mmomm25 commited on
Commit
ad3e29c
·
verified ·
1 Parent(s): 8955e55

Training in progress, epoch 3

Browse files
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:677eacd964420203b3977b93c7673ef38ffd1b1026647495e41c8bbeea224d46
3
  size 95943916
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1252ff9c13c63d9984f41594c3c093d891f057215437c77f5f8af922270a56d2
3
  size 95943916
tmp-checkpoint-5781/config.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "Xrenya/pvt-small-224",
3
+ "architectures": [
4
+ "PvtForImageClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "depths": [
8
+ 3,
9
+ 4,
10
+ 6,
11
+ 3
12
+ ],
13
+ "drop_path_rate": 0.0,
14
+ "hidden_act": "gelu",
15
+ "hidden_dropout_prob": 0.0,
16
+ "hidden_sizes": [
17
+ 64,
18
+ 128,
19
+ 320,
20
+ 512
21
+ ],
22
+ "id2label": {
23
+ "0": "Cracking",
24
+ "1": "For Review",
25
+ "2": "Joint Defect",
26
+ "3": "Non Distressed",
27
+ "4": "Patches",
28
+ "5": "Potholes",
29
+ "6": "Wearing Surface"
30
+ },
31
+ "image_size": 224,
32
+ "initializer_range": 0.02,
33
+ "label2id": {
34
+ "Cracking": 0,
35
+ "For Review": 1,
36
+ "Joint Defect": 2,
37
+ "Non Distressed": 3,
38
+ "Patches": 4,
39
+ "Potholes": 5,
40
+ "Wearing Surface": 6
41
+ },
42
+ "layer_norm_eps": 1e-06,
43
+ "mlp_ratios": [
44
+ 8,
45
+ 8,
46
+ 4,
47
+ 4
48
+ ],
49
+ "model_type": "pvt",
50
+ "num_attention_heads": [
51
+ 1,
52
+ 2,
53
+ 5,
54
+ 8
55
+ ],
56
+ "num_channels": 3,
57
+ "num_encoder_blocks": 4,
58
+ "patch_sizes": [
59
+ 4,
60
+ 2,
61
+ 2,
62
+ 2
63
+ ],
64
+ "problem_type": "single_label_classification",
65
+ "qkv_bias": true,
66
+ "reshape_last_stage": true,
67
+ "sequence_reduction_ratios": [
68
+ 8,
69
+ 4,
70
+ 2,
71
+ 1
72
+ ],
73
+ "strides": [
74
+ 4,
75
+ 2,
76
+ 2,
77
+ 2
78
+ ],
79
+ "torch_dtype": "float32",
80
+ "transformers_version": "4.37.2"
81
+ }
tmp-checkpoint-5781/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1252ff9c13c63d9984f41594c3c093d891f057215437c77f5f8af922270a56d2
3
+ size 95943916
tmp-checkpoint-5781/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3873b609a1605beaf3c0355dc73b172e205b50f87f05cf2c27ed6c3611dac943
3
+ size 192089466
tmp-checkpoint-5781/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
+ 0.485,
7
+ 0.456,
8
+ 0.406
9
+ ],
10
+ "image_processor_type": "PvtImageProcessor",
11
+ "image_std": [
12
+ 0.229,
13
+ 0.224,
14
+ 0.225
15
+ ],
16
+ "resample": 2,
17
+ "rescale_factor": 0.00392156862745098,
18
+ "size": {
19
+ "height": 224,
20
+ "width": 224
21
+ }
22
+ }
tmp-checkpoint-5781/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34717d334cc5bd8109b0bf4c290aaabf502e3582783876983a2e41d9d5850607
3
+ size 14244
tmp-checkpoint-5781/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc66456095ec06fe0d39f9ffb176822a948a3c363588440f235808926c73fe9c
3
+ size 1064
tmp-checkpoint-5781/trainer_state.json ADDED
@@ -0,0 +1,3549 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.9446961879730225,
3
+ "best_model_checkpoint": "pvt-small-224-ConcreteClassifier-PVT\\checkpoint-3854",
4
+ "epoch": 3.0,
5
+ "eval_steps": 500,
6
+ "global_step": 5781,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 0.001,
14
+ "loss": 2.7631,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 0.001,
20
+ "loss": 2.0945,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.02,
25
+ "learning_rate": 0.001,
26
+ "loss": 1.9938,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.02,
31
+ "learning_rate": 0.001,
32
+ "loss": 1.9829,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.03,
37
+ "learning_rate": 0.001,
38
+ "loss": 2.0147,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.03,
43
+ "learning_rate": 0.001,
44
+ "loss": 2.096,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.04,
49
+ "learning_rate": 0.001,
50
+ "loss": 2.006,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.04,
55
+ "learning_rate": 0.001,
56
+ "loss": 1.9784,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.05,
61
+ "learning_rate": 0.001,
62
+ "loss": 1.9937,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.05,
67
+ "learning_rate": 0.001,
68
+ "loss": 1.9689,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.06,
73
+ "learning_rate": 0.001,
74
+ "loss": 2.0182,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 0.06,
79
+ "learning_rate": 0.001,
80
+ "loss": 1.9828,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 0.07,
85
+ "learning_rate": 0.001,
86
+ "loss": 1.9499,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 0.07,
91
+ "learning_rate": 0.001,
92
+ "loss": 1.9696,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 0.08,
97
+ "learning_rate": 0.001,
98
+ "loss": 1.9334,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 0.08,
103
+ "learning_rate": 0.001,
104
+ "loss": 1.9834,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 0.09,
109
+ "learning_rate": 0.001,
110
+ "loss": 1.9474,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 0.09,
115
+ "learning_rate": 0.001,
116
+ "loss": 1.9558,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 0.1,
121
+ "learning_rate": 0.001,
122
+ "loss": 1.9509,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 0.1,
127
+ "learning_rate": 0.001,
128
+ "loss": 1.9384,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 0.11,
133
+ "learning_rate": 0.001,
134
+ "loss": 1.9978,
135
+ "step": 210
136
+ },
137
+ {
138
+ "epoch": 0.11,
139
+ "learning_rate": 0.001,
140
+ "loss": 1.9604,
141
+ "step": 220
142
+ },
143
+ {
144
+ "epoch": 0.12,
145
+ "learning_rate": 0.001,
146
+ "loss": 1.971,
147
+ "step": 230
148
+ },
149
+ {
150
+ "epoch": 0.12,
151
+ "learning_rate": 0.001,
152
+ "loss": 1.9705,
153
+ "step": 240
154
+ },
155
+ {
156
+ "epoch": 0.13,
157
+ "learning_rate": 0.001,
158
+ "loss": 1.9544,
159
+ "step": 250
160
+ },
161
+ {
162
+ "epoch": 0.13,
163
+ "learning_rate": 0.001,
164
+ "loss": 1.9343,
165
+ "step": 260
166
+ },
167
+ {
168
+ "epoch": 0.14,
169
+ "learning_rate": 0.001,
170
+ "loss": 1.9263,
171
+ "step": 270
172
+ },
173
+ {
174
+ "epoch": 0.15,
175
+ "learning_rate": 0.001,
176
+ "loss": 1.9695,
177
+ "step": 280
178
+ },
179
+ {
180
+ "epoch": 0.15,
181
+ "learning_rate": 0.001,
182
+ "loss": 1.9348,
183
+ "step": 290
184
+ },
185
+ {
186
+ "epoch": 0.16,
187
+ "learning_rate": 0.001,
188
+ "loss": 1.9506,
189
+ "step": 300
190
+ },
191
+ {
192
+ "epoch": 0.16,
193
+ "learning_rate": 0.001,
194
+ "loss": 1.9976,
195
+ "step": 310
196
+ },
197
+ {
198
+ "epoch": 0.17,
199
+ "learning_rate": 0.001,
200
+ "loss": 1.9527,
201
+ "step": 320
202
+ },
203
+ {
204
+ "epoch": 0.17,
205
+ "learning_rate": 0.001,
206
+ "loss": 1.9526,
207
+ "step": 330
208
+ },
209
+ {
210
+ "epoch": 0.18,
211
+ "learning_rate": 0.001,
212
+ "loss": 1.9623,
213
+ "step": 340
214
+ },
215
+ {
216
+ "epoch": 0.18,
217
+ "learning_rate": 0.001,
218
+ "loss": 1.9455,
219
+ "step": 350
220
+ },
221
+ {
222
+ "epoch": 0.19,
223
+ "learning_rate": 0.001,
224
+ "loss": 1.9286,
225
+ "step": 360
226
+ },
227
+ {
228
+ "epoch": 0.19,
229
+ "learning_rate": 0.001,
230
+ "loss": 1.9476,
231
+ "step": 370
232
+ },
233
+ {
234
+ "epoch": 0.2,
235
+ "learning_rate": 0.001,
236
+ "loss": 1.9484,
237
+ "step": 380
238
+ },
239
+ {
240
+ "epoch": 0.2,
241
+ "learning_rate": 0.001,
242
+ "loss": 1.9564,
243
+ "step": 390
244
+ },
245
+ {
246
+ "epoch": 0.21,
247
+ "learning_rate": 0.001,
248
+ "loss": 1.9471,
249
+ "step": 400
250
+ },
251
+ {
252
+ "epoch": 0.21,
253
+ "learning_rate": 0.001,
254
+ "loss": 1.9434,
255
+ "step": 410
256
+ },
257
+ {
258
+ "epoch": 0.22,
259
+ "learning_rate": 0.001,
260
+ "loss": 1.928,
261
+ "step": 420
262
+ },
263
+ {
264
+ "epoch": 0.22,
265
+ "learning_rate": 0.001,
266
+ "loss": 1.9543,
267
+ "step": 430
268
+ },
269
+ {
270
+ "epoch": 0.23,
271
+ "learning_rate": 0.001,
272
+ "loss": 1.8994,
273
+ "step": 440
274
+ },
275
+ {
276
+ "epoch": 0.23,
277
+ "learning_rate": 0.001,
278
+ "loss": 2.029,
279
+ "step": 450
280
+ },
281
+ {
282
+ "epoch": 0.24,
283
+ "learning_rate": 0.001,
284
+ "loss": 1.9449,
285
+ "step": 460
286
+ },
287
+ {
288
+ "epoch": 0.24,
289
+ "learning_rate": 0.001,
290
+ "loss": 1.9539,
291
+ "step": 470
292
+ },
293
+ {
294
+ "epoch": 0.25,
295
+ "learning_rate": 0.001,
296
+ "loss": 1.9541,
297
+ "step": 480
298
+ },
299
+ {
300
+ "epoch": 0.25,
301
+ "learning_rate": 0.001,
302
+ "loss": 1.956,
303
+ "step": 490
304
+ },
305
+ {
306
+ "epoch": 0.26,
307
+ "learning_rate": 0.001,
308
+ "loss": 1.9707,
309
+ "step": 500
310
+ },
311
+ {
312
+ "epoch": 0.26,
313
+ "learning_rate": 0.001,
314
+ "loss": 1.9402,
315
+ "step": 510
316
+ },
317
+ {
318
+ "epoch": 0.27,
319
+ "learning_rate": 0.001,
320
+ "loss": 1.9656,
321
+ "step": 520
322
+ },
323
+ {
324
+ "epoch": 0.28,
325
+ "learning_rate": 0.001,
326
+ "loss": 1.9372,
327
+ "step": 530
328
+ },
329
+ {
330
+ "epoch": 0.28,
331
+ "learning_rate": 0.001,
332
+ "loss": 1.955,
333
+ "step": 540
334
+ },
335
+ {
336
+ "epoch": 0.29,
337
+ "learning_rate": 0.001,
338
+ "loss": 1.9288,
339
+ "step": 550
340
+ },
341
+ {
342
+ "epoch": 0.29,
343
+ "learning_rate": 0.001,
344
+ "loss": 1.9574,
345
+ "step": 560
346
+ },
347
+ {
348
+ "epoch": 0.3,
349
+ "learning_rate": 0.001,
350
+ "loss": 1.956,
351
+ "step": 570
352
+ },
353
+ {
354
+ "epoch": 0.3,
355
+ "learning_rate": 0.001,
356
+ "loss": 1.9587,
357
+ "step": 580
358
+ },
359
+ {
360
+ "epoch": 0.31,
361
+ "learning_rate": 0.001,
362
+ "loss": 1.9473,
363
+ "step": 590
364
+ },
365
+ {
366
+ "epoch": 0.31,
367
+ "learning_rate": 0.001,
368
+ "loss": 1.9452,
369
+ "step": 600
370
+ },
371
+ {
372
+ "epoch": 0.32,
373
+ "learning_rate": 0.001,
374
+ "loss": 1.9648,
375
+ "step": 610
376
+ },
377
+ {
378
+ "epoch": 0.32,
379
+ "learning_rate": 0.001,
380
+ "loss": 1.9314,
381
+ "step": 620
382
+ },
383
+ {
384
+ "epoch": 0.33,
385
+ "learning_rate": 0.001,
386
+ "loss": 1.961,
387
+ "step": 630
388
+ },
389
+ {
390
+ "epoch": 0.33,
391
+ "learning_rate": 0.001,
392
+ "loss": 1.9386,
393
+ "step": 640
394
+ },
395
+ {
396
+ "epoch": 0.34,
397
+ "learning_rate": 0.001,
398
+ "loss": 1.9546,
399
+ "step": 650
400
+ },
401
+ {
402
+ "epoch": 0.34,
403
+ "learning_rate": 0.001,
404
+ "loss": 1.967,
405
+ "step": 660
406
+ },
407
+ {
408
+ "epoch": 0.35,
409
+ "learning_rate": 0.001,
410
+ "loss": 1.9451,
411
+ "step": 670
412
+ },
413
+ {
414
+ "epoch": 0.35,
415
+ "learning_rate": 0.001,
416
+ "loss": 1.9475,
417
+ "step": 680
418
+ },
419
+ {
420
+ "epoch": 0.36,
421
+ "learning_rate": 0.001,
422
+ "loss": 1.9518,
423
+ "step": 690
424
+ },
425
+ {
426
+ "epoch": 0.36,
427
+ "learning_rate": 0.001,
428
+ "loss": 1.9518,
429
+ "step": 700
430
+ },
431
+ {
432
+ "epoch": 0.37,
433
+ "learning_rate": 0.001,
434
+ "loss": 1.9481,
435
+ "step": 710
436
+ },
437
+ {
438
+ "epoch": 0.37,
439
+ "learning_rate": 0.001,
440
+ "loss": 1.9456,
441
+ "step": 720
442
+ },
443
+ {
444
+ "epoch": 0.38,
445
+ "learning_rate": 0.001,
446
+ "loss": 1.9362,
447
+ "step": 730
448
+ },
449
+ {
450
+ "epoch": 0.38,
451
+ "learning_rate": 0.001,
452
+ "loss": 1.9188,
453
+ "step": 740
454
+ },
455
+ {
456
+ "epoch": 0.39,
457
+ "learning_rate": 0.001,
458
+ "loss": 2.0077,
459
+ "step": 750
460
+ },
461
+ {
462
+ "epoch": 0.39,
463
+ "learning_rate": 0.001,
464
+ "loss": 1.9442,
465
+ "step": 760
466
+ },
467
+ {
468
+ "epoch": 0.4,
469
+ "learning_rate": 0.001,
470
+ "loss": 1.9348,
471
+ "step": 770
472
+ },
473
+ {
474
+ "epoch": 0.4,
475
+ "learning_rate": 0.001,
476
+ "loss": 1.9512,
477
+ "step": 780
478
+ },
479
+ {
480
+ "epoch": 0.41,
481
+ "learning_rate": 0.001,
482
+ "loss": 1.9409,
483
+ "step": 790
484
+ },
485
+ {
486
+ "epoch": 0.42,
487
+ "learning_rate": 0.001,
488
+ "loss": 1.9404,
489
+ "step": 800
490
+ },
491
+ {
492
+ "epoch": 0.42,
493
+ "learning_rate": 0.001,
494
+ "loss": 1.9656,
495
+ "step": 810
496
+ },
497
+ {
498
+ "epoch": 0.43,
499
+ "learning_rate": 0.001,
500
+ "loss": 1.948,
501
+ "step": 820
502
+ },
503
+ {
504
+ "epoch": 0.43,
505
+ "learning_rate": 0.001,
506
+ "loss": 1.9471,
507
+ "step": 830
508
+ },
509
+ {
510
+ "epoch": 0.44,
511
+ "learning_rate": 0.001,
512
+ "loss": 1.9453,
513
+ "step": 840
514
+ },
515
+ {
516
+ "epoch": 0.44,
517
+ "learning_rate": 0.001,
518
+ "loss": 1.9386,
519
+ "step": 850
520
+ },
521
+ {
522
+ "epoch": 0.45,
523
+ "learning_rate": 0.001,
524
+ "loss": 1.9568,
525
+ "step": 860
526
+ },
527
+ {
528
+ "epoch": 0.45,
529
+ "learning_rate": 0.001,
530
+ "loss": 1.9498,
531
+ "step": 870
532
+ },
533
+ {
534
+ "epoch": 0.46,
535
+ "learning_rate": 0.001,
536
+ "loss": 1.9621,
537
+ "step": 880
538
+ },
539
+ {
540
+ "epoch": 0.46,
541
+ "learning_rate": 0.001,
542
+ "loss": 1.9444,
543
+ "step": 890
544
+ },
545
+ {
546
+ "epoch": 0.47,
547
+ "learning_rate": 0.001,
548
+ "loss": 1.9515,
549
+ "step": 900
550
+ },
551
+ {
552
+ "epoch": 0.47,
553
+ "learning_rate": 0.001,
554
+ "loss": 1.9419,
555
+ "step": 910
556
+ },
557
+ {
558
+ "epoch": 0.48,
559
+ "learning_rate": 0.001,
560
+ "loss": 1.948,
561
+ "step": 920
562
+ },
563
+ {
564
+ "epoch": 0.48,
565
+ "learning_rate": 0.001,
566
+ "loss": 1.9427,
567
+ "step": 930
568
+ },
569
+ {
570
+ "epoch": 0.49,
571
+ "learning_rate": 0.001,
572
+ "loss": 1.9586,
573
+ "step": 940
574
+ },
575
+ {
576
+ "epoch": 0.49,
577
+ "learning_rate": 0.001,
578
+ "loss": 1.9588,
579
+ "step": 950
580
+ },
581
+ {
582
+ "epoch": 0.5,
583
+ "learning_rate": 0.001,
584
+ "loss": 1.9511,
585
+ "step": 960
586
+ },
587
+ {
588
+ "epoch": 0.5,
589
+ "learning_rate": 0.001,
590
+ "loss": 1.9505,
591
+ "step": 970
592
+ },
593
+ {
594
+ "epoch": 0.51,
595
+ "learning_rate": 0.001,
596
+ "loss": 1.9424,
597
+ "step": 980
598
+ },
599
+ {
600
+ "epoch": 0.51,
601
+ "learning_rate": 0.001,
602
+ "loss": 1.9378,
603
+ "step": 990
604
+ },
605
+ {
606
+ "epoch": 0.52,
607
+ "learning_rate": 0.001,
608
+ "loss": 1.9501,
609
+ "step": 1000
610
+ },
611
+ {
612
+ "epoch": 0.52,
613
+ "learning_rate": 0.001,
614
+ "loss": 1.9352,
615
+ "step": 1010
616
+ },
617
+ {
618
+ "epoch": 0.53,
619
+ "learning_rate": 0.001,
620
+ "loss": 1.9338,
621
+ "step": 1020
622
+ },
623
+ {
624
+ "epoch": 0.53,
625
+ "learning_rate": 0.001,
626
+ "loss": 1.9546,
627
+ "step": 1030
628
+ },
629
+ {
630
+ "epoch": 0.54,
631
+ "learning_rate": 0.001,
632
+ "loss": 1.9436,
633
+ "step": 1040
634
+ },
635
+ {
636
+ "epoch": 0.54,
637
+ "learning_rate": 0.001,
638
+ "loss": 1.9347,
639
+ "step": 1050
640
+ },
641
+ {
642
+ "epoch": 0.55,
643
+ "learning_rate": 0.001,
644
+ "loss": 1.9458,
645
+ "step": 1060
646
+ },
647
+ {
648
+ "epoch": 0.56,
649
+ "learning_rate": 0.001,
650
+ "loss": 1.9389,
651
+ "step": 1070
652
+ },
653
+ {
654
+ "epoch": 0.56,
655
+ "learning_rate": 0.001,
656
+ "loss": 1.9445,
657
+ "step": 1080
658
+ },
659
+ {
660
+ "epoch": 0.57,
661
+ "learning_rate": 0.001,
662
+ "loss": 1.9594,
663
+ "step": 1090
664
+ },
665
+ {
666
+ "epoch": 0.57,
667
+ "learning_rate": 0.001,
668
+ "loss": 1.9666,
669
+ "step": 1100
670
+ },
671
+ {
672
+ "epoch": 0.58,
673
+ "learning_rate": 0.001,
674
+ "loss": 1.9567,
675
+ "step": 1110
676
+ },
677
+ {
678
+ "epoch": 0.58,
679
+ "learning_rate": 0.001,
680
+ "loss": 1.9527,
681
+ "step": 1120
682
+ },
683
+ {
684
+ "epoch": 0.59,
685
+ "learning_rate": 0.001,
686
+ "loss": 1.9434,
687
+ "step": 1130
688
+ },
689
+ {
690
+ "epoch": 0.59,
691
+ "learning_rate": 0.001,
692
+ "loss": 1.9252,
693
+ "step": 1140
694
+ },
695
+ {
696
+ "epoch": 0.6,
697
+ "learning_rate": 0.001,
698
+ "loss": 1.9541,
699
+ "step": 1150
700
+ },
701
+ {
702
+ "epoch": 0.6,
703
+ "learning_rate": 0.001,
704
+ "loss": 1.9619,
705
+ "step": 1160
706
+ },
707
+ {
708
+ "epoch": 0.61,
709
+ "learning_rate": 0.001,
710
+ "loss": 1.9531,
711
+ "step": 1170
712
+ },
713
+ {
714
+ "epoch": 0.61,
715
+ "learning_rate": 0.001,
716
+ "loss": 1.9419,
717
+ "step": 1180
718
+ },
719
+ {
720
+ "epoch": 0.62,
721
+ "learning_rate": 0.001,
722
+ "loss": 1.9521,
723
+ "step": 1190
724
+ },
725
+ {
726
+ "epoch": 0.62,
727
+ "learning_rate": 0.001,
728
+ "loss": 1.9358,
729
+ "step": 1200
730
+ },
731
+ {
732
+ "epoch": 0.63,
733
+ "learning_rate": 0.001,
734
+ "loss": 1.9166,
735
+ "step": 1210
736
+ },
737
+ {
738
+ "epoch": 0.63,
739
+ "learning_rate": 0.001,
740
+ "loss": 2.0,
741
+ "step": 1220
742
+ },
743
+ {
744
+ "epoch": 0.64,
745
+ "learning_rate": 0.001,
746
+ "loss": 1.9459,
747
+ "step": 1230
748
+ },
749
+ {
750
+ "epoch": 0.64,
751
+ "learning_rate": 0.001,
752
+ "loss": 1.9397,
753
+ "step": 1240
754
+ },
755
+ {
756
+ "epoch": 0.65,
757
+ "learning_rate": 0.001,
758
+ "loss": 1.9722,
759
+ "step": 1250
760
+ },
761
+ {
762
+ "epoch": 0.65,
763
+ "learning_rate": 0.001,
764
+ "loss": 1.9704,
765
+ "step": 1260
766
+ },
767
+ {
768
+ "epoch": 0.66,
769
+ "learning_rate": 0.001,
770
+ "loss": 1.957,
771
+ "step": 1270
772
+ },
773
+ {
774
+ "epoch": 0.66,
775
+ "learning_rate": 0.001,
776
+ "loss": 1.9551,
777
+ "step": 1280
778
+ },
779
+ {
780
+ "epoch": 0.67,
781
+ "learning_rate": 0.001,
782
+ "loss": 1.9534,
783
+ "step": 1290
784
+ },
785
+ {
786
+ "epoch": 0.67,
787
+ "learning_rate": 0.001,
788
+ "loss": 1.9718,
789
+ "step": 1300
790
+ },
791
+ {
792
+ "epoch": 0.68,
793
+ "learning_rate": 0.001,
794
+ "loss": 1.9597,
795
+ "step": 1310
796
+ },
797
+ {
798
+ "epoch": 0.69,
799
+ "learning_rate": 0.001,
800
+ "loss": 1.9527,
801
+ "step": 1320
802
+ },
803
+ {
804
+ "epoch": 0.69,
805
+ "learning_rate": 0.001,
806
+ "loss": 1.9362,
807
+ "step": 1330
808
+ },
809
+ {
810
+ "epoch": 0.7,
811
+ "learning_rate": 0.001,
812
+ "loss": 1.9636,
813
+ "step": 1340
814
+ },
815
+ {
816
+ "epoch": 0.7,
817
+ "learning_rate": 0.001,
818
+ "loss": 1.95,
819
+ "step": 1350
820
+ },
821
+ {
822
+ "epoch": 0.71,
823
+ "learning_rate": 0.001,
824
+ "loss": 1.9551,
825
+ "step": 1360
826
+ },
827
+ {
828
+ "epoch": 0.71,
829
+ "learning_rate": 0.001,
830
+ "loss": 2.0172,
831
+ "step": 1370
832
+ },
833
+ {
834
+ "epoch": 0.72,
835
+ "learning_rate": 0.001,
836
+ "loss": 1.943,
837
+ "step": 1380
838
+ },
839
+ {
840
+ "epoch": 0.72,
841
+ "learning_rate": 0.001,
842
+ "loss": 1.9486,
843
+ "step": 1390
844
+ },
845
+ {
846
+ "epoch": 0.73,
847
+ "learning_rate": 0.001,
848
+ "loss": 1.9877,
849
+ "step": 1400
850
+ },
851
+ {
852
+ "epoch": 0.73,
853
+ "learning_rate": 0.001,
854
+ "loss": 1.976,
855
+ "step": 1410
856
+ },
857
+ {
858
+ "epoch": 0.74,
859
+ "learning_rate": 0.001,
860
+ "loss": 1.9371,
861
+ "step": 1420
862
+ },
863
+ {
864
+ "epoch": 0.74,
865
+ "learning_rate": 0.001,
866
+ "loss": 1.9797,
867
+ "step": 1430
868
+ },
869
+ {
870
+ "epoch": 0.75,
871
+ "learning_rate": 0.001,
872
+ "loss": 1.9719,
873
+ "step": 1440
874
+ },
875
+ {
876
+ "epoch": 0.75,
877
+ "learning_rate": 0.001,
878
+ "loss": 1.925,
879
+ "step": 1450
880
+ },
881
+ {
882
+ "epoch": 0.76,
883
+ "learning_rate": 0.001,
884
+ "loss": 1.9756,
885
+ "step": 1460
886
+ },
887
+ {
888
+ "epoch": 0.76,
889
+ "learning_rate": 0.001,
890
+ "loss": 1.9553,
891
+ "step": 1470
892
+ },
893
+ {
894
+ "epoch": 0.77,
895
+ "learning_rate": 0.001,
896
+ "loss": 1.949,
897
+ "step": 1480
898
+ },
899
+ {
900
+ "epoch": 0.77,
901
+ "learning_rate": 0.001,
902
+ "loss": 1.9494,
903
+ "step": 1490
904
+ },
905
+ {
906
+ "epoch": 0.78,
907
+ "learning_rate": 0.001,
908
+ "loss": 1.9454,
909
+ "step": 1500
910
+ },
911
+ {
912
+ "epoch": 0.78,
913
+ "learning_rate": 0.001,
914
+ "loss": 1.9347,
915
+ "step": 1510
916
+ },
917
+ {
918
+ "epoch": 0.79,
919
+ "learning_rate": 0.001,
920
+ "loss": 1.9342,
921
+ "step": 1520
922
+ },
923
+ {
924
+ "epoch": 0.79,
925
+ "learning_rate": 0.001,
926
+ "loss": 1.9762,
927
+ "step": 1530
928
+ },
929
+ {
930
+ "epoch": 0.8,
931
+ "learning_rate": 0.001,
932
+ "loss": 1.9467,
933
+ "step": 1540
934
+ },
935
+ {
936
+ "epoch": 0.8,
937
+ "learning_rate": 0.001,
938
+ "loss": 1.9493,
939
+ "step": 1550
940
+ },
941
+ {
942
+ "epoch": 0.81,
943
+ "learning_rate": 0.001,
944
+ "loss": 1.9309,
945
+ "step": 1560
946
+ },
947
+ {
948
+ "epoch": 0.81,
949
+ "learning_rate": 0.001,
950
+ "loss": 1.9875,
951
+ "step": 1570
952
+ },
953
+ {
954
+ "epoch": 0.82,
955
+ "learning_rate": 0.001,
956
+ "loss": 1.9379,
957
+ "step": 1580
958
+ },
959
+ {
960
+ "epoch": 0.83,
961
+ "learning_rate": 0.001,
962
+ "loss": 1.9231,
963
+ "step": 1590
964
+ },
965
+ {
966
+ "epoch": 0.83,
967
+ "learning_rate": 0.001,
968
+ "loss": 1.9163,
969
+ "step": 1600
970
+ },
971
+ {
972
+ "epoch": 0.84,
973
+ "learning_rate": 0.001,
974
+ "loss": 1.9894,
975
+ "step": 1610
976
+ },
977
+ {
978
+ "epoch": 0.84,
979
+ "learning_rate": 0.001,
980
+ "loss": 1.9543,
981
+ "step": 1620
982
+ },
983
+ {
984
+ "epoch": 0.85,
985
+ "learning_rate": 0.001,
986
+ "loss": 1.9237,
987
+ "step": 1630
988
+ },
989
+ {
990
+ "epoch": 0.85,
991
+ "learning_rate": 0.001,
992
+ "loss": 2.1552,
993
+ "step": 1640
994
+ },
995
+ {
996
+ "epoch": 0.86,
997
+ "learning_rate": 0.001,
998
+ "loss": 2.0399,
999
+ "step": 1650
1000
+ },
1001
+ {
1002
+ "epoch": 0.86,
1003
+ "learning_rate": 0.001,
1004
+ "loss": 1.971,
1005
+ "step": 1660
1006
+ },
1007
+ {
1008
+ "epoch": 0.87,
1009
+ "learning_rate": 0.001,
1010
+ "loss": 2.0006,
1011
+ "step": 1670
1012
+ },
1013
+ {
1014
+ "epoch": 0.87,
1015
+ "learning_rate": 0.001,
1016
+ "loss": 1.9451,
1017
+ "step": 1680
1018
+ },
1019
+ {
1020
+ "epoch": 0.88,
1021
+ "learning_rate": 0.001,
1022
+ "loss": 1.993,
1023
+ "step": 1690
1024
+ },
1025
+ {
1026
+ "epoch": 0.88,
1027
+ "learning_rate": 0.001,
1028
+ "loss": 1.9586,
1029
+ "step": 1700
1030
+ },
1031
+ {
1032
+ "epoch": 0.89,
1033
+ "learning_rate": 0.001,
1034
+ "loss": 1.963,
1035
+ "step": 1710
1036
+ },
1037
+ {
1038
+ "epoch": 0.89,
1039
+ "learning_rate": 0.001,
1040
+ "loss": 1.9641,
1041
+ "step": 1720
1042
+ },
1043
+ {
1044
+ "epoch": 0.9,
1045
+ "learning_rate": 0.001,
1046
+ "loss": 1.9555,
1047
+ "step": 1730
1048
+ },
1049
+ {
1050
+ "epoch": 0.9,
1051
+ "learning_rate": 0.001,
1052
+ "loss": 1.9381,
1053
+ "step": 1740
1054
+ },
1055
+ {
1056
+ "epoch": 0.91,
1057
+ "learning_rate": 0.001,
1058
+ "loss": 1.934,
1059
+ "step": 1750
1060
+ },
1061
+ {
1062
+ "epoch": 0.91,
1063
+ "learning_rate": 0.001,
1064
+ "loss": 2.0062,
1065
+ "step": 1760
1066
+ },
1067
+ {
1068
+ "epoch": 0.92,
1069
+ "learning_rate": 0.001,
1070
+ "loss": 1.9315,
1071
+ "step": 1770
1072
+ },
1073
+ {
1074
+ "epoch": 0.92,
1075
+ "learning_rate": 0.001,
1076
+ "loss": 1.9626,
1077
+ "step": 1780
1078
+ },
1079
+ {
1080
+ "epoch": 0.93,
1081
+ "learning_rate": 0.001,
1082
+ "loss": 1.9135,
1083
+ "step": 1790
1084
+ },
1085
+ {
1086
+ "epoch": 0.93,
1087
+ "learning_rate": 0.001,
1088
+ "loss": 1.9756,
1089
+ "step": 1800
1090
+ },
1091
+ {
1092
+ "epoch": 0.94,
1093
+ "learning_rate": 0.001,
1094
+ "loss": 1.9628,
1095
+ "step": 1810
1096
+ },
1097
+ {
1098
+ "epoch": 0.94,
1099
+ "learning_rate": 0.001,
1100
+ "loss": 1.952,
1101
+ "step": 1820
1102
+ },
1103
+ {
1104
+ "epoch": 0.95,
1105
+ "learning_rate": 0.001,
1106
+ "loss": 1.9326,
1107
+ "step": 1830
1108
+ },
1109
+ {
1110
+ "epoch": 0.95,
1111
+ "learning_rate": 0.001,
1112
+ "loss": 1.9552,
1113
+ "step": 1840
1114
+ },
1115
+ {
1116
+ "epoch": 0.96,
1117
+ "learning_rate": 0.001,
1118
+ "loss": 1.9527,
1119
+ "step": 1850
1120
+ },
1121
+ {
1122
+ "epoch": 0.97,
1123
+ "learning_rate": 0.001,
1124
+ "loss": 1.938,
1125
+ "step": 1860
1126
+ },
1127
+ {
1128
+ "epoch": 0.97,
1129
+ "learning_rate": 0.001,
1130
+ "loss": 1.9452,
1131
+ "step": 1870
1132
+ },
1133
+ {
1134
+ "epoch": 0.98,
1135
+ "learning_rate": 0.001,
1136
+ "loss": 1.9515,
1137
+ "step": 1880
1138
+ },
1139
+ {
1140
+ "epoch": 0.98,
1141
+ "learning_rate": 0.001,
1142
+ "loss": 1.9465,
1143
+ "step": 1890
1144
+ },
1145
+ {
1146
+ "epoch": 0.99,
1147
+ "learning_rate": 0.001,
1148
+ "loss": 1.9524,
1149
+ "step": 1900
1150
+ },
1151
+ {
1152
+ "epoch": 0.99,
1153
+ "learning_rate": 0.001,
1154
+ "loss": 1.9271,
1155
+ "step": 1910
1156
+ },
1157
+ {
1158
+ "epoch": 1.0,
1159
+ "learning_rate": 0.001,
1160
+ "loss": 1.981,
1161
+ "step": 1920
1162
+ },
1163
+ {
1164
+ "epoch": 1.0,
1165
+ "eval_accuracy": {
1166
+ "accuracy": 0.1556420233463035
1167
+ },
1168
+ "eval_f1": {
1169
+ "f1": 0.03848003848003848
1170
+ },
1171
+ "eval_loss": 1.9584157466888428,
1172
+ "eval_precision": {
1173
+ "precision": 0.022234574763757644
1174
+ },
1175
+ "eval_recall": {
1176
+ "recall": 0.14285714285714285
1177
+ },
1178
+ "eval_runtime": 201.5846,
1179
+ "eval_samples_per_second": 19.123,
1180
+ "eval_steps_per_second": 9.564,
1181
+ "step": 1927
1182
+ },
1183
+ {
1184
+ "epoch": 1.0,
1185
+ "learning_rate": 0.001,
1186
+ "loss": 1.9356,
1187
+ "step": 1930
1188
+ },
1189
+ {
1190
+ "epoch": 1.01,
1191
+ "learning_rate": 0.001,
1192
+ "loss": 1.9557,
1193
+ "step": 1940
1194
+ },
1195
+ {
1196
+ "epoch": 1.01,
1197
+ "learning_rate": 0.001,
1198
+ "loss": 1.9435,
1199
+ "step": 1950
1200
+ },
1201
+ {
1202
+ "epoch": 1.02,
1203
+ "learning_rate": 0.001,
1204
+ "loss": 1.9192,
1205
+ "step": 1960
1206
+ },
1207
+ {
1208
+ "epoch": 1.02,
1209
+ "learning_rate": 0.001,
1210
+ "loss": 1.9903,
1211
+ "step": 1970
1212
+ },
1213
+ {
1214
+ "epoch": 1.03,
1215
+ "learning_rate": 0.001,
1216
+ "loss": 1.9343,
1217
+ "step": 1980
1218
+ },
1219
+ {
1220
+ "epoch": 1.03,
1221
+ "learning_rate": 0.001,
1222
+ "loss": 1.9753,
1223
+ "step": 1990
1224
+ },
1225
+ {
1226
+ "epoch": 1.04,
1227
+ "learning_rate": 0.001,
1228
+ "loss": 1.9178,
1229
+ "step": 2000
1230
+ },
1231
+ {
1232
+ "epoch": 1.04,
1233
+ "learning_rate": 0.001,
1234
+ "loss": 1.925,
1235
+ "step": 2010
1236
+ },
1237
+ {
1238
+ "epoch": 1.05,
1239
+ "learning_rate": 0.001,
1240
+ "loss": 1.955,
1241
+ "step": 2020
1242
+ },
1243
+ {
1244
+ "epoch": 1.05,
1245
+ "learning_rate": 0.001,
1246
+ "loss": 1.9774,
1247
+ "step": 2030
1248
+ },
1249
+ {
1250
+ "epoch": 1.06,
1251
+ "learning_rate": 0.001,
1252
+ "loss": 1.9351,
1253
+ "step": 2040
1254
+ },
1255
+ {
1256
+ "epoch": 1.06,
1257
+ "learning_rate": 0.001,
1258
+ "loss": 1.9794,
1259
+ "step": 2050
1260
+ },
1261
+ {
1262
+ "epoch": 1.07,
1263
+ "learning_rate": 0.001,
1264
+ "loss": 1.9379,
1265
+ "step": 2060
1266
+ },
1267
+ {
1268
+ "epoch": 1.07,
1269
+ "learning_rate": 0.001,
1270
+ "loss": 1.9512,
1271
+ "step": 2070
1272
+ },
1273
+ {
1274
+ "epoch": 1.08,
1275
+ "learning_rate": 0.001,
1276
+ "loss": 1.9483,
1277
+ "step": 2080
1278
+ },
1279
+ {
1280
+ "epoch": 1.08,
1281
+ "learning_rate": 0.001,
1282
+ "loss": 1.9376,
1283
+ "step": 2090
1284
+ },
1285
+ {
1286
+ "epoch": 1.09,
1287
+ "learning_rate": 0.001,
1288
+ "loss": 1.9282,
1289
+ "step": 2100
1290
+ },
1291
+ {
1292
+ "epoch": 1.09,
1293
+ "learning_rate": 0.001,
1294
+ "loss": 1.936,
1295
+ "step": 2110
1296
+ },
1297
+ {
1298
+ "epoch": 1.1,
1299
+ "learning_rate": 0.001,
1300
+ "loss": 1.9811,
1301
+ "step": 2120
1302
+ },
1303
+ {
1304
+ "epoch": 1.11,
1305
+ "learning_rate": 0.001,
1306
+ "loss": 1.9467,
1307
+ "step": 2130
1308
+ },
1309
+ {
1310
+ "epoch": 1.11,
1311
+ "learning_rate": 0.001,
1312
+ "loss": 1.9484,
1313
+ "step": 2140
1314
+ },
1315
+ {
1316
+ "epoch": 1.12,
1317
+ "learning_rate": 0.001,
1318
+ "loss": 1.9356,
1319
+ "step": 2150
1320
+ },
1321
+ {
1322
+ "epoch": 1.12,
1323
+ "learning_rate": 0.001,
1324
+ "loss": 1.9418,
1325
+ "step": 2160
1326
+ },
1327
+ {
1328
+ "epoch": 1.13,
1329
+ "learning_rate": 0.001,
1330
+ "loss": 1.944,
1331
+ "step": 2170
1332
+ },
1333
+ {
1334
+ "epoch": 1.13,
1335
+ "learning_rate": 0.001,
1336
+ "loss": 1.9451,
1337
+ "step": 2180
1338
+ },
1339
+ {
1340
+ "epoch": 1.14,
1341
+ "learning_rate": 0.001,
1342
+ "loss": 1.9423,
1343
+ "step": 2190
1344
+ },
1345
+ {
1346
+ "epoch": 1.14,
1347
+ "learning_rate": 0.001,
1348
+ "loss": 1.9435,
1349
+ "step": 2200
1350
+ },
1351
+ {
1352
+ "epoch": 1.15,
1353
+ "learning_rate": 0.001,
1354
+ "loss": 1.9374,
1355
+ "step": 2210
1356
+ },
1357
+ {
1358
+ "epoch": 1.15,
1359
+ "learning_rate": 0.001,
1360
+ "loss": 1.9436,
1361
+ "step": 2220
1362
+ },
1363
+ {
1364
+ "epoch": 1.16,
1365
+ "learning_rate": 0.001,
1366
+ "loss": 1.9377,
1367
+ "step": 2230
1368
+ },
1369
+ {
1370
+ "epoch": 1.16,
1371
+ "learning_rate": 0.001,
1372
+ "loss": 1.9339,
1373
+ "step": 2240
1374
+ },
1375
+ {
1376
+ "epoch": 1.17,
1377
+ "learning_rate": 0.001,
1378
+ "loss": 1.9285,
1379
+ "step": 2250
1380
+ },
1381
+ {
1382
+ "epoch": 1.17,
1383
+ "learning_rate": 0.001,
1384
+ "loss": 1.9482,
1385
+ "step": 2260
1386
+ },
1387
+ {
1388
+ "epoch": 1.18,
1389
+ "learning_rate": 0.001,
1390
+ "loss": 1.9719,
1391
+ "step": 2270
1392
+ },
1393
+ {
1394
+ "epoch": 1.18,
1395
+ "learning_rate": 0.001,
1396
+ "loss": 1.9526,
1397
+ "step": 2280
1398
+ },
1399
+ {
1400
+ "epoch": 1.19,
1401
+ "learning_rate": 0.001,
1402
+ "loss": 1.9551,
1403
+ "step": 2290
1404
+ },
1405
+ {
1406
+ "epoch": 1.19,
1407
+ "learning_rate": 0.001,
1408
+ "loss": 1.9423,
1409
+ "step": 2300
1410
+ },
1411
+ {
1412
+ "epoch": 1.2,
1413
+ "learning_rate": 0.001,
1414
+ "loss": 1.9501,
1415
+ "step": 2310
1416
+ },
1417
+ {
1418
+ "epoch": 1.2,
1419
+ "learning_rate": 0.001,
1420
+ "loss": 1.9412,
1421
+ "step": 2320
1422
+ },
1423
+ {
1424
+ "epoch": 1.21,
1425
+ "learning_rate": 0.001,
1426
+ "loss": 1.9403,
1427
+ "step": 2330
1428
+ },
1429
+ {
1430
+ "epoch": 1.21,
1431
+ "learning_rate": 0.001,
1432
+ "loss": 1.9462,
1433
+ "step": 2340
1434
+ },
1435
+ {
1436
+ "epoch": 1.22,
1437
+ "learning_rate": 0.001,
1438
+ "loss": 1.9344,
1439
+ "step": 2350
1440
+ },
1441
+ {
1442
+ "epoch": 1.22,
1443
+ "learning_rate": 0.001,
1444
+ "loss": 1.9686,
1445
+ "step": 2360
1446
+ },
1447
+ {
1448
+ "epoch": 1.23,
1449
+ "learning_rate": 0.001,
1450
+ "loss": 1.9476,
1451
+ "step": 2370
1452
+ },
1453
+ {
1454
+ "epoch": 1.24,
1455
+ "learning_rate": 0.001,
1456
+ "loss": 1.955,
1457
+ "step": 2380
1458
+ },
1459
+ {
1460
+ "epoch": 1.24,
1461
+ "learning_rate": 0.001,
1462
+ "loss": 1.9482,
1463
+ "step": 2390
1464
+ },
1465
+ {
1466
+ "epoch": 1.25,
1467
+ "learning_rate": 0.001,
1468
+ "loss": 1.947,
1469
+ "step": 2400
1470
+ },
1471
+ {
1472
+ "epoch": 1.25,
1473
+ "learning_rate": 0.001,
1474
+ "loss": 1.9498,
1475
+ "step": 2410
1476
+ },
1477
+ {
1478
+ "epoch": 1.26,
1479
+ "learning_rate": 0.001,
1480
+ "loss": 1.9463,
1481
+ "step": 2420
1482
+ },
1483
+ {
1484
+ "epoch": 1.26,
1485
+ "learning_rate": 0.001,
1486
+ "loss": 1.9486,
1487
+ "step": 2430
1488
+ },
1489
+ {
1490
+ "epoch": 1.27,
1491
+ "learning_rate": 0.001,
1492
+ "loss": 1.9521,
1493
+ "step": 2440
1494
+ },
1495
+ {
1496
+ "epoch": 1.27,
1497
+ "learning_rate": 0.001,
1498
+ "loss": 1.9454,
1499
+ "step": 2450
1500
+ },
1501
+ {
1502
+ "epoch": 1.28,
1503
+ "learning_rate": 0.001,
1504
+ "loss": 1.939,
1505
+ "step": 2460
1506
+ },
1507
+ {
1508
+ "epoch": 1.28,
1509
+ "learning_rate": 0.001,
1510
+ "loss": 1.9401,
1511
+ "step": 2470
1512
+ },
1513
+ {
1514
+ "epoch": 1.29,
1515
+ "learning_rate": 0.001,
1516
+ "loss": 1.9452,
1517
+ "step": 2480
1518
+ },
1519
+ {
1520
+ "epoch": 1.29,
1521
+ "learning_rate": 0.001,
1522
+ "loss": 1.9474,
1523
+ "step": 2490
1524
+ },
1525
+ {
1526
+ "epoch": 1.3,
1527
+ "learning_rate": 0.001,
1528
+ "loss": 1.9452,
1529
+ "step": 2500
1530
+ },
1531
+ {
1532
+ "epoch": 1.3,
1533
+ "learning_rate": 0.001,
1534
+ "loss": 1.9416,
1535
+ "step": 2510
1536
+ },
1537
+ {
1538
+ "epoch": 1.31,
1539
+ "learning_rate": 0.001,
1540
+ "loss": 1.9523,
1541
+ "step": 2520
1542
+ },
1543
+ {
1544
+ "epoch": 1.31,
1545
+ "learning_rate": 0.001,
1546
+ "loss": 1.9497,
1547
+ "step": 2530
1548
+ },
1549
+ {
1550
+ "epoch": 1.32,
1551
+ "learning_rate": 0.001,
1552
+ "loss": 1.9332,
1553
+ "step": 2540
1554
+ },
1555
+ {
1556
+ "epoch": 1.32,
1557
+ "learning_rate": 0.001,
1558
+ "loss": 1.9372,
1559
+ "step": 2550
1560
+ },
1561
+ {
1562
+ "epoch": 1.33,
1563
+ "learning_rate": 0.001,
1564
+ "loss": 1.9374,
1565
+ "step": 2560
1566
+ },
1567
+ {
1568
+ "epoch": 1.33,
1569
+ "learning_rate": 0.001,
1570
+ "loss": 1.9349,
1571
+ "step": 2570
1572
+ },
1573
+ {
1574
+ "epoch": 1.34,
1575
+ "learning_rate": 0.001,
1576
+ "loss": 1.9223,
1577
+ "step": 2580
1578
+ },
1579
+ {
1580
+ "epoch": 1.34,
1581
+ "learning_rate": 0.001,
1582
+ "loss": 1.9679,
1583
+ "step": 2590
1584
+ },
1585
+ {
1586
+ "epoch": 1.35,
1587
+ "learning_rate": 0.001,
1588
+ "loss": 1.9427,
1589
+ "step": 2600
1590
+ },
1591
+ {
1592
+ "epoch": 1.35,
1593
+ "learning_rate": 0.001,
1594
+ "loss": 1.9377,
1595
+ "step": 2610
1596
+ },
1597
+ {
1598
+ "epoch": 1.36,
1599
+ "learning_rate": 0.001,
1600
+ "loss": 1.9311,
1601
+ "step": 2620
1602
+ },
1603
+ {
1604
+ "epoch": 1.36,
1605
+ "learning_rate": 0.001,
1606
+ "loss": 1.951,
1607
+ "step": 2630
1608
+ },
1609
+ {
1610
+ "epoch": 1.37,
1611
+ "learning_rate": 0.001,
1612
+ "loss": 1.9154,
1613
+ "step": 2640
1614
+ },
1615
+ {
1616
+ "epoch": 1.38,
1617
+ "learning_rate": 0.001,
1618
+ "loss": 1.965,
1619
+ "step": 2650
1620
+ },
1621
+ {
1622
+ "epoch": 1.38,
1623
+ "learning_rate": 0.001,
1624
+ "loss": 1.9611,
1625
+ "step": 2660
1626
+ },
1627
+ {
1628
+ "epoch": 1.39,
1629
+ "learning_rate": 0.001,
1630
+ "loss": 1.9356,
1631
+ "step": 2670
1632
+ },
1633
+ {
1634
+ "epoch": 1.39,
1635
+ "learning_rate": 0.001,
1636
+ "loss": 1.9417,
1637
+ "step": 2680
1638
+ },
1639
+ {
1640
+ "epoch": 1.4,
1641
+ "learning_rate": 0.001,
1642
+ "loss": 1.9597,
1643
+ "step": 2690
1644
+ },
1645
+ {
1646
+ "epoch": 1.4,
1647
+ "learning_rate": 0.001,
1648
+ "loss": 1.9337,
1649
+ "step": 2700
1650
+ },
1651
+ {
1652
+ "epoch": 1.41,
1653
+ "learning_rate": 0.001,
1654
+ "loss": 1.9329,
1655
+ "step": 2710
1656
+ },
1657
+ {
1658
+ "epoch": 1.41,
1659
+ "learning_rate": 0.001,
1660
+ "loss": 1.934,
1661
+ "step": 2720
1662
+ },
1663
+ {
1664
+ "epoch": 1.42,
1665
+ "learning_rate": 0.001,
1666
+ "loss": 1.921,
1667
+ "step": 2730
1668
+ },
1669
+ {
1670
+ "epoch": 1.42,
1671
+ "learning_rate": 0.001,
1672
+ "loss": 1.9422,
1673
+ "step": 2740
1674
+ },
1675
+ {
1676
+ "epoch": 1.43,
1677
+ "learning_rate": 0.001,
1678
+ "loss": 1.9407,
1679
+ "step": 2750
1680
+ },
1681
+ {
1682
+ "epoch": 1.43,
1683
+ "learning_rate": 0.001,
1684
+ "loss": 1.9546,
1685
+ "step": 2760
1686
+ },
1687
+ {
1688
+ "epoch": 1.44,
1689
+ "learning_rate": 0.001,
1690
+ "loss": 1.9527,
1691
+ "step": 2770
1692
+ },
1693
+ {
1694
+ "epoch": 1.44,
1695
+ "learning_rate": 0.001,
1696
+ "loss": 1.9362,
1697
+ "step": 2780
1698
+ },
1699
+ {
1700
+ "epoch": 1.45,
1701
+ "learning_rate": 0.001,
1702
+ "loss": 1.9506,
1703
+ "step": 2790
1704
+ },
1705
+ {
1706
+ "epoch": 1.45,
1707
+ "learning_rate": 0.001,
1708
+ "loss": 1.9293,
1709
+ "step": 2800
1710
+ },
1711
+ {
1712
+ "epoch": 1.46,
1713
+ "learning_rate": 0.001,
1714
+ "loss": 1.9577,
1715
+ "step": 2810
1716
+ },
1717
+ {
1718
+ "epoch": 1.46,
1719
+ "learning_rate": 0.001,
1720
+ "loss": 1.9549,
1721
+ "step": 2820
1722
+ },
1723
+ {
1724
+ "epoch": 1.47,
1725
+ "learning_rate": 0.001,
1726
+ "loss": 1.9555,
1727
+ "step": 2830
1728
+ },
1729
+ {
1730
+ "epoch": 1.47,
1731
+ "learning_rate": 0.001,
1732
+ "loss": 1.9427,
1733
+ "step": 2840
1734
+ },
1735
+ {
1736
+ "epoch": 1.48,
1737
+ "learning_rate": 0.001,
1738
+ "loss": 1.9417,
1739
+ "step": 2850
1740
+ },
1741
+ {
1742
+ "epoch": 1.48,
1743
+ "learning_rate": 0.001,
1744
+ "loss": 1.9328,
1745
+ "step": 2860
1746
+ },
1747
+ {
1748
+ "epoch": 1.49,
1749
+ "learning_rate": 0.001,
1750
+ "loss": 1.948,
1751
+ "step": 2870
1752
+ },
1753
+ {
1754
+ "epoch": 1.49,
1755
+ "learning_rate": 0.001,
1756
+ "loss": 1.9619,
1757
+ "step": 2880
1758
+ },
1759
+ {
1760
+ "epoch": 1.5,
1761
+ "learning_rate": 0.001,
1762
+ "loss": 1.9359,
1763
+ "step": 2890
1764
+ },
1765
+ {
1766
+ "epoch": 1.5,
1767
+ "learning_rate": 0.001,
1768
+ "loss": 1.9323,
1769
+ "step": 2900
1770
+ },
1771
+ {
1772
+ "epoch": 1.51,
1773
+ "learning_rate": 0.001,
1774
+ "loss": 1.9226,
1775
+ "step": 2910
1776
+ },
1777
+ {
1778
+ "epoch": 1.52,
1779
+ "learning_rate": 0.001,
1780
+ "loss": 1.9836,
1781
+ "step": 2920
1782
+ },
1783
+ {
1784
+ "epoch": 1.52,
1785
+ "learning_rate": 0.001,
1786
+ "loss": 1.9518,
1787
+ "step": 2930
1788
+ },
1789
+ {
1790
+ "epoch": 1.53,
1791
+ "learning_rate": 0.001,
1792
+ "loss": 1.9512,
1793
+ "step": 2940
1794
+ },
1795
+ {
1796
+ "epoch": 1.53,
1797
+ "learning_rate": 0.001,
1798
+ "loss": 1.9445,
1799
+ "step": 2950
1800
+ },
1801
+ {
1802
+ "epoch": 1.54,
1803
+ "learning_rate": 0.001,
1804
+ "loss": 1.9528,
1805
+ "step": 2960
1806
+ },
1807
+ {
1808
+ "epoch": 1.54,
1809
+ "learning_rate": 0.001,
1810
+ "loss": 1.9518,
1811
+ "step": 2970
1812
+ },
1813
+ {
1814
+ "epoch": 1.55,
1815
+ "learning_rate": 0.001,
1816
+ "loss": 1.942,
1817
+ "step": 2980
1818
+ },
1819
+ {
1820
+ "epoch": 1.55,
1821
+ "learning_rate": 0.001,
1822
+ "loss": 1.9563,
1823
+ "step": 2990
1824
+ },
1825
+ {
1826
+ "epoch": 1.56,
1827
+ "learning_rate": 0.001,
1828
+ "loss": 1.9378,
1829
+ "step": 3000
1830
+ },
1831
+ {
1832
+ "epoch": 1.56,
1833
+ "learning_rate": 0.001,
1834
+ "loss": 1.9591,
1835
+ "step": 3010
1836
+ },
1837
+ {
1838
+ "epoch": 1.57,
1839
+ "learning_rate": 0.001,
1840
+ "loss": 1.9421,
1841
+ "step": 3020
1842
+ },
1843
+ {
1844
+ "epoch": 1.57,
1845
+ "learning_rate": 0.001,
1846
+ "loss": 1.9447,
1847
+ "step": 3030
1848
+ },
1849
+ {
1850
+ "epoch": 1.58,
1851
+ "learning_rate": 0.001,
1852
+ "loss": 1.9345,
1853
+ "step": 3040
1854
+ },
1855
+ {
1856
+ "epoch": 1.58,
1857
+ "learning_rate": 0.001,
1858
+ "loss": 1.9399,
1859
+ "step": 3050
1860
+ },
1861
+ {
1862
+ "epoch": 1.59,
1863
+ "learning_rate": 0.001,
1864
+ "loss": 1.9321,
1865
+ "step": 3060
1866
+ },
1867
+ {
1868
+ "epoch": 1.59,
1869
+ "learning_rate": 0.001,
1870
+ "loss": 1.9367,
1871
+ "step": 3070
1872
+ },
1873
+ {
1874
+ "epoch": 1.6,
1875
+ "learning_rate": 0.001,
1876
+ "loss": 1.9368,
1877
+ "step": 3080
1878
+ },
1879
+ {
1880
+ "epoch": 1.6,
1881
+ "learning_rate": 0.001,
1882
+ "loss": 1.9531,
1883
+ "step": 3090
1884
+ },
1885
+ {
1886
+ "epoch": 1.61,
1887
+ "learning_rate": 0.001,
1888
+ "loss": 1.9573,
1889
+ "step": 3100
1890
+ },
1891
+ {
1892
+ "epoch": 1.61,
1893
+ "learning_rate": 0.001,
1894
+ "loss": 1.9359,
1895
+ "step": 3110
1896
+ },
1897
+ {
1898
+ "epoch": 1.62,
1899
+ "learning_rate": 0.001,
1900
+ "loss": 1.9541,
1901
+ "step": 3120
1902
+ },
1903
+ {
1904
+ "epoch": 1.62,
1905
+ "learning_rate": 0.001,
1906
+ "loss": 1.9457,
1907
+ "step": 3130
1908
+ },
1909
+ {
1910
+ "epoch": 1.63,
1911
+ "learning_rate": 0.001,
1912
+ "loss": 1.9443,
1913
+ "step": 3140
1914
+ },
1915
+ {
1916
+ "epoch": 1.63,
1917
+ "learning_rate": 0.001,
1918
+ "loss": 1.9431,
1919
+ "step": 3150
1920
+ },
1921
+ {
1922
+ "epoch": 1.64,
1923
+ "learning_rate": 0.001,
1924
+ "loss": 1.9507,
1925
+ "step": 3160
1926
+ },
1927
+ {
1928
+ "epoch": 1.65,
1929
+ "learning_rate": 0.001,
1930
+ "loss": 1.9282,
1931
+ "step": 3170
1932
+ },
1933
+ {
1934
+ "epoch": 1.65,
1935
+ "learning_rate": 0.001,
1936
+ "loss": 1.9731,
1937
+ "step": 3180
1938
+ },
1939
+ {
1940
+ "epoch": 1.66,
1941
+ "learning_rate": 0.001,
1942
+ "loss": 1.9422,
1943
+ "step": 3190
1944
+ },
1945
+ {
1946
+ "epoch": 1.66,
1947
+ "learning_rate": 0.001,
1948
+ "loss": 1.9436,
1949
+ "step": 3200
1950
+ },
1951
+ {
1952
+ "epoch": 1.67,
1953
+ "learning_rate": 0.001,
1954
+ "loss": 1.9439,
1955
+ "step": 3210
1956
+ },
1957
+ {
1958
+ "epoch": 1.67,
1959
+ "learning_rate": 0.001,
1960
+ "loss": 1.9339,
1961
+ "step": 3220
1962
+ },
1963
+ {
1964
+ "epoch": 1.68,
1965
+ "learning_rate": 0.001,
1966
+ "loss": 1.9413,
1967
+ "step": 3230
1968
+ },
1969
+ {
1970
+ "epoch": 1.68,
1971
+ "learning_rate": 0.001,
1972
+ "loss": 1.9429,
1973
+ "step": 3240
1974
+ },
1975
+ {
1976
+ "epoch": 1.69,
1977
+ "learning_rate": 0.001,
1978
+ "loss": 1.9539,
1979
+ "step": 3250
1980
+ },
1981
+ {
1982
+ "epoch": 1.69,
1983
+ "learning_rate": 0.001,
1984
+ "loss": 1.9412,
1985
+ "step": 3260
1986
+ },
1987
+ {
1988
+ "epoch": 1.7,
1989
+ "learning_rate": 0.001,
1990
+ "loss": 1.9442,
1991
+ "step": 3270
1992
+ },
1993
+ {
1994
+ "epoch": 1.7,
1995
+ "learning_rate": 0.001,
1996
+ "loss": 1.953,
1997
+ "step": 3280
1998
+ },
1999
+ {
2000
+ "epoch": 1.71,
2001
+ "learning_rate": 0.001,
2002
+ "loss": 1.9504,
2003
+ "step": 3290
2004
+ },
2005
+ {
2006
+ "epoch": 1.71,
2007
+ "learning_rate": 0.001,
2008
+ "loss": 1.9493,
2009
+ "step": 3300
2010
+ },
2011
+ {
2012
+ "epoch": 1.72,
2013
+ "learning_rate": 0.001,
2014
+ "loss": 1.9371,
2015
+ "step": 3310
2016
+ },
2017
+ {
2018
+ "epoch": 1.72,
2019
+ "learning_rate": 0.001,
2020
+ "loss": 1.9394,
2021
+ "step": 3320
2022
+ },
2023
+ {
2024
+ "epoch": 1.73,
2025
+ "learning_rate": 0.001,
2026
+ "loss": 1.9654,
2027
+ "step": 3330
2028
+ },
2029
+ {
2030
+ "epoch": 1.73,
2031
+ "learning_rate": 0.001,
2032
+ "loss": 1.9607,
2033
+ "step": 3340
2034
+ },
2035
+ {
2036
+ "epoch": 1.74,
2037
+ "learning_rate": 0.001,
2038
+ "loss": 1.956,
2039
+ "step": 3350
2040
+ },
2041
+ {
2042
+ "epoch": 1.74,
2043
+ "learning_rate": 0.001,
2044
+ "loss": 1.9358,
2045
+ "step": 3360
2046
+ },
2047
+ {
2048
+ "epoch": 1.75,
2049
+ "learning_rate": 0.001,
2050
+ "loss": 1.9138,
2051
+ "step": 3370
2052
+ },
2053
+ {
2054
+ "epoch": 1.75,
2055
+ "learning_rate": 0.001,
2056
+ "loss": 1.928,
2057
+ "step": 3380
2058
+ },
2059
+ {
2060
+ "epoch": 1.76,
2061
+ "learning_rate": 0.001,
2062
+ "loss": 1.9845,
2063
+ "step": 3390
2064
+ },
2065
+ {
2066
+ "epoch": 1.76,
2067
+ "learning_rate": 0.001,
2068
+ "loss": 1.9536,
2069
+ "step": 3400
2070
+ },
2071
+ {
2072
+ "epoch": 1.77,
2073
+ "learning_rate": 0.001,
2074
+ "loss": 1.9458,
2075
+ "step": 3410
2076
+ },
2077
+ {
2078
+ "epoch": 1.77,
2079
+ "learning_rate": 0.001,
2080
+ "loss": 1.9565,
2081
+ "step": 3420
2082
+ },
2083
+ {
2084
+ "epoch": 1.78,
2085
+ "learning_rate": 0.001,
2086
+ "loss": 1.935,
2087
+ "step": 3430
2088
+ },
2089
+ {
2090
+ "epoch": 1.79,
2091
+ "learning_rate": 0.001,
2092
+ "loss": 1.9366,
2093
+ "step": 3440
2094
+ },
2095
+ {
2096
+ "epoch": 1.79,
2097
+ "learning_rate": 0.001,
2098
+ "loss": 1.931,
2099
+ "step": 3450
2100
+ },
2101
+ {
2102
+ "epoch": 1.8,
2103
+ "learning_rate": 0.001,
2104
+ "loss": 1.9534,
2105
+ "step": 3460
2106
+ },
2107
+ {
2108
+ "epoch": 1.8,
2109
+ "learning_rate": 0.001,
2110
+ "loss": 1.949,
2111
+ "step": 3470
2112
+ },
2113
+ {
2114
+ "epoch": 1.81,
2115
+ "learning_rate": 0.001,
2116
+ "loss": 1.9455,
2117
+ "step": 3480
2118
+ },
2119
+ {
2120
+ "epoch": 1.81,
2121
+ "learning_rate": 0.001,
2122
+ "loss": 1.9437,
2123
+ "step": 3490
2124
+ },
2125
+ {
2126
+ "epoch": 1.82,
2127
+ "learning_rate": 0.001,
2128
+ "loss": 1.9459,
2129
+ "step": 3500
2130
+ },
2131
+ {
2132
+ "epoch": 1.82,
2133
+ "learning_rate": 0.001,
2134
+ "loss": 1.9354,
2135
+ "step": 3510
2136
+ },
2137
+ {
2138
+ "epoch": 1.83,
2139
+ "learning_rate": 0.001,
2140
+ "loss": 1.9056,
2141
+ "step": 3520
2142
+ },
2143
+ {
2144
+ "epoch": 1.83,
2145
+ "learning_rate": 0.001,
2146
+ "loss": 1.9871,
2147
+ "step": 3530
2148
+ },
2149
+ {
2150
+ "epoch": 1.84,
2151
+ "learning_rate": 0.001,
2152
+ "loss": 1.9884,
2153
+ "step": 3540
2154
+ },
2155
+ {
2156
+ "epoch": 1.84,
2157
+ "learning_rate": 0.001,
2158
+ "loss": 1.944,
2159
+ "step": 3550
2160
+ },
2161
+ {
2162
+ "epoch": 1.85,
2163
+ "learning_rate": 0.001,
2164
+ "loss": 1.9365,
2165
+ "step": 3560
2166
+ },
2167
+ {
2168
+ "epoch": 1.85,
2169
+ "learning_rate": 0.001,
2170
+ "loss": 1.9536,
2171
+ "step": 3570
2172
+ },
2173
+ {
2174
+ "epoch": 1.86,
2175
+ "learning_rate": 0.001,
2176
+ "loss": 1.9269,
2177
+ "step": 3580
2178
+ },
2179
+ {
2180
+ "epoch": 1.86,
2181
+ "learning_rate": 0.001,
2182
+ "loss": 1.9259,
2183
+ "step": 3590
2184
+ },
2185
+ {
2186
+ "epoch": 1.87,
2187
+ "learning_rate": 0.001,
2188
+ "loss": 1.9648,
2189
+ "step": 3600
2190
+ },
2191
+ {
2192
+ "epoch": 1.87,
2193
+ "learning_rate": 0.001,
2194
+ "loss": 1.9642,
2195
+ "step": 3610
2196
+ },
2197
+ {
2198
+ "epoch": 1.88,
2199
+ "learning_rate": 0.001,
2200
+ "loss": 1.9677,
2201
+ "step": 3620
2202
+ },
2203
+ {
2204
+ "epoch": 1.88,
2205
+ "learning_rate": 0.001,
2206
+ "loss": 1.9543,
2207
+ "step": 3630
2208
+ },
2209
+ {
2210
+ "epoch": 1.89,
2211
+ "learning_rate": 0.001,
2212
+ "loss": 1.9517,
2213
+ "step": 3640
2214
+ },
2215
+ {
2216
+ "epoch": 1.89,
2217
+ "learning_rate": 0.001,
2218
+ "loss": 1.9523,
2219
+ "step": 3650
2220
+ },
2221
+ {
2222
+ "epoch": 1.9,
2223
+ "learning_rate": 0.001,
2224
+ "loss": 1.9372,
2225
+ "step": 3660
2226
+ },
2227
+ {
2228
+ "epoch": 1.9,
2229
+ "learning_rate": 0.001,
2230
+ "loss": 1.9453,
2231
+ "step": 3670
2232
+ },
2233
+ {
2234
+ "epoch": 1.91,
2235
+ "learning_rate": 0.001,
2236
+ "loss": 1.9463,
2237
+ "step": 3680
2238
+ },
2239
+ {
2240
+ "epoch": 1.91,
2241
+ "learning_rate": 0.001,
2242
+ "loss": 1.9392,
2243
+ "step": 3690
2244
+ },
2245
+ {
2246
+ "epoch": 1.92,
2247
+ "learning_rate": 0.001,
2248
+ "loss": 1.9467,
2249
+ "step": 3700
2250
+ },
2251
+ {
2252
+ "epoch": 1.93,
2253
+ "learning_rate": 0.001,
2254
+ "loss": 1.9575,
2255
+ "step": 3710
2256
+ },
2257
+ {
2258
+ "epoch": 1.93,
2259
+ "learning_rate": 0.001,
2260
+ "loss": 1.9441,
2261
+ "step": 3720
2262
+ },
2263
+ {
2264
+ "epoch": 1.94,
2265
+ "learning_rate": 0.001,
2266
+ "loss": 1.9534,
2267
+ "step": 3730
2268
+ },
2269
+ {
2270
+ "epoch": 1.94,
2271
+ "learning_rate": 0.001,
2272
+ "loss": 1.9484,
2273
+ "step": 3740
2274
+ },
2275
+ {
2276
+ "epoch": 1.95,
2277
+ "learning_rate": 0.001,
2278
+ "loss": 1.9277,
2279
+ "step": 3750
2280
+ },
2281
+ {
2282
+ "epoch": 1.95,
2283
+ "learning_rate": 0.001,
2284
+ "loss": 1.9326,
2285
+ "step": 3760
2286
+ },
2287
+ {
2288
+ "epoch": 1.96,
2289
+ "learning_rate": 0.001,
2290
+ "loss": 1.9501,
2291
+ "step": 3770
2292
+ },
2293
+ {
2294
+ "epoch": 1.96,
2295
+ "learning_rate": 0.001,
2296
+ "loss": 1.9568,
2297
+ "step": 3780
2298
+ },
2299
+ {
2300
+ "epoch": 1.97,
2301
+ "learning_rate": 0.001,
2302
+ "loss": 1.9535,
2303
+ "step": 3790
2304
+ },
2305
+ {
2306
+ "epoch": 1.97,
2307
+ "learning_rate": 0.001,
2308
+ "loss": 1.9523,
2309
+ "step": 3800
2310
+ },
2311
+ {
2312
+ "epoch": 1.98,
2313
+ "learning_rate": 0.001,
2314
+ "loss": 1.943,
2315
+ "step": 3810
2316
+ },
2317
+ {
2318
+ "epoch": 1.98,
2319
+ "learning_rate": 0.001,
2320
+ "loss": 1.9607,
2321
+ "step": 3820
2322
+ },
2323
+ {
2324
+ "epoch": 1.99,
2325
+ "learning_rate": 0.001,
2326
+ "loss": 1.9499,
2327
+ "step": 3830
2328
+ },
2329
+ {
2330
+ "epoch": 1.99,
2331
+ "learning_rate": 0.001,
2332
+ "loss": 1.9524,
2333
+ "step": 3840
2334
+ },
2335
+ {
2336
+ "epoch": 2.0,
2337
+ "learning_rate": 0.001,
2338
+ "loss": 1.951,
2339
+ "step": 3850
2340
+ },
2341
+ {
2342
+ "epoch": 2.0,
2343
+ "eval_accuracy": {
2344
+ "accuracy": 0.1556420233463035
2345
+ },
2346
+ "eval_f1": {
2347
+ "f1": 0.03848003848003848
2348
+ },
2349
+ "eval_loss": 1.9446961879730225,
2350
+ "eval_precision": {
2351
+ "precision": 0.022234574763757644
2352
+ },
2353
+ "eval_recall": {
2354
+ "recall": 0.14285714285714285
2355
+ },
2356
+ "eval_runtime": 78.0133,
2357
+ "eval_samples_per_second": 49.415,
2358
+ "eval_steps_per_second": 24.714,
2359
+ "step": 3854
2360
+ },
2361
+ {
2362
+ "epoch": 2.0,
2363
+ "learning_rate": 0.001,
2364
+ "loss": 1.9442,
2365
+ "step": 3860
2366
+ },
2367
+ {
2368
+ "epoch": 2.01,
2369
+ "learning_rate": 0.001,
2370
+ "loss": 1.9355,
2371
+ "step": 3870
2372
+ },
2373
+ {
2374
+ "epoch": 2.01,
2375
+ "learning_rate": 0.001,
2376
+ "loss": 1.9457,
2377
+ "step": 3880
2378
+ },
2379
+ {
2380
+ "epoch": 2.02,
2381
+ "learning_rate": 0.001,
2382
+ "loss": 1.9509,
2383
+ "step": 3890
2384
+ },
2385
+ {
2386
+ "epoch": 2.02,
2387
+ "learning_rate": 0.001,
2388
+ "loss": 1.9332,
2389
+ "step": 3900
2390
+ },
2391
+ {
2392
+ "epoch": 2.03,
2393
+ "learning_rate": 0.001,
2394
+ "loss": 1.9269,
2395
+ "step": 3910
2396
+ },
2397
+ {
2398
+ "epoch": 2.03,
2399
+ "learning_rate": 0.001,
2400
+ "loss": 1.9283,
2401
+ "step": 3920
2402
+ },
2403
+ {
2404
+ "epoch": 2.04,
2405
+ "learning_rate": 0.001,
2406
+ "loss": 1.9559,
2407
+ "step": 3930
2408
+ },
2409
+ {
2410
+ "epoch": 2.04,
2411
+ "learning_rate": 0.001,
2412
+ "loss": 1.9414,
2413
+ "step": 3940
2414
+ },
2415
+ {
2416
+ "epoch": 2.05,
2417
+ "learning_rate": 0.001,
2418
+ "loss": 1.9505,
2419
+ "step": 3950
2420
+ },
2421
+ {
2422
+ "epoch": 2.06,
2423
+ "learning_rate": 0.001,
2424
+ "loss": 1.9506,
2425
+ "step": 3960
2426
+ },
2427
+ {
2428
+ "epoch": 2.06,
2429
+ "learning_rate": 0.001,
2430
+ "loss": 1.9453,
2431
+ "step": 3970
2432
+ },
2433
+ {
2434
+ "epoch": 2.07,
2435
+ "learning_rate": 0.001,
2436
+ "loss": 1.9479,
2437
+ "step": 3980
2438
+ },
2439
+ {
2440
+ "epoch": 2.07,
2441
+ "learning_rate": 0.001,
2442
+ "loss": 1.9362,
2443
+ "step": 3990
2444
+ },
2445
+ {
2446
+ "epoch": 2.08,
2447
+ "learning_rate": 0.001,
2448
+ "loss": 1.949,
2449
+ "step": 4000
2450
+ },
2451
+ {
2452
+ "epoch": 2.08,
2453
+ "learning_rate": 0.001,
2454
+ "loss": 1.9581,
2455
+ "step": 4010
2456
+ },
2457
+ {
2458
+ "epoch": 2.09,
2459
+ "learning_rate": 0.001,
2460
+ "loss": 1.9388,
2461
+ "step": 4020
2462
+ },
2463
+ {
2464
+ "epoch": 2.09,
2465
+ "learning_rate": 0.001,
2466
+ "loss": 1.935,
2467
+ "step": 4030
2468
+ },
2469
+ {
2470
+ "epoch": 2.1,
2471
+ "learning_rate": 0.001,
2472
+ "loss": 1.9466,
2473
+ "step": 4040
2474
+ },
2475
+ {
2476
+ "epoch": 2.1,
2477
+ "learning_rate": 0.001,
2478
+ "loss": 1.9417,
2479
+ "step": 4050
2480
+ },
2481
+ {
2482
+ "epoch": 2.11,
2483
+ "learning_rate": 0.001,
2484
+ "loss": 1.9401,
2485
+ "step": 4060
2486
+ },
2487
+ {
2488
+ "epoch": 2.11,
2489
+ "learning_rate": 0.001,
2490
+ "loss": 1.9596,
2491
+ "step": 4070
2492
+ },
2493
+ {
2494
+ "epoch": 2.12,
2495
+ "learning_rate": 0.001,
2496
+ "loss": 1.9571,
2497
+ "step": 4080
2498
+ },
2499
+ {
2500
+ "epoch": 2.12,
2501
+ "learning_rate": 0.001,
2502
+ "loss": 1.9452,
2503
+ "step": 4090
2504
+ },
2505
+ {
2506
+ "epoch": 2.13,
2507
+ "learning_rate": 0.001,
2508
+ "loss": 1.9431,
2509
+ "step": 4100
2510
+ },
2511
+ {
2512
+ "epoch": 2.13,
2513
+ "learning_rate": 0.001,
2514
+ "loss": 1.942,
2515
+ "step": 4110
2516
+ },
2517
+ {
2518
+ "epoch": 2.14,
2519
+ "learning_rate": 0.001,
2520
+ "loss": 1.9499,
2521
+ "step": 4120
2522
+ },
2523
+ {
2524
+ "epoch": 2.14,
2525
+ "learning_rate": 0.001,
2526
+ "loss": 1.94,
2527
+ "step": 4130
2528
+ },
2529
+ {
2530
+ "epoch": 2.15,
2531
+ "learning_rate": 0.001,
2532
+ "loss": 1.9363,
2533
+ "step": 4140
2534
+ },
2535
+ {
2536
+ "epoch": 2.15,
2537
+ "learning_rate": 0.001,
2538
+ "loss": 1.9478,
2539
+ "step": 4150
2540
+ },
2541
+ {
2542
+ "epoch": 2.16,
2543
+ "learning_rate": 0.001,
2544
+ "loss": 1.9543,
2545
+ "step": 4160
2546
+ },
2547
+ {
2548
+ "epoch": 2.16,
2549
+ "learning_rate": 0.001,
2550
+ "loss": 1.9266,
2551
+ "step": 4170
2552
+ },
2553
+ {
2554
+ "epoch": 2.17,
2555
+ "learning_rate": 0.001,
2556
+ "loss": 1.9416,
2557
+ "step": 4180
2558
+ },
2559
+ {
2560
+ "epoch": 2.17,
2561
+ "learning_rate": 0.001,
2562
+ "loss": 1.9441,
2563
+ "step": 4190
2564
+ },
2565
+ {
2566
+ "epoch": 2.18,
2567
+ "learning_rate": 0.001,
2568
+ "loss": 1.9429,
2569
+ "step": 4200
2570
+ },
2571
+ {
2572
+ "epoch": 2.18,
2573
+ "learning_rate": 0.001,
2574
+ "loss": 1.9465,
2575
+ "step": 4210
2576
+ },
2577
+ {
2578
+ "epoch": 2.19,
2579
+ "learning_rate": 0.001,
2580
+ "loss": 1.952,
2581
+ "step": 4220
2582
+ },
2583
+ {
2584
+ "epoch": 2.2,
2585
+ "learning_rate": 0.001,
2586
+ "loss": 1.9493,
2587
+ "step": 4230
2588
+ },
2589
+ {
2590
+ "epoch": 2.2,
2591
+ "learning_rate": 0.001,
2592
+ "loss": 1.9393,
2593
+ "step": 4240
2594
+ },
2595
+ {
2596
+ "epoch": 2.21,
2597
+ "learning_rate": 0.001,
2598
+ "loss": 1.9531,
2599
+ "step": 4250
2600
+ },
2601
+ {
2602
+ "epoch": 2.21,
2603
+ "learning_rate": 0.001,
2604
+ "loss": 1.9412,
2605
+ "step": 4260
2606
+ },
2607
+ {
2608
+ "epoch": 2.22,
2609
+ "learning_rate": 0.001,
2610
+ "loss": 1.9493,
2611
+ "step": 4270
2612
+ },
2613
+ {
2614
+ "epoch": 2.22,
2615
+ "learning_rate": 0.001,
2616
+ "loss": 1.944,
2617
+ "step": 4280
2618
+ },
2619
+ {
2620
+ "epoch": 2.23,
2621
+ "learning_rate": 0.001,
2622
+ "loss": 1.9572,
2623
+ "step": 4290
2624
+ },
2625
+ {
2626
+ "epoch": 2.23,
2627
+ "learning_rate": 0.001,
2628
+ "loss": 1.9311,
2629
+ "step": 4300
2630
+ },
2631
+ {
2632
+ "epoch": 2.24,
2633
+ "learning_rate": 0.001,
2634
+ "loss": 1.9446,
2635
+ "step": 4310
2636
+ },
2637
+ {
2638
+ "epoch": 2.24,
2639
+ "learning_rate": 0.001,
2640
+ "loss": 1.9557,
2641
+ "step": 4320
2642
+ },
2643
+ {
2644
+ "epoch": 2.25,
2645
+ "learning_rate": 0.001,
2646
+ "loss": 1.9396,
2647
+ "step": 4330
2648
+ },
2649
+ {
2650
+ "epoch": 2.25,
2651
+ "learning_rate": 0.001,
2652
+ "loss": 1.9253,
2653
+ "step": 4340
2654
+ },
2655
+ {
2656
+ "epoch": 2.26,
2657
+ "learning_rate": 0.001,
2658
+ "loss": 1.9697,
2659
+ "step": 4350
2660
+ },
2661
+ {
2662
+ "epoch": 2.26,
2663
+ "learning_rate": 0.001,
2664
+ "loss": 1.9489,
2665
+ "step": 4360
2666
+ },
2667
+ {
2668
+ "epoch": 2.27,
2669
+ "learning_rate": 0.001,
2670
+ "loss": 1.9489,
2671
+ "step": 4370
2672
+ },
2673
+ {
2674
+ "epoch": 2.27,
2675
+ "learning_rate": 0.001,
2676
+ "loss": 1.9589,
2677
+ "step": 4380
2678
+ },
2679
+ {
2680
+ "epoch": 2.28,
2681
+ "learning_rate": 0.001,
2682
+ "loss": 1.95,
2683
+ "step": 4390
2684
+ },
2685
+ {
2686
+ "epoch": 2.28,
2687
+ "learning_rate": 0.001,
2688
+ "loss": 1.9422,
2689
+ "step": 4400
2690
+ },
2691
+ {
2692
+ "epoch": 2.29,
2693
+ "learning_rate": 0.001,
2694
+ "loss": 1.958,
2695
+ "step": 4410
2696
+ },
2697
+ {
2698
+ "epoch": 2.29,
2699
+ "learning_rate": 0.001,
2700
+ "loss": 1.9288,
2701
+ "step": 4420
2702
+ },
2703
+ {
2704
+ "epoch": 2.3,
2705
+ "learning_rate": 0.001,
2706
+ "loss": 1.9225,
2707
+ "step": 4430
2708
+ },
2709
+ {
2710
+ "epoch": 2.3,
2711
+ "learning_rate": 0.001,
2712
+ "loss": 1.9194,
2713
+ "step": 4440
2714
+ },
2715
+ {
2716
+ "epoch": 2.31,
2717
+ "learning_rate": 0.001,
2718
+ "loss": 1.9816,
2719
+ "step": 4450
2720
+ },
2721
+ {
2722
+ "epoch": 2.31,
2723
+ "learning_rate": 0.001,
2724
+ "loss": 1.9352,
2725
+ "step": 4460
2726
+ },
2727
+ {
2728
+ "epoch": 2.32,
2729
+ "learning_rate": 0.001,
2730
+ "loss": 1.939,
2731
+ "step": 4470
2732
+ },
2733
+ {
2734
+ "epoch": 2.32,
2735
+ "learning_rate": 0.001,
2736
+ "loss": 1.9567,
2737
+ "step": 4480
2738
+ },
2739
+ {
2740
+ "epoch": 2.33,
2741
+ "learning_rate": 0.001,
2742
+ "loss": 1.955,
2743
+ "step": 4490
2744
+ },
2745
+ {
2746
+ "epoch": 2.34,
2747
+ "learning_rate": 0.001,
2748
+ "loss": 1.9545,
2749
+ "step": 4500
2750
+ },
2751
+ {
2752
+ "epoch": 2.34,
2753
+ "learning_rate": 0.001,
2754
+ "loss": 1.9386,
2755
+ "step": 4510
2756
+ },
2757
+ {
2758
+ "epoch": 2.35,
2759
+ "learning_rate": 0.001,
2760
+ "loss": 1.9513,
2761
+ "step": 4520
2762
+ },
2763
+ {
2764
+ "epoch": 2.35,
2765
+ "learning_rate": 0.001,
2766
+ "loss": 1.9269,
2767
+ "step": 4530
2768
+ },
2769
+ {
2770
+ "epoch": 2.36,
2771
+ "learning_rate": 0.001,
2772
+ "loss": 1.9262,
2773
+ "step": 4540
2774
+ },
2775
+ {
2776
+ "epoch": 2.36,
2777
+ "learning_rate": 0.001,
2778
+ "loss": 1.9299,
2779
+ "step": 4550
2780
+ },
2781
+ {
2782
+ "epoch": 2.37,
2783
+ "learning_rate": 0.001,
2784
+ "loss": 1.9358,
2785
+ "step": 4560
2786
+ },
2787
+ {
2788
+ "epoch": 2.37,
2789
+ "learning_rate": 0.001,
2790
+ "loss": 1.9701,
2791
+ "step": 4570
2792
+ },
2793
+ {
2794
+ "epoch": 2.38,
2795
+ "learning_rate": 0.001,
2796
+ "loss": 1.9508,
2797
+ "step": 4580
2798
+ },
2799
+ {
2800
+ "epoch": 2.38,
2801
+ "learning_rate": 0.001,
2802
+ "loss": 1.9402,
2803
+ "step": 4590
2804
+ },
2805
+ {
2806
+ "epoch": 2.39,
2807
+ "learning_rate": 0.001,
2808
+ "loss": 1.9614,
2809
+ "step": 4600
2810
+ },
2811
+ {
2812
+ "epoch": 2.39,
2813
+ "learning_rate": 0.001,
2814
+ "loss": 1.9515,
2815
+ "step": 4610
2816
+ },
2817
+ {
2818
+ "epoch": 2.4,
2819
+ "learning_rate": 0.001,
2820
+ "loss": 1.9401,
2821
+ "step": 4620
2822
+ },
2823
+ {
2824
+ "epoch": 2.4,
2825
+ "learning_rate": 0.001,
2826
+ "loss": 1.9388,
2827
+ "step": 4630
2828
+ },
2829
+ {
2830
+ "epoch": 2.41,
2831
+ "learning_rate": 0.001,
2832
+ "loss": 1.9659,
2833
+ "step": 4640
2834
+ },
2835
+ {
2836
+ "epoch": 2.41,
2837
+ "learning_rate": 0.001,
2838
+ "loss": 1.9382,
2839
+ "step": 4650
2840
+ },
2841
+ {
2842
+ "epoch": 2.42,
2843
+ "learning_rate": 0.001,
2844
+ "loss": 1.9402,
2845
+ "step": 4660
2846
+ },
2847
+ {
2848
+ "epoch": 2.42,
2849
+ "learning_rate": 0.001,
2850
+ "loss": 1.9567,
2851
+ "step": 4670
2852
+ },
2853
+ {
2854
+ "epoch": 2.43,
2855
+ "learning_rate": 0.001,
2856
+ "loss": 1.9468,
2857
+ "step": 4680
2858
+ },
2859
+ {
2860
+ "epoch": 2.43,
2861
+ "learning_rate": 0.001,
2862
+ "loss": 1.9426,
2863
+ "step": 4690
2864
+ },
2865
+ {
2866
+ "epoch": 2.44,
2867
+ "learning_rate": 0.001,
2868
+ "loss": 1.9659,
2869
+ "step": 4700
2870
+ },
2871
+ {
2872
+ "epoch": 2.44,
2873
+ "learning_rate": 0.001,
2874
+ "loss": 1.9491,
2875
+ "step": 4710
2876
+ },
2877
+ {
2878
+ "epoch": 2.45,
2879
+ "learning_rate": 0.001,
2880
+ "loss": 1.9372,
2881
+ "step": 4720
2882
+ },
2883
+ {
2884
+ "epoch": 2.45,
2885
+ "learning_rate": 0.001,
2886
+ "loss": 1.9482,
2887
+ "step": 4730
2888
+ },
2889
+ {
2890
+ "epoch": 2.46,
2891
+ "learning_rate": 0.001,
2892
+ "loss": 1.9546,
2893
+ "step": 4740
2894
+ },
2895
+ {
2896
+ "epoch": 2.46,
2897
+ "learning_rate": 0.001,
2898
+ "loss": 1.9404,
2899
+ "step": 4750
2900
+ },
2901
+ {
2902
+ "epoch": 2.47,
2903
+ "learning_rate": 0.001,
2904
+ "loss": 1.94,
2905
+ "step": 4760
2906
+ },
2907
+ {
2908
+ "epoch": 2.48,
2909
+ "learning_rate": 0.001,
2910
+ "loss": 1.9248,
2911
+ "step": 4770
2912
+ },
2913
+ {
2914
+ "epoch": 2.48,
2915
+ "learning_rate": 0.001,
2916
+ "loss": 1.9324,
2917
+ "step": 4780
2918
+ },
2919
+ {
2920
+ "epoch": 2.49,
2921
+ "learning_rate": 0.001,
2922
+ "loss": 1.9833,
2923
+ "step": 4790
2924
+ },
2925
+ {
2926
+ "epoch": 2.49,
2927
+ "learning_rate": 0.001,
2928
+ "loss": 1.9301,
2929
+ "step": 4800
2930
+ },
2931
+ {
2932
+ "epoch": 2.5,
2933
+ "learning_rate": 0.001,
2934
+ "loss": 1.939,
2935
+ "step": 4810
2936
+ },
2937
+ {
2938
+ "epoch": 2.5,
2939
+ "learning_rate": 0.001,
2940
+ "loss": 1.9267,
2941
+ "step": 4820
2942
+ },
2943
+ {
2944
+ "epoch": 2.51,
2945
+ "learning_rate": 0.001,
2946
+ "loss": 1.9304,
2947
+ "step": 4830
2948
+ },
2949
+ {
2950
+ "epoch": 2.51,
2951
+ "learning_rate": 0.001,
2952
+ "loss": 1.9502,
2953
+ "step": 4840
2954
+ },
2955
+ {
2956
+ "epoch": 2.52,
2957
+ "learning_rate": 0.001,
2958
+ "loss": 1.9482,
2959
+ "step": 4850
2960
+ },
2961
+ {
2962
+ "epoch": 2.52,
2963
+ "learning_rate": 0.001,
2964
+ "loss": 1.9399,
2965
+ "step": 4860
2966
+ },
2967
+ {
2968
+ "epoch": 2.53,
2969
+ "learning_rate": 0.001,
2970
+ "loss": 1.9452,
2971
+ "step": 4870
2972
+ },
2973
+ {
2974
+ "epoch": 2.53,
2975
+ "learning_rate": 0.001,
2976
+ "loss": 1.92,
2977
+ "step": 4880
2978
+ },
2979
+ {
2980
+ "epoch": 2.54,
2981
+ "learning_rate": 0.001,
2982
+ "loss": 1.9378,
2983
+ "step": 4890
2984
+ },
2985
+ {
2986
+ "epoch": 2.54,
2987
+ "learning_rate": 0.001,
2988
+ "loss": 1.9481,
2989
+ "step": 4900
2990
+ },
2991
+ {
2992
+ "epoch": 2.55,
2993
+ "learning_rate": 0.001,
2994
+ "loss": 1.9496,
2995
+ "step": 4910
2996
+ },
2997
+ {
2998
+ "epoch": 2.55,
2999
+ "learning_rate": 0.001,
3000
+ "loss": 1.9494,
3001
+ "step": 4920
3002
+ },
3003
+ {
3004
+ "epoch": 2.56,
3005
+ "learning_rate": 0.001,
3006
+ "loss": 1.9501,
3007
+ "step": 4930
3008
+ },
3009
+ {
3010
+ "epoch": 2.56,
3011
+ "learning_rate": 0.001,
3012
+ "loss": 1.9553,
3013
+ "step": 4940
3014
+ },
3015
+ {
3016
+ "epoch": 2.57,
3017
+ "learning_rate": 0.001,
3018
+ "loss": 1.9423,
3019
+ "step": 4950
3020
+ },
3021
+ {
3022
+ "epoch": 2.57,
3023
+ "learning_rate": 0.001,
3024
+ "loss": 1.951,
3025
+ "step": 4960
3026
+ },
3027
+ {
3028
+ "epoch": 2.58,
3029
+ "learning_rate": 0.001,
3030
+ "loss": 1.8918,
3031
+ "step": 4970
3032
+ },
3033
+ {
3034
+ "epoch": 2.58,
3035
+ "learning_rate": 0.001,
3036
+ "loss": 1.94,
3037
+ "step": 4980
3038
+ },
3039
+ {
3040
+ "epoch": 2.59,
3041
+ "learning_rate": 0.001,
3042
+ "loss": 1.9264,
3043
+ "step": 4990
3044
+ },
3045
+ {
3046
+ "epoch": 2.59,
3047
+ "learning_rate": 0.001,
3048
+ "loss": 1.9679,
3049
+ "step": 5000
3050
+ },
3051
+ {
3052
+ "epoch": 2.6,
3053
+ "learning_rate": 0.001,
3054
+ "loss": 1.9513,
3055
+ "step": 5010
3056
+ },
3057
+ {
3058
+ "epoch": 2.61,
3059
+ "learning_rate": 0.001,
3060
+ "loss": 1.9212,
3061
+ "step": 5020
3062
+ },
3063
+ {
3064
+ "epoch": 2.61,
3065
+ "learning_rate": 0.001,
3066
+ "loss": 1.929,
3067
+ "step": 5030
3068
+ },
3069
+ {
3070
+ "epoch": 2.62,
3071
+ "learning_rate": 0.001,
3072
+ "loss": 1.9478,
3073
+ "step": 5040
3074
+ },
3075
+ {
3076
+ "epoch": 2.62,
3077
+ "learning_rate": 0.001,
3078
+ "loss": 1.9475,
3079
+ "step": 5050
3080
+ },
3081
+ {
3082
+ "epoch": 2.63,
3083
+ "learning_rate": 0.001,
3084
+ "loss": 1.9432,
3085
+ "step": 5060
3086
+ },
3087
+ {
3088
+ "epoch": 2.63,
3089
+ "learning_rate": 0.001,
3090
+ "loss": 1.9235,
3091
+ "step": 5070
3092
+ },
3093
+ {
3094
+ "epoch": 2.64,
3095
+ "learning_rate": 0.001,
3096
+ "loss": 1.9589,
3097
+ "step": 5080
3098
+ },
3099
+ {
3100
+ "epoch": 2.64,
3101
+ "learning_rate": 0.001,
3102
+ "loss": 1.9806,
3103
+ "step": 5090
3104
+ },
3105
+ {
3106
+ "epoch": 2.65,
3107
+ "learning_rate": 0.001,
3108
+ "loss": 1.9347,
3109
+ "step": 5100
3110
+ },
3111
+ {
3112
+ "epoch": 2.65,
3113
+ "learning_rate": 0.001,
3114
+ "loss": 1.9781,
3115
+ "step": 5110
3116
+ },
3117
+ {
3118
+ "epoch": 2.66,
3119
+ "learning_rate": 0.001,
3120
+ "loss": 1.9489,
3121
+ "step": 5120
3122
+ },
3123
+ {
3124
+ "epoch": 2.66,
3125
+ "learning_rate": 0.001,
3126
+ "loss": 1.9359,
3127
+ "step": 5130
3128
+ },
3129
+ {
3130
+ "epoch": 2.67,
3131
+ "learning_rate": 0.001,
3132
+ "loss": 1.9579,
3133
+ "step": 5140
3134
+ },
3135
+ {
3136
+ "epoch": 2.67,
3137
+ "learning_rate": 0.001,
3138
+ "loss": 1.967,
3139
+ "step": 5150
3140
+ },
3141
+ {
3142
+ "epoch": 2.68,
3143
+ "learning_rate": 0.001,
3144
+ "loss": 1.9475,
3145
+ "step": 5160
3146
+ },
3147
+ {
3148
+ "epoch": 2.68,
3149
+ "learning_rate": 0.001,
3150
+ "loss": 1.9492,
3151
+ "step": 5170
3152
+ },
3153
+ {
3154
+ "epoch": 2.69,
3155
+ "learning_rate": 0.001,
3156
+ "loss": 1.9345,
3157
+ "step": 5180
3158
+ },
3159
+ {
3160
+ "epoch": 2.69,
3161
+ "learning_rate": 0.001,
3162
+ "loss": 1.9456,
3163
+ "step": 5190
3164
+ },
3165
+ {
3166
+ "epoch": 2.7,
3167
+ "learning_rate": 0.001,
3168
+ "loss": 1.9515,
3169
+ "step": 5200
3170
+ },
3171
+ {
3172
+ "epoch": 2.7,
3173
+ "learning_rate": 0.001,
3174
+ "loss": 1.96,
3175
+ "step": 5210
3176
+ },
3177
+ {
3178
+ "epoch": 2.71,
3179
+ "learning_rate": 0.001,
3180
+ "loss": 1.9477,
3181
+ "step": 5220
3182
+ },
3183
+ {
3184
+ "epoch": 2.71,
3185
+ "learning_rate": 0.001,
3186
+ "loss": 1.9319,
3187
+ "step": 5230
3188
+ },
3189
+ {
3190
+ "epoch": 2.72,
3191
+ "learning_rate": 0.001,
3192
+ "loss": 1.9564,
3193
+ "step": 5240
3194
+ },
3195
+ {
3196
+ "epoch": 2.72,
3197
+ "learning_rate": 0.001,
3198
+ "loss": 1.9389,
3199
+ "step": 5250
3200
+ },
3201
+ {
3202
+ "epoch": 2.73,
3203
+ "learning_rate": 0.001,
3204
+ "loss": 1.9474,
3205
+ "step": 5260
3206
+ },
3207
+ {
3208
+ "epoch": 2.73,
3209
+ "learning_rate": 0.001,
3210
+ "loss": 1.9454,
3211
+ "step": 5270
3212
+ },
3213
+ {
3214
+ "epoch": 2.74,
3215
+ "learning_rate": 0.001,
3216
+ "loss": 1.9403,
3217
+ "step": 5280
3218
+ },
3219
+ {
3220
+ "epoch": 2.75,
3221
+ "learning_rate": 0.001,
3222
+ "loss": 1.9624,
3223
+ "step": 5290
3224
+ },
3225
+ {
3226
+ "epoch": 2.75,
3227
+ "learning_rate": 0.001,
3228
+ "loss": 1.9514,
3229
+ "step": 5300
3230
+ },
3231
+ {
3232
+ "epoch": 2.76,
3233
+ "learning_rate": 0.001,
3234
+ "loss": 1.9217,
3235
+ "step": 5310
3236
+ },
3237
+ {
3238
+ "epoch": 2.76,
3239
+ "learning_rate": 0.001,
3240
+ "loss": 1.9347,
3241
+ "step": 5320
3242
+ },
3243
+ {
3244
+ "epoch": 2.77,
3245
+ "learning_rate": 0.001,
3246
+ "loss": 1.9502,
3247
+ "step": 5330
3248
+ },
3249
+ {
3250
+ "epoch": 2.77,
3251
+ "learning_rate": 0.001,
3252
+ "loss": 1.99,
3253
+ "step": 5340
3254
+ },
3255
+ {
3256
+ "epoch": 2.78,
3257
+ "learning_rate": 0.001,
3258
+ "loss": 1.9212,
3259
+ "step": 5350
3260
+ },
3261
+ {
3262
+ "epoch": 2.78,
3263
+ "learning_rate": 0.001,
3264
+ "loss": 1.9377,
3265
+ "step": 5360
3266
+ },
3267
+ {
3268
+ "epoch": 2.79,
3269
+ "learning_rate": 0.001,
3270
+ "loss": 1.9435,
3271
+ "step": 5370
3272
+ },
3273
+ {
3274
+ "epoch": 2.79,
3275
+ "learning_rate": 0.001,
3276
+ "loss": 1.9398,
3277
+ "step": 5380
3278
+ },
3279
+ {
3280
+ "epoch": 2.8,
3281
+ "learning_rate": 0.001,
3282
+ "loss": 1.9381,
3283
+ "step": 5390
3284
+ },
3285
+ {
3286
+ "epoch": 2.8,
3287
+ "learning_rate": 0.001,
3288
+ "loss": 1.9239,
3289
+ "step": 5400
3290
+ },
3291
+ {
3292
+ "epoch": 2.81,
3293
+ "learning_rate": 0.001,
3294
+ "loss": 1.9821,
3295
+ "step": 5410
3296
+ },
3297
+ {
3298
+ "epoch": 2.81,
3299
+ "learning_rate": 0.001,
3300
+ "loss": 1.934,
3301
+ "step": 5420
3302
+ },
3303
+ {
3304
+ "epoch": 2.82,
3305
+ "learning_rate": 0.001,
3306
+ "loss": 1.9505,
3307
+ "step": 5430
3308
+ },
3309
+ {
3310
+ "epoch": 2.82,
3311
+ "learning_rate": 0.001,
3312
+ "loss": 1.9296,
3313
+ "step": 5440
3314
+ },
3315
+ {
3316
+ "epoch": 2.83,
3317
+ "learning_rate": 0.001,
3318
+ "loss": 1.9525,
3319
+ "step": 5450
3320
+ },
3321
+ {
3322
+ "epoch": 2.83,
3323
+ "learning_rate": 0.001,
3324
+ "loss": 1.9197,
3325
+ "step": 5460
3326
+ },
3327
+ {
3328
+ "epoch": 2.84,
3329
+ "learning_rate": 0.001,
3330
+ "loss": 1.9298,
3331
+ "step": 5470
3332
+ },
3333
+ {
3334
+ "epoch": 2.84,
3335
+ "learning_rate": 0.001,
3336
+ "loss": 1.9653,
3337
+ "step": 5480
3338
+ },
3339
+ {
3340
+ "epoch": 2.85,
3341
+ "learning_rate": 0.001,
3342
+ "loss": 1.947,
3343
+ "step": 5490
3344
+ },
3345
+ {
3346
+ "epoch": 2.85,
3347
+ "learning_rate": 0.001,
3348
+ "loss": 1.9438,
3349
+ "step": 5500
3350
+ },
3351
+ {
3352
+ "epoch": 2.86,
3353
+ "learning_rate": 0.001,
3354
+ "loss": 1.9581,
3355
+ "step": 5510
3356
+ },
3357
+ {
3358
+ "epoch": 2.86,
3359
+ "learning_rate": 0.001,
3360
+ "loss": 1.9582,
3361
+ "step": 5520
3362
+ },
3363
+ {
3364
+ "epoch": 2.87,
3365
+ "learning_rate": 0.001,
3366
+ "loss": 1.9202,
3367
+ "step": 5530
3368
+ },
3369
+ {
3370
+ "epoch": 2.87,
3371
+ "learning_rate": 0.001,
3372
+ "loss": 1.9561,
3373
+ "step": 5540
3374
+ },
3375
+ {
3376
+ "epoch": 2.88,
3377
+ "learning_rate": 0.001,
3378
+ "loss": 1.939,
3379
+ "step": 5550
3380
+ },
3381
+ {
3382
+ "epoch": 2.89,
3383
+ "learning_rate": 0.001,
3384
+ "loss": 1.9475,
3385
+ "step": 5560
3386
+ },
3387
+ {
3388
+ "epoch": 2.89,
3389
+ "learning_rate": 0.001,
3390
+ "loss": 1.9328,
3391
+ "step": 5570
3392
+ },
3393
+ {
3394
+ "epoch": 2.9,
3395
+ "learning_rate": 0.001,
3396
+ "loss": 1.9546,
3397
+ "step": 5580
3398
+ },
3399
+ {
3400
+ "epoch": 2.9,
3401
+ "learning_rate": 0.001,
3402
+ "loss": 1.9488,
3403
+ "step": 5590
3404
+ },
3405
+ {
3406
+ "epoch": 2.91,
3407
+ "learning_rate": 0.001,
3408
+ "loss": 1.9404,
3409
+ "step": 5600
3410
+ },
3411
+ {
3412
+ "epoch": 2.91,
3413
+ "learning_rate": 0.001,
3414
+ "loss": 1.959,
3415
+ "step": 5610
3416
+ },
3417
+ {
3418
+ "epoch": 2.92,
3419
+ "learning_rate": 0.001,
3420
+ "loss": 1.9649,
3421
+ "step": 5620
3422
+ },
3423
+ {
3424
+ "epoch": 2.92,
3425
+ "learning_rate": 0.001,
3426
+ "loss": 1.9498,
3427
+ "step": 5630
3428
+ },
3429
+ {
3430
+ "epoch": 2.93,
3431
+ "learning_rate": 0.001,
3432
+ "loss": 1.9353,
3433
+ "step": 5640
3434
+ },
3435
+ {
3436
+ "epoch": 2.93,
3437
+ "learning_rate": 0.001,
3438
+ "loss": 1.9521,
3439
+ "step": 5650
3440
+ },
3441
+ {
3442
+ "epoch": 2.94,
3443
+ "learning_rate": 0.001,
3444
+ "loss": 1.9564,
3445
+ "step": 5660
3446
+ },
3447
+ {
3448
+ "epoch": 2.94,
3449
+ "learning_rate": 0.001,
3450
+ "loss": 1.949,
3451
+ "step": 5670
3452
+ },
3453
+ {
3454
+ "epoch": 2.95,
3455
+ "learning_rate": 0.001,
3456
+ "loss": 1.9463,
3457
+ "step": 5680
3458
+ },
3459
+ {
3460
+ "epoch": 2.95,
3461
+ "learning_rate": 0.001,
3462
+ "loss": 1.95,
3463
+ "step": 5690
3464
+ },
3465
+ {
3466
+ "epoch": 2.96,
3467
+ "learning_rate": 0.001,
3468
+ "loss": 1.9389,
3469
+ "step": 5700
3470
+ },
3471
+ {
3472
+ "epoch": 2.96,
3473
+ "learning_rate": 0.001,
3474
+ "loss": 1.9575,
3475
+ "step": 5710
3476
+ },
3477
+ {
3478
+ "epoch": 2.97,
3479
+ "learning_rate": 0.001,
3480
+ "loss": 1.9573,
3481
+ "step": 5720
3482
+ },
3483
+ {
3484
+ "epoch": 2.97,
3485
+ "learning_rate": 0.001,
3486
+ "loss": 1.9435,
3487
+ "step": 5730
3488
+ },
3489
+ {
3490
+ "epoch": 2.98,
3491
+ "learning_rate": 0.001,
3492
+ "loss": 1.9379,
3493
+ "step": 5740
3494
+ },
3495
+ {
3496
+ "epoch": 2.98,
3497
+ "learning_rate": 0.001,
3498
+ "loss": 1.9365,
3499
+ "step": 5750
3500
+ },
3501
+ {
3502
+ "epoch": 2.99,
3503
+ "learning_rate": 0.001,
3504
+ "loss": 1.9192,
3505
+ "step": 5760
3506
+ },
3507
+ {
3508
+ "epoch": 2.99,
3509
+ "learning_rate": 0.001,
3510
+ "loss": 1.9526,
3511
+ "step": 5770
3512
+ },
3513
+ {
3514
+ "epoch": 3.0,
3515
+ "learning_rate": 0.001,
3516
+ "loss": 1.9799,
3517
+ "step": 5780
3518
+ },
3519
+ {
3520
+ "epoch": 3.0,
3521
+ "eval_accuracy": {
3522
+ "accuracy": 0.13618677042801555
3523
+ },
3524
+ "eval_f1": {
3525
+ "f1": 0.03424657534246575
3526
+ },
3527
+ "eval_loss": 1.9497852325439453,
3528
+ "eval_precision": {
3529
+ "precision": 0.019455252918287935
3530
+ },
3531
+ "eval_recall": {
3532
+ "recall": 0.14285714285714285
3533
+ },
3534
+ "eval_runtime": 77.5559,
3535
+ "eval_samples_per_second": 49.706,
3536
+ "eval_steps_per_second": 24.859,
3537
+ "step": 5781
3538
+ }
3539
+ ],
3540
+ "logging_steps": 10,
3541
+ "max_steps": 57810,
3542
+ "num_input_tokens_seen": 0,
3543
+ "num_train_epochs": 30,
3544
+ "save_steps": 500,
3545
+ "total_flos": 1.0014896600277811e+18,
3546
+ "train_batch_size": 2,
3547
+ "trial_name": null,
3548
+ "trial_params": null
3549
+ }
tmp-checkpoint-5781/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:272c41f466cf3b84b824960f5e302e07abd54eebf988ff66ec17995a20320f97
3
+ size 4728