| 1065 |
+
"step": 1560
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 6.16,
|
| 1069 |
+
"learning_rate": 4.415395787944808e-05,
|
| 1070 |
+
"loss": 0.262,
|
| 1071 |
+
"step": 1570
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 6.2,
|
| 1075 |
+
"learning_rate": 4.4081336238198986e-05,
|
| 1076 |
+
"loss": 0.2064,
|
| 1077 |
+
"step": 1580
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 6.24,
|
| 1081 |
+
"learning_rate": 4.400871459694989e-05,
|
| 1082 |
+
"loss": 0.1815,
|
| 1083 |
+
"step": 1590
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 6.27,
|
| 1087 |
+
"learning_rate": 4.39360929557008e-05,
|
| 1088 |
+
"loss": 0.1959,
|
| 1089 |
+
"step": 1600
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 6.31,
|
| 1093 |
+
"learning_rate": 4.386347131445171e-05,
|
| 1094 |
+
"loss": 0.1959,
|
| 1095 |
+
"step": 1610
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.35,
|
| 1099 |
+
"learning_rate": 4.379084967320262e-05,
|
| 1100 |
+
"loss": 0.2398,
|
| 1101 |
+
"step": 1620
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.39,
|
| 1105 |
+
"learning_rate": 4.3718228031953526e-05,
|
| 1106 |
+
"loss": 0.2038,
|
| 1107 |
+
"step": 1630
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 6.43,
|
| 1111 |
+
"learning_rate": 4.364560639070443e-05,
|
| 1112 |
+
"loss": 0.2004,
|
| 1113 |
+
"step": 1640
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.47,
|
| 1117 |
+
"learning_rate": 4.357298474945534e-05,
|
| 1118 |
+
"loss": 0.0377,
|
| 1119 |
+
"step": 1650
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 6.51,
|
| 1123 |
+
"learning_rate": 4.350036310820625e-05,
|
| 1124 |
+
"loss": 0.2022,
|
| 1125 |
+
"step": 1660
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 6.55,
|
| 1129 |
+
"learning_rate": 4.342774146695716e-05,
|
| 1130 |
+
"loss": 0.4403,
|
| 1131 |
+
"step": 1670
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 6.59,
|
| 1135 |
+
"learning_rate": 4.3355119825708066e-05,
|
| 1136 |
+
"loss": 0.2577,
|
| 1137 |
+
"step": 1680
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.63,
|
| 1141 |
+
"learning_rate": 4.328249818445897e-05,
|
| 1142 |
+
"loss": 0.3312,
|
| 1143 |
+
"step": 1690
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.67,
|
| 1147 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 1148 |
+
"loss": 0.307,
|
| 1149 |
+
"step": 1700
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 6.71,
|
| 1153 |
+
"learning_rate": 4.313725490196079e-05,
|
| 1154 |
+
"loss": 0.0961,
|
| 1155 |
+
"step": 1710
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.75,
|
| 1159 |
+
"learning_rate": 4.30646332607117e-05,
|
| 1160 |
+
"loss": 0.3266,
|
| 1161 |
+
"step": 1720
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 6.78,
|
| 1165 |
+
"learning_rate": 4.2992011619462605e-05,
|
| 1166 |
+
"loss": 0.2601,
|
| 1167 |
+
"step": 1730
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 6.82,
|
| 1171 |
+
"learning_rate": 4.291938997821351e-05,
|
| 1172 |
+
"loss": 0.2382,
|
| 1173 |
+
"step": 1740
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 6.86,
|
| 1177 |
+
"learning_rate": 4.2846768336964415e-05,
|
| 1178 |
+
"loss": 0.1469,
|
| 1179 |
+
"step": 1750
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.9,
|
| 1183 |
+
"learning_rate": 4.277414669571532e-05,
|
| 1184 |
+
"loss": 0.2158,
|
| 1185 |
+
"step": 1760
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.94,
|
| 1189 |
+
"learning_rate": 4.270152505446624e-05,
|
| 1190 |
+
"loss": 0.4392,
|
| 1191 |
+
"step": 1770
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 6.98,
|
| 1195 |
+
"learning_rate": 4.262890341321714e-05,
|
| 1196 |
+
"loss": 0.4321,
|
| 1197 |
+
"step": 1780
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 7.0,
|
| 1201 |
+
"eval_accuracy": {
|
| 1202 |
+
"accuracy": 0.8901960784313725
|
| 1203 |
+
},
|
| 1204 |
+
"eval_f1": {
|
| 1205 |
+
"f1": 0.8896787277785286
|
| 1206 |
+
},
|
| 1207 |
+
"eval_loss": 0.36420056223869324,
|
| 1208 |
+
"eval_precision": {
|
| 1209 |
+
"precision": 0.9002265766283616
|
| 1210 |
+
},
|
| 1211 |
+
"eval_recall": {
|
| 1212 |
+
"recall": 0.8878911425453978
|
| 1213 |
+
},
|
| 1214 |
+
"eval_runtime": 10.2347,
|
| 1215 |
+
"eval_samples_per_second": 49.83,
|
| 1216 |
+
"eval_steps_per_second": 24.915,
|
| 1217 |
+
"step": 1785
|
| 1218 |
+
}
|
| 1219 |
+
],
|
| 1220 |
+
"logging_steps": 10,
|
| 1221 |
+
"max_steps": 7650,
|
| 1222 |
+
"num_input_tokens_seen": 0,
|
| 1223 |
+
"num_train_epochs": 30,
|
| 1224 |
+
"save_steps": 500,
|
| 1225 |
+
"total_flos": 1.106615365626839e+18,
|
| 1226 |
+
"train_batch_size": 2,
|
| 1227 |
+
"trial_name": null,
|
| 1228 |
+
"trial_params": null
|
| 1229 |
+
}
|
tmp-checkpoint-1785/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-2040/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-2040/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:acd9b607bc70bc4e9b48c0b7651e2d81495db1f6a17fad66e9d224a77db290dc
|
| 3 |
+
size 343233204
|
tmp-checkpoint-2040/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63003ab30e89038a1e5d663edf0023f1f1517f180708d31b0188ec66bbdc1779
|
| 3 |
+
size 686587322
|
tmp-checkpoint-2040/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-2040/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6c7e9136c4fd326a1bc79722347af2b4c1ef317163712d8002b16cd82c8df5f3
|
| 3 |
+
size 14244
|
tmp-checkpoint-2040/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bf914d353649793424be575e67eded8a2eb9a15aecbbd475e56c75be1491f22d
|
| 3 |
+
size 1064
|
tmp-checkpoint-2040/trainer_state.json
ADDED
|
@@ -0,0 +1,1405 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.36420056223869324,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-1785",
|
| 4 |
+
"epoch": 8.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2040,
|
| 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 |
+
"epoch": 3.02,
|
| 529 |
+
"learning_rate": 4.9963689179375456e-05,
|
| 530 |
+
"loss": 0.3347,
|
| 531 |
+
"step": 770
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 3.06,
|
| 535 |
+
"learning_rate": 4.9891067538126364e-05,
|
| 536 |
+
"loss": 0.3339,
|
| 537 |
+
"step": 780
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 3.1,
|
| 541 |
+
"learning_rate": 4.981844589687727e-05,
|
| 542 |
+
"loss": 0.5573,
|
| 543 |
+
"step": 790
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 3.14,
|
| 547 |
+
"learning_rate": 4.974582425562818e-05,
|
| 548 |
+
"loss": 0.5463,
|
| 549 |
+
"step": 800
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.18,
|
| 553 |
+
"learning_rate": 4.967320261437909e-05,
|
| 554 |
+
"loss": 0.3453,
|
| 555 |
+
"step": 810
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.22,
|
| 559 |
+
"learning_rate": 4.9600580973129996e-05,
|
| 560 |
+
"loss": 0.3814,
|
| 561 |
+
"step": 820
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.25,
|
| 565 |
+
"learning_rate": 4.9527959331880904e-05,
|
| 566 |
+
"loss": 0.4434,
|
| 567 |
+
"step": 830
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 3.29,
|
| 571 |
+
"learning_rate": 4.945533769063181e-05,
|
| 572 |
+
"loss": 0.3675,
|
| 573 |
+
"step": 840
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 3.33,
|
| 577 |
+
"learning_rate": 4.938271604938271e-05,
|
| 578 |
+
"loss": 0.5386,
|
| 579 |
+
"step": 850
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 3.37,
|
| 583 |
+
"learning_rate": 4.931009440813363e-05,
|
| 584 |
+
"loss": 0.2692,
|
| 585 |
+
"step": 860
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.41,
|
| 589 |
+
"learning_rate": 4.9237472766884536e-05,
|
| 590 |
+
"loss": 0.4815,
|
| 591 |
+
"step": 870
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.45,
|
| 595 |
+
"learning_rate": 4.9164851125635444e-05,
|
| 596 |
+
"loss": 0.4178,
|
| 597 |
+
"step": 880
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.49,
|
| 601 |
+
"learning_rate": 4.909222948438635e-05,
|
| 602 |
+
"loss": 0.3618,
|
| 603 |
+
"step": 890
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.53,
|
| 607 |
+
"learning_rate": 4.901960784313725e-05,
|
| 608 |
+
"loss": 0.6841,
|
| 609 |
+
"step": 900
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 3.57,
|
| 613 |
+
"learning_rate": 4.894698620188817e-05,
|
| 614 |
+
"loss": 0.471,
|
| 615 |
+
"step": 910
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.61,
|
| 619 |
+
"learning_rate": 4.8874364560639076e-05,
|
| 620 |
+
"loss": 0.313,
|
| 621 |
+
"step": 920
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 3.65,
|
| 625 |
+
"learning_rate": 4.8801742919389983e-05,
|
| 626 |
+
"loss": 0.2437,
|
| 627 |
+
"step": 930
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 3.69,
|
| 631 |
+
"learning_rate": 4.8729121278140885e-05,
|
| 632 |
+
"loss": 0.4748,
|
| 633 |
+
"step": 940
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.73,
|
| 637 |
+
"learning_rate": 4.865649963689179e-05,
|
| 638 |
+
"loss": 0.5519,
|
| 639 |
+
"step": 950
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.76,
|
| 643 |
+
"learning_rate": 4.85838779956427e-05,
|
| 644 |
+
"loss": 0.3051,
|
| 645 |
+
"step": 960
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.8,
|
| 649 |
+
"learning_rate": 4.8511256354393615e-05,
|
| 650 |
+
"loss": 0.3847,
|
| 651 |
+
"step": 970
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 3.84,
|
| 655 |
+
"learning_rate": 4.843863471314452e-05,
|
| 656 |
+
"loss": 0.2454,
|
| 657 |
+
"step": 980
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.88,
|
| 661 |
+
"learning_rate": 4.8366013071895424e-05,
|
| 662 |
+
"loss": 0.379,
|
| 663 |
+
"step": 990
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 3.92,
|
| 667 |
+
"learning_rate": 4.829339143064633e-05,
|
| 668 |
+
"loss": 0.4485,
|
| 669 |
+
"step": 1000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 3.96,
|
| 673 |
+
"learning_rate": 4.822076978939724e-05,
|
| 674 |
+
"loss": 0.2484,
|
| 675 |
+
"step": 1010
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.0,
|
| 679 |
+
"learning_rate": 4.814814814814815e-05,
|
| 680 |
+
"loss": 0.4402,
|
| 681 |
+
"step": 1020
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 4.0,
|
| 685 |
+
"eval_accuracy": {
|
| 686 |
+
"accuracy": 0.8784313725490196
|
| 687 |
+
},
|
| 688 |
+
"eval_f1": {
|
| 689 |
+
"f1": 0.880471094803992
|
| 690 |
+
},
|
| 691 |
+
"eval_loss": 0.3946979343891144,
|
| 692 |
+
"eval_precision": {
|
| 693 |
+
"precision": 0.8952890487430221
|
| 694 |
+
},
|
| 695 |
+
"eval_recall": {
|
| 696 |
+
"recall": 0.8776488582339645
|
| 697 |
+
},
|
| 698 |
+
"eval_runtime": 10.1771,
|
| 699 |
+
"eval_samples_per_second": 50.113,
|
| 700 |
+
"eval_steps_per_second": 25.056,
|
| 701 |
+
"step": 1020
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 4.04,
|
| 705 |
+
"learning_rate": 4.807552650689906e-05,
|
| 706 |
+
"loss": 0.3494,
|
| 707 |
+
"step": 1030
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 4.08,
|
| 711 |
+
"learning_rate": 4.8002904865649964e-05,
|
| 712 |
+
"loss": 0.3327,
|
| 713 |
+
"step": 1040
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 4.12,
|
| 717 |
+
"learning_rate": 4.793028322440087e-05,
|
| 718 |
+
"loss": 0.2898,
|
| 719 |
+
"step": 1050
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 4.16,
|
| 723 |
+
"learning_rate": 4.785766158315178e-05,
|
| 724 |
+
"loss": 0.3965,
|
| 725 |
+
"step": 1060
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 4.2,
|
| 729 |
+
"learning_rate": 4.778503994190269e-05,
|
| 730 |
+
"loss": 0.3607,
|
| 731 |
+
"step": 1070
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.24,
|
| 735 |
+
"learning_rate": 4.77124183006536e-05,
|
| 736 |
+
"loss": 0.3714,
|
| 737 |
+
"step": 1080
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.27,
|
| 741 |
+
"learning_rate": 4.7639796659404504e-05,
|
| 742 |
+
"loss": 0.1796,
|
| 743 |
+
"step": 1090
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 4.31,
|
| 747 |
+
"learning_rate": 4.756717501815541e-05,
|
| 748 |
+
"loss": 0.1874,
|
| 749 |
+
"step": 1100
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 4.35,
|
| 753 |
+
"learning_rate": 4.749455337690632e-05,
|
| 754 |
+
"loss": 0.3462,
|
| 755 |
+
"step": 1110
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 4.39,
|
| 759 |
+
"learning_rate": 4.742193173565723e-05,
|
| 760 |
+
"loss": 0.2082,
|
| 761 |
+
"step": 1120
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 4.43,
|
| 765 |
+
"learning_rate": 4.7349310094408136e-05,
|
| 766 |
+
"loss": 0.4531,
|
| 767 |
+
"step": 1130
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 4.47,
|
| 771 |
+
"learning_rate": 4.7276688453159044e-05,
|
| 772 |
+
"loss": 0.3951,
|
| 773 |
+
"step": 1140
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.51,
|
| 777 |
+
"learning_rate": 4.720406681190995e-05,
|
| 778 |
+
"loss": 0.2337,
|
| 779 |
+
"step": 1150
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.55,
|
| 783 |
+
"learning_rate": 4.713144517066086e-05,
|
| 784 |
+
"loss": 0.2118,
|
| 785 |
+
"step": 1160
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 4.59,
|
| 789 |
+
"learning_rate": 4.705882352941177e-05,
|
| 790 |
+
"loss": 0.373,
|
| 791 |
+
"step": 1170
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 4.63,
|
| 795 |
+
"learning_rate": 4.6986201888162675e-05,
|
| 796 |
+
"loss": 0.1484,
|
| 797 |
+
"step": 1180
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 4.67,
|
| 801 |
+
"learning_rate": 4.691358024691358e-05,
|
| 802 |
+
"loss": 0.2568,
|
| 803 |
+
"step": 1190
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 4.71,
|
| 807 |
+
"learning_rate": 4.684095860566449e-05,
|
| 808 |
+
"loss": 0.3067,
|
| 809 |
+
"step": 1200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 4.75,
|
| 813 |
+
"learning_rate": 4.67683369644154e-05,
|
| 814 |
+
"loss": 0.2007,
|
| 815 |
+
"step": 1210
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.78,
|
| 819 |
+
"learning_rate": 4.669571532316631e-05,
|
| 820 |
+
"loss": 0.2493,
|
| 821 |
+
"step": 1220
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.82,
|
| 825 |
+
"learning_rate": 4.6623093681917215e-05,
|
| 826 |
+
"loss": 0.2717,
|
| 827 |
+
"step": 1230
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 4.86,
|
| 831 |
+
"learning_rate": 4.6550472040668116e-05,
|
| 832 |
+
"loss": 0.4134,
|
| 833 |
+
"step": 1240
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 4.9,
|
| 837 |
+
"learning_rate": 4.647785039941903e-05,
|
| 838 |
+
"loss": 0.2417,
|
| 839 |
+
"step": 1250
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 4.94,
|
| 843 |
+
"learning_rate": 4.640522875816994e-05,
|
| 844 |
+
"loss": 0.2573,
|
| 845 |
+
"step": 1260
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 4.98,
|
| 849 |
+
"learning_rate": 4.633260711692085e-05,
|
| 850 |
+
"loss": 0.3937,
|
| 851 |
+
"step": 1270
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 5.0,
|
| 855 |
+
"eval_accuracy": {
|
| 856 |
+
"accuracy": 0.8725490196078431
|
| 857 |
+
},
|
| 858 |
+
"eval_f1": {
|
| 859 |
+
"f1": 0.8703780158419334
|
| 860 |
+
},
|
| 861 |
+
"eval_loss": 0.4377364218235016,
|
| 862 |
+
"eval_precision": {
|
| 863 |
+
"precision": 0.8900760323128502
|
| 864 |
+
},
|
| 865 |
+
"eval_recall": {
|
| 866 |
+
"recall": 0.8691138648585458
|
| 867 |
+
},
|
| 868 |
+
"eval_runtime": 10.1762,
|
| 869 |
+
"eval_samples_per_second": 50.117,
|
| 870 |
+
"eval_steps_per_second": 25.058,
|
| 871 |
+
"step": 1275
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 5.02,
|
| 875 |
+
"learning_rate": 4.6259985475671755e-05,
|
| 876 |
+
"loss": 0.165,
|
| 877 |
+
"step": 1280
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 5.06,
|
| 881 |
+
"learning_rate": 4.6187363834422656e-05,
|
| 882 |
+
"loss": 0.1452,
|
| 883 |
+
"step": 1290
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 5.1,
|
| 887 |
+
"learning_rate": 4.6114742193173564e-05,
|
| 888 |
+
"loss": 0.1548,
|
| 889 |
+
"step": 1300
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 5.14,
|
| 893 |
+
"learning_rate": 4.604212055192448e-05,
|
| 894 |
+
"loss": 0.4443,
|
| 895 |
+
"step": 1310
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 5.18,
|
| 899 |
+
"learning_rate": 4.5969498910675387e-05,
|
| 900 |
+
"loss": 0.4322,
|
| 901 |
+
"step": 1320
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 5.22,
|
| 905 |
+
"learning_rate": 4.5896877269426294e-05,
|
| 906 |
+
"loss": 0.2355,
|
| 907 |
+
"step": 1330
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 5.25,
|
| 911 |
+
"learning_rate": 4.5824255628177196e-05,
|
| 912 |
+
"loss": 0.1949,
|
| 913 |
+
"step": 1340
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 5.29,
|
| 917 |
+
"learning_rate": 4.5751633986928104e-05,
|
| 918 |
+
"loss": 0.1959,
|
| 919 |
+
"step": 1350
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.33,
|
| 923 |
+
"learning_rate": 4.567901234567901e-05,
|
| 924 |
+
"loss": 0.3841,
|
| 925 |
+
"step": 1360
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 5.37,
|
| 929 |
+
"learning_rate": 4.5606390704429926e-05,
|
| 930 |
+
"loss": 0.1401,
|
| 931 |
+
"step": 1370
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 5.41,
|
| 935 |
+
"learning_rate": 4.5533769063180834e-05,
|
| 936 |
+
"loss": 0.1148,
|
| 937 |
+
"step": 1380
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 5.45,
|
| 941 |
+
"learning_rate": 4.5461147421931735e-05,
|
| 942 |
+
"loss": 0.234,
|
| 943 |
+
"step": 1390
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 5.49,
|
| 947 |
+
"learning_rate": 4.538852578068264e-05,
|
| 948 |
+
"loss": 0.3899,
|
| 949 |
+
"step": 1400
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 5.53,
|
| 953 |
+
"learning_rate": 4.531590413943355e-05,
|
| 954 |
+
"loss": 0.4873,
|
| 955 |
+
"step": 1410
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 5.57,
|
| 959 |
+
"learning_rate": 4.5243282498184466e-05,
|
| 960 |
+
"loss": 0.2483,
|
| 961 |
+
"step": 1420
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.61,
|
| 965 |
+
"learning_rate": 4.5170660856935374e-05,
|
| 966 |
+
"loss": 0.2654,
|
| 967 |
+
"step": 1430
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 5.65,
|
| 971 |
+
"learning_rate": 4.5098039215686275e-05,
|
| 972 |
+
"loss": 0.2035,
|
| 973 |
+
"step": 1440
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 5.69,
|
| 977 |
+
"learning_rate": 4.502541757443718e-05,
|
| 978 |
+
"loss": 0.3744,
|
| 979 |
+
"step": 1450
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 5.73,
|
| 983 |
+
"learning_rate": 4.495279593318809e-05,
|
| 984 |
+
"loss": 0.2139,
|
| 985 |
+
"step": 1460
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 5.76,
|
| 989 |
+
"learning_rate": 4.4880174291939e-05,
|
| 990 |
+
"loss": 0.2212,
|
| 991 |
+
"step": 1470
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 5.8,
|
| 995 |
+
"learning_rate": 4.480755265068991e-05,
|
| 996 |
+
"loss": 0.376,
|
| 997 |
+
"step": 1480
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 5.84,
|
| 1001 |
+
"learning_rate": 4.4734931009440815e-05,
|
| 1002 |
+
"loss": 0.2399,
|
| 1003 |
+
"step": 1490
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.88,
|
| 1007 |
+
"learning_rate": 4.466230936819172e-05,
|
| 1008 |
+
"loss": 0.3534,
|
| 1009 |
+
"step": 1500
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 5.92,
|
| 1013 |
+
"learning_rate": 4.458968772694263e-05,
|
| 1014 |
+
"loss": 0.3667,
|
| 1015 |
+
"step": 1510
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 5.96,
|
| 1019 |
+
"learning_rate": 4.451706608569354e-05,
|
| 1020 |
+
"loss": 0.496,
|
| 1021 |
+
"step": 1520
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 6.0,
|
| 1025 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 1026 |
+
"loss": 0.4042,
|
| 1027 |
+
"step": 1530
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 6.0,
|
| 1031 |
+
"eval_accuracy": {
|
| 1032 |
+
"accuracy": 0.8862745098039215
|
| 1033 |
+
},
|
| 1034 |
+
"eval_f1": {
|
| 1035 |
+
"f1": 0.8859196308355288
|
| 1036 |
+
},
|
| 1037 |
+
"eval_loss": 0.39891302585601807,
|
| 1038 |
+
"eval_precision": {
|
| 1039 |
+
"precision": 0.8972093144987638
|
| 1040 |
+
},
|
| 1041 |
+
"eval_recall": {
|
| 1042 |
+
"recall": 0.8834955381497934
|
| 1043 |
+
},
|
| 1044 |
+
"eval_runtime": 10.1174,
|
| 1045 |
+
"eval_samples_per_second": 50.408,
|
| 1046 |
+
"eval_steps_per_second": 25.204,
|
| 1047 |
+
"step": 1530
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 6.04,
|
| 1051 |
+
"learning_rate": 4.4371822803195355e-05,
|
| 1052 |
+
"loss": 0.2637,
|
| 1053 |
+
"step": 1540
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 6.08,
|
| 1057 |
+
"learning_rate": 4.429920116194626e-05,
|
| 1058 |
+
"loss": 0.2821,
|
| 1059 |
+
"step": 1550
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 6.12,
|
| 1063 |
+
"learning_rate": 4.422657952069717e-05,
|
| 1064 |
+
"loss": 0.1201,
|
| 1065 |
+
"step": 1560
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 6.16,
|
| 1069 |
+
"learning_rate": 4.415395787944808e-05,
|
| 1070 |
+
"loss": 0.262,
|
| 1071 |
+
"step": 1570
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 6.2,
|
| 1075 |
+
"learning_rate": 4.4081336238198986e-05,
|
| 1076 |
+
"loss": 0.2064,
|
| 1077 |
+
"step": 1580
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 6.24,
|
| 1081 |
+
"learning_rate": 4.400871459694989e-05,
|
| 1082 |
+
"loss": 0.1815,
|
| 1083 |
+
"step": 1590
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 6.27,
|
| 1087 |
+
"learning_rate": 4.39360929557008e-05,
|
| 1088 |
+
"loss": 0.1959,
|
| 1089 |
+
"step": 1600
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 6.31,
|
| 1093 |
+
"learning_rate": 4.386347131445171e-05,
|
| 1094 |
+
"loss": 0.1959,
|
| 1095 |
+
"step": 1610
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.35,
|
| 1099 |
+
"learning_rate": 4.379084967320262e-05,
|
| 1100 |
+
"loss": 0.2398,
|
| 1101 |
+
"step": 1620
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.39,
|
| 1105 |
+
"learning_rate": 4.3718228031953526e-05,
|
| 1106 |
+
"loss": 0.2038,
|
| 1107 |
+
"step": 1630
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 6.43,
|
| 1111 |
+
"learning_rate": 4.364560639070443e-05,
|
| 1112 |
+
"loss": 0.2004,
|
| 1113 |
+
"step": 1640
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.47,
|
| 1117 |
+
"learning_rate": 4.357298474945534e-05,
|
| 1118 |
+
"loss": 0.0377,
|
| 1119 |
+
"step": 1650
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 6.51,
|
| 1123 |
+
"learning_rate": 4.350036310820625e-05,
|
| 1124 |
+
"loss": 0.2022,
|
| 1125 |
+
"step": 1660
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 6.55,
|
| 1129 |
+
"learning_rate": 4.342774146695716e-05,
|
| 1130 |
+
"loss": 0.4403,
|
| 1131 |
+
"step": 1670
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 6.59,
|
| 1135 |
+
"learning_rate": 4.3355119825708066e-05,
|
| 1136 |
+
"loss": 0.2577,
|
| 1137 |
+
"step": 1680
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.63,
|
| 1141 |
+
"learning_rate": 4.328249818445897e-05,
|
| 1142 |
+
"loss": 0.3312,
|
| 1143 |
+
"step": 1690
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.67,
|
| 1147 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 1148 |
+
"loss": 0.307,
|
| 1149 |
+
"step": 1700
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 6.71,
|
| 1153 |
+
"learning_rate": 4.313725490196079e-05,
|
| 1154 |
+
"loss": 0.0961,
|
| 1155 |
+
"step": 1710
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.75,
|
| 1159 |
+
"learning_rate": 4.30646332607117e-05,
|
| 1160 |
+
"loss": 0.3266,
|
| 1161 |
+
"step": 1720
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 6.78,
|
| 1165 |
+
"learning_rate": 4.2992011619462605e-05,
|
| 1166 |
+
"loss": 0.2601,
|
| 1167 |
+
"step": 1730
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 6.82,
|
| 1171 |
+
"learning_rate": 4.291938997821351e-05,
|
| 1172 |
+
"loss": 0.2382,
|
| 1173 |
+
"step": 1740
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 6.86,
|
| 1177 |
+
"learning_rate": 4.2846768336964415e-05,
|
| 1178 |
+
"loss": 0.1469,
|
| 1179 |
+
"step": 1750
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.9,
|
| 1183 |
+
"learning_rate": 4.277414669571532e-05,
|
| 1184 |
+
"loss": 0.2158,
|
| 1185 |
+
"step": 1760
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.94,
|
| 1189 |
+
"learning_rate": 4.270152505446624e-05,
|
| 1190 |
+
"loss": 0.4392,
|
| 1191 |
+
"step": 1770
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 6.98,
|
| 1195 |
+
"learning_rate": 4.262890341321714e-05,
|
| 1196 |
+
"loss": 0.4321,
|
| 1197 |
+
"step": 1780
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 7.0,
|
| 1201 |
+
"eval_accuracy": {
|
| 1202 |
+
"accuracy": 0.8901960784313725
|
| 1203 |
+
},
|
| 1204 |
+
"eval_f1": {
|
| 1205 |
+
"f1": 0.8896787277785286
|
| 1206 |
+
},
|
| 1207 |
+
"eval_loss": 0.36420056223869324,
|
| 1208 |
+
"eval_precision": {
|
| 1209 |
+
"precision": 0.9002265766283616
|
| 1210 |
+
},
|
| 1211 |
+
"eval_recall": {
|
| 1212 |
+
"recall": 0.8878911425453978
|
| 1213 |
+
},
|
| 1214 |
+
"eval_runtime": 10.2347,
|
| 1215 |
+
"eval_samples_per_second": 49.83,
|
| 1216 |
+
"eval_steps_per_second": 24.915,
|
| 1217 |
+
"step": 1785
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 7.02,
|
| 1221 |
+
"learning_rate": 4.2556281771968046e-05,
|
| 1222 |
+
"loss": 0.5162,
|
| 1223 |
+
"step": 1790
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 7.06,
|
| 1227 |
+
"learning_rate": 4.2483660130718954e-05,
|
| 1228 |
+
"loss": 0.2491,
|
| 1229 |
+
"step": 1800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 7.1,
|
| 1233 |
+
"learning_rate": 4.241103848946986e-05,
|
| 1234 |
+
"loss": 0.2267,
|
| 1235 |
+
"step": 1810
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 7.14,
|
| 1239 |
+
"learning_rate": 4.233841684822078e-05,
|
| 1240 |
+
"loss": 0.2395,
|
| 1241 |
+
"step": 1820
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 7.18,
|
| 1245 |
+
"learning_rate": 4.226579520697168e-05,
|
| 1246 |
+
"loss": 0.3007,
|
| 1247 |
+
"step": 1830
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 7.22,
|
| 1251 |
+
"learning_rate": 4.2193173565722586e-05,
|
| 1252 |
+
"loss": 0.1316,
|
| 1253 |
+
"step": 1840
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 7.25,
|
| 1257 |
+
"learning_rate": 4.2120551924473494e-05,
|
| 1258 |
+
"loss": 0.1592,
|
| 1259 |
+
"step": 1850
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 7.29,
|
| 1263 |
+
"learning_rate": 4.20479302832244e-05,
|
| 1264 |
+
"loss": 0.3467,
|
| 1265 |
+
"step": 1860
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 7.33,
|
| 1269 |
+
"learning_rate": 4.197530864197531e-05,
|
| 1270 |
+
"loss": 0.4195,
|
| 1271 |
+
"step": 1870
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 7.37,
|
| 1275 |
+
"learning_rate": 4.190268700072622e-05,
|
| 1276 |
+
"loss": 0.3561,
|
| 1277 |
+
"step": 1880
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 7.41,
|
| 1281 |
+
"learning_rate": 4.1830065359477126e-05,
|
| 1282 |
+
"loss": 0.1572,
|
| 1283 |
+
"step": 1890
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.45,
|
| 1287 |
+
"learning_rate": 4.1757443718228034e-05,
|
| 1288 |
+
"loss": 0.2104,
|
| 1289 |
+
"step": 1900
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 7.49,
|
| 1293 |
+
"learning_rate": 4.168482207697894e-05,
|
| 1294 |
+
"loss": 0.3536,
|
| 1295 |
+
"step": 1910
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 7.53,
|
| 1299 |
+
"learning_rate": 4.161220043572985e-05,
|
| 1300 |
+
"loss": 0.345,
|
| 1301 |
+
"step": 1920
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 7.57,
|
| 1305 |
+
"learning_rate": 4.153957879448076e-05,
|
| 1306 |
+
"loss": 0.1083,
|
| 1307 |
+
"step": 1930
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 7.61,
|
| 1311 |
+
"learning_rate": 4.1466957153231666e-05,
|
| 1312 |
+
"loss": 0.1895,
|
| 1313 |
+
"step": 1940
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 7.65,
|
| 1317 |
+
"learning_rate": 4.1394335511982573e-05,
|
| 1318 |
+
"loss": 0.3232,
|
| 1319 |
+
"step": 1950
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 7.69,
|
| 1323 |
+
"learning_rate": 4.132171387073348e-05,
|
| 1324 |
+
"loss": 0.2939,
|
| 1325 |
+
"step": 1960
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.73,
|
| 1329 |
+
"learning_rate": 4.124909222948439e-05,
|
| 1330 |
+
"loss": 0.0918,
|
| 1331 |
+
"step": 1970
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 7.76,
|
| 1335 |
+
"learning_rate": 4.11764705882353e-05,
|
| 1336 |
+
"loss": 0.1722,
|
| 1337 |
+
"step": 1980
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 7.8,
|
| 1341 |
+
"learning_rate": 4.1103848946986205e-05,
|
| 1342 |
+
"loss": 0.3229,
|
| 1343 |
+
"step": 1990
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 7.84,
|
| 1347 |
+
"learning_rate": 4.103122730573711e-05,
|
| 1348 |
+
"loss": 0.1501,
|
| 1349 |
+
"step": 2000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 7.88,
|
| 1353 |
+
"learning_rate": 4.095860566448802e-05,
|
| 1354 |
+
"loss": 0.1468,
|
| 1355 |
+
"step": 2010
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 7.92,
|
| 1359 |
+
"learning_rate": 4.088598402323893e-05,
|
| 1360 |
+
"loss": 0.1665,
|
| 1361 |
+
"step": 2020
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 7.96,
|
| 1365 |
+
"learning_rate": 4.081336238198984e-05,
|
| 1366 |
+
"loss": 0.1324,
|
| 1367 |
+
"step": 2030
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 8.0,
|
| 1371 |
+
"learning_rate": 4.074074074074074e-05,
|
| 1372 |
+
"loss": 0.2618,
|
| 1373 |
+
"step": 2040
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 8.0,
|
| 1377 |
+
"eval_accuracy": {
|
| 1378 |
+
"accuracy": 0.8960784313725491
|
| 1379 |
+
},
|
| 1380 |
+
"eval_f1": {
|
| 1381 |
+
"f1": 0.8950352545633242
|
| 1382 |
+
},
|
| 1383 |
+
"eval_loss": 0.40923169255256653,
|
| 1384 |
+
"eval_precision": {
|
| 1385 |
+
"precision": 0.9033601037449228
|
| 1386 |
+
},
|
| 1387 |
+
"eval_recall": {
|
| 1388 |
+
"recall": 0.8953584093211753
|
| 1389 |
+
},
|
| 1390 |
+
"eval_runtime": 10.3235,
|
| 1391 |
+
"eval_samples_per_second": 49.402,
|
| 1392 |
+
"eval_steps_per_second": 24.701,
|
| 1393 |
+
"step": 2040
|
| 1394 |
+
}
|
| 1395 |
+
],
|
| 1396 |
+
"logging_steps": 10,
|
| 1397 |
+
"max_steps": 7650,
|
| 1398 |
+
"num_input_tokens_seen": 0,
|
| 1399 |
+
"num_train_epochs": 30,
|
| 1400 |
+
"save_steps": 500,
|
| 1401 |
+
"total_flos": 1.2647032750021018e+18,
|
| 1402 |
+
"train_batch_size": 2,
|
| 1403 |
+
"trial_name": null,
|
| 1404 |
+
"trial_params": null
|
| 1405 |
+
}
|
tmp-checkpoint-2040/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-2295/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-2295/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:75a1b6c5d56a7db190736effc0e273a38a5cc5c9df4add9dbe4d7dc439421c36
|
| 3 |
+
size 343233204
|
tmp-checkpoint-2295/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d06fa340ed808675d3e8cd604aa01a999f4cecd040eee4075a7bed2c4da35a74
|
| 3 |
+
size 686587322
|
tmp-checkpoint-2295/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-2295/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0510755e0e32136c5b6ebf9f8dc1b5ed0c8f817bc47fa251a7b4155e22dc0d95
|
| 3 |
+
size 14244
|
tmp-checkpoint-2295/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:00056fa36abe1f9819ac0d9ccd4d68a127325805e9526e152156ac69398d571e
|
| 3 |
+
size 1064
|
tmp-checkpoint-2295/trainer_state.json
ADDED
|
@@ -0,0 +1,1575 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.36420056223869324,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-1785",
|
| 4 |
+
"epoch": 9.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2295,
|
| 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 |
+
"epoch": 3.02,
|
| 529 |
+
"learning_rate": 4.9963689179375456e-05,
|
| 530 |
+
"loss": 0.3347,
|
| 531 |
+
"step": 770
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 3.06,
|
| 535 |
+
"learning_rate": 4.9891067538126364e-05,
|
| 536 |
+
"loss": 0.3339,
|
| 537 |
+
"step": 780
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 3.1,
|
| 541 |
+
"learning_rate": 4.981844589687727e-05,
|
| 542 |
+
"loss": 0.5573,
|
| 543 |
+
"step": 790
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 3.14,
|
| 547 |
+
"learning_rate": 4.974582425562818e-05,
|
| 548 |
+
"loss": 0.5463,
|
| 549 |
+
"step": 800
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.18,
|
| 553 |
+
"learning_rate": 4.967320261437909e-05,
|
| 554 |
+
"loss": 0.3453,
|
| 555 |
+
"step": 810
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.22,
|
| 559 |
+
"learning_rate": 4.9600580973129996e-05,
|
| 560 |
+
"loss": 0.3814,
|
| 561 |
+
"step": 820
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.25,
|
| 565 |
+
"learning_rate": 4.9527959331880904e-05,
|
| 566 |
+
"loss": 0.4434,
|
| 567 |
+
"step": 830
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 3.29,
|
| 571 |
+
"learning_rate": 4.945533769063181e-05,
|
| 572 |
+
"loss": 0.3675,
|
| 573 |
+
"step": 840
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 3.33,
|
| 577 |
+
"learning_rate": 4.938271604938271e-05,
|
| 578 |
+
"loss": 0.5386,
|
| 579 |
+
"step": 850
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 3.37,
|
| 583 |
+
"learning_rate": 4.931009440813363e-05,
|
| 584 |
+
"loss": 0.2692,
|
| 585 |
+
"step": 860
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.41,
|
| 589 |
+
"learning_rate": 4.9237472766884536e-05,
|
| 590 |
+
"loss": 0.4815,
|
| 591 |
+
"step": 870
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.45,
|
| 595 |
+
"learning_rate": 4.9164851125635444e-05,
|
| 596 |
+
"loss": 0.4178,
|
| 597 |
+
"step": 880
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.49,
|
| 601 |
+
"learning_rate": 4.909222948438635e-05,
|
| 602 |
+
"loss": 0.3618,
|
| 603 |
+
"step": 890
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.53,
|
| 607 |
+
"learning_rate": 4.901960784313725e-05,
|
| 608 |
+
"loss": 0.6841,
|
| 609 |
+
"step": 900
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 3.57,
|
| 613 |
+
"learning_rate": 4.894698620188817e-05,
|
| 614 |
+
"loss": 0.471,
|
| 615 |
+
"step": 910
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.61,
|
| 619 |
+
"learning_rate": 4.8874364560639076e-05,
|
| 620 |
+
"loss": 0.313,
|
| 621 |
+
"step": 920
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 3.65,
|
| 625 |
+
"learning_rate": 4.8801742919389983e-05,
|
| 626 |
+
"loss": 0.2437,
|
| 627 |
+
"step": 930
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 3.69,
|
| 631 |
+
"learning_rate": 4.8729121278140885e-05,
|
| 632 |
+
"loss": 0.4748,
|
| 633 |
+
"step": 940
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.73,
|
| 637 |
+
"learning_rate": 4.865649963689179e-05,
|
| 638 |
+
"loss": 0.5519,
|
| 639 |
+
"step": 950
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.76,
|
| 643 |
+
"learning_rate": 4.85838779956427e-05,
|
| 644 |
+
"loss": 0.3051,
|
| 645 |
+
"step": 960
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.8,
|
| 649 |
+
"learning_rate": 4.8511256354393615e-05,
|
| 650 |
+
"loss": 0.3847,
|
| 651 |
+
"step": 970
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 3.84,
|
| 655 |
+
"learning_rate": 4.843863471314452e-05,
|
| 656 |
+
"loss": 0.2454,
|
| 657 |
+
"step": 980
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.88,
|
| 661 |
+
"learning_rate": 4.8366013071895424e-05,
|
| 662 |
+
"loss": 0.379,
|
| 663 |
+
"step": 990
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 3.92,
|
| 667 |
+
"learning_rate": 4.829339143064633e-05,
|
| 668 |
+
"loss": 0.4485,
|
| 669 |
+
"step": 1000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 3.96,
|
| 673 |
+
"learning_rate": 4.822076978939724e-05,
|
| 674 |
+
"loss": 0.2484,
|
| 675 |
+
"step": 1010
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.0,
|
| 679 |
+
"learning_rate": 4.814814814814815e-05,
|
| 680 |
+
"loss": 0.4402,
|
| 681 |
+
"step": 1020
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 4.0,
|
| 685 |
+
"eval_accuracy": {
|
| 686 |
+
"accuracy": 0.8784313725490196
|
| 687 |
+
},
|
| 688 |
+
"eval_f1": {
|
| 689 |
+
"f1": 0.880471094803992
|
| 690 |
+
},
|
| 691 |
+
"eval_loss": 0.3946979343891144,
|
| 692 |
+
"eval_precision": {
|
| 693 |
+
"precision": 0.8952890487430221
|
| 694 |
+
},
|
| 695 |
+
"eval_recall": {
|
| 696 |
+
"recall": 0.8776488582339645
|
| 697 |
+
},
|
| 698 |
+
"eval_runtime": 10.1771,
|
| 699 |
+
"eval_samples_per_second": 50.113,
|
| 700 |
+
"eval_steps_per_second": 25.056,
|
| 701 |
+
"step": 1020
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 4.04,
|
| 705 |
+
"learning_rate": 4.807552650689906e-05,
|
| 706 |
+
"loss": 0.3494,
|
| 707 |
+
"step": 1030
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 4.08,
|
| 711 |
+
"learning_rate": 4.8002904865649964e-05,
|
| 712 |
+
"loss": 0.3327,
|
| 713 |
+
"step": 1040
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 4.12,
|
| 717 |
+
"learning_rate": 4.793028322440087e-05,
|
| 718 |
+
"loss": 0.2898,
|
| 719 |
+
"step": 1050
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 4.16,
|
| 723 |
+
"learning_rate": 4.785766158315178e-05,
|
| 724 |
+
"loss": 0.3965,
|
| 725 |
+
"step": 1060
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 4.2,
|
| 729 |
+
"learning_rate": 4.778503994190269e-05,
|
| 730 |
+
"loss": 0.3607,
|
| 731 |
+
"step": 1070
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.24,
|
| 735 |
+
"learning_rate": 4.77124183006536e-05,
|
| 736 |
+
"loss": 0.3714,
|
| 737 |
+
"step": 1080
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.27,
|
| 741 |
+
"learning_rate": 4.7639796659404504e-05,
|
| 742 |
+
"loss": 0.1796,
|
| 743 |
+
"step": 1090
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 4.31,
|
| 747 |
+
"learning_rate": 4.756717501815541e-05,
|
| 748 |
+
"loss": 0.1874,
|
| 749 |
+
"step": 1100
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 4.35,
|
| 753 |
+
"learning_rate": 4.749455337690632e-05,
|
| 754 |
+
"loss": 0.3462,
|
| 755 |
+
"step": 1110
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 4.39,
|
| 759 |
+
"learning_rate": 4.742193173565723e-05,
|
| 760 |
+
"loss": 0.2082,
|
| 761 |
+
"step": 1120
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 4.43,
|
| 765 |
+
"learning_rate": 4.7349310094408136e-05,
|
| 766 |
+
"loss": 0.4531,
|
| 767 |
+
"step": 1130
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 4.47,
|
| 771 |
+
"learning_rate": 4.7276688453159044e-05,
|
| 772 |
+
"loss": 0.3951,
|
| 773 |
+
"step": 1140
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.51,
|
| 777 |
+
"learning_rate": 4.720406681190995e-05,
|
| 778 |
+
"loss": 0.2337,
|
| 779 |
+
"step": 1150
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.55,
|
| 783 |
+
"learning_rate": 4.713144517066086e-05,
|
| 784 |
+
"loss": 0.2118,
|
| 785 |
+
"step": 1160
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 4.59,
|
| 789 |
+
"learning_rate": 4.705882352941177e-05,
|
| 790 |
+
"loss": 0.373,
|
| 791 |
+
"step": 1170
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 4.63,
|
| 795 |
+
"learning_rate": 4.6986201888162675e-05,
|
| 796 |
+
"loss": 0.1484,
|
| 797 |
+
"step": 1180
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 4.67,
|
| 801 |
+
"learning_rate": 4.691358024691358e-05,
|
| 802 |
+
"loss": 0.2568,
|
| 803 |
+
"step": 1190
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 4.71,
|
| 807 |
+
"learning_rate": 4.684095860566449e-05,
|
| 808 |
+
"loss": 0.3067,
|
| 809 |
+
"step": 1200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 4.75,
|
| 813 |
+
"learning_rate": 4.67683369644154e-05,
|
| 814 |
+
"loss": 0.2007,
|
| 815 |
+
"step": 1210
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.78,
|
| 819 |
+
"learning_rate": 4.669571532316631e-05,
|
| 820 |
+
"loss": 0.2493,
|
| 821 |
+
"step": 1220
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.82,
|
| 825 |
+
"learning_rate": 4.6623093681917215e-05,
|
| 826 |
+
"loss": 0.2717,
|
| 827 |
+
"step": 1230
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 4.86,
|
| 831 |
+
"learning_rate": 4.6550472040668116e-05,
|
| 832 |
+
"loss": 0.4134,
|
| 833 |
+
"step": 1240
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 4.9,
|
| 837 |
+
"learning_rate": 4.647785039941903e-05,
|
| 838 |
+
"loss": 0.2417,
|
| 839 |
+
"step": 1250
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 4.94,
|
| 843 |
+
"learning_rate": 4.640522875816994e-05,
|
| 844 |
+
"loss": 0.2573,
|
| 845 |
+
"step": 1260
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 4.98,
|
| 849 |
+
"learning_rate": 4.633260711692085e-05,
|
| 850 |
+
"loss": 0.3937,
|
| 851 |
+
"step": 1270
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 5.0,
|
| 855 |
+
"eval_accuracy": {
|
| 856 |
+
"accuracy": 0.8725490196078431
|
| 857 |
+
},
|
| 858 |
+
"eval_f1": {
|
| 859 |
+
"f1": 0.8703780158419334
|
| 860 |
+
},
|
| 861 |
+
"eval_loss": 0.4377364218235016,
|
| 862 |
+
"eval_precision": {
|
| 863 |
+
"precision": 0.8900760323128502
|
| 864 |
+
},
|
| 865 |
+
"eval_recall": {
|
| 866 |
+
"recall": 0.8691138648585458
|
| 867 |
+
},
|
| 868 |
+
"eval_runtime": 10.1762,
|
| 869 |
+
"eval_samples_per_second": 50.117,
|
| 870 |
+
"eval_steps_per_second": 25.058,
|
| 871 |
+
"step": 1275
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 5.02,
|
| 875 |
+
"learning_rate": 4.6259985475671755e-05,
|
| 876 |
+
"loss": 0.165,
|
| 877 |
+
"step": 1280
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 5.06,
|
| 881 |
+
"learning_rate": 4.6187363834422656e-05,
|
| 882 |
+
"loss": 0.1452,
|
| 883 |
+
"step": 1290
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 5.1,
|
| 887 |
+
"learning_rate": 4.6114742193173564e-05,
|
| 888 |
+
"loss": 0.1548,
|
| 889 |
+
"step": 1300
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 5.14,
|
| 893 |
+
"learning_rate": 4.604212055192448e-05,
|
| 894 |
+
"loss": 0.4443,
|
| 895 |
+
"step": 1310
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 5.18,
|
| 899 |
+
"learning_rate": 4.5969498910675387e-05,
|
| 900 |
+
"loss": 0.4322,
|
| 901 |
+
"step": 1320
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 5.22,
|
| 905 |
+
"learning_rate": 4.5896877269426294e-05,
|
| 906 |
+
"loss": 0.2355,
|
| 907 |
+
"step": 1330
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 5.25,
|
| 911 |
+
"learning_rate": 4.5824255628177196e-05,
|
| 912 |
+
"loss": 0.1949,
|
| 913 |
+
"step": 1340
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 5.29,
|
| 917 |
+
"learning_rate": 4.5751633986928104e-05,
|
| 918 |
+
"loss": 0.1959,
|
| 919 |
+
"step": 1350
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.33,
|
| 923 |
+
"learning_rate": 4.567901234567901e-05,
|
| 924 |
+
"loss": 0.3841,
|
| 925 |
+
"step": 1360
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 5.37,
|
| 929 |
+
"learning_rate": 4.5606390704429926e-05,
|
| 930 |
+
"loss": 0.1401,
|
| 931 |
+
"step": 1370
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 5.41,
|
| 935 |
+
"learning_rate": 4.5533769063180834e-05,
|
| 936 |
+
"loss": 0.1148,
|
| 937 |
+
"step": 1380
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 5.45,
|
| 941 |
+
"learning_rate": 4.5461147421931735e-05,
|
| 942 |
+
"loss": 0.234,
|
| 943 |
+
"step": 1390
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 5.49,
|
| 947 |
+
"learning_rate": 4.538852578068264e-05,
|
| 948 |
+
"loss": 0.3899,
|
| 949 |
+
"step": 1400
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 5.53,
|
| 953 |
+
"learning_rate": 4.531590413943355e-05,
|
| 954 |
+
"loss": 0.4873,
|
| 955 |
+
"step": 1410
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 5.57,
|
| 959 |
+
"learning_rate": 4.5243282498184466e-05,
|
| 960 |
+
"loss": 0.2483,
|
| 961 |
+
"step": 1420
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.61,
|
| 965 |
+
"learning_rate": 4.5170660856935374e-05,
|
| 966 |
+
"loss": 0.2654,
|
| 967 |
+
"step": 1430
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 5.65,
|
| 971 |
+
"learning_rate": 4.5098039215686275e-05,
|
| 972 |
+
"loss": 0.2035,
|
| 973 |
+
"step": 1440
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 5.69,
|
| 977 |
+
"learning_rate": 4.502541757443718e-05,
|
| 978 |
+
"loss": 0.3744,
|
| 979 |
+
"step": 1450
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 5.73,
|
| 983 |
+
"learning_rate": 4.495279593318809e-05,
|
| 984 |
+
"loss": 0.2139,
|
| 985 |
+
"step": 1460
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 5.76,
|
| 989 |
+
"learning_rate": 4.4880174291939e-05,
|
| 990 |
+
"loss": 0.2212,
|
| 991 |
+
"step": 1470
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 5.8,
|
| 995 |
+
"learning_rate": 4.480755265068991e-05,
|
| 996 |
+
"loss": 0.376,
|
| 997 |
+
"step": 1480
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 5.84,
|
| 1001 |
+
"learning_rate": 4.4734931009440815e-05,
|
| 1002 |
+
"loss": 0.2399,
|
| 1003 |
+
"step": 1490
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.88,
|
| 1007 |
+
"learning_rate": 4.466230936819172e-05,
|
| 1008 |
+
"loss": 0.3534,
|
| 1009 |
+
"step": 1500
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 5.92,
|
| 1013 |
+
"learning_rate": 4.458968772694263e-05,
|
| 1014 |
+
"loss": 0.3667,
|
| 1015 |
+
"step": 1510
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 5.96,
|
| 1019 |
+
"learning_rate": 4.451706608569354e-05,
|
| 1020 |
+
"loss": 0.496,
|
| 1021 |
+
"step": 1520
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 6.0,
|
| 1025 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 1026 |
+
"loss": 0.4042,
|
| 1027 |
+
"step": 1530
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 6.0,
|
| 1031 |
+
"eval_accuracy": {
|
| 1032 |
+
"accuracy": 0.8862745098039215
|
| 1033 |
+
},
|
| 1034 |
+
"eval_f1": {
|
| 1035 |
+
"f1": 0.8859196308355288
|
| 1036 |
+
},
|
| 1037 |
+
"eval_loss": 0.39891302585601807,
|
| 1038 |
+
"eval_precision": {
|
| 1039 |
+
"precision": 0.8972093144987638
|
| 1040 |
+
},
|
| 1041 |
+
"eval_recall": {
|
| 1042 |
+
"recall": 0.8834955381497934
|
| 1043 |
+
},
|
| 1044 |
+
"eval_runtime": 10.1174,
|
| 1045 |
+
"eval_samples_per_second": 50.408,
|
| 1046 |
+
"eval_steps_per_second": 25.204,
|
| 1047 |
+
"step": 1530
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 6.04,
|
| 1051 |
+
"learning_rate": 4.4371822803195355e-05,
|
| 1052 |
+
"loss": 0.2637,
|
| 1053 |
+
"step": 1540
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 6.08,
|
| 1057 |
+
"learning_rate": 4.429920116194626e-05,
|
| 1058 |
+
"loss": 0.2821,
|
| 1059 |
+
"step": 1550
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 6.12,
|
| 1063 |
+
"learning_rate": 4.422657952069717e-05,
|
| 1064 |
+
"loss": 0.1201,
|
| 1065 |
+
"step": 1560
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 6.16,
|
| 1069 |
+
"learning_rate": 4.415395787944808e-05,
|
| 1070 |
+
"loss": 0.262,
|
| 1071 |
+
"step": 1570
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 6.2,
|
| 1075 |
+
"learning_rate": 4.4081336238198986e-05,
|
| 1076 |
+
"loss": 0.2064,
|
| 1077 |
+
"step": 1580
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 6.24,
|
| 1081 |
+
"learning_rate": 4.400871459694989e-05,
|
| 1082 |
+
"loss": 0.1815,
|
| 1083 |
+
"step": 1590
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 6.27,
|
| 1087 |
+
"learning_rate": 4.39360929557008e-05,
|
| 1088 |
+
"loss": 0.1959,
|
| 1089 |
+
"step": 1600
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 6.31,
|
| 1093 |
+
"learning_rate": 4.386347131445171e-05,
|
| 1094 |
+
"loss": 0.1959,
|
| 1095 |
+
"step": 1610
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.35,
|
| 1099 |
+
"learning_rate": 4.379084967320262e-05,
|
| 1100 |
+
"loss": 0.2398,
|
| 1101 |
+
"step": 1620
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.39,
|
| 1105 |
+
"learning_rate": 4.3718228031953526e-05,
|
| 1106 |
+
"loss": 0.2038,
|
| 1107 |
+
"step": 1630
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 6.43,
|
| 1111 |
+
"learning_rate": 4.364560639070443e-05,
|
| 1112 |
+
"loss": 0.2004,
|
| 1113 |
+
"step": 1640
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.47,
|
| 1117 |
+
"learning_rate": 4.357298474945534e-05,
|
| 1118 |
+
"loss": 0.0377,
|
| 1119 |
+
"step": 1650
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 6.51,
|
| 1123 |
+
"learning_rate": 4.350036310820625e-05,
|
| 1124 |
+
"loss": 0.2022,
|
| 1125 |
+
"step": 1660
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 6.55,
|
| 1129 |
+
"learning_rate": 4.342774146695716e-05,
|
| 1130 |
+
"loss": 0.4403,
|
| 1131 |
+
"step": 1670
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 6.59,
|
| 1135 |
+
"learning_rate": 4.3355119825708066e-05,
|
| 1136 |
+
"loss": 0.2577,
|
| 1137 |
+
"step": 1680
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.63,
|
| 1141 |
+
"learning_rate": 4.328249818445897e-05,
|
| 1142 |
+
"loss": 0.3312,
|
| 1143 |
+
"step": 1690
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.67,
|
| 1147 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 1148 |
+
"loss": 0.307,
|
| 1149 |
+
"step": 1700
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 6.71,
|
| 1153 |
+
"learning_rate": 4.313725490196079e-05,
|
| 1154 |
+
"loss": 0.0961,
|
| 1155 |
+
"step": 1710
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.75,
|
| 1159 |
+
"learning_rate": 4.30646332607117e-05,
|
| 1160 |
+
"loss": 0.3266,
|
| 1161 |
+
"step": 1720
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 6.78,
|
| 1165 |
+
"learning_rate": 4.2992011619462605e-05,
|
| 1166 |
+
"loss": 0.2601,
|
| 1167 |
+
"step": 1730
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 6.82,
|
| 1171 |
+
"learning_rate": 4.291938997821351e-05,
|
| 1172 |
+
"loss": 0.2382,
|
| 1173 |
+
"step": 1740
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 6.86,
|
| 1177 |
+
"learning_rate": 4.2846768336964415e-05,
|
| 1178 |
+
"loss": 0.1469,
|
| 1179 |
+
"step": 1750
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.9,
|
| 1183 |
+
"learning_rate": 4.277414669571532e-05,
|
| 1184 |
+
"loss": 0.2158,
|
| 1185 |
+
"step": 1760
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.94,
|
| 1189 |
+
"learning_rate": 4.270152505446624e-05,
|
| 1190 |
+
"loss": 0.4392,
|
| 1191 |
+
"step": 1770
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 6.98,
|
| 1195 |
+
"learning_rate": 4.262890341321714e-05,
|
| 1196 |
+
"loss": 0.4321,
|
| 1197 |
+
"step": 1780
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 7.0,
|
| 1201 |
+
"eval_accuracy": {
|
| 1202 |
+
"accuracy": 0.8901960784313725
|
| 1203 |
+
},
|
| 1204 |
+
"eval_f1": {
|
| 1205 |
+
"f1": 0.8896787277785286
|
| 1206 |
+
},
|
| 1207 |
+
"eval_loss": 0.36420056223869324,
|
| 1208 |
+
"eval_precision": {
|
| 1209 |
+
"precision": 0.9002265766283616
|
| 1210 |
+
},
|
| 1211 |
+
"eval_recall": {
|
| 1212 |
+
"recall": 0.8878911425453978
|
| 1213 |
+
},
|
| 1214 |
+
"eval_runtime": 10.2347,
|
| 1215 |
+
"eval_samples_per_second": 49.83,
|
| 1216 |
+
"eval_steps_per_second": 24.915,
|
| 1217 |
+
"step": 1785
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 7.02,
|
| 1221 |
+
"learning_rate": 4.2556281771968046e-05,
|
| 1222 |
+
"loss": 0.5162,
|
| 1223 |
+
"step": 1790
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 7.06,
|
| 1227 |
+
"learning_rate": 4.2483660130718954e-05,
|
| 1228 |
+
"loss": 0.2491,
|
| 1229 |
+
"step": 1800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 7.1,
|
| 1233 |
+
"learning_rate": 4.241103848946986e-05,
|
| 1234 |
+
"loss": 0.2267,
|
| 1235 |
+
"step": 1810
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 7.14,
|
| 1239 |
+
"learning_rate": 4.233841684822078e-05,
|
| 1240 |
+
"loss": 0.2395,
|
| 1241 |
+
"step": 1820
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 7.18,
|
| 1245 |
+
"learning_rate": 4.226579520697168e-05,
|
| 1246 |
+
"loss": 0.3007,
|
| 1247 |
+
"step": 1830
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 7.22,
|
| 1251 |
+
"learning_rate": 4.2193173565722586e-05,
|
| 1252 |
+
"loss": 0.1316,
|
| 1253 |
+
"step": 1840
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 7.25,
|
| 1257 |
+
"learning_rate": 4.2120551924473494e-05,
|
| 1258 |
+
"loss": 0.1592,
|
| 1259 |
+
"step": 1850
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 7.29,
|
| 1263 |
+
"learning_rate": 4.20479302832244e-05,
|
| 1264 |
+
"loss": 0.3467,
|
| 1265 |
+
"step": 1860
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 7.33,
|
| 1269 |
+
"learning_rate": 4.197530864197531e-05,
|
| 1270 |
+
"loss": 0.4195,
|
| 1271 |
+
"step": 1870
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 7.37,
|
| 1275 |
+
"learning_rate": 4.190268700072622e-05,
|
| 1276 |
+
"loss": 0.3561,
|
| 1277 |
+
"step": 1880
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 7.41,
|
| 1281 |
+
"learning_rate": 4.1830065359477126e-05,
|
| 1282 |
+
"loss": 0.1572,
|
| 1283 |
+
"step": 1890
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.45,
|
| 1287 |
+
"learning_rate": 4.1757443718228034e-05,
|
| 1288 |
+
"loss": 0.2104,
|
| 1289 |
+
"step": 1900
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 7.49,
|
| 1293 |
+
"learning_rate": 4.168482207697894e-05,
|
| 1294 |
+
"loss": 0.3536,
|
| 1295 |
+
"step": 1910
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 7.53,
|
| 1299 |
+
"learning_rate": 4.161220043572985e-05,
|
| 1300 |
+
"loss": 0.345,
|
| 1301 |
+
"step": 1920
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 7.57,
|
| 1305 |
+
"learning_rate": 4.153957879448076e-05,
|
| 1306 |
+
"loss": 0.1083,
|
| 1307 |
+
"step": 1930
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 7.61,
|
| 1311 |
+
"learning_rate": 4.1466957153231666e-05,
|
| 1312 |
+
"loss": 0.1895,
|
| 1313 |
+
"step": 1940
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 7.65,
|
| 1317 |
+
"learning_rate": 4.1394335511982573e-05,
|
| 1318 |
+
"loss": 0.3232,
|
| 1319 |
+
"step": 1950
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 7.69,
|
| 1323 |
+
"learning_rate": 4.132171387073348e-05,
|
| 1324 |
+
"loss": 0.2939,
|
| 1325 |
+
"step": 1960
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.73,
|
| 1329 |
+
"learning_rate": 4.124909222948439e-05,
|
| 1330 |
+
"loss": 0.0918,
|
| 1331 |
+
"step": 1970
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 7.76,
|
| 1335 |
+
"learning_rate": 4.11764705882353e-05,
|
| 1336 |
+
"loss": 0.1722,
|
| 1337 |
+
"step": 1980
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 7.8,
|
| 1341 |
+
"learning_rate": 4.1103848946986205e-05,
|
| 1342 |
+
"loss": 0.3229,
|
| 1343 |
+
"step": 1990
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 7.84,
|
| 1347 |
+
"learning_rate": 4.103122730573711e-05,
|
| 1348 |
+
"loss": 0.1501,
|
| 1349 |
+
"step": 2000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 7.88,
|
| 1353 |
+
"learning_rate": 4.095860566448802e-05,
|
| 1354 |
+
"loss": 0.1468,
|
| 1355 |
+
"step": 2010
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 7.92,
|
| 1359 |
+
"learning_rate": 4.088598402323893e-05,
|
| 1360 |
+
"loss": 0.1665,
|
| 1361 |
+
"step": 2020
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 7.96,
|
| 1365 |
+
"learning_rate": 4.081336238198984e-05,
|
| 1366 |
+
"loss": 0.1324,
|
| 1367 |
+
"step": 2030
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 8.0,
|
| 1371 |
+
"learning_rate": 4.074074074074074e-05,
|
| 1372 |
+
"loss": 0.2618,
|
| 1373 |
+
"step": 2040
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 8.0,
|
| 1377 |
+
"eval_accuracy": {
|
| 1378 |
+
"accuracy": 0.8960784313725491
|
| 1379 |
+
},
|
| 1380 |
+
"eval_f1": {
|
| 1381 |
+
"f1": 0.8950352545633242
|
| 1382 |
+
},
|
| 1383 |
+
"eval_loss": 0.40923169255256653,
|
| 1384 |
+
"eval_precision": {
|
| 1385 |
+
"precision": 0.9033601037449228
|
| 1386 |
+
},
|
| 1387 |
+
"eval_recall": {
|
| 1388 |
+
"recall": 0.8953584093211753
|
| 1389 |
+
},
|
| 1390 |
+
"eval_runtime": 10.3235,
|
| 1391 |
+
"eval_samples_per_second": 49.402,
|
| 1392 |
+
"eval_steps_per_second": 24.701,
|
| 1393 |
+
"step": 2040
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 8.04,
|
| 1397 |
+
"learning_rate": 4.066811909949165e-05,
|
| 1398 |
+
"loss": 0.1887,
|
| 1399 |
+
"step": 2050
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 8.08,
|
| 1403 |
+
"learning_rate": 4.059549745824256e-05,
|
| 1404 |
+
"loss": 0.4153,
|
| 1405 |
+
"step": 2060
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 8.12,
|
| 1409 |
+
"learning_rate": 4.052287581699347e-05,
|
| 1410 |
+
"loss": 0.2771,
|
| 1411 |
+
"step": 2070
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 8.16,
|
| 1415 |
+
"learning_rate": 4.045025417574437e-05,
|
| 1416 |
+
"loss": 0.3316,
|
| 1417 |
+
"step": 2080
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 8.2,
|
| 1421 |
+
"learning_rate": 4.037763253449528e-05,
|
| 1422 |
+
"loss": 0.1125,
|
| 1423 |
+
"step": 2090
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 8.24,
|
| 1427 |
+
"learning_rate": 4.0305010893246186e-05,
|
| 1428 |
+
"loss": 0.3016,
|
| 1429 |
+
"step": 2100
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 8.27,
|
| 1433 |
+
"learning_rate": 4.02323892519971e-05,
|
| 1434 |
+
"loss": 0.2458,
|
| 1435 |
+
"step": 2110
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 8.31,
|
| 1439 |
+
"learning_rate": 4.015976761074801e-05,
|
| 1440 |
+
"loss": 0.179,
|
| 1441 |
+
"step": 2120
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 8.35,
|
| 1445 |
+
"learning_rate": 4.008714596949891e-05,
|
| 1446 |
+
"loss": 0.224,
|
| 1447 |
+
"step": 2130
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 8.39,
|
| 1451 |
+
"learning_rate": 4.001452432824982e-05,
|
| 1452 |
+
"loss": 0.0803,
|
| 1453 |
+
"step": 2140
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 8.43,
|
| 1457 |
+
"learning_rate": 3.9941902687000726e-05,
|
| 1458 |
+
"loss": 0.1121,
|
| 1459 |
+
"step": 2150
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 8.47,
|
| 1463 |
+
"learning_rate": 3.986928104575164e-05,
|
| 1464 |
+
"loss": 0.3751,
|
| 1465 |
+
"step": 2160
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 8.51,
|
| 1469 |
+
"learning_rate": 3.979665940450255e-05,
|
| 1470 |
+
"loss": 0.1354,
|
| 1471 |
+
"step": 2170
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 8.55,
|
| 1475 |
+
"learning_rate": 3.972403776325345e-05,
|
| 1476 |
+
"loss": 0.3105,
|
| 1477 |
+
"step": 2180
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 8.59,
|
| 1481 |
+
"learning_rate": 3.965141612200436e-05,
|
| 1482 |
+
"loss": 0.1937,
|
| 1483 |
+
"step": 2190
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 8.63,
|
| 1487 |
+
"learning_rate": 3.9578794480755265e-05,
|
| 1488 |
+
"loss": 0.3028,
|
| 1489 |
+
"step": 2200
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 8.67,
|
| 1493 |
+
"learning_rate": 3.950617283950617e-05,
|
| 1494 |
+
"loss": 0.1851,
|
| 1495 |
+
"step": 2210
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 8.71,
|
| 1499 |
+
"learning_rate": 3.943355119825709e-05,
|
| 1500 |
+
"loss": 0.074,
|
| 1501 |
+
"step": 2220
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 8.75,
|
| 1505 |
+
"learning_rate": 3.936092955700799e-05,
|
| 1506 |
+
"loss": 0.2098,
|
| 1507 |
+
"step": 2230
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 8.78,
|
| 1511 |
+
"learning_rate": 3.92883079157589e-05,
|
| 1512 |
+
"loss": 0.3167,
|
| 1513 |
+
"step": 2240
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 8.82,
|
| 1517 |
+
"learning_rate": 3.9215686274509805e-05,
|
| 1518 |
+
"loss": 0.2527,
|
| 1519 |
+
"step": 2250
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 8.86,
|
| 1523 |
+
"learning_rate": 3.914306463326071e-05,
|
| 1524 |
+
"loss": 0.1521,
|
| 1525 |
+
"step": 2260
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 8.9,
|
| 1529 |
+
"learning_rate": 3.907044299201162e-05,
|
| 1530 |
+
"loss": 0.1393,
|
| 1531 |
+
"step": 2270
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 8.94,
|
| 1535 |
+
"learning_rate": 3.899782135076253e-05,
|
| 1536 |
+
"loss": 0.1155,
|
| 1537 |
+
"step": 2280
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 8.98,
|
| 1541 |
+
"learning_rate": 3.892519970951344e-05,
|
| 1542 |
+
"loss": 0.3376,
|
| 1543 |
+
"step": 2290
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 9.0,
|
| 1547 |
+
"eval_accuracy": {
|
| 1548 |
+
"accuracy": 0.8941176470588236
|
| 1549 |
+
},
|
| 1550 |
+
"eval_f1": {
|
| 1551 |
+
"f1": 0.8933906521569781
|
| 1552 |
+
},
|
| 1553 |
+
"eval_loss": 0.4055422246456146,
|
| 1554 |
+
"eval_precision": {
|
| 1555 |
+
"precision": 0.9014239716079142
|
| 1556 |
+
},
|
| 1557 |
+
"eval_recall": {
|
| 1558 |
+
"recall": 0.8936214051905541
|
| 1559 |
+
},
|
| 1560 |
+
"eval_runtime": 10.1735,
|
| 1561 |
+
"eval_samples_per_second": 50.13,
|
| 1562 |
+
"eval_steps_per_second": 25.065,
|
| 1563 |
+
"step": 2295
|
| 1564 |
+
}
|
| 1565 |
+
],
|
| 1566 |
+
"logging_steps": 10,
|
| 1567 |
+
"max_steps": 7650,
|
| 1568 |
+
"num_input_tokens_seen": 0,
|
| 1569 |
+
"num_train_epochs": 30,
|
| 1570 |
+
"save_steps": 500,
|
| 1571 |
+
"total_flos": 1.4227911843773645e+18,
|
| 1572 |
+
"train_batch_size": 2,
|
| 1573 |
+
"trial_name": null,
|
| 1574 |
+
"trial_params": null
|
| 1575 |
+
}
|
tmp-checkpoint-2295/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-2550/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-2550/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:71220e4f8d8fc91d4e6453cf368927ac3ea0f9337c62964d4e2c2c8b93954159
|
| 3 |
+
size 343233204
|
tmp-checkpoint-2550/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ca9597fe8d53bdc92fc4bf5143c2a8c05da5f4bc67f58f86fced483dd732064
|
| 3 |
+
size 686587322
|
tmp-checkpoint-2550/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-2550/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ee10e3884ffd0e97647f33450eef155a972520ba2cbb17efa6399958cf9560b
|
| 3 |
+
size 14244
|
tmp-checkpoint-2550/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0cd9636c0bfbde324a68f3a372e5329c5955a222a38bd15ff030366b8b2fb72e
|
| 3 |
+
size 1064
|
tmp-checkpoint-2550/trainer_state.json
ADDED
|
@@ -0,0 +1,1751 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.36420056223869324,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-1785",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2550,
|
| 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 |
+
"epoch": 3.02,
|
| 529 |
+
"learning_rate": 4.9963689179375456e-05,
|
| 530 |
+
"loss": 0.3347,
|
| 531 |
+
"step": 770
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 3.06,
|
| 535 |
+
"learning_rate": 4.9891067538126364e-05,
|
| 536 |
+
"loss": 0.3339,
|
| 537 |
+
"step": 780
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 3.1,
|
| 541 |
+
"learning_rate": 4.981844589687727e-05,
|
| 542 |
+
"loss": 0.5573,
|
| 543 |
+
"step": 790
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 3.14,
|
| 547 |
+
"learning_rate": 4.974582425562818e-05,
|
| 548 |
+
"loss": 0.5463,
|
| 549 |
+
"step": 800
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.18,
|
| 553 |
+
"learning_rate": 4.967320261437909e-05,
|
| 554 |
+
"loss": 0.3453,
|
| 555 |
+
"step": 810
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.22,
|
| 559 |
+
"learning_rate": 4.9600580973129996e-05,
|
| 560 |
+
"loss": 0.3814,
|
| 561 |
+
"step": 820
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.25,
|
| 565 |
+
"learning_rate": 4.9527959331880904e-05,
|
| 566 |
+
"loss": 0.4434,
|
| 567 |
+
"step": 830
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 3.29,
|
| 571 |
+
"learning_rate": 4.945533769063181e-05,
|
| 572 |
+
"loss": 0.3675,
|
| 573 |
+
"step": 840
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 3.33,
|
| 577 |
+
"learning_rate": 4.938271604938271e-05,
|
| 578 |
+
"loss": 0.5386,
|
| 579 |
+
"step": 850
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 3.37,
|
| 583 |
+
"learning_rate": 4.931009440813363e-05,
|
| 584 |
+
"loss": 0.2692,
|
| 585 |
+
"step": 860
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.41,
|
| 589 |
+
"learning_rate": 4.9237472766884536e-05,
|
| 590 |
+
"loss": 0.4815,
|
| 591 |
+
"step": 870
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.45,
|
| 595 |
+
"learning_rate": 4.9164851125635444e-05,
|
| 596 |
+
"loss": 0.4178,
|
| 597 |
+
"step": 880
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.49,
|
| 601 |
+
"learning_rate": 4.909222948438635e-05,
|
| 602 |
+
"loss": 0.3618,
|
| 603 |
+
"step": 890
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.53,
|
| 607 |
+
"learning_rate": 4.901960784313725e-05,
|
| 608 |
+
"loss": 0.6841,
|
| 609 |
+
"step": 900
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 3.57,
|
| 613 |
+
"learning_rate": 4.894698620188817e-05,
|
| 614 |
+
"loss": 0.471,
|
| 615 |
+
"step": 910
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.61,
|
| 619 |
+
"learning_rate": 4.8874364560639076e-05,
|
| 620 |
+
"loss": 0.313,
|
| 621 |
+
"step": 920
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 3.65,
|
| 625 |
+
"learning_rate": 4.8801742919389983e-05,
|
| 626 |
+
"loss": 0.2437,
|
| 627 |
+
"step": 930
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 3.69,
|
| 631 |
+
"learning_rate": 4.8729121278140885e-05,
|
| 632 |
+
"loss": 0.4748,
|
| 633 |
+
"step": 940
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.73,
|
| 637 |
+
"learning_rate": 4.865649963689179e-05,
|
| 638 |
+
"loss": 0.5519,
|
| 639 |
+
"step": 950
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.76,
|
| 643 |
+
"learning_rate": 4.85838779956427e-05,
|
| 644 |
+
"loss": 0.3051,
|
| 645 |
+
"step": 960
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.8,
|
| 649 |
+
"learning_rate": 4.8511256354393615e-05,
|
| 650 |
+
"loss": 0.3847,
|
| 651 |
+
"step": 970
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 3.84,
|
| 655 |
+
"learning_rate": 4.843863471314452e-05,
|
| 656 |
+
"loss": 0.2454,
|
| 657 |
+
"step": 980
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.88,
|
| 661 |
+
"learning_rate": 4.8366013071895424e-05,
|
| 662 |
+
"loss": 0.379,
|
| 663 |
+
"step": 990
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 3.92,
|
| 667 |
+
"learning_rate": 4.829339143064633e-05,
|
| 668 |
+
"loss": 0.4485,
|
| 669 |
+
"step": 1000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 3.96,
|
| 673 |
+
"learning_rate": 4.822076978939724e-05,
|
| 674 |
+
"loss": 0.2484,
|
| 675 |
+
"step": 1010
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.0,
|
| 679 |
+
"learning_rate": 4.814814814814815e-05,
|
| 680 |
+
"loss": 0.4402,
|
| 681 |
+
"step": 1020
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 4.0,
|
| 685 |
+
"eval_accuracy": {
|
| 686 |
+
"accuracy": 0.8784313725490196
|
| 687 |
+
},
|
| 688 |
+
"eval_f1": {
|
| 689 |
+
"f1": 0.880471094803992
|
| 690 |
+
},
|
| 691 |
+
"eval_loss": 0.3946979343891144,
|
| 692 |
+
"eval_precision": {
|
| 693 |
+
"precision": 0.8952890487430221
|
| 694 |
+
},
|
| 695 |
+
"eval_recall": {
|
| 696 |
+
"recall": 0.8776488582339645
|
| 697 |
+
},
|
| 698 |
+
"eval_runtime": 10.1771,
|
| 699 |
+
"eval_samples_per_second": 50.113,
|
| 700 |
+
"eval_steps_per_second": 25.056,
|
| 701 |
+
"step": 1020
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 4.04,
|
| 705 |
+
"learning_rate": 4.807552650689906e-05,
|
| 706 |
+
"loss": 0.3494,
|
| 707 |
+
"step": 1030
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 4.08,
|
| 711 |
+
"learning_rate": 4.8002904865649964e-05,
|
| 712 |
+
"loss": 0.3327,
|
| 713 |
+
"step": 1040
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 4.12,
|
| 717 |
+
"learning_rate": 4.793028322440087e-05,
|
| 718 |
+
"loss": 0.2898,
|
| 719 |
+
"step": 1050
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 4.16,
|
| 723 |
+
"learning_rate": 4.785766158315178e-05,
|
| 724 |
+
"loss": 0.3965,
|
| 725 |
+
"step": 1060
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 4.2,
|
| 729 |
+
"learning_rate": 4.778503994190269e-05,
|
| 730 |
+
"loss": 0.3607,
|
| 731 |
+
"step": 1070
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.24,
|
| 735 |
+
"learning_rate": 4.77124183006536e-05,
|
| 736 |
+
"loss": 0.3714,
|
| 737 |
+
"step": 1080
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.27,
|
| 741 |
+
"learning_rate": 4.7639796659404504e-05,
|
| 742 |
+
"loss": 0.1796,
|
| 743 |
+
"step": 1090
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 4.31,
|
| 747 |
+
"learning_rate": 4.756717501815541e-05,
|
| 748 |
+
"loss": 0.1874,
|
| 749 |
+
"step": 1100
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 4.35,
|
| 753 |
+
"learning_rate": 4.749455337690632e-05,
|
| 754 |
+
"loss": 0.3462,
|
| 755 |
+
"step": 1110
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 4.39,
|
| 759 |
+
"learning_rate": 4.742193173565723e-05,
|
| 760 |
+
"loss": 0.2082,
|
| 761 |
+
"step": 1120
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 4.43,
|
| 765 |
+
"learning_rate": 4.7349310094408136e-05,
|
| 766 |
+
"loss": 0.4531,
|
| 767 |
+
"step": 1130
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 4.47,
|
| 771 |
+
"learning_rate": 4.7276688453159044e-05,
|
| 772 |
+
"loss": 0.3951,
|
| 773 |
+
"step": 1140
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.51,
|
| 777 |
+
"learning_rate": 4.720406681190995e-05,
|
| 778 |
+
"loss": 0.2337,
|
| 779 |
+
"step": 1150
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.55,
|
| 783 |
+
"learning_rate": 4.713144517066086e-05,
|
| 784 |
+
"loss": 0.2118,
|
| 785 |
+
"step": 1160
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 4.59,
|
| 789 |
+
"learning_rate": 4.705882352941177e-05,
|
| 790 |
+
"loss": 0.373,
|
| 791 |
+
"step": 1170
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 4.63,
|
| 795 |
+
"learning_rate": 4.6986201888162675e-05,
|
| 796 |
+
"loss": 0.1484,
|
| 797 |
+
"step": 1180
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 4.67,
|
| 801 |
+
"learning_rate": 4.691358024691358e-05,
|
| 802 |
+
"loss": 0.2568,
|
| 803 |
+
"step": 1190
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 4.71,
|
| 807 |
+
"learning_rate": 4.684095860566449e-05,
|
| 808 |
+
"loss": 0.3067,
|
| 809 |
+
"step": 1200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 4.75,
|
| 813 |
+
"learning_rate": 4.67683369644154e-05,
|
| 814 |
+
"loss": 0.2007,
|
| 815 |
+
"step": 1210
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.78,
|
| 819 |
+
"learning_rate": 4.669571532316631e-05,
|
| 820 |
+
"loss": 0.2493,
|
| 821 |
+
"step": 1220
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.82,
|
| 825 |
+
"learning_rate": 4.6623093681917215e-05,
|
| 826 |
+
"loss": 0.2717,
|
| 827 |
+
"step": 1230
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 4.86,
|
| 831 |
+
"learning_rate": 4.6550472040668116e-05,
|
| 832 |
+
"loss": 0.4134,
|
| 833 |
+
"step": 1240
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 4.9,
|
| 837 |
+
"learning_rate": 4.647785039941903e-05,
|
| 838 |
+
"loss": 0.2417,
|
| 839 |
+
"step": 1250
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 4.94,
|
| 843 |
+
"learning_rate": 4.640522875816994e-05,
|
| 844 |
+
"loss": 0.2573,
|
| 845 |
+
"step": 1260
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 4.98,
|
| 849 |
+
"learning_rate": 4.633260711692085e-05,
|
| 850 |
+
"loss": 0.3937,
|
| 851 |
+
"step": 1270
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 5.0,
|
| 855 |
+
"eval_accuracy": {
|
| 856 |
+
"accuracy": 0.8725490196078431
|
| 857 |
+
},
|
| 858 |
+
"eval_f1": {
|
| 859 |
+
"f1": 0.8703780158419334
|
| 860 |
+
},
|
| 861 |
+
"eval_loss": 0.4377364218235016,
|
| 862 |
+
"eval_precision": {
|
| 863 |
+
"precision": 0.8900760323128502
|
| 864 |
+
},
|
| 865 |
+
"eval_recall": {
|
| 866 |
+
"recall": 0.8691138648585458
|
| 867 |
+
},
|
| 868 |
+
"eval_runtime": 10.1762,
|
| 869 |
+
"eval_samples_per_second": 50.117,
|
| 870 |
+
"eval_steps_per_second": 25.058,
|
| 871 |
+
"step": 1275
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 5.02,
|
| 875 |
+
"learning_rate": 4.6259985475671755e-05,
|
| 876 |
+
"loss": 0.165,
|
| 877 |
+
"step": 1280
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 5.06,
|
| 881 |
+
"learning_rate": 4.6187363834422656e-05,
|
| 882 |
+
"loss": 0.1452,
|
| 883 |
+
"step": 1290
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 5.1,
|
| 887 |
+
"learning_rate": 4.6114742193173564e-05,
|
| 888 |
+
"loss": 0.1548,
|
| 889 |
+
"step": 1300
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 5.14,
|
| 893 |
+
"learning_rate": 4.604212055192448e-05,
|
| 894 |
+
"loss": 0.4443,
|
| 895 |
+
"step": 1310
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 5.18,
|
| 899 |
+
"learning_rate": 4.5969498910675387e-05,
|
| 900 |
+
"loss": 0.4322,
|
| 901 |
+
"step": 1320
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 5.22,
|
| 905 |
+
"learning_rate": 4.5896877269426294e-05,
|
| 906 |
+
"loss": 0.2355,
|
| 907 |
+
"step": 1330
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 5.25,
|
| 911 |
+
"learning_rate": 4.5824255628177196e-05,
|
| 912 |
+
"loss": 0.1949,
|
| 913 |
+
"step": 1340
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 5.29,
|
| 917 |
+
"learning_rate": 4.5751633986928104e-05,
|
| 918 |
+
"loss": 0.1959,
|
| 919 |
+
"step": 1350
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.33,
|
| 923 |
+
"learning_rate": 4.567901234567901e-05,
|
| 924 |
+
"loss": 0.3841,
|
| 925 |
+
"step": 1360
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 5.37,
|
| 929 |
+
"learning_rate": 4.5606390704429926e-05,
|
| 930 |
+
"loss": 0.1401,
|
| 931 |
+
"step": 1370
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 5.41,
|
| 935 |
+
"learning_rate": 4.5533769063180834e-05,
|
| 936 |
+
"loss": 0.1148,
|
| 937 |
+
"step": 1380
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 5.45,
|
| 941 |
+
"learning_rate": 4.5461147421931735e-05,
|
| 942 |
+
"loss": 0.234,
|
| 943 |
+
"step": 1390
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 5.49,
|
| 947 |
+
"learning_rate": 4.538852578068264e-05,
|
| 948 |
+
"loss": 0.3899,
|
| 949 |
+
"step": 1400
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 5.53,
|
| 953 |
+
"learning_rate": 4.531590413943355e-05,
|
| 954 |
+
"loss": 0.4873,
|
| 955 |
+
"step": 1410
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 5.57,
|
| 959 |
+
"learning_rate": 4.5243282498184466e-05,
|
| 960 |
+
"loss": 0.2483,
|
| 961 |
+
"step": 1420
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.61,
|
| 965 |
+
"learning_rate": 4.5170660856935374e-05,
|
| 966 |
+
"loss": 0.2654,
|
| 967 |
+
"step": 1430
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 5.65,
|
| 971 |
+
"learning_rate": 4.5098039215686275e-05,
|
| 972 |
+
"loss": 0.2035,
|
| 973 |
+
"step": 1440
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 5.69,
|
| 977 |
+
"learning_rate": 4.502541757443718e-05,
|
| 978 |
+
"loss": 0.3744,
|
| 979 |
+
"step": 1450
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 5.73,
|
| 983 |
+
"learning_rate": 4.495279593318809e-05,
|
| 984 |
+
"loss": 0.2139,
|
| 985 |
+
"step": 1460
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 5.76,
|
| 989 |
+
"learning_rate": 4.4880174291939e-05,
|
| 990 |
+
"loss": 0.2212,
|
| 991 |
+
"step": 1470
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 5.8,
|
| 995 |
+
"learning_rate": 4.480755265068991e-05,
|
| 996 |
+
"loss": 0.376,
|
| 997 |
+
"step": 1480
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 5.84,
|
| 1001 |
+
"learning_rate": 4.4734931009440815e-05,
|
| 1002 |
+
"loss": 0.2399,
|
| 1003 |
+
"step": 1490
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.88,
|
| 1007 |
+
"learning_rate": 4.466230936819172e-05,
|
| 1008 |
+
"loss": 0.3534,
|
| 1009 |
+
"step": 1500
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 5.92,
|
| 1013 |
+
"learning_rate": 4.458968772694263e-05,
|
| 1014 |
+
"loss": 0.3667,
|
| 1015 |
+
"step": 1510
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 5.96,
|
| 1019 |
+
"learning_rate": 4.451706608569354e-05,
|
| 1020 |
+
"loss": 0.496,
|
| 1021 |
+
"step": 1520
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 6.0,
|
| 1025 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 1026 |
+
"loss": 0.4042,
|
| 1027 |
+
"step": 1530
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 6.0,
|
| 1031 |
+
"eval_accuracy": {
|
| 1032 |
+
"accuracy": 0.8862745098039215
|
| 1033 |
+
},
|
| 1034 |
+
"eval_f1": {
|
| 1035 |
+
"f1": 0.8859196308355288
|
| 1036 |
+
},
|
| 1037 |
+
"eval_loss": 0.39891302585601807,
|
| 1038 |
+
"eval_precision": {
|
| 1039 |
+
"precision": 0.8972093144987638
|
| 1040 |
+
},
|
| 1041 |
+
"eval_recall": {
|
| 1042 |
+
"recall": 0.8834955381497934
|
| 1043 |
+
},
|
| 1044 |
+
"eval_runtime": 10.1174,
|
| 1045 |
+
"eval_samples_per_second": 50.408,
|
| 1046 |
+
"eval_steps_per_second": 25.204,
|
| 1047 |
+
"step": 1530
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 6.04,
|
| 1051 |
+
"learning_rate": 4.4371822803195355e-05,
|
| 1052 |
+
"loss": 0.2637,
|
| 1053 |
+
"step": 1540
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 6.08,
|
| 1057 |
+
"learning_rate": 4.429920116194626e-05,
|
| 1058 |
+
"loss": 0.2821,
|
| 1059 |
+
"step": 1550
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 6.12,
|
| 1063 |
+
"learning_rate": 4.422657952069717e-05,
|
| 1064 |
+
"loss": 0.1201,
|
| 1065 |
+
"step": 1560
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 6.16,
|
| 1069 |
+
"learning_rate": 4.415395787944808e-05,
|
| 1070 |
+
"loss": 0.262,
|
| 1071 |
+
"step": 1570
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 6.2,
|
| 1075 |
+
"learning_rate": 4.4081336238198986e-05,
|
| 1076 |
+
"loss": 0.2064,
|
| 1077 |
+
"step": 1580
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 6.24,
|
| 1081 |
+
"learning_rate": 4.400871459694989e-05,
|
| 1082 |
+
"loss": 0.1815,
|
| 1083 |
+
"step": 1590
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 6.27,
|
| 1087 |
+
"learning_rate": 4.39360929557008e-05,
|
| 1088 |
+
"loss": 0.1959,
|
| 1089 |
+
"step": 1600
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 6.31,
|
| 1093 |
+
"learning_rate": 4.386347131445171e-05,
|
| 1094 |
+
"loss": 0.1959,
|
| 1095 |
+
"step": 1610
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.35,
|
| 1099 |
+
"learning_rate": 4.379084967320262e-05,
|
| 1100 |
+
"loss": 0.2398,
|
| 1101 |
+
"step": 1620
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.39,
|
| 1105 |
+
"learning_rate": 4.3718228031953526e-05,
|
| 1106 |
+
"loss": 0.2038,
|
| 1107 |
+
"step": 1630
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 6.43,
|
| 1111 |
+
"learning_rate": 4.364560639070443e-05,
|
| 1112 |
+
"loss": 0.2004,
|
| 1113 |
+
"step": 1640
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.47,
|
| 1117 |
+
"learning_rate": 4.357298474945534e-05,
|
| 1118 |
+
"loss": 0.0377,
|
| 1119 |
+
"step": 1650
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 6.51,
|
| 1123 |
+
"learning_rate": 4.350036310820625e-05,
|
| 1124 |
+
"loss": 0.2022,
|
| 1125 |
+
"step": 1660
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 6.55,
|
| 1129 |
+
"learning_rate": 4.342774146695716e-05,
|
| 1130 |
+
"loss": 0.4403,
|
| 1131 |
+
"step": 1670
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 6.59,
|
| 1135 |
+
"learning_rate": 4.3355119825708066e-05,
|
| 1136 |
+
"loss": 0.2577,
|
| 1137 |
+
"step": 1680
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.63,
|
| 1141 |
+
"learning_rate": 4.328249818445897e-05,
|
| 1142 |
+
"loss": 0.3312,
|
| 1143 |
+
"step": 1690
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.67,
|
| 1147 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 1148 |
+
"loss": 0.307,
|
| 1149 |
+
"step": 1700
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 6.71,
|
| 1153 |
+
"learning_rate": 4.313725490196079e-05,
|
| 1154 |
+
"loss": 0.0961,
|
| 1155 |
+
"step": 1710
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.75,
|
| 1159 |
+
"learning_rate": 4.30646332607117e-05,
|
| 1160 |
+
"loss": 0.3266,
|
| 1161 |
+
"step": 1720
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 6.78,
|
| 1165 |
+
"learning_rate": 4.2992011619462605e-05,
|
| 1166 |
+
"loss": 0.2601,
|
| 1167 |
+
"step": 1730
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 6.82,
|
| 1171 |
+
"learning_rate": 4.291938997821351e-05,
|
| 1172 |
+
"loss": 0.2382,
|
| 1173 |
+
"step": 1740
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 6.86,
|
| 1177 |
+
"learning_rate": 4.2846768336964415e-05,
|
| 1178 |
+
"loss": 0.1469,
|
| 1179 |
+
"step": 1750
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.9,
|
| 1183 |
+
"learning_rate": 4.277414669571532e-05,
|
| 1184 |
+
"loss": 0.2158,
|
| 1185 |
+
"step": 1760
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.94,
|
| 1189 |
+
"learning_rate": 4.270152505446624e-05,
|
| 1190 |
+
"loss": 0.4392,
|
| 1191 |
+
"step": 1770
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 6.98,
|
| 1195 |
+
"learning_rate": 4.262890341321714e-05,
|
| 1196 |
+
"loss": 0.4321,
|
| 1197 |
+
"step": 1780
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 7.0,
|
| 1201 |
+
"eval_accuracy": {
|
| 1202 |
+
"accuracy": 0.8901960784313725
|
| 1203 |
+
},
|
| 1204 |
+
"eval_f1": {
|
| 1205 |
+
"f1": 0.8896787277785286
|
| 1206 |
+
},
|
| 1207 |
+
"eval_loss": 0.36420056223869324,
|
| 1208 |
+
"eval_precision": {
|
| 1209 |
+
"precision": 0.9002265766283616
|
| 1210 |
+
},
|
| 1211 |
+
"eval_recall": {
|
| 1212 |
+
"recall": 0.8878911425453978
|
| 1213 |
+
},
|
| 1214 |
+
"eval_runtime": 10.2347,
|
| 1215 |
+
"eval_samples_per_second": 49.83,
|
| 1216 |
+
"eval_steps_per_second": 24.915,
|
| 1217 |
+
"step": 1785
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 7.02,
|
| 1221 |
+
"learning_rate": 4.2556281771968046e-05,
|
| 1222 |
+
"loss": 0.5162,
|
| 1223 |
+
"step": 1790
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 7.06,
|
| 1227 |
+
"learning_rate": 4.2483660130718954e-05,
|
| 1228 |
+
"loss": 0.2491,
|
| 1229 |
+
"step": 1800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 7.1,
|
| 1233 |
+
"learning_rate": 4.241103848946986e-05,
|
| 1234 |
+
"loss": 0.2267,
|
| 1235 |
+
"step": 1810
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 7.14,
|
| 1239 |
+
"learning_rate": 4.233841684822078e-05,
|
| 1240 |
+
"loss": 0.2395,
|
| 1241 |
+
"step": 1820
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 7.18,
|
| 1245 |
+
"learning_rate": 4.226579520697168e-05,
|
| 1246 |
+
"loss": 0.3007,
|
| 1247 |
+
"step": 1830
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 7.22,
|
| 1251 |
+
"learning_rate": 4.2193173565722586e-05,
|
| 1252 |
+
"loss": 0.1316,
|
| 1253 |
+
"step": 1840
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 7.25,
|
| 1257 |
+
"learning_rate": 4.2120551924473494e-05,
|
| 1258 |
+
"loss": 0.1592,
|
| 1259 |
+
"step": 1850
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 7.29,
|
| 1263 |
+
"learning_rate": 4.20479302832244e-05,
|
| 1264 |
+
"loss": 0.3467,
|
| 1265 |
+
"step": 1860
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 7.33,
|
| 1269 |
+
"learning_rate": 4.197530864197531e-05,
|
| 1270 |
+
"loss": 0.4195,
|
| 1271 |
+
"step": 1870
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 7.37,
|
| 1275 |
+
"learning_rate": 4.190268700072622e-05,
|
| 1276 |
+
"loss": 0.3561,
|
| 1277 |
+
"step": 1880
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 7.41,
|
| 1281 |
+
"learning_rate": 4.1830065359477126e-05,
|
| 1282 |
+
"loss": 0.1572,
|
| 1283 |
+
"step": 1890
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.45,
|
| 1287 |
+
"learning_rate": 4.1757443718228034e-05,
|
| 1288 |
+
"loss": 0.2104,
|
| 1289 |
+
"step": 1900
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 7.49,
|
| 1293 |
+
"learning_rate": 4.168482207697894e-05,
|
| 1294 |
+
"loss": 0.3536,
|
| 1295 |
+
"step": 1910
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 7.53,
|
| 1299 |
+
"learning_rate": 4.161220043572985e-05,
|
| 1300 |
+
"loss": 0.345,
|
| 1301 |
+
"step": 1920
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 7.57,
|
| 1305 |
+
"learning_rate": 4.153957879448076e-05,
|
| 1306 |
+
"loss": 0.1083,
|
| 1307 |
+
"step": 1930
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 7.61,
|
| 1311 |
+
"learning_rate": 4.1466957153231666e-05,
|
| 1312 |
+
"loss": 0.1895,
|
| 1313 |
+
"step": 1940
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 7.65,
|
| 1317 |
+
"learning_rate": 4.1394335511982573e-05,
|
| 1318 |
+
"loss": 0.3232,
|
| 1319 |
+
"step": 1950
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 7.69,
|
| 1323 |
+
"learning_rate": 4.132171387073348e-05,
|
| 1324 |
+
"loss": 0.2939,
|
| 1325 |
+
"step": 1960
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.73,
|
| 1329 |
+
"learning_rate": 4.124909222948439e-05,
|
| 1330 |
+
"loss": 0.0918,
|
| 1331 |
+
"step": 1970
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 7.76,
|
| 1335 |
+
"learning_rate": 4.11764705882353e-05,
|
| 1336 |
+
"loss": 0.1722,
|
| 1337 |
+
"step": 1980
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 7.8,
|
| 1341 |
+
"learning_rate": 4.1103848946986205e-05,
|
| 1342 |
+
"loss": 0.3229,
|
| 1343 |
+
"step": 1990
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 7.84,
|
| 1347 |
+
"learning_rate": 4.103122730573711e-05,
|
| 1348 |
+
"loss": 0.1501,
|
| 1349 |
+
"step": 2000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 7.88,
|
| 1353 |
+
"learning_rate": 4.095860566448802e-05,
|
| 1354 |
+
"loss": 0.1468,
|
| 1355 |
+
"step": 2010
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 7.92,
|
| 1359 |
+
"learning_rate": 4.088598402323893e-05,
|
| 1360 |
+
"loss": 0.1665,
|
| 1361 |
+
"step": 2020
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 7.96,
|
| 1365 |
+
"learning_rate": 4.081336238198984e-05,
|
| 1366 |
+
"loss": 0.1324,
|
| 1367 |
+
"step": 2030
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 8.0,
|
| 1371 |
+
"learning_rate": 4.074074074074074e-05,
|
| 1372 |
+
"loss": 0.2618,
|
| 1373 |
+
"step": 2040
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 8.0,
|
| 1377 |
+
"eval_accuracy": {
|
| 1378 |
+
"accuracy": 0.8960784313725491
|
| 1379 |
+
},
|
| 1380 |
+
"eval_f1": {
|
| 1381 |
+
"f1": 0.8950352545633242
|
| 1382 |
+
},
|
| 1383 |
+
"eval_loss": 0.40923169255256653,
|
| 1384 |
+
"eval_precision": {
|
| 1385 |
+
"precision": 0.9033601037449228
|
| 1386 |
+
},
|
| 1387 |
+
"eval_recall": {
|
| 1388 |
+
"recall": 0.8953584093211753
|
| 1389 |
+
},
|
| 1390 |
+
"eval_runtime": 10.3235,
|
| 1391 |
+
"eval_samples_per_second": 49.402,
|
| 1392 |
+
"eval_steps_per_second": 24.701,
|
| 1393 |
+
"step": 2040
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 8.04,
|
| 1397 |
+
"learning_rate": 4.066811909949165e-05,
|
| 1398 |
+
"loss": 0.1887,
|
| 1399 |
+
"step": 2050
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 8.08,
|
| 1403 |
+
"learning_rate": 4.059549745824256e-05,
|
| 1404 |
+
"loss": 0.4153,
|
| 1405 |
+
"step": 2060
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 8.12,
|
| 1409 |
+
"learning_rate": 4.052287581699347e-05,
|
| 1410 |
+
"loss": 0.2771,
|
| 1411 |
+
"step": 2070
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 8.16,
|
| 1415 |
+
"learning_rate": 4.045025417574437e-05,
|
| 1416 |
+
"loss": 0.3316,
|
| 1417 |
+
"step": 2080
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 8.2,
|
| 1421 |
+
"learning_rate": 4.037763253449528e-05,
|
| 1422 |
+
"loss": 0.1125,
|
| 1423 |
+
"step": 2090
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 8.24,
|
| 1427 |
+
"learning_rate": 4.0305010893246186e-05,
|
| 1428 |
+
"loss": 0.3016,
|
| 1429 |
+
"step": 2100
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 8.27,
|
| 1433 |
+
"learning_rate": 4.02323892519971e-05,
|
| 1434 |
+
"loss": 0.2458,
|
| 1435 |
+
"step": 2110
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 8.31,
|
| 1439 |
+
"learning_rate": 4.015976761074801e-05,
|
| 1440 |
+
"loss": 0.179,
|
| 1441 |
+
"step": 2120
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 8.35,
|
| 1445 |
+
"learning_rate": 4.008714596949891e-05,
|
| 1446 |
+
"loss": 0.224,
|
| 1447 |
+
"step": 2130
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 8.39,
|
| 1451 |
+
"learning_rate": 4.001452432824982e-05,
|
| 1452 |
+
"loss": 0.0803,
|
| 1453 |
+
"step": 2140
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 8.43,
|
| 1457 |
+
"learning_rate": 3.9941902687000726e-05,
|
| 1458 |
+
"loss": 0.1121,
|
| 1459 |
+
"step": 2150
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 8.47,
|
| 1463 |
+
"learning_rate": 3.986928104575164e-05,
|
| 1464 |
+
"loss": 0.3751,
|
| 1465 |
+
"step": 2160
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 8.51,
|
| 1469 |
+
"learning_rate": 3.979665940450255e-05,
|
| 1470 |
+
"loss": 0.1354,
|
| 1471 |
+
"step": 2170
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 8.55,
|
| 1475 |
+
"learning_rate": 3.972403776325345e-05,
|
| 1476 |
+
"loss": 0.3105,
|
| 1477 |
+
"step": 2180
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 8.59,
|
| 1481 |
+
"learning_rate": 3.965141612200436e-05,
|
| 1482 |
+
"loss": 0.1937,
|
| 1483 |
+
"step": 2190
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 8.63,
|
| 1487 |
+
"learning_rate": 3.9578794480755265e-05,
|
| 1488 |
+
"loss": 0.3028,
|
| 1489 |
+
"step": 2200
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 8.67,
|
| 1493 |
+
"learning_rate": 3.950617283950617e-05,
|
| 1494 |
+
"loss": 0.1851,
|
| 1495 |
+
"step": 2210
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 8.71,
|
| 1499 |
+
"learning_rate": 3.943355119825709e-05,
|
| 1500 |
+
"loss": 0.074,
|
| 1501 |
+
"step": 2220
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 8.75,
|
| 1505 |
+
"learning_rate": 3.936092955700799e-05,
|
| 1506 |
+
"loss": 0.2098,
|
| 1507 |
+
"step": 2230
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 8.78,
|
| 1511 |
+
"learning_rate": 3.92883079157589e-05,
|
| 1512 |
+
"loss": 0.3167,
|
| 1513 |
+
"step": 2240
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 8.82,
|
| 1517 |
+
"learning_rate": 3.9215686274509805e-05,
|
| 1518 |
+
"loss": 0.2527,
|
| 1519 |
+
"step": 2250
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 8.86,
|
| 1523 |
+
"learning_rate": 3.914306463326071e-05,
|
| 1524 |
+
"loss": 0.1521,
|
| 1525 |
+
"step": 2260
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 8.9,
|
| 1529 |
+
"learning_rate": 3.907044299201162e-05,
|
| 1530 |
+
"loss": 0.1393,
|
| 1531 |
+
"step": 2270
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 8.94,
|
| 1535 |
+
"learning_rate": 3.899782135076253e-05,
|
| 1536 |
+
"loss": 0.1155,
|
| 1537 |
+
"step": 2280
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 8.98,
|
| 1541 |
+
"learning_rate": 3.892519970951344e-05,
|
| 1542 |
+
"loss": 0.3376,
|
| 1543 |
+
"step": 2290
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 9.0,
|
| 1547 |
+
"eval_accuracy": {
|
| 1548 |
+
"accuracy": 0.8941176470588236
|
| 1549 |
+
},
|
| 1550 |
+
"eval_f1": {
|
| 1551 |
+
"f1": 0.8933906521569781
|
| 1552 |
+
},
|
| 1553 |
+
"eval_loss": 0.4055422246456146,
|
| 1554 |
+
"eval_precision": {
|
| 1555 |
+
"precision": 0.9014239716079142
|
| 1556 |
+
},
|
| 1557 |
+
"eval_recall": {
|
| 1558 |
+
"recall": 0.8936214051905541
|
| 1559 |
+
},
|
| 1560 |
+
"eval_runtime": 10.1735,
|
| 1561 |
+
"eval_samples_per_second": 50.13,
|
| 1562 |
+
"eval_steps_per_second": 25.065,
|
| 1563 |
+
"step": 2295
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 9.02,
|
| 1567 |
+
"learning_rate": 3.8852578068264345e-05,
|
| 1568 |
+
"loss": 0.1807,
|
| 1569 |
+
"step": 2300
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 9.06,
|
| 1573 |
+
"learning_rate": 3.877995642701525e-05,
|
| 1574 |
+
"loss": 0.1127,
|
| 1575 |
+
"step": 2310
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 9.1,
|
| 1579 |
+
"learning_rate": 3.870733478576616e-05,
|
| 1580 |
+
"loss": 0.0503,
|
| 1581 |
+
"step": 2320
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 9.14,
|
| 1585 |
+
"learning_rate": 3.863471314451707e-05,
|
| 1586 |
+
"loss": 0.0748,
|
| 1587 |
+
"step": 2330
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 9.18,
|
| 1591 |
+
"learning_rate": 3.8562091503267977e-05,
|
| 1592 |
+
"loss": 0.2657,
|
| 1593 |
+
"step": 2340
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 9.22,
|
| 1597 |
+
"learning_rate": 3.8489469862018884e-05,
|
| 1598 |
+
"loss": 0.1785,
|
| 1599 |
+
"step": 2350
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 9.25,
|
| 1603 |
+
"learning_rate": 3.841684822076979e-05,
|
| 1604 |
+
"loss": 0.0808,
|
| 1605 |
+
"step": 2360
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 9.29,
|
| 1609 |
+
"learning_rate": 3.83442265795207e-05,
|
| 1610 |
+
"loss": 0.3348,
|
| 1611 |
+
"step": 2370
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 9.33,
|
| 1615 |
+
"learning_rate": 3.82716049382716e-05,
|
| 1616 |
+
"loss": 0.259,
|
| 1617 |
+
"step": 2380
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 9.37,
|
| 1621 |
+
"learning_rate": 3.8198983297022516e-05,
|
| 1622 |
+
"loss": 0.1707,
|
| 1623 |
+
"step": 2390
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 9.41,
|
| 1627 |
+
"learning_rate": 3.8126361655773424e-05,
|
| 1628 |
+
"loss": 0.0915,
|
| 1629 |
+
"step": 2400
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 9.45,
|
| 1633 |
+
"learning_rate": 3.805374001452433e-05,
|
| 1634 |
+
"loss": 0.0987,
|
| 1635 |
+
"step": 2410
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 9.49,
|
| 1639 |
+
"learning_rate": 3.798111837327524e-05,
|
| 1640 |
+
"loss": 0.0906,
|
| 1641 |
+
"step": 2420
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 9.53,
|
| 1645 |
+
"learning_rate": 3.790849673202614e-05,
|
| 1646 |
+
"loss": 0.1955,
|
| 1647 |
+
"step": 2430
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 9.57,
|
| 1651 |
+
"learning_rate": 3.783587509077705e-05,
|
| 1652 |
+
"loss": 0.1936,
|
| 1653 |
+
"step": 2440
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 9.61,
|
| 1657 |
+
"learning_rate": 3.7763253449527964e-05,
|
| 1658 |
+
"loss": 0.1479,
|
| 1659 |
+
"step": 2450
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 9.65,
|
| 1663 |
+
"learning_rate": 3.769063180827887e-05,
|
| 1664 |
+
"loss": 0.1662,
|
| 1665 |
+
"step": 2460
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 9.69,
|
| 1669 |
+
"learning_rate": 3.761801016702978e-05,
|
| 1670 |
+
"loss": 0.0815,
|
| 1671 |
+
"step": 2470
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 9.73,
|
| 1675 |
+
"learning_rate": 3.754538852578068e-05,
|
| 1676 |
+
"loss": 0.1128,
|
| 1677 |
+
"step": 2480
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 9.76,
|
| 1681 |
+
"learning_rate": 3.747276688453159e-05,
|
| 1682 |
+
"loss": 0.0891,
|
| 1683 |
+
"step": 2490
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 9.8,
|
| 1687 |
+
"learning_rate": 3.74001452432825e-05,
|
| 1688 |
+
"loss": 0.0853,
|
| 1689 |
+
"step": 2500
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 9.84,
|
| 1693 |
+
"learning_rate": 3.732752360203341e-05,
|
| 1694 |
+
"loss": 0.2403,
|
| 1695 |
+
"step": 2510
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 9.88,
|
| 1699 |
+
"learning_rate": 3.725490196078432e-05,
|
| 1700 |
+
"loss": 0.2393,
|
| 1701 |
+
"step": 2520
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 9.92,
|
| 1705 |
+
"learning_rate": 3.718228031953522e-05,
|
| 1706 |
+
"loss": 0.113,
|
| 1707 |
+
"step": 2530
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 9.96,
|
| 1711 |
+
"learning_rate": 3.710965867828613e-05,
|
| 1712 |
+
"loss": 0.1374,
|
| 1713 |
+
"step": 2540
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 10.0,
|
| 1717 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 1718 |
+
"loss": 0.2592,
|
| 1719 |
+
"step": 2550
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 10.0,
|
| 1723 |
+
"eval_accuracy": {
|
| 1724 |
+
"accuracy": 0.8627450980392157
|
| 1725 |
+
},
|
| 1726 |
+
"eval_f1": {
|
| 1727 |
+
"f1": 0.8639402375974159
|
| 1728 |
+
},
|
| 1729 |
+
"eval_loss": 0.49467796087265015,
|
| 1730 |
+
"eval_precision": {
|
| 1731 |
+
"precision": 0.8837020302136581
|
| 1732 |
+
},
|
| 1733 |
+
"eval_recall": {
|
| 1734 |
+
"recall": 0.8620484568622867
|
| 1735 |
+
},
|
| 1736 |
+
"eval_runtime": 10.09,
|
| 1737 |
+
"eval_samples_per_second": 50.545,
|
| 1738 |
+
"eval_steps_per_second": 25.273,
|
| 1739 |
+
"step": 2550
|
| 1740 |
+
}
|
| 1741 |
+
],
|
| 1742 |
+
"logging_steps": 10,
|
| 1743 |
+
"max_steps": 7650,
|
| 1744 |
+
"num_input_tokens_seen": 0,
|
| 1745 |
+
"num_train_epochs": 30,
|
| 1746 |
+
"save_steps": 500,
|
| 1747 |
+
"total_flos": 1.5808790937526272e+18,
|
| 1748 |
+
"train_batch_size": 2,
|
| 1749 |
+
"trial_name": null,
|
| 1750 |
+
"trial_params": null
|
| 1751 |
+
}
|
tmp-checkpoint-2550/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-2805/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-2805/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd0bdc5483d786e2b3f7c85777a9c669be94de7087b07a6141776eb245d18eb9
|
| 3 |
+
size 343233204
|
tmp-checkpoint-2805/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d96bf8b85f46c40763b7cb5c6ad6b01d8c384b2eb116a0dfcff6a89a7709bdec
|
| 3 |
+
size 686587322
|
tmp-checkpoint-2805/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-2805/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aba3144b727566b7ec2cbf79929760faa2bb06960b408a20f856e9be2d6e9236
|
| 3 |
+
size 14244
|
tmp-checkpoint-2805/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cdb4ad3bd86c02310f9ed1408b6f87f39e03f51a5a0e8b7a08019883769769ff
|
| 3 |
+
size 1064
|
tmp-checkpoint-2805/trainer_state.json
ADDED
|
@@ -0,0 +1,1921 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.36420056223869324,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-1785",
|
| 4 |
+
"epoch": 11.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2805,
|
| 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 |
+
"epoch": 3.02,
|
| 529 |
+
"learning_rate": 4.9963689179375456e-05,
|
| 530 |
+
"loss": 0.3347,
|
| 531 |
+
"step": 770
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 3.06,
|
| 535 |
+
"learning_rate": 4.9891067538126364e-05,
|
| 536 |
+
"loss": 0.3339,
|
| 537 |
+
"step": 780
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 3.1,
|
| 541 |
+
"learning_rate": 4.981844589687727e-05,
|
| 542 |
+
"loss": 0.5573,
|
| 543 |
+
"step": 790
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 3.14,
|
| 547 |
+
"learning_rate": 4.974582425562818e-05,
|
| 548 |
+
"loss": 0.5463,
|
| 549 |
+
"step": 800
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.18,
|
| 553 |
+
"learning_rate": 4.967320261437909e-05,
|
| 554 |
+
"loss": 0.3453,
|
| 555 |
+
"step": 810
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.22,
|
| 559 |
+
"learning_rate": 4.9600580973129996e-05,
|
| 560 |
+
"loss": 0.3814,
|
| 561 |
+
"step": 820
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.25,
|
| 565 |
+
"learning_rate": 4.9527959331880904e-05,
|
| 566 |
+
"loss": 0.4434,
|
| 567 |
+
"step": 830
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 3.29,
|
| 571 |
+
"learning_rate": 4.945533769063181e-05,
|
| 572 |
+
"loss": 0.3675,
|
| 573 |
+
"step": 840
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 3.33,
|
| 577 |
+
"learning_rate": 4.938271604938271e-05,
|
| 578 |
+
"loss": 0.5386,
|
| 579 |
+
"step": 850
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 3.37,
|
| 583 |
+
"learning_rate": 4.931009440813363e-05,
|
| 584 |
+
"loss": 0.2692,
|
| 585 |
+
"step": 860
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.41,
|
| 589 |
+
"learning_rate": 4.9237472766884536e-05,
|
| 590 |
+
"loss": 0.4815,
|
| 591 |
+
"step": 870
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.45,
|
| 595 |
+
"learning_rate": 4.9164851125635444e-05,
|
| 596 |
+
"loss": 0.4178,
|
| 597 |
+
"step": 880
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.49,
|
| 601 |
+
"learning_rate": 4.909222948438635e-05,
|
| 602 |
+
"loss": 0.3618,
|
| 603 |
+
"step": 890
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.53,
|
| 607 |
+
"learning_rate": 4.901960784313725e-05,
|
| 608 |
+
"loss": 0.6841,
|
| 609 |
+
"step": 900
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 3.57,
|
| 613 |
+
"learning_rate": 4.894698620188817e-05,
|
| 614 |
+
"loss": 0.471,
|
| 615 |
+
"step": 910
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.61,
|
| 619 |
+
"learning_rate": 4.8874364560639076e-05,
|
| 620 |
+
"loss": 0.313,
|
| 621 |
+
"step": 920
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 3.65,
|
| 625 |
+
"learning_rate": 4.8801742919389983e-05,
|
| 626 |
+
"loss": 0.2437,
|
| 627 |
+
"step": 930
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 3.69,
|
| 631 |
+
"learning_rate": 4.8729121278140885e-05,
|
| 632 |
+
"loss": 0.4748,
|
| 633 |
+
"step": 940
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.73,
|
| 637 |
+
"learning_rate": 4.865649963689179e-05,
|
| 638 |
+
"loss": 0.5519,
|
| 639 |
+
"step": 950
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.76,
|
| 643 |
+
"learning_rate": 4.85838779956427e-05,
|
| 644 |
+
"loss": 0.3051,
|
| 645 |
+
"step": 960
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.8,
|
| 649 |
+
"learning_rate": 4.8511256354393615e-05,
|
| 650 |
+
"loss": 0.3847,
|
| 651 |
+
"step": 970
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 3.84,
|
| 655 |
+
"learning_rate": 4.843863471314452e-05,
|
| 656 |
+
"loss": 0.2454,
|
| 657 |
+
"step": 980
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.88,
|
| 661 |
+
"learning_rate": 4.8366013071895424e-05,
|
| 662 |
+
"loss": 0.379,
|
| 663 |
+
"step": 990
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 3.92,
|
| 667 |
+
"learning_rate": 4.829339143064633e-05,
|
| 668 |
+
"loss": 0.4485,
|
| 669 |
+
"step": 1000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 3.96,
|
| 673 |
+
"learning_rate": 4.822076978939724e-05,
|
| 674 |
+
"loss": 0.2484,
|
| 675 |
+
"step": 1010
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.0,
|
| 679 |
+
"learning_rate": 4.814814814814815e-05,
|
| 680 |
+
"loss": 0.4402,
|
| 681 |
+
"step": 1020
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 4.0,
|
| 685 |
+
"eval_accuracy": {
|
| 686 |
+
"accuracy": 0.8784313725490196
|
| 687 |
+
},
|
| 688 |
+
"eval_f1": {
|
| 689 |
+
"f1": 0.880471094803992
|
| 690 |
+
},
|
| 691 |
+
"eval_loss": 0.3946979343891144,
|
| 692 |
+
"eval_precision": {
|
| 693 |
+
"precision": 0.8952890487430221
|
| 694 |
+
},
|
| 695 |
+
"eval_recall": {
|
| 696 |
+
"recall": 0.8776488582339645
|
| 697 |
+
},
|
| 698 |
+
"eval_runtime": 10.1771,
|
| 699 |
+
"eval_samples_per_second": 50.113,
|
| 700 |
+
"eval_steps_per_second": 25.056,
|
| 701 |
+
"step": 1020
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 4.04,
|
| 705 |
+
"learning_rate": 4.807552650689906e-05,
|
| 706 |
+
"loss": 0.3494,
|
| 707 |
+
"step": 1030
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 4.08,
|
| 711 |
+
"learning_rate": 4.8002904865649964e-05,
|
| 712 |
+
"loss": 0.3327,
|
| 713 |
+
"step": 1040
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 4.12,
|
| 717 |
+
"learning_rate": 4.793028322440087e-05,
|
| 718 |
+
"loss": 0.2898,
|
| 719 |
+
"step": 1050
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 4.16,
|
| 723 |
+
"learning_rate": 4.785766158315178e-05,
|
| 724 |
+
"loss": 0.3965,
|
| 725 |
+
"step": 1060
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 4.2,
|
| 729 |
+
"learning_rate": 4.778503994190269e-05,
|
| 730 |
+
"loss": 0.3607,
|
| 731 |
+
"step": 1070
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.24,
|
| 735 |
+
"learning_rate": 4.77124183006536e-05,
|
| 736 |
+
"loss": 0.3714,
|
| 737 |
+
"step": 1080
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.27,
|
| 741 |
+
"learning_rate": 4.7639796659404504e-05,
|
| 742 |
+
"loss": 0.1796,
|
| 743 |
+
"step": 1090
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 4.31,
|
| 747 |
+
"learning_rate": 4.756717501815541e-05,
|
| 748 |
+
"loss": 0.1874,
|
| 749 |
+
"step": 1100
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 4.35,
|
| 753 |
+
"learning_rate": 4.749455337690632e-05,
|
| 754 |
+
"loss": 0.3462,
|
| 755 |
+
"step": 1110
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 4.39,
|
| 759 |
+
"learning_rate": 4.742193173565723e-05,
|
| 760 |
+
"loss": 0.2082,
|
| 761 |
+
"step": 1120
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 4.43,
|
| 765 |
+
"learning_rate": 4.7349310094408136e-05,
|
| 766 |
+
"loss": 0.4531,
|
| 767 |
+
"step": 1130
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 4.47,
|
| 771 |
+
"learning_rate": 4.7276688453159044e-05,
|
| 772 |
+
"loss": 0.3951,
|
| 773 |
+
"step": 1140
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.51,
|
| 777 |
+
"learning_rate": 4.720406681190995e-05,
|
| 778 |
+
"loss": 0.2337,
|
| 779 |
+
"step": 1150
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.55,
|
| 783 |
+
"learning_rate": 4.713144517066086e-05,
|
| 784 |
+
"loss": 0.2118,
|
| 785 |
+
"step": 1160
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 4.59,
|
| 789 |
+
"learning_rate": 4.705882352941177e-05,
|
| 790 |
+
"loss": 0.373,
|
| 791 |
+
"step": 1170
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 4.63,
|
| 795 |
+
"learning_rate": 4.6986201888162675e-05,
|
| 796 |
+
"loss": 0.1484,
|
| 797 |
+
"step": 1180
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 4.67,
|
| 801 |
+
"learning_rate": 4.691358024691358e-05,
|
| 802 |
+
"loss": 0.2568,
|
| 803 |
+
"step": 1190
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 4.71,
|
| 807 |
+
"learning_rate": 4.684095860566449e-05,
|
| 808 |
+
"loss": 0.3067,
|
| 809 |
+
"step": 1200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 4.75,
|
| 813 |
+
"learning_rate": 4.67683369644154e-05,
|
| 814 |
+
"loss": 0.2007,
|
| 815 |
+
"step": 1210
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.78,
|
| 819 |
+
"learning_rate": 4.669571532316631e-05,
|
| 820 |
+
"loss": 0.2493,
|
| 821 |
+
"step": 1220
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.82,
|
| 825 |
+
"learning_rate": 4.6623093681917215e-05,
|
| 826 |
+
"loss": 0.2717,
|
| 827 |
+
"step": 1230
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 4.86,
|
| 831 |
+
"learning_rate": 4.6550472040668116e-05,
|
| 832 |
+
"loss": 0.4134,
|
| 833 |
+
"step": 1240
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 4.9,
|
| 837 |
+
"learning_rate": 4.647785039941903e-05,
|
| 838 |
+
"loss": 0.2417,
|
| 839 |
+
"step": 1250
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 4.94,
|
| 843 |
+
"learning_rate": 4.640522875816994e-05,
|
| 844 |
+
"loss": 0.2573,
|
| 845 |
+
"step": 1260
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 4.98,
|
| 849 |
+
"learning_rate": 4.633260711692085e-05,
|
| 850 |
+
"loss": 0.3937,
|
| 851 |
+
"step": 1270
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 5.0,
|
| 855 |
+
"eval_accuracy": {
|
| 856 |
+
"accuracy": 0.8725490196078431
|
| 857 |
+
},
|
| 858 |
+
"eval_f1": {
|
| 859 |
+
"f1": 0.8703780158419334
|
| 860 |
+
},
|
| 861 |
+
"eval_loss": 0.4377364218235016,
|
| 862 |
+
"eval_precision": {
|
| 863 |
+
"precision": 0.8900760323128502
|
| 864 |
+
},
|
| 865 |
+
"eval_recall": {
|
| 866 |
+
"recall": 0.8691138648585458
|
| 867 |
+
},
|
| 868 |
+
"eval_runtime": 10.1762,
|
| 869 |
+
"eval_samples_per_second": 50.117,
|
| 870 |
+
"eval_steps_per_second": 25.058,
|
| 871 |
+
"step": 1275
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 5.02,
|
| 875 |
+
"learning_rate": 4.6259985475671755e-05,
|
| 876 |
+
"loss": 0.165,
|
| 877 |
+
"step": 1280
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 5.06,
|
| 881 |
+
"learning_rate": 4.6187363834422656e-05,
|
| 882 |
+
"loss": 0.1452,
|
| 883 |
+
"step": 1290
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 5.1,
|
| 887 |
+
"learning_rate": 4.6114742193173564e-05,
|
| 888 |
+
"loss": 0.1548,
|
| 889 |
+
"step": 1300
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 5.14,
|
| 893 |
+
"learning_rate": 4.604212055192448e-05,
|
| 894 |
+
"loss": 0.4443,
|
| 895 |
+
"step": 1310
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 5.18,
|
| 899 |
+
"learning_rate": 4.5969498910675387e-05,
|
| 900 |
+
"loss": 0.4322,
|
| 901 |
+
"step": 1320
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 5.22,
|
| 905 |
+
"learning_rate": 4.5896877269426294e-05,
|
| 906 |
+
"loss": 0.2355,
|
| 907 |
+
"step": 1330
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 5.25,
|
| 911 |
+
"learning_rate": 4.5824255628177196e-05,
|
| 912 |
+
"loss": 0.1949,
|
| 913 |
+
"step": 1340
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 5.29,
|
| 917 |
+
"learning_rate": 4.5751633986928104e-05,
|
| 918 |
+
"loss": 0.1959,
|
| 919 |
+
"step": 1350
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.33,
|
| 923 |
+
"learning_rate": 4.567901234567901e-05,
|
| 924 |
+
"loss": 0.3841,
|
| 925 |
+
"step": 1360
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 5.37,
|
| 929 |
+
"learning_rate": 4.5606390704429926e-05,
|
| 930 |
+
"loss": 0.1401,
|
| 931 |
+
"step": 1370
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 5.41,
|
| 935 |
+
"learning_rate": 4.5533769063180834e-05,
|
| 936 |
+
"loss": 0.1148,
|
| 937 |
+
"step": 1380
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 5.45,
|
| 941 |
+
"learning_rate": 4.5461147421931735e-05,
|
| 942 |
+
"loss": 0.234,
|
| 943 |
+
"step": 1390
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 5.49,
|
| 947 |
+
"learning_rate": 4.538852578068264e-05,
|
| 948 |
+
"loss": 0.3899,
|
| 949 |
+
"step": 1400
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 5.53,
|
| 953 |
+
"learning_rate": 4.531590413943355e-05,
|
| 954 |
+
"loss": 0.4873,
|
| 955 |
+
"step": 1410
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 5.57,
|
| 959 |
+
"learning_rate": 4.5243282498184466e-05,
|
| 960 |
+
"loss": 0.2483,
|
| 961 |
+
"step": 1420
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.61,
|
| 965 |
+
"learning_rate": 4.5170660856935374e-05,
|
| 966 |
+
"loss": 0.2654,
|
| 967 |
+
"step": 1430
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 5.65,
|
| 971 |
+
"learning_rate": 4.5098039215686275e-05,
|
| 972 |
+
"loss": 0.2035,
|
| 973 |
+
"step": 1440
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 5.69,
|
| 977 |
+
"learning_rate": 4.502541757443718e-05,
|
| 978 |
+
"loss": 0.3744,
|
| 979 |
+
"step": 1450
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 5.73,
|
| 983 |
+
"learning_rate": 4.495279593318809e-05,
|
| 984 |
+
"loss": 0.2139,
|
| 985 |
+
"step": 1460
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 5.76,
|
| 989 |
+
"learning_rate": 4.4880174291939e-05,
|
| 990 |
+
"loss": 0.2212,
|
| 991 |
+
"step": 1470
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 5.8,
|
| 995 |
+
"learning_rate": 4.480755265068991e-05,
|
| 996 |
+
"loss": 0.376,
|
| 997 |
+
"step": 1480
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 5.84,
|
| 1001 |
+
"learning_rate": 4.4734931009440815e-05,
|
| 1002 |
+
"loss": 0.2399,
|
| 1003 |
+
"step": 1490
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.88,
|
| 1007 |
+
"learning_rate": 4.466230936819172e-05,
|
| 1008 |
+
"loss": 0.3534,
|
| 1009 |
+
"step": 1500
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 5.92,
|
| 1013 |
+
"learning_rate": 4.458968772694263e-05,
|
| 1014 |
+
"loss": 0.3667,
|
| 1015 |
+
"step": 1510
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 5.96,
|
| 1019 |
+
"learning_rate": 4.451706608569354e-05,
|
| 1020 |
+
"loss": 0.496,
|
| 1021 |
+
"step": 1520
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 6.0,
|
| 1025 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 1026 |
+
"loss": 0.4042,
|
| 1027 |
+
"step": 1530
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 6.0,
|
| 1031 |
+
"eval_accuracy": {
|
| 1032 |
+
"accuracy": 0.8862745098039215
|
| 1033 |
+
},
|
| 1034 |
+
"eval_f1": {
|
| 1035 |
+
"f1": 0.8859196308355288
|
| 1036 |
+
},
|
| 1037 |
+
"eval_loss": 0.39891302585601807,
|
| 1038 |
+
"eval_precision": {
|
| 1039 |
+
"precision": 0.8972093144987638
|
| 1040 |
+
},
|
| 1041 |
+
"eval_recall": {
|
| 1042 |
+
"recall": 0.8834955381497934
|
| 1043 |
+
},
|
| 1044 |
+
"eval_runtime": 10.1174,
|
| 1045 |
+
"eval_samples_per_second": 50.408,
|
| 1046 |
+
"eval_steps_per_second": 25.204,
|
| 1047 |
+
"step": 1530
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 6.04,
|
| 1051 |
+
"learning_rate": 4.4371822803195355e-05,
|
| 1052 |
+
"loss": 0.2637,
|
| 1053 |
+
"step": 1540
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 6.08,
|
| 1057 |
+
"learning_rate": 4.429920116194626e-05,
|
| 1058 |
+
"loss": 0.2821,
|
| 1059 |
+
"step": 1550
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 6.12,
|
| 1063 |
+
"learning_rate": 4.422657952069717e-05,
|
| 1064 |
+
"loss": 0.1201,
|
| 1065 |
+
"step": 1560
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 6.16,
|
| 1069 |
+
"learning_rate": 4.415395787944808e-05,
|
| 1070 |
+
"loss": 0.262,
|
| 1071 |
+
"step": 1570
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 6.2,
|
| 1075 |
+
"learning_rate": 4.4081336238198986e-05,
|
| 1076 |
+
"loss": 0.2064,
|
| 1077 |
+
"step": 1580
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 6.24,
|
| 1081 |
+
"learning_rate": 4.400871459694989e-05,
|
| 1082 |
+
"loss": 0.1815,
|
| 1083 |
+
"step": 1590
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 6.27,
|
| 1087 |
+
"learning_rate": 4.39360929557008e-05,
|
| 1088 |
+
"loss": 0.1959,
|
| 1089 |
+
"step": 1600
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 6.31,
|
| 1093 |
+
"learning_rate": 4.386347131445171e-05,
|
| 1094 |
+
"loss": 0.1959,
|
| 1095 |
+
"step": 1610
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.35,
|
| 1099 |
+
"learning_rate": 4.379084967320262e-05,
|
| 1100 |
+
"loss": 0.2398,
|
| 1101 |
+
"step": 1620
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.39,
|
| 1105 |
+
"learning_rate": 4.3718228031953526e-05,
|
| 1106 |
+
"loss": 0.2038,
|
| 1107 |
+
"step": 1630
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 6.43,
|
| 1111 |
+
"learning_rate": 4.364560639070443e-05,
|
| 1112 |
+
"loss": 0.2004,
|
| 1113 |
+
"step": 1640
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.47,
|
| 1117 |
+
"learning_rate": 4.357298474945534e-05,
|
| 1118 |
+
"loss": 0.0377,
|
| 1119 |
+
"step": 1650
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 6.51,
|
| 1123 |
+
"learning_rate": 4.350036310820625e-05,
|
| 1124 |
+
"loss": 0.2022,
|
| 1125 |
+
"step": 1660
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 6.55,
|
| 1129 |
+
"learning_rate": 4.342774146695716e-05,
|
| 1130 |
+
"loss": 0.4403,
|
| 1131 |
+
"step": 1670
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 6.59,
|
| 1135 |
+
"learning_rate": 4.3355119825708066e-05,
|
| 1136 |
+
"loss": 0.2577,
|
| 1137 |
+
"step": 1680
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.63,
|
| 1141 |
+
"learning_rate": 4.328249818445897e-05,
|
| 1142 |
+
"loss": 0.3312,
|
| 1143 |
+
"step": 1690
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.67,
|
| 1147 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 1148 |
+
"loss": 0.307,
|
| 1149 |
+
"step": 1700
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 6.71,
|
| 1153 |
+
"learning_rate": 4.313725490196079e-05,
|
| 1154 |
+
"loss": 0.0961,
|
| 1155 |
+
"step": 1710
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.75,
|
| 1159 |
+
"learning_rate": 4.30646332607117e-05,
|
| 1160 |
+
"loss": 0.3266,
|
| 1161 |
+
"step": 1720
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 6.78,
|
| 1165 |
+
"learning_rate": 4.2992011619462605e-05,
|
| 1166 |
+
"loss": 0.2601,
|
| 1167 |
+
"step": 1730
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 6.82,
|
| 1171 |
+
"learning_rate": 4.291938997821351e-05,
|
| 1172 |
+
"loss": 0.2382,
|
| 1173 |
+
"step": 1740
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 6.86,
|
| 1177 |
+
"learning_rate": 4.2846768336964415e-05,
|
| 1178 |
+
"loss": 0.1469,
|
| 1179 |
+
"step": 1750
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.9,
|
| 1183 |
+
"learning_rate": 4.277414669571532e-05,
|
| 1184 |
+
"loss": 0.2158,
|
| 1185 |
+
"step": 1760
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.94,
|
| 1189 |
+
"learning_rate": 4.270152505446624e-05,
|
| 1190 |
+
"loss": 0.4392,
|
| 1191 |
+
"step": 1770
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 6.98,
|
| 1195 |
+
"learning_rate": 4.262890341321714e-05,
|
| 1196 |
+
"loss": 0.4321,
|
| 1197 |
+
"step": 1780
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 7.0,
|
| 1201 |
+
"eval_accuracy": {
|
| 1202 |
+
"accuracy": 0.8901960784313725
|
| 1203 |
+
},
|
| 1204 |
+
"eval_f1": {
|
| 1205 |
+
"f1": 0.8896787277785286
|
| 1206 |
+
},
|
| 1207 |
+
"eval_loss": 0.36420056223869324,
|
| 1208 |
+
"eval_precision": {
|
| 1209 |
+
"precision": 0.9002265766283616
|
| 1210 |
+
},
|
| 1211 |
+
"eval_recall": {
|
| 1212 |
+
"recall": 0.8878911425453978
|
| 1213 |
+
},
|
| 1214 |
+
"eval_runtime": 10.2347,
|
| 1215 |
+
"eval_samples_per_second": 49.83,
|
| 1216 |
+
"eval_steps_per_second": 24.915,
|
| 1217 |
+
"step": 1785
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 7.02,
|
| 1221 |
+
"learning_rate": 4.2556281771968046e-05,
|
| 1222 |
+
"loss": 0.5162,
|
| 1223 |
+
"step": 1790
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 7.06,
|
| 1227 |
+
"learning_rate": 4.2483660130718954e-05,
|
| 1228 |
+
"loss": 0.2491,
|
| 1229 |
+
"step": 1800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 7.1,
|
| 1233 |
+
"learning_rate": 4.241103848946986e-05,
|
| 1234 |
+
"loss": 0.2267,
|
| 1235 |
+
"step": 1810
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 7.14,
|
| 1239 |
+
"learning_rate": 4.233841684822078e-05,
|
| 1240 |
+
"loss": 0.2395,
|
| 1241 |
+
"step": 1820
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 7.18,
|
| 1245 |
+
"learning_rate": 4.226579520697168e-05,
|
| 1246 |
+
"loss": 0.3007,
|
| 1247 |
+
"step": 1830
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 7.22,
|
| 1251 |
+
"learning_rate": 4.2193173565722586e-05,
|
| 1252 |
+
"loss": 0.1316,
|
| 1253 |
+
"step": 1840
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 7.25,
|
| 1257 |
+
"learning_rate": 4.2120551924473494e-05,
|
| 1258 |
+
"loss": 0.1592,
|
| 1259 |
+
"step": 1850
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 7.29,
|
| 1263 |
+
"learning_rate": 4.20479302832244e-05,
|
| 1264 |
+
"loss": 0.3467,
|
| 1265 |
+
"step": 1860
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 7.33,
|
| 1269 |
+
"learning_rate": 4.197530864197531e-05,
|
| 1270 |
+
"loss": 0.4195,
|
| 1271 |
+
"step": 1870
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 7.37,
|
| 1275 |
+
"learning_rate": 4.190268700072622e-05,
|
| 1276 |
+
"loss": 0.3561,
|
| 1277 |
+
"step": 1880
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 7.41,
|
| 1281 |
+
"learning_rate": 4.1830065359477126e-05,
|
| 1282 |
+
"loss": 0.1572,
|
| 1283 |
+
"step": 1890
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.45,
|
| 1287 |
+
"learning_rate": 4.1757443718228034e-05,
|
| 1288 |
+
"loss": 0.2104,
|
| 1289 |
+
"step": 1900
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 7.49,
|
| 1293 |
+
"learning_rate": 4.168482207697894e-05,
|
| 1294 |
+
"loss": 0.3536,
|
| 1295 |
+
"step": 1910
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 7.53,
|
| 1299 |
+
"learning_rate": 4.161220043572985e-05,
|
| 1300 |
+
"loss": 0.345,
|
| 1301 |
+
"step": 1920
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 7.57,
|
| 1305 |
+
"learning_rate": 4.153957879448076e-05,
|
| 1306 |
+
"loss": 0.1083,
|
| 1307 |
+
"step": 1930
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 7.61,
|
| 1311 |
+
"learning_rate": 4.1466957153231666e-05,
|
| 1312 |
+
"loss": 0.1895,
|
| 1313 |
+
"step": 1940
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 7.65,
|
| 1317 |
+
"learning_rate": 4.1394335511982573e-05,
|
| 1318 |
+
"loss": 0.3232,
|
| 1319 |
+
"step": 1950
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 7.69,
|
| 1323 |
+
"learning_rate": 4.132171387073348e-05,
|
| 1324 |
+
"loss": 0.2939,
|
| 1325 |
+
"step": 1960
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.73,
|
| 1329 |
+
"learning_rate": 4.124909222948439e-05,
|
| 1330 |
+
"loss": 0.0918,
|
| 1331 |
+
"step": 1970
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 7.76,
|
| 1335 |
+
"learning_rate": 4.11764705882353e-05,
|
| 1336 |
+
"loss": 0.1722,
|
| 1337 |
+
"step": 1980
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 7.8,
|
| 1341 |
+
"learning_rate": 4.1103848946986205e-05,
|
| 1342 |
+
"loss": 0.3229,
|
| 1343 |
+
"step": 1990
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 7.84,
|
| 1347 |
+
"learning_rate": 4.103122730573711e-05,
|
| 1348 |
+
"loss": 0.1501,
|
| 1349 |
+
"step": 2000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 7.88,
|
| 1353 |
+
"learning_rate": 4.095860566448802e-05,
|
| 1354 |
+
"loss": 0.1468,
|
| 1355 |
+
"step": 2010
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 7.92,
|
| 1359 |
+
"learning_rate": 4.088598402323893e-05,
|
| 1360 |
+
"loss": 0.1665,
|
| 1361 |
+
"step": 2020
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 7.96,
|
| 1365 |
+
"learning_rate": 4.081336238198984e-05,
|
| 1366 |
+
"loss": 0.1324,
|
| 1367 |
+
"step": 2030
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 8.0,
|
| 1371 |
+
"learning_rate": 4.074074074074074e-05,
|
| 1372 |
+
"loss": 0.2618,
|
| 1373 |
+
"step": 2040
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 8.0,
|
| 1377 |
+
"eval_accuracy": {
|
| 1378 |
+
"accuracy": 0.8960784313725491
|
| 1379 |
+
},
|
| 1380 |
+
"eval_f1": {
|
| 1381 |
+
"f1": 0.8950352545633242
|
| 1382 |
+
},
|
| 1383 |
+
"eval_loss": 0.40923169255256653,
|
| 1384 |
+
"eval_precision": {
|
| 1385 |
+
"precision": 0.9033601037449228
|
| 1386 |
+
},
|
| 1387 |
+
"eval_recall": {
|
| 1388 |
+
"recall": 0.8953584093211753
|
| 1389 |
+
},
|
| 1390 |
+
"eval_runtime": 10.3235,
|
| 1391 |
+
"eval_samples_per_second": 49.402,
|
| 1392 |
+
"eval_steps_per_second": 24.701,
|
| 1393 |
+
"step": 2040
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 8.04,
|
| 1397 |
+
"learning_rate": 4.066811909949165e-05,
|
| 1398 |
+
"loss": 0.1887,
|
| 1399 |
+
"step": 2050
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 8.08,
|
| 1403 |
+
"learning_rate": 4.059549745824256e-05,
|
| 1404 |
+
"loss": 0.4153,
|
| 1405 |
+
"step": 2060
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 8.12,
|
| 1409 |
+
"learning_rate": 4.052287581699347e-05,
|
| 1410 |
+
"loss": 0.2771,
|
| 1411 |
+
"step": 2070
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 8.16,
|
| 1415 |
+
"learning_rate": 4.045025417574437e-05,
|
| 1416 |
+
"loss": 0.3316,
|
| 1417 |
+
"step": 2080
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 8.2,
|
| 1421 |
+
"learning_rate": 4.037763253449528e-05,
|
| 1422 |
+
"loss": 0.1125,
|
| 1423 |
+
"step": 2090
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 8.24,
|
| 1427 |
+
"learning_rate": 4.0305010893246186e-05,
|
| 1428 |
+
"loss": 0.3016,
|
| 1429 |
+
"step": 2100
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 8.27,
|
| 1433 |
+
"learning_rate": 4.02323892519971e-05,
|
| 1434 |
+
"loss": 0.2458,
|
| 1435 |
+
"step": 2110
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 8.31,
|
| 1439 |
+
"learning_rate": 4.015976761074801e-05,
|
| 1440 |
+
"loss": 0.179,
|
| 1441 |
+
"step": 2120
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 8.35,
|
| 1445 |
+
"learning_rate": 4.008714596949891e-05,
|
| 1446 |
+
"loss": 0.224,
|
| 1447 |
+
"step": 2130
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 8.39,
|
| 1451 |
+
"learning_rate": 4.001452432824982e-05,
|
| 1452 |
+
"loss": 0.0803,
|
| 1453 |
+
"step": 2140
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 8.43,
|
| 1457 |
+
"learning_rate": 3.9941902687000726e-05,
|
| 1458 |
+
"loss": 0.1121,
|
| 1459 |
+
"step": 2150
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 8.47,
|
| 1463 |
+
"learning_rate": 3.986928104575164e-05,
|
| 1464 |
+
"loss": 0.3751,
|
| 1465 |
+
"step": 2160
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 8.51,
|
| 1469 |
+
"learning_rate": 3.979665940450255e-05,
|
| 1470 |
+
"loss": 0.1354,
|
| 1471 |
+
"step": 2170
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 8.55,
|
| 1475 |
+
"learning_rate": 3.972403776325345e-05,
|
| 1476 |
+
"loss": 0.3105,
|
| 1477 |
+
"step": 2180
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 8.59,
|
| 1481 |
+
"learning_rate": 3.965141612200436e-05,
|
| 1482 |
+
"loss": 0.1937,
|
| 1483 |
+
"step": 2190
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 8.63,
|
| 1487 |
+
"learning_rate": 3.9578794480755265e-05,
|
| 1488 |
+
"loss": 0.3028,
|
| 1489 |
+
"step": 2200
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 8.67,
|
| 1493 |
+
"learning_rate": 3.950617283950617e-05,
|
| 1494 |
+
"loss": 0.1851,
|
| 1495 |
+
"step": 2210
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 8.71,
|
| 1499 |
+
"learning_rate": 3.943355119825709e-05,
|
| 1500 |
+
"loss": 0.074,
|
| 1501 |
+
"step": 2220
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 8.75,
|
| 1505 |
+
"learning_rate": 3.936092955700799e-05,
|
| 1506 |
+
"loss": 0.2098,
|
| 1507 |
+
"step": 2230
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 8.78,
|
| 1511 |
+
"learning_rate": 3.92883079157589e-05,
|
| 1512 |
+
"loss": 0.3167,
|
| 1513 |
+
"step": 2240
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 8.82,
|
| 1517 |
+
"learning_rate": 3.9215686274509805e-05,
|
| 1518 |
+
"loss": 0.2527,
|
| 1519 |
+
"step": 2250
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 8.86,
|
| 1523 |
+
"learning_rate": 3.914306463326071e-05,
|
| 1524 |
+
"loss": 0.1521,
|
| 1525 |
+
"step": 2260
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 8.9,
|
| 1529 |
+
"learning_rate": 3.907044299201162e-05,
|
| 1530 |
+
"loss": 0.1393,
|
| 1531 |
+
"step": 2270
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 8.94,
|
| 1535 |
+
"learning_rate": 3.899782135076253e-05,
|
| 1536 |
+
"loss": 0.1155,
|
| 1537 |
+
"step": 2280
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 8.98,
|
| 1541 |
+
"learning_rate": 3.892519970951344e-05,
|
| 1542 |
+
"loss": 0.3376,
|
| 1543 |
+
"step": 2290
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 9.0,
|
| 1547 |
+
"eval_accuracy": {
|
| 1548 |
+
"accuracy": 0.8941176470588236
|
| 1549 |
+
},
|
| 1550 |
+
"eval_f1": {
|
| 1551 |
+
"f1": 0.8933906521569781
|
| 1552 |
+
},
|
| 1553 |
+
"eval_loss": 0.4055422246456146,
|
| 1554 |
+
"eval_precision": {
|
| 1555 |
+
"precision": 0.9014239716079142
|
| 1556 |
+
},
|
| 1557 |
+
"eval_recall": {
|
| 1558 |
+
"recall": 0.8936214051905541
|
| 1559 |
+
},
|
| 1560 |
+
"eval_runtime": 10.1735,
|
| 1561 |
+
"eval_samples_per_second": 50.13,
|
| 1562 |
+
"eval_steps_per_second": 25.065,
|
| 1563 |
+
"step": 2295
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 9.02,
|
| 1567 |
+
"learning_rate": 3.8852578068264345e-05,
|
| 1568 |
+
"loss": 0.1807,
|
| 1569 |
+
"step": 2300
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 9.06,
|
| 1573 |
+
"learning_rate": 3.877995642701525e-05,
|
| 1574 |
+
"loss": 0.1127,
|
| 1575 |
+
"step": 2310
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 9.1,
|
| 1579 |
+
"learning_rate": 3.870733478576616e-05,
|
| 1580 |
+
"loss": 0.0503,
|
| 1581 |
+
"step": 2320
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 9.14,
|
| 1585 |
+
"learning_rate": 3.863471314451707e-05,
|
| 1586 |
+
"loss": 0.0748,
|
| 1587 |
+
"step": 2330
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 9.18,
|
| 1591 |
+
"learning_rate": 3.8562091503267977e-05,
|
| 1592 |
+
"loss": 0.2657,
|
| 1593 |
+
"step": 2340
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 9.22,
|
| 1597 |
+
"learning_rate": 3.8489469862018884e-05,
|
| 1598 |
+
"loss": 0.1785,
|
| 1599 |
+
"step": 2350
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 9.25,
|
| 1603 |
+
"learning_rate": 3.841684822076979e-05,
|
| 1604 |
+
"loss": 0.0808,
|
| 1605 |
+
"step": 2360
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 9.29,
|
| 1609 |
+
"learning_rate": 3.83442265795207e-05,
|
| 1610 |
+
"loss": 0.3348,
|
| 1611 |
+
"step": 2370
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 9.33,
|
| 1615 |
+
"learning_rate": 3.82716049382716e-05,
|
| 1616 |
+
"loss": 0.259,
|
| 1617 |
+
"step": 2380
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 9.37,
|
| 1621 |
+
"learning_rate": 3.8198983297022516e-05,
|
| 1622 |
+
"loss": 0.1707,
|
| 1623 |
+
"step": 2390
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 9.41,
|
| 1627 |
+
"learning_rate": 3.8126361655773424e-05,
|
| 1628 |
+
"loss": 0.0915,
|
| 1629 |
+
"step": 2400
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 9.45,
|
| 1633 |
+
"learning_rate": 3.805374001452433e-05,
|
| 1634 |
+
"loss": 0.0987,
|
| 1635 |
+
"step": 2410
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 9.49,
|
| 1639 |
+
"learning_rate": 3.798111837327524e-05,
|
| 1640 |
+
"loss": 0.0906,
|
| 1641 |
+
"step": 2420
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 9.53,
|
| 1645 |
+
"learning_rate": 3.790849673202614e-05,
|
| 1646 |
+
"loss": 0.1955,
|
| 1647 |
+
"step": 2430
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 9.57,
|
| 1651 |
+
"learning_rate": 3.783587509077705e-05,
|
| 1652 |
+
"loss": 0.1936,
|
| 1653 |
+
"step": 2440
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 9.61,
|
| 1657 |
+
"learning_rate": 3.7763253449527964e-05,
|
| 1658 |
+
"loss": 0.1479,
|
| 1659 |
+
"step": 2450
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 9.65,
|
| 1663 |
+
"learning_rate": 3.769063180827887e-05,
|
| 1664 |
+
"loss": 0.1662,
|
| 1665 |
+
"step": 2460
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 9.69,
|
| 1669 |
+
"learning_rate": 3.761801016702978e-05,
|
| 1670 |
+
"loss": 0.0815,
|
| 1671 |
+
"step": 2470
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 9.73,
|
| 1675 |
+
"learning_rate": 3.754538852578068e-05,
|
| 1676 |
+
"loss": 0.1128,
|
| 1677 |
+
"step": 2480
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 9.76,
|
| 1681 |
+
"learning_rate": 3.747276688453159e-05,
|
| 1682 |
+
"loss": 0.0891,
|
| 1683 |
+
"step": 2490
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 9.8,
|
| 1687 |
+
"learning_rate": 3.74001452432825e-05,
|
| 1688 |
+
"loss": 0.0853,
|
| 1689 |
+
"step": 2500
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 9.84,
|
| 1693 |
+
"learning_rate": 3.732752360203341e-05,
|
| 1694 |
+
"loss": 0.2403,
|
| 1695 |
+
"step": 2510
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 9.88,
|
| 1699 |
+
"learning_rate": 3.725490196078432e-05,
|
| 1700 |
+
"loss": 0.2393,
|
| 1701 |
+
"step": 2520
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 9.92,
|
| 1705 |
+
"learning_rate": 3.718228031953522e-05,
|
| 1706 |
+
"loss": 0.113,
|
| 1707 |
+
"step": 2530
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 9.96,
|
| 1711 |
+
"learning_rate": 3.710965867828613e-05,
|
| 1712 |
+
"loss": 0.1374,
|
| 1713 |
+
"step": 2540
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 10.0,
|
| 1717 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 1718 |
+
"loss": 0.2592,
|
| 1719 |
+
"step": 2550
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 10.0,
|
| 1723 |
+
"eval_accuracy": {
|
| 1724 |
+
"accuracy": 0.8627450980392157
|
| 1725 |
+
},
|
| 1726 |
+
"eval_f1": {
|
| 1727 |
+
"f1": 0.8639402375974159
|
| 1728 |
+
},
|
| 1729 |
+
"eval_loss": 0.49467796087265015,
|
| 1730 |
+
"eval_precision": {
|
| 1731 |
+
"precision": 0.8837020302136581
|
| 1732 |
+
},
|
| 1733 |
+
"eval_recall": {
|
| 1734 |
+
"recall": 0.8620484568622867
|
| 1735 |
+
},
|
| 1736 |
+
"eval_runtime": 10.09,
|
| 1737 |
+
"eval_samples_per_second": 50.545,
|
| 1738 |
+
"eval_steps_per_second": 25.273,
|
| 1739 |
+
"step": 2550
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 10.04,
|
| 1743 |
+
"learning_rate": 3.696441539578795e-05,
|
| 1744 |
+
"loss": 0.1971,
|
| 1745 |
+
"step": 2560
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 10.08,
|
| 1749 |
+
"learning_rate": 3.689179375453886e-05,
|
| 1750 |
+
"loss": 0.1467,
|
| 1751 |
+
"step": 2570
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 10.12,
|
| 1755 |
+
"learning_rate": 3.681917211328976e-05,
|
| 1756 |
+
"loss": 0.2304,
|
| 1757 |
+
"step": 2580
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 10.16,
|
| 1761 |
+
"learning_rate": 3.674655047204067e-05,
|
| 1762 |
+
"loss": 0.1111,
|
| 1763 |
+
"step": 2590
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 10.2,
|
| 1767 |
+
"learning_rate": 3.6673928830791576e-05,
|
| 1768 |
+
"loss": 0.2193,
|
| 1769 |
+
"step": 2600
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 10.24,
|
| 1773 |
+
"learning_rate": 3.6601307189542484e-05,
|
| 1774 |
+
"loss": 0.1701,
|
| 1775 |
+
"step": 2610
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 10.27,
|
| 1779 |
+
"learning_rate": 3.652868554829339e-05,
|
| 1780 |
+
"loss": 0.0251,
|
| 1781 |
+
"step": 2620
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 10.31,
|
| 1785 |
+
"learning_rate": 3.64560639070443e-05,
|
| 1786 |
+
"loss": 0.0138,
|
| 1787 |
+
"step": 2630
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 10.35,
|
| 1791 |
+
"learning_rate": 3.638344226579521e-05,
|
| 1792 |
+
"loss": 0.191,
|
| 1793 |
+
"step": 2640
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 10.39,
|
| 1797 |
+
"learning_rate": 3.6310820624546116e-05,
|
| 1798 |
+
"loss": 0.1112,
|
| 1799 |
+
"step": 2650
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 10.43,
|
| 1803 |
+
"learning_rate": 3.6238198983297024e-05,
|
| 1804 |
+
"loss": 0.1115,
|
| 1805 |
+
"step": 2660
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 10.47,
|
| 1809 |
+
"learning_rate": 3.616557734204793e-05,
|
| 1810 |
+
"loss": 0.1842,
|
| 1811 |
+
"step": 2670
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 10.51,
|
| 1815 |
+
"learning_rate": 3.609295570079884e-05,
|
| 1816 |
+
"loss": 0.1397,
|
| 1817 |
+
"step": 2680
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 10.55,
|
| 1821 |
+
"learning_rate": 3.602033405954975e-05,
|
| 1822 |
+
"loss": 0.1044,
|
| 1823 |
+
"step": 2690
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 10.59,
|
| 1827 |
+
"learning_rate": 3.5947712418300656e-05,
|
| 1828 |
+
"loss": 0.1015,
|
| 1829 |
+
"step": 2700
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 10.63,
|
| 1833 |
+
"learning_rate": 3.5875090777051564e-05,
|
| 1834 |
+
"loss": 0.2076,
|
| 1835 |
+
"step": 2710
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 10.67,
|
| 1839 |
+
"learning_rate": 3.580246913580247e-05,
|
| 1840 |
+
"loss": 0.1405,
|
| 1841 |
+
"step": 2720
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 10.71,
|
| 1845 |
+
"learning_rate": 3.572984749455338e-05,
|
| 1846 |
+
"loss": 0.154,
|
| 1847 |
+
"step": 2730
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 10.75,
|
| 1851 |
+
"learning_rate": 3.565722585330429e-05,
|
| 1852 |
+
"loss": 0.2566,
|
| 1853 |
+
"step": 2740
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 10.78,
|
| 1857 |
+
"learning_rate": 3.5584604212055195e-05,
|
| 1858 |
+
"loss": 0.1589,
|
| 1859 |
+
"step": 2750
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 10.82,
|
| 1863 |
+
"learning_rate": 3.55119825708061e-05,
|
| 1864 |
+
"loss": 0.2138,
|
| 1865 |
+
"step": 2760
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 10.86,
|
| 1869 |
+
"learning_rate": 3.543936092955701e-05,
|
| 1870 |
+
"loss": 0.1734,
|
| 1871 |
+
"step": 2770
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 10.9,
|
| 1875 |
+
"learning_rate": 3.536673928830791e-05,
|
| 1876 |
+
"loss": 0.0846,
|
| 1877 |
+
"step": 2780
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 10.94,
|
| 1881 |
+
"learning_rate": 3.529411764705883e-05,
|
| 1882 |
+
"loss": 0.2093,
|
| 1883 |
+
"step": 2790
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 10.98,
|
| 1887 |
+
"learning_rate": 3.5221496005809735e-05,
|
| 1888 |
+
"loss": 0.363,
|
| 1889 |
+
"step": 2800
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 11.0,
|
| 1893 |
+
"eval_accuracy": {
|
| 1894 |
+
"accuracy": 0.8862745098039215
|
| 1895 |
+
},
|
| 1896 |
+
"eval_f1": {
|
| 1897 |
+
"f1": 0.8871534026726227
|
| 1898 |
+
},
|
| 1899 |
+
"eval_loss": 0.45857715606689453,
|
| 1900 |
+
"eval_precision": {
|
| 1901 |
+
"precision": 0.89776885319054
|
| 1902 |
+
},
|
| 1903 |
+
"eval_recall": {
|
| 1904 |
+
"recall": 0.8850722858701582
|
| 1905 |
+
},
|
| 1906 |
+
"eval_runtime": 9.9484,
|
| 1907 |
+
"eval_samples_per_second": 51.265,
|
| 1908 |
+
"eval_steps_per_second": 25.632,
|
| 1909 |
+
"step": 2805
|
| 1910 |
+
}
|
| 1911 |
+
],
|
| 1912 |
+
"logging_steps": 10,
|
| 1913 |
+
"max_steps": 7650,
|
| 1914 |
+
"num_input_tokens_seen": 0,
|
| 1915 |
+
"num_train_epochs": 30,
|
| 1916 |
+
"save_steps": 500,
|
| 1917 |
+
"total_flos": 1.73896700312789e+18,
|
| 1918 |
+
"train_batch_size": 2,
|
| 1919 |
+
"trial_name": null,
|
| 1920 |
+
"trial_params": null
|
| 1921 |
+
}
|
tmp-checkpoint-2805/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:418b85f35316fefe4bc777808f9dff13bb59b607bf93a8504dc8b5709313507f
|
| 3 |
+
size 4728
|