Commit
·
001da99
1
Parent(s):
7cb6c30
Upload folder using huggingface_hub (#1)
Browse files- Upload folder using huggingface_hub (6433aa441a34a8eb17e29cbe190218176639119f)
- all_results.json +8 -0
- checkpoint-1480/config.json +86 -0
- checkpoint-1480/optimizer.pt +3 -0
- checkpoint-1480/preprocessor_config.json +22 -0
- checkpoint-1480/pytorch_model.bin +3 -0
- checkpoint-1480/rng_state.pth +3 -0
- checkpoint-1480/scheduler.pt +3 -0
- checkpoint-1480/trainer_state.json +1237 -0
- checkpoint-1480/training_args.bin +3 -0
- checkpoint-1840/config.json +86 -0
- checkpoint-1840/optimizer.pt +3 -0
- checkpoint-1840/preprocessor_config.json +22 -0
- checkpoint-1840/pytorch_model.bin +3 -0
- checkpoint-1840/rng_state.pth +3 -0
- checkpoint-1840/scheduler.pt +3 -0
- checkpoint-1840/trainer_state.json +1534 -0
- checkpoint-1840/training_args.bin +3 -0
- config.json +86 -0
- preprocessor_config.json +22 -0
- pytorch_model.bin +3 -0
- train_results.json +8 -0
- trainer_state.json +1543 -0
- training_args.bin +3 -0
all_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"total_flos": 3.419773941089157e+18,
|
4 |
+
"train_loss": 0.33101742866894474,
|
5 |
+
"train_runtime": 1771.2395,
|
6 |
+
"train_samples_per_second": 24.909,
|
7 |
+
"train_steps_per_second": 1.039
|
8 |
+
}
|
checkpoint-1480/config.json
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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": "a",
|
13 |
+
"1": "b",
|
14 |
+
"10": "j",
|
15 |
+
"11": "k",
|
16 |
+
"12": "l",
|
17 |
+
"13": "m",
|
18 |
+
"14": "n",
|
19 |
+
"15": "o",
|
20 |
+
"16": "p",
|
21 |
+
"17": "period",
|
22 |
+
"18": "q",
|
23 |
+
"19": "question%20mark",
|
24 |
+
"2": "c",
|
25 |
+
"20": "r",
|
26 |
+
"21": "s",
|
27 |
+
"22": "t",
|
28 |
+
"23": "u",
|
29 |
+
"24": "v",
|
30 |
+
"25": "w",
|
31 |
+
"26": "x",
|
32 |
+
"27": "y",
|
33 |
+
"28": "z",
|
34 |
+
"3": "capital",
|
35 |
+
"4": "d",
|
36 |
+
"5": "e",
|
37 |
+
"6": "f",
|
38 |
+
"7": "g",
|
39 |
+
"8": "h",
|
40 |
+
"9": "i"
|
41 |
+
},
|
42 |
+
"image_size": 224,
|
43 |
+
"initializer_range": 0.02,
|
44 |
+
"intermediate_size": 3072,
|
45 |
+
"label2id": {
|
46 |
+
"a": "0",
|
47 |
+
"b": "1",
|
48 |
+
"c": "2",
|
49 |
+
"capital": "3",
|
50 |
+
"d": "4",
|
51 |
+
"e": "5",
|
52 |
+
"f": "6",
|
53 |
+
"g": "7",
|
54 |
+
"h": "8",
|
55 |
+
"i": "9",
|
56 |
+
"j": "10",
|
57 |
+
"k": "11",
|
58 |
+
"l": "12",
|
59 |
+
"m": "13",
|
60 |
+
"n": "14",
|
61 |
+
"o": "15",
|
62 |
+
"p": "16",
|
63 |
+
"period": "17",
|
64 |
+
"q": "18",
|
65 |
+
"question%20mark": "19",
|
66 |
+
"r": "20",
|
67 |
+
"s": "21",
|
68 |
+
"t": "22",
|
69 |
+
"u": "23",
|
70 |
+
"v": "24",
|
71 |
+
"w": "25",
|
72 |
+
"x": "26",
|
73 |
+
"y": "27",
|
74 |
+
"z": "28"
|
75 |
+
},
|
76 |
+
"layer_norm_eps": 1e-12,
|
77 |
+
"model_type": "vit",
|
78 |
+
"num_attention_heads": 12,
|
79 |
+
"num_channels": 3,
|
80 |
+
"num_hidden_layers": 12,
|
81 |
+
"patch_size": 16,
|
82 |
+
"problem_type": "single_label_classification",
|
83 |
+
"qkv_bias": true,
|
84 |
+
"torch_dtype": "float32",
|
85 |
+
"transformers_version": "4.30.2"
|
86 |
+
}
|
checkpoint-1480/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:59886c09c1d0edec7e6befd8910c482777935e25c3aeda028d2662c66786d606
|
3 |
+
size 686684933
|
checkpoint-1480/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": "ViTFeatureExtractor",
|
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 |
+
}
|
checkpoint-1480/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df31f9efd884fa58db6a3ad678a6079dd27b1b6e32f1cf200e7eb1ccc50f8620
|
3 |
+
size 343351725
|
checkpoint-1480/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a2c7741b65ea9eb8477d41d3b8e9abd9eff658c5093a9d462a289b5ebaf90d71
|
3 |
+
size 14575
|
checkpoint-1480/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3d38c128d017daf2ebc6b59d26743e5c488021504f208ff1244552944d067a1d
|
3 |
+
size 627
|
checkpoint-1480/trainer_state.json
ADDED
@@ -0,0 +1,1237 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.04137137532234192,
|
3 |
+
"best_model_checkpoint": "./vit-base-beans/checkpoint-1480",
|
4 |
+
"epoch": 16.08695652173913,
|
5 |
+
"global_step": 1480,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.11,
|
12 |
+
"learning_rate": 9.945652173913043e-05,
|
13 |
+
"loss": 3.3812,
|
14 |
+
"step": 10
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.22,
|
18 |
+
"learning_rate": 9.891304347826087e-05,
|
19 |
+
"loss": 3.3288,
|
20 |
+
"step": 20
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.33,
|
24 |
+
"learning_rate": 9.836956521739132e-05,
|
25 |
+
"loss": 3.3101,
|
26 |
+
"step": 30
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.43,
|
30 |
+
"learning_rate": 9.782608695652174e-05,
|
31 |
+
"loss": 3.2579,
|
32 |
+
"step": 40
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.43,
|
36 |
+
"eval_accuracy": 0.17882919005613473,
|
37 |
+
"eval_loss": 3.1847527027130127,
|
38 |
+
"eval_runtime": 13.4342,
|
39 |
+
"eval_samples_per_second": 92.823,
|
40 |
+
"eval_steps_per_second": 11.612,
|
41 |
+
"step": 40
|
42 |
+
},
|
43 |
+
{
|
44 |
+
"epoch": 0.54,
|
45 |
+
"learning_rate": 9.728260869565217e-05,
|
46 |
+
"loss": 3.1255,
|
47 |
+
"step": 50
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"epoch": 0.65,
|
51 |
+
"learning_rate": 9.673913043478261e-05,
|
52 |
+
"loss": 3.0162,
|
53 |
+
"step": 60
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 0.76,
|
57 |
+
"learning_rate": 9.619565217391306e-05,
|
58 |
+
"loss": 2.8706,
|
59 |
+
"step": 70
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"epoch": 0.87,
|
63 |
+
"learning_rate": 9.565217391304348e-05,
|
64 |
+
"loss": 2.7157,
|
65 |
+
"step": 80
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.87,
|
69 |
+
"eval_accuracy": 0.43785084202085006,
|
70 |
+
"eval_loss": 2.5922651290893555,
|
71 |
+
"eval_runtime": 13.9416,
|
72 |
+
"eval_samples_per_second": 89.445,
|
73 |
+
"eval_steps_per_second": 11.19,
|
74 |
+
"step": 80
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.98,
|
78 |
+
"learning_rate": 9.510869565217391e-05,
|
79 |
+
"loss": 2.5322,
|
80 |
+
"step": 90
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 1.09,
|
84 |
+
"learning_rate": 9.456521739130435e-05,
|
85 |
+
"loss": 2.3504,
|
86 |
+
"step": 100
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 1.2,
|
90 |
+
"learning_rate": 9.402173913043478e-05,
|
91 |
+
"loss": 2.1887,
|
92 |
+
"step": 110
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 1.3,
|
96 |
+
"learning_rate": 9.347826086956522e-05,
|
97 |
+
"loss": 2.0664,
|
98 |
+
"step": 120
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 1.3,
|
102 |
+
"eval_accuracy": 0.6696070569366479,
|
103 |
+
"eval_loss": 1.9748882055282593,
|
104 |
+
"eval_runtime": 13.4691,
|
105 |
+
"eval_samples_per_second": 92.582,
|
106 |
+
"eval_steps_per_second": 11.582,
|
107 |
+
"step": 120
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 1.41,
|
111 |
+
"learning_rate": 9.293478260869566e-05,
|
112 |
+
"loss": 1.8705,
|
113 |
+
"step": 130
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 1.52,
|
117 |
+
"learning_rate": 9.239130434782609e-05,
|
118 |
+
"loss": 1.688,
|
119 |
+
"step": 140
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"epoch": 1.63,
|
123 |
+
"learning_rate": 9.184782608695652e-05,
|
124 |
+
"loss": 1.5939,
|
125 |
+
"step": 150
|
126 |
+
},
|
127 |
+
{
|
128 |
+
"epoch": 1.74,
|
129 |
+
"learning_rate": 9.130434782608696e-05,
|
130 |
+
"loss": 1.4765,
|
131 |
+
"step": 160
|
132 |
+
},
|
133 |
+
{
|
134 |
+
"epoch": 1.74,
|
135 |
+
"eval_accuracy": 0.917401764234162,
|
136 |
+
"eval_loss": 1.3413872718811035,
|
137 |
+
"eval_runtime": 13.9419,
|
138 |
+
"eval_samples_per_second": 89.442,
|
139 |
+
"eval_steps_per_second": 11.189,
|
140 |
+
"step": 160
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 1.85,
|
144 |
+
"learning_rate": 9.07608695652174e-05,
|
145 |
+
"loss": 1.3014,
|
146 |
+
"step": 170
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 1.96,
|
150 |
+
"learning_rate": 9.021739130434783e-05,
|
151 |
+
"loss": 1.201,
|
152 |
+
"step": 180
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 2.07,
|
156 |
+
"learning_rate": 8.967391304347826e-05,
|
157 |
+
"loss": 1.0165,
|
158 |
+
"step": 190
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 2.17,
|
162 |
+
"learning_rate": 8.91304347826087e-05,
|
163 |
+
"loss": 0.965,
|
164 |
+
"step": 200
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 2.17,
|
168 |
+
"eval_accuracy": 0.9615076182838813,
|
169 |
+
"eval_loss": 0.9264132380485535,
|
170 |
+
"eval_runtime": 13.1542,
|
171 |
+
"eval_samples_per_second": 94.798,
|
172 |
+
"eval_steps_per_second": 11.859,
|
173 |
+
"step": 200
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 2.28,
|
177 |
+
"learning_rate": 8.858695652173914e-05,
|
178 |
+
"loss": 0.9096,
|
179 |
+
"step": 210
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 2.39,
|
183 |
+
"learning_rate": 8.804347826086957e-05,
|
184 |
+
"loss": 0.8527,
|
185 |
+
"step": 220
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 2.5,
|
189 |
+
"learning_rate": 8.75e-05,
|
190 |
+
"loss": 0.759,
|
191 |
+
"step": 230
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 2.61,
|
195 |
+
"learning_rate": 8.695652173913044e-05,
|
196 |
+
"loss": 0.7163,
|
197 |
+
"step": 240
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"epoch": 2.61,
|
201 |
+
"eval_accuracy": 0.9647153167602245,
|
202 |
+
"eval_loss": 0.6652109622955322,
|
203 |
+
"eval_runtime": 14.0687,
|
204 |
+
"eval_samples_per_second": 88.637,
|
205 |
+
"eval_steps_per_second": 11.088,
|
206 |
+
"step": 240
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 2.72,
|
210 |
+
"learning_rate": 8.641304347826087e-05,
|
211 |
+
"loss": 0.6403,
|
212 |
+
"step": 250
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 2.83,
|
216 |
+
"learning_rate": 8.586956521739131e-05,
|
217 |
+
"loss": 0.5857,
|
218 |
+
"step": 260
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 2.93,
|
222 |
+
"learning_rate": 8.532608695652174e-05,
|
223 |
+
"loss": 0.5406,
|
224 |
+
"step": 270
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 3.04,
|
228 |
+
"learning_rate": 8.478260869565218e-05,
|
229 |
+
"loss": 0.5061,
|
230 |
+
"step": 280
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 3.04,
|
234 |
+
"eval_accuracy": 0.9687249398556536,
|
235 |
+
"eval_loss": 0.5080122947692871,
|
236 |
+
"eval_runtime": 13.0785,
|
237 |
+
"eval_samples_per_second": 95.347,
|
238 |
+
"eval_steps_per_second": 11.928,
|
239 |
+
"step": 280
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 3.15,
|
243 |
+
"learning_rate": 8.423913043478261e-05,
|
244 |
+
"loss": 0.4622,
|
245 |
+
"step": 290
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"epoch": 3.26,
|
249 |
+
"learning_rate": 8.369565217391305e-05,
|
250 |
+
"loss": 0.4919,
|
251 |
+
"step": 300
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"epoch": 3.37,
|
255 |
+
"learning_rate": 8.315217391304349e-05,
|
256 |
+
"loss": 0.371,
|
257 |
+
"step": 310
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 3.48,
|
261 |
+
"learning_rate": 8.260869565217392e-05,
|
262 |
+
"loss": 0.3883,
|
263 |
+
"step": 320
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 3.48,
|
267 |
+
"eval_accuracy": 0.9759422614274258,
|
268 |
+
"eval_loss": 0.3574630916118622,
|
269 |
+
"eval_runtime": 13.9479,
|
270 |
+
"eval_samples_per_second": 89.404,
|
271 |
+
"eval_steps_per_second": 11.184,
|
272 |
+
"step": 320
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 3.59,
|
276 |
+
"learning_rate": 8.206521739130435e-05,
|
277 |
+
"loss": 0.3831,
|
278 |
+
"step": 330
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 3.7,
|
282 |
+
"learning_rate": 8.152173913043478e-05,
|
283 |
+
"loss": 0.3329,
|
284 |
+
"step": 340
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 3.8,
|
288 |
+
"learning_rate": 8.097826086956523e-05,
|
289 |
+
"loss": 0.3383,
|
290 |
+
"step": 350
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 3.91,
|
294 |
+
"learning_rate": 8.043478260869566e-05,
|
295 |
+
"loss": 0.3328,
|
296 |
+
"step": 360
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 3.91,
|
300 |
+
"eval_accuracy": 0.9839615076182838,
|
301 |
+
"eval_loss": 0.27629122138023376,
|
302 |
+
"eval_runtime": 13.7308,
|
303 |
+
"eval_samples_per_second": 90.818,
|
304 |
+
"eval_steps_per_second": 11.361,
|
305 |
+
"step": 360
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"epoch": 4.02,
|
309 |
+
"learning_rate": 7.989130434782609e-05,
|
310 |
+
"loss": 0.2387,
|
311 |
+
"step": 370
|
312 |
+
},
|
313 |
+
{
|
314 |
+
"epoch": 4.13,
|
315 |
+
"learning_rate": 7.934782608695653e-05,
|
316 |
+
"loss": 0.2509,
|
317 |
+
"step": 380
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 4.24,
|
321 |
+
"learning_rate": 7.880434782608696e-05,
|
322 |
+
"loss": 0.2259,
|
323 |
+
"step": 390
|
324 |
+
},
|
325 |
+
{
|
326 |
+
"epoch": 4.35,
|
327 |
+
"learning_rate": 7.82608695652174e-05,
|
328 |
+
"loss": 0.2049,
|
329 |
+
"step": 400
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 4.35,
|
333 |
+
"eval_accuracy": 0.9855653568564555,
|
334 |
+
"eval_loss": 0.2094665914773941,
|
335 |
+
"eval_runtime": 13.8999,
|
336 |
+
"eval_samples_per_second": 89.713,
|
337 |
+
"eval_steps_per_second": 11.223,
|
338 |
+
"step": 400
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 4.46,
|
342 |
+
"learning_rate": 7.771739130434783e-05,
|
343 |
+
"loss": 0.1979,
|
344 |
+
"step": 410
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 4.57,
|
348 |
+
"learning_rate": 7.717391304347827e-05,
|
349 |
+
"loss": 0.1703,
|
350 |
+
"step": 420
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 4.67,
|
354 |
+
"learning_rate": 7.66304347826087e-05,
|
355 |
+
"loss": 0.1771,
|
356 |
+
"step": 430
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 4.78,
|
360 |
+
"learning_rate": 7.608695652173914e-05,
|
361 |
+
"loss": 0.2078,
|
362 |
+
"step": 440
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 4.78,
|
366 |
+
"eval_accuracy": 0.9871692060946271,
|
367 |
+
"eval_loss": 0.19693857431411743,
|
368 |
+
"eval_runtime": 13.7091,
|
369 |
+
"eval_samples_per_second": 90.961,
|
370 |
+
"eval_steps_per_second": 11.379,
|
371 |
+
"step": 440
|
372 |
+
},
|
373 |
+
{
|
374 |
+
"epoch": 4.89,
|
375 |
+
"learning_rate": 7.554347826086957e-05,
|
376 |
+
"loss": 0.1564,
|
377 |
+
"step": 450
|
378 |
+
},
|
379 |
+
{
|
380 |
+
"epoch": 5.0,
|
381 |
+
"learning_rate": 7.500000000000001e-05,
|
382 |
+
"loss": 0.1512,
|
383 |
+
"step": 460
|
384 |
+
},
|
385 |
+
{
|
386 |
+
"epoch": 5.11,
|
387 |
+
"learning_rate": 7.445652173913044e-05,
|
388 |
+
"loss": 0.1339,
|
389 |
+
"step": 470
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 5.22,
|
393 |
+
"learning_rate": 7.391304347826086e-05,
|
394 |
+
"loss": 0.1447,
|
395 |
+
"step": 480
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 5.22,
|
399 |
+
"eval_accuracy": 0.9871692060946271,
|
400 |
+
"eval_loss": 0.14835722744464874,
|
401 |
+
"eval_runtime": 13.9008,
|
402 |
+
"eval_samples_per_second": 89.707,
|
403 |
+
"eval_steps_per_second": 11.222,
|
404 |
+
"step": 480
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 5.33,
|
408 |
+
"learning_rate": 7.336956521739132e-05,
|
409 |
+
"loss": 0.1207,
|
410 |
+
"step": 490
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 5.43,
|
414 |
+
"learning_rate": 7.282608695652175e-05,
|
415 |
+
"loss": 0.1294,
|
416 |
+
"step": 500
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 5.54,
|
420 |
+
"learning_rate": 7.228260869565217e-05,
|
421 |
+
"loss": 0.116,
|
422 |
+
"step": 510
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 5.65,
|
426 |
+
"learning_rate": 7.17391304347826e-05,
|
427 |
+
"loss": 0.1401,
|
428 |
+
"step": 520
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 5.65,
|
432 |
+
"eval_accuracy": 0.9839615076182838,
|
433 |
+
"eval_loss": 0.14811548590660095,
|
434 |
+
"eval_runtime": 13.5209,
|
435 |
+
"eval_samples_per_second": 92.228,
|
436 |
+
"eval_steps_per_second": 11.538,
|
437 |
+
"step": 520
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 5.76,
|
441 |
+
"learning_rate": 7.119565217391306e-05,
|
442 |
+
"loss": 0.0904,
|
443 |
+
"step": 530
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 5.87,
|
447 |
+
"learning_rate": 7.065217391304349e-05,
|
448 |
+
"loss": 0.1099,
|
449 |
+
"step": 540
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 5.98,
|
453 |
+
"learning_rate": 7.010869565217391e-05,
|
454 |
+
"loss": 0.1599,
|
455 |
+
"step": 550
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 6.09,
|
459 |
+
"learning_rate": 6.956521739130436e-05,
|
460 |
+
"loss": 0.1232,
|
461 |
+
"step": 560
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 6.09,
|
465 |
+
"eval_accuracy": 0.991980753809142,
|
466 |
+
"eval_loss": 0.11416751146316528,
|
467 |
+
"eval_runtime": 13.9361,
|
468 |
+
"eval_samples_per_second": 89.48,
|
469 |
+
"eval_steps_per_second": 11.194,
|
470 |
+
"step": 560
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 6.2,
|
474 |
+
"learning_rate": 6.902173913043478e-05,
|
475 |
+
"loss": 0.1381,
|
476 |
+
"step": 570
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 6.3,
|
480 |
+
"learning_rate": 6.847826086956522e-05,
|
481 |
+
"loss": 0.1001,
|
482 |
+
"step": 580
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 6.41,
|
486 |
+
"learning_rate": 6.793478260869565e-05,
|
487 |
+
"loss": 0.0823,
|
488 |
+
"step": 590
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 6.52,
|
492 |
+
"learning_rate": 6.73913043478261e-05,
|
493 |
+
"loss": 0.0725,
|
494 |
+
"step": 600
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 6.52,
|
498 |
+
"eval_accuracy": 0.9879711307137129,
|
499 |
+
"eval_loss": 0.10076911747455597,
|
500 |
+
"eval_runtime": 13.8114,
|
501 |
+
"eval_samples_per_second": 90.288,
|
502 |
+
"eval_steps_per_second": 11.295,
|
503 |
+
"step": 600
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"epoch": 6.63,
|
507 |
+
"learning_rate": 6.684782608695652e-05,
|
508 |
+
"loss": 0.0852,
|
509 |
+
"step": 610
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 6.74,
|
513 |
+
"learning_rate": 6.630434782608695e-05,
|
514 |
+
"loss": 0.0723,
|
515 |
+
"step": 620
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 6.85,
|
519 |
+
"learning_rate": 6.576086956521739e-05,
|
520 |
+
"loss": 0.0881,
|
521 |
+
"step": 630
|
522 |
+
},
|
523 |
+
{
|
524 |
+
"epoch": 6.96,
|
525 |
+
"learning_rate": 6.521739130434783e-05,
|
526 |
+
"loss": 0.0934,
|
527 |
+
"step": 640
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 6.96,
|
531 |
+
"eval_accuracy": 0.9895749799518845,
|
532 |
+
"eval_loss": 0.09398525953292847,
|
533 |
+
"eval_runtime": 14.2025,
|
534 |
+
"eval_samples_per_second": 87.801,
|
535 |
+
"eval_steps_per_second": 10.984,
|
536 |
+
"step": 640
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 7.07,
|
540 |
+
"learning_rate": 6.467391304347826e-05,
|
541 |
+
"loss": 0.0668,
|
542 |
+
"step": 650
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 7.17,
|
546 |
+
"learning_rate": 6.413043478260869e-05,
|
547 |
+
"loss": 0.0586,
|
548 |
+
"step": 660
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 7.28,
|
552 |
+
"learning_rate": 6.358695652173913e-05,
|
553 |
+
"loss": 0.0543,
|
554 |
+
"step": 670
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 7.39,
|
558 |
+
"learning_rate": 6.304347826086957e-05,
|
559 |
+
"loss": 0.053,
|
560 |
+
"step": 680
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 7.39,
|
564 |
+
"eval_accuracy": 0.9895749799518845,
|
565 |
+
"eval_loss": 0.08539092540740967,
|
566 |
+
"eval_runtime": 13.9817,
|
567 |
+
"eval_samples_per_second": 89.188,
|
568 |
+
"eval_steps_per_second": 11.157,
|
569 |
+
"step": 680
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 7.5,
|
573 |
+
"learning_rate": 6.25e-05,
|
574 |
+
"loss": 0.0514,
|
575 |
+
"step": 690
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"epoch": 7.61,
|
579 |
+
"learning_rate": 6.195652173913043e-05,
|
580 |
+
"loss": 0.0491,
|
581 |
+
"step": 700
|
582 |
+
},
|
583 |
+
{
|
584 |
+
"epoch": 7.72,
|
585 |
+
"learning_rate": 6.141304347826087e-05,
|
586 |
+
"loss": 0.0481,
|
587 |
+
"step": 710
|
588 |
+
},
|
589 |
+
{
|
590 |
+
"epoch": 7.83,
|
591 |
+
"learning_rate": 6.086956521739131e-05,
|
592 |
+
"loss": 0.0469,
|
593 |
+
"step": 720
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 7.83,
|
597 |
+
"eval_accuracy": 0.9903769045709703,
|
598 |
+
"eval_loss": 0.06862174719572067,
|
599 |
+
"eval_runtime": 14.4287,
|
600 |
+
"eval_samples_per_second": 86.425,
|
601 |
+
"eval_steps_per_second": 10.812,
|
602 |
+
"step": 720
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 7.93,
|
606 |
+
"learning_rate": 6.032608695652174e-05,
|
607 |
+
"loss": 0.0693,
|
608 |
+
"step": 730
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 8.04,
|
612 |
+
"learning_rate": 5.9782608695652175e-05,
|
613 |
+
"loss": 0.0664,
|
614 |
+
"step": 740
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 8.15,
|
618 |
+
"learning_rate": 5.923913043478261e-05,
|
619 |
+
"loss": 0.0502,
|
620 |
+
"step": 750
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 8.26,
|
624 |
+
"learning_rate": 5.869565217391305e-05,
|
625 |
+
"loss": 0.0429,
|
626 |
+
"step": 760
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 8.26,
|
630 |
+
"eval_accuracy": 0.9863672814755413,
|
631 |
+
"eval_loss": 0.0824466422200203,
|
632 |
+
"eval_runtime": 13.8977,
|
633 |
+
"eval_samples_per_second": 89.727,
|
634 |
+
"eval_steps_per_second": 11.225,
|
635 |
+
"step": 760
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"epoch": 8.37,
|
639 |
+
"learning_rate": 5.815217391304349e-05,
|
640 |
+
"loss": 0.0622,
|
641 |
+
"step": 770
|
642 |
+
},
|
643 |
+
{
|
644 |
+
"epoch": 8.48,
|
645 |
+
"learning_rate": 5.7608695652173915e-05,
|
646 |
+
"loss": 0.0394,
|
647 |
+
"step": 780
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 8.59,
|
651 |
+
"learning_rate": 5.706521739130435e-05,
|
652 |
+
"loss": 0.0375,
|
653 |
+
"step": 790
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 8.7,
|
657 |
+
"learning_rate": 5.652173913043478e-05,
|
658 |
+
"loss": 0.0371,
|
659 |
+
"step": 800
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"epoch": 8.7,
|
663 |
+
"eval_accuracy": 0.991980753809142,
|
664 |
+
"eval_loss": 0.07010400295257568,
|
665 |
+
"eval_runtime": 13.4894,
|
666 |
+
"eval_samples_per_second": 92.443,
|
667 |
+
"eval_steps_per_second": 11.565,
|
668 |
+
"step": 800
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 8.8,
|
672 |
+
"learning_rate": 5.5978260869565226e-05,
|
673 |
+
"loss": 0.036,
|
674 |
+
"step": 810
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 8.91,
|
678 |
+
"learning_rate": 5.5434782608695654e-05,
|
679 |
+
"loss": 0.0352,
|
680 |
+
"step": 820
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 9.02,
|
684 |
+
"learning_rate": 5.489130434782609e-05,
|
685 |
+
"loss": 0.0344,
|
686 |
+
"step": 830
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 9.13,
|
690 |
+
"learning_rate": 5.4347826086956524e-05,
|
691 |
+
"loss": 0.033,
|
692 |
+
"step": 840
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 9.13,
|
696 |
+
"eval_accuracy": 0.991980753809142,
|
697 |
+
"eval_loss": 0.06847481429576874,
|
698 |
+
"eval_runtime": 13.9465,
|
699 |
+
"eval_samples_per_second": 89.413,
|
700 |
+
"eval_steps_per_second": 11.186,
|
701 |
+
"step": 840
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 9.24,
|
705 |
+
"learning_rate": 5.380434782608695e-05,
|
706 |
+
"loss": 0.0327,
|
707 |
+
"step": 850
|
708 |
+
},
|
709 |
+
{
|
710 |
+
"epoch": 9.35,
|
711 |
+
"learning_rate": 5.32608695652174e-05,
|
712 |
+
"loss": 0.0318,
|
713 |
+
"step": 860
|
714 |
+
},
|
715 |
+
{
|
716 |
+
"epoch": 9.46,
|
717 |
+
"learning_rate": 5.271739130434783e-05,
|
718 |
+
"loss": 0.0315,
|
719 |
+
"step": 870
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 9.57,
|
723 |
+
"learning_rate": 5.217391304347826e-05,
|
724 |
+
"loss": 0.0308,
|
725 |
+
"step": 880
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"epoch": 9.57,
|
729 |
+
"eval_accuracy": 0.991980753809142,
|
730 |
+
"eval_loss": 0.06314855068922043,
|
731 |
+
"eval_runtime": 13.4895,
|
732 |
+
"eval_samples_per_second": 92.442,
|
733 |
+
"eval_steps_per_second": 11.565,
|
734 |
+
"step": 880
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 9.67,
|
738 |
+
"learning_rate": 5.163043478260869e-05,
|
739 |
+
"loss": 0.0502,
|
740 |
+
"step": 890
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 9.78,
|
744 |
+
"learning_rate": 5.108695652173914e-05,
|
745 |
+
"loss": 0.03,
|
746 |
+
"step": 900
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 9.89,
|
750 |
+
"learning_rate": 5.054347826086957e-05,
|
751 |
+
"loss": 0.0294,
|
752 |
+
"step": 910
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 10.0,
|
756 |
+
"learning_rate": 5e-05,
|
757 |
+
"loss": 0.0398,
|
758 |
+
"step": 920
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 10.0,
|
762 |
+
"eval_accuracy": 0.9927826784282278,
|
763 |
+
"eval_loss": 0.05900084227323532,
|
764 |
+
"eval_runtime": 14.0073,
|
765 |
+
"eval_samples_per_second": 89.025,
|
766 |
+
"eval_steps_per_second": 11.137,
|
767 |
+
"step": 920
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 10.11,
|
771 |
+
"learning_rate": 4.945652173913044e-05,
|
772 |
+
"loss": 0.03,
|
773 |
+
"step": 930
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 10.22,
|
777 |
+
"learning_rate": 4.891304347826087e-05,
|
778 |
+
"loss": 0.029,
|
779 |
+
"step": 940
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 10.33,
|
783 |
+
"learning_rate": 4.836956521739131e-05,
|
784 |
+
"loss": 0.0273,
|
785 |
+
"step": 950
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 10.43,
|
789 |
+
"learning_rate": 4.782608695652174e-05,
|
790 |
+
"loss": 0.0453,
|
791 |
+
"step": 960
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 10.43,
|
795 |
+
"eval_accuracy": 0.9895749799518845,
|
796 |
+
"eval_loss": 0.062146905809640884,
|
797 |
+
"eval_runtime": 14.1053,
|
798 |
+
"eval_samples_per_second": 88.406,
|
799 |
+
"eval_steps_per_second": 11.06,
|
800 |
+
"step": 960
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 10.54,
|
804 |
+
"learning_rate": 4.7282608695652177e-05,
|
805 |
+
"loss": 0.0415,
|
806 |
+
"step": 970
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 10.65,
|
810 |
+
"learning_rate": 4.673913043478261e-05,
|
811 |
+
"loss": 0.0268,
|
812 |
+
"step": 980
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 10.76,
|
816 |
+
"learning_rate": 4.6195652173913046e-05,
|
817 |
+
"loss": 0.0282,
|
818 |
+
"step": 990
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 10.87,
|
822 |
+
"learning_rate": 4.565217391304348e-05,
|
823 |
+
"loss": 0.026,
|
824 |
+
"step": 1000
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 10.87,
|
828 |
+
"eval_accuracy": 0.9855653568564555,
|
829 |
+
"eval_loss": 0.0649920180439949,
|
830 |
+
"eval_runtime": 13.8769,
|
831 |
+
"eval_samples_per_second": 89.861,
|
832 |
+
"eval_steps_per_second": 11.242,
|
833 |
+
"step": 1000
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 10.98,
|
837 |
+
"learning_rate": 4.5108695652173916e-05,
|
838 |
+
"loss": 0.0255,
|
839 |
+
"step": 1010
|
840 |
+
},
|
841 |
+
{
|
842 |
+
"epoch": 11.09,
|
843 |
+
"learning_rate": 4.456521739130435e-05,
|
844 |
+
"loss": 0.0246,
|
845 |
+
"step": 1020
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 11.2,
|
849 |
+
"learning_rate": 4.4021739130434786e-05,
|
850 |
+
"loss": 0.0264,
|
851 |
+
"step": 1030
|
852 |
+
},
|
853 |
+
{
|
854 |
+
"epoch": 11.3,
|
855 |
+
"learning_rate": 4.347826086956522e-05,
|
856 |
+
"loss": 0.0257,
|
857 |
+
"step": 1040
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"epoch": 11.3,
|
861 |
+
"eval_accuracy": 0.9927826784282278,
|
862 |
+
"eval_loss": 0.04654848575592041,
|
863 |
+
"eval_runtime": 13.6877,
|
864 |
+
"eval_samples_per_second": 91.103,
|
865 |
+
"eval_steps_per_second": 11.397,
|
866 |
+
"step": 1040
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 11.41,
|
870 |
+
"learning_rate": 4.2934782608695655e-05,
|
871 |
+
"loss": 0.0237,
|
872 |
+
"step": 1050
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 11.52,
|
876 |
+
"learning_rate": 4.239130434782609e-05,
|
877 |
+
"loss": 0.0233,
|
878 |
+
"step": 1060
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 11.63,
|
882 |
+
"learning_rate": 4.1847826086956525e-05,
|
883 |
+
"loss": 0.0231,
|
884 |
+
"step": 1070
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 11.74,
|
888 |
+
"learning_rate": 4.130434782608696e-05,
|
889 |
+
"loss": 0.041,
|
890 |
+
"step": 1080
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 11.74,
|
894 |
+
"eval_accuracy": 0.9927826784282278,
|
895 |
+
"eval_loss": 0.04421408474445343,
|
896 |
+
"eval_runtime": 14.1229,
|
897 |
+
"eval_samples_per_second": 88.296,
|
898 |
+
"eval_steps_per_second": 11.046,
|
899 |
+
"step": 1080
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 11.85,
|
903 |
+
"learning_rate": 4.076086956521739e-05,
|
904 |
+
"loss": 0.0234,
|
905 |
+
"step": 1090
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 11.96,
|
909 |
+
"learning_rate": 4.021739130434783e-05,
|
910 |
+
"loss": 0.0221,
|
911 |
+
"step": 1100
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 12.07,
|
915 |
+
"learning_rate": 3.9673913043478264e-05,
|
916 |
+
"loss": 0.0251,
|
917 |
+
"step": 1110
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 12.17,
|
921 |
+
"learning_rate": 3.91304347826087e-05,
|
922 |
+
"loss": 0.0223,
|
923 |
+
"step": 1120
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 12.17,
|
927 |
+
"eval_accuracy": 0.9863672814755413,
|
928 |
+
"eval_loss": 0.06379802525043488,
|
929 |
+
"eval_runtime": 13.3726,
|
930 |
+
"eval_samples_per_second": 93.25,
|
931 |
+
"eval_steps_per_second": 11.666,
|
932 |
+
"step": 1120
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 12.28,
|
936 |
+
"learning_rate": 3.8586956521739134e-05,
|
937 |
+
"loss": 0.0222,
|
938 |
+
"step": 1130
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 12.39,
|
942 |
+
"learning_rate": 3.804347826086957e-05,
|
943 |
+
"loss": 0.0208,
|
944 |
+
"step": 1140
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 12.5,
|
948 |
+
"learning_rate": 3.7500000000000003e-05,
|
949 |
+
"loss": 0.0207,
|
950 |
+
"step": 1150
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 12.61,
|
954 |
+
"learning_rate": 3.695652173913043e-05,
|
955 |
+
"loss": 0.0205,
|
956 |
+
"step": 1160
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 12.61,
|
960 |
+
"eval_accuracy": 0.9911788291900562,
|
961 |
+
"eval_loss": 0.050300538539886475,
|
962 |
+
"eval_runtime": 14.0669,
|
963 |
+
"eval_samples_per_second": 88.648,
|
964 |
+
"eval_steps_per_second": 11.09,
|
965 |
+
"step": 1160
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 12.72,
|
969 |
+
"learning_rate": 3.641304347826087e-05,
|
970 |
+
"loss": 0.0331,
|
971 |
+
"step": 1170
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 12.83,
|
975 |
+
"learning_rate": 3.58695652173913e-05,
|
976 |
+
"loss": 0.021,
|
977 |
+
"step": 1180
|
978 |
+
},
|
979 |
+
{
|
980 |
+
"epoch": 12.93,
|
981 |
+
"learning_rate": 3.532608695652174e-05,
|
982 |
+
"loss": 0.0203,
|
983 |
+
"step": 1190
|
984 |
+
},
|
985 |
+
{
|
986 |
+
"epoch": 13.04,
|
987 |
+
"learning_rate": 3.478260869565218e-05,
|
988 |
+
"loss": 0.0221,
|
989 |
+
"step": 1200
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 13.04,
|
993 |
+
"eval_accuracy": 0.991980753809142,
|
994 |
+
"eval_loss": 0.047799013555049896,
|
995 |
+
"eval_runtime": 13.3712,
|
996 |
+
"eval_samples_per_second": 93.26,
|
997 |
+
"eval_steps_per_second": 11.667,
|
998 |
+
"step": 1200
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 13.15,
|
1002 |
+
"learning_rate": 3.423913043478261e-05,
|
1003 |
+
"loss": 0.0191,
|
1004 |
+
"step": 1210
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 13.26,
|
1008 |
+
"learning_rate": 3.369565217391305e-05,
|
1009 |
+
"loss": 0.0195,
|
1010 |
+
"step": 1220
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 13.37,
|
1014 |
+
"learning_rate": 3.3152173913043475e-05,
|
1015 |
+
"loss": 0.0188,
|
1016 |
+
"step": 1230
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 13.48,
|
1020 |
+
"learning_rate": 3.260869565217392e-05,
|
1021 |
+
"loss": 0.0188,
|
1022 |
+
"step": 1240
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 13.48,
|
1026 |
+
"eval_accuracy": 0.9911788291900562,
|
1027 |
+
"eval_loss": 0.04699365794658661,
|
1028 |
+
"eval_runtime": 13.8942,
|
1029 |
+
"eval_samples_per_second": 89.75,
|
1030 |
+
"eval_steps_per_second": 11.228,
|
1031 |
+
"step": 1240
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 13.59,
|
1035 |
+
"learning_rate": 3.2065217391304345e-05,
|
1036 |
+
"loss": 0.019,
|
1037 |
+
"step": 1250
|
1038 |
+
},
|
1039 |
+
{
|
1040 |
+
"epoch": 13.7,
|
1041 |
+
"learning_rate": 3.152173913043479e-05,
|
1042 |
+
"loss": 0.0184,
|
1043 |
+
"step": 1260
|
1044 |
+
},
|
1045 |
+
{
|
1046 |
+
"epoch": 13.8,
|
1047 |
+
"learning_rate": 3.0978260869565215e-05,
|
1048 |
+
"loss": 0.0179,
|
1049 |
+
"step": 1270
|
1050 |
+
},
|
1051 |
+
{
|
1052 |
+
"epoch": 13.91,
|
1053 |
+
"learning_rate": 3.0434782608695656e-05,
|
1054 |
+
"loss": 0.0302,
|
1055 |
+
"step": 1280
|
1056 |
+
},
|
1057 |
+
{
|
1058 |
+
"epoch": 13.91,
|
1059 |
+
"eval_accuracy": 0.9927826784282278,
|
1060 |
+
"eval_loss": 0.04419828951358795,
|
1061 |
+
"eval_runtime": 13.9931,
|
1062 |
+
"eval_samples_per_second": 89.115,
|
1063 |
+
"eval_steps_per_second": 11.148,
|
1064 |
+
"step": 1280
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 14.02,
|
1068 |
+
"learning_rate": 2.9891304347826088e-05,
|
1069 |
+
"loss": 0.0182,
|
1070 |
+
"step": 1290
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 14.13,
|
1074 |
+
"learning_rate": 2.9347826086956526e-05,
|
1075 |
+
"loss": 0.0216,
|
1076 |
+
"step": 1300
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 14.24,
|
1080 |
+
"learning_rate": 2.8804347826086957e-05,
|
1081 |
+
"loss": 0.0174,
|
1082 |
+
"step": 1310
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 14.35,
|
1086 |
+
"learning_rate": 2.826086956521739e-05,
|
1087 |
+
"loss": 0.0171,
|
1088 |
+
"step": 1320
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 14.35,
|
1092 |
+
"eval_accuracy": 0.9935846030473136,
|
1093 |
+
"eval_loss": 0.04177280142903328,
|
1094 |
+
"eval_runtime": 13.9993,
|
1095 |
+
"eval_samples_per_second": 89.076,
|
1096 |
+
"eval_steps_per_second": 11.143,
|
1097 |
+
"step": 1320
|
1098 |
+
},
|
1099 |
+
{
|
1100 |
+
"epoch": 14.46,
|
1101 |
+
"learning_rate": 2.7717391304347827e-05,
|
1102 |
+
"loss": 0.0172,
|
1103 |
+
"step": 1330
|
1104 |
+
},
|
1105 |
+
{
|
1106 |
+
"epoch": 14.57,
|
1107 |
+
"learning_rate": 2.7173913043478262e-05,
|
1108 |
+
"loss": 0.0173,
|
1109 |
+
"step": 1340
|
1110 |
+
},
|
1111 |
+
{
|
1112 |
+
"epoch": 14.67,
|
1113 |
+
"learning_rate": 2.66304347826087e-05,
|
1114 |
+
"loss": 0.0259,
|
1115 |
+
"step": 1350
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 14.78,
|
1119 |
+
"learning_rate": 2.608695652173913e-05,
|
1120 |
+
"loss": 0.0197,
|
1121 |
+
"step": 1360
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 14.78,
|
1125 |
+
"eval_accuracy": 0.991980753809142,
|
1126 |
+
"eval_loss": 0.04225374758243561,
|
1127 |
+
"eval_runtime": 14.4748,
|
1128 |
+
"eval_samples_per_second": 86.15,
|
1129 |
+
"eval_steps_per_second": 10.777,
|
1130 |
+
"step": 1360
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 14.89,
|
1134 |
+
"learning_rate": 2.554347826086957e-05,
|
1135 |
+
"loss": 0.0166,
|
1136 |
+
"step": 1370
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 15.0,
|
1140 |
+
"learning_rate": 2.5e-05,
|
1141 |
+
"loss": 0.0163,
|
1142 |
+
"step": 1380
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 15.11,
|
1146 |
+
"learning_rate": 2.4456521739130436e-05,
|
1147 |
+
"loss": 0.0164,
|
1148 |
+
"step": 1390
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 15.22,
|
1152 |
+
"learning_rate": 2.391304347826087e-05,
|
1153 |
+
"loss": 0.0162,
|
1154 |
+
"step": 1400
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 15.22,
|
1158 |
+
"eval_accuracy": 0.9927826784282278,
|
1159 |
+
"eval_loss": 0.04216426983475685,
|
1160 |
+
"eval_runtime": 14.0671,
|
1161 |
+
"eval_samples_per_second": 88.646,
|
1162 |
+
"eval_steps_per_second": 11.09,
|
1163 |
+
"step": 1400
|
1164 |
+
},
|
1165 |
+
{
|
1166 |
+
"epoch": 15.33,
|
1167 |
+
"learning_rate": 2.3369565217391306e-05,
|
1168 |
+
"loss": 0.0172,
|
1169 |
+
"step": 1410
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 15.43,
|
1173 |
+
"learning_rate": 2.282608695652174e-05,
|
1174 |
+
"loss": 0.016,
|
1175 |
+
"step": 1420
|
1176 |
+
},
|
1177 |
+
{
|
1178 |
+
"epoch": 15.54,
|
1179 |
+
"learning_rate": 2.2282608695652175e-05,
|
1180 |
+
"loss": 0.0158,
|
1181 |
+
"step": 1430
|
1182 |
+
},
|
1183 |
+
{
|
1184 |
+
"epoch": 15.65,
|
1185 |
+
"learning_rate": 2.173913043478261e-05,
|
1186 |
+
"loss": 0.0159,
|
1187 |
+
"step": 1440
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 15.65,
|
1191 |
+
"eval_accuracy": 0.991980753809142,
|
1192 |
+
"eval_loss": 0.043235816061496735,
|
1193 |
+
"eval_runtime": 13.435,
|
1194 |
+
"eval_samples_per_second": 92.817,
|
1195 |
+
"eval_steps_per_second": 11.611,
|
1196 |
+
"step": 1440
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 15.76,
|
1200 |
+
"learning_rate": 2.1195652173913045e-05,
|
1201 |
+
"loss": 0.0158,
|
1202 |
+
"step": 1450
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 15.87,
|
1206 |
+
"learning_rate": 2.065217391304348e-05,
|
1207 |
+
"loss": 0.0252,
|
1208 |
+
"step": 1460
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 15.98,
|
1212 |
+
"learning_rate": 2.0108695652173915e-05,
|
1213 |
+
"loss": 0.0156,
|
1214 |
+
"step": 1470
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 16.09,
|
1218 |
+
"learning_rate": 1.956521739130435e-05,
|
1219 |
+
"loss": 0.0155,
|
1220 |
+
"step": 1480
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 16.09,
|
1224 |
+
"eval_accuracy": 0.9911788291900562,
|
1225 |
+
"eval_loss": 0.04137137532234192,
|
1226 |
+
"eval_runtime": 13.9815,
|
1227 |
+
"eval_samples_per_second": 89.189,
|
1228 |
+
"eval_steps_per_second": 11.158,
|
1229 |
+
"step": 1480
|
1230 |
+
}
|
1231 |
+
],
|
1232 |
+
"max_steps": 1840,
|
1233 |
+
"num_train_epochs": 20,
|
1234 |
+
"total_flos": 2.7507012153529467e+18,
|
1235 |
+
"trial_name": null,
|
1236 |
+
"trial_params": null
|
1237 |
+
}
|
checkpoint-1480/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8e63cb3581b9d82130890094c7fc9d777da84a1f291cbc24b62807505befaa6e
|
3 |
+
size 3899
|
checkpoint-1840/config.json
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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": "a",
|
13 |
+
"1": "b",
|
14 |
+
"10": "j",
|
15 |
+
"11": "k",
|
16 |
+
"12": "l",
|
17 |
+
"13": "m",
|
18 |
+
"14": "n",
|
19 |
+
"15": "o",
|
20 |
+
"16": "p",
|
21 |
+
"17": "period",
|
22 |
+
"18": "q",
|
23 |
+
"19": "question%20mark",
|
24 |
+
"2": "c",
|
25 |
+
"20": "r",
|
26 |
+
"21": "s",
|
27 |
+
"22": "t",
|
28 |
+
"23": "u",
|
29 |
+
"24": "v",
|
30 |
+
"25": "w",
|
31 |
+
"26": "x",
|
32 |
+
"27": "y",
|
33 |
+
"28": "z",
|
34 |
+
"3": "capital",
|
35 |
+
"4": "d",
|
36 |
+
"5": "e",
|
37 |
+
"6": "f",
|
38 |
+
"7": "g",
|
39 |
+
"8": "h",
|
40 |
+
"9": "i"
|
41 |
+
},
|
42 |
+
"image_size": 224,
|
43 |
+
"initializer_range": 0.02,
|
44 |
+
"intermediate_size": 3072,
|
45 |
+
"label2id": {
|
46 |
+
"a": "0",
|
47 |
+
"b": "1",
|
48 |
+
"c": "2",
|
49 |
+
"capital": "3",
|
50 |
+
"d": "4",
|
51 |
+
"e": "5",
|
52 |
+
"f": "6",
|
53 |
+
"g": "7",
|
54 |
+
"h": "8",
|
55 |
+
"i": "9",
|
56 |
+
"j": "10",
|
57 |
+
"k": "11",
|
58 |
+
"l": "12",
|
59 |
+
"m": "13",
|
60 |
+
"n": "14",
|
61 |
+
"o": "15",
|
62 |
+
"p": "16",
|
63 |
+
"period": "17",
|
64 |
+
"q": "18",
|
65 |
+
"question%20mark": "19",
|
66 |
+
"r": "20",
|
67 |
+
"s": "21",
|
68 |
+
"t": "22",
|
69 |
+
"u": "23",
|
70 |
+
"v": "24",
|
71 |
+
"w": "25",
|
72 |
+
"x": "26",
|
73 |
+
"y": "27",
|
74 |
+
"z": "28"
|
75 |
+
},
|
76 |
+
"layer_norm_eps": 1e-12,
|
77 |
+
"model_type": "vit",
|
78 |
+
"num_attention_heads": 12,
|
79 |
+
"num_channels": 3,
|
80 |
+
"num_hidden_layers": 12,
|
81 |
+
"patch_size": 16,
|
82 |
+
"problem_type": "single_label_classification",
|
83 |
+
"qkv_bias": true,
|
84 |
+
"torch_dtype": "float32",
|
85 |
+
"transformers_version": "4.30.2"
|
86 |
+
}
|
checkpoint-1840/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a757af4b0570107d7b36712511c0099bde5f8e5b053bd421121249270dd37aac
|
3 |
+
size 686684933
|
checkpoint-1840/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": "ViTFeatureExtractor",
|
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 |
+
}
|
checkpoint-1840/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d8a1095586e7e7963dc048bc6b0b40c7112af3d050fbc514aec9c9e58f681103
|
3 |
+
size 343351725
|
checkpoint-1840/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f05bfe4d4e69df16d1f272028d7714dff1f9f853f352dd41e3c82b1c4919791e
|
3 |
+
size 14575
|
checkpoint-1840/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d4d9af862f27d0b1ab39b0cf1250933f6abc8b4d08d033c50b3a88516bf2393d
|
3 |
+
size 627
|
checkpoint-1840/trainer_state.json
ADDED
@@ -0,0 +1,1534 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.04137137532234192,
|
3 |
+
"best_model_checkpoint": "./vit-base-beans/checkpoint-1480",
|
4 |
+
"epoch": 20.0,
|
5 |
+
"global_step": 1840,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.11,
|
12 |
+
"learning_rate": 9.945652173913043e-05,
|
13 |
+
"loss": 3.3812,
|
14 |
+
"step": 10
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.22,
|
18 |
+
"learning_rate": 9.891304347826087e-05,
|
19 |
+
"loss": 3.3288,
|
20 |
+
"step": 20
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.33,
|
24 |
+
"learning_rate": 9.836956521739132e-05,
|
25 |
+
"loss": 3.3101,
|
26 |
+
"step": 30
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.43,
|
30 |
+
"learning_rate": 9.782608695652174e-05,
|
31 |
+
"loss": 3.2579,
|
32 |
+
"step": 40
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.43,
|
36 |
+
"eval_accuracy": 0.17882919005613473,
|
37 |
+
"eval_loss": 3.1847527027130127,
|
38 |
+
"eval_runtime": 13.4342,
|
39 |
+
"eval_samples_per_second": 92.823,
|
40 |
+
"eval_steps_per_second": 11.612,
|
41 |
+
"step": 40
|
42 |
+
},
|
43 |
+
{
|
44 |
+
"epoch": 0.54,
|
45 |
+
"learning_rate": 9.728260869565217e-05,
|
46 |
+
"loss": 3.1255,
|
47 |
+
"step": 50
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"epoch": 0.65,
|
51 |
+
"learning_rate": 9.673913043478261e-05,
|
52 |
+
"loss": 3.0162,
|
53 |
+
"step": 60
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 0.76,
|
57 |
+
"learning_rate": 9.619565217391306e-05,
|
58 |
+
"loss": 2.8706,
|
59 |
+
"step": 70
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"epoch": 0.87,
|
63 |
+
"learning_rate": 9.565217391304348e-05,
|
64 |
+
"loss": 2.7157,
|
65 |
+
"step": 80
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.87,
|
69 |
+
"eval_accuracy": 0.43785084202085006,
|
70 |
+
"eval_loss": 2.5922651290893555,
|
71 |
+
"eval_runtime": 13.9416,
|
72 |
+
"eval_samples_per_second": 89.445,
|
73 |
+
"eval_steps_per_second": 11.19,
|
74 |
+
"step": 80
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.98,
|
78 |
+
"learning_rate": 9.510869565217391e-05,
|
79 |
+
"loss": 2.5322,
|
80 |
+
"step": 90
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 1.09,
|
84 |
+
"learning_rate": 9.456521739130435e-05,
|
85 |
+
"loss": 2.3504,
|
86 |
+
"step": 100
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 1.2,
|
90 |
+
"learning_rate": 9.402173913043478e-05,
|
91 |
+
"loss": 2.1887,
|
92 |
+
"step": 110
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 1.3,
|
96 |
+
"learning_rate": 9.347826086956522e-05,
|
97 |
+
"loss": 2.0664,
|
98 |
+
"step": 120
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 1.3,
|
102 |
+
"eval_accuracy": 0.6696070569366479,
|
103 |
+
"eval_loss": 1.9748882055282593,
|
104 |
+
"eval_runtime": 13.4691,
|
105 |
+
"eval_samples_per_second": 92.582,
|
106 |
+
"eval_steps_per_second": 11.582,
|
107 |
+
"step": 120
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 1.41,
|
111 |
+
"learning_rate": 9.293478260869566e-05,
|
112 |
+
"loss": 1.8705,
|
113 |
+
"step": 130
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 1.52,
|
117 |
+
"learning_rate": 9.239130434782609e-05,
|
118 |
+
"loss": 1.688,
|
119 |
+
"step": 140
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"epoch": 1.63,
|
123 |
+
"learning_rate": 9.184782608695652e-05,
|
124 |
+
"loss": 1.5939,
|
125 |
+
"step": 150
|
126 |
+
},
|
127 |
+
{
|
128 |
+
"epoch": 1.74,
|
129 |
+
"learning_rate": 9.130434782608696e-05,
|
130 |
+
"loss": 1.4765,
|
131 |
+
"step": 160
|
132 |
+
},
|
133 |
+
{
|
134 |
+
"epoch": 1.74,
|
135 |
+
"eval_accuracy": 0.917401764234162,
|
136 |
+
"eval_loss": 1.3413872718811035,
|
137 |
+
"eval_runtime": 13.9419,
|
138 |
+
"eval_samples_per_second": 89.442,
|
139 |
+
"eval_steps_per_second": 11.189,
|
140 |
+
"step": 160
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 1.85,
|
144 |
+
"learning_rate": 9.07608695652174e-05,
|
145 |
+
"loss": 1.3014,
|
146 |
+
"step": 170
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 1.96,
|
150 |
+
"learning_rate": 9.021739130434783e-05,
|
151 |
+
"loss": 1.201,
|
152 |
+
"step": 180
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 2.07,
|
156 |
+
"learning_rate": 8.967391304347826e-05,
|
157 |
+
"loss": 1.0165,
|
158 |
+
"step": 190
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 2.17,
|
162 |
+
"learning_rate": 8.91304347826087e-05,
|
163 |
+
"loss": 0.965,
|
164 |
+
"step": 200
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 2.17,
|
168 |
+
"eval_accuracy": 0.9615076182838813,
|
169 |
+
"eval_loss": 0.9264132380485535,
|
170 |
+
"eval_runtime": 13.1542,
|
171 |
+
"eval_samples_per_second": 94.798,
|
172 |
+
"eval_steps_per_second": 11.859,
|
173 |
+
"step": 200
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 2.28,
|
177 |
+
"learning_rate": 8.858695652173914e-05,
|
178 |
+
"loss": 0.9096,
|
179 |
+
"step": 210
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 2.39,
|
183 |
+
"learning_rate": 8.804347826086957e-05,
|
184 |
+
"loss": 0.8527,
|
185 |
+
"step": 220
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 2.5,
|
189 |
+
"learning_rate": 8.75e-05,
|
190 |
+
"loss": 0.759,
|
191 |
+
"step": 230
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 2.61,
|
195 |
+
"learning_rate": 8.695652173913044e-05,
|
196 |
+
"loss": 0.7163,
|
197 |
+
"step": 240
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"epoch": 2.61,
|
201 |
+
"eval_accuracy": 0.9647153167602245,
|
202 |
+
"eval_loss": 0.6652109622955322,
|
203 |
+
"eval_runtime": 14.0687,
|
204 |
+
"eval_samples_per_second": 88.637,
|
205 |
+
"eval_steps_per_second": 11.088,
|
206 |
+
"step": 240
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 2.72,
|
210 |
+
"learning_rate": 8.641304347826087e-05,
|
211 |
+
"loss": 0.6403,
|
212 |
+
"step": 250
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 2.83,
|
216 |
+
"learning_rate": 8.586956521739131e-05,
|
217 |
+
"loss": 0.5857,
|
218 |
+
"step": 260
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 2.93,
|
222 |
+
"learning_rate": 8.532608695652174e-05,
|
223 |
+
"loss": 0.5406,
|
224 |
+
"step": 270
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 3.04,
|
228 |
+
"learning_rate": 8.478260869565218e-05,
|
229 |
+
"loss": 0.5061,
|
230 |
+
"step": 280
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 3.04,
|
234 |
+
"eval_accuracy": 0.9687249398556536,
|
235 |
+
"eval_loss": 0.5080122947692871,
|
236 |
+
"eval_runtime": 13.0785,
|
237 |
+
"eval_samples_per_second": 95.347,
|
238 |
+
"eval_steps_per_second": 11.928,
|
239 |
+
"step": 280
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 3.15,
|
243 |
+
"learning_rate": 8.423913043478261e-05,
|
244 |
+
"loss": 0.4622,
|
245 |
+
"step": 290
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"epoch": 3.26,
|
249 |
+
"learning_rate": 8.369565217391305e-05,
|
250 |
+
"loss": 0.4919,
|
251 |
+
"step": 300
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"epoch": 3.37,
|
255 |
+
"learning_rate": 8.315217391304349e-05,
|
256 |
+
"loss": 0.371,
|
257 |
+
"step": 310
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 3.48,
|
261 |
+
"learning_rate": 8.260869565217392e-05,
|
262 |
+
"loss": 0.3883,
|
263 |
+
"step": 320
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 3.48,
|
267 |
+
"eval_accuracy": 0.9759422614274258,
|
268 |
+
"eval_loss": 0.3574630916118622,
|
269 |
+
"eval_runtime": 13.9479,
|
270 |
+
"eval_samples_per_second": 89.404,
|
271 |
+
"eval_steps_per_second": 11.184,
|
272 |
+
"step": 320
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 3.59,
|
276 |
+
"learning_rate": 8.206521739130435e-05,
|
277 |
+
"loss": 0.3831,
|
278 |
+
"step": 330
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 3.7,
|
282 |
+
"learning_rate": 8.152173913043478e-05,
|
283 |
+
"loss": 0.3329,
|
284 |
+
"step": 340
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 3.8,
|
288 |
+
"learning_rate": 8.097826086956523e-05,
|
289 |
+
"loss": 0.3383,
|
290 |
+
"step": 350
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 3.91,
|
294 |
+
"learning_rate": 8.043478260869566e-05,
|
295 |
+
"loss": 0.3328,
|
296 |
+
"step": 360
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 3.91,
|
300 |
+
"eval_accuracy": 0.9839615076182838,
|
301 |
+
"eval_loss": 0.27629122138023376,
|
302 |
+
"eval_runtime": 13.7308,
|
303 |
+
"eval_samples_per_second": 90.818,
|
304 |
+
"eval_steps_per_second": 11.361,
|
305 |
+
"step": 360
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"epoch": 4.02,
|
309 |
+
"learning_rate": 7.989130434782609e-05,
|
310 |
+
"loss": 0.2387,
|
311 |
+
"step": 370
|
312 |
+
},
|
313 |
+
{
|
314 |
+
"epoch": 4.13,
|
315 |
+
"learning_rate": 7.934782608695653e-05,
|
316 |
+
"loss": 0.2509,
|
317 |
+
"step": 380
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 4.24,
|
321 |
+
"learning_rate": 7.880434782608696e-05,
|
322 |
+
"loss": 0.2259,
|
323 |
+
"step": 390
|
324 |
+
},
|
325 |
+
{
|
326 |
+
"epoch": 4.35,
|
327 |
+
"learning_rate": 7.82608695652174e-05,
|
328 |
+
"loss": 0.2049,
|
329 |
+
"step": 400
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 4.35,
|
333 |
+
"eval_accuracy": 0.9855653568564555,
|
334 |
+
"eval_loss": 0.2094665914773941,
|
335 |
+
"eval_runtime": 13.8999,
|
336 |
+
"eval_samples_per_second": 89.713,
|
337 |
+
"eval_steps_per_second": 11.223,
|
338 |
+
"step": 400
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 4.46,
|
342 |
+
"learning_rate": 7.771739130434783e-05,
|
343 |
+
"loss": 0.1979,
|
344 |
+
"step": 410
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 4.57,
|
348 |
+
"learning_rate": 7.717391304347827e-05,
|
349 |
+
"loss": 0.1703,
|
350 |
+
"step": 420
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 4.67,
|
354 |
+
"learning_rate": 7.66304347826087e-05,
|
355 |
+
"loss": 0.1771,
|
356 |
+
"step": 430
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 4.78,
|
360 |
+
"learning_rate": 7.608695652173914e-05,
|
361 |
+
"loss": 0.2078,
|
362 |
+
"step": 440
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 4.78,
|
366 |
+
"eval_accuracy": 0.9871692060946271,
|
367 |
+
"eval_loss": 0.19693857431411743,
|
368 |
+
"eval_runtime": 13.7091,
|
369 |
+
"eval_samples_per_second": 90.961,
|
370 |
+
"eval_steps_per_second": 11.379,
|
371 |
+
"step": 440
|
372 |
+
},
|
373 |
+
{
|
374 |
+
"epoch": 4.89,
|
375 |
+
"learning_rate": 7.554347826086957e-05,
|
376 |
+
"loss": 0.1564,
|
377 |
+
"step": 450
|
378 |
+
},
|
379 |
+
{
|
380 |
+
"epoch": 5.0,
|
381 |
+
"learning_rate": 7.500000000000001e-05,
|
382 |
+
"loss": 0.1512,
|
383 |
+
"step": 460
|
384 |
+
},
|
385 |
+
{
|
386 |
+
"epoch": 5.11,
|
387 |
+
"learning_rate": 7.445652173913044e-05,
|
388 |
+
"loss": 0.1339,
|
389 |
+
"step": 470
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 5.22,
|
393 |
+
"learning_rate": 7.391304347826086e-05,
|
394 |
+
"loss": 0.1447,
|
395 |
+
"step": 480
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 5.22,
|
399 |
+
"eval_accuracy": 0.9871692060946271,
|
400 |
+
"eval_loss": 0.14835722744464874,
|
401 |
+
"eval_runtime": 13.9008,
|
402 |
+
"eval_samples_per_second": 89.707,
|
403 |
+
"eval_steps_per_second": 11.222,
|
404 |
+
"step": 480
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 5.33,
|
408 |
+
"learning_rate": 7.336956521739132e-05,
|
409 |
+
"loss": 0.1207,
|
410 |
+
"step": 490
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 5.43,
|
414 |
+
"learning_rate": 7.282608695652175e-05,
|
415 |
+
"loss": 0.1294,
|
416 |
+
"step": 500
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 5.54,
|
420 |
+
"learning_rate": 7.228260869565217e-05,
|
421 |
+
"loss": 0.116,
|
422 |
+
"step": 510
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 5.65,
|
426 |
+
"learning_rate": 7.17391304347826e-05,
|
427 |
+
"loss": 0.1401,
|
428 |
+
"step": 520
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 5.65,
|
432 |
+
"eval_accuracy": 0.9839615076182838,
|
433 |
+
"eval_loss": 0.14811548590660095,
|
434 |
+
"eval_runtime": 13.5209,
|
435 |
+
"eval_samples_per_second": 92.228,
|
436 |
+
"eval_steps_per_second": 11.538,
|
437 |
+
"step": 520
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 5.76,
|
441 |
+
"learning_rate": 7.119565217391306e-05,
|
442 |
+
"loss": 0.0904,
|
443 |
+
"step": 530
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 5.87,
|
447 |
+
"learning_rate": 7.065217391304349e-05,
|
448 |
+
"loss": 0.1099,
|
449 |
+
"step": 540
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 5.98,
|
453 |
+
"learning_rate": 7.010869565217391e-05,
|
454 |
+
"loss": 0.1599,
|
455 |
+
"step": 550
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 6.09,
|
459 |
+
"learning_rate": 6.956521739130436e-05,
|
460 |
+
"loss": 0.1232,
|
461 |
+
"step": 560
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 6.09,
|
465 |
+
"eval_accuracy": 0.991980753809142,
|
466 |
+
"eval_loss": 0.11416751146316528,
|
467 |
+
"eval_runtime": 13.9361,
|
468 |
+
"eval_samples_per_second": 89.48,
|
469 |
+
"eval_steps_per_second": 11.194,
|
470 |
+
"step": 560
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 6.2,
|
474 |
+
"learning_rate": 6.902173913043478e-05,
|
475 |
+
"loss": 0.1381,
|
476 |
+
"step": 570
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 6.3,
|
480 |
+
"learning_rate": 6.847826086956522e-05,
|
481 |
+
"loss": 0.1001,
|
482 |
+
"step": 580
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 6.41,
|
486 |
+
"learning_rate": 6.793478260869565e-05,
|
487 |
+
"loss": 0.0823,
|
488 |
+
"step": 590
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 6.52,
|
492 |
+
"learning_rate": 6.73913043478261e-05,
|
493 |
+
"loss": 0.0725,
|
494 |
+
"step": 600
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 6.52,
|
498 |
+
"eval_accuracy": 0.9879711307137129,
|
499 |
+
"eval_loss": 0.10076911747455597,
|
500 |
+
"eval_runtime": 13.8114,
|
501 |
+
"eval_samples_per_second": 90.288,
|
502 |
+
"eval_steps_per_second": 11.295,
|
503 |
+
"step": 600
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"epoch": 6.63,
|
507 |
+
"learning_rate": 6.684782608695652e-05,
|
508 |
+
"loss": 0.0852,
|
509 |
+
"step": 610
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 6.74,
|
513 |
+
"learning_rate": 6.630434782608695e-05,
|
514 |
+
"loss": 0.0723,
|
515 |
+
"step": 620
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 6.85,
|
519 |
+
"learning_rate": 6.576086956521739e-05,
|
520 |
+
"loss": 0.0881,
|
521 |
+
"step": 630
|
522 |
+
},
|
523 |
+
{
|
524 |
+
"epoch": 6.96,
|
525 |
+
"learning_rate": 6.521739130434783e-05,
|
526 |
+
"loss": 0.0934,
|
527 |
+
"step": 640
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 6.96,
|
531 |
+
"eval_accuracy": 0.9895749799518845,
|
532 |
+
"eval_loss": 0.09398525953292847,
|
533 |
+
"eval_runtime": 14.2025,
|
534 |
+
"eval_samples_per_second": 87.801,
|
535 |
+
"eval_steps_per_second": 10.984,
|
536 |
+
"step": 640
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 7.07,
|
540 |
+
"learning_rate": 6.467391304347826e-05,
|
541 |
+
"loss": 0.0668,
|
542 |
+
"step": 650
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 7.17,
|
546 |
+
"learning_rate": 6.413043478260869e-05,
|
547 |
+
"loss": 0.0586,
|
548 |
+
"step": 660
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 7.28,
|
552 |
+
"learning_rate": 6.358695652173913e-05,
|
553 |
+
"loss": 0.0543,
|
554 |
+
"step": 670
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 7.39,
|
558 |
+
"learning_rate": 6.304347826086957e-05,
|
559 |
+
"loss": 0.053,
|
560 |
+
"step": 680
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 7.39,
|
564 |
+
"eval_accuracy": 0.9895749799518845,
|
565 |
+
"eval_loss": 0.08539092540740967,
|
566 |
+
"eval_runtime": 13.9817,
|
567 |
+
"eval_samples_per_second": 89.188,
|
568 |
+
"eval_steps_per_second": 11.157,
|
569 |
+
"step": 680
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 7.5,
|
573 |
+
"learning_rate": 6.25e-05,
|
574 |
+
"loss": 0.0514,
|
575 |
+
"step": 690
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"epoch": 7.61,
|
579 |
+
"learning_rate": 6.195652173913043e-05,
|
580 |
+
"loss": 0.0491,
|
581 |
+
"step": 700
|
582 |
+
},
|
583 |
+
{
|
584 |
+
"epoch": 7.72,
|
585 |
+
"learning_rate": 6.141304347826087e-05,
|
586 |
+
"loss": 0.0481,
|
587 |
+
"step": 710
|
588 |
+
},
|
589 |
+
{
|
590 |
+
"epoch": 7.83,
|
591 |
+
"learning_rate": 6.086956521739131e-05,
|
592 |
+
"loss": 0.0469,
|
593 |
+
"step": 720
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 7.83,
|
597 |
+
"eval_accuracy": 0.9903769045709703,
|
598 |
+
"eval_loss": 0.06862174719572067,
|
599 |
+
"eval_runtime": 14.4287,
|
600 |
+
"eval_samples_per_second": 86.425,
|
601 |
+
"eval_steps_per_second": 10.812,
|
602 |
+
"step": 720
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 7.93,
|
606 |
+
"learning_rate": 6.032608695652174e-05,
|
607 |
+
"loss": 0.0693,
|
608 |
+
"step": 730
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 8.04,
|
612 |
+
"learning_rate": 5.9782608695652175e-05,
|
613 |
+
"loss": 0.0664,
|
614 |
+
"step": 740
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 8.15,
|
618 |
+
"learning_rate": 5.923913043478261e-05,
|
619 |
+
"loss": 0.0502,
|
620 |
+
"step": 750
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 8.26,
|
624 |
+
"learning_rate": 5.869565217391305e-05,
|
625 |
+
"loss": 0.0429,
|
626 |
+
"step": 760
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 8.26,
|
630 |
+
"eval_accuracy": 0.9863672814755413,
|
631 |
+
"eval_loss": 0.0824466422200203,
|
632 |
+
"eval_runtime": 13.8977,
|
633 |
+
"eval_samples_per_second": 89.727,
|
634 |
+
"eval_steps_per_second": 11.225,
|
635 |
+
"step": 760
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"epoch": 8.37,
|
639 |
+
"learning_rate": 5.815217391304349e-05,
|
640 |
+
"loss": 0.0622,
|
641 |
+
"step": 770
|
642 |
+
},
|
643 |
+
{
|
644 |
+
"epoch": 8.48,
|
645 |
+
"learning_rate": 5.7608695652173915e-05,
|
646 |
+
"loss": 0.0394,
|
647 |
+
"step": 780
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 8.59,
|
651 |
+
"learning_rate": 5.706521739130435e-05,
|
652 |
+
"loss": 0.0375,
|
653 |
+
"step": 790
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 8.7,
|
657 |
+
"learning_rate": 5.652173913043478e-05,
|
658 |
+
"loss": 0.0371,
|
659 |
+
"step": 800
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"epoch": 8.7,
|
663 |
+
"eval_accuracy": 0.991980753809142,
|
664 |
+
"eval_loss": 0.07010400295257568,
|
665 |
+
"eval_runtime": 13.4894,
|
666 |
+
"eval_samples_per_second": 92.443,
|
667 |
+
"eval_steps_per_second": 11.565,
|
668 |
+
"step": 800
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 8.8,
|
672 |
+
"learning_rate": 5.5978260869565226e-05,
|
673 |
+
"loss": 0.036,
|
674 |
+
"step": 810
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 8.91,
|
678 |
+
"learning_rate": 5.5434782608695654e-05,
|
679 |
+
"loss": 0.0352,
|
680 |
+
"step": 820
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 9.02,
|
684 |
+
"learning_rate": 5.489130434782609e-05,
|
685 |
+
"loss": 0.0344,
|
686 |
+
"step": 830
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 9.13,
|
690 |
+
"learning_rate": 5.4347826086956524e-05,
|
691 |
+
"loss": 0.033,
|
692 |
+
"step": 840
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 9.13,
|
696 |
+
"eval_accuracy": 0.991980753809142,
|
697 |
+
"eval_loss": 0.06847481429576874,
|
698 |
+
"eval_runtime": 13.9465,
|
699 |
+
"eval_samples_per_second": 89.413,
|
700 |
+
"eval_steps_per_second": 11.186,
|
701 |
+
"step": 840
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 9.24,
|
705 |
+
"learning_rate": 5.380434782608695e-05,
|
706 |
+
"loss": 0.0327,
|
707 |
+
"step": 850
|
708 |
+
},
|
709 |
+
{
|
710 |
+
"epoch": 9.35,
|
711 |
+
"learning_rate": 5.32608695652174e-05,
|
712 |
+
"loss": 0.0318,
|
713 |
+
"step": 860
|
714 |
+
},
|
715 |
+
{
|
716 |
+
"epoch": 9.46,
|
717 |
+
"learning_rate": 5.271739130434783e-05,
|
718 |
+
"loss": 0.0315,
|
719 |
+
"step": 870
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 9.57,
|
723 |
+
"learning_rate": 5.217391304347826e-05,
|
724 |
+
"loss": 0.0308,
|
725 |
+
"step": 880
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"epoch": 9.57,
|
729 |
+
"eval_accuracy": 0.991980753809142,
|
730 |
+
"eval_loss": 0.06314855068922043,
|
731 |
+
"eval_runtime": 13.4895,
|
732 |
+
"eval_samples_per_second": 92.442,
|
733 |
+
"eval_steps_per_second": 11.565,
|
734 |
+
"step": 880
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 9.67,
|
738 |
+
"learning_rate": 5.163043478260869e-05,
|
739 |
+
"loss": 0.0502,
|
740 |
+
"step": 890
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 9.78,
|
744 |
+
"learning_rate": 5.108695652173914e-05,
|
745 |
+
"loss": 0.03,
|
746 |
+
"step": 900
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 9.89,
|
750 |
+
"learning_rate": 5.054347826086957e-05,
|
751 |
+
"loss": 0.0294,
|
752 |
+
"step": 910
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 10.0,
|
756 |
+
"learning_rate": 5e-05,
|
757 |
+
"loss": 0.0398,
|
758 |
+
"step": 920
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 10.0,
|
762 |
+
"eval_accuracy": 0.9927826784282278,
|
763 |
+
"eval_loss": 0.05900084227323532,
|
764 |
+
"eval_runtime": 14.0073,
|
765 |
+
"eval_samples_per_second": 89.025,
|
766 |
+
"eval_steps_per_second": 11.137,
|
767 |
+
"step": 920
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 10.11,
|
771 |
+
"learning_rate": 4.945652173913044e-05,
|
772 |
+
"loss": 0.03,
|
773 |
+
"step": 930
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 10.22,
|
777 |
+
"learning_rate": 4.891304347826087e-05,
|
778 |
+
"loss": 0.029,
|
779 |
+
"step": 940
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 10.33,
|
783 |
+
"learning_rate": 4.836956521739131e-05,
|
784 |
+
"loss": 0.0273,
|
785 |
+
"step": 950
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 10.43,
|
789 |
+
"learning_rate": 4.782608695652174e-05,
|
790 |
+
"loss": 0.0453,
|
791 |
+
"step": 960
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 10.43,
|
795 |
+
"eval_accuracy": 0.9895749799518845,
|
796 |
+
"eval_loss": 0.062146905809640884,
|
797 |
+
"eval_runtime": 14.1053,
|
798 |
+
"eval_samples_per_second": 88.406,
|
799 |
+
"eval_steps_per_second": 11.06,
|
800 |
+
"step": 960
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 10.54,
|
804 |
+
"learning_rate": 4.7282608695652177e-05,
|
805 |
+
"loss": 0.0415,
|
806 |
+
"step": 970
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 10.65,
|
810 |
+
"learning_rate": 4.673913043478261e-05,
|
811 |
+
"loss": 0.0268,
|
812 |
+
"step": 980
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 10.76,
|
816 |
+
"learning_rate": 4.6195652173913046e-05,
|
817 |
+
"loss": 0.0282,
|
818 |
+
"step": 990
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 10.87,
|
822 |
+
"learning_rate": 4.565217391304348e-05,
|
823 |
+
"loss": 0.026,
|
824 |
+
"step": 1000
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 10.87,
|
828 |
+
"eval_accuracy": 0.9855653568564555,
|
829 |
+
"eval_loss": 0.0649920180439949,
|
830 |
+
"eval_runtime": 13.8769,
|
831 |
+
"eval_samples_per_second": 89.861,
|
832 |
+
"eval_steps_per_second": 11.242,
|
833 |
+
"step": 1000
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 10.98,
|
837 |
+
"learning_rate": 4.5108695652173916e-05,
|
838 |
+
"loss": 0.0255,
|
839 |
+
"step": 1010
|
840 |
+
},
|
841 |
+
{
|
842 |
+
"epoch": 11.09,
|
843 |
+
"learning_rate": 4.456521739130435e-05,
|
844 |
+
"loss": 0.0246,
|
845 |
+
"step": 1020
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 11.2,
|
849 |
+
"learning_rate": 4.4021739130434786e-05,
|
850 |
+
"loss": 0.0264,
|
851 |
+
"step": 1030
|
852 |
+
},
|
853 |
+
{
|
854 |
+
"epoch": 11.3,
|
855 |
+
"learning_rate": 4.347826086956522e-05,
|
856 |
+
"loss": 0.0257,
|
857 |
+
"step": 1040
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"epoch": 11.3,
|
861 |
+
"eval_accuracy": 0.9927826784282278,
|
862 |
+
"eval_loss": 0.04654848575592041,
|
863 |
+
"eval_runtime": 13.6877,
|
864 |
+
"eval_samples_per_second": 91.103,
|
865 |
+
"eval_steps_per_second": 11.397,
|
866 |
+
"step": 1040
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 11.41,
|
870 |
+
"learning_rate": 4.2934782608695655e-05,
|
871 |
+
"loss": 0.0237,
|
872 |
+
"step": 1050
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 11.52,
|
876 |
+
"learning_rate": 4.239130434782609e-05,
|
877 |
+
"loss": 0.0233,
|
878 |
+
"step": 1060
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 11.63,
|
882 |
+
"learning_rate": 4.1847826086956525e-05,
|
883 |
+
"loss": 0.0231,
|
884 |
+
"step": 1070
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 11.74,
|
888 |
+
"learning_rate": 4.130434782608696e-05,
|
889 |
+
"loss": 0.041,
|
890 |
+
"step": 1080
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 11.74,
|
894 |
+
"eval_accuracy": 0.9927826784282278,
|
895 |
+
"eval_loss": 0.04421408474445343,
|
896 |
+
"eval_runtime": 14.1229,
|
897 |
+
"eval_samples_per_second": 88.296,
|
898 |
+
"eval_steps_per_second": 11.046,
|
899 |
+
"step": 1080
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 11.85,
|
903 |
+
"learning_rate": 4.076086956521739e-05,
|
904 |
+
"loss": 0.0234,
|
905 |
+
"step": 1090
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 11.96,
|
909 |
+
"learning_rate": 4.021739130434783e-05,
|
910 |
+
"loss": 0.0221,
|
911 |
+
"step": 1100
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 12.07,
|
915 |
+
"learning_rate": 3.9673913043478264e-05,
|
916 |
+
"loss": 0.0251,
|
917 |
+
"step": 1110
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 12.17,
|
921 |
+
"learning_rate": 3.91304347826087e-05,
|
922 |
+
"loss": 0.0223,
|
923 |
+
"step": 1120
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 12.17,
|
927 |
+
"eval_accuracy": 0.9863672814755413,
|
928 |
+
"eval_loss": 0.06379802525043488,
|
929 |
+
"eval_runtime": 13.3726,
|
930 |
+
"eval_samples_per_second": 93.25,
|
931 |
+
"eval_steps_per_second": 11.666,
|
932 |
+
"step": 1120
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 12.28,
|
936 |
+
"learning_rate": 3.8586956521739134e-05,
|
937 |
+
"loss": 0.0222,
|
938 |
+
"step": 1130
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 12.39,
|
942 |
+
"learning_rate": 3.804347826086957e-05,
|
943 |
+
"loss": 0.0208,
|
944 |
+
"step": 1140
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 12.5,
|
948 |
+
"learning_rate": 3.7500000000000003e-05,
|
949 |
+
"loss": 0.0207,
|
950 |
+
"step": 1150
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 12.61,
|
954 |
+
"learning_rate": 3.695652173913043e-05,
|
955 |
+
"loss": 0.0205,
|
956 |
+
"step": 1160
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 12.61,
|
960 |
+
"eval_accuracy": 0.9911788291900562,
|
961 |
+
"eval_loss": 0.050300538539886475,
|
962 |
+
"eval_runtime": 14.0669,
|
963 |
+
"eval_samples_per_second": 88.648,
|
964 |
+
"eval_steps_per_second": 11.09,
|
965 |
+
"step": 1160
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 12.72,
|
969 |
+
"learning_rate": 3.641304347826087e-05,
|
970 |
+
"loss": 0.0331,
|
971 |
+
"step": 1170
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 12.83,
|
975 |
+
"learning_rate": 3.58695652173913e-05,
|
976 |
+
"loss": 0.021,
|
977 |
+
"step": 1180
|
978 |
+
},
|
979 |
+
{
|
980 |
+
"epoch": 12.93,
|
981 |
+
"learning_rate": 3.532608695652174e-05,
|
982 |
+
"loss": 0.0203,
|
983 |
+
"step": 1190
|
984 |
+
},
|
985 |
+
{
|
986 |
+
"epoch": 13.04,
|
987 |
+
"learning_rate": 3.478260869565218e-05,
|
988 |
+
"loss": 0.0221,
|
989 |
+
"step": 1200
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 13.04,
|
993 |
+
"eval_accuracy": 0.991980753809142,
|
994 |
+
"eval_loss": 0.047799013555049896,
|
995 |
+
"eval_runtime": 13.3712,
|
996 |
+
"eval_samples_per_second": 93.26,
|
997 |
+
"eval_steps_per_second": 11.667,
|
998 |
+
"step": 1200
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 13.15,
|
1002 |
+
"learning_rate": 3.423913043478261e-05,
|
1003 |
+
"loss": 0.0191,
|
1004 |
+
"step": 1210
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 13.26,
|
1008 |
+
"learning_rate": 3.369565217391305e-05,
|
1009 |
+
"loss": 0.0195,
|
1010 |
+
"step": 1220
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 13.37,
|
1014 |
+
"learning_rate": 3.3152173913043475e-05,
|
1015 |
+
"loss": 0.0188,
|
1016 |
+
"step": 1230
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 13.48,
|
1020 |
+
"learning_rate": 3.260869565217392e-05,
|
1021 |
+
"loss": 0.0188,
|
1022 |
+
"step": 1240
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 13.48,
|
1026 |
+
"eval_accuracy": 0.9911788291900562,
|
1027 |
+
"eval_loss": 0.04699365794658661,
|
1028 |
+
"eval_runtime": 13.8942,
|
1029 |
+
"eval_samples_per_second": 89.75,
|
1030 |
+
"eval_steps_per_second": 11.228,
|
1031 |
+
"step": 1240
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 13.59,
|
1035 |
+
"learning_rate": 3.2065217391304345e-05,
|
1036 |
+
"loss": 0.019,
|
1037 |
+
"step": 1250
|
1038 |
+
},
|
1039 |
+
{
|
1040 |
+
"epoch": 13.7,
|
1041 |
+
"learning_rate": 3.152173913043479e-05,
|
1042 |
+
"loss": 0.0184,
|
1043 |
+
"step": 1260
|
1044 |
+
},
|
1045 |
+
{
|
1046 |
+
"epoch": 13.8,
|
1047 |
+
"learning_rate": 3.0978260869565215e-05,
|
1048 |
+
"loss": 0.0179,
|
1049 |
+
"step": 1270
|
1050 |
+
},
|
1051 |
+
{
|
1052 |
+
"epoch": 13.91,
|
1053 |
+
"learning_rate": 3.0434782608695656e-05,
|
1054 |
+
"loss": 0.0302,
|
1055 |
+
"step": 1280
|
1056 |
+
},
|
1057 |
+
{
|
1058 |
+
"epoch": 13.91,
|
1059 |
+
"eval_accuracy": 0.9927826784282278,
|
1060 |
+
"eval_loss": 0.04419828951358795,
|
1061 |
+
"eval_runtime": 13.9931,
|
1062 |
+
"eval_samples_per_second": 89.115,
|
1063 |
+
"eval_steps_per_second": 11.148,
|
1064 |
+
"step": 1280
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 14.02,
|
1068 |
+
"learning_rate": 2.9891304347826088e-05,
|
1069 |
+
"loss": 0.0182,
|
1070 |
+
"step": 1290
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 14.13,
|
1074 |
+
"learning_rate": 2.9347826086956526e-05,
|
1075 |
+
"loss": 0.0216,
|
1076 |
+
"step": 1300
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 14.24,
|
1080 |
+
"learning_rate": 2.8804347826086957e-05,
|
1081 |
+
"loss": 0.0174,
|
1082 |
+
"step": 1310
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 14.35,
|
1086 |
+
"learning_rate": 2.826086956521739e-05,
|
1087 |
+
"loss": 0.0171,
|
1088 |
+
"step": 1320
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 14.35,
|
1092 |
+
"eval_accuracy": 0.9935846030473136,
|
1093 |
+
"eval_loss": 0.04177280142903328,
|
1094 |
+
"eval_runtime": 13.9993,
|
1095 |
+
"eval_samples_per_second": 89.076,
|
1096 |
+
"eval_steps_per_second": 11.143,
|
1097 |
+
"step": 1320
|
1098 |
+
},
|
1099 |
+
{
|
1100 |
+
"epoch": 14.46,
|
1101 |
+
"learning_rate": 2.7717391304347827e-05,
|
1102 |
+
"loss": 0.0172,
|
1103 |
+
"step": 1330
|
1104 |
+
},
|
1105 |
+
{
|
1106 |
+
"epoch": 14.57,
|
1107 |
+
"learning_rate": 2.7173913043478262e-05,
|
1108 |
+
"loss": 0.0173,
|
1109 |
+
"step": 1340
|
1110 |
+
},
|
1111 |
+
{
|
1112 |
+
"epoch": 14.67,
|
1113 |
+
"learning_rate": 2.66304347826087e-05,
|
1114 |
+
"loss": 0.0259,
|
1115 |
+
"step": 1350
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 14.78,
|
1119 |
+
"learning_rate": 2.608695652173913e-05,
|
1120 |
+
"loss": 0.0197,
|
1121 |
+
"step": 1360
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 14.78,
|
1125 |
+
"eval_accuracy": 0.991980753809142,
|
1126 |
+
"eval_loss": 0.04225374758243561,
|
1127 |
+
"eval_runtime": 14.4748,
|
1128 |
+
"eval_samples_per_second": 86.15,
|
1129 |
+
"eval_steps_per_second": 10.777,
|
1130 |
+
"step": 1360
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 14.89,
|
1134 |
+
"learning_rate": 2.554347826086957e-05,
|
1135 |
+
"loss": 0.0166,
|
1136 |
+
"step": 1370
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 15.0,
|
1140 |
+
"learning_rate": 2.5e-05,
|
1141 |
+
"loss": 0.0163,
|
1142 |
+
"step": 1380
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 15.11,
|
1146 |
+
"learning_rate": 2.4456521739130436e-05,
|
1147 |
+
"loss": 0.0164,
|
1148 |
+
"step": 1390
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 15.22,
|
1152 |
+
"learning_rate": 2.391304347826087e-05,
|
1153 |
+
"loss": 0.0162,
|
1154 |
+
"step": 1400
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 15.22,
|
1158 |
+
"eval_accuracy": 0.9927826784282278,
|
1159 |
+
"eval_loss": 0.04216426983475685,
|
1160 |
+
"eval_runtime": 14.0671,
|
1161 |
+
"eval_samples_per_second": 88.646,
|
1162 |
+
"eval_steps_per_second": 11.09,
|
1163 |
+
"step": 1400
|
1164 |
+
},
|
1165 |
+
{
|
1166 |
+
"epoch": 15.33,
|
1167 |
+
"learning_rate": 2.3369565217391306e-05,
|
1168 |
+
"loss": 0.0172,
|
1169 |
+
"step": 1410
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 15.43,
|
1173 |
+
"learning_rate": 2.282608695652174e-05,
|
1174 |
+
"loss": 0.016,
|
1175 |
+
"step": 1420
|
1176 |
+
},
|
1177 |
+
{
|
1178 |
+
"epoch": 15.54,
|
1179 |
+
"learning_rate": 2.2282608695652175e-05,
|
1180 |
+
"loss": 0.0158,
|
1181 |
+
"step": 1430
|
1182 |
+
},
|
1183 |
+
{
|
1184 |
+
"epoch": 15.65,
|
1185 |
+
"learning_rate": 2.173913043478261e-05,
|
1186 |
+
"loss": 0.0159,
|
1187 |
+
"step": 1440
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 15.65,
|
1191 |
+
"eval_accuracy": 0.991980753809142,
|
1192 |
+
"eval_loss": 0.043235816061496735,
|
1193 |
+
"eval_runtime": 13.435,
|
1194 |
+
"eval_samples_per_second": 92.817,
|
1195 |
+
"eval_steps_per_second": 11.611,
|
1196 |
+
"step": 1440
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 15.76,
|
1200 |
+
"learning_rate": 2.1195652173913045e-05,
|
1201 |
+
"loss": 0.0158,
|
1202 |
+
"step": 1450
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 15.87,
|
1206 |
+
"learning_rate": 2.065217391304348e-05,
|
1207 |
+
"loss": 0.0252,
|
1208 |
+
"step": 1460
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 15.98,
|
1212 |
+
"learning_rate": 2.0108695652173915e-05,
|
1213 |
+
"loss": 0.0156,
|
1214 |
+
"step": 1470
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 16.09,
|
1218 |
+
"learning_rate": 1.956521739130435e-05,
|
1219 |
+
"loss": 0.0155,
|
1220 |
+
"step": 1480
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 16.09,
|
1224 |
+
"eval_accuracy": 0.9911788291900562,
|
1225 |
+
"eval_loss": 0.04137137532234192,
|
1226 |
+
"eval_runtime": 13.9815,
|
1227 |
+
"eval_samples_per_second": 89.189,
|
1228 |
+
"eval_steps_per_second": 11.158,
|
1229 |
+
"step": 1480
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 16.2,
|
1233 |
+
"learning_rate": 1.9021739130434784e-05,
|
1234 |
+
"loss": 0.0175,
|
1235 |
+
"step": 1490
|
1236 |
+
},
|
1237 |
+
{
|
1238 |
+
"epoch": 16.3,
|
1239 |
+
"learning_rate": 1.8478260869565216e-05,
|
1240 |
+
"loss": 0.0155,
|
1241 |
+
"step": 1500
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 16.41,
|
1245 |
+
"learning_rate": 1.793478260869565e-05,
|
1246 |
+
"loss": 0.0258,
|
1247 |
+
"step": 1510
|
1248 |
+
},
|
1249 |
+
{
|
1250 |
+
"epoch": 16.52,
|
1251 |
+
"learning_rate": 1.739130434782609e-05,
|
1252 |
+
"loss": 0.015,
|
1253 |
+
"step": 1520
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 16.52,
|
1257 |
+
"eval_accuracy": 0.9911788291900562,
|
1258 |
+
"eval_loss": 0.0487416572868824,
|
1259 |
+
"eval_runtime": 13.4779,
|
1260 |
+
"eval_samples_per_second": 92.522,
|
1261 |
+
"eval_steps_per_second": 11.575,
|
1262 |
+
"step": 1520
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 16.63,
|
1266 |
+
"learning_rate": 1.6847826086956524e-05,
|
1267 |
+
"loss": 0.0152,
|
1268 |
+
"step": 1530
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 16.74,
|
1272 |
+
"learning_rate": 1.630434782608696e-05,
|
1273 |
+
"loss": 0.0174,
|
1274 |
+
"step": 1540
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 16.85,
|
1278 |
+
"learning_rate": 1.5760869565217393e-05,
|
1279 |
+
"loss": 0.0147,
|
1280 |
+
"step": 1550
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 16.96,
|
1284 |
+
"learning_rate": 1.5217391304347828e-05,
|
1285 |
+
"loss": 0.015,
|
1286 |
+
"step": 1560
|
1287 |
+
},
|
1288 |
+
{
|
1289 |
+
"epoch": 16.96,
|
1290 |
+
"eval_accuracy": 0.991980753809142,
|
1291 |
+
"eval_loss": 0.04399973526597023,
|
1292 |
+
"eval_runtime": 14.0057,
|
1293 |
+
"eval_samples_per_second": 89.035,
|
1294 |
+
"eval_steps_per_second": 11.138,
|
1295 |
+
"step": 1560
|
1296 |
+
},
|
1297 |
+
{
|
1298 |
+
"epoch": 17.07,
|
1299 |
+
"learning_rate": 1.4673913043478263e-05,
|
1300 |
+
"loss": 0.0148,
|
1301 |
+
"step": 1570
|
1302 |
+
},
|
1303 |
+
{
|
1304 |
+
"epoch": 17.17,
|
1305 |
+
"learning_rate": 1.4130434782608694e-05,
|
1306 |
+
"loss": 0.0147,
|
1307 |
+
"step": 1580
|
1308 |
+
},
|
1309 |
+
{
|
1310 |
+
"epoch": 17.28,
|
1311 |
+
"learning_rate": 1.3586956521739131e-05,
|
1312 |
+
"loss": 0.0148,
|
1313 |
+
"step": 1590
|
1314 |
+
},
|
1315 |
+
{
|
1316 |
+
"epoch": 17.39,
|
1317 |
+
"learning_rate": 1.3043478260869566e-05,
|
1318 |
+
"loss": 0.0146,
|
1319 |
+
"step": 1600
|
1320 |
+
},
|
1321 |
+
{
|
1322 |
+
"epoch": 17.39,
|
1323 |
+
"eval_accuracy": 0.991980753809142,
|
1324 |
+
"eval_loss": 0.04343697056174278,
|
1325 |
+
"eval_runtime": 14.4989,
|
1326 |
+
"eval_samples_per_second": 86.007,
|
1327 |
+
"eval_steps_per_second": 10.759,
|
1328 |
+
"step": 1600
|
1329 |
+
},
|
1330 |
+
{
|
1331 |
+
"epoch": 17.5,
|
1332 |
+
"learning_rate": 1.25e-05,
|
1333 |
+
"loss": 0.0145,
|
1334 |
+
"step": 1610
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 17.61,
|
1338 |
+
"learning_rate": 1.1956521739130435e-05,
|
1339 |
+
"loss": 0.0144,
|
1340 |
+
"step": 1620
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 17.72,
|
1344 |
+
"learning_rate": 1.141304347826087e-05,
|
1345 |
+
"loss": 0.0149,
|
1346 |
+
"step": 1630
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 17.83,
|
1350 |
+
"learning_rate": 1.0869565217391305e-05,
|
1351 |
+
"loss": 0.0143,
|
1352 |
+
"step": 1640
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 17.83,
|
1356 |
+
"eval_accuracy": 0.991980753809142,
|
1357 |
+
"eval_loss": 0.042883455753326416,
|
1358 |
+
"eval_runtime": 14.0877,
|
1359 |
+
"eval_samples_per_second": 88.517,
|
1360 |
+
"eval_steps_per_second": 11.073,
|
1361 |
+
"step": 1640
|
1362 |
+
},
|
1363 |
+
{
|
1364 |
+
"epoch": 17.93,
|
1365 |
+
"learning_rate": 1.032608695652174e-05,
|
1366 |
+
"loss": 0.0142,
|
1367 |
+
"step": 1650
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 18.04,
|
1371 |
+
"learning_rate": 9.782608695652175e-06,
|
1372 |
+
"loss": 0.0225,
|
1373 |
+
"step": 1660
|
1374 |
+
},
|
1375 |
+
{
|
1376 |
+
"epoch": 18.15,
|
1377 |
+
"learning_rate": 9.239130434782608e-06,
|
1378 |
+
"loss": 0.0146,
|
1379 |
+
"step": 1670
|
1380 |
+
},
|
1381 |
+
{
|
1382 |
+
"epoch": 18.26,
|
1383 |
+
"learning_rate": 8.695652173913044e-06,
|
1384 |
+
"loss": 0.0143,
|
1385 |
+
"step": 1680
|
1386 |
+
},
|
1387 |
+
{
|
1388 |
+
"epoch": 18.26,
|
1389 |
+
"eval_accuracy": 0.9911788291900562,
|
1390 |
+
"eval_loss": 0.04524253308773041,
|
1391 |
+
"eval_runtime": 14.4999,
|
1392 |
+
"eval_samples_per_second": 86.001,
|
1393 |
+
"eval_steps_per_second": 10.759,
|
1394 |
+
"step": 1680
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 18.37,
|
1398 |
+
"learning_rate": 8.15217391304348e-06,
|
1399 |
+
"loss": 0.0154,
|
1400 |
+
"step": 1690
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"epoch": 18.48,
|
1404 |
+
"learning_rate": 7.608695652173914e-06,
|
1405 |
+
"loss": 0.0144,
|
1406 |
+
"step": 1700
|
1407 |
+
},
|
1408 |
+
{
|
1409 |
+
"epoch": 18.59,
|
1410 |
+
"learning_rate": 7.065217391304347e-06,
|
1411 |
+
"loss": 0.014,
|
1412 |
+
"step": 1710
|
1413 |
+
},
|
1414 |
+
{
|
1415 |
+
"epoch": 18.7,
|
1416 |
+
"learning_rate": 6.521739130434783e-06,
|
1417 |
+
"loss": 0.014,
|
1418 |
+
"step": 1720
|
1419 |
+
},
|
1420 |
+
{
|
1421 |
+
"epoch": 18.7,
|
1422 |
+
"eval_accuracy": 0.9911788291900562,
|
1423 |
+
"eval_loss": 0.04453733563423157,
|
1424 |
+
"eval_runtime": 14.0735,
|
1425 |
+
"eval_samples_per_second": 88.606,
|
1426 |
+
"eval_steps_per_second": 11.085,
|
1427 |
+
"step": 1720
|
1428 |
+
},
|
1429 |
+
{
|
1430 |
+
"epoch": 18.8,
|
1431 |
+
"learning_rate": 5.978260869565218e-06,
|
1432 |
+
"loss": 0.018,
|
1433 |
+
"step": 1730
|
1434 |
+
},
|
1435 |
+
{
|
1436 |
+
"epoch": 18.91,
|
1437 |
+
"learning_rate": 5.4347826086956525e-06,
|
1438 |
+
"loss": 0.014,
|
1439 |
+
"step": 1740
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 19.02,
|
1443 |
+
"learning_rate": 4.891304347826087e-06,
|
1444 |
+
"loss": 0.0155,
|
1445 |
+
"step": 1750
|
1446 |
+
},
|
1447 |
+
{
|
1448 |
+
"epoch": 19.13,
|
1449 |
+
"learning_rate": 4.347826086956522e-06,
|
1450 |
+
"loss": 0.0141,
|
1451 |
+
"step": 1760
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 19.13,
|
1455 |
+
"eval_accuracy": 0.9911788291900562,
|
1456 |
+
"eval_loss": 0.048826370388269424,
|
1457 |
+
"eval_runtime": 14.2162,
|
1458 |
+
"eval_samples_per_second": 87.717,
|
1459 |
+
"eval_steps_per_second": 10.973,
|
1460 |
+
"step": 1760
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 19.24,
|
1464 |
+
"learning_rate": 3.804347826086957e-06,
|
1465 |
+
"loss": 0.0139,
|
1466 |
+
"step": 1770
|
1467 |
+
},
|
1468 |
+
{
|
1469 |
+
"epoch": 19.35,
|
1470 |
+
"learning_rate": 3.2608695652173914e-06,
|
1471 |
+
"loss": 0.0139,
|
1472 |
+
"step": 1780
|
1473 |
+
},
|
1474 |
+
{
|
1475 |
+
"epoch": 19.46,
|
1476 |
+
"learning_rate": 2.7173913043478263e-06,
|
1477 |
+
"loss": 0.0139,
|
1478 |
+
"step": 1790
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 19.57,
|
1482 |
+
"learning_rate": 2.173913043478261e-06,
|
1483 |
+
"loss": 0.0138,
|
1484 |
+
"step": 1800
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"epoch": 19.57,
|
1488 |
+
"eval_accuracy": 0.9911788291900562,
|
1489 |
+
"eval_loss": 0.048504043370485306,
|
1490 |
+
"eval_runtime": 13.6564,
|
1491 |
+
"eval_samples_per_second": 91.312,
|
1492 |
+
"eval_steps_per_second": 11.423,
|
1493 |
+
"step": 1800
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 19.67,
|
1497 |
+
"learning_rate": 1.6304347826086957e-06,
|
1498 |
+
"loss": 0.0144,
|
1499 |
+
"step": 1810
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 19.78,
|
1503 |
+
"learning_rate": 1.0869565217391306e-06,
|
1504 |
+
"loss": 0.0155,
|
1505 |
+
"step": 1820
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"epoch": 19.89,
|
1509 |
+
"learning_rate": 5.434782608695653e-07,
|
1510 |
+
"loss": 0.0141,
|
1511 |
+
"step": 1830
|
1512 |
+
},
|
1513 |
+
{
|
1514 |
+
"epoch": 20.0,
|
1515 |
+
"learning_rate": 0.0,
|
1516 |
+
"loss": 0.0138,
|
1517 |
+
"step": 1840
|
1518 |
+
},
|
1519 |
+
{
|
1520 |
+
"epoch": 20.0,
|
1521 |
+
"eval_accuracy": 0.9911788291900562,
|
1522 |
+
"eval_loss": 0.0495075099170208,
|
1523 |
+
"eval_runtime": 14.095,
|
1524 |
+
"eval_samples_per_second": 88.471,
|
1525 |
+
"eval_steps_per_second": 11.068,
|
1526 |
+
"step": 1840
|
1527 |
+
}
|
1528 |
+
],
|
1529 |
+
"max_steps": 1840,
|
1530 |
+
"num_train_epochs": 20,
|
1531 |
+
"total_flos": 3.419773941089157e+18,
|
1532 |
+
"trial_name": null,
|
1533 |
+
"trial_params": null
|
1534 |
+
}
|
checkpoint-1840/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8e63cb3581b9d82130890094c7fc9d777da84a1f291cbc24b62807505befaa6e
|
3 |
+
size 3899
|
config.json
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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": "a",
|
13 |
+
"1": "b",
|
14 |
+
"10": "j",
|
15 |
+
"11": "k",
|
16 |
+
"12": "l",
|
17 |
+
"13": "m",
|
18 |
+
"14": "n",
|
19 |
+
"15": "o",
|
20 |
+
"16": "p",
|
21 |
+
"17": "period",
|
22 |
+
"18": "q",
|
23 |
+
"19": "question%20mark",
|
24 |
+
"2": "c",
|
25 |
+
"20": "r",
|
26 |
+
"21": "s",
|
27 |
+
"22": "t",
|
28 |
+
"23": "u",
|
29 |
+
"24": "v",
|
30 |
+
"25": "w",
|
31 |
+
"26": "x",
|
32 |
+
"27": "y",
|
33 |
+
"28": "z",
|
34 |
+
"3": "capital",
|
35 |
+
"4": "d",
|
36 |
+
"5": "e",
|
37 |
+
"6": "f",
|
38 |
+
"7": "g",
|
39 |
+
"8": "h",
|
40 |
+
"9": "i"
|
41 |
+
},
|
42 |
+
"image_size": 224,
|
43 |
+
"initializer_range": 0.02,
|
44 |
+
"intermediate_size": 3072,
|
45 |
+
"label2id": {
|
46 |
+
"a": "0",
|
47 |
+
"b": "1",
|
48 |
+
"c": "2",
|
49 |
+
"capital": "3",
|
50 |
+
"d": "4",
|
51 |
+
"e": "5",
|
52 |
+
"f": "6",
|
53 |
+
"g": "7",
|
54 |
+
"h": "8",
|
55 |
+
"i": "9",
|
56 |
+
"j": "10",
|
57 |
+
"k": "11",
|
58 |
+
"l": "12",
|
59 |
+
"m": "13",
|
60 |
+
"n": "14",
|
61 |
+
"o": "15",
|
62 |
+
"p": "16",
|
63 |
+
"period": "17",
|
64 |
+
"q": "18",
|
65 |
+
"question%20mark": "19",
|
66 |
+
"r": "20",
|
67 |
+
"s": "21",
|
68 |
+
"t": "22",
|
69 |
+
"u": "23",
|
70 |
+
"v": "24",
|
71 |
+
"w": "25",
|
72 |
+
"x": "26",
|
73 |
+
"y": "27",
|
74 |
+
"z": "28"
|
75 |
+
},
|
76 |
+
"layer_norm_eps": 1e-12,
|
77 |
+
"model_type": "vit",
|
78 |
+
"num_attention_heads": 12,
|
79 |
+
"num_channels": 3,
|
80 |
+
"num_hidden_layers": 12,
|
81 |
+
"patch_size": 16,
|
82 |
+
"problem_type": "single_label_classification",
|
83 |
+
"qkv_bias": true,
|
84 |
+
"torch_dtype": "float32",
|
85 |
+
"transformers_version": "4.30.2"
|
86 |
+
}
|
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": "ViTFeatureExtractor",
|
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 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df31f9efd884fa58db6a3ad678a6079dd27b1b6e32f1cf200e7eb1ccc50f8620
|
3 |
+
size 343351725
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"total_flos": 3.419773941089157e+18,
|
4 |
+
"train_loss": 0.33101742866894474,
|
5 |
+
"train_runtime": 1771.2395,
|
6 |
+
"train_samples_per_second": 24.909,
|
7 |
+
"train_steps_per_second": 1.039
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1543 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.04137137532234192,
|
3 |
+
"best_model_checkpoint": "./vit-base-beans/checkpoint-1480",
|
4 |
+
"epoch": 20.0,
|
5 |
+
"global_step": 1840,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.11,
|
12 |
+
"learning_rate": 9.945652173913043e-05,
|
13 |
+
"loss": 3.3812,
|
14 |
+
"step": 10
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.22,
|
18 |
+
"learning_rate": 9.891304347826087e-05,
|
19 |
+
"loss": 3.3288,
|
20 |
+
"step": 20
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.33,
|
24 |
+
"learning_rate": 9.836956521739132e-05,
|
25 |
+
"loss": 3.3101,
|
26 |
+
"step": 30
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.43,
|
30 |
+
"learning_rate": 9.782608695652174e-05,
|
31 |
+
"loss": 3.2579,
|
32 |
+
"step": 40
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.43,
|
36 |
+
"eval_accuracy": 0.17882919005613473,
|
37 |
+
"eval_loss": 3.1847527027130127,
|
38 |
+
"eval_runtime": 13.4342,
|
39 |
+
"eval_samples_per_second": 92.823,
|
40 |
+
"eval_steps_per_second": 11.612,
|
41 |
+
"step": 40
|
42 |
+
},
|
43 |
+
{
|
44 |
+
"epoch": 0.54,
|
45 |
+
"learning_rate": 9.728260869565217e-05,
|
46 |
+
"loss": 3.1255,
|
47 |
+
"step": 50
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"epoch": 0.65,
|
51 |
+
"learning_rate": 9.673913043478261e-05,
|
52 |
+
"loss": 3.0162,
|
53 |
+
"step": 60
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 0.76,
|
57 |
+
"learning_rate": 9.619565217391306e-05,
|
58 |
+
"loss": 2.8706,
|
59 |
+
"step": 70
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"epoch": 0.87,
|
63 |
+
"learning_rate": 9.565217391304348e-05,
|
64 |
+
"loss": 2.7157,
|
65 |
+
"step": 80
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.87,
|
69 |
+
"eval_accuracy": 0.43785084202085006,
|
70 |
+
"eval_loss": 2.5922651290893555,
|
71 |
+
"eval_runtime": 13.9416,
|
72 |
+
"eval_samples_per_second": 89.445,
|
73 |
+
"eval_steps_per_second": 11.19,
|
74 |
+
"step": 80
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.98,
|
78 |
+
"learning_rate": 9.510869565217391e-05,
|
79 |
+
"loss": 2.5322,
|
80 |
+
"step": 90
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 1.09,
|
84 |
+
"learning_rate": 9.456521739130435e-05,
|
85 |
+
"loss": 2.3504,
|
86 |
+
"step": 100
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 1.2,
|
90 |
+
"learning_rate": 9.402173913043478e-05,
|
91 |
+
"loss": 2.1887,
|
92 |
+
"step": 110
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 1.3,
|
96 |
+
"learning_rate": 9.347826086956522e-05,
|
97 |
+
"loss": 2.0664,
|
98 |
+
"step": 120
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 1.3,
|
102 |
+
"eval_accuracy": 0.6696070569366479,
|
103 |
+
"eval_loss": 1.9748882055282593,
|
104 |
+
"eval_runtime": 13.4691,
|
105 |
+
"eval_samples_per_second": 92.582,
|
106 |
+
"eval_steps_per_second": 11.582,
|
107 |
+
"step": 120
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 1.41,
|
111 |
+
"learning_rate": 9.293478260869566e-05,
|
112 |
+
"loss": 1.8705,
|
113 |
+
"step": 130
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 1.52,
|
117 |
+
"learning_rate": 9.239130434782609e-05,
|
118 |
+
"loss": 1.688,
|
119 |
+
"step": 140
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"epoch": 1.63,
|
123 |
+
"learning_rate": 9.184782608695652e-05,
|
124 |
+
"loss": 1.5939,
|
125 |
+
"step": 150
|
126 |
+
},
|
127 |
+
{
|
128 |
+
"epoch": 1.74,
|
129 |
+
"learning_rate": 9.130434782608696e-05,
|
130 |
+
"loss": 1.4765,
|
131 |
+
"step": 160
|
132 |
+
},
|
133 |
+
{
|
134 |
+
"epoch": 1.74,
|
135 |
+
"eval_accuracy": 0.917401764234162,
|
136 |
+
"eval_loss": 1.3413872718811035,
|
137 |
+
"eval_runtime": 13.9419,
|
138 |
+
"eval_samples_per_second": 89.442,
|
139 |
+
"eval_steps_per_second": 11.189,
|
140 |
+
"step": 160
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 1.85,
|
144 |
+
"learning_rate": 9.07608695652174e-05,
|
145 |
+
"loss": 1.3014,
|
146 |
+
"step": 170
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 1.96,
|
150 |
+
"learning_rate": 9.021739130434783e-05,
|
151 |
+
"loss": 1.201,
|
152 |
+
"step": 180
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 2.07,
|
156 |
+
"learning_rate": 8.967391304347826e-05,
|
157 |
+
"loss": 1.0165,
|
158 |
+
"step": 190
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 2.17,
|
162 |
+
"learning_rate": 8.91304347826087e-05,
|
163 |
+
"loss": 0.965,
|
164 |
+
"step": 200
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 2.17,
|
168 |
+
"eval_accuracy": 0.9615076182838813,
|
169 |
+
"eval_loss": 0.9264132380485535,
|
170 |
+
"eval_runtime": 13.1542,
|
171 |
+
"eval_samples_per_second": 94.798,
|
172 |
+
"eval_steps_per_second": 11.859,
|
173 |
+
"step": 200
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 2.28,
|
177 |
+
"learning_rate": 8.858695652173914e-05,
|
178 |
+
"loss": 0.9096,
|
179 |
+
"step": 210
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 2.39,
|
183 |
+
"learning_rate": 8.804347826086957e-05,
|
184 |
+
"loss": 0.8527,
|
185 |
+
"step": 220
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 2.5,
|
189 |
+
"learning_rate": 8.75e-05,
|
190 |
+
"loss": 0.759,
|
191 |
+
"step": 230
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 2.61,
|
195 |
+
"learning_rate": 8.695652173913044e-05,
|
196 |
+
"loss": 0.7163,
|
197 |
+
"step": 240
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"epoch": 2.61,
|
201 |
+
"eval_accuracy": 0.9647153167602245,
|
202 |
+
"eval_loss": 0.6652109622955322,
|
203 |
+
"eval_runtime": 14.0687,
|
204 |
+
"eval_samples_per_second": 88.637,
|
205 |
+
"eval_steps_per_second": 11.088,
|
206 |
+
"step": 240
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 2.72,
|
210 |
+
"learning_rate": 8.641304347826087e-05,
|
211 |
+
"loss": 0.6403,
|
212 |
+
"step": 250
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 2.83,
|
216 |
+
"learning_rate": 8.586956521739131e-05,
|
217 |
+
"loss": 0.5857,
|
218 |
+
"step": 260
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 2.93,
|
222 |
+
"learning_rate": 8.532608695652174e-05,
|
223 |
+
"loss": 0.5406,
|
224 |
+
"step": 270
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 3.04,
|
228 |
+
"learning_rate": 8.478260869565218e-05,
|
229 |
+
"loss": 0.5061,
|
230 |
+
"step": 280
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 3.04,
|
234 |
+
"eval_accuracy": 0.9687249398556536,
|
235 |
+
"eval_loss": 0.5080122947692871,
|
236 |
+
"eval_runtime": 13.0785,
|
237 |
+
"eval_samples_per_second": 95.347,
|
238 |
+
"eval_steps_per_second": 11.928,
|
239 |
+
"step": 280
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 3.15,
|
243 |
+
"learning_rate": 8.423913043478261e-05,
|
244 |
+
"loss": 0.4622,
|
245 |
+
"step": 290
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"epoch": 3.26,
|
249 |
+
"learning_rate": 8.369565217391305e-05,
|
250 |
+
"loss": 0.4919,
|
251 |
+
"step": 300
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"epoch": 3.37,
|
255 |
+
"learning_rate": 8.315217391304349e-05,
|
256 |
+
"loss": 0.371,
|
257 |
+
"step": 310
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 3.48,
|
261 |
+
"learning_rate": 8.260869565217392e-05,
|
262 |
+
"loss": 0.3883,
|
263 |
+
"step": 320
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 3.48,
|
267 |
+
"eval_accuracy": 0.9759422614274258,
|
268 |
+
"eval_loss": 0.3574630916118622,
|
269 |
+
"eval_runtime": 13.9479,
|
270 |
+
"eval_samples_per_second": 89.404,
|
271 |
+
"eval_steps_per_second": 11.184,
|
272 |
+
"step": 320
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 3.59,
|
276 |
+
"learning_rate": 8.206521739130435e-05,
|
277 |
+
"loss": 0.3831,
|
278 |
+
"step": 330
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 3.7,
|
282 |
+
"learning_rate": 8.152173913043478e-05,
|
283 |
+
"loss": 0.3329,
|
284 |
+
"step": 340
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 3.8,
|
288 |
+
"learning_rate": 8.097826086956523e-05,
|
289 |
+
"loss": 0.3383,
|
290 |
+
"step": 350
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 3.91,
|
294 |
+
"learning_rate": 8.043478260869566e-05,
|
295 |
+
"loss": 0.3328,
|
296 |
+
"step": 360
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 3.91,
|
300 |
+
"eval_accuracy": 0.9839615076182838,
|
301 |
+
"eval_loss": 0.27629122138023376,
|
302 |
+
"eval_runtime": 13.7308,
|
303 |
+
"eval_samples_per_second": 90.818,
|
304 |
+
"eval_steps_per_second": 11.361,
|
305 |
+
"step": 360
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"epoch": 4.02,
|
309 |
+
"learning_rate": 7.989130434782609e-05,
|
310 |
+
"loss": 0.2387,
|
311 |
+
"step": 370
|
312 |
+
},
|
313 |
+
{
|
314 |
+
"epoch": 4.13,
|
315 |
+
"learning_rate": 7.934782608695653e-05,
|
316 |
+
"loss": 0.2509,
|
317 |
+
"step": 380
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 4.24,
|
321 |
+
"learning_rate": 7.880434782608696e-05,
|
322 |
+
"loss": 0.2259,
|
323 |
+
"step": 390
|
324 |
+
},
|
325 |
+
{
|
326 |
+
"epoch": 4.35,
|
327 |
+
"learning_rate": 7.82608695652174e-05,
|
328 |
+
"loss": 0.2049,
|
329 |
+
"step": 400
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 4.35,
|
333 |
+
"eval_accuracy": 0.9855653568564555,
|
334 |
+
"eval_loss": 0.2094665914773941,
|
335 |
+
"eval_runtime": 13.8999,
|
336 |
+
"eval_samples_per_second": 89.713,
|
337 |
+
"eval_steps_per_second": 11.223,
|
338 |
+
"step": 400
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 4.46,
|
342 |
+
"learning_rate": 7.771739130434783e-05,
|
343 |
+
"loss": 0.1979,
|
344 |
+
"step": 410
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 4.57,
|
348 |
+
"learning_rate": 7.717391304347827e-05,
|
349 |
+
"loss": 0.1703,
|
350 |
+
"step": 420
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 4.67,
|
354 |
+
"learning_rate": 7.66304347826087e-05,
|
355 |
+
"loss": 0.1771,
|
356 |
+
"step": 430
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 4.78,
|
360 |
+
"learning_rate": 7.608695652173914e-05,
|
361 |
+
"loss": 0.2078,
|
362 |
+
"step": 440
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 4.78,
|
366 |
+
"eval_accuracy": 0.9871692060946271,
|
367 |
+
"eval_loss": 0.19693857431411743,
|
368 |
+
"eval_runtime": 13.7091,
|
369 |
+
"eval_samples_per_second": 90.961,
|
370 |
+
"eval_steps_per_second": 11.379,
|
371 |
+
"step": 440
|
372 |
+
},
|
373 |
+
{
|
374 |
+
"epoch": 4.89,
|
375 |
+
"learning_rate": 7.554347826086957e-05,
|
376 |
+
"loss": 0.1564,
|
377 |
+
"step": 450
|
378 |
+
},
|
379 |
+
{
|
380 |
+
"epoch": 5.0,
|
381 |
+
"learning_rate": 7.500000000000001e-05,
|
382 |
+
"loss": 0.1512,
|
383 |
+
"step": 460
|
384 |
+
},
|
385 |
+
{
|
386 |
+
"epoch": 5.11,
|
387 |
+
"learning_rate": 7.445652173913044e-05,
|
388 |
+
"loss": 0.1339,
|
389 |
+
"step": 470
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 5.22,
|
393 |
+
"learning_rate": 7.391304347826086e-05,
|
394 |
+
"loss": 0.1447,
|
395 |
+
"step": 480
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 5.22,
|
399 |
+
"eval_accuracy": 0.9871692060946271,
|
400 |
+
"eval_loss": 0.14835722744464874,
|
401 |
+
"eval_runtime": 13.9008,
|
402 |
+
"eval_samples_per_second": 89.707,
|
403 |
+
"eval_steps_per_second": 11.222,
|
404 |
+
"step": 480
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 5.33,
|
408 |
+
"learning_rate": 7.336956521739132e-05,
|
409 |
+
"loss": 0.1207,
|
410 |
+
"step": 490
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 5.43,
|
414 |
+
"learning_rate": 7.282608695652175e-05,
|
415 |
+
"loss": 0.1294,
|
416 |
+
"step": 500
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 5.54,
|
420 |
+
"learning_rate": 7.228260869565217e-05,
|
421 |
+
"loss": 0.116,
|
422 |
+
"step": 510
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 5.65,
|
426 |
+
"learning_rate": 7.17391304347826e-05,
|
427 |
+
"loss": 0.1401,
|
428 |
+
"step": 520
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 5.65,
|
432 |
+
"eval_accuracy": 0.9839615076182838,
|
433 |
+
"eval_loss": 0.14811548590660095,
|
434 |
+
"eval_runtime": 13.5209,
|
435 |
+
"eval_samples_per_second": 92.228,
|
436 |
+
"eval_steps_per_second": 11.538,
|
437 |
+
"step": 520
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 5.76,
|
441 |
+
"learning_rate": 7.119565217391306e-05,
|
442 |
+
"loss": 0.0904,
|
443 |
+
"step": 530
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 5.87,
|
447 |
+
"learning_rate": 7.065217391304349e-05,
|
448 |
+
"loss": 0.1099,
|
449 |
+
"step": 540
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 5.98,
|
453 |
+
"learning_rate": 7.010869565217391e-05,
|
454 |
+
"loss": 0.1599,
|
455 |
+
"step": 550
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 6.09,
|
459 |
+
"learning_rate": 6.956521739130436e-05,
|
460 |
+
"loss": 0.1232,
|
461 |
+
"step": 560
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 6.09,
|
465 |
+
"eval_accuracy": 0.991980753809142,
|
466 |
+
"eval_loss": 0.11416751146316528,
|
467 |
+
"eval_runtime": 13.9361,
|
468 |
+
"eval_samples_per_second": 89.48,
|
469 |
+
"eval_steps_per_second": 11.194,
|
470 |
+
"step": 560
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 6.2,
|
474 |
+
"learning_rate": 6.902173913043478e-05,
|
475 |
+
"loss": 0.1381,
|
476 |
+
"step": 570
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 6.3,
|
480 |
+
"learning_rate": 6.847826086956522e-05,
|
481 |
+
"loss": 0.1001,
|
482 |
+
"step": 580
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 6.41,
|
486 |
+
"learning_rate": 6.793478260869565e-05,
|
487 |
+
"loss": 0.0823,
|
488 |
+
"step": 590
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 6.52,
|
492 |
+
"learning_rate": 6.73913043478261e-05,
|
493 |
+
"loss": 0.0725,
|
494 |
+
"step": 600
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 6.52,
|
498 |
+
"eval_accuracy": 0.9879711307137129,
|
499 |
+
"eval_loss": 0.10076911747455597,
|
500 |
+
"eval_runtime": 13.8114,
|
501 |
+
"eval_samples_per_second": 90.288,
|
502 |
+
"eval_steps_per_second": 11.295,
|
503 |
+
"step": 600
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"epoch": 6.63,
|
507 |
+
"learning_rate": 6.684782608695652e-05,
|
508 |
+
"loss": 0.0852,
|
509 |
+
"step": 610
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 6.74,
|
513 |
+
"learning_rate": 6.630434782608695e-05,
|
514 |
+
"loss": 0.0723,
|
515 |
+
"step": 620
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 6.85,
|
519 |
+
"learning_rate": 6.576086956521739e-05,
|
520 |
+
"loss": 0.0881,
|
521 |
+
"step": 630
|
522 |
+
},
|
523 |
+
{
|
524 |
+
"epoch": 6.96,
|
525 |
+
"learning_rate": 6.521739130434783e-05,
|
526 |
+
"loss": 0.0934,
|
527 |
+
"step": 640
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 6.96,
|
531 |
+
"eval_accuracy": 0.9895749799518845,
|
532 |
+
"eval_loss": 0.09398525953292847,
|
533 |
+
"eval_runtime": 14.2025,
|
534 |
+
"eval_samples_per_second": 87.801,
|
535 |
+
"eval_steps_per_second": 10.984,
|
536 |
+
"step": 640
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 7.07,
|
540 |
+
"learning_rate": 6.467391304347826e-05,
|
541 |
+
"loss": 0.0668,
|
542 |
+
"step": 650
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 7.17,
|
546 |
+
"learning_rate": 6.413043478260869e-05,
|
547 |
+
"loss": 0.0586,
|
548 |
+
"step": 660
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 7.28,
|
552 |
+
"learning_rate": 6.358695652173913e-05,
|
553 |
+
"loss": 0.0543,
|
554 |
+
"step": 670
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 7.39,
|
558 |
+
"learning_rate": 6.304347826086957e-05,
|
559 |
+
"loss": 0.053,
|
560 |
+
"step": 680
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 7.39,
|
564 |
+
"eval_accuracy": 0.9895749799518845,
|
565 |
+
"eval_loss": 0.08539092540740967,
|
566 |
+
"eval_runtime": 13.9817,
|
567 |
+
"eval_samples_per_second": 89.188,
|
568 |
+
"eval_steps_per_second": 11.157,
|
569 |
+
"step": 680
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 7.5,
|
573 |
+
"learning_rate": 6.25e-05,
|
574 |
+
"loss": 0.0514,
|
575 |
+
"step": 690
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"epoch": 7.61,
|
579 |
+
"learning_rate": 6.195652173913043e-05,
|
580 |
+
"loss": 0.0491,
|
581 |
+
"step": 700
|
582 |
+
},
|
583 |
+
{
|
584 |
+
"epoch": 7.72,
|
585 |
+
"learning_rate": 6.141304347826087e-05,
|
586 |
+
"loss": 0.0481,
|
587 |
+
"step": 710
|
588 |
+
},
|
589 |
+
{
|
590 |
+
"epoch": 7.83,
|
591 |
+
"learning_rate": 6.086956521739131e-05,
|
592 |
+
"loss": 0.0469,
|
593 |
+
"step": 720
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 7.83,
|
597 |
+
"eval_accuracy": 0.9903769045709703,
|
598 |
+
"eval_loss": 0.06862174719572067,
|
599 |
+
"eval_runtime": 14.4287,
|
600 |
+
"eval_samples_per_second": 86.425,
|
601 |
+
"eval_steps_per_second": 10.812,
|
602 |
+
"step": 720
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 7.93,
|
606 |
+
"learning_rate": 6.032608695652174e-05,
|
607 |
+
"loss": 0.0693,
|
608 |
+
"step": 730
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 8.04,
|
612 |
+
"learning_rate": 5.9782608695652175e-05,
|
613 |
+
"loss": 0.0664,
|
614 |
+
"step": 740
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 8.15,
|
618 |
+
"learning_rate": 5.923913043478261e-05,
|
619 |
+
"loss": 0.0502,
|
620 |
+
"step": 750
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 8.26,
|
624 |
+
"learning_rate": 5.869565217391305e-05,
|
625 |
+
"loss": 0.0429,
|
626 |
+
"step": 760
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 8.26,
|
630 |
+
"eval_accuracy": 0.9863672814755413,
|
631 |
+
"eval_loss": 0.0824466422200203,
|
632 |
+
"eval_runtime": 13.8977,
|
633 |
+
"eval_samples_per_second": 89.727,
|
634 |
+
"eval_steps_per_second": 11.225,
|
635 |
+
"step": 760
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"epoch": 8.37,
|
639 |
+
"learning_rate": 5.815217391304349e-05,
|
640 |
+
"loss": 0.0622,
|
641 |
+
"step": 770
|
642 |
+
},
|
643 |
+
{
|
644 |
+
"epoch": 8.48,
|
645 |
+
"learning_rate": 5.7608695652173915e-05,
|
646 |
+
"loss": 0.0394,
|
647 |
+
"step": 780
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 8.59,
|
651 |
+
"learning_rate": 5.706521739130435e-05,
|
652 |
+
"loss": 0.0375,
|
653 |
+
"step": 790
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 8.7,
|
657 |
+
"learning_rate": 5.652173913043478e-05,
|
658 |
+
"loss": 0.0371,
|
659 |
+
"step": 800
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"epoch": 8.7,
|
663 |
+
"eval_accuracy": 0.991980753809142,
|
664 |
+
"eval_loss": 0.07010400295257568,
|
665 |
+
"eval_runtime": 13.4894,
|
666 |
+
"eval_samples_per_second": 92.443,
|
667 |
+
"eval_steps_per_second": 11.565,
|
668 |
+
"step": 800
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 8.8,
|
672 |
+
"learning_rate": 5.5978260869565226e-05,
|
673 |
+
"loss": 0.036,
|
674 |
+
"step": 810
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 8.91,
|
678 |
+
"learning_rate": 5.5434782608695654e-05,
|
679 |
+
"loss": 0.0352,
|
680 |
+
"step": 820
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 9.02,
|
684 |
+
"learning_rate": 5.489130434782609e-05,
|
685 |
+
"loss": 0.0344,
|
686 |
+
"step": 830
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 9.13,
|
690 |
+
"learning_rate": 5.4347826086956524e-05,
|
691 |
+
"loss": 0.033,
|
692 |
+
"step": 840
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 9.13,
|
696 |
+
"eval_accuracy": 0.991980753809142,
|
697 |
+
"eval_loss": 0.06847481429576874,
|
698 |
+
"eval_runtime": 13.9465,
|
699 |
+
"eval_samples_per_second": 89.413,
|
700 |
+
"eval_steps_per_second": 11.186,
|
701 |
+
"step": 840
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 9.24,
|
705 |
+
"learning_rate": 5.380434782608695e-05,
|
706 |
+
"loss": 0.0327,
|
707 |
+
"step": 850
|
708 |
+
},
|
709 |
+
{
|
710 |
+
"epoch": 9.35,
|
711 |
+
"learning_rate": 5.32608695652174e-05,
|
712 |
+
"loss": 0.0318,
|
713 |
+
"step": 860
|
714 |
+
},
|
715 |
+
{
|
716 |
+
"epoch": 9.46,
|
717 |
+
"learning_rate": 5.271739130434783e-05,
|
718 |
+
"loss": 0.0315,
|
719 |
+
"step": 870
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 9.57,
|
723 |
+
"learning_rate": 5.217391304347826e-05,
|
724 |
+
"loss": 0.0308,
|
725 |
+
"step": 880
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"epoch": 9.57,
|
729 |
+
"eval_accuracy": 0.991980753809142,
|
730 |
+
"eval_loss": 0.06314855068922043,
|
731 |
+
"eval_runtime": 13.4895,
|
732 |
+
"eval_samples_per_second": 92.442,
|
733 |
+
"eval_steps_per_second": 11.565,
|
734 |
+
"step": 880
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 9.67,
|
738 |
+
"learning_rate": 5.163043478260869e-05,
|
739 |
+
"loss": 0.0502,
|
740 |
+
"step": 890
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 9.78,
|
744 |
+
"learning_rate": 5.108695652173914e-05,
|
745 |
+
"loss": 0.03,
|
746 |
+
"step": 900
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 9.89,
|
750 |
+
"learning_rate": 5.054347826086957e-05,
|
751 |
+
"loss": 0.0294,
|
752 |
+
"step": 910
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 10.0,
|
756 |
+
"learning_rate": 5e-05,
|
757 |
+
"loss": 0.0398,
|
758 |
+
"step": 920
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 10.0,
|
762 |
+
"eval_accuracy": 0.9927826784282278,
|
763 |
+
"eval_loss": 0.05900084227323532,
|
764 |
+
"eval_runtime": 14.0073,
|
765 |
+
"eval_samples_per_second": 89.025,
|
766 |
+
"eval_steps_per_second": 11.137,
|
767 |
+
"step": 920
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 10.11,
|
771 |
+
"learning_rate": 4.945652173913044e-05,
|
772 |
+
"loss": 0.03,
|
773 |
+
"step": 930
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 10.22,
|
777 |
+
"learning_rate": 4.891304347826087e-05,
|
778 |
+
"loss": 0.029,
|
779 |
+
"step": 940
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 10.33,
|
783 |
+
"learning_rate": 4.836956521739131e-05,
|
784 |
+
"loss": 0.0273,
|
785 |
+
"step": 950
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 10.43,
|
789 |
+
"learning_rate": 4.782608695652174e-05,
|
790 |
+
"loss": 0.0453,
|
791 |
+
"step": 960
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 10.43,
|
795 |
+
"eval_accuracy": 0.9895749799518845,
|
796 |
+
"eval_loss": 0.062146905809640884,
|
797 |
+
"eval_runtime": 14.1053,
|
798 |
+
"eval_samples_per_second": 88.406,
|
799 |
+
"eval_steps_per_second": 11.06,
|
800 |
+
"step": 960
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 10.54,
|
804 |
+
"learning_rate": 4.7282608695652177e-05,
|
805 |
+
"loss": 0.0415,
|
806 |
+
"step": 970
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 10.65,
|
810 |
+
"learning_rate": 4.673913043478261e-05,
|
811 |
+
"loss": 0.0268,
|
812 |
+
"step": 980
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 10.76,
|
816 |
+
"learning_rate": 4.6195652173913046e-05,
|
817 |
+
"loss": 0.0282,
|
818 |
+
"step": 990
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 10.87,
|
822 |
+
"learning_rate": 4.565217391304348e-05,
|
823 |
+
"loss": 0.026,
|
824 |
+
"step": 1000
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 10.87,
|
828 |
+
"eval_accuracy": 0.9855653568564555,
|
829 |
+
"eval_loss": 0.0649920180439949,
|
830 |
+
"eval_runtime": 13.8769,
|
831 |
+
"eval_samples_per_second": 89.861,
|
832 |
+
"eval_steps_per_second": 11.242,
|
833 |
+
"step": 1000
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 10.98,
|
837 |
+
"learning_rate": 4.5108695652173916e-05,
|
838 |
+
"loss": 0.0255,
|
839 |
+
"step": 1010
|
840 |
+
},
|
841 |
+
{
|
842 |
+
"epoch": 11.09,
|
843 |
+
"learning_rate": 4.456521739130435e-05,
|
844 |
+
"loss": 0.0246,
|
845 |
+
"step": 1020
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 11.2,
|
849 |
+
"learning_rate": 4.4021739130434786e-05,
|
850 |
+
"loss": 0.0264,
|
851 |
+
"step": 1030
|
852 |
+
},
|
853 |
+
{
|
854 |
+
"epoch": 11.3,
|
855 |
+
"learning_rate": 4.347826086956522e-05,
|
856 |
+
"loss": 0.0257,
|
857 |
+
"step": 1040
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"epoch": 11.3,
|
861 |
+
"eval_accuracy": 0.9927826784282278,
|
862 |
+
"eval_loss": 0.04654848575592041,
|
863 |
+
"eval_runtime": 13.6877,
|
864 |
+
"eval_samples_per_second": 91.103,
|
865 |
+
"eval_steps_per_second": 11.397,
|
866 |
+
"step": 1040
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 11.41,
|
870 |
+
"learning_rate": 4.2934782608695655e-05,
|
871 |
+
"loss": 0.0237,
|
872 |
+
"step": 1050
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 11.52,
|
876 |
+
"learning_rate": 4.239130434782609e-05,
|
877 |
+
"loss": 0.0233,
|
878 |
+
"step": 1060
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 11.63,
|
882 |
+
"learning_rate": 4.1847826086956525e-05,
|
883 |
+
"loss": 0.0231,
|
884 |
+
"step": 1070
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 11.74,
|
888 |
+
"learning_rate": 4.130434782608696e-05,
|
889 |
+
"loss": 0.041,
|
890 |
+
"step": 1080
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 11.74,
|
894 |
+
"eval_accuracy": 0.9927826784282278,
|
895 |
+
"eval_loss": 0.04421408474445343,
|
896 |
+
"eval_runtime": 14.1229,
|
897 |
+
"eval_samples_per_second": 88.296,
|
898 |
+
"eval_steps_per_second": 11.046,
|
899 |
+
"step": 1080
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 11.85,
|
903 |
+
"learning_rate": 4.076086956521739e-05,
|
904 |
+
"loss": 0.0234,
|
905 |
+
"step": 1090
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 11.96,
|
909 |
+
"learning_rate": 4.021739130434783e-05,
|
910 |
+
"loss": 0.0221,
|
911 |
+
"step": 1100
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 12.07,
|
915 |
+
"learning_rate": 3.9673913043478264e-05,
|
916 |
+
"loss": 0.0251,
|
917 |
+
"step": 1110
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 12.17,
|
921 |
+
"learning_rate": 3.91304347826087e-05,
|
922 |
+
"loss": 0.0223,
|
923 |
+
"step": 1120
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 12.17,
|
927 |
+
"eval_accuracy": 0.9863672814755413,
|
928 |
+
"eval_loss": 0.06379802525043488,
|
929 |
+
"eval_runtime": 13.3726,
|
930 |
+
"eval_samples_per_second": 93.25,
|
931 |
+
"eval_steps_per_second": 11.666,
|
932 |
+
"step": 1120
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 12.28,
|
936 |
+
"learning_rate": 3.8586956521739134e-05,
|
937 |
+
"loss": 0.0222,
|
938 |
+
"step": 1130
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 12.39,
|
942 |
+
"learning_rate": 3.804347826086957e-05,
|
943 |
+
"loss": 0.0208,
|
944 |
+
"step": 1140
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 12.5,
|
948 |
+
"learning_rate": 3.7500000000000003e-05,
|
949 |
+
"loss": 0.0207,
|
950 |
+
"step": 1150
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 12.61,
|
954 |
+
"learning_rate": 3.695652173913043e-05,
|
955 |
+
"loss": 0.0205,
|
956 |
+
"step": 1160
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 12.61,
|
960 |
+
"eval_accuracy": 0.9911788291900562,
|
961 |
+
"eval_loss": 0.050300538539886475,
|
962 |
+
"eval_runtime": 14.0669,
|
963 |
+
"eval_samples_per_second": 88.648,
|
964 |
+
"eval_steps_per_second": 11.09,
|
965 |
+
"step": 1160
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 12.72,
|
969 |
+
"learning_rate": 3.641304347826087e-05,
|
970 |
+
"loss": 0.0331,
|
971 |
+
"step": 1170
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 12.83,
|
975 |
+
"learning_rate": 3.58695652173913e-05,
|
976 |
+
"loss": 0.021,
|
977 |
+
"step": 1180
|
978 |
+
},
|
979 |
+
{
|
980 |
+
"epoch": 12.93,
|
981 |
+
"learning_rate": 3.532608695652174e-05,
|
982 |
+
"loss": 0.0203,
|
983 |
+
"step": 1190
|
984 |
+
},
|
985 |
+
{
|
986 |
+
"epoch": 13.04,
|
987 |
+
"learning_rate": 3.478260869565218e-05,
|
988 |
+
"loss": 0.0221,
|
989 |
+
"step": 1200
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 13.04,
|
993 |
+
"eval_accuracy": 0.991980753809142,
|
994 |
+
"eval_loss": 0.047799013555049896,
|
995 |
+
"eval_runtime": 13.3712,
|
996 |
+
"eval_samples_per_second": 93.26,
|
997 |
+
"eval_steps_per_second": 11.667,
|
998 |
+
"step": 1200
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 13.15,
|
1002 |
+
"learning_rate": 3.423913043478261e-05,
|
1003 |
+
"loss": 0.0191,
|
1004 |
+
"step": 1210
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 13.26,
|
1008 |
+
"learning_rate": 3.369565217391305e-05,
|
1009 |
+
"loss": 0.0195,
|
1010 |
+
"step": 1220
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 13.37,
|
1014 |
+
"learning_rate": 3.3152173913043475e-05,
|
1015 |
+
"loss": 0.0188,
|
1016 |
+
"step": 1230
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 13.48,
|
1020 |
+
"learning_rate": 3.260869565217392e-05,
|
1021 |
+
"loss": 0.0188,
|
1022 |
+
"step": 1240
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 13.48,
|
1026 |
+
"eval_accuracy": 0.9911788291900562,
|
1027 |
+
"eval_loss": 0.04699365794658661,
|
1028 |
+
"eval_runtime": 13.8942,
|
1029 |
+
"eval_samples_per_second": 89.75,
|
1030 |
+
"eval_steps_per_second": 11.228,
|
1031 |
+
"step": 1240
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 13.59,
|
1035 |
+
"learning_rate": 3.2065217391304345e-05,
|
1036 |
+
"loss": 0.019,
|
1037 |
+
"step": 1250
|
1038 |
+
},
|
1039 |
+
{
|
1040 |
+
"epoch": 13.7,
|
1041 |
+
"learning_rate": 3.152173913043479e-05,
|
1042 |
+
"loss": 0.0184,
|
1043 |
+
"step": 1260
|
1044 |
+
},
|
1045 |
+
{
|
1046 |
+
"epoch": 13.8,
|
1047 |
+
"learning_rate": 3.0978260869565215e-05,
|
1048 |
+
"loss": 0.0179,
|
1049 |
+
"step": 1270
|
1050 |
+
},
|
1051 |
+
{
|
1052 |
+
"epoch": 13.91,
|
1053 |
+
"learning_rate": 3.0434782608695656e-05,
|
1054 |
+
"loss": 0.0302,
|
1055 |
+
"step": 1280
|
1056 |
+
},
|
1057 |
+
{
|
1058 |
+
"epoch": 13.91,
|
1059 |
+
"eval_accuracy": 0.9927826784282278,
|
1060 |
+
"eval_loss": 0.04419828951358795,
|
1061 |
+
"eval_runtime": 13.9931,
|
1062 |
+
"eval_samples_per_second": 89.115,
|
1063 |
+
"eval_steps_per_second": 11.148,
|
1064 |
+
"step": 1280
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 14.02,
|
1068 |
+
"learning_rate": 2.9891304347826088e-05,
|
1069 |
+
"loss": 0.0182,
|
1070 |
+
"step": 1290
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 14.13,
|
1074 |
+
"learning_rate": 2.9347826086956526e-05,
|
1075 |
+
"loss": 0.0216,
|
1076 |
+
"step": 1300
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 14.24,
|
1080 |
+
"learning_rate": 2.8804347826086957e-05,
|
1081 |
+
"loss": 0.0174,
|
1082 |
+
"step": 1310
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 14.35,
|
1086 |
+
"learning_rate": 2.826086956521739e-05,
|
1087 |
+
"loss": 0.0171,
|
1088 |
+
"step": 1320
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 14.35,
|
1092 |
+
"eval_accuracy": 0.9935846030473136,
|
1093 |
+
"eval_loss": 0.04177280142903328,
|
1094 |
+
"eval_runtime": 13.9993,
|
1095 |
+
"eval_samples_per_second": 89.076,
|
1096 |
+
"eval_steps_per_second": 11.143,
|
1097 |
+
"step": 1320
|
1098 |
+
},
|
1099 |
+
{
|
1100 |
+
"epoch": 14.46,
|
1101 |
+
"learning_rate": 2.7717391304347827e-05,
|
1102 |
+
"loss": 0.0172,
|
1103 |
+
"step": 1330
|
1104 |
+
},
|
1105 |
+
{
|
1106 |
+
"epoch": 14.57,
|
1107 |
+
"learning_rate": 2.7173913043478262e-05,
|
1108 |
+
"loss": 0.0173,
|
1109 |
+
"step": 1340
|
1110 |
+
},
|
1111 |
+
{
|
1112 |
+
"epoch": 14.67,
|
1113 |
+
"learning_rate": 2.66304347826087e-05,
|
1114 |
+
"loss": 0.0259,
|
1115 |
+
"step": 1350
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 14.78,
|
1119 |
+
"learning_rate": 2.608695652173913e-05,
|
1120 |
+
"loss": 0.0197,
|
1121 |
+
"step": 1360
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 14.78,
|
1125 |
+
"eval_accuracy": 0.991980753809142,
|
1126 |
+
"eval_loss": 0.04225374758243561,
|
1127 |
+
"eval_runtime": 14.4748,
|
1128 |
+
"eval_samples_per_second": 86.15,
|
1129 |
+
"eval_steps_per_second": 10.777,
|
1130 |
+
"step": 1360
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 14.89,
|
1134 |
+
"learning_rate": 2.554347826086957e-05,
|
1135 |
+
"loss": 0.0166,
|
1136 |
+
"step": 1370
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 15.0,
|
1140 |
+
"learning_rate": 2.5e-05,
|
1141 |
+
"loss": 0.0163,
|
1142 |
+
"step": 1380
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 15.11,
|
1146 |
+
"learning_rate": 2.4456521739130436e-05,
|
1147 |
+
"loss": 0.0164,
|
1148 |
+
"step": 1390
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 15.22,
|
1152 |
+
"learning_rate": 2.391304347826087e-05,
|
1153 |
+
"loss": 0.0162,
|
1154 |
+
"step": 1400
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 15.22,
|
1158 |
+
"eval_accuracy": 0.9927826784282278,
|
1159 |
+
"eval_loss": 0.04216426983475685,
|
1160 |
+
"eval_runtime": 14.0671,
|
1161 |
+
"eval_samples_per_second": 88.646,
|
1162 |
+
"eval_steps_per_second": 11.09,
|
1163 |
+
"step": 1400
|
1164 |
+
},
|
1165 |
+
{
|
1166 |
+
"epoch": 15.33,
|
1167 |
+
"learning_rate": 2.3369565217391306e-05,
|
1168 |
+
"loss": 0.0172,
|
1169 |
+
"step": 1410
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 15.43,
|
1173 |
+
"learning_rate": 2.282608695652174e-05,
|
1174 |
+
"loss": 0.016,
|
1175 |
+
"step": 1420
|
1176 |
+
},
|
1177 |
+
{
|
1178 |
+
"epoch": 15.54,
|
1179 |
+
"learning_rate": 2.2282608695652175e-05,
|
1180 |
+
"loss": 0.0158,
|
1181 |
+
"step": 1430
|
1182 |
+
},
|
1183 |
+
{
|
1184 |
+
"epoch": 15.65,
|
1185 |
+
"learning_rate": 2.173913043478261e-05,
|
1186 |
+
"loss": 0.0159,
|
1187 |
+
"step": 1440
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 15.65,
|
1191 |
+
"eval_accuracy": 0.991980753809142,
|
1192 |
+
"eval_loss": 0.043235816061496735,
|
1193 |
+
"eval_runtime": 13.435,
|
1194 |
+
"eval_samples_per_second": 92.817,
|
1195 |
+
"eval_steps_per_second": 11.611,
|
1196 |
+
"step": 1440
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 15.76,
|
1200 |
+
"learning_rate": 2.1195652173913045e-05,
|
1201 |
+
"loss": 0.0158,
|
1202 |
+
"step": 1450
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 15.87,
|
1206 |
+
"learning_rate": 2.065217391304348e-05,
|
1207 |
+
"loss": 0.0252,
|
1208 |
+
"step": 1460
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 15.98,
|
1212 |
+
"learning_rate": 2.0108695652173915e-05,
|
1213 |
+
"loss": 0.0156,
|
1214 |
+
"step": 1470
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 16.09,
|
1218 |
+
"learning_rate": 1.956521739130435e-05,
|
1219 |
+
"loss": 0.0155,
|
1220 |
+
"step": 1480
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 16.09,
|
1224 |
+
"eval_accuracy": 0.9911788291900562,
|
1225 |
+
"eval_loss": 0.04137137532234192,
|
1226 |
+
"eval_runtime": 13.9815,
|
1227 |
+
"eval_samples_per_second": 89.189,
|
1228 |
+
"eval_steps_per_second": 11.158,
|
1229 |
+
"step": 1480
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 16.2,
|
1233 |
+
"learning_rate": 1.9021739130434784e-05,
|
1234 |
+
"loss": 0.0175,
|
1235 |
+
"step": 1490
|
1236 |
+
},
|
1237 |
+
{
|
1238 |
+
"epoch": 16.3,
|
1239 |
+
"learning_rate": 1.8478260869565216e-05,
|
1240 |
+
"loss": 0.0155,
|
1241 |
+
"step": 1500
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 16.41,
|
1245 |
+
"learning_rate": 1.793478260869565e-05,
|
1246 |
+
"loss": 0.0258,
|
1247 |
+
"step": 1510
|
1248 |
+
},
|
1249 |
+
{
|
1250 |
+
"epoch": 16.52,
|
1251 |
+
"learning_rate": 1.739130434782609e-05,
|
1252 |
+
"loss": 0.015,
|
1253 |
+
"step": 1520
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 16.52,
|
1257 |
+
"eval_accuracy": 0.9911788291900562,
|
1258 |
+
"eval_loss": 0.0487416572868824,
|
1259 |
+
"eval_runtime": 13.4779,
|
1260 |
+
"eval_samples_per_second": 92.522,
|
1261 |
+
"eval_steps_per_second": 11.575,
|
1262 |
+
"step": 1520
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 16.63,
|
1266 |
+
"learning_rate": 1.6847826086956524e-05,
|
1267 |
+
"loss": 0.0152,
|
1268 |
+
"step": 1530
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 16.74,
|
1272 |
+
"learning_rate": 1.630434782608696e-05,
|
1273 |
+
"loss": 0.0174,
|
1274 |
+
"step": 1540
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 16.85,
|
1278 |
+
"learning_rate": 1.5760869565217393e-05,
|
1279 |
+
"loss": 0.0147,
|
1280 |
+
"step": 1550
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 16.96,
|
1284 |
+
"learning_rate": 1.5217391304347828e-05,
|
1285 |
+
"loss": 0.015,
|
1286 |
+
"step": 1560
|
1287 |
+
},
|
1288 |
+
{
|
1289 |
+
"epoch": 16.96,
|
1290 |
+
"eval_accuracy": 0.991980753809142,
|
1291 |
+
"eval_loss": 0.04399973526597023,
|
1292 |
+
"eval_runtime": 14.0057,
|
1293 |
+
"eval_samples_per_second": 89.035,
|
1294 |
+
"eval_steps_per_second": 11.138,
|
1295 |
+
"step": 1560
|
1296 |
+
},
|
1297 |
+
{
|
1298 |
+
"epoch": 17.07,
|
1299 |
+
"learning_rate": 1.4673913043478263e-05,
|
1300 |
+
"loss": 0.0148,
|
1301 |
+
"step": 1570
|
1302 |
+
},
|
1303 |
+
{
|
1304 |
+
"epoch": 17.17,
|
1305 |
+
"learning_rate": 1.4130434782608694e-05,
|
1306 |
+
"loss": 0.0147,
|
1307 |
+
"step": 1580
|
1308 |
+
},
|
1309 |
+
{
|
1310 |
+
"epoch": 17.28,
|
1311 |
+
"learning_rate": 1.3586956521739131e-05,
|
1312 |
+
"loss": 0.0148,
|
1313 |
+
"step": 1590
|
1314 |
+
},
|
1315 |
+
{
|
1316 |
+
"epoch": 17.39,
|
1317 |
+
"learning_rate": 1.3043478260869566e-05,
|
1318 |
+
"loss": 0.0146,
|
1319 |
+
"step": 1600
|
1320 |
+
},
|
1321 |
+
{
|
1322 |
+
"epoch": 17.39,
|
1323 |
+
"eval_accuracy": 0.991980753809142,
|
1324 |
+
"eval_loss": 0.04343697056174278,
|
1325 |
+
"eval_runtime": 14.4989,
|
1326 |
+
"eval_samples_per_second": 86.007,
|
1327 |
+
"eval_steps_per_second": 10.759,
|
1328 |
+
"step": 1600
|
1329 |
+
},
|
1330 |
+
{
|
1331 |
+
"epoch": 17.5,
|
1332 |
+
"learning_rate": 1.25e-05,
|
1333 |
+
"loss": 0.0145,
|
1334 |
+
"step": 1610
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 17.61,
|
1338 |
+
"learning_rate": 1.1956521739130435e-05,
|
1339 |
+
"loss": 0.0144,
|
1340 |
+
"step": 1620
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 17.72,
|
1344 |
+
"learning_rate": 1.141304347826087e-05,
|
1345 |
+
"loss": 0.0149,
|
1346 |
+
"step": 1630
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 17.83,
|
1350 |
+
"learning_rate": 1.0869565217391305e-05,
|
1351 |
+
"loss": 0.0143,
|
1352 |
+
"step": 1640
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 17.83,
|
1356 |
+
"eval_accuracy": 0.991980753809142,
|
1357 |
+
"eval_loss": 0.042883455753326416,
|
1358 |
+
"eval_runtime": 14.0877,
|
1359 |
+
"eval_samples_per_second": 88.517,
|
1360 |
+
"eval_steps_per_second": 11.073,
|
1361 |
+
"step": 1640
|
1362 |
+
},
|
1363 |
+
{
|
1364 |
+
"epoch": 17.93,
|
1365 |
+
"learning_rate": 1.032608695652174e-05,
|
1366 |
+
"loss": 0.0142,
|
1367 |
+
"step": 1650
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 18.04,
|
1371 |
+
"learning_rate": 9.782608695652175e-06,
|
1372 |
+
"loss": 0.0225,
|
1373 |
+
"step": 1660
|
1374 |
+
},
|
1375 |
+
{
|
1376 |
+
"epoch": 18.15,
|
1377 |
+
"learning_rate": 9.239130434782608e-06,
|
1378 |
+
"loss": 0.0146,
|
1379 |
+
"step": 1670
|
1380 |
+
},
|
1381 |
+
{
|
1382 |
+
"epoch": 18.26,
|
1383 |
+
"learning_rate": 8.695652173913044e-06,
|
1384 |
+
"loss": 0.0143,
|
1385 |
+
"step": 1680
|
1386 |
+
},
|
1387 |
+
{
|
1388 |
+
"epoch": 18.26,
|
1389 |
+
"eval_accuracy": 0.9911788291900562,
|
1390 |
+
"eval_loss": 0.04524253308773041,
|
1391 |
+
"eval_runtime": 14.4999,
|
1392 |
+
"eval_samples_per_second": 86.001,
|
1393 |
+
"eval_steps_per_second": 10.759,
|
1394 |
+
"step": 1680
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 18.37,
|
1398 |
+
"learning_rate": 8.15217391304348e-06,
|
1399 |
+
"loss": 0.0154,
|
1400 |
+
"step": 1690
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"epoch": 18.48,
|
1404 |
+
"learning_rate": 7.608695652173914e-06,
|
1405 |
+
"loss": 0.0144,
|
1406 |
+
"step": 1700
|
1407 |
+
},
|
1408 |
+
{
|
1409 |
+
"epoch": 18.59,
|
1410 |
+
"learning_rate": 7.065217391304347e-06,
|
1411 |
+
"loss": 0.014,
|
1412 |
+
"step": 1710
|
1413 |
+
},
|
1414 |
+
{
|
1415 |
+
"epoch": 18.7,
|
1416 |
+
"learning_rate": 6.521739130434783e-06,
|
1417 |
+
"loss": 0.014,
|
1418 |
+
"step": 1720
|
1419 |
+
},
|
1420 |
+
{
|
1421 |
+
"epoch": 18.7,
|
1422 |
+
"eval_accuracy": 0.9911788291900562,
|
1423 |
+
"eval_loss": 0.04453733563423157,
|
1424 |
+
"eval_runtime": 14.0735,
|
1425 |
+
"eval_samples_per_second": 88.606,
|
1426 |
+
"eval_steps_per_second": 11.085,
|
1427 |
+
"step": 1720
|
1428 |
+
},
|
1429 |
+
{
|
1430 |
+
"epoch": 18.8,
|
1431 |
+
"learning_rate": 5.978260869565218e-06,
|
1432 |
+
"loss": 0.018,
|
1433 |
+
"step": 1730
|
1434 |
+
},
|
1435 |
+
{
|
1436 |
+
"epoch": 18.91,
|
1437 |
+
"learning_rate": 5.4347826086956525e-06,
|
1438 |
+
"loss": 0.014,
|
1439 |
+
"step": 1740
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 19.02,
|
1443 |
+
"learning_rate": 4.891304347826087e-06,
|
1444 |
+
"loss": 0.0155,
|
1445 |
+
"step": 1750
|
1446 |
+
},
|
1447 |
+
{
|
1448 |
+
"epoch": 19.13,
|
1449 |
+
"learning_rate": 4.347826086956522e-06,
|
1450 |
+
"loss": 0.0141,
|
1451 |
+
"step": 1760
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 19.13,
|
1455 |
+
"eval_accuracy": 0.9911788291900562,
|
1456 |
+
"eval_loss": 0.048826370388269424,
|
1457 |
+
"eval_runtime": 14.2162,
|
1458 |
+
"eval_samples_per_second": 87.717,
|
1459 |
+
"eval_steps_per_second": 10.973,
|
1460 |
+
"step": 1760
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 19.24,
|
1464 |
+
"learning_rate": 3.804347826086957e-06,
|
1465 |
+
"loss": 0.0139,
|
1466 |
+
"step": 1770
|
1467 |
+
},
|
1468 |
+
{
|
1469 |
+
"epoch": 19.35,
|
1470 |
+
"learning_rate": 3.2608695652173914e-06,
|
1471 |
+
"loss": 0.0139,
|
1472 |
+
"step": 1780
|
1473 |
+
},
|
1474 |
+
{
|
1475 |
+
"epoch": 19.46,
|
1476 |
+
"learning_rate": 2.7173913043478263e-06,
|
1477 |
+
"loss": 0.0139,
|
1478 |
+
"step": 1790
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 19.57,
|
1482 |
+
"learning_rate": 2.173913043478261e-06,
|
1483 |
+
"loss": 0.0138,
|
1484 |
+
"step": 1800
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"epoch": 19.57,
|
1488 |
+
"eval_accuracy": 0.9911788291900562,
|
1489 |
+
"eval_loss": 0.048504043370485306,
|
1490 |
+
"eval_runtime": 13.6564,
|
1491 |
+
"eval_samples_per_second": 91.312,
|
1492 |
+
"eval_steps_per_second": 11.423,
|
1493 |
+
"step": 1800
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 19.67,
|
1497 |
+
"learning_rate": 1.6304347826086957e-06,
|
1498 |
+
"loss": 0.0144,
|
1499 |
+
"step": 1810
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 19.78,
|
1503 |
+
"learning_rate": 1.0869565217391306e-06,
|
1504 |
+
"loss": 0.0155,
|
1505 |
+
"step": 1820
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"epoch": 19.89,
|
1509 |
+
"learning_rate": 5.434782608695653e-07,
|
1510 |
+
"loss": 0.0141,
|
1511 |
+
"step": 1830
|
1512 |
+
},
|
1513 |
+
{
|
1514 |
+
"epoch": 20.0,
|
1515 |
+
"learning_rate": 0.0,
|
1516 |
+
"loss": 0.0138,
|
1517 |
+
"step": 1840
|
1518 |
+
},
|
1519 |
+
{
|
1520 |
+
"epoch": 20.0,
|
1521 |
+
"eval_accuracy": 0.9911788291900562,
|
1522 |
+
"eval_loss": 0.0495075099170208,
|
1523 |
+
"eval_runtime": 14.095,
|
1524 |
+
"eval_samples_per_second": 88.471,
|
1525 |
+
"eval_steps_per_second": 11.068,
|
1526 |
+
"step": 1840
|
1527 |
+
},
|
1528 |
+
{
|
1529 |
+
"epoch": 20.0,
|
1530 |
+
"step": 1840,
|
1531 |
+
"total_flos": 3.419773941089157e+18,
|
1532 |
+
"train_loss": 0.33101742866894474,
|
1533 |
+
"train_runtime": 1771.2395,
|
1534 |
+
"train_samples_per_second": 24.909,
|
1535 |
+
"train_steps_per_second": 1.039
|
1536 |
+
}
|
1537 |
+
],
|
1538 |
+
"max_steps": 1840,
|
1539 |
+
"num_train_epochs": 20,
|
1540 |
+
"total_flos": 3.419773941089157e+18,
|
1541 |
+
"trial_name": null,
|
1542 |
+
"trial_params": null
|
1543 |
+
}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8e63cb3581b9d82130890094c7fc9d777da84a1f291cbc24b62807505befaa6e
|
3 |
+
size 3899
|