Upload folder using huggingface_hub
Browse files- checkpoint-594/config.json +113 -0
- checkpoint-594/optimizer.pt +3 -0
- checkpoint-594/preprocessor_config.json +9 -0
- checkpoint-594/pytorch_model.bin +3 -0
- checkpoint-594/rng_state.pth +3 -0
- checkpoint-594/scheduler.pt +3 -0
- checkpoint-594/trainer_state.json +3661 -0
- checkpoint-594/training_args.bin +3 -0
- config.json +113 -0
- preprocessor_config.json +9 -0
- pytorch_model.bin +3 -0
- training_args.bin +3 -0
checkpoint-594/config.json
ADDED
@@ -0,0 +1,113 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "facebook/wav2vec2-base-960h",
|
3 |
+
"activation_dropout": 0.1,
|
4 |
+
"adapter_kernel_size": 3,
|
5 |
+
"adapter_stride": 2,
|
6 |
+
"add_adapter": false,
|
7 |
+
"apply_spec_augment": true,
|
8 |
+
"architectures": [
|
9 |
+
"Wav2Vec2ForSequenceClassification"
|
10 |
+
],
|
11 |
+
"attention_dropout": 0.1,
|
12 |
+
"bos_token_id": 1,
|
13 |
+
"classifier_proj_size": 256,
|
14 |
+
"codevector_dim": 256,
|
15 |
+
"contrastive_logits_temperature": 0.1,
|
16 |
+
"conv_bias": false,
|
17 |
+
"conv_dim": [
|
18 |
+
512,
|
19 |
+
512,
|
20 |
+
512,
|
21 |
+
512,
|
22 |
+
512,
|
23 |
+
512,
|
24 |
+
512
|
25 |
+
],
|
26 |
+
"conv_kernel": [
|
27 |
+
10,
|
28 |
+
3,
|
29 |
+
3,
|
30 |
+
3,
|
31 |
+
3,
|
32 |
+
2,
|
33 |
+
2
|
34 |
+
],
|
35 |
+
"conv_stride": [
|
36 |
+
5,
|
37 |
+
2,
|
38 |
+
2,
|
39 |
+
2,
|
40 |
+
2,
|
41 |
+
2,
|
42 |
+
2
|
43 |
+
],
|
44 |
+
"ctc_loss_reduction": "sum",
|
45 |
+
"ctc_zero_infinity": false,
|
46 |
+
"diversity_loss_weight": 0.1,
|
47 |
+
"do_stable_layer_norm": false,
|
48 |
+
"eos_token_id": 2,
|
49 |
+
"feat_extract_activation": "gelu",
|
50 |
+
"feat_extract_dropout": 0.0,
|
51 |
+
"feat_extract_norm": "group",
|
52 |
+
"feat_proj_dropout": 0.1,
|
53 |
+
"feat_quantizer_dropout": 0.0,
|
54 |
+
"final_dropout": 0.1,
|
55 |
+
"gradient_checkpointing": false,
|
56 |
+
"hidden_act": "gelu",
|
57 |
+
"hidden_dropout": 0.1,
|
58 |
+
"hidden_dropout_prob": 0.1,
|
59 |
+
"hidden_size": 768,
|
60 |
+
"id2label": {
|
61 |
+
"0": "cat",
|
62 |
+
"1": "dog"
|
63 |
+
},
|
64 |
+
"initializer_range": 0.02,
|
65 |
+
"intermediate_size": 3072,
|
66 |
+
"layer_norm_eps": 1e-05,
|
67 |
+
"layerdrop": 0.1,
|
68 |
+
"mask_feature_length": 10,
|
69 |
+
"mask_feature_min_masks": 0,
|
70 |
+
"mask_feature_prob": 0.0,
|
71 |
+
"mask_time_length": 10,
|
72 |
+
"mask_time_min_masks": 2,
|
73 |
+
"mask_time_prob": 0.05,
|
74 |
+
"model_type": "wav2vec2",
|
75 |
+
"num_adapter_layers": 3,
|
76 |
+
"num_attention_heads": 12,
|
77 |
+
"num_codevector_groups": 2,
|
78 |
+
"num_codevectors_per_group": 320,
|
79 |
+
"num_conv_pos_embedding_groups": 16,
|
80 |
+
"num_conv_pos_embeddings": 128,
|
81 |
+
"num_feat_extract_layers": 7,
|
82 |
+
"num_hidden_layers": 12,
|
83 |
+
"num_negatives": 100,
|
84 |
+
"output_hidden_size": 768,
|
85 |
+
"pad_token_id": 0,
|
86 |
+
"proj_codevector_dim": 256,
|
87 |
+
"tdnn_dilation": [
|
88 |
+
1,
|
89 |
+
2,
|
90 |
+
3,
|
91 |
+
1,
|
92 |
+
1
|
93 |
+
],
|
94 |
+
"tdnn_dim": [
|
95 |
+
512,
|
96 |
+
512,
|
97 |
+
512,
|
98 |
+
512,
|
99 |
+
1500
|
100 |
+
],
|
101 |
+
"tdnn_kernel": [
|
102 |
+
5,
|
103 |
+
3,
|
104 |
+
3,
|
105 |
+
1,
|
106 |
+
1
|
107 |
+
],
|
108 |
+
"torch_dtype": "float32",
|
109 |
+
"transformers_version": "4.28.1",
|
110 |
+
"use_weighted_layer_sum": false,
|
111 |
+
"vocab_size": 32,
|
112 |
+
"xvector_output_dim": 512
|
113 |
+
}
|
checkpoint-594/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:87a1e9ee5da6b81fe6fe27afd9fc04412d15489f3574047c3034b6ec5273afa9
|
3 |
+
size 756677477
|
checkpoint-594/preprocessor_config.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"do_normalize": true,
|
3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
4 |
+
"feature_size": 1,
|
5 |
+
"padding_side": "right",
|
6 |
+
"padding_value": 0.0,
|
7 |
+
"return_attention_mask": false,
|
8 |
+
"sampling_rate": 16000
|
9 |
+
}
|
checkpoint-594/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1c7703931a4da1e56d95c6aa0a0b88792476b5f3de2ecae89d6337234dec2b08
|
3 |
+
size 378348563
|
checkpoint-594/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b31587509b5a4440ac695f871f03a0d3d046b09f6ddf5c5a685d16e2ab39c77e
|
3 |
+
size 14567
|
checkpoint-594/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:87126e6c4e3c926adca49c3c7813426a12d25b65fb1a19b3e4cc4187aabd6806
|
3 |
+
size 623
|
checkpoint-594/trainer_state.json
ADDED
@@ -0,0 +1,3661 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.18054482340812683,
|
3 |
+
"best_model_checkpoint": "cat_dog_sounds_classification/checkpoint-594",
|
4 |
+
"epoch": 9.0,
|
5 |
+
"global_step": 594,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.02,
|
12 |
+
"learning_rate": 1.0000000000000001e-07,
|
13 |
+
"loss": 0.7073,
|
14 |
+
"step": 1
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.03,
|
18 |
+
"learning_rate": 2.0000000000000002e-07,
|
19 |
+
"loss": 0.6877,
|
20 |
+
"step": 2
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.05,
|
24 |
+
"learning_rate": 3.0000000000000004e-07,
|
25 |
+
"loss": 0.7053,
|
26 |
+
"step": 3
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.06,
|
30 |
+
"learning_rate": 4.0000000000000003e-07,
|
31 |
+
"loss": 0.6912,
|
32 |
+
"step": 4
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.08,
|
36 |
+
"learning_rate": 5.000000000000001e-07,
|
37 |
+
"loss": 0.6998,
|
38 |
+
"step": 5
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.09,
|
42 |
+
"learning_rate": 6.000000000000001e-07,
|
43 |
+
"loss": 0.6961,
|
44 |
+
"step": 6
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.11,
|
48 |
+
"learning_rate": 7.000000000000001e-07,
|
49 |
+
"loss": 0.7068,
|
50 |
+
"step": 7
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.12,
|
54 |
+
"learning_rate": 8.000000000000001e-07,
|
55 |
+
"loss": 0.7004,
|
56 |
+
"step": 8
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.14,
|
60 |
+
"learning_rate": 9.000000000000001e-07,
|
61 |
+
"loss": 0.6771,
|
62 |
+
"step": 9
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.15,
|
66 |
+
"learning_rate": 1.0000000000000002e-06,
|
67 |
+
"loss": 0.6778,
|
68 |
+
"step": 10
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.17,
|
72 |
+
"learning_rate": 1.1e-06,
|
73 |
+
"loss": 0.6763,
|
74 |
+
"step": 11
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.18,
|
78 |
+
"learning_rate": 1.2000000000000002e-06,
|
79 |
+
"loss": 0.6894,
|
80 |
+
"step": 12
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.2,
|
84 |
+
"learning_rate": 1.3e-06,
|
85 |
+
"loss": 0.6848,
|
86 |
+
"step": 13
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.21,
|
90 |
+
"learning_rate": 1.4000000000000001e-06,
|
91 |
+
"loss": 0.6854,
|
92 |
+
"step": 14
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.23,
|
96 |
+
"learning_rate": 1.5e-06,
|
97 |
+
"loss": 0.6875,
|
98 |
+
"step": 15
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.24,
|
102 |
+
"learning_rate": 1.6000000000000001e-06,
|
103 |
+
"loss": 0.7022,
|
104 |
+
"step": 16
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.26,
|
108 |
+
"learning_rate": 1.7000000000000002e-06,
|
109 |
+
"loss": 0.6683,
|
110 |
+
"step": 17
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.27,
|
114 |
+
"learning_rate": 1.8000000000000001e-06,
|
115 |
+
"loss": 0.7008,
|
116 |
+
"step": 18
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.29,
|
120 |
+
"learning_rate": 1.9000000000000002e-06,
|
121 |
+
"loss": 0.6789,
|
122 |
+
"step": 19
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.3,
|
126 |
+
"learning_rate": 2.0000000000000003e-06,
|
127 |
+
"loss": 0.7054,
|
128 |
+
"step": 20
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.32,
|
132 |
+
"learning_rate": 2.1000000000000002e-06,
|
133 |
+
"loss": 0.6824,
|
134 |
+
"step": 21
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 0.33,
|
138 |
+
"learning_rate": 2.2e-06,
|
139 |
+
"loss": 0.6826,
|
140 |
+
"step": 22
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 0.35,
|
144 |
+
"learning_rate": 2.3000000000000004e-06,
|
145 |
+
"loss": 0.688,
|
146 |
+
"step": 23
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 0.36,
|
150 |
+
"learning_rate": 2.4000000000000003e-06,
|
151 |
+
"loss": 0.6835,
|
152 |
+
"step": 24
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 0.38,
|
156 |
+
"learning_rate": 2.5e-06,
|
157 |
+
"loss": 0.7085,
|
158 |
+
"step": 25
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 0.39,
|
162 |
+
"learning_rate": 2.6e-06,
|
163 |
+
"loss": 0.7053,
|
164 |
+
"step": 26
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 0.41,
|
168 |
+
"learning_rate": 2.7000000000000004e-06,
|
169 |
+
"loss": 0.6684,
|
170 |
+
"step": 27
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.42,
|
174 |
+
"learning_rate": 2.8000000000000003e-06,
|
175 |
+
"loss": 0.6799,
|
176 |
+
"step": 28
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 0.44,
|
180 |
+
"learning_rate": 2.9e-06,
|
181 |
+
"loss": 0.6832,
|
182 |
+
"step": 29
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 0.45,
|
186 |
+
"learning_rate": 3e-06,
|
187 |
+
"loss": 0.6846,
|
188 |
+
"step": 30
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 0.47,
|
192 |
+
"learning_rate": 3.1000000000000004e-06,
|
193 |
+
"loss": 0.6922,
|
194 |
+
"step": 31
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 0.48,
|
198 |
+
"learning_rate": 3.2000000000000003e-06,
|
199 |
+
"loss": 0.6649,
|
200 |
+
"step": 32
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 0.5,
|
204 |
+
"learning_rate": 3.3000000000000006e-06,
|
205 |
+
"loss": 0.6956,
|
206 |
+
"step": 33
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 0.52,
|
210 |
+
"learning_rate": 3.4000000000000005e-06,
|
211 |
+
"loss": 0.6748,
|
212 |
+
"step": 34
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.53,
|
216 |
+
"learning_rate": 3.5e-06,
|
217 |
+
"loss": 0.686,
|
218 |
+
"step": 35
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 0.55,
|
222 |
+
"learning_rate": 3.6000000000000003e-06,
|
223 |
+
"loss": 0.6987,
|
224 |
+
"step": 36
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 0.56,
|
228 |
+
"learning_rate": 3.7e-06,
|
229 |
+
"loss": 0.7159,
|
230 |
+
"step": 37
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 0.58,
|
234 |
+
"learning_rate": 3.8000000000000005e-06,
|
235 |
+
"loss": 0.7099,
|
236 |
+
"step": 38
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"epoch": 0.59,
|
240 |
+
"learning_rate": 3.900000000000001e-06,
|
241 |
+
"loss": 0.6821,
|
242 |
+
"step": 39
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 0.61,
|
246 |
+
"learning_rate": 4.000000000000001e-06,
|
247 |
+
"loss": 0.7196,
|
248 |
+
"step": 40
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 0.62,
|
252 |
+
"learning_rate": 4.1e-06,
|
253 |
+
"loss": 0.6929,
|
254 |
+
"step": 41
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.64,
|
258 |
+
"learning_rate": 4.2000000000000004e-06,
|
259 |
+
"loss": 0.6609,
|
260 |
+
"step": 42
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 0.65,
|
264 |
+
"learning_rate": 4.3e-06,
|
265 |
+
"loss": 0.6723,
|
266 |
+
"step": 43
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 0.67,
|
270 |
+
"learning_rate": 4.4e-06,
|
271 |
+
"loss": 0.6687,
|
272 |
+
"step": 44
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 0.68,
|
276 |
+
"learning_rate": 4.5e-06,
|
277 |
+
"loss": 0.6905,
|
278 |
+
"step": 45
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 0.7,
|
282 |
+
"learning_rate": 4.600000000000001e-06,
|
283 |
+
"loss": 0.703,
|
284 |
+
"step": 46
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 0.71,
|
288 |
+
"learning_rate": 4.7e-06,
|
289 |
+
"loss": 0.6729,
|
290 |
+
"step": 47
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 0.73,
|
294 |
+
"learning_rate": 4.800000000000001e-06,
|
295 |
+
"loss": 0.6754,
|
296 |
+
"step": 48
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.74,
|
300 |
+
"learning_rate": 4.9000000000000005e-06,
|
301 |
+
"loss": 0.6966,
|
302 |
+
"step": 49
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 0.76,
|
306 |
+
"learning_rate": 5e-06,
|
307 |
+
"loss": 0.6825,
|
308 |
+
"step": 50
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 0.77,
|
312 |
+
"learning_rate": 4.996062992125984e-06,
|
313 |
+
"loss": 0.6912,
|
314 |
+
"step": 51
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 0.79,
|
318 |
+
"learning_rate": 4.992125984251969e-06,
|
319 |
+
"loss": 0.6679,
|
320 |
+
"step": 52
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 0.8,
|
324 |
+
"learning_rate": 4.9881889763779535e-06,
|
325 |
+
"loss": 0.6755,
|
326 |
+
"step": 53
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 0.82,
|
330 |
+
"learning_rate": 4.984251968503937e-06,
|
331 |
+
"loss": 0.6728,
|
332 |
+
"step": 54
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 0.83,
|
336 |
+
"learning_rate": 4.980314960629922e-06,
|
337 |
+
"loss": 0.6923,
|
338 |
+
"step": 55
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.85,
|
342 |
+
"learning_rate": 4.976377952755906e-06,
|
343 |
+
"loss": 0.6701,
|
344 |
+
"step": 56
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 0.86,
|
348 |
+
"learning_rate": 4.97244094488189e-06,
|
349 |
+
"loss": 0.6872,
|
350 |
+
"step": 57
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 0.88,
|
354 |
+
"learning_rate": 4.968503937007874e-06,
|
355 |
+
"loss": 0.6747,
|
356 |
+
"step": 58
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 0.89,
|
360 |
+
"learning_rate": 4.964566929133859e-06,
|
361 |
+
"loss": 0.6365,
|
362 |
+
"step": 59
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 0.91,
|
366 |
+
"learning_rate": 4.960629921259843e-06,
|
367 |
+
"loss": 0.6946,
|
368 |
+
"step": 60
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 0.92,
|
372 |
+
"learning_rate": 4.956692913385827e-06,
|
373 |
+
"loss": 0.6871,
|
374 |
+
"step": 61
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 0.94,
|
378 |
+
"learning_rate": 4.952755905511811e-06,
|
379 |
+
"loss": 0.6927,
|
380 |
+
"step": 62
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.95,
|
384 |
+
"learning_rate": 4.948818897637796e-06,
|
385 |
+
"loss": 0.6347,
|
386 |
+
"step": 63
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 0.97,
|
390 |
+
"learning_rate": 4.94488188976378e-06,
|
391 |
+
"loss": 0.6879,
|
392 |
+
"step": 64
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 0.98,
|
396 |
+
"learning_rate": 4.940944881889764e-06,
|
397 |
+
"loss": 0.664,
|
398 |
+
"step": 65
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 1.0,
|
402 |
+
"learning_rate": 4.937007874015749e-06,
|
403 |
+
"loss": 0.6729,
|
404 |
+
"step": 66
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 1.0,
|
408 |
+
"eval_accuracy": 0.6060606060606061,
|
409 |
+
"eval_loss": 0.6856966614723206,
|
410 |
+
"eval_runtime": 6.2798,
|
411 |
+
"eval_samples_per_second": 10.51,
|
412 |
+
"eval_steps_per_second": 2.707,
|
413 |
+
"step": 66
|
414 |
+
},
|
415 |
+
{
|
416 |
+
"epoch": 1.02,
|
417 |
+
"learning_rate": 4.9330708661417326e-06,
|
418 |
+
"loss": 0.6889,
|
419 |
+
"step": 67
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"epoch": 1.03,
|
423 |
+
"learning_rate": 4.929133858267716e-06,
|
424 |
+
"loss": 0.6719,
|
425 |
+
"step": 68
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"epoch": 1.05,
|
429 |
+
"learning_rate": 4.925196850393702e-06,
|
430 |
+
"loss": 0.6741,
|
431 |
+
"step": 69
|
432 |
+
},
|
433 |
+
{
|
434 |
+
"epoch": 1.06,
|
435 |
+
"learning_rate": 4.921259842519686e-06,
|
436 |
+
"loss": 0.6327,
|
437 |
+
"step": 70
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 1.08,
|
441 |
+
"learning_rate": 4.9173228346456694e-06,
|
442 |
+
"loss": 0.6793,
|
443 |
+
"step": 71
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 1.09,
|
447 |
+
"learning_rate": 4.913385826771654e-06,
|
448 |
+
"loss": 0.672,
|
449 |
+
"step": 72
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 1.11,
|
453 |
+
"learning_rate": 4.909448818897638e-06,
|
454 |
+
"loss": 0.6736,
|
455 |
+
"step": 73
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 1.12,
|
459 |
+
"learning_rate": 4.9055118110236225e-06,
|
460 |
+
"loss": 0.6702,
|
461 |
+
"step": 74
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 1.14,
|
465 |
+
"learning_rate": 4.901574803149607e-06,
|
466 |
+
"loss": 0.6439,
|
467 |
+
"step": 75
|
468 |
+
},
|
469 |
+
{
|
470 |
+
"epoch": 1.15,
|
471 |
+
"learning_rate": 4.897637795275591e-06,
|
472 |
+
"loss": 0.6697,
|
473 |
+
"step": 76
|
474 |
+
},
|
475 |
+
{
|
476 |
+
"epoch": 1.17,
|
477 |
+
"learning_rate": 4.893700787401575e-06,
|
478 |
+
"loss": 0.6766,
|
479 |
+
"step": 77
|
480 |
+
},
|
481 |
+
{
|
482 |
+
"epoch": 1.18,
|
483 |
+
"learning_rate": 4.889763779527559e-06,
|
484 |
+
"loss": 0.6173,
|
485 |
+
"step": 78
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 1.2,
|
489 |
+
"learning_rate": 4.885826771653543e-06,
|
490 |
+
"loss": 0.6802,
|
491 |
+
"step": 79
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"epoch": 1.21,
|
495 |
+
"learning_rate": 4.881889763779528e-06,
|
496 |
+
"loss": 0.6312,
|
497 |
+
"step": 80
|
498 |
+
},
|
499 |
+
{
|
500 |
+
"epoch": 1.23,
|
501 |
+
"learning_rate": 4.8779527559055125e-06,
|
502 |
+
"loss": 0.6695,
|
503 |
+
"step": 81
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"epoch": 1.24,
|
507 |
+
"learning_rate": 4.874015748031496e-06,
|
508 |
+
"loss": 0.6467,
|
509 |
+
"step": 82
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 1.26,
|
513 |
+
"learning_rate": 4.870078740157481e-06,
|
514 |
+
"loss": 0.6835,
|
515 |
+
"step": 83
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 1.27,
|
519 |
+
"learning_rate": 4.866141732283465e-06,
|
520 |
+
"loss": 0.638,
|
521 |
+
"step": 84
|
522 |
+
},
|
523 |
+
{
|
524 |
+
"epoch": 1.29,
|
525 |
+
"learning_rate": 4.862204724409449e-06,
|
526 |
+
"loss": 0.6197,
|
527 |
+
"step": 85
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 1.3,
|
531 |
+
"learning_rate": 4.858267716535434e-06,
|
532 |
+
"loss": 0.6398,
|
533 |
+
"step": 86
|
534 |
+
},
|
535 |
+
{
|
536 |
+
"epoch": 1.32,
|
537 |
+
"learning_rate": 4.854330708661418e-06,
|
538 |
+
"loss": 0.6617,
|
539 |
+
"step": 87
|
540 |
+
},
|
541 |
+
{
|
542 |
+
"epoch": 1.33,
|
543 |
+
"learning_rate": 4.850393700787402e-06,
|
544 |
+
"loss": 0.6413,
|
545 |
+
"step": 88
|
546 |
+
},
|
547 |
+
{
|
548 |
+
"epoch": 1.35,
|
549 |
+
"learning_rate": 4.846456692913386e-06,
|
550 |
+
"loss": 0.6386,
|
551 |
+
"step": 89
|
552 |
+
},
|
553 |
+
{
|
554 |
+
"epoch": 1.36,
|
555 |
+
"learning_rate": 4.84251968503937e-06,
|
556 |
+
"loss": 0.6712,
|
557 |
+
"step": 90
|
558 |
+
},
|
559 |
+
{
|
560 |
+
"epoch": 1.38,
|
561 |
+
"learning_rate": 4.838582677165355e-06,
|
562 |
+
"loss": 0.6349,
|
563 |
+
"step": 91
|
564 |
+
},
|
565 |
+
{
|
566 |
+
"epoch": 1.39,
|
567 |
+
"learning_rate": 4.834645669291339e-06,
|
568 |
+
"loss": 0.659,
|
569 |
+
"step": 92
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 1.41,
|
573 |
+
"learning_rate": 4.830708661417323e-06,
|
574 |
+
"loss": 0.6665,
|
575 |
+
"step": 93
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"epoch": 1.42,
|
579 |
+
"learning_rate": 4.826771653543308e-06,
|
580 |
+
"loss": 0.6042,
|
581 |
+
"step": 94
|
582 |
+
},
|
583 |
+
{
|
584 |
+
"epoch": 1.44,
|
585 |
+
"learning_rate": 4.8228346456692916e-06,
|
586 |
+
"loss": 0.6762,
|
587 |
+
"step": 95
|
588 |
+
},
|
589 |
+
{
|
590 |
+
"epoch": 1.45,
|
591 |
+
"learning_rate": 4.818897637795276e-06,
|
592 |
+
"loss": 0.5887,
|
593 |
+
"step": 96
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 1.47,
|
597 |
+
"learning_rate": 4.81496062992126e-06,
|
598 |
+
"loss": 0.6614,
|
599 |
+
"step": 97
|
600 |
+
},
|
601 |
+
{
|
602 |
+
"epoch": 1.48,
|
603 |
+
"learning_rate": 4.811023622047245e-06,
|
604 |
+
"loss": 0.6639,
|
605 |
+
"step": 98
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"epoch": 1.5,
|
609 |
+
"learning_rate": 4.8070866141732284e-06,
|
610 |
+
"loss": 0.6283,
|
611 |
+
"step": 99
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 1.52,
|
615 |
+
"learning_rate": 4.803149606299213e-06,
|
616 |
+
"loss": 0.6996,
|
617 |
+
"step": 100
|
618 |
+
},
|
619 |
+
{
|
620 |
+
"epoch": 1.53,
|
621 |
+
"learning_rate": 4.799212598425198e-06,
|
622 |
+
"loss": 0.6321,
|
623 |
+
"step": 101
|
624 |
+
},
|
625 |
+
{
|
626 |
+
"epoch": 1.55,
|
627 |
+
"learning_rate": 4.7952755905511815e-06,
|
628 |
+
"loss": 0.6023,
|
629 |
+
"step": 102
|
630 |
+
},
|
631 |
+
{
|
632 |
+
"epoch": 1.56,
|
633 |
+
"learning_rate": 4.791338582677166e-06,
|
634 |
+
"loss": 0.6921,
|
635 |
+
"step": 103
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"epoch": 1.58,
|
639 |
+
"learning_rate": 4.78740157480315e-06,
|
640 |
+
"loss": 0.6572,
|
641 |
+
"step": 104
|
642 |
+
},
|
643 |
+
{
|
644 |
+
"epoch": 1.59,
|
645 |
+
"learning_rate": 4.783464566929134e-06,
|
646 |
+
"loss": 0.6463,
|
647 |
+
"step": 105
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 1.61,
|
651 |
+
"learning_rate": 4.779527559055118e-06,
|
652 |
+
"loss": 0.5887,
|
653 |
+
"step": 106
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 1.62,
|
657 |
+
"learning_rate": 4.775590551181103e-06,
|
658 |
+
"loss": 0.6794,
|
659 |
+
"step": 107
|
660 |
+
},
|
661 |
+
{
|
662 |
+
"epoch": 1.64,
|
663 |
+
"learning_rate": 4.771653543307087e-06,
|
664 |
+
"loss": 0.6947,
|
665 |
+
"step": 108
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"epoch": 1.65,
|
669 |
+
"learning_rate": 4.7677165354330715e-06,
|
670 |
+
"loss": 0.6205,
|
671 |
+
"step": 109
|
672 |
+
},
|
673 |
+
{
|
674 |
+
"epoch": 1.67,
|
675 |
+
"learning_rate": 4.763779527559055e-06,
|
676 |
+
"loss": 0.6743,
|
677 |
+
"step": 110
|
678 |
+
},
|
679 |
+
{
|
680 |
+
"epoch": 1.68,
|
681 |
+
"learning_rate": 4.75984251968504e-06,
|
682 |
+
"loss": 0.6543,
|
683 |
+
"step": 111
|
684 |
+
},
|
685 |
+
{
|
686 |
+
"epoch": 1.7,
|
687 |
+
"learning_rate": 4.7559055118110246e-06,
|
688 |
+
"loss": 0.6636,
|
689 |
+
"step": 112
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"epoch": 1.71,
|
693 |
+
"learning_rate": 4.751968503937008e-06,
|
694 |
+
"loss": 0.6826,
|
695 |
+
"step": 113
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 1.73,
|
699 |
+
"learning_rate": 4.748031496062992e-06,
|
700 |
+
"loss": 0.6454,
|
701 |
+
"step": 114
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 1.74,
|
705 |
+
"learning_rate": 4.744094488188977e-06,
|
706 |
+
"loss": 0.6453,
|
707 |
+
"step": 115
|
708 |
+
},
|
709 |
+
{
|
710 |
+
"epoch": 1.76,
|
711 |
+
"learning_rate": 4.740157480314961e-06,
|
712 |
+
"loss": 0.681,
|
713 |
+
"step": 116
|
714 |
+
},
|
715 |
+
{
|
716 |
+
"epoch": 1.77,
|
717 |
+
"learning_rate": 4.736220472440945e-06,
|
718 |
+
"loss": 0.5993,
|
719 |
+
"step": 117
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 1.79,
|
723 |
+
"learning_rate": 4.73228346456693e-06,
|
724 |
+
"loss": 0.6114,
|
725 |
+
"step": 118
|
726 |
+
},
|
727 |
+
{
|
728 |
+
"epoch": 1.8,
|
729 |
+
"learning_rate": 4.728346456692914e-06,
|
730 |
+
"loss": 0.6765,
|
731 |
+
"step": 119
|
732 |
+
},
|
733 |
+
{
|
734 |
+
"epoch": 1.82,
|
735 |
+
"learning_rate": 4.724409448818898e-06,
|
736 |
+
"loss": 0.7033,
|
737 |
+
"step": 120
|
738 |
+
},
|
739 |
+
{
|
740 |
+
"epoch": 1.83,
|
741 |
+
"learning_rate": 4.720472440944882e-06,
|
742 |
+
"loss": 0.64,
|
743 |
+
"step": 121
|
744 |
+
},
|
745 |
+
{
|
746 |
+
"epoch": 1.85,
|
747 |
+
"learning_rate": 4.716535433070866e-06,
|
748 |
+
"loss": 0.6224,
|
749 |
+
"step": 122
|
750 |
+
},
|
751 |
+
{
|
752 |
+
"epoch": 1.86,
|
753 |
+
"learning_rate": 4.7125984251968506e-06,
|
754 |
+
"loss": 0.5752,
|
755 |
+
"step": 123
|
756 |
+
},
|
757 |
+
{
|
758 |
+
"epoch": 1.88,
|
759 |
+
"learning_rate": 4.708661417322835e-06,
|
760 |
+
"loss": 0.6047,
|
761 |
+
"step": 124
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"epoch": 1.89,
|
765 |
+
"learning_rate": 4.704724409448819e-06,
|
766 |
+
"loss": 0.6562,
|
767 |
+
"step": 125
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 1.91,
|
771 |
+
"learning_rate": 4.700787401574804e-06,
|
772 |
+
"loss": 0.6247,
|
773 |
+
"step": 126
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 1.92,
|
777 |
+
"learning_rate": 4.6968503937007874e-06,
|
778 |
+
"loss": 0.6994,
|
779 |
+
"step": 127
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 1.94,
|
783 |
+
"learning_rate": 4.692913385826772e-06,
|
784 |
+
"loss": 0.7222,
|
785 |
+
"step": 128
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 1.95,
|
789 |
+
"learning_rate": 4.688976377952757e-06,
|
790 |
+
"loss": 0.6639,
|
791 |
+
"step": 129
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 1.97,
|
795 |
+
"learning_rate": 4.6850393700787405e-06,
|
796 |
+
"loss": 0.6009,
|
797 |
+
"step": 130
|
798 |
+
},
|
799 |
+
{
|
800 |
+
"epoch": 1.98,
|
801 |
+
"learning_rate": 4.681102362204724e-06,
|
802 |
+
"loss": 0.6718,
|
803 |
+
"step": 131
|
804 |
+
},
|
805 |
+
{
|
806 |
+
"epoch": 2.0,
|
807 |
+
"learning_rate": 4.677165354330709e-06,
|
808 |
+
"loss": 0.5929,
|
809 |
+
"step": 132
|
810 |
+
},
|
811 |
+
{
|
812 |
+
"epoch": 2.0,
|
813 |
+
"eval_accuracy": 0.7121212121212122,
|
814 |
+
"eval_loss": 0.6547654867172241,
|
815 |
+
"eval_runtime": 6.4368,
|
816 |
+
"eval_samples_per_second": 10.254,
|
817 |
+
"eval_steps_per_second": 2.641,
|
818 |
+
"step": 132
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 2.02,
|
822 |
+
"learning_rate": 4.673228346456694e-06,
|
823 |
+
"loss": 0.5657,
|
824 |
+
"step": 133
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 2.03,
|
828 |
+
"learning_rate": 4.669291338582677e-06,
|
829 |
+
"loss": 0.6252,
|
830 |
+
"step": 134
|
831 |
+
},
|
832 |
+
{
|
833 |
+
"epoch": 2.05,
|
834 |
+
"learning_rate": 4.665354330708662e-06,
|
835 |
+
"loss": 0.5993,
|
836 |
+
"step": 135
|
837 |
+
},
|
838 |
+
{
|
839 |
+
"epoch": 2.06,
|
840 |
+
"learning_rate": 4.661417322834646e-06,
|
841 |
+
"loss": 0.6276,
|
842 |
+
"step": 136
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 2.08,
|
846 |
+
"learning_rate": 4.6574803149606305e-06,
|
847 |
+
"loss": 0.5883,
|
848 |
+
"step": 137
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 2.09,
|
852 |
+
"learning_rate": 4.653543307086614e-06,
|
853 |
+
"loss": 0.6206,
|
854 |
+
"step": 138
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 2.11,
|
858 |
+
"learning_rate": 4.649606299212599e-06,
|
859 |
+
"loss": 0.5486,
|
860 |
+
"step": 139
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 2.12,
|
864 |
+
"learning_rate": 4.645669291338583e-06,
|
865 |
+
"loss": 0.5415,
|
866 |
+
"step": 140
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 2.14,
|
870 |
+
"learning_rate": 4.641732283464567e-06,
|
871 |
+
"loss": 0.6378,
|
872 |
+
"step": 141
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 2.15,
|
876 |
+
"learning_rate": 4.637795275590551e-06,
|
877 |
+
"loss": 0.5229,
|
878 |
+
"step": 142
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 2.17,
|
882 |
+
"learning_rate": 4.633858267716536e-06,
|
883 |
+
"loss": 0.6823,
|
884 |
+
"step": 143
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 2.18,
|
888 |
+
"learning_rate": 4.6299212598425204e-06,
|
889 |
+
"loss": 0.7012,
|
890 |
+
"step": 144
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 2.2,
|
894 |
+
"learning_rate": 4.625984251968504e-06,
|
895 |
+
"loss": 0.6254,
|
896 |
+
"step": 145
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 2.21,
|
900 |
+
"learning_rate": 4.622047244094489e-06,
|
901 |
+
"loss": 0.6011,
|
902 |
+
"step": 146
|
903 |
+
},
|
904 |
+
{
|
905 |
+
"epoch": 2.23,
|
906 |
+
"learning_rate": 4.618110236220473e-06,
|
907 |
+
"loss": 0.5934,
|
908 |
+
"step": 147
|
909 |
+
},
|
910 |
+
{
|
911 |
+
"epoch": 2.24,
|
912 |
+
"learning_rate": 4.6141732283464565e-06,
|
913 |
+
"loss": 0.7028,
|
914 |
+
"step": 148
|
915 |
+
},
|
916 |
+
{
|
917 |
+
"epoch": 2.26,
|
918 |
+
"learning_rate": 4.610236220472442e-06,
|
919 |
+
"loss": 0.5105,
|
920 |
+
"step": 149
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 2.27,
|
924 |
+
"learning_rate": 4.606299212598426e-06,
|
925 |
+
"loss": 0.6125,
|
926 |
+
"step": 150
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 2.29,
|
930 |
+
"learning_rate": 4.6023622047244096e-06,
|
931 |
+
"loss": 0.6041,
|
932 |
+
"step": 151
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 2.3,
|
936 |
+
"learning_rate": 4.598425196850394e-06,
|
937 |
+
"loss": 0.5185,
|
938 |
+
"step": 152
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 2.32,
|
942 |
+
"learning_rate": 4.594488188976378e-06,
|
943 |
+
"loss": 0.6009,
|
944 |
+
"step": 153
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 2.33,
|
948 |
+
"learning_rate": 4.590551181102363e-06,
|
949 |
+
"loss": 0.6171,
|
950 |
+
"step": 154
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 2.35,
|
954 |
+
"learning_rate": 4.586614173228347e-06,
|
955 |
+
"loss": 0.5724,
|
956 |
+
"step": 155
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 2.36,
|
960 |
+
"learning_rate": 4.582677165354331e-06,
|
961 |
+
"loss": 0.5557,
|
962 |
+
"step": 156
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 2.38,
|
966 |
+
"learning_rate": 4.578740157480316e-06,
|
967 |
+
"loss": 0.5141,
|
968 |
+
"step": 157
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 2.39,
|
972 |
+
"learning_rate": 4.5748031496062995e-06,
|
973 |
+
"loss": 0.5375,
|
974 |
+
"step": 158
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 2.41,
|
978 |
+
"learning_rate": 4.570866141732283e-06,
|
979 |
+
"loss": 0.5167,
|
980 |
+
"step": 159
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 2.42,
|
984 |
+
"learning_rate": 4.566929133858268e-06,
|
985 |
+
"loss": 0.7095,
|
986 |
+
"step": 160
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 2.44,
|
990 |
+
"learning_rate": 4.562992125984253e-06,
|
991 |
+
"loss": 0.6252,
|
992 |
+
"step": 161
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 2.45,
|
996 |
+
"learning_rate": 4.559055118110236e-06,
|
997 |
+
"loss": 0.5686,
|
998 |
+
"step": 162
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 2.47,
|
1002 |
+
"learning_rate": 4.555118110236221e-06,
|
1003 |
+
"loss": 0.5789,
|
1004 |
+
"step": 163
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 2.48,
|
1008 |
+
"learning_rate": 4.551181102362205e-06,
|
1009 |
+
"loss": 0.4726,
|
1010 |
+
"step": 164
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 2.5,
|
1014 |
+
"learning_rate": 4.5472440944881895e-06,
|
1015 |
+
"loss": 0.6467,
|
1016 |
+
"step": 165
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 2.52,
|
1020 |
+
"learning_rate": 4.543307086614174e-06,
|
1021 |
+
"loss": 0.4756,
|
1022 |
+
"step": 166
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 2.53,
|
1026 |
+
"learning_rate": 4.539370078740158e-06,
|
1027 |
+
"loss": 0.5845,
|
1028 |
+
"step": 167
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 2.55,
|
1032 |
+
"learning_rate": 4.535433070866142e-06,
|
1033 |
+
"loss": 0.5993,
|
1034 |
+
"step": 168
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 2.56,
|
1038 |
+
"learning_rate": 4.531496062992126e-06,
|
1039 |
+
"loss": 0.6704,
|
1040 |
+
"step": 169
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 2.58,
|
1044 |
+
"learning_rate": 4.52755905511811e-06,
|
1045 |
+
"loss": 0.5848,
|
1046 |
+
"step": 170
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 2.59,
|
1050 |
+
"learning_rate": 4.523622047244095e-06,
|
1051 |
+
"loss": 0.459,
|
1052 |
+
"step": 171
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 2.61,
|
1056 |
+
"learning_rate": 4.5196850393700794e-06,
|
1057 |
+
"loss": 0.7056,
|
1058 |
+
"step": 172
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 2.62,
|
1062 |
+
"learning_rate": 4.515748031496063e-06,
|
1063 |
+
"loss": 0.6121,
|
1064 |
+
"step": 173
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 2.64,
|
1068 |
+
"learning_rate": 4.511811023622048e-06,
|
1069 |
+
"loss": 0.641,
|
1070 |
+
"step": 174
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 2.65,
|
1074 |
+
"learning_rate": 4.507874015748032e-06,
|
1075 |
+
"loss": 0.5589,
|
1076 |
+
"step": 175
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 2.67,
|
1080 |
+
"learning_rate": 4.503937007874016e-06,
|
1081 |
+
"loss": 0.4841,
|
1082 |
+
"step": 176
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 2.68,
|
1086 |
+
"learning_rate": 4.5e-06,
|
1087 |
+
"loss": 0.5499,
|
1088 |
+
"step": 177
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 2.7,
|
1092 |
+
"learning_rate": 4.496062992125985e-06,
|
1093 |
+
"loss": 0.6007,
|
1094 |
+
"step": 178
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 2.71,
|
1098 |
+
"learning_rate": 4.4921259842519686e-06,
|
1099 |
+
"loss": 0.6004,
|
1100 |
+
"step": 179
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 2.73,
|
1104 |
+
"learning_rate": 4.488188976377953e-06,
|
1105 |
+
"loss": 0.5338,
|
1106 |
+
"step": 180
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 2.74,
|
1110 |
+
"learning_rate": 4.484251968503938e-06,
|
1111 |
+
"loss": 0.4604,
|
1112 |
+
"step": 181
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 2.76,
|
1116 |
+
"learning_rate": 4.480314960629922e-06,
|
1117 |
+
"loss": 0.5016,
|
1118 |
+
"step": 182
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 2.77,
|
1122 |
+
"learning_rate": 4.476377952755906e-06,
|
1123 |
+
"loss": 0.477,
|
1124 |
+
"step": 183
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 2.79,
|
1128 |
+
"learning_rate": 4.47244094488189e-06,
|
1129 |
+
"loss": 0.5267,
|
1130 |
+
"step": 184
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 2.8,
|
1134 |
+
"learning_rate": 4.468503937007874e-06,
|
1135 |
+
"loss": 0.5974,
|
1136 |
+
"step": 185
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 2.82,
|
1140 |
+
"learning_rate": 4.4645669291338585e-06,
|
1141 |
+
"loss": 0.6451,
|
1142 |
+
"step": 186
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 2.83,
|
1146 |
+
"learning_rate": 4.460629921259843e-06,
|
1147 |
+
"loss": 0.4955,
|
1148 |
+
"step": 187
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 2.85,
|
1152 |
+
"learning_rate": 4.456692913385827e-06,
|
1153 |
+
"loss": 0.5164,
|
1154 |
+
"step": 188
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 2.86,
|
1158 |
+
"learning_rate": 4.452755905511812e-06,
|
1159 |
+
"loss": 0.6009,
|
1160 |
+
"step": 189
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"epoch": 2.88,
|
1164 |
+
"learning_rate": 4.448818897637795e-06,
|
1165 |
+
"loss": 0.4405,
|
1166 |
+
"step": 190
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 2.89,
|
1170 |
+
"learning_rate": 4.44488188976378e-06,
|
1171 |
+
"loss": 0.6577,
|
1172 |
+
"step": 191
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 2.91,
|
1176 |
+
"learning_rate": 4.440944881889765e-06,
|
1177 |
+
"loss": 0.5552,
|
1178 |
+
"step": 192
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 2.92,
|
1182 |
+
"learning_rate": 4.4370078740157485e-06,
|
1183 |
+
"loss": 0.4946,
|
1184 |
+
"step": 193
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 2.94,
|
1188 |
+
"learning_rate": 4.433070866141732e-06,
|
1189 |
+
"loss": 0.6111,
|
1190 |
+
"step": 194
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 2.95,
|
1194 |
+
"learning_rate": 4.429133858267717e-06,
|
1195 |
+
"loss": 0.6654,
|
1196 |
+
"step": 195
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 2.97,
|
1200 |
+
"learning_rate": 4.425196850393701e-06,
|
1201 |
+
"loss": 0.5907,
|
1202 |
+
"step": 196
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 2.98,
|
1206 |
+
"learning_rate": 4.421259842519685e-06,
|
1207 |
+
"loss": 0.4857,
|
1208 |
+
"step": 197
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 3.0,
|
1212 |
+
"learning_rate": 4.41732283464567e-06,
|
1213 |
+
"loss": 0.3732,
|
1214 |
+
"step": 198
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 3.0,
|
1218 |
+
"eval_accuracy": 0.803030303030303,
|
1219 |
+
"eval_loss": 0.567893385887146,
|
1220 |
+
"eval_runtime": 6.3298,
|
1221 |
+
"eval_samples_per_second": 10.427,
|
1222 |
+
"eval_steps_per_second": 2.686,
|
1223 |
+
"step": 198
|
1224 |
+
},
|
1225 |
+
{
|
1226 |
+
"epoch": 3.02,
|
1227 |
+
"learning_rate": 4.413385826771654e-06,
|
1228 |
+
"loss": 0.5889,
|
1229 |
+
"step": 199
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 3.03,
|
1233 |
+
"learning_rate": 4.4094488188976384e-06,
|
1234 |
+
"loss": 0.5622,
|
1235 |
+
"step": 200
|
1236 |
+
},
|
1237 |
+
{
|
1238 |
+
"epoch": 3.05,
|
1239 |
+
"learning_rate": 4.405511811023622e-06,
|
1240 |
+
"loss": 0.5293,
|
1241 |
+
"step": 201
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 3.06,
|
1245 |
+
"learning_rate": 4.401574803149606e-06,
|
1246 |
+
"loss": 0.5314,
|
1247 |
+
"step": 202
|
1248 |
+
},
|
1249 |
+
{
|
1250 |
+
"epoch": 3.08,
|
1251 |
+
"learning_rate": 4.397637795275591e-06,
|
1252 |
+
"loss": 0.558,
|
1253 |
+
"step": 203
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 3.09,
|
1257 |
+
"learning_rate": 4.393700787401575e-06,
|
1258 |
+
"loss": 0.6215,
|
1259 |
+
"step": 204
|
1260 |
+
},
|
1261 |
+
{
|
1262 |
+
"epoch": 3.11,
|
1263 |
+
"learning_rate": 4.389763779527559e-06,
|
1264 |
+
"loss": 0.5632,
|
1265 |
+
"step": 205
|
1266 |
+
},
|
1267 |
+
{
|
1268 |
+
"epoch": 3.12,
|
1269 |
+
"learning_rate": 4.385826771653544e-06,
|
1270 |
+
"loss": 0.4325,
|
1271 |
+
"step": 206
|
1272 |
+
},
|
1273 |
+
{
|
1274 |
+
"epoch": 3.14,
|
1275 |
+
"learning_rate": 4.3818897637795276e-06,
|
1276 |
+
"loss": 0.4294,
|
1277 |
+
"step": 207
|
1278 |
+
},
|
1279 |
+
{
|
1280 |
+
"epoch": 3.15,
|
1281 |
+
"learning_rate": 4.377952755905512e-06,
|
1282 |
+
"loss": 0.5966,
|
1283 |
+
"step": 208
|
1284 |
+
},
|
1285 |
+
{
|
1286 |
+
"epoch": 3.17,
|
1287 |
+
"learning_rate": 4.374015748031497e-06,
|
1288 |
+
"loss": 0.4259,
|
1289 |
+
"step": 209
|
1290 |
+
},
|
1291 |
+
{
|
1292 |
+
"epoch": 3.18,
|
1293 |
+
"learning_rate": 4.370078740157481e-06,
|
1294 |
+
"loss": 0.5215,
|
1295 |
+
"step": 210
|
1296 |
+
},
|
1297 |
+
{
|
1298 |
+
"epoch": 3.2,
|
1299 |
+
"learning_rate": 4.3661417322834644e-06,
|
1300 |
+
"loss": 0.5925,
|
1301 |
+
"step": 211
|
1302 |
+
},
|
1303 |
+
{
|
1304 |
+
"epoch": 3.21,
|
1305 |
+
"learning_rate": 4.362204724409449e-06,
|
1306 |
+
"loss": 0.6151,
|
1307 |
+
"step": 212
|
1308 |
+
},
|
1309 |
+
{
|
1310 |
+
"epoch": 3.23,
|
1311 |
+
"learning_rate": 4.358267716535434e-06,
|
1312 |
+
"loss": 0.4599,
|
1313 |
+
"step": 213
|
1314 |
+
},
|
1315 |
+
{
|
1316 |
+
"epoch": 3.24,
|
1317 |
+
"learning_rate": 4.3543307086614175e-06,
|
1318 |
+
"loss": 0.4949,
|
1319 |
+
"step": 214
|
1320 |
+
},
|
1321 |
+
{
|
1322 |
+
"epoch": 3.26,
|
1323 |
+
"learning_rate": 4.350393700787402e-06,
|
1324 |
+
"loss": 0.5712,
|
1325 |
+
"step": 215
|
1326 |
+
},
|
1327 |
+
{
|
1328 |
+
"epoch": 3.27,
|
1329 |
+
"learning_rate": 4.346456692913386e-06,
|
1330 |
+
"loss": 0.4125,
|
1331 |
+
"step": 216
|
1332 |
+
},
|
1333 |
+
{
|
1334 |
+
"epoch": 3.29,
|
1335 |
+
"learning_rate": 4.342519685039371e-06,
|
1336 |
+
"loss": 0.5483,
|
1337 |
+
"step": 217
|
1338 |
+
},
|
1339 |
+
{
|
1340 |
+
"epoch": 3.3,
|
1341 |
+
"learning_rate": 4.338582677165354e-06,
|
1342 |
+
"loss": 0.4794,
|
1343 |
+
"step": 218
|
1344 |
+
},
|
1345 |
+
{
|
1346 |
+
"epoch": 3.32,
|
1347 |
+
"learning_rate": 4.334645669291339e-06,
|
1348 |
+
"loss": 0.5533,
|
1349 |
+
"step": 219
|
1350 |
+
},
|
1351 |
+
{
|
1352 |
+
"epoch": 3.33,
|
1353 |
+
"learning_rate": 4.330708661417324e-06,
|
1354 |
+
"loss": 0.6201,
|
1355 |
+
"step": 220
|
1356 |
+
},
|
1357 |
+
{
|
1358 |
+
"epoch": 3.35,
|
1359 |
+
"learning_rate": 4.3267716535433075e-06,
|
1360 |
+
"loss": 0.3873,
|
1361 |
+
"step": 221
|
1362 |
+
},
|
1363 |
+
{
|
1364 |
+
"epoch": 3.36,
|
1365 |
+
"learning_rate": 4.322834645669291e-06,
|
1366 |
+
"loss": 0.3827,
|
1367 |
+
"step": 222
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 3.38,
|
1371 |
+
"learning_rate": 4.318897637795276e-06,
|
1372 |
+
"loss": 0.5698,
|
1373 |
+
"step": 223
|
1374 |
+
},
|
1375 |
+
{
|
1376 |
+
"epoch": 3.39,
|
1377 |
+
"learning_rate": 4.3149606299212606e-06,
|
1378 |
+
"loss": 0.3458,
|
1379 |
+
"step": 224
|
1380 |
+
},
|
1381 |
+
{
|
1382 |
+
"epoch": 3.41,
|
1383 |
+
"learning_rate": 4.311023622047244e-06,
|
1384 |
+
"loss": 0.5704,
|
1385 |
+
"step": 225
|
1386 |
+
},
|
1387 |
+
{
|
1388 |
+
"epoch": 3.42,
|
1389 |
+
"learning_rate": 4.307086614173229e-06,
|
1390 |
+
"loss": 0.432,
|
1391 |
+
"step": 226
|
1392 |
+
},
|
1393 |
+
{
|
1394 |
+
"epoch": 3.44,
|
1395 |
+
"learning_rate": 4.303149606299213e-06,
|
1396 |
+
"loss": 0.706,
|
1397 |
+
"step": 227
|
1398 |
+
},
|
1399 |
+
{
|
1400 |
+
"epoch": 3.45,
|
1401 |
+
"learning_rate": 4.299212598425197e-06,
|
1402 |
+
"loss": 0.5699,
|
1403 |
+
"step": 228
|
1404 |
+
},
|
1405 |
+
{
|
1406 |
+
"epoch": 3.47,
|
1407 |
+
"learning_rate": 4.295275590551182e-06,
|
1408 |
+
"loss": 0.5042,
|
1409 |
+
"step": 229
|
1410 |
+
},
|
1411 |
+
{
|
1412 |
+
"epoch": 3.48,
|
1413 |
+
"learning_rate": 4.291338582677166e-06,
|
1414 |
+
"loss": 0.3555,
|
1415 |
+
"step": 230
|
1416 |
+
},
|
1417 |
+
{
|
1418 |
+
"epoch": 3.5,
|
1419 |
+
"learning_rate": 4.28740157480315e-06,
|
1420 |
+
"loss": 0.4435,
|
1421 |
+
"step": 231
|
1422 |
+
},
|
1423 |
+
{
|
1424 |
+
"epoch": 3.52,
|
1425 |
+
"learning_rate": 4.283464566929134e-06,
|
1426 |
+
"loss": 0.516,
|
1427 |
+
"step": 232
|
1428 |
+
},
|
1429 |
+
{
|
1430 |
+
"epoch": 3.53,
|
1431 |
+
"learning_rate": 4.279527559055118e-06,
|
1432 |
+
"loss": 0.3766,
|
1433 |
+
"step": 233
|
1434 |
+
},
|
1435 |
+
{
|
1436 |
+
"epoch": 3.55,
|
1437 |
+
"learning_rate": 4.275590551181103e-06,
|
1438 |
+
"loss": 0.3667,
|
1439 |
+
"step": 234
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 3.56,
|
1443 |
+
"learning_rate": 4.271653543307087e-06,
|
1444 |
+
"loss": 0.4945,
|
1445 |
+
"step": 235
|
1446 |
+
},
|
1447 |
+
{
|
1448 |
+
"epoch": 3.58,
|
1449 |
+
"learning_rate": 4.267716535433071e-06,
|
1450 |
+
"loss": 0.3392,
|
1451 |
+
"step": 236
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 3.59,
|
1455 |
+
"learning_rate": 4.263779527559056e-06,
|
1456 |
+
"loss": 0.5006,
|
1457 |
+
"step": 237
|
1458 |
+
},
|
1459 |
+
{
|
1460 |
+
"epoch": 3.61,
|
1461 |
+
"learning_rate": 4.25984251968504e-06,
|
1462 |
+
"loss": 0.3513,
|
1463 |
+
"step": 238
|
1464 |
+
},
|
1465 |
+
{
|
1466 |
+
"epoch": 3.62,
|
1467 |
+
"learning_rate": 4.2559055118110234e-06,
|
1468 |
+
"loss": 0.5244,
|
1469 |
+
"step": 239
|
1470 |
+
},
|
1471 |
+
{
|
1472 |
+
"epoch": 3.64,
|
1473 |
+
"learning_rate": 4.251968503937008e-06,
|
1474 |
+
"loss": 0.5114,
|
1475 |
+
"step": 240
|
1476 |
+
},
|
1477 |
+
{
|
1478 |
+
"epoch": 3.65,
|
1479 |
+
"learning_rate": 4.248031496062993e-06,
|
1480 |
+
"loss": 0.2958,
|
1481 |
+
"step": 241
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 3.67,
|
1485 |
+
"learning_rate": 4.2440944881889765e-06,
|
1486 |
+
"loss": 0.3114,
|
1487 |
+
"step": 242
|
1488 |
+
},
|
1489 |
+
{
|
1490 |
+
"epoch": 3.68,
|
1491 |
+
"learning_rate": 4.240157480314961e-06,
|
1492 |
+
"loss": 0.3446,
|
1493 |
+
"step": 243
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 3.7,
|
1497 |
+
"learning_rate": 4.236220472440945e-06,
|
1498 |
+
"loss": 0.4279,
|
1499 |
+
"step": 244
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 3.71,
|
1503 |
+
"learning_rate": 4.23228346456693e-06,
|
1504 |
+
"loss": 0.3878,
|
1505 |
+
"step": 245
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"epoch": 3.73,
|
1509 |
+
"learning_rate": 4.228346456692914e-06,
|
1510 |
+
"loss": 0.7247,
|
1511 |
+
"step": 246
|
1512 |
+
},
|
1513 |
+
{
|
1514 |
+
"epoch": 3.74,
|
1515 |
+
"learning_rate": 4.224409448818898e-06,
|
1516 |
+
"loss": 0.3515,
|
1517 |
+
"step": 247
|
1518 |
+
},
|
1519 |
+
{
|
1520 |
+
"epoch": 3.76,
|
1521 |
+
"learning_rate": 4.220472440944882e-06,
|
1522 |
+
"loss": 0.4386,
|
1523 |
+
"step": 248
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 3.77,
|
1527 |
+
"learning_rate": 4.2165354330708665e-06,
|
1528 |
+
"loss": 0.3985,
|
1529 |
+
"step": 249
|
1530 |
+
},
|
1531 |
+
{
|
1532 |
+
"epoch": 3.79,
|
1533 |
+
"learning_rate": 4.21259842519685e-06,
|
1534 |
+
"loss": 0.2859,
|
1535 |
+
"step": 250
|
1536 |
+
},
|
1537 |
+
{
|
1538 |
+
"epoch": 3.8,
|
1539 |
+
"learning_rate": 4.208661417322835e-06,
|
1540 |
+
"loss": 0.6352,
|
1541 |
+
"step": 251
|
1542 |
+
},
|
1543 |
+
{
|
1544 |
+
"epoch": 3.82,
|
1545 |
+
"learning_rate": 4.2047244094488196e-06,
|
1546 |
+
"loss": 0.3076,
|
1547 |
+
"step": 252
|
1548 |
+
},
|
1549 |
+
{
|
1550 |
+
"epoch": 3.83,
|
1551 |
+
"learning_rate": 4.200787401574803e-06,
|
1552 |
+
"loss": 0.2918,
|
1553 |
+
"step": 253
|
1554 |
+
},
|
1555 |
+
{
|
1556 |
+
"epoch": 3.85,
|
1557 |
+
"learning_rate": 4.196850393700788e-06,
|
1558 |
+
"loss": 0.5486,
|
1559 |
+
"step": 254
|
1560 |
+
},
|
1561 |
+
{
|
1562 |
+
"epoch": 3.86,
|
1563 |
+
"learning_rate": 4.192913385826772e-06,
|
1564 |
+
"loss": 0.6028,
|
1565 |
+
"step": 255
|
1566 |
+
},
|
1567 |
+
{
|
1568 |
+
"epoch": 3.88,
|
1569 |
+
"learning_rate": 4.1889763779527564e-06,
|
1570 |
+
"loss": 0.5627,
|
1571 |
+
"step": 256
|
1572 |
+
},
|
1573 |
+
{
|
1574 |
+
"epoch": 3.89,
|
1575 |
+
"learning_rate": 4.18503937007874e-06,
|
1576 |
+
"loss": 0.6596,
|
1577 |
+
"step": 257
|
1578 |
+
},
|
1579 |
+
{
|
1580 |
+
"epoch": 3.91,
|
1581 |
+
"learning_rate": 4.181102362204725e-06,
|
1582 |
+
"loss": 0.4707,
|
1583 |
+
"step": 258
|
1584 |
+
},
|
1585 |
+
{
|
1586 |
+
"epoch": 3.92,
|
1587 |
+
"learning_rate": 4.177165354330709e-06,
|
1588 |
+
"loss": 0.3549,
|
1589 |
+
"step": 259
|
1590 |
+
},
|
1591 |
+
{
|
1592 |
+
"epoch": 3.94,
|
1593 |
+
"learning_rate": 4.173228346456693e-06,
|
1594 |
+
"loss": 0.5753,
|
1595 |
+
"step": 260
|
1596 |
+
},
|
1597 |
+
{
|
1598 |
+
"epoch": 3.95,
|
1599 |
+
"learning_rate": 4.169291338582678e-06,
|
1600 |
+
"loss": 0.5335,
|
1601 |
+
"step": 261
|
1602 |
+
},
|
1603 |
+
{
|
1604 |
+
"epoch": 3.97,
|
1605 |
+
"learning_rate": 4.165354330708662e-06,
|
1606 |
+
"loss": 0.3238,
|
1607 |
+
"step": 262
|
1608 |
+
},
|
1609 |
+
{
|
1610 |
+
"epoch": 3.98,
|
1611 |
+
"learning_rate": 4.161417322834646e-06,
|
1612 |
+
"loss": 0.4037,
|
1613 |
+
"step": 263
|
1614 |
+
},
|
1615 |
+
{
|
1616 |
+
"epoch": 4.0,
|
1617 |
+
"learning_rate": 4.15748031496063e-06,
|
1618 |
+
"loss": 0.4637,
|
1619 |
+
"step": 264
|
1620 |
+
},
|
1621 |
+
{
|
1622 |
+
"epoch": 4.0,
|
1623 |
+
"eval_accuracy": 0.8181818181818182,
|
1624 |
+
"eval_loss": 0.4488632380962372,
|
1625 |
+
"eval_runtime": 6.8798,
|
1626 |
+
"eval_samples_per_second": 9.593,
|
1627 |
+
"eval_steps_per_second": 2.471,
|
1628 |
+
"step": 264
|
1629 |
+
},
|
1630 |
+
{
|
1631 |
+
"epoch": 4.02,
|
1632 |
+
"learning_rate": 4.153543307086614e-06,
|
1633 |
+
"loss": 0.3581,
|
1634 |
+
"step": 265
|
1635 |
+
},
|
1636 |
+
{
|
1637 |
+
"epoch": 4.03,
|
1638 |
+
"learning_rate": 4.149606299212599e-06,
|
1639 |
+
"loss": 0.3822,
|
1640 |
+
"step": 266
|
1641 |
+
},
|
1642 |
+
{
|
1643 |
+
"epoch": 4.05,
|
1644 |
+
"learning_rate": 4.145669291338583e-06,
|
1645 |
+
"loss": 0.5748,
|
1646 |
+
"step": 267
|
1647 |
+
},
|
1648 |
+
{
|
1649 |
+
"epoch": 4.06,
|
1650 |
+
"learning_rate": 4.141732283464567e-06,
|
1651 |
+
"loss": 0.2967,
|
1652 |
+
"step": 268
|
1653 |
+
},
|
1654 |
+
{
|
1655 |
+
"epoch": 4.08,
|
1656 |
+
"learning_rate": 4.137795275590552e-06,
|
1657 |
+
"loss": 0.588,
|
1658 |
+
"step": 269
|
1659 |
+
},
|
1660 |
+
{
|
1661 |
+
"epoch": 4.09,
|
1662 |
+
"learning_rate": 4.1338582677165355e-06,
|
1663 |
+
"loss": 0.5945,
|
1664 |
+
"step": 270
|
1665 |
+
},
|
1666 |
+
{
|
1667 |
+
"epoch": 4.11,
|
1668 |
+
"learning_rate": 4.12992125984252e-06,
|
1669 |
+
"loss": 0.4901,
|
1670 |
+
"step": 271
|
1671 |
+
},
|
1672 |
+
{
|
1673 |
+
"epoch": 4.12,
|
1674 |
+
"learning_rate": 4.125984251968505e-06,
|
1675 |
+
"loss": 0.4632,
|
1676 |
+
"step": 272
|
1677 |
+
},
|
1678 |
+
{
|
1679 |
+
"epoch": 4.14,
|
1680 |
+
"learning_rate": 4.122047244094489e-06,
|
1681 |
+
"loss": 0.3848,
|
1682 |
+
"step": 273
|
1683 |
+
},
|
1684 |
+
{
|
1685 |
+
"epoch": 4.15,
|
1686 |
+
"learning_rate": 4.118110236220472e-06,
|
1687 |
+
"loss": 0.425,
|
1688 |
+
"step": 274
|
1689 |
+
},
|
1690 |
+
{
|
1691 |
+
"epoch": 4.17,
|
1692 |
+
"learning_rate": 4.114173228346457e-06,
|
1693 |
+
"loss": 0.5269,
|
1694 |
+
"step": 275
|
1695 |
+
},
|
1696 |
+
{
|
1697 |
+
"epoch": 4.18,
|
1698 |
+
"learning_rate": 4.110236220472441e-06,
|
1699 |
+
"loss": 0.2838,
|
1700 |
+
"step": 276
|
1701 |
+
},
|
1702 |
+
{
|
1703 |
+
"epoch": 4.2,
|
1704 |
+
"learning_rate": 4.1062992125984255e-06,
|
1705 |
+
"loss": 0.2754,
|
1706 |
+
"step": 277
|
1707 |
+
},
|
1708 |
+
{
|
1709 |
+
"epoch": 4.21,
|
1710 |
+
"learning_rate": 4.10236220472441e-06,
|
1711 |
+
"loss": 0.4213,
|
1712 |
+
"step": 278
|
1713 |
+
},
|
1714 |
+
{
|
1715 |
+
"epoch": 4.23,
|
1716 |
+
"learning_rate": 4.098425196850394e-06,
|
1717 |
+
"loss": 0.286,
|
1718 |
+
"step": 279
|
1719 |
+
},
|
1720 |
+
{
|
1721 |
+
"epoch": 4.24,
|
1722 |
+
"learning_rate": 4.0944881889763785e-06,
|
1723 |
+
"loss": 0.5367,
|
1724 |
+
"step": 280
|
1725 |
+
},
|
1726 |
+
{
|
1727 |
+
"epoch": 4.26,
|
1728 |
+
"learning_rate": 4.090551181102362e-06,
|
1729 |
+
"loss": 0.5445,
|
1730 |
+
"step": 281
|
1731 |
+
},
|
1732 |
+
{
|
1733 |
+
"epoch": 4.27,
|
1734 |
+
"learning_rate": 4.086614173228346e-06,
|
1735 |
+
"loss": 0.4772,
|
1736 |
+
"step": 282
|
1737 |
+
},
|
1738 |
+
{
|
1739 |
+
"epoch": 4.29,
|
1740 |
+
"learning_rate": 4.082677165354332e-06,
|
1741 |
+
"loss": 0.2597,
|
1742 |
+
"step": 283
|
1743 |
+
},
|
1744 |
+
{
|
1745 |
+
"epoch": 4.3,
|
1746 |
+
"learning_rate": 4.0787401574803154e-06,
|
1747 |
+
"loss": 0.378,
|
1748 |
+
"step": 284
|
1749 |
+
},
|
1750 |
+
{
|
1751 |
+
"epoch": 4.32,
|
1752 |
+
"learning_rate": 4.074803149606299e-06,
|
1753 |
+
"loss": 0.4049,
|
1754 |
+
"step": 285
|
1755 |
+
},
|
1756 |
+
{
|
1757 |
+
"epoch": 4.33,
|
1758 |
+
"learning_rate": 4.070866141732284e-06,
|
1759 |
+
"loss": 0.408,
|
1760 |
+
"step": 286
|
1761 |
+
},
|
1762 |
+
{
|
1763 |
+
"epoch": 4.35,
|
1764 |
+
"learning_rate": 4.066929133858268e-06,
|
1765 |
+
"loss": 0.5702,
|
1766 |
+
"step": 287
|
1767 |
+
},
|
1768 |
+
{
|
1769 |
+
"epoch": 4.36,
|
1770 |
+
"learning_rate": 4.062992125984252e-06,
|
1771 |
+
"loss": 0.2844,
|
1772 |
+
"step": 288
|
1773 |
+
},
|
1774 |
+
{
|
1775 |
+
"epoch": 4.38,
|
1776 |
+
"learning_rate": 4.059055118110237e-06,
|
1777 |
+
"loss": 0.2546,
|
1778 |
+
"step": 289
|
1779 |
+
},
|
1780 |
+
{
|
1781 |
+
"epoch": 4.39,
|
1782 |
+
"learning_rate": 4.055118110236221e-06,
|
1783 |
+
"loss": 0.26,
|
1784 |
+
"step": 290
|
1785 |
+
},
|
1786 |
+
{
|
1787 |
+
"epoch": 4.41,
|
1788 |
+
"learning_rate": 4.051181102362205e-06,
|
1789 |
+
"loss": 0.548,
|
1790 |
+
"step": 291
|
1791 |
+
},
|
1792 |
+
{
|
1793 |
+
"epoch": 4.42,
|
1794 |
+
"learning_rate": 4.047244094488189e-06,
|
1795 |
+
"loss": 0.5079,
|
1796 |
+
"step": 292
|
1797 |
+
},
|
1798 |
+
{
|
1799 |
+
"epoch": 4.44,
|
1800 |
+
"learning_rate": 4.043307086614174e-06,
|
1801 |
+
"loss": 0.3426,
|
1802 |
+
"step": 293
|
1803 |
+
},
|
1804 |
+
{
|
1805 |
+
"epoch": 4.45,
|
1806 |
+
"learning_rate": 4.039370078740158e-06,
|
1807 |
+
"loss": 0.2459,
|
1808 |
+
"step": 294
|
1809 |
+
},
|
1810 |
+
{
|
1811 |
+
"epoch": 4.47,
|
1812 |
+
"learning_rate": 4.035433070866142e-06,
|
1813 |
+
"loss": 0.6087,
|
1814 |
+
"step": 295
|
1815 |
+
},
|
1816 |
+
{
|
1817 |
+
"epoch": 4.48,
|
1818 |
+
"learning_rate": 4.031496062992126e-06,
|
1819 |
+
"loss": 0.303,
|
1820 |
+
"step": 296
|
1821 |
+
},
|
1822 |
+
{
|
1823 |
+
"epoch": 4.5,
|
1824 |
+
"learning_rate": 4.027559055118111e-06,
|
1825 |
+
"loss": 0.2649,
|
1826 |
+
"step": 297
|
1827 |
+
},
|
1828 |
+
{
|
1829 |
+
"epoch": 4.52,
|
1830 |
+
"learning_rate": 4.0236220472440945e-06,
|
1831 |
+
"loss": 0.4792,
|
1832 |
+
"step": 298
|
1833 |
+
},
|
1834 |
+
{
|
1835 |
+
"epoch": 4.53,
|
1836 |
+
"learning_rate": 4.019685039370079e-06,
|
1837 |
+
"loss": 0.353,
|
1838 |
+
"step": 299
|
1839 |
+
},
|
1840 |
+
{
|
1841 |
+
"epoch": 4.55,
|
1842 |
+
"learning_rate": 4.015748031496064e-06,
|
1843 |
+
"loss": 0.2387,
|
1844 |
+
"step": 300
|
1845 |
+
},
|
1846 |
+
{
|
1847 |
+
"epoch": 4.56,
|
1848 |
+
"learning_rate": 4.011811023622048e-06,
|
1849 |
+
"loss": 0.5596,
|
1850 |
+
"step": 301
|
1851 |
+
},
|
1852 |
+
{
|
1853 |
+
"epoch": 4.58,
|
1854 |
+
"learning_rate": 4.007874015748031e-06,
|
1855 |
+
"loss": 0.2684,
|
1856 |
+
"step": 302
|
1857 |
+
},
|
1858 |
+
{
|
1859 |
+
"epoch": 4.59,
|
1860 |
+
"learning_rate": 4.003937007874016e-06,
|
1861 |
+
"loss": 0.3403,
|
1862 |
+
"step": 303
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 4.61,
|
1866 |
+
"learning_rate": 4.000000000000001e-06,
|
1867 |
+
"loss": 0.2609,
|
1868 |
+
"step": 304
|
1869 |
+
},
|
1870 |
+
{
|
1871 |
+
"epoch": 4.62,
|
1872 |
+
"learning_rate": 3.9960629921259845e-06,
|
1873 |
+
"loss": 0.2897,
|
1874 |
+
"step": 305
|
1875 |
+
},
|
1876 |
+
{
|
1877 |
+
"epoch": 4.64,
|
1878 |
+
"learning_rate": 3.992125984251969e-06,
|
1879 |
+
"loss": 0.4744,
|
1880 |
+
"step": 306
|
1881 |
+
},
|
1882 |
+
{
|
1883 |
+
"epoch": 4.65,
|
1884 |
+
"learning_rate": 3.988188976377953e-06,
|
1885 |
+
"loss": 0.2779,
|
1886 |
+
"step": 307
|
1887 |
+
},
|
1888 |
+
{
|
1889 |
+
"epoch": 4.67,
|
1890 |
+
"learning_rate": 3.9842519685039375e-06,
|
1891 |
+
"loss": 0.2217,
|
1892 |
+
"step": 308
|
1893 |
+
},
|
1894 |
+
{
|
1895 |
+
"epoch": 4.68,
|
1896 |
+
"learning_rate": 3.980314960629922e-06,
|
1897 |
+
"loss": 0.4549,
|
1898 |
+
"step": 309
|
1899 |
+
},
|
1900 |
+
{
|
1901 |
+
"epoch": 4.7,
|
1902 |
+
"learning_rate": 3.976377952755906e-06,
|
1903 |
+
"loss": 0.5696,
|
1904 |
+
"step": 310
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 4.71,
|
1908 |
+
"learning_rate": 3.97244094488189e-06,
|
1909 |
+
"loss": 0.2175,
|
1910 |
+
"step": 311
|
1911 |
+
},
|
1912 |
+
{
|
1913 |
+
"epoch": 4.73,
|
1914 |
+
"learning_rate": 3.968503937007874e-06,
|
1915 |
+
"loss": 0.2406,
|
1916 |
+
"step": 312
|
1917 |
+
},
|
1918 |
+
{
|
1919 |
+
"epoch": 4.74,
|
1920 |
+
"learning_rate": 3.964566929133858e-06,
|
1921 |
+
"loss": 0.5512,
|
1922 |
+
"step": 313
|
1923 |
+
},
|
1924 |
+
{
|
1925 |
+
"epoch": 4.76,
|
1926 |
+
"learning_rate": 3.960629921259843e-06,
|
1927 |
+
"loss": 0.2105,
|
1928 |
+
"step": 314
|
1929 |
+
},
|
1930 |
+
{
|
1931 |
+
"epoch": 4.77,
|
1932 |
+
"learning_rate": 3.9566929133858275e-06,
|
1933 |
+
"loss": 0.8069,
|
1934 |
+
"step": 315
|
1935 |
+
},
|
1936 |
+
{
|
1937 |
+
"epoch": 4.79,
|
1938 |
+
"learning_rate": 3.952755905511811e-06,
|
1939 |
+
"loss": 0.4629,
|
1940 |
+
"step": 316
|
1941 |
+
},
|
1942 |
+
{
|
1943 |
+
"epoch": 4.8,
|
1944 |
+
"learning_rate": 3.948818897637796e-06,
|
1945 |
+
"loss": 0.2115,
|
1946 |
+
"step": 317
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 4.82,
|
1950 |
+
"learning_rate": 3.94488188976378e-06,
|
1951 |
+
"loss": 0.2883,
|
1952 |
+
"step": 318
|
1953 |
+
},
|
1954 |
+
{
|
1955 |
+
"epoch": 4.83,
|
1956 |
+
"learning_rate": 3.9409448818897635e-06,
|
1957 |
+
"loss": 0.5092,
|
1958 |
+
"step": 319
|
1959 |
+
},
|
1960 |
+
{
|
1961 |
+
"epoch": 4.85,
|
1962 |
+
"learning_rate": 3.937007874015748e-06,
|
1963 |
+
"loss": 0.3486,
|
1964 |
+
"step": 320
|
1965 |
+
},
|
1966 |
+
{
|
1967 |
+
"epoch": 4.86,
|
1968 |
+
"learning_rate": 3.933070866141733e-06,
|
1969 |
+
"loss": 0.5374,
|
1970 |
+
"step": 321
|
1971 |
+
},
|
1972 |
+
{
|
1973 |
+
"epoch": 4.88,
|
1974 |
+
"learning_rate": 3.929133858267717e-06,
|
1975 |
+
"loss": 0.3302,
|
1976 |
+
"step": 322
|
1977 |
+
},
|
1978 |
+
{
|
1979 |
+
"epoch": 4.89,
|
1980 |
+
"learning_rate": 3.925196850393701e-06,
|
1981 |
+
"loss": 0.5962,
|
1982 |
+
"step": 323
|
1983 |
+
},
|
1984 |
+
{
|
1985 |
+
"epoch": 4.91,
|
1986 |
+
"learning_rate": 3.921259842519685e-06,
|
1987 |
+
"loss": 0.2028,
|
1988 |
+
"step": 324
|
1989 |
+
},
|
1990 |
+
{
|
1991 |
+
"epoch": 4.92,
|
1992 |
+
"learning_rate": 3.91732283464567e-06,
|
1993 |
+
"loss": 0.4314,
|
1994 |
+
"step": 325
|
1995 |
+
},
|
1996 |
+
{
|
1997 |
+
"epoch": 4.94,
|
1998 |
+
"learning_rate": 3.913385826771654e-06,
|
1999 |
+
"loss": 0.2254,
|
2000 |
+
"step": 326
|
2001 |
+
},
|
2002 |
+
{
|
2003 |
+
"epoch": 4.95,
|
2004 |
+
"learning_rate": 3.909448818897638e-06,
|
2005 |
+
"loss": 0.5416,
|
2006 |
+
"step": 327
|
2007 |
+
},
|
2008 |
+
{
|
2009 |
+
"epoch": 4.97,
|
2010 |
+
"learning_rate": 3.905511811023622e-06,
|
2011 |
+
"loss": 0.1982,
|
2012 |
+
"step": 328
|
2013 |
+
},
|
2014 |
+
{
|
2015 |
+
"epoch": 4.98,
|
2016 |
+
"learning_rate": 3.901574803149607e-06,
|
2017 |
+
"loss": 0.2117,
|
2018 |
+
"step": 329
|
2019 |
+
},
|
2020 |
+
{
|
2021 |
+
"epoch": 5.0,
|
2022 |
+
"learning_rate": 3.89763779527559e-06,
|
2023 |
+
"loss": 0.2166,
|
2024 |
+
"step": 330
|
2025 |
+
},
|
2026 |
+
{
|
2027 |
+
"epoch": 5.0,
|
2028 |
+
"eval_accuracy": 0.8939393939393939,
|
2029 |
+
"eval_loss": 0.3404465913772583,
|
2030 |
+
"eval_runtime": 6.3639,
|
2031 |
+
"eval_samples_per_second": 10.371,
|
2032 |
+
"eval_steps_per_second": 2.671,
|
2033 |
+
"step": 330
|
2034 |
+
},
|
2035 |
+
{
|
2036 |
+
"epoch": 5.02,
|
2037 |
+
"learning_rate": 3.893700787401575e-06,
|
2038 |
+
"loss": 0.3983,
|
2039 |
+
"step": 331
|
2040 |
+
},
|
2041 |
+
{
|
2042 |
+
"epoch": 5.03,
|
2043 |
+
"learning_rate": 3.88976377952756e-06,
|
2044 |
+
"loss": 0.566,
|
2045 |
+
"step": 332
|
2046 |
+
},
|
2047 |
+
{
|
2048 |
+
"epoch": 5.05,
|
2049 |
+
"learning_rate": 3.8858267716535435e-06,
|
2050 |
+
"loss": 0.2756,
|
2051 |
+
"step": 333
|
2052 |
+
},
|
2053 |
+
{
|
2054 |
+
"epoch": 5.06,
|
2055 |
+
"learning_rate": 3.881889763779528e-06,
|
2056 |
+
"loss": 0.2214,
|
2057 |
+
"step": 334
|
2058 |
+
},
|
2059 |
+
{
|
2060 |
+
"epoch": 5.08,
|
2061 |
+
"learning_rate": 3.877952755905512e-06,
|
2062 |
+
"loss": 0.21,
|
2063 |
+
"step": 335
|
2064 |
+
},
|
2065 |
+
{
|
2066 |
+
"epoch": 5.09,
|
2067 |
+
"learning_rate": 3.8740157480314965e-06,
|
2068 |
+
"loss": 0.2232,
|
2069 |
+
"step": 336
|
2070 |
+
},
|
2071 |
+
{
|
2072 |
+
"epoch": 5.11,
|
2073 |
+
"learning_rate": 3.87007874015748e-06,
|
2074 |
+
"loss": 0.2061,
|
2075 |
+
"step": 337
|
2076 |
+
},
|
2077 |
+
{
|
2078 |
+
"epoch": 5.12,
|
2079 |
+
"learning_rate": 3.866141732283465e-06,
|
2080 |
+
"loss": 0.311,
|
2081 |
+
"step": 338
|
2082 |
+
},
|
2083 |
+
{
|
2084 |
+
"epoch": 5.14,
|
2085 |
+
"learning_rate": 3.862204724409449e-06,
|
2086 |
+
"loss": 0.1845,
|
2087 |
+
"step": 339
|
2088 |
+
},
|
2089 |
+
{
|
2090 |
+
"epoch": 5.15,
|
2091 |
+
"learning_rate": 3.858267716535433e-06,
|
2092 |
+
"loss": 0.2081,
|
2093 |
+
"step": 340
|
2094 |
+
},
|
2095 |
+
{
|
2096 |
+
"epoch": 5.17,
|
2097 |
+
"learning_rate": 3.854330708661418e-06,
|
2098 |
+
"loss": 0.4295,
|
2099 |
+
"step": 341
|
2100 |
+
},
|
2101 |
+
{
|
2102 |
+
"epoch": 5.18,
|
2103 |
+
"learning_rate": 3.850393700787402e-06,
|
2104 |
+
"loss": 0.1863,
|
2105 |
+
"step": 342
|
2106 |
+
},
|
2107 |
+
{
|
2108 |
+
"epoch": 5.2,
|
2109 |
+
"learning_rate": 3.8464566929133865e-06,
|
2110 |
+
"loss": 0.2167,
|
2111 |
+
"step": 343
|
2112 |
+
},
|
2113 |
+
{
|
2114 |
+
"epoch": 5.21,
|
2115 |
+
"learning_rate": 3.84251968503937e-06,
|
2116 |
+
"loss": 0.5119,
|
2117 |
+
"step": 344
|
2118 |
+
},
|
2119 |
+
{
|
2120 |
+
"epoch": 5.23,
|
2121 |
+
"learning_rate": 3.838582677165354e-06,
|
2122 |
+
"loss": 0.4946,
|
2123 |
+
"step": 345
|
2124 |
+
},
|
2125 |
+
{
|
2126 |
+
"epoch": 5.24,
|
2127 |
+
"learning_rate": 3.834645669291339e-06,
|
2128 |
+
"loss": 0.2194,
|
2129 |
+
"step": 346
|
2130 |
+
},
|
2131 |
+
{
|
2132 |
+
"epoch": 5.26,
|
2133 |
+
"learning_rate": 3.830708661417323e-06,
|
2134 |
+
"loss": 0.1861,
|
2135 |
+
"step": 347
|
2136 |
+
},
|
2137 |
+
{
|
2138 |
+
"epoch": 5.27,
|
2139 |
+
"learning_rate": 3.826771653543307e-06,
|
2140 |
+
"loss": 0.1871,
|
2141 |
+
"step": 348
|
2142 |
+
},
|
2143 |
+
{
|
2144 |
+
"epoch": 5.29,
|
2145 |
+
"learning_rate": 3.822834645669292e-06,
|
2146 |
+
"loss": 0.293,
|
2147 |
+
"step": 349
|
2148 |
+
},
|
2149 |
+
{
|
2150 |
+
"epoch": 5.3,
|
2151 |
+
"learning_rate": 3.818897637795276e-06,
|
2152 |
+
"loss": 0.183,
|
2153 |
+
"step": 350
|
2154 |
+
},
|
2155 |
+
{
|
2156 |
+
"epoch": 5.32,
|
2157 |
+
"learning_rate": 3.81496062992126e-06,
|
2158 |
+
"loss": 0.4737,
|
2159 |
+
"step": 351
|
2160 |
+
},
|
2161 |
+
{
|
2162 |
+
"epoch": 5.33,
|
2163 |
+
"learning_rate": 3.8110236220472445e-06,
|
2164 |
+
"loss": 0.1968,
|
2165 |
+
"step": 352
|
2166 |
+
},
|
2167 |
+
{
|
2168 |
+
"epoch": 5.35,
|
2169 |
+
"learning_rate": 3.8070866141732287e-06,
|
2170 |
+
"loss": 0.5645,
|
2171 |
+
"step": 353
|
2172 |
+
},
|
2173 |
+
{
|
2174 |
+
"epoch": 5.36,
|
2175 |
+
"learning_rate": 3.803149606299213e-06,
|
2176 |
+
"loss": 0.2107,
|
2177 |
+
"step": 354
|
2178 |
+
},
|
2179 |
+
{
|
2180 |
+
"epoch": 5.38,
|
2181 |
+
"learning_rate": 3.799212598425197e-06,
|
2182 |
+
"loss": 0.6736,
|
2183 |
+
"step": 355
|
2184 |
+
},
|
2185 |
+
{
|
2186 |
+
"epoch": 5.39,
|
2187 |
+
"learning_rate": 3.7952755905511814e-06,
|
2188 |
+
"loss": 0.1696,
|
2189 |
+
"step": 356
|
2190 |
+
},
|
2191 |
+
{
|
2192 |
+
"epoch": 5.41,
|
2193 |
+
"learning_rate": 3.791338582677166e-06,
|
2194 |
+
"loss": 0.204,
|
2195 |
+
"step": 357
|
2196 |
+
},
|
2197 |
+
{
|
2198 |
+
"epoch": 5.42,
|
2199 |
+
"learning_rate": 3.7874015748031502e-06,
|
2200 |
+
"loss": 0.1744,
|
2201 |
+
"step": 358
|
2202 |
+
},
|
2203 |
+
{
|
2204 |
+
"epoch": 5.44,
|
2205 |
+
"learning_rate": 3.783464566929134e-06,
|
2206 |
+
"loss": 0.1769,
|
2207 |
+
"step": 359
|
2208 |
+
},
|
2209 |
+
{
|
2210 |
+
"epoch": 5.45,
|
2211 |
+
"learning_rate": 3.7795275590551182e-06,
|
2212 |
+
"loss": 0.164,
|
2213 |
+
"step": 360
|
2214 |
+
},
|
2215 |
+
{
|
2216 |
+
"epoch": 5.47,
|
2217 |
+
"learning_rate": 3.7755905511811025e-06,
|
2218 |
+
"loss": 0.7071,
|
2219 |
+
"step": 361
|
2220 |
+
},
|
2221 |
+
{
|
2222 |
+
"epoch": 5.48,
|
2223 |
+
"learning_rate": 3.7716535433070867e-06,
|
2224 |
+
"loss": 0.1572,
|
2225 |
+
"step": 362
|
2226 |
+
},
|
2227 |
+
{
|
2228 |
+
"epoch": 5.5,
|
2229 |
+
"learning_rate": 3.7677165354330713e-06,
|
2230 |
+
"loss": 0.3369,
|
2231 |
+
"step": 363
|
2232 |
+
},
|
2233 |
+
{
|
2234 |
+
"epoch": 5.52,
|
2235 |
+
"learning_rate": 3.7637795275590555e-06,
|
2236 |
+
"loss": 0.1589,
|
2237 |
+
"step": 364
|
2238 |
+
},
|
2239 |
+
{
|
2240 |
+
"epoch": 5.53,
|
2241 |
+
"learning_rate": 3.7598425196850398e-06,
|
2242 |
+
"loss": 0.1886,
|
2243 |
+
"step": 365
|
2244 |
+
},
|
2245 |
+
{
|
2246 |
+
"epoch": 5.55,
|
2247 |
+
"learning_rate": 3.755905511811024e-06,
|
2248 |
+
"loss": 0.2163,
|
2249 |
+
"step": 366
|
2250 |
+
},
|
2251 |
+
{
|
2252 |
+
"epoch": 5.56,
|
2253 |
+
"learning_rate": 3.7519685039370078e-06,
|
2254 |
+
"loss": 0.1526,
|
2255 |
+
"step": 367
|
2256 |
+
},
|
2257 |
+
{
|
2258 |
+
"epoch": 5.58,
|
2259 |
+
"learning_rate": 3.748031496062993e-06,
|
2260 |
+
"loss": 0.5631,
|
2261 |
+
"step": 368
|
2262 |
+
},
|
2263 |
+
{
|
2264 |
+
"epoch": 5.59,
|
2265 |
+
"learning_rate": 3.7440944881889766e-06,
|
2266 |
+
"loss": 0.1584,
|
2267 |
+
"step": 369
|
2268 |
+
},
|
2269 |
+
{
|
2270 |
+
"epoch": 5.61,
|
2271 |
+
"learning_rate": 3.740157480314961e-06,
|
2272 |
+
"loss": 0.5094,
|
2273 |
+
"step": 370
|
2274 |
+
},
|
2275 |
+
{
|
2276 |
+
"epoch": 5.62,
|
2277 |
+
"learning_rate": 3.736220472440945e-06,
|
2278 |
+
"loss": 0.4171,
|
2279 |
+
"step": 371
|
2280 |
+
},
|
2281 |
+
{
|
2282 |
+
"epoch": 5.64,
|
2283 |
+
"learning_rate": 3.7322834645669293e-06,
|
2284 |
+
"loss": 0.5874,
|
2285 |
+
"step": 372
|
2286 |
+
},
|
2287 |
+
{
|
2288 |
+
"epoch": 5.65,
|
2289 |
+
"learning_rate": 3.728346456692914e-06,
|
2290 |
+
"loss": 0.177,
|
2291 |
+
"step": 373
|
2292 |
+
},
|
2293 |
+
{
|
2294 |
+
"epoch": 5.67,
|
2295 |
+
"learning_rate": 3.724409448818898e-06,
|
2296 |
+
"loss": 0.1753,
|
2297 |
+
"step": 374
|
2298 |
+
},
|
2299 |
+
{
|
2300 |
+
"epoch": 5.68,
|
2301 |
+
"learning_rate": 3.7204724409448824e-06,
|
2302 |
+
"loss": 0.158,
|
2303 |
+
"step": 375
|
2304 |
+
},
|
2305 |
+
{
|
2306 |
+
"epoch": 5.7,
|
2307 |
+
"learning_rate": 3.7165354330708666e-06,
|
2308 |
+
"loss": 0.1736,
|
2309 |
+
"step": 376
|
2310 |
+
},
|
2311 |
+
{
|
2312 |
+
"epoch": 5.71,
|
2313 |
+
"learning_rate": 3.7125984251968504e-06,
|
2314 |
+
"loss": 0.2736,
|
2315 |
+
"step": 377
|
2316 |
+
},
|
2317 |
+
{
|
2318 |
+
"epoch": 5.73,
|
2319 |
+
"learning_rate": 3.7086614173228346e-06,
|
2320 |
+
"loss": 0.2054,
|
2321 |
+
"step": 378
|
2322 |
+
},
|
2323 |
+
{
|
2324 |
+
"epoch": 5.74,
|
2325 |
+
"learning_rate": 3.7047244094488193e-06,
|
2326 |
+
"loss": 0.383,
|
2327 |
+
"step": 379
|
2328 |
+
},
|
2329 |
+
{
|
2330 |
+
"epoch": 5.76,
|
2331 |
+
"learning_rate": 3.7007874015748035e-06,
|
2332 |
+
"loss": 0.146,
|
2333 |
+
"step": 380
|
2334 |
+
},
|
2335 |
+
{
|
2336 |
+
"epoch": 5.77,
|
2337 |
+
"learning_rate": 3.6968503937007877e-06,
|
2338 |
+
"loss": 0.1785,
|
2339 |
+
"step": 381
|
2340 |
+
},
|
2341 |
+
{
|
2342 |
+
"epoch": 5.79,
|
2343 |
+
"learning_rate": 3.692913385826772e-06,
|
2344 |
+
"loss": 0.599,
|
2345 |
+
"step": 382
|
2346 |
+
},
|
2347 |
+
{
|
2348 |
+
"epoch": 5.8,
|
2349 |
+
"learning_rate": 3.688976377952756e-06,
|
2350 |
+
"loss": 0.5044,
|
2351 |
+
"step": 383
|
2352 |
+
},
|
2353 |
+
{
|
2354 |
+
"epoch": 5.82,
|
2355 |
+
"learning_rate": 3.6850393700787408e-06,
|
2356 |
+
"loss": 0.1508,
|
2357 |
+
"step": 384
|
2358 |
+
},
|
2359 |
+
{
|
2360 |
+
"epoch": 5.83,
|
2361 |
+
"learning_rate": 3.681102362204725e-06,
|
2362 |
+
"loss": 0.1437,
|
2363 |
+
"step": 385
|
2364 |
+
},
|
2365 |
+
{
|
2366 |
+
"epoch": 5.85,
|
2367 |
+
"learning_rate": 3.677165354330709e-06,
|
2368 |
+
"loss": 0.2467,
|
2369 |
+
"step": 386
|
2370 |
+
},
|
2371 |
+
{
|
2372 |
+
"epoch": 5.86,
|
2373 |
+
"learning_rate": 3.673228346456693e-06,
|
2374 |
+
"loss": 0.1497,
|
2375 |
+
"step": 387
|
2376 |
+
},
|
2377 |
+
{
|
2378 |
+
"epoch": 5.88,
|
2379 |
+
"learning_rate": 3.6692913385826772e-06,
|
2380 |
+
"loss": 0.3545,
|
2381 |
+
"step": 388
|
2382 |
+
},
|
2383 |
+
{
|
2384 |
+
"epoch": 5.89,
|
2385 |
+
"learning_rate": 3.665354330708662e-06,
|
2386 |
+
"loss": 0.1441,
|
2387 |
+
"step": 389
|
2388 |
+
},
|
2389 |
+
{
|
2390 |
+
"epoch": 5.91,
|
2391 |
+
"learning_rate": 3.661417322834646e-06,
|
2392 |
+
"loss": 0.1408,
|
2393 |
+
"step": 390
|
2394 |
+
},
|
2395 |
+
{
|
2396 |
+
"epoch": 5.92,
|
2397 |
+
"learning_rate": 3.6574803149606303e-06,
|
2398 |
+
"loss": 0.156,
|
2399 |
+
"step": 391
|
2400 |
+
},
|
2401 |
+
{
|
2402 |
+
"epoch": 5.94,
|
2403 |
+
"learning_rate": 3.6535433070866145e-06,
|
2404 |
+
"loss": 0.1311,
|
2405 |
+
"step": 392
|
2406 |
+
},
|
2407 |
+
{
|
2408 |
+
"epoch": 5.95,
|
2409 |
+
"learning_rate": 3.6496062992125988e-06,
|
2410 |
+
"loss": 0.6291,
|
2411 |
+
"step": 393
|
2412 |
+
},
|
2413 |
+
{
|
2414 |
+
"epoch": 5.97,
|
2415 |
+
"learning_rate": 3.6456692913385826e-06,
|
2416 |
+
"loss": 0.408,
|
2417 |
+
"step": 394
|
2418 |
+
},
|
2419 |
+
{
|
2420 |
+
"epoch": 5.98,
|
2421 |
+
"learning_rate": 3.6417322834645676e-06,
|
2422 |
+
"loss": 0.1383,
|
2423 |
+
"step": 395
|
2424 |
+
},
|
2425 |
+
{
|
2426 |
+
"epoch": 6.0,
|
2427 |
+
"learning_rate": 3.6377952755905514e-06,
|
2428 |
+
"loss": 0.2411,
|
2429 |
+
"step": 396
|
2430 |
+
},
|
2431 |
+
{
|
2432 |
+
"epoch": 6.0,
|
2433 |
+
"eval_accuracy": 0.9393939393939394,
|
2434 |
+
"eval_loss": 0.2369035929441452,
|
2435 |
+
"eval_runtime": 6.2358,
|
2436 |
+
"eval_samples_per_second": 10.584,
|
2437 |
+
"eval_steps_per_second": 2.726,
|
2438 |
+
"step": 396
|
2439 |
+
},
|
2440 |
+
{
|
2441 |
+
"epoch": 6.02,
|
2442 |
+
"learning_rate": 3.6338582677165356e-06,
|
2443 |
+
"loss": 0.5985,
|
2444 |
+
"step": 397
|
2445 |
+
},
|
2446 |
+
{
|
2447 |
+
"epoch": 6.03,
|
2448 |
+
"learning_rate": 3.62992125984252e-06,
|
2449 |
+
"loss": 0.1448,
|
2450 |
+
"step": 398
|
2451 |
+
},
|
2452 |
+
{
|
2453 |
+
"epoch": 6.05,
|
2454 |
+
"learning_rate": 3.625984251968504e-06,
|
2455 |
+
"loss": 0.5026,
|
2456 |
+
"step": 399
|
2457 |
+
},
|
2458 |
+
{
|
2459 |
+
"epoch": 6.06,
|
2460 |
+
"learning_rate": 3.6220472440944887e-06,
|
2461 |
+
"loss": 0.1389,
|
2462 |
+
"step": 400
|
2463 |
+
},
|
2464 |
+
{
|
2465 |
+
"epoch": 6.08,
|
2466 |
+
"learning_rate": 3.618110236220473e-06,
|
2467 |
+
"loss": 0.5432,
|
2468 |
+
"step": 401
|
2469 |
+
},
|
2470 |
+
{
|
2471 |
+
"epoch": 6.09,
|
2472 |
+
"learning_rate": 3.614173228346457e-06,
|
2473 |
+
"loss": 0.5761,
|
2474 |
+
"step": 402
|
2475 |
+
},
|
2476 |
+
{
|
2477 |
+
"epoch": 6.11,
|
2478 |
+
"learning_rate": 3.610236220472441e-06,
|
2479 |
+
"loss": 0.2165,
|
2480 |
+
"step": 403
|
2481 |
+
},
|
2482 |
+
{
|
2483 |
+
"epoch": 6.12,
|
2484 |
+
"learning_rate": 3.606299212598425e-06,
|
2485 |
+
"loss": 0.1467,
|
2486 |
+
"step": 404
|
2487 |
+
},
|
2488 |
+
{
|
2489 |
+
"epoch": 6.14,
|
2490 |
+
"learning_rate": 3.60236220472441e-06,
|
2491 |
+
"loss": 0.1382,
|
2492 |
+
"step": 405
|
2493 |
+
},
|
2494 |
+
{
|
2495 |
+
"epoch": 6.15,
|
2496 |
+
"learning_rate": 3.598425196850394e-06,
|
2497 |
+
"loss": 0.3128,
|
2498 |
+
"step": 406
|
2499 |
+
},
|
2500 |
+
{
|
2501 |
+
"epoch": 6.17,
|
2502 |
+
"learning_rate": 3.5944881889763783e-06,
|
2503 |
+
"loss": 0.1911,
|
2504 |
+
"step": 407
|
2505 |
+
},
|
2506 |
+
{
|
2507 |
+
"epoch": 6.18,
|
2508 |
+
"learning_rate": 3.5905511811023625e-06,
|
2509 |
+
"loss": 0.1306,
|
2510 |
+
"step": 408
|
2511 |
+
},
|
2512 |
+
{
|
2513 |
+
"epoch": 6.2,
|
2514 |
+
"learning_rate": 3.5866141732283467e-06,
|
2515 |
+
"loss": 0.1925,
|
2516 |
+
"step": 409
|
2517 |
+
},
|
2518 |
+
{
|
2519 |
+
"epoch": 6.21,
|
2520 |
+
"learning_rate": 3.582677165354331e-06,
|
2521 |
+
"loss": 0.1275,
|
2522 |
+
"step": 410
|
2523 |
+
},
|
2524 |
+
{
|
2525 |
+
"epoch": 6.23,
|
2526 |
+
"learning_rate": 3.5787401574803156e-06,
|
2527 |
+
"loss": 0.1296,
|
2528 |
+
"step": 411
|
2529 |
+
},
|
2530 |
+
{
|
2531 |
+
"epoch": 6.24,
|
2532 |
+
"learning_rate": 3.5748031496062998e-06,
|
2533 |
+
"loss": 0.1237,
|
2534 |
+
"step": 412
|
2535 |
+
},
|
2536 |
+
{
|
2537 |
+
"epoch": 6.26,
|
2538 |
+
"learning_rate": 3.5708661417322836e-06,
|
2539 |
+
"loss": 1.1349,
|
2540 |
+
"step": 413
|
2541 |
+
},
|
2542 |
+
{
|
2543 |
+
"epoch": 6.27,
|
2544 |
+
"learning_rate": 3.566929133858268e-06,
|
2545 |
+
"loss": 0.1208,
|
2546 |
+
"step": 414
|
2547 |
+
},
|
2548 |
+
{
|
2549 |
+
"epoch": 6.29,
|
2550 |
+
"learning_rate": 3.562992125984252e-06,
|
2551 |
+
"loss": 0.1221,
|
2552 |
+
"step": 415
|
2553 |
+
},
|
2554 |
+
{
|
2555 |
+
"epoch": 6.3,
|
2556 |
+
"learning_rate": 3.5590551181102367e-06,
|
2557 |
+
"loss": 0.6214,
|
2558 |
+
"step": 416
|
2559 |
+
},
|
2560 |
+
{
|
2561 |
+
"epoch": 6.32,
|
2562 |
+
"learning_rate": 3.555118110236221e-06,
|
2563 |
+
"loss": 0.1193,
|
2564 |
+
"step": 417
|
2565 |
+
},
|
2566 |
+
{
|
2567 |
+
"epoch": 6.33,
|
2568 |
+
"learning_rate": 3.551181102362205e-06,
|
2569 |
+
"loss": 0.1218,
|
2570 |
+
"step": 418
|
2571 |
+
},
|
2572 |
+
{
|
2573 |
+
"epoch": 6.35,
|
2574 |
+
"learning_rate": 3.5472440944881893e-06,
|
2575 |
+
"loss": 0.1493,
|
2576 |
+
"step": 419
|
2577 |
+
},
|
2578 |
+
{
|
2579 |
+
"epoch": 6.36,
|
2580 |
+
"learning_rate": 3.5433070866141735e-06,
|
2581 |
+
"loss": 0.1247,
|
2582 |
+
"step": 420
|
2583 |
+
},
|
2584 |
+
{
|
2585 |
+
"epoch": 6.38,
|
2586 |
+
"learning_rate": 3.539370078740158e-06,
|
2587 |
+
"loss": 0.1158,
|
2588 |
+
"step": 421
|
2589 |
+
},
|
2590 |
+
{
|
2591 |
+
"epoch": 6.39,
|
2592 |
+
"learning_rate": 3.535433070866142e-06,
|
2593 |
+
"loss": 0.1201,
|
2594 |
+
"step": 422
|
2595 |
+
},
|
2596 |
+
{
|
2597 |
+
"epoch": 6.41,
|
2598 |
+
"learning_rate": 3.531496062992126e-06,
|
2599 |
+
"loss": 0.6259,
|
2600 |
+
"step": 423
|
2601 |
+
},
|
2602 |
+
{
|
2603 |
+
"epoch": 6.42,
|
2604 |
+
"learning_rate": 3.5275590551181104e-06,
|
2605 |
+
"loss": 0.1146,
|
2606 |
+
"step": 424
|
2607 |
+
},
|
2608 |
+
{
|
2609 |
+
"epoch": 6.44,
|
2610 |
+
"learning_rate": 3.5236220472440946e-06,
|
2611 |
+
"loss": 0.1311,
|
2612 |
+
"step": 425
|
2613 |
+
},
|
2614 |
+
{
|
2615 |
+
"epoch": 6.45,
|
2616 |
+
"learning_rate": 3.519685039370079e-06,
|
2617 |
+
"loss": 0.1097,
|
2618 |
+
"step": 426
|
2619 |
+
},
|
2620 |
+
{
|
2621 |
+
"epoch": 6.47,
|
2622 |
+
"learning_rate": 3.5157480314960635e-06,
|
2623 |
+
"loss": 0.4229,
|
2624 |
+
"step": 427
|
2625 |
+
},
|
2626 |
+
{
|
2627 |
+
"epoch": 6.48,
|
2628 |
+
"learning_rate": 3.5118110236220477e-06,
|
2629 |
+
"loss": 0.1245,
|
2630 |
+
"step": 428
|
2631 |
+
},
|
2632 |
+
{
|
2633 |
+
"epoch": 6.5,
|
2634 |
+
"learning_rate": 3.507874015748032e-06,
|
2635 |
+
"loss": 0.1066,
|
2636 |
+
"step": 429
|
2637 |
+
},
|
2638 |
+
{
|
2639 |
+
"epoch": 6.52,
|
2640 |
+
"learning_rate": 3.5039370078740157e-06,
|
2641 |
+
"loss": 0.1215,
|
2642 |
+
"step": 430
|
2643 |
+
},
|
2644 |
+
{
|
2645 |
+
"epoch": 6.53,
|
2646 |
+
"learning_rate": 3.5e-06,
|
2647 |
+
"loss": 0.1178,
|
2648 |
+
"step": 431
|
2649 |
+
},
|
2650 |
+
{
|
2651 |
+
"epoch": 6.55,
|
2652 |
+
"learning_rate": 3.4960629921259846e-06,
|
2653 |
+
"loss": 0.5898,
|
2654 |
+
"step": 432
|
2655 |
+
},
|
2656 |
+
{
|
2657 |
+
"epoch": 6.56,
|
2658 |
+
"learning_rate": 3.492125984251969e-06,
|
2659 |
+
"loss": 0.1037,
|
2660 |
+
"step": 433
|
2661 |
+
},
|
2662 |
+
{
|
2663 |
+
"epoch": 6.58,
|
2664 |
+
"learning_rate": 3.488188976377953e-06,
|
2665 |
+
"loss": 0.1185,
|
2666 |
+
"step": 434
|
2667 |
+
},
|
2668 |
+
{
|
2669 |
+
"epoch": 6.59,
|
2670 |
+
"learning_rate": 3.4842519685039372e-06,
|
2671 |
+
"loss": 0.4492,
|
2672 |
+
"step": 435
|
2673 |
+
},
|
2674 |
+
{
|
2675 |
+
"epoch": 6.61,
|
2676 |
+
"learning_rate": 3.4803149606299215e-06,
|
2677 |
+
"loss": 0.4031,
|
2678 |
+
"step": 436
|
2679 |
+
},
|
2680 |
+
{
|
2681 |
+
"epoch": 6.62,
|
2682 |
+
"learning_rate": 3.476377952755906e-06,
|
2683 |
+
"loss": 0.2397,
|
2684 |
+
"step": 437
|
2685 |
+
},
|
2686 |
+
{
|
2687 |
+
"epoch": 6.64,
|
2688 |
+
"learning_rate": 3.4724409448818903e-06,
|
2689 |
+
"loss": 0.1238,
|
2690 |
+
"step": 438
|
2691 |
+
},
|
2692 |
+
{
|
2693 |
+
"epoch": 6.65,
|
2694 |
+
"learning_rate": 3.4685039370078746e-06,
|
2695 |
+
"loss": 0.6626,
|
2696 |
+
"step": 439
|
2697 |
+
},
|
2698 |
+
{
|
2699 |
+
"epoch": 6.67,
|
2700 |
+
"learning_rate": 3.4645669291338583e-06,
|
2701 |
+
"loss": 0.1303,
|
2702 |
+
"step": 440
|
2703 |
+
},
|
2704 |
+
{
|
2705 |
+
"epoch": 6.68,
|
2706 |
+
"learning_rate": 3.4606299212598426e-06,
|
2707 |
+
"loss": 0.1235,
|
2708 |
+
"step": 441
|
2709 |
+
},
|
2710 |
+
{
|
2711 |
+
"epoch": 6.7,
|
2712 |
+
"learning_rate": 3.4566929133858268e-06,
|
2713 |
+
"loss": 0.1008,
|
2714 |
+
"step": 442
|
2715 |
+
},
|
2716 |
+
{
|
2717 |
+
"epoch": 6.71,
|
2718 |
+
"learning_rate": 3.4527559055118114e-06,
|
2719 |
+
"loss": 0.1055,
|
2720 |
+
"step": 443
|
2721 |
+
},
|
2722 |
+
{
|
2723 |
+
"epoch": 6.73,
|
2724 |
+
"learning_rate": 3.4488188976377956e-06,
|
2725 |
+
"loss": 0.0986,
|
2726 |
+
"step": 444
|
2727 |
+
},
|
2728 |
+
{
|
2729 |
+
"epoch": 6.74,
|
2730 |
+
"learning_rate": 3.44488188976378e-06,
|
2731 |
+
"loss": 0.0998,
|
2732 |
+
"step": 445
|
2733 |
+
},
|
2734 |
+
{
|
2735 |
+
"epoch": 6.76,
|
2736 |
+
"learning_rate": 3.440944881889764e-06,
|
2737 |
+
"loss": 0.1114,
|
2738 |
+
"step": 446
|
2739 |
+
},
|
2740 |
+
{
|
2741 |
+
"epoch": 6.77,
|
2742 |
+
"learning_rate": 3.437007874015748e-06,
|
2743 |
+
"loss": 0.1749,
|
2744 |
+
"step": 447
|
2745 |
+
},
|
2746 |
+
{
|
2747 |
+
"epoch": 6.79,
|
2748 |
+
"learning_rate": 3.433070866141733e-06,
|
2749 |
+
"loss": 0.0985,
|
2750 |
+
"step": 448
|
2751 |
+
},
|
2752 |
+
{
|
2753 |
+
"epoch": 6.8,
|
2754 |
+
"learning_rate": 3.4291338582677167e-06,
|
2755 |
+
"loss": 0.0948,
|
2756 |
+
"step": 449
|
2757 |
+
},
|
2758 |
+
{
|
2759 |
+
"epoch": 6.82,
|
2760 |
+
"learning_rate": 3.425196850393701e-06,
|
2761 |
+
"loss": 0.0961,
|
2762 |
+
"step": 450
|
2763 |
+
},
|
2764 |
+
{
|
2765 |
+
"epoch": 6.83,
|
2766 |
+
"learning_rate": 3.421259842519685e-06,
|
2767 |
+
"loss": 0.6144,
|
2768 |
+
"step": 451
|
2769 |
+
},
|
2770 |
+
{
|
2771 |
+
"epoch": 6.85,
|
2772 |
+
"learning_rate": 3.4173228346456694e-06,
|
2773 |
+
"loss": 0.7958,
|
2774 |
+
"step": 452
|
2775 |
+
},
|
2776 |
+
{
|
2777 |
+
"epoch": 6.86,
|
2778 |
+
"learning_rate": 3.413385826771654e-06,
|
2779 |
+
"loss": 0.2617,
|
2780 |
+
"step": 453
|
2781 |
+
},
|
2782 |
+
{
|
2783 |
+
"epoch": 6.88,
|
2784 |
+
"learning_rate": 3.4094488188976383e-06,
|
2785 |
+
"loss": 0.0962,
|
2786 |
+
"step": 454
|
2787 |
+
},
|
2788 |
+
{
|
2789 |
+
"epoch": 6.89,
|
2790 |
+
"learning_rate": 3.4055118110236225e-06,
|
2791 |
+
"loss": 0.0918,
|
2792 |
+
"step": 455
|
2793 |
+
},
|
2794 |
+
{
|
2795 |
+
"epoch": 6.91,
|
2796 |
+
"learning_rate": 3.4015748031496067e-06,
|
2797 |
+
"loss": 0.3138,
|
2798 |
+
"step": 456
|
2799 |
+
},
|
2800 |
+
{
|
2801 |
+
"epoch": 6.92,
|
2802 |
+
"learning_rate": 3.3976377952755905e-06,
|
2803 |
+
"loss": 0.0919,
|
2804 |
+
"step": 457
|
2805 |
+
},
|
2806 |
+
{
|
2807 |
+
"epoch": 6.94,
|
2808 |
+
"learning_rate": 3.3937007874015747e-06,
|
2809 |
+
"loss": 0.0918,
|
2810 |
+
"step": 458
|
2811 |
+
},
|
2812 |
+
{
|
2813 |
+
"epoch": 6.95,
|
2814 |
+
"learning_rate": 3.3897637795275594e-06,
|
2815 |
+
"loss": 0.0895,
|
2816 |
+
"step": 459
|
2817 |
+
},
|
2818 |
+
{
|
2819 |
+
"epoch": 6.97,
|
2820 |
+
"learning_rate": 3.3858267716535436e-06,
|
2821 |
+
"loss": 0.0912,
|
2822 |
+
"step": 460
|
2823 |
+
},
|
2824 |
+
{
|
2825 |
+
"epoch": 6.98,
|
2826 |
+
"learning_rate": 3.381889763779528e-06,
|
2827 |
+
"loss": 0.0922,
|
2828 |
+
"step": 461
|
2829 |
+
},
|
2830 |
+
{
|
2831 |
+
"epoch": 7.0,
|
2832 |
+
"learning_rate": 3.377952755905512e-06,
|
2833 |
+
"loss": 0.1105,
|
2834 |
+
"step": 462
|
2835 |
+
},
|
2836 |
+
{
|
2837 |
+
"epoch": 7.0,
|
2838 |
+
"eval_accuracy": 0.9393939393939394,
|
2839 |
+
"eval_loss": 0.2016882747411728,
|
2840 |
+
"eval_runtime": 6.2203,
|
2841 |
+
"eval_samples_per_second": 10.61,
|
2842 |
+
"eval_steps_per_second": 2.733,
|
2843 |
+
"step": 462
|
2844 |
+
},
|
2845 |
+
{
|
2846 |
+
"epoch": 7.02,
|
2847 |
+
"learning_rate": 3.3740157480314962e-06,
|
2848 |
+
"loss": 0.0905,
|
2849 |
+
"step": 463
|
2850 |
+
},
|
2851 |
+
{
|
2852 |
+
"epoch": 7.03,
|
2853 |
+
"learning_rate": 3.370078740157481e-06,
|
2854 |
+
"loss": 0.2391,
|
2855 |
+
"step": 464
|
2856 |
+
},
|
2857 |
+
{
|
2858 |
+
"epoch": 7.05,
|
2859 |
+
"learning_rate": 3.366141732283465e-06,
|
2860 |
+
"loss": 0.0839,
|
2861 |
+
"step": 465
|
2862 |
+
},
|
2863 |
+
{
|
2864 |
+
"epoch": 7.06,
|
2865 |
+
"learning_rate": 3.362204724409449e-06,
|
2866 |
+
"loss": 0.0859,
|
2867 |
+
"step": 466
|
2868 |
+
},
|
2869 |
+
{
|
2870 |
+
"epoch": 7.08,
|
2871 |
+
"learning_rate": 3.358267716535433e-06,
|
2872 |
+
"loss": 0.0863,
|
2873 |
+
"step": 467
|
2874 |
+
},
|
2875 |
+
{
|
2876 |
+
"epoch": 7.09,
|
2877 |
+
"learning_rate": 3.3543307086614173e-06,
|
2878 |
+
"loss": 0.169,
|
2879 |
+
"step": 468
|
2880 |
+
},
|
2881 |
+
{
|
2882 |
+
"epoch": 7.11,
|
2883 |
+
"learning_rate": 3.350393700787402e-06,
|
2884 |
+
"loss": 0.4396,
|
2885 |
+
"step": 469
|
2886 |
+
},
|
2887 |
+
{
|
2888 |
+
"epoch": 7.12,
|
2889 |
+
"learning_rate": 3.346456692913386e-06,
|
2890 |
+
"loss": 0.3005,
|
2891 |
+
"step": 470
|
2892 |
+
},
|
2893 |
+
{
|
2894 |
+
"epoch": 7.14,
|
2895 |
+
"learning_rate": 3.3425196850393704e-06,
|
2896 |
+
"loss": 0.0832,
|
2897 |
+
"step": 471
|
2898 |
+
},
|
2899 |
+
{
|
2900 |
+
"epoch": 7.15,
|
2901 |
+
"learning_rate": 3.3385826771653546e-06,
|
2902 |
+
"loss": 0.4338,
|
2903 |
+
"step": 472
|
2904 |
+
},
|
2905 |
+
{
|
2906 |
+
"epoch": 7.17,
|
2907 |
+
"learning_rate": 3.334645669291339e-06,
|
2908 |
+
"loss": 0.0898,
|
2909 |
+
"step": 473
|
2910 |
+
},
|
2911 |
+
{
|
2912 |
+
"epoch": 7.18,
|
2913 |
+
"learning_rate": 3.3307086614173227e-06,
|
2914 |
+
"loss": 0.0873,
|
2915 |
+
"step": 474
|
2916 |
+
},
|
2917 |
+
{
|
2918 |
+
"epoch": 7.2,
|
2919 |
+
"learning_rate": 3.3267716535433077e-06,
|
2920 |
+
"loss": 0.0856,
|
2921 |
+
"step": 475
|
2922 |
+
},
|
2923 |
+
{
|
2924 |
+
"epoch": 7.21,
|
2925 |
+
"learning_rate": 3.3228346456692915e-06,
|
2926 |
+
"loss": 0.6076,
|
2927 |
+
"step": 476
|
2928 |
+
},
|
2929 |
+
{
|
2930 |
+
"epoch": 7.23,
|
2931 |
+
"learning_rate": 3.3188976377952757e-06,
|
2932 |
+
"loss": 0.8815,
|
2933 |
+
"step": 477
|
2934 |
+
},
|
2935 |
+
{
|
2936 |
+
"epoch": 7.24,
|
2937 |
+
"learning_rate": 3.31496062992126e-06,
|
2938 |
+
"loss": 0.3032,
|
2939 |
+
"step": 478
|
2940 |
+
},
|
2941 |
+
{
|
2942 |
+
"epoch": 7.26,
|
2943 |
+
"learning_rate": 3.311023622047244e-06,
|
2944 |
+
"loss": 0.0893,
|
2945 |
+
"step": 479
|
2946 |
+
},
|
2947 |
+
{
|
2948 |
+
"epoch": 7.27,
|
2949 |
+
"learning_rate": 3.307086614173229e-06,
|
2950 |
+
"loss": 0.0791,
|
2951 |
+
"step": 480
|
2952 |
+
},
|
2953 |
+
{
|
2954 |
+
"epoch": 7.29,
|
2955 |
+
"learning_rate": 3.303149606299213e-06,
|
2956 |
+
"loss": 0.0757,
|
2957 |
+
"step": 481
|
2958 |
+
},
|
2959 |
+
{
|
2960 |
+
"epoch": 7.3,
|
2961 |
+
"learning_rate": 3.2992125984251973e-06,
|
2962 |
+
"loss": 0.0902,
|
2963 |
+
"step": 482
|
2964 |
+
},
|
2965 |
+
{
|
2966 |
+
"epoch": 7.32,
|
2967 |
+
"learning_rate": 3.2952755905511815e-06,
|
2968 |
+
"loss": 0.0793,
|
2969 |
+
"step": 483
|
2970 |
+
},
|
2971 |
+
{
|
2972 |
+
"epoch": 7.33,
|
2973 |
+
"learning_rate": 3.2913385826771653e-06,
|
2974 |
+
"loss": 0.7039,
|
2975 |
+
"step": 484
|
2976 |
+
},
|
2977 |
+
{
|
2978 |
+
"epoch": 7.35,
|
2979 |
+
"learning_rate": 3.2874015748031503e-06,
|
2980 |
+
"loss": 0.0776,
|
2981 |
+
"step": 485
|
2982 |
+
},
|
2983 |
+
{
|
2984 |
+
"epoch": 7.36,
|
2985 |
+
"learning_rate": 3.283464566929134e-06,
|
2986 |
+
"loss": 0.0861,
|
2987 |
+
"step": 486
|
2988 |
+
},
|
2989 |
+
{
|
2990 |
+
"epoch": 7.38,
|
2991 |
+
"learning_rate": 3.2795275590551184e-06,
|
2992 |
+
"loss": 0.9843,
|
2993 |
+
"step": 487
|
2994 |
+
},
|
2995 |
+
{
|
2996 |
+
"epoch": 7.39,
|
2997 |
+
"learning_rate": 3.2755905511811026e-06,
|
2998 |
+
"loss": 0.2719,
|
2999 |
+
"step": 488
|
3000 |
+
},
|
3001 |
+
{
|
3002 |
+
"epoch": 7.41,
|
3003 |
+
"learning_rate": 3.271653543307087e-06,
|
3004 |
+
"loss": 0.0826,
|
3005 |
+
"step": 489
|
3006 |
+
},
|
3007 |
+
{
|
3008 |
+
"epoch": 7.42,
|
3009 |
+
"learning_rate": 3.267716535433071e-06,
|
3010 |
+
"loss": 0.0747,
|
3011 |
+
"step": 490
|
3012 |
+
},
|
3013 |
+
{
|
3014 |
+
"epoch": 7.44,
|
3015 |
+
"learning_rate": 3.2637795275590557e-06,
|
3016 |
+
"loss": 0.0754,
|
3017 |
+
"step": 491
|
3018 |
+
},
|
3019 |
+
{
|
3020 |
+
"epoch": 7.45,
|
3021 |
+
"learning_rate": 3.25984251968504e-06,
|
3022 |
+
"loss": 0.0743,
|
3023 |
+
"step": 492
|
3024 |
+
},
|
3025 |
+
{
|
3026 |
+
"epoch": 7.47,
|
3027 |
+
"learning_rate": 3.2559055118110237e-06,
|
3028 |
+
"loss": 0.0739,
|
3029 |
+
"step": 493
|
3030 |
+
},
|
3031 |
+
{
|
3032 |
+
"epoch": 7.48,
|
3033 |
+
"learning_rate": 3.251968503937008e-06,
|
3034 |
+
"loss": 0.073,
|
3035 |
+
"step": 494
|
3036 |
+
},
|
3037 |
+
{
|
3038 |
+
"epoch": 7.5,
|
3039 |
+
"learning_rate": 3.248031496062992e-06,
|
3040 |
+
"loss": 0.7082,
|
3041 |
+
"step": 495
|
3042 |
+
},
|
3043 |
+
{
|
3044 |
+
"epoch": 7.52,
|
3045 |
+
"learning_rate": 3.2440944881889768e-06,
|
3046 |
+
"loss": 0.0758,
|
3047 |
+
"step": 496
|
3048 |
+
},
|
3049 |
+
{
|
3050 |
+
"epoch": 7.53,
|
3051 |
+
"learning_rate": 3.240157480314961e-06,
|
3052 |
+
"loss": 0.0714,
|
3053 |
+
"step": 497
|
3054 |
+
},
|
3055 |
+
{
|
3056 |
+
"epoch": 7.55,
|
3057 |
+
"learning_rate": 3.236220472440945e-06,
|
3058 |
+
"loss": 0.079,
|
3059 |
+
"step": 498
|
3060 |
+
},
|
3061 |
+
{
|
3062 |
+
"epoch": 7.56,
|
3063 |
+
"learning_rate": 3.2322834645669294e-06,
|
3064 |
+
"loss": 0.0703,
|
3065 |
+
"step": 499
|
3066 |
+
},
|
3067 |
+
{
|
3068 |
+
"epoch": 7.58,
|
3069 |
+
"learning_rate": 3.2283464566929136e-06,
|
3070 |
+
"loss": 0.0716,
|
3071 |
+
"step": 500
|
3072 |
+
},
|
3073 |
+
{
|
3074 |
+
"epoch": 7.59,
|
3075 |
+
"learning_rate": 3.2244094488188983e-06,
|
3076 |
+
"loss": 0.0742,
|
3077 |
+
"step": 501
|
3078 |
+
},
|
3079 |
+
{
|
3080 |
+
"epoch": 7.61,
|
3081 |
+
"learning_rate": 3.2204724409448825e-06,
|
3082 |
+
"loss": 0.0691,
|
3083 |
+
"step": 502
|
3084 |
+
},
|
3085 |
+
{
|
3086 |
+
"epoch": 7.62,
|
3087 |
+
"learning_rate": 3.2165354330708663e-06,
|
3088 |
+
"loss": 0.0697,
|
3089 |
+
"step": 503
|
3090 |
+
},
|
3091 |
+
{
|
3092 |
+
"epoch": 7.64,
|
3093 |
+
"learning_rate": 3.2125984251968505e-06,
|
3094 |
+
"loss": 0.0791,
|
3095 |
+
"step": 504
|
3096 |
+
},
|
3097 |
+
{
|
3098 |
+
"epoch": 7.65,
|
3099 |
+
"learning_rate": 3.2086614173228347e-06,
|
3100 |
+
"loss": 0.2838,
|
3101 |
+
"step": 505
|
3102 |
+
},
|
3103 |
+
{
|
3104 |
+
"epoch": 7.67,
|
3105 |
+
"learning_rate": 3.204724409448819e-06,
|
3106 |
+
"loss": 0.0683,
|
3107 |
+
"step": 506
|
3108 |
+
},
|
3109 |
+
{
|
3110 |
+
"epoch": 7.68,
|
3111 |
+
"learning_rate": 3.2007874015748036e-06,
|
3112 |
+
"loss": 0.4993,
|
3113 |
+
"step": 507
|
3114 |
+
},
|
3115 |
+
{
|
3116 |
+
"epoch": 7.7,
|
3117 |
+
"learning_rate": 3.196850393700788e-06,
|
3118 |
+
"loss": 0.2051,
|
3119 |
+
"step": 508
|
3120 |
+
},
|
3121 |
+
{
|
3122 |
+
"epoch": 7.71,
|
3123 |
+
"learning_rate": 3.192913385826772e-06,
|
3124 |
+
"loss": 0.0673,
|
3125 |
+
"step": 509
|
3126 |
+
},
|
3127 |
+
{
|
3128 |
+
"epoch": 7.73,
|
3129 |
+
"learning_rate": 3.1889763779527563e-06,
|
3130 |
+
"loss": 0.0676,
|
3131 |
+
"step": 510
|
3132 |
+
},
|
3133 |
+
{
|
3134 |
+
"epoch": 7.74,
|
3135 |
+
"learning_rate": 3.18503937007874e-06,
|
3136 |
+
"loss": 0.7316,
|
3137 |
+
"step": 511
|
3138 |
+
},
|
3139 |
+
{
|
3140 |
+
"epoch": 7.76,
|
3141 |
+
"learning_rate": 3.1811023622047247e-06,
|
3142 |
+
"loss": 0.065,
|
3143 |
+
"step": 512
|
3144 |
+
},
|
3145 |
+
{
|
3146 |
+
"epoch": 7.77,
|
3147 |
+
"learning_rate": 3.177165354330709e-06,
|
3148 |
+
"loss": 0.2066,
|
3149 |
+
"step": 513
|
3150 |
+
},
|
3151 |
+
{
|
3152 |
+
"epoch": 7.79,
|
3153 |
+
"learning_rate": 3.173228346456693e-06,
|
3154 |
+
"loss": 0.1082,
|
3155 |
+
"step": 514
|
3156 |
+
},
|
3157 |
+
{
|
3158 |
+
"epoch": 7.8,
|
3159 |
+
"learning_rate": 3.1692913385826774e-06,
|
3160 |
+
"loss": 0.1039,
|
3161 |
+
"step": 515
|
3162 |
+
},
|
3163 |
+
{
|
3164 |
+
"epoch": 7.82,
|
3165 |
+
"learning_rate": 3.1653543307086616e-06,
|
3166 |
+
"loss": 0.0752,
|
3167 |
+
"step": 516
|
3168 |
+
},
|
3169 |
+
{
|
3170 |
+
"epoch": 7.83,
|
3171 |
+
"learning_rate": 3.1614173228346462e-06,
|
3172 |
+
"loss": 0.0647,
|
3173 |
+
"step": 517
|
3174 |
+
},
|
3175 |
+
{
|
3176 |
+
"epoch": 7.85,
|
3177 |
+
"learning_rate": 3.1574803149606304e-06,
|
3178 |
+
"loss": 0.0636,
|
3179 |
+
"step": 518
|
3180 |
+
},
|
3181 |
+
{
|
3182 |
+
"epoch": 7.86,
|
3183 |
+
"learning_rate": 3.1535433070866147e-06,
|
3184 |
+
"loss": 0.2198,
|
3185 |
+
"step": 519
|
3186 |
+
},
|
3187 |
+
{
|
3188 |
+
"epoch": 7.88,
|
3189 |
+
"learning_rate": 3.1496062992125985e-06,
|
3190 |
+
"loss": 0.0719,
|
3191 |
+
"step": 520
|
3192 |
+
},
|
3193 |
+
{
|
3194 |
+
"epoch": 7.89,
|
3195 |
+
"learning_rate": 3.1456692913385827e-06,
|
3196 |
+
"loss": 0.0642,
|
3197 |
+
"step": 521
|
3198 |
+
},
|
3199 |
+
{
|
3200 |
+
"epoch": 7.91,
|
3201 |
+
"learning_rate": 3.141732283464567e-06,
|
3202 |
+
"loss": 0.0711,
|
3203 |
+
"step": 522
|
3204 |
+
},
|
3205 |
+
{
|
3206 |
+
"epoch": 7.92,
|
3207 |
+
"learning_rate": 3.1377952755905515e-06,
|
3208 |
+
"loss": 0.0645,
|
3209 |
+
"step": 523
|
3210 |
+
},
|
3211 |
+
{
|
3212 |
+
"epoch": 7.94,
|
3213 |
+
"learning_rate": 3.1338582677165358e-06,
|
3214 |
+
"loss": 0.1827,
|
3215 |
+
"step": 524
|
3216 |
+
},
|
3217 |
+
{
|
3218 |
+
"epoch": 7.95,
|
3219 |
+
"learning_rate": 3.12992125984252e-06,
|
3220 |
+
"loss": 0.0636,
|
3221 |
+
"step": 525
|
3222 |
+
},
|
3223 |
+
{
|
3224 |
+
"epoch": 7.97,
|
3225 |
+
"learning_rate": 3.125984251968504e-06,
|
3226 |
+
"loss": 0.0634,
|
3227 |
+
"step": 526
|
3228 |
+
},
|
3229 |
+
{
|
3230 |
+
"epoch": 7.98,
|
3231 |
+
"learning_rate": 3.1220472440944884e-06,
|
3232 |
+
"loss": 0.2832,
|
3233 |
+
"step": 527
|
3234 |
+
},
|
3235 |
+
{
|
3236 |
+
"epoch": 8.0,
|
3237 |
+
"learning_rate": 3.118110236220473e-06,
|
3238 |
+
"loss": 0.0683,
|
3239 |
+
"step": 528
|
3240 |
+
},
|
3241 |
+
{
|
3242 |
+
"epoch": 8.0,
|
3243 |
+
"eval_accuracy": 0.9545454545454546,
|
3244 |
+
"eval_loss": 0.18883031606674194,
|
3245 |
+
"eval_runtime": 6.9079,
|
3246 |
+
"eval_samples_per_second": 9.554,
|
3247 |
+
"eval_steps_per_second": 2.461,
|
3248 |
+
"step": 528
|
3249 |
+
},
|
3250 |
+
{
|
3251 |
+
"epoch": 8.02,
|
3252 |
+
"learning_rate": 3.1141732283464573e-06,
|
3253 |
+
"loss": 0.0693,
|
3254 |
+
"step": 529
|
3255 |
+
},
|
3256 |
+
{
|
3257 |
+
"epoch": 8.03,
|
3258 |
+
"learning_rate": 3.110236220472441e-06,
|
3259 |
+
"loss": 0.7524,
|
3260 |
+
"step": 530
|
3261 |
+
},
|
3262 |
+
{
|
3263 |
+
"epoch": 8.05,
|
3264 |
+
"learning_rate": 3.1062992125984253e-06,
|
3265 |
+
"loss": 0.1929,
|
3266 |
+
"step": 531
|
3267 |
+
},
|
3268 |
+
{
|
3269 |
+
"epoch": 8.06,
|
3270 |
+
"learning_rate": 3.1023622047244095e-06,
|
3271 |
+
"loss": 0.0604,
|
3272 |
+
"step": 532
|
3273 |
+
},
|
3274 |
+
{
|
3275 |
+
"epoch": 8.08,
|
3276 |
+
"learning_rate": 3.098425196850394e-06,
|
3277 |
+
"loss": 0.0603,
|
3278 |
+
"step": 533
|
3279 |
+
},
|
3280 |
+
{
|
3281 |
+
"epoch": 8.09,
|
3282 |
+
"learning_rate": 3.0944881889763784e-06,
|
3283 |
+
"loss": 0.1681,
|
3284 |
+
"step": 534
|
3285 |
+
},
|
3286 |
+
{
|
3287 |
+
"epoch": 8.11,
|
3288 |
+
"learning_rate": 3.0905511811023626e-06,
|
3289 |
+
"loss": 0.0663,
|
3290 |
+
"step": 535
|
3291 |
+
},
|
3292 |
+
{
|
3293 |
+
"epoch": 8.12,
|
3294 |
+
"learning_rate": 3.086614173228347e-06,
|
3295 |
+
"loss": 0.0625,
|
3296 |
+
"step": 536
|
3297 |
+
},
|
3298 |
+
{
|
3299 |
+
"epoch": 8.14,
|
3300 |
+
"learning_rate": 3.0826771653543306e-06,
|
3301 |
+
"loss": 0.0799,
|
3302 |
+
"step": 537
|
3303 |
+
},
|
3304 |
+
{
|
3305 |
+
"epoch": 8.15,
|
3306 |
+
"learning_rate": 3.078740157480315e-06,
|
3307 |
+
"loss": 0.29,
|
3308 |
+
"step": 538
|
3309 |
+
},
|
3310 |
+
{
|
3311 |
+
"epoch": 8.17,
|
3312 |
+
"learning_rate": 3.0748031496062995e-06,
|
3313 |
+
"loss": 0.5023,
|
3314 |
+
"step": 539
|
3315 |
+
},
|
3316 |
+
{
|
3317 |
+
"epoch": 8.18,
|
3318 |
+
"learning_rate": 3.0708661417322837e-06,
|
3319 |
+
"loss": 0.0811,
|
3320 |
+
"step": 540
|
3321 |
+
},
|
3322 |
+
{
|
3323 |
+
"epoch": 8.2,
|
3324 |
+
"learning_rate": 3.066929133858268e-06,
|
3325 |
+
"loss": 0.0576,
|
3326 |
+
"step": 541
|
3327 |
+
},
|
3328 |
+
{
|
3329 |
+
"epoch": 8.21,
|
3330 |
+
"learning_rate": 3.062992125984252e-06,
|
3331 |
+
"loss": 0.0572,
|
3332 |
+
"step": 542
|
3333 |
+
},
|
3334 |
+
{
|
3335 |
+
"epoch": 8.23,
|
3336 |
+
"learning_rate": 3.0590551181102364e-06,
|
3337 |
+
"loss": 0.0571,
|
3338 |
+
"step": 543
|
3339 |
+
},
|
3340 |
+
{
|
3341 |
+
"epoch": 8.24,
|
3342 |
+
"learning_rate": 3.055118110236221e-06,
|
3343 |
+
"loss": 0.0626,
|
3344 |
+
"step": 544
|
3345 |
+
},
|
3346 |
+
{
|
3347 |
+
"epoch": 8.26,
|
3348 |
+
"learning_rate": 3.0511811023622052e-06,
|
3349 |
+
"loss": 0.5849,
|
3350 |
+
"step": 545
|
3351 |
+
},
|
3352 |
+
{
|
3353 |
+
"epoch": 8.27,
|
3354 |
+
"learning_rate": 3.0472440944881894e-06,
|
3355 |
+
"loss": 0.7242,
|
3356 |
+
"step": 546
|
3357 |
+
},
|
3358 |
+
{
|
3359 |
+
"epoch": 8.29,
|
3360 |
+
"learning_rate": 3.0433070866141732e-06,
|
3361 |
+
"loss": 0.0564,
|
3362 |
+
"step": 547
|
3363 |
+
},
|
3364 |
+
{
|
3365 |
+
"epoch": 8.3,
|
3366 |
+
"learning_rate": 3.0393700787401575e-06,
|
3367 |
+
"loss": 0.0649,
|
3368 |
+
"step": 548
|
3369 |
+
},
|
3370 |
+
{
|
3371 |
+
"epoch": 8.32,
|
3372 |
+
"learning_rate": 3.035433070866142e-06,
|
3373 |
+
"loss": 0.0547,
|
3374 |
+
"step": 549
|
3375 |
+
},
|
3376 |
+
{
|
3377 |
+
"epoch": 8.33,
|
3378 |
+
"learning_rate": 3.0314960629921263e-06,
|
3379 |
+
"loss": 0.2119,
|
3380 |
+
"step": 550
|
3381 |
+
},
|
3382 |
+
{
|
3383 |
+
"epoch": 8.35,
|
3384 |
+
"learning_rate": 3.0275590551181105e-06,
|
3385 |
+
"loss": 0.0541,
|
3386 |
+
"step": 551
|
3387 |
+
},
|
3388 |
+
{
|
3389 |
+
"epoch": 8.36,
|
3390 |
+
"learning_rate": 3.0236220472440948e-06,
|
3391 |
+
"loss": 0.0551,
|
3392 |
+
"step": 552
|
3393 |
+
},
|
3394 |
+
{
|
3395 |
+
"epoch": 8.38,
|
3396 |
+
"learning_rate": 3.019685039370079e-06,
|
3397 |
+
"loss": 0.0596,
|
3398 |
+
"step": 553
|
3399 |
+
},
|
3400 |
+
{
|
3401 |
+
"epoch": 8.39,
|
3402 |
+
"learning_rate": 3.015748031496063e-06,
|
3403 |
+
"loss": 0.0533,
|
3404 |
+
"step": 554
|
3405 |
+
},
|
3406 |
+
{
|
3407 |
+
"epoch": 8.41,
|
3408 |
+
"learning_rate": 3.011811023622048e-06,
|
3409 |
+
"loss": 0.4296,
|
3410 |
+
"step": 555
|
3411 |
+
},
|
3412 |
+
{
|
3413 |
+
"epoch": 8.42,
|
3414 |
+
"learning_rate": 3.0078740157480316e-06,
|
3415 |
+
"loss": 0.1105,
|
3416 |
+
"step": 556
|
3417 |
+
},
|
3418 |
+
{
|
3419 |
+
"epoch": 8.44,
|
3420 |
+
"learning_rate": 3.003937007874016e-06,
|
3421 |
+
"loss": 0.054,
|
3422 |
+
"step": 557
|
3423 |
+
},
|
3424 |
+
{
|
3425 |
+
"epoch": 8.45,
|
3426 |
+
"learning_rate": 3e-06,
|
3427 |
+
"loss": 0.055,
|
3428 |
+
"step": 558
|
3429 |
+
},
|
3430 |
+
{
|
3431 |
+
"epoch": 8.47,
|
3432 |
+
"learning_rate": 2.9960629921259843e-06,
|
3433 |
+
"loss": 0.0517,
|
3434 |
+
"step": 559
|
3435 |
+
},
|
3436 |
+
{
|
3437 |
+
"epoch": 8.48,
|
3438 |
+
"learning_rate": 2.992125984251969e-06,
|
3439 |
+
"loss": 0.7577,
|
3440 |
+
"step": 560
|
3441 |
+
},
|
3442 |
+
{
|
3443 |
+
"epoch": 8.5,
|
3444 |
+
"learning_rate": 2.988188976377953e-06,
|
3445 |
+
"loss": 0.0598,
|
3446 |
+
"step": 561
|
3447 |
+
},
|
3448 |
+
{
|
3449 |
+
"epoch": 8.52,
|
3450 |
+
"learning_rate": 2.9842519685039374e-06,
|
3451 |
+
"loss": 0.0517,
|
3452 |
+
"step": 562
|
3453 |
+
},
|
3454 |
+
{
|
3455 |
+
"epoch": 8.53,
|
3456 |
+
"learning_rate": 2.9803149606299216e-06,
|
3457 |
+
"loss": 0.5103,
|
3458 |
+
"step": 563
|
3459 |
+
},
|
3460 |
+
{
|
3461 |
+
"epoch": 8.55,
|
3462 |
+
"learning_rate": 2.9763779527559054e-06,
|
3463 |
+
"loss": 0.0534,
|
3464 |
+
"step": 564
|
3465 |
+
},
|
3466 |
+
{
|
3467 |
+
"epoch": 8.56,
|
3468 |
+
"learning_rate": 2.9724409448818905e-06,
|
3469 |
+
"loss": 0.0527,
|
3470 |
+
"step": 565
|
3471 |
+
},
|
3472 |
+
{
|
3473 |
+
"epoch": 8.58,
|
3474 |
+
"learning_rate": 2.9685039370078743e-06,
|
3475 |
+
"loss": 0.0585,
|
3476 |
+
"step": 566
|
3477 |
+
},
|
3478 |
+
{
|
3479 |
+
"epoch": 8.59,
|
3480 |
+
"learning_rate": 2.9645669291338585e-06,
|
3481 |
+
"loss": 0.0834,
|
3482 |
+
"step": 567
|
3483 |
+
},
|
3484 |
+
{
|
3485 |
+
"epoch": 8.61,
|
3486 |
+
"learning_rate": 2.9606299212598427e-06,
|
3487 |
+
"loss": 0.7881,
|
3488 |
+
"step": 568
|
3489 |
+
},
|
3490 |
+
{
|
3491 |
+
"epoch": 8.62,
|
3492 |
+
"learning_rate": 2.956692913385827e-06,
|
3493 |
+
"loss": 0.071,
|
3494 |
+
"step": 569
|
3495 |
+
},
|
3496 |
+
{
|
3497 |
+
"epoch": 8.64,
|
3498 |
+
"learning_rate": 2.952755905511811e-06,
|
3499 |
+
"loss": 0.0518,
|
3500 |
+
"step": 570
|
3501 |
+
},
|
3502 |
+
{
|
3503 |
+
"epoch": 8.65,
|
3504 |
+
"learning_rate": 2.9488188976377958e-06,
|
3505 |
+
"loss": 0.0574,
|
3506 |
+
"step": 571
|
3507 |
+
},
|
3508 |
+
{
|
3509 |
+
"epoch": 8.67,
|
3510 |
+
"learning_rate": 2.94488188976378e-06,
|
3511 |
+
"loss": 0.0522,
|
3512 |
+
"step": 572
|
3513 |
+
},
|
3514 |
+
{
|
3515 |
+
"epoch": 8.68,
|
3516 |
+
"learning_rate": 2.9409448818897642e-06,
|
3517 |
+
"loss": 0.0486,
|
3518 |
+
"step": 573
|
3519 |
+
},
|
3520 |
+
{
|
3521 |
+
"epoch": 8.7,
|
3522 |
+
"learning_rate": 2.937007874015748e-06,
|
3523 |
+
"loss": 0.1236,
|
3524 |
+
"step": 574
|
3525 |
+
},
|
3526 |
+
{
|
3527 |
+
"epoch": 8.71,
|
3528 |
+
"learning_rate": 2.9330708661417322e-06,
|
3529 |
+
"loss": 0.0501,
|
3530 |
+
"step": 575
|
3531 |
+
},
|
3532 |
+
{
|
3533 |
+
"epoch": 8.73,
|
3534 |
+
"learning_rate": 2.929133858267717e-06,
|
3535 |
+
"loss": 0.0509,
|
3536 |
+
"step": 576
|
3537 |
+
},
|
3538 |
+
{
|
3539 |
+
"epoch": 8.74,
|
3540 |
+
"learning_rate": 2.925196850393701e-06,
|
3541 |
+
"loss": 0.0497,
|
3542 |
+
"step": 577
|
3543 |
+
},
|
3544 |
+
{
|
3545 |
+
"epoch": 8.76,
|
3546 |
+
"learning_rate": 2.9212598425196853e-06,
|
3547 |
+
"loss": 0.3165,
|
3548 |
+
"step": 578
|
3549 |
+
},
|
3550 |
+
{
|
3551 |
+
"epoch": 8.77,
|
3552 |
+
"learning_rate": 2.9173228346456695e-06,
|
3553 |
+
"loss": 0.0532,
|
3554 |
+
"step": 579
|
3555 |
+
},
|
3556 |
+
{
|
3557 |
+
"epoch": 8.79,
|
3558 |
+
"learning_rate": 2.9133858267716538e-06,
|
3559 |
+
"loss": 0.1789,
|
3560 |
+
"step": 580
|
3561 |
+
},
|
3562 |
+
{
|
3563 |
+
"epoch": 8.8,
|
3564 |
+
"learning_rate": 2.9094488188976384e-06,
|
3565 |
+
"loss": 0.119,
|
3566 |
+
"step": 581
|
3567 |
+
},
|
3568 |
+
{
|
3569 |
+
"epoch": 8.82,
|
3570 |
+
"learning_rate": 2.9055118110236226e-06,
|
3571 |
+
"loss": 0.0476,
|
3572 |
+
"step": 582
|
3573 |
+
},
|
3574 |
+
{
|
3575 |
+
"epoch": 8.83,
|
3576 |
+
"learning_rate": 2.9015748031496064e-06,
|
3577 |
+
"loss": 0.0482,
|
3578 |
+
"step": 583
|
3579 |
+
},
|
3580 |
+
{
|
3581 |
+
"epoch": 8.85,
|
3582 |
+
"learning_rate": 2.8976377952755906e-06,
|
3583 |
+
"loss": 0.2127,
|
3584 |
+
"step": 584
|
3585 |
+
},
|
3586 |
+
{
|
3587 |
+
"epoch": 8.86,
|
3588 |
+
"learning_rate": 2.893700787401575e-06,
|
3589 |
+
"loss": 0.0508,
|
3590 |
+
"step": 585
|
3591 |
+
},
|
3592 |
+
{
|
3593 |
+
"epoch": 8.88,
|
3594 |
+
"learning_rate": 2.889763779527559e-06,
|
3595 |
+
"loss": 0.0935,
|
3596 |
+
"step": 586
|
3597 |
+
},
|
3598 |
+
{
|
3599 |
+
"epoch": 8.89,
|
3600 |
+
"learning_rate": 2.8858267716535437e-06,
|
3601 |
+
"loss": 0.0468,
|
3602 |
+
"step": 587
|
3603 |
+
},
|
3604 |
+
{
|
3605 |
+
"epoch": 8.91,
|
3606 |
+
"learning_rate": 2.881889763779528e-06,
|
3607 |
+
"loss": 0.0463,
|
3608 |
+
"step": 588
|
3609 |
+
},
|
3610 |
+
{
|
3611 |
+
"epoch": 8.92,
|
3612 |
+
"learning_rate": 2.877952755905512e-06,
|
3613 |
+
"loss": 0.078,
|
3614 |
+
"step": 589
|
3615 |
+
},
|
3616 |
+
{
|
3617 |
+
"epoch": 8.94,
|
3618 |
+
"learning_rate": 2.8740157480314964e-06,
|
3619 |
+
"loss": 0.0486,
|
3620 |
+
"step": 590
|
3621 |
+
},
|
3622 |
+
{
|
3623 |
+
"epoch": 8.95,
|
3624 |
+
"learning_rate": 2.87007874015748e-06,
|
3625 |
+
"loss": 0.1974,
|
3626 |
+
"step": 591
|
3627 |
+
},
|
3628 |
+
{
|
3629 |
+
"epoch": 8.97,
|
3630 |
+
"learning_rate": 2.8661417322834652e-06,
|
3631 |
+
"loss": 1.5522,
|
3632 |
+
"step": 592
|
3633 |
+
},
|
3634 |
+
{
|
3635 |
+
"epoch": 8.98,
|
3636 |
+
"learning_rate": 2.862204724409449e-06,
|
3637 |
+
"loss": 0.046,
|
3638 |
+
"step": 593
|
3639 |
+
},
|
3640 |
+
{
|
3641 |
+
"epoch": 9.0,
|
3642 |
+
"learning_rate": 2.8582677165354333e-06,
|
3643 |
+
"loss": 0.0572,
|
3644 |
+
"step": 594
|
3645 |
+
},
|
3646 |
+
{
|
3647 |
+
"epoch": 9.0,
|
3648 |
+
"eval_accuracy": 0.9545454545454546,
|
3649 |
+
"eval_loss": 0.18054482340812683,
|
3650 |
+
"eval_runtime": 6.2699,
|
3651 |
+
"eval_samples_per_second": 10.526,
|
3652 |
+
"eval_steps_per_second": 2.711,
|
3653 |
+
"step": 594
|
3654 |
+
}
|
3655 |
+
],
|
3656 |
+
"max_steps": 1320,
|
3657 |
+
"num_train_epochs": 20,
|
3658 |
+
"total_flos": 2.0278844132667123e+17,
|
3659 |
+
"trial_name": null,
|
3660 |
+
"trial_params": null
|
3661 |
+
}
|
checkpoint-594/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d4fba6b704dfb45312fb9c8b716fbccad21fadbd01864cb25088260308dbeef3
|
3 |
+
size 3567
|
config.json
ADDED
@@ -0,0 +1,113 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "facebook/wav2vec2-base-960h",
|
3 |
+
"activation_dropout": 0.1,
|
4 |
+
"adapter_kernel_size": 3,
|
5 |
+
"adapter_stride": 2,
|
6 |
+
"add_adapter": false,
|
7 |
+
"apply_spec_augment": true,
|
8 |
+
"architectures": [
|
9 |
+
"Wav2Vec2ForSequenceClassification"
|
10 |
+
],
|
11 |
+
"attention_dropout": 0.1,
|
12 |
+
"bos_token_id": 1,
|
13 |
+
"classifier_proj_size": 256,
|
14 |
+
"codevector_dim": 256,
|
15 |
+
"contrastive_logits_temperature": 0.1,
|
16 |
+
"conv_bias": false,
|
17 |
+
"conv_dim": [
|
18 |
+
512,
|
19 |
+
512,
|
20 |
+
512,
|
21 |
+
512,
|
22 |
+
512,
|
23 |
+
512,
|
24 |
+
512
|
25 |
+
],
|
26 |
+
"conv_kernel": [
|
27 |
+
10,
|
28 |
+
3,
|
29 |
+
3,
|
30 |
+
3,
|
31 |
+
3,
|
32 |
+
2,
|
33 |
+
2
|
34 |
+
],
|
35 |
+
"conv_stride": [
|
36 |
+
5,
|
37 |
+
2,
|
38 |
+
2,
|
39 |
+
2,
|
40 |
+
2,
|
41 |
+
2,
|
42 |
+
2
|
43 |
+
],
|
44 |
+
"ctc_loss_reduction": "sum",
|
45 |
+
"ctc_zero_infinity": false,
|
46 |
+
"diversity_loss_weight": 0.1,
|
47 |
+
"do_stable_layer_norm": false,
|
48 |
+
"eos_token_id": 2,
|
49 |
+
"feat_extract_activation": "gelu",
|
50 |
+
"feat_extract_dropout": 0.0,
|
51 |
+
"feat_extract_norm": "group",
|
52 |
+
"feat_proj_dropout": 0.1,
|
53 |
+
"feat_quantizer_dropout": 0.0,
|
54 |
+
"final_dropout": 0.1,
|
55 |
+
"gradient_checkpointing": false,
|
56 |
+
"hidden_act": "gelu",
|
57 |
+
"hidden_dropout": 0.1,
|
58 |
+
"hidden_dropout_prob": 0.1,
|
59 |
+
"hidden_size": 768,
|
60 |
+
"id2label": {
|
61 |
+
"0": "cat",
|
62 |
+
"1": "dog"
|
63 |
+
},
|
64 |
+
"initializer_range": 0.02,
|
65 |
+
"intermediate_size": 3072,
|
66 |
+
"layer_norm_eps": 1e-05,
|
67 |
+
"layerdrop": 0.1,
|
68 |
+
"mask_feature_length": 10,
|
69 |
+
"mask_feature_min_masks": 0,
|
70 |
+
"mask_feature_prob": 0.0,
|
71 |
+
"mask_time_length": 10,
|
72 |
+
"mask_time_min_masks": 2,
|
73 |
+
"mask_time_prob": 0.05,
|
74 |
+
"model_type": "wav2vec2",
|
75 |
+
"num_adapter_layers": 3,
|
76 |
+
"num_attention_heads": 12,
|
77 |
+
"num_codevector_groups": 2,
|
78 |
+
"num_codevectors_per_group": 320,
|
79 |
+
"num_conv_pos_embedding_groups": 16,
|
80 |
+
"num_conv_pos_embeddings": 128,
|
81 |
+
"num_feat_extract_layers": 7,
|
82 |
+
"num_hidden_layers": 12,
|
83 |
+
"num_negatives": 100,
|
84 |
+
"output_hidden_size": 768,
|
85 |
+
"pad_token_id": 0,
|
86 |
+
"proj_codevector_dim": 256,
|
87 |
+
"tdnn_dilation": [
|
88 |
+
1,
|
89 |
+
2,
|
90 |
+
3,
|
91 |
+
1,
|
92 |
+
1
|
93 |
+
],
|
94 |
+
"tdnn_dim": [
|
95 |
+
512,
|
96 |
+
512,
|
97 |
+
512,
|
98 |
+
512,
|
99 |
+
1500
|
100 |
+
],
|
101 |
+
"tdnn_kernel": [
|
102 |
+
5,
|
103 |
+
3,
|
104 |
+
3,
|
105 |
+
1,
|
106 |
+
1
|
107 |
+
],
|
108 |
+
"torch_dtype": "float32",
|
109 |
+
"transformers_version": "4.28.1",
|
110 |
+
"use_weighted_layer_sum": false,
|
111 |
+
"vocab_size": 32,
|
112 |
+
"xvector_output_dim": 512
|
113 |
+
}
|
preprocessor_config.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"do_normalize": true,
|
3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
4 |
+
"feature_size": 1,
|
5 |
+
"padding_side": "right",
|
6 |
+
"padding_value": 0.0,
|
7 |
+
"return_attention_mask": false,
|
8 |
+
"sampling_rate": 16000
|
9 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1c7703931a4da1e56d95c6aa0a0b88792476b5f3de2ecae89d6337234dec2b08
|
3 |
+
size 378348563
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d4fba6b704dfb45312fb9c8b716fbccad21fadbd01864cb25088260308dbeef3
|
3 |
+
size 3567
|