Hugging Face's logo

aleandrananu
/
pegasus-dialogue

Text2Text Generation
Transformers
PyTorch
pegasus
Model card Files Community
aleandrananu commited on
Commit
3ce7ec5
·
1 Parent(s): e590b21

Upload 11 files

Browse files
Files changed (11) hide show
  1. config.json +55 -0
  2. generation_config.json +12 -0
  3. optimizer.pt +3 -0
  4. pytorch_model.bin +3 -0
  5. rng_state.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +110 -0
  8. spiece.model +3 -0
  9. tokenizer_config.json +117 -0
  10. trainer_state.json +4096 -0
  11. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "tuner007/pegasus_paraphrase",
3
+ "activation_dropout": 0.1,
4
+ "activation_function": "relu",
5
+ "add_bias_logits": false,
6
+ "add_final_layer_norm": true,
7
+ "architectures": [
8
+ "PegasusForConditionalGeneration"
9
+ ],
10
+ "attention_dropout": 0.1,
11
+ "bos_token_id": 0,
12
+ "classif_dropout": 0.0,
13
+ "d_model": 1024,
14
+ "decoder_attention_heads": 16,
15
+ "decoder_ffn_dim": 4096,
16
+ "decoder_layerdrop": 0.0,
17
+ "decoder_layers": 16,
18
+ "decoder_start_token_id": 0,
19
+ "dropout": 0.1,
20
+ "encoder_attention_heads": 16,
21
+ "encoder_ffn_dim": 4096,
22
+ "encoder_layerdrop": 0.0,
23
+ "encoder_layers": 16,
24
+ "eos_token_id": 1,
25
+ "extra_pos_embeddings": 1,
26
+ "force_bos_token_to_be_generated": false,
27
+ "forced_eos_token_id": 1,
28
+ "id2label": {
29
+ "0": "LABEL_0",
30
+ "1": "LABEL_1",
31
+ "2": "LABEL_2"
32
+ },
33
+ "init_std": 0.02,
34
+ "is_encoder_decoder": true,
35
+ "label2id": {
36
+ "LABEL_0": 0,
37
+ "LABEL_1": 1,
38
+ "LABEL_2": 2
39
+ },
40
+ "length_penalty": 0.8,
41
+ "max_length": 60,
42
+ "max_position_embeddings": 60,
43
+ "model_type": "pegasus",
44
+ "normalize_before": true,
45
+ "normalize_embedding": false,
46
+ "num_beams": 8,
47
+ "num_hidden_layers": 16,
48
+ "pad_token_id": 0,
49
+ "scale_embedding": true,
50
+ "static_position_embeddings": true,
51
+ "torch_dtype": "float32",
52
+ "transformers_version": "4.29.1",
53
+ "use_cache": true,
54
+ "vocab_size": 96103
55
+ }
generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "decoder_start_token_id": 0,
5
+ "eos_token_id": 1,
6
+ "forced_eos_token_id": 1,
7
+ "length_penalty": 0.8,
8
+ "max_length": 60,
9
+ "num_beams": 8,
10
+ "pad_token_id": 0,
11
+ "transformers_version": "4.29.1"
12
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b0d19a53563a8f61d8c9d10352b5cb5e4b25466bbbc67235cc2e64b6630eefb
3
+ size 393216
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5821b6f515c22cb21b04fde1525f04595a18fdc927acfd293b1656a7c2916b23
3
+ size 393216
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8cfb5da9ea4ec797d30806f1ac2fd0637158600f937ecb588a729c5f577ee4d
3
+ size 14575
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd5738d7616124d06402c192261d1d2f0f61928ab62267f674a0a5ec98eb462a
3
+ size 627
special_tokens_map.json ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<mask_1>",
4
+ "<unk_2>",
5
+ "<unk_3>",
6
+ "<unk_4>",
7
+ "<unk_5>",
8
+ "<unk_6>",
9
+ "<unk_7>",
10
+ "<unk_8>",
11
+ "<unk_9>",
12
+ "<unk_10>",
13
+ "<unk_11>",
14
+ "<unk_12>",
15
+ "<unk_13>",
16
+ "<unk_14>",
17
+ "<unk_15>",
18
+ "<unk_16>",
19
+ "<unk_17>",
20
+ "<unk_18>",
21
+ "<unk_19>",
22
+ "<unk_20>",
23
+ "<unk_21>",
24
+ "<unk_22>",
25
+ "<unk_23>",
26
+ "<unk_24>",
27
+ "<unk_25>",
28
+ "<unk_26>",
29
+ "<unk_27>",
30
+ "<unk_28>",
31
+ "<unk_29>",
32
+ "<unk_30>",
33
+ "<unk_31>",
34
+ "<unk_32>",
35
+ "<unk_33>",
36
+ "<unk_34>",
37
+ "<unk_35>",
38
+ "<unk_36>",
39
+ "<unk_37>",
40
+ "<unk_38>",
41
+ "<unk_39>",
42
+ "<unk_40>",
43
+ "<unk_41>",
44
+ "<unk_42>",
45
+ "<unk_43>",
46
+ "<unk_44>",
47
+ "<unk_45>",
48
+ "<unk_46>",
49
+ "<unk_47>",
50
+ "<unk_48>",
51
+ "<unk_49>",
52
+ "<unk_50>",
53
+ "<unk_51>",
54
+ "<unk_52>",
55
+ "<unk_53>",
56
+ "<unk_54>",
57
+ "<unk_55>",
58
+ "<unk_56>",
59
+ "<unk_57>",
60
+ "<unk_58>",
61
+ "<unk_59>",
62
+ "<unk_60>",
63
+ "<unk_61>",
64
+ "<unk_62>",
65
+ "<unk_63>",
66
+ "<unk_64>",
67
+ "<unk_65>",
68
+ "<unk_66>",
69
+ "<unk_67>",
70
+ "<unk_68>",
71
+ "<unk_69>",
72
+ "<unk_70>",
73
+ "<unk_71>",
74
+ "<unk_72>",
75
+ "<unk_73>",
76
+ "<unk_74>",
77
+ "<unk_75>",
78
+ "<unk_76>",
79
+ "<unk_77>",
80
+ "<unk_78>",
81
+ "<unk_79>",
82
+ "<unk_80>",
83
+ "<unk_81>",
84
+ "<unk_82>",
85
+ "<unk_83>",
86
+ "<unk_84>",
87
+ "<unk_85>",
88
+ "<unk_86>",
89
+ "<unk_87>",
90
+ "<unk_88>",
91
+ "<unk_89>",
92
+ "<unk_90>",
93
+ "<unk_91>",
94
+ "<unk_92>",
95
+ "<unk_93>",
96
+ "<unk_94>",
97
+ "<unk_95>",
98
+ "<unk_96>",
99
+ "<unk_97>",
100
+ "<unk_98>",
101
+ "<unk_99>",
102
+ "<unk_100>",
103
+ "<unk_101>",
104
+ "<unk_102>"
105
+ ],
106
+ "eos_token": "</s>",
107
+ "mask_token": "<mask_2>",
108
+ "pad_token": "<pad>",
109
+ "unk_token": "<unk>"
110
+ }
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6b9fc5c9298ff73a5a3005e5edc58a475c3961887ab9b60e11a69fdafe9edb8
3
+ size 393216
tokenizer_config.json ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<mask_1>",
4
+ "<unk_2>",
5
+ "<unk_3>",
6
+ "<unk_4>",
7
+ "<unk_5>",
8
+ "<unk_6>",
9
+ "<unk_7>",
10
+ "<unk_8>",
11
+ "<unk_9>",
12
+ "<unk_10>",
13
+ "<unk_11>",
14
+ "<unk_12>",
15
+ "<unk_13>",
16
+ "<unk_14>",
17
+ "<unk_15>",
18
+ "<unk_16>",
19
+ "<unk_17>",
20
+ "<unk_18>",
21
+ "<unk_19>",
22
+ "<unk_20>",
23
+ "<unk_21>",
24
+ "<unk_22>",
25
+ "<unk_23>",
26
+ "<unk_24>",
27
+ "<unk_25>",
28
+ "<unk_26>",
29
+ "<unk_27>",
30
+ "<unk_28>",
31
+ "<unk_29>",
32
+ "<unk_30>",
33
+ "<unk_31>",
34
+ "<unk_32>",
35
+ "<unk_33>",
36
+ "<unk_34>",
37
+ "<unk_35>",
38
+ "<unk_36>",
39
+ "<unk_37>",
40
+ "<unk_38>",
41
+ "<unk_39>",
42
+ "<unk_40>",
43
+ "<unk_41>",
44
+ "<unk_42>",
45
+ "<unk_43>",
46
+ "<unk_44>",
47
+ "<unk_45>",
48
+ "<unk_46>",
49
+ "<unk_47>",
50
+ "<unk_48>",
51
+ "<unk_49>",
52
+ "<unk_50>",
53
+ "<unk_51>",
54
+ "<unk_52>",
55
+ "<unk_53>",
56
+ "<unk_54>",
57
+ "<unk_55>",
58
+ "<unk_56>",
59
+ "<unk_57>",
60
+ "<unk_58>",
61
+ "<unk_59>",
62
+ "<unk_60>",
63
+ "<unk_61>",
64
+ "<unk_62>",
65
+ "<unk_63>",
66
+ "<unk_64>",
67
+ "<unk_65>",
68
+ "<unk_66>",
69
+ "<unk_67>",
70
+ "<unk_68>",
71
+ "<unk_69>",
72
+ "<unk_70>",
73
+ "<unk_71>",
74
+ "<unk_72>",
75
+ "<unk_73>",
76
+ "<unk_74>",
77
+ "<unk_75>",
78
+ "<unk_76>",
79
+ "<unk_77>",
80
+ "<unk_78>",
81
+ "<unk_79>",
82
+ "<unk_80>",
83
+ "<unk_81>",
84
+ "<unk_82>",
85
+ "<unk_83>",
86
+ "<unk_84>",
87
+ "<unk_85>",
88
+ "<unk_86>",
89
+ "<unk_87>",
90
+ "<unk_88>",
91
+ "<unk_89>",
92
+ "<unk_90>",
93
+ "<unk_91>",
94
+ "<unk_92>",
95
+ "<unk_93>",
96
+ "<unk_94>",
97
+ "<unk_95>",
98
+ "<unk_96>",
99
+ "<unk_97>",
100
+ "<unk_98>",
101
+ "<unk_99>",
102
+ "<unk_100>",
103
+ "<unk_101>",
104
+ "<unk_102>"
105
+ ],
106
+ "clean_up_tokenization_spaces": true,
107
+ "eos_token": "</s>",
108
+ "full_tokenizer_file": null,
109
+ "mask_token": "<mask_2>",
110
+ "mask_token_sent": "<mask_1>",
111
+ "model_max_length": 60,
112
+ "offset": 103,
113
+ "pad_token": "<pad>",
114
+ "sp_model_kwargs": {},
115
+ "tokenizer_class": "PegasusTokenizer",
116
+ "unk_token": "<unk>"
117
+ }
trainer_state.json ADDED
@@ -0,0 +1,4096 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9765625,
5
+ "global_step": 6000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 1.0000000000000002e-06,
13
+ "loss": 9.5302,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 2.0000000000000003e-06,
19
+ "loss": 9.5978,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 3e-06,
25
+ "loss": 9.6702,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 4.000000000000001e-06,
31
+ "loss": 9.4757,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 5e-06,
37
+ "loss": 9.4232,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 6e-06,
43
+ "loss": 9.3929,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 7.000000000000001e-06,
49
+ "loss": 9.0641,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "learning_rate": 8.000000000000001e-06,
55
+ "loss": 9.1413,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.01,
60
+ "learning_rate": 9e-06,
61
+ "loss": 8.7706,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.02,
66
+ "learning_rate": 1e-05,
67
+ "loss": 8.7685,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.02,
72
+ "eval_loss": 8.517706871032715,
73
+ "eval_runtime": 138.145,
74
+ "eval_samples_per_second": 203.648,
75
+ "eval_steps_per_second": 6.37,
76
+ "step": 100
77
+ },
78
+ {
79
+ "epoch": 0.02,
80
+ "learning_rate": 1.1000000000000001e-05,
81
+ "loss": 8.6076,
82
+ "step": 110
83
+ },
84
+ {
85
+ "epoch": 0.02,
86
+ "learning_rate": 1.2e-05,
87
+ "loss": 8.5143,
88
+ "step": 120
89
+ },
90
+ {
91
+ "epoch": 0.02,
92
+ "learning_rate": 1.3000000000000001e-05,
93
+ "loss": 8.4501,
94
+ "step": 130
95
+ },
96
+ {
97
+ "epoch": 0.02,
98
+ "learning_rate": 1.4000000000000001e-05,
99
+ "loss": 8.3463,
100
+ "step": 140
101
+ },
102
+ {
103
+ "epoch": 0.02,
104
+ "learning_rate": 1.5e-05,
105
+ "loss": 8.2408,
106
+ "step": 150
107
+ },
108
+ {
109
+ "epoch": 0.03,
110
+ "learning_rate": 1.6000000000000003e-05,
111
+ "loss": 8.1665,
112
+ "step": 160
113
+ },
114
+ {
115
+ "epoch": 0.03,
116
+ "learning_rate": 1.7000000000000003e-05,
117
+ "loss": 7.8703,
118
+ "step": 170
119
+ },
120
+ {
121
+ "epoch": 0.03,
122
+ "learning_rate": 1.8e-05,
123
+ "loss": 7.8592,
124
+ "step": 180
125
+ },
126
+ {
127
+ "epoch": 0.03,
128
+ "learning_rate": 1.9e-05,
129
+ "loss": 7.7984,
130
+ "step": 190
131
+ },
132
+ {
133
+ "epoch": 0.03,
134
+ "learning_rate": 2e-05,
135
+ "loss": 7.7686,
136
+ "step": 200
137
+ },
138
+ {
139
+ "epoch": 0.03,
140
+ "eval_loss": 7.766874313354492,
141
+ "eval_runtime": 138.1211,
142
+ "eval_samples_per_second": 203.684,
143
+ "eval_steps_per_second": 6.371,
144
+ "step": 200
145
+ },
146
+ {
147
+ "epoch": 0.03,
148
+ "learning_rate": 2.1e-05,
149
+ "loss": 7.8618,
150
+ "step": 210
151
+ },
152
+ {
153
+ "epoch": 0.04,
154
+ "learning_rate": 2.2000000000000003e-05,
155
+ "loss": 7.6736,
156
+ "step": 220
157
+ },
158
+ {
159
+ "epoch": 0.04,
160
+ "learning_rate": 2.3000000000000003e-05,
161
+ "loss": 7.6716,
162
+ "step": 230
163
+ },
164
+ {
165
+ "epoch": 0.04,
166
+ "learning_rate": 2.4e-05,
167
+ "loss": 7.4289,
168
+ "step": 240
169
+ },
170
+ {
171
+ "epoch": 0.04,
172
+ "learning_rate": 2.5e-05,
173
+ "loss": 7.3674,
174
+ "step": 250
175
+ },
176
+ {
177
+ "epoch": 0.04,
178
+ "learning_rate": 2.6000000000000002e-05,
179
+ "loss": 7.2308,
180
+ "step": 260
181
+ },
182
+ {
183
+ "epoch": 0.04,
184
+ "learning_rate": 2.7000000000000002e-05,
185
+ "loss": 7.0519,
186
+ "step": 270
187
+ },
188
+ {
189
+ "epoch": 0.05,
190
+ "learning_rate": 2.8000000000000003e-05,
191
+ "loss": 6.8731,
192
+ "step": 280
193
+ },
194
+ {
195
+ "epoch": 0.05,
196
+ "learning_rate": 2.9e-05,
197
+ "loss": 6.676,
198
+ "step": 290
199
+ },
200
+ {
201
+ "epoch": 0.05,
202
+ "learning_rate": 3e-05,
203
+ "loss": 6.496,
204
+ "step": 300
205
+ },
206
+ {
207
+ "epoch": 0.05,
208
+ "eval_loss": 5.7125983238220215,
209
+ "eval_runtime": 138.1818,
210
+ "eval_samples_per_second": 203.594,
211
+ "eval_steps_per_second": 6.368,
212
+ "step": 300
213
+ },
214
+ {
215
+ "epoch": 0.05,
216
+ "learning_rate": 3.1e-05,
217
+ "loss": 6.0497,
218
+ "step": 310
219
+ },
220
+ {
221
+ "epoch": 0.05,
222
+ "learning_rate": 3.2000000000000005e-05,
223
+ "loss": 5.4927,
224
+ "step": 320
225
+ },
226
+ {
227
+ "epoch": 0.05,
228
+ "learning_rate": 3.3e-05,
229
+ "loss": 5.0328,
230
+ "step": 330
231
+ },
232
+ {
233
+ "epoch": 0.06,
234
+ "learning_rate": 3.4000000000000007e-05,
235
+ "loss": 4.5907,
236
+ "step": 340
237
+ },
238
+ {
239
+ "epoch": 0.06,
240
+ "learning_rate": 3.5e-05,
241
+ "loss": 3.9888,
242
+ "step": 350
243
+ },
244
+ {
245
+ "epoch": 0.06,
246
+ "learning_rate": 3.6e-05,
247
+ "loss": 3.4476,
248
+ "step": 360
249
+ },
250
+ {
251
+ "epoch": 0.06,
252
+ "learning_rate": 3.7e-05,
253
+ "loss": 2.8771,
254
+ "step": 370
255
+ },
256
+ {
257
+ "epoch": 0.06,
258
+ "learning_rate": 3.8e-05,
259
+ "loss": 2.3019,
260
+ "step": 380
261
+ },
262
+ {
263
+ "epoch": 0.06,
264
+ "learning_rate": 3.9000000000000006e-05,
265
+ "loss": 1.7391,
266
+ "step": 390
267
+ },
268
+ {
269
+ "epoch": 0.07,
270
+ "learning_rate": 4e-05,
271
+ "loss": 1.3205,
272
+ "step": 400
273
+ },
274
+ {
275
+ "epoch": 0.07,
276
+ "eval_loss": 0.6095399260520935,
277
+ "eval_runtime": 138.1438,
278
+ "eval_samples_per_second": 203.65,
279
+ "eval_steps_per_second": 6.37,
280
+ "step": 400
281
+ },
282
+ {
283
+ "epoch": 0.07,
284
+ "learning_rate": 4.1e-05,
285
+ "loss": 0.9919,
286
+ "step": 410
287
+ },
288
+ {
289
+ "epoch": 0.07,
290
+ "learning_rate": 4.2e-05,
291
+ "loss": 0.7867,
292
+ "step": 420
293
+ },
294
+ {
295
+ "epoch": 0.07,
296
+ "learning_rate": 4.3e-05,
297
+ "loss": 0.6837,
298
+ "step": 430
299
+ },
300
+ {
301
+ "epoch": 0.07,
302
+ "learning_rate": 4.4000000000000006e-05,
303
+ "loss": 0.6113,
304
+ "step": 440
305
+ },
306
+ {
307
+ "epoch": 0.07,
308
+ "learning_rate": 4.5e-05,
309
+ "loss": 0.6006,
310
+ "step": 450
311
+ },
312
+ {
313
+ "epoch": 0.07,
314
+ "learning_rate": 4.600000000000001e-05,
315
+ "loss": 0.5477,
316
+ "step": 460
317
+ },
318
+ {
319
+ "epoch": 0.08,
320
+ "learning_rate": 4.7e-05,
321
+ "loss": 0.5483,
322
+ "step": 470
323
+ },
324
+ {
325
+ "epoch": 0.08,
326
+ "learning_rate": 4.8e-05,
327
+ "loss": 0.5149,
328
+ "step": 480
329
+ },
330
+ {
331
+ "epoch": 0.08,
332
+ "learning_rate": 4.9e-05,
333
+ "loss": 0.5213,
334
+ "step": 490
335
+ },
336
+ {
337
+ "epoch": 0.08,
338
+ "learning_rate": 5e-05,
339
+ "loss": 0.5004,
340
+ "step": 500
341
+ },
342
+ {
343
+ "epoch": 0.08,
344
+ "eval_loss": 0.4818989634513855,
345
+ "eval_runtime": 138.1607,
346
+ "eval_samples_per_second": 203.625,
347
+ "eval_steps_per_second": 6.369,
348
+ "step": 500
349
+ },
350
+ {
351
+ "epoch": 0.08,
352
+ "learning_rate": 4.9911410347271445e-05,
353
+ "loss": 0.5115,
354
+ "step": 510
355
+ },
356
+ {
357
+ "epoch": 0.08,
358
+ "learning_rate": 4.982282069454288e-05,
359
+ "loss": 0.5135,
360
+ "step": 520
361
+ },
362
+ {
363
+ "epoch": 0.09,
364
+ "learning_rate": 4.973423104181432e-05,
365
+ "loss": 0.4966,
366
+ "step": 530
367
+ },
368
+ {
369
+ "epoch": 0.09,
370
+ "learning_rate": 4.964564138908575e-05,
371
+ "loss": 0.4864,
372
+ "step": 540
373
+ },
374
+ {
375
+ "epoch": 0.09,
376
+ "learning_rate": 4.9557051736357196e-05,
377
+ "loss": 0.4763,
378
+ "step": 550
379
+ },
380
+ {
381
+ "epoch": 0.09,
382
+ "learning_rate": 4.946846208362864e-05,
383
+ "loss": 0.5101,
384
+ "step": 560
385
+ },
386
+ {
387
+ "epoch": 0.09,
388
+ "learning_rate": 4.9379872430900075e-05,
389
+ "loss": 0.4859,
390
+ "step": 570
391
+ },
392
+ {
393
+ "epoch": 0.09,
394
+ "learning_rate": 4.929128277817151e-05,
395
+ "loss": 0.5046,
396
+ "step": 580
397
+ },
398
+ {
399
+ "epoch": 0.1,
400
+ "learning_rate": 4.920269312544295e-05,
401
+ "loss": 0.475,
402
+ "step": 590
403
+ },
404
+ {
405
+ "epoch": 0.1,
406
+ "learning_rate": 4.911410347271439e-05,
407
+ "loss": 0.4774,
408
+ "step": 600
409
+ },
410
+ {
411
+ "epoch": 0.1,
412
+ "eval_loss": 0.45798131823539734,
413
+ "eval_runtime": 138.1059,
414
+ "eval_samples_per_second": 203.706,
415
+ "eval_steps_per_second": 6.372,
416
+ "step": 600
417
+ },
418
+ {
419
+ "epoch": 0.1,
420
+ "learning_rate": 4.902551381998583e-05,
421
+ "loss": 0.4966,
422
+ "step": 610
423
+ },
424
+ {
425
+ "epoch": 0.1,
426
+ "learning_rate": 4.893692416725726e-05,
427
+ "loss": 0.4836,
428
+ "step": 620
429
+ },
430
+ {
431
+ "epoch": 0.1,
432
+ "learning_rate": 4.8848334514528704e-05,
433
+ "loss": 0.4713,
434
+ "step": 630
435
+ },
436
+ {
437
+ "epoch": 0.1,
438
+ "learning_rate": 4.875974486180015e-05,
439
+ "loss": 0.4489,
440
+ "step": 640
441
+ },
442
+ {
443
+ "epoch": 0.11,
444
+ "learning_rate": 4.867115520907158e-05,
445
+ "loss": 0.4846,
446
+ "step": 650
447
+ },
448
+ {
449
+ "epoch": 0.11,
450
+ "learning_rate": 4.8582565556343026e-05,
451
+ "loss": 0.465,
452
+ "step": 660
453
+ },
454
+ {
455
+ "epoch": 0.11,
456
+ "learning_rate": 4.8493975903614455e-05,
457
+ "loss": 0.4256,
458
+ "step": 670
459
+ },
460
+ {
461
+ "epoch": 0.11,
462
+ "learning_rate": 4.84053862508859e-05,
463
+ "loss": 0.4559,
464
+ "step": 680
465
+ },
466
+ {
467
+ "epoch": 0.11,
468
+ "learning_rate": 4.831679659815734e-05,
469
+ "loss": 0.4599,
470
+ "step": 690
471
+ },
472
+ {
473
+ "epoch": 0.11,
474
+ "learning_rate": 4.822820694542878e-05,
475
+ "loss": 0.4545,
476
+ "step": 700
477
+ },
478
+ {
479
+ "epoch": 0.11,
480
+ "eval_loss": 0.4460126459598541,
481
+ "eval_runtime": 138.1094,
482
+ "eval_samples_per_second": 203.701,
483
+ "eval_steps_per_second": 6.372,
484
+ "step": 700
485
+ },
486
+ {
487
+ "epoch": 0.12,
488
+ "learning_rate": 4.813961729270022e-05,
489
+ "loss": 0.459,
490
+ "step": 710
491
+ },
492
+ {
493
+ "epoch": 0.12,
494
+ "learning_rate": 4.805102763997165e-05,
495
+ "loss": 0.4604,
496
+ "step": 720
497
+ },
498
+ {
499
+ "epoch": 0.12,
500
+ "learning_rate": 4.796243798724309e-05,
501
+ "loss": 0.4259,
502
+ "step": 730
503
+ },
504
+ {
505
+ "epoch": 0.12,
506
+ "learning_rate": 4.7873848334514534e-05,
507
+ "loss": 0.4564,
508
+ "step": 740
509
+ },
510
+ {
511
+ "epoch": 0.12,
512
+ "learning_rate": 4.778525868178597e-05,
513
+ "loss": 0.434,
514
+ "step": 750
515
+ },
516
+ {
517
+ "epoch": 0.12,
518
+ "learning_rate": 4.7696669029057406e-05,
519
+ "loss": 0.429,
520
+ "step": 760
521
+ },
522
+ {
523
+ "epoch": 0.13,
524
+ "learning_rate": 4.760807937632884e-05,
525
+ "loss": 0.4524,
526
+ "step": 770
527
+ },
528
+ {
529
+ "epoch": 0.13,
530
+ "learning_rate": 4.7519489723600285e-05,
531
+ "loss": 0.4491,
532
+ "step": 780
533
+ },
534
+ {
535
+ "epoch": 0.13,
536
+ "learning_rate": 4.743090007087173e-05,
537
+ "loss": 0.4377,
538
+ "step": 790
539
+ },
540
+ {
541
+ "epoch": 0.13,
542
+ "learning_rate": 4.7342310418143164e-05,
543
+ "loss": 0.4395,
544
+ "step": 800
545
+ },
546
+ {
547
+ "epoch": 0.13,
548
+ "eval_loss": 0.43905675411224365,
549
+ "eval_runtime": 138.1109,
550
+ "eval_samples_per_second": 203.699,
551
+ "eval_steps_per_second": 6.372,
552
+ "step": 800
553
+ },
554
+ {
555
+ "epoch": 0.13,
556
+ "learning_rate": 4.72537207654146e-05,
557
+ "loss": 0.4382,
558
+ "step": 810
559
+ },
560
+ {
561
+ "epoch": 0.13,
562
+ "learning_rate": 4.716513111268604e-05,
563
+ "loss": 0.4452,
564
+ "step": 820
565
+ },
566
+ {
567
+ "epoch": 0.14,
568
+ "learning_rate": 4.707654145995748e-05,
569
+ "loss": 0.4266,
570
+ "step": 830
571
+ },
572
+ {
573
+ "epoch": 0.14,
574
+ "learning_rate": 4.698795180722892e-05,
575
+ "loss": 0.4595,
576
+ "step": 840
577
+ },
578
+ {
579
+ "epoch": 0.14,
580
+ "learning_rate": 4.689936215450036e-05,
581
+ "loss": 0.4168,
582
+ "step": 850
583
+ },
584
+ {
585
+ "epoch": 0.14,
586
+ "learning_rate": 4.6810772501771794e-05,
587
+ "loss": 0.4291,
588
+ "step": 860
589
+ },
590
+ {
591
+ "epoch": 0.14,
592
+ "learning_rate": 4.6722182849043236e-05,
593
+ "loss": 0.4807,
594
+ "step": 870
595
+ },
596
+ {
597
+ "epoch": 0.14,
598
+ "learning_rate": 4.663359319631467e-05,
599
+ "loss": 0.4171,
600
+ "step": 880
601
+ },
602
+ {
603
+ "epoch": 0.14,
604
+ "learning_rate": 4.6545003543586115e-05,
605
+ "loss": 0.4447,
606
+ "step": 890
607
+ },
608
+ {
609
+ "epoch": 0.15,
610
+ "learning_rate": 4.6456413890857544e-05,
611
+ "loss": 0.4277,
612
+ "step": 900
613
+ },
614
+ {
615
+ "epoch": 0.15,
616
+ "eval_loss": 0.43294572830200195,
617
+ "eval_runtime": 138.1242,
618
+ "eval_samples_per_second": 203.679,
619
+ "eval_steps_per_second": 6.371,
620
+ "step": 900
621
+ },
622
+ {
623
+ "epoch": 0.15,
624
+ "learning_rate": 4.636782423812899e-05,
625
+ "loss": 0.4357,
626
+ "step": 910
627
+ },
628
+ {
629
+ "epoch": 0.15,
630
+ "learning_rate": 4.627923458540043e-05,
631
+ "loss": 0.4469,
632
+ "step": 920
633
+ },
634
+ {
635
+ "epoch": 0.15,
636
+ "learning_rate": 4.6190644932671866e-05,
637
+ "loss": 0.4047,
638
+ "step": 930
639
+ },
640
+ {
641
+ "epoch": 0.15,
642
+ "learning_rate": 4.610205527994331e-05,
643
+ "loss": 0.4343,
644
+ "step": 940
645
+ },
646
+ {
647
+ "epoch": 0.15,
648
+ "learning_rate": 4.601346562721474e-05,
649
+ "loss": 0.4256,
650
+ "step": 950
651
+ },
652
+ {
653
+ "epoch": 0.16,
654
+ "learning_rate": 4.592487597448618e-05,
655
+ "loss": 0.4331,
656
+ "step": 960
657
+ },
658
+ {
659
+ "epoch": 0.16,
660
+ "learning_rate": 4.5836286321757624e-05,
661
+ "loss": 0.4525,
662
+ "step": 970
663
+ },
664
+ {
665
+ "epoch": 0.16,
666
+ "learning_rate": 4.574769666902906e-05,
667
+ "loss": 0.411,
668
+ "step": 980
669
+ },
670
+ {
671
+ "epoch": 0.16,
672
+ "learning_rate": 4.56591070163005e-05,
673
+ "loss": 0.4485,
674
+ "step": 990
675
+ },
676
+ {
677
+ "epoch": 0.16,
678
+ "learning_rate": 4.557051736357194e-05,
679
+ "loss": 0.4114,
680
+ "step": 1000
681
+ },
682
+ {
683
+ "epoch": 0.16,
684
+ "eval_loss": 0.4274757206439972,
685
+ "eval_runtime": 138.0935,
686
+ "eval_samples_per_second": 203.724,
687
+ "eval_steps_per_second": 6.372,
688
+ "step": 1000
689
+ },
690
+ {
691
+ "epoch": 0.16,
692
+ "learning_rate": 4.5481927710843374e-05,
693
+ "loss": 0.4273,
694
+ "step": 1010
695
+ },
696
+ {
697
+ "epoch": 0.17,
698
+ "learning_rate": 4.539333805811482e-05,
699
+ "loss": 0.4555,
700
+ "step": 1020
701
+ },
702
+ {
703
+ "epoch": 0.17,
704
+ "learning_rate": 4.530474840538625e-05,
705
+ "loss": 0.4108,
706
+ "step": 1030
707
+ },
708
+ {
709
+ "epoch": 0.17,
710
+ "learning_rate": 4.521615875265769e-05,
711
+ "loss": 0.4002,
712
+ "step": 1040
713
+ },
714
+ {
715
+ "epoch": 0.17,
716
+ "learning_rate": 4.512756909992913e-05,
717
+ "loss": 0.4244,
718
+ "step": 1050
719
+ },
720
+ {
721
+ "epoch": 0.17,
722
+ "learning_rate": 4.503897944720057e-05,
723
+ "loss": 0.4337,
724
+ "step": 1060
725
+ },
726
+ {
727
+ "epoch": 0.17,
728
+ "learning_rate": 4.495038979447201e-05,
729
+ "loss": 0.3895,
730
+ "step": 1070
731
+ },
732
+ {
733
+ "epoch": 0.18,
734
+ "learning_rate": 4.486180014174345e-05,
735
+ "loss": 0.4,
736
+ "step": 1080
737
+ },
738
+ {
739
+ "epoch": 0.18,
740
+ "learning_rate": 4.477321048901488e-05,
741
+ "loss": 0.4197,
742
+ "step": 1090
743
+ },
744
+ {
745
+ "epoch": 0.18,
746
+ "learning_rate": 4.4684620836286326e-05,
747
+ "loss": 0.4048,
748
+ "step": 1100
749
+ },
750
+ {
751
+ "epoch": 0.18,
752
+ "eval_loss": 0.4227732717990875,
753
+ "eval_runtime": 138.0996,
754
+ "eval_samples_per_second": 203.715,
755
+ "eval_steps_per_second": 6.372,
756
+ "step": 1100
757
+ },
758
+ {
759
+ "epoch": 0.18,
760
+ "learning_rate": 4.459603118355776e-05,
761
+ "loss": 0.4341,
762
+ "step": 1110
763
+ },
764
+ {
765
+ "epoch": 0.18,
766
+ "learning_rate": 4.4507441530829204e-05,
767
+ "loss": 0.4447,
768
+ "step": 1120
769
+ },
770
+ {
771
+ "epoch": 0.18,
772
+ "learning_rate": 4.441885187810064e-05,
773
+ "loss": 0.4316,
774
+ "step": 1130
775
+ },
776
+ {
777
+ "epoch": 0.19,
778
+ "learning_rate": 4.4330262225372076e-05,
779
+ "loss": 0.4324,
780
+ "step": 1140
781
+ },
782
+ {
783
+ "epoch": 0.19,
784
+ "learning_rate": 4.424167257264352e-05,
785
+ "loss": 0.4169,
786
+ "step": 1150
787
+ },
788
+ {
789
+ "epoch": 0.19,
790
+ "learning_rate": 4.4153082919914955e-05,
791
+ "loss": 0.4061,
792
+ "step": 1160
793
+ },
794
+ {
795
+ "epoch": 0.19,
796
+ "learning_rate": 4.40644932671864e-05,
797
+ "loss": 0.3898,
798
+ "step": 1170
799
+ },
800
+ {
801
+ "epoch": 0.19,
802
+ "learning_rate": 4.3975903614457834e-05,
803
+ "loss": 0.4044,
804
+ "step": 1180
805
+ },
806
+ {
807
+ "epoch": 0.19,
808
+ "learning_rate": 4.388731396172927e-05,
809
+ "loss": 0.4002,
810
+ "step": 1190
811
+ },
812
+ {
813
+ "epoch": 0.2,
814
+ "learning_rate": 4.379872430900071e-05,
815
+ "loss": 0.4012,
816
+ "step": 1200
817
+ },
818
+ {
819
+ "epoch": 0.2,
820
+ "eval_loss": 0.41991087794303894,
821
+ "eval_runtime": 138.1063,
822
+ "eval_samples_per_second": 203.705,
823
+ "eval_steps_per_second": 6.372,
824
+ "step": 1200
825
+ },
826
+ {
827
+ "epoch": 0.2,
828
+ "learning_rate": 4.371013465627215e-05,
829
+ "loss": 0.4263,
830
+ "step": 1210
831
+ },
832
+ {
833
+ "epoch": 0.2,
834
+ "learning_rate": 4.362154500354359e-05,
835
+ "loss": 0.4256,
836
+ "step": 1220
837
+ },
838
+ {
839
+ "epoch": 0.2,
840
+ "learning_rate": 4.353295535081503e-05,
841
+ "loss": 0.4282,
842
+ "step": 1230
843
+ },
844
+ {
845
+ "epoch": 0.2,
846
+ "learning_rate": 4.3444365698086464e-05,
847
+ "loss": 0.4265,
848
+ "step": 1240
849
+ },
850
+ {
851
+ "epoch": 0.2,
852
+ "learning_rate": 4.3355776045357906e-05,
853
+ "loss": 0.4326,
854
+ "step": 1250
855
+ },
856
+ {
857
+ "epoch": 0.21,
858
+ "learning_rate": 4.326718639262934e-05,
859
+ "loss": 0.4188,
860
+ "step": 1260
861
+ },
862
+ {
863
+ "epoch": 0.21,
864
+ "learning_rate": 4.3178596739900785e-05,
865
+ "loss": 0.3904,
866
+ "step": 1270
867
+ },
868
+ {
869
+ "epoch": 0.21,
870
+ "learning_rate": 4.309000708717222e-05,
871
+ "loss": 0.3901,
872
+ "step": 1280
873
+ },
874
+ {
875
+ "epoch": 0.21,
876
+ "learning_rate": 4.300141743444366e-05,
877
+ "loss": 0.3984,
878
+ "step": 1290
879
+ },
880
+ {
881
+ "epoch": 0.21,
882
+ "learning_rate": 4.29128277817151e-05,
883
+ "loss": 0.4071,
884
+ "step": 1300
885
+ },
886
+ {
887
+ "epoch": 0.21,
888
+ "eval_loss": 0.4168910086154938,
889
+ "eval_runtime": 138.0833,
890
+ "eval_samples_per_second": 203.739,
891
+ "eval_steps_per_second": 6.373,
892
+ "step": 1300
893
+ },
894
+ {
895
+ "epoch": 0.21,
896
+ "learning_rate": 4.2824238128986536e-05,
897
+ "loss": 0.3994,
898
+ "step": 1310
899
+ },
900
+ {
901
+ "epoch": 0.21,
902
+ "learning_rate": 4.273564847625798e-05,
903
+ "loss": 0.4146,
904
+ "step": 1320
905
+ },
906
+ {
907
+ "epoch": 0.22,
908
+ "learning_rate": 4.2647058823529415e-05,
909
+ "loss": 0.4319,
910
+ "step": 1330
911
+ },
912
+ {
913
+ "epoch": 0.22,
914
+ "learning_rate": 4.255846917080085e-05,
915
+ "loss": 0.4141,
916
+ "step": 1340
917
+ },
918
+ {
919
+ "epoch": 0.22,
920
+ "learning_rate": 4.2469879518072294e-05,
921
+ "loss": 0.4355,
922
+ "step": 1350
923
+ },
924
+ {
925
+ "epoch": 0.22,
926
+ "learning_rate": 4.238128986534373e-05,
927
+ "loss": 0.4084,
928
+ "step": 1360
929
+ },
930
+ {
931
+ "epoch": 0.22,
932
+ "learning_rate": 4.2292700212615166e-05,
933
+ "loss": 0.4016,
934
+ "step": 1370
935
+ },
936
+ {
937
+ "epoch": 0.22,
938
+ "learning_rate": 4.220411055988661e-05,
939
+ "loss": 0.4154,
940
+ "step": 1380
941
+ },
942
+ {
943
+ "epoch": 0.23,
944
+ "learning_rate": 4.2115520907158044e-05,
945
+ "loss": 0.4022,
946
+ "step": 1390
947
+ },
948
+ {
949
+ "epoch": 0.23,
950
+ "learning_rate": 4.202693125442949e-05,
951
+ "loss": 0.4143,
952
+ "step": 1400
953
+ },
954
+ {
955
+ "epoch": 0.23,
956
+ "eval_loss": 0.41502994298934937,
957
+ "eval_runtime": 138.0512,
958
+ "eval_samples_per_second": 203.787,
959
+ "eval_steps_per_second": 6.374,
960
+ "step": 1400
961
+ },
962
+ {
963
+ "epoch": 0.23,
964
+ "learning_rate": 4.193834160170092e-05,
965
+ "loss": 0.4101,
966
+ "step": 1410
967
+ },
968
+ {
969
+ "epoch": 0.23,
970
+ "learning_rate": 4.184975194897236e-05,
971
+ "loss": 0.4162,
972
+ "step": 1420
973
+ },
974
+ {
975
+ "epoch": 0.23,
976
+ "learning_rate": 4.17611622962438e-05,
977
+ "loss": 0.4304,
978
+ "step": 1430
979
+ },
980
+ {
981
+ "epoch": 0.23,
982
+ "learning_rate": 4.167257264351524e-05,
983
+ "loss": 0.4037,
984
+ "step": 1440
985
+ },
986
+ {
987
+ "epoch": 0.24,
988
+ "learning_rate": 4.158398299078668e-05,
989
+ "loss": 0.3952,
990
+ "step": 1450
991
+ },
992
+ {
993
+ "epoch": 0.24,
994
+ "learning_rate": 4.149539333805812e-05,
995
+ "loss": 0.4238,
996
+ "step": 1460
997
+ },
998
+ {
999
+ "epoch": 0.24,
1000
+ "learning_rate": 4.140680368532955e-05,
1001
+ "loss": 0.4158,
1002
+ "step": 1470
1003
+ },
1004
+ {
1005
+ "epoch": 0.24,
1006
+ "learning_rate": 4.1318214032600996e-05,
1007
+ "loss": 0.4139,
1008
+ "step": 1480
1009
+ },
1010
+ {
1011
+ "epoch": 0.24,
1012
+ "learning_rate": 4.122962437987243e-05,
1013
+ "loss": 0.4069,
1014
+ "step": 1490
1015
+ },
1016
+ {
1017
+ "epoch": 0.24,
1018
+ "learning_rate": 4.1141034727143874e-05,
1019
+ "loss": 0.4028,
1020
+ "step": 1500
1021
+ },
1022
+ {
1023
+ "epoch": 0.24,
1024
+ "eval_loss": 0.4133906662464142,
1025
+ "eval_runtime": 138.1272,
1026
+ "eval_samples_per_second": 203.675,
1027
+ "eval_steps_per_second": 6.371,
1028
+ "step": 1500
1029
+ },
1030
+ {
1031
+ "epoch": 0.25,
1032
+ "learning_rate": 4.105244507441531e-05,
1033
+ "loss": 0.4127,
1034
+ "step": 1510
1035
+ },
1036
+ {
1037
+ "epoch": 0.25,
1038
+ "learning_rate": 4.0963855421686746e-05,
1039
+ "loss": 0.3997,
1040
+ "step": 1520
1041
+ },
1042
+ {
1043
+ "epoch": 0.25,
1044
+ "learning_rate": 4.087526576895819e-05,
1045
+ "loss": 0.4113,
1046
+ "step": 1530
1047
+ },
1048
+ {
1049
+ "epoch": 0.25,
1050
+ "learning_rate": 4.0786676116229625e-05,
1051
+ "loss": 0.4004,
1052
+ "step": 1540
1053
+ },
1054
+ {
1055
+ "epoch": 0.25,
1056
+ "learning_rate": 4.069808646350107e-05,
1057
+ "loss": 0.405,
1058
+ "step": 1550
1059
+ },
1060
+ {
1061
+ "epoch": 0.25,
1062
+ "learning_rate": 4.0609496810772504e-05,
1063
+ "loss": 0.3972,
1064
+ "step": 1560
1065
+ },
1066
+ {
1067
+ "epoch": 0.26,
1068
+ "learning_rate": 4.052090715804394e-05,
1069
+ "loss": 0.3986,
1070
+ "step": 1570
1071
+ },
1072
+ {
1073
+ "epoch": 0.26,
1074
+ "learning_rate": 4.043231750531538e-05,
1075
+ "loss": 0.3942,
1076
+ "step": 1580
1077
+ },
1078
+ {
1079
+ "epoch": 0.26,
1080
+ "learning_rate": 4.034372785258682e-05,
1081
+ "loss": 0.4,
1082
+ "step": 1590
1083
+ },
1084
+ {
1085
+ "epoch": 0.26,
1086
+ "learning_rate": 4.025513819985826e-05,
1087
+ "loss": 0.4257,
1088
+ "step": 1600
1089
+ },
1090
+ {
1091
+ "epoch": 0.26,
1092
+ "eval_loss": 0.41290345788002014,
1093
+ "eval_runtime": 138.0689,
1094
+ "eval_samples_per_second": 203.761,
1095
+ "eval_steps_per_second": 6.374,
1096
+ "step": 1600
1097
+ },
1098
+ {
1099
+ "epoch": 0.26,
1100
+ "learning_rate": 4.01665485471297e-05,
1101
+ "loss": 0.3852,
1102
+ "step": 1610
1103
+ },
1104
+ {
1105
+ "epoch": 0.26,
1106
+ "learning_rate": 4.0077958894401134e-05,
1107
+ "loss": 0.4232,
1108
+ "step": 1620
1109
+ },
1110
+ {
1111
+ "epoch": 0.27,
1112
+ "learning_rate": 3.9989369241672576e-05,
1113
+ "loss": 0.4041,
1114
+ "step": 1630
1115
+ },
1116
+ {
1117
+ "epoch": 0.27,
1118
+ "learning_rate": 3.990077958894401e-05,
1119
+ "loss": 0.3955,
1120
+ "step": 1640
1121
+ },
1122
+ {
1123
+ "epoch": 0.27,
1124
+ "learning_rate": 3.981218993621545e-05,
1125
+ "loss": 0.4095,
1126
+ "step": 1650
1127
+ },
1128
+ {
1129
+ "epoch": 0.27,
1130
+ "learning_rate": 3.972360028348689e-05,
1131
+ "loss": 0.3863,
1132
+ "step": 1660
1133
+ },
1134
+ {
1135
+ "epoch": 0.27,
1136
+ "learning_rate": 3.963501063075833e-05,
1137
+ "loss": 0.4058,
1138
+ "step": 1670
1139
+ },
1140
+ {
1141
+ "epoch": 0.27,
1142
+ "learning_rate": 3.954642097802977e-05,
1143
+ "loss": 0.4147,
1144
+ "step": 1680
1145
+ },
1146
+ {
1147
+ "epoch": 0.28,
1148
+ "learning_rate": 3.9457831325301206e-05,
1149
+ "loss": 0.3819,
1150
+ "step": 1690
1151
+ },
1152
+ {
1153
+ "epoch": 0.28,
1154
+ "learning_rate": 3.936924167257264e-05,
1155
+ "loss": 0.3786,
1156
+ "step": 1700
1157
+ },
1158
+ {
1159
+ "epoch": 0.28,
1160
+ "eval_loss": 0.4112655222415924,
1161
+ "eval_runtime": 138.0649,
1162
+ "eval_samples_per_second": 203.766,
1163
+ "eval_steps_per_second": 6.374,
1164
+ "step": 1700
1165
+ },
1166
+ {
1167
+ "epoch": 0.28,
1168
+ "learning_rate": 3.9280652019844085e-05,
1169
+ "loss": 0.3672,
1170
+ "step": 1710
1171
+ },
1172
+ {
1173
+ "epoch": 0.28,
1174
+ "learning_rate": 3.919206236711552e-05,
1175
+ "loss": 0.3881,
1176
+ "step": 1720
1177
+ },
1178
+ {
1179
+ "epoch": 0.28,
1180
+ "learning_rate": 3.9103472714386964e-05,
1181
+ "loss": 0.3904,
1182
+ "step": 1730
1183
+ },
1184
+ {
1185
+ "epoch": 0.28,
1186
+ "learning_rate": 3.90148830616584e-05,
1187
+ "loss": 0.4021,
1188
+ "step": 1740
1189
+ },
1190
+ {
1191
+ "epoch": 0.28,
1192
+ "learning_rate": 3.8926293408929836e-05,
1193
+ "loss": 0.385,
1194
+ "step": 1750
1195
+ },
1196
+ {
1197
+ "epoch": 0.29,
1198
+ "learning_rate": 3.883770375620128e-05,
1199
+ "loss": 0.4194,
1200
+ "step": 1760
1201
+ },
1202
+ {
1203
+ "epoch": 0.29,
1204
+ "learning_rate": 3.8749114103472714e-05,
1205
+ "loss": 0.4043,
1206
+ "step": 1770
1207
+ },
1208
+ {
1209
+ "epoch": 0.29,
1210
+ "learning_rate": 3.866052445074416e-05,
1211
+ "loss": 0.391,
1212
+ "step": 1780
1213
+ },
1214
+ {
1215
+ "epoch": 0.29,
1216
+ "learning_rate": 3.857193479801559e-05,
1217
+ "loss": 0.3976,
1218
+ "step": 1790
1219
+ },
1220
+ {
1221
+ "epoch": 0.29,
1222
+ "learning_rate": 3.848334514528703e-05,
1223
+ "loss": 0.4006,
1224
+ "step": 1800
1225
+ },
1226
+ {
1227
+ "epoch": 0.29,
1228
+ "eval_loss": 0.40891483426094055,
1229
+ "eval_runtime": 138.0756,
1230
+ "eval_samples_per_second": 203.751,
1231
+ "eval_steps_per_second": 6.373,
1232
+ "step": 1800
1233
+ },
1234
+ {
1235
+ "epoch": 0.29,
1236
+ "learning_rate": 3.839475549255847e-05,
1237
+ "loss": 0.3823,
1238
+ "step": 1810
1239
+ },
1240
+ {
1241
+ "epoch": 0.3,
1242
+ "learning_rate": 3.830616583982991e-05,
1243
+ "loss": 0.3779,
1244
+ "step": 1820
1245
+ },
1246
+ {
1247
+ "epoch": 0.3,
1248
+ "learning_rate": 3.821757618710135e-05,
1249
+ "loss": 0.447,
1250
+ "step": 1830
1251
+ },
1252
+ {
1253
+ "epoch": 0.3,
1254
+ "learning_rate": 3.812898653437279e-05,
1255
+ "loss": 0.387,
1256
+ "step": 1840
1257
+ },
1258
+ {
1259
+ "epoch": 0.3,
1260
+ "learning_rate": 3.804039688164422e-05,
1261
+ "loss": 0.3958,
1262
+ "step": 1850
1263
+ },
1264
+ {
1265
+ "epoch": 0.3,
1266
+ "learning_rate": 3.7951807228915666e-05,
1267
+ "loss": 0.3903,
1268
+ "step": 1860
1269
+ },
1270
+ {
1271
+ "epoch": 0.3,
1272
+ "learning_rate": 3.78632175761871e-05,
1273
+ "loss": 0.4026,
1274
+ "step": 1870
1275
+ },
1276
+ {
1277
+ "epoch": 0.31,
1278
+ "learning_rate": 3.7774627923458544e-05,
1279
+ "loss": 0.4113,
1280
+ "step": 1880
1281
+ },
1282
+ {
1283
+ "epoch": 0.31,
1284
+ "learning_rate": 3.768603827072998e-05,
1285
+ "loss": 0.3806,
1286
+ "step": 1890
1287
+ },
1288
+ {
1289
+ "epoch": 0.31,
1290
+ "learning_rate": 3.7597448618001416e-05,
1291
+ "loss": 0.4016,
1292
+ "step": 1900
1293
+ },
1294
+ {
1295
+ "epoch": 0.31,
1296
+ "eval_loss": 0.40891408920288086,
1297
+ "eval_runtime": 138.1603,
1298
+ "eval_samples_per_second": 203.626,
1299
+ "eval_steps_per_second": 6.369,
1300
+ "step": 1900
1301
+ },
1302
+ {
1303
+ "epoch": 0.31,
1304
+ "learning_rate": 3.750885896527286e-05,
1305
+ "loss": 0.399,
1306
+ "step": 1910
1307
+ },
1308
+ {
1309
+ "epoch": 0.31,
1310
+ "learning_rate": 3.7420269312544295e-05,
1311
+ "loss": 0.4028,
1312
+ "step": 1920
1313
+ },
1314
+ {
1315
+ "epoch": 0.31,
1316
+ "learning_rate": 3.733167965981573e-05,
1317
+ "loss": 0.4029,
1318
+ "step": 1930
1319
+ },
1320
+ {
1321
+ "epoch": 0.32,
1322
+ "learning_rate": 3.7243090007087174e-05,
1323
+ "loss": 0.4031,
1324
+ "step": 1940
1325
+ },
1326
+ {
1327
+ "epoch": 0.32,
1328
+ "learning_rate": 3.715450035435861e-05,
1329
+ "loss": 0.3945,
1330
+ "step": 1950
1331
+ },
1332
+ {
1333
+ "epoch": 0.32,
1334
+ "learning_rate": 3.706591070163005e-05,
1335
+ "loss": 0.3965,
1336
+ "step": 1960
1337
+ },
1338
+ {
1339
+ "epoch": 0.32,
1340
+ "learning_rate": 3.697732104890149e-05,
1341
+ "loss": 0.3928,
1342
+ "step": 1970
1343
+ },
1344
+ {
1345
+ "epoch": 0.32,
1346
+ "learning_rate": 3.6888731396172925e-05,
1347
+ "loss": 0.3866,
1348
+ "step": 1980
1349
+ },
1350
+ {
1351
+ "epoch": 0.32,
1352
+ "learning_rate": 3.680014174344437e-05,
1353
+ "loss": 0.3722,
1354
+ "step": 1990
1355
+ },
1356
+ {
1357
+ "epoch": 0.33,
1358
+ "learning_rate": 3.6711552090715804e-05,
1359
+ "loss": 0.3963,
1360
+ "step": 2000
1361
+ },
1362
+ {
1363
+ "epoch": 0.33,
1364
+ "eval_loss": 0.40644311904907227,
1365
+ "eval_runtime": 138.3589,
1366
+ "eval_samples_per_second": 203.334,
1367
+ "eval_steps_per_second": 6.36,
1368
+ "step": 2000
1369
+ },
1370
+ {
1371
+ "epoch": 0.33,
1372
+ "learning_rate": 3.6622962437987246e-05,
1373
+ "loss": 0.3999,
1374
+ "step": 2010
1375
+ },
1376
+ {
1377
+ "epoch": 0.33,
1378
+ "learning_rate": 3.653437278525869e-05,
1379
+ "loss": 0.3749,
1380
+ "step": 2020
1381
+ },
1382
+ {
1383
+ "epoch": 0.33,
1384
+ "learning_rate": 3.644578313253012e-05,
1385
+ "loss": 0.4157,
1386
+ "step": 2030
1387
+ },
1388
+ {
1389
+ "epoch": 0.33,
1390
+ "learning_rate": 3.635719347980156e-05,
1391
+ "loss": 0.4045,
1392
+ "step": 2040
1393
+ },
1394
+ {
1395
+ "epoch": 0.33,
1396
+ "learning_rate": 3.6268603827073e-05,
1397
+ "loss": 0.4307,
1398
+ "step": 2050
1399
+ },
1400
+ {
1401
+ "epoch": 0.34,
1402
+ "learning_rate": 3.618001417434444e-05,
1403
+ "loss": 0.392,
1404
+ "step": 2060
1405
+ },
1406
+ {
1407
+ "epoch": 0.34,
1408
+ "learning_rate": 3.609142452161588e-05,
1409
+ "loss": 0.4116,
1410
+ "step": 2070
1411
+ },
1412
+ {
1413
+ "epoch": 0.34,
1414
+ "learning_rate": 3.600283486888731e-05,
1415
+ "loss": 0.4095,
1416
+ "step": 2080
1417
+ },
1418
+ {
1419
+ "epoch": 0.34,
1420
+ "learning_rate": 3.5914245216158755e-05,
1421
+ "loss": 0.4091,
1422
+ "step": 2090
1423
+ },
1424
+ {
1425
+ "epoch": 0.34,
1426
+ "learning_rate": 3.582565556343019e-05,
1427
+ "loss": 0.3591,
1428
+ "step": 2100
1429
+ },
1430
+ {
1431
+ "epoch": 0.34,
1432
+ "eval_loss": 0.4051670730113983,
1433
+ "eval_runtime": 138.2882,
1434
+ "eval_samples_per_second": 203.437,
1435
+ "eval_steps_per_second": 6.364,
1436
+ "step": 2100
1437
+ },
1438
+ {
1439
+ "epoch": 0.34,
1440
+ "learning_rate": 3.5737065910701634e-05,
1441
+ "loss": 0.4117,
1442
+ "step": 2110
1443
+ },
1444
+ {
1445
+ "epoch": 0.35,
1446
+ "learning_rate": 3.564847625797307e-05,
1447
+ "loss": 0.387,
1448
+ "step": 2120
1449
+ },
1450
+ {
1451
+ "epoch": 0.35,
1452
+ "learning_rate": 3.5559886605244506e-05,
1453
+ "loss": 0.4078,
1454
+ "step": 2130
1455
+ },
1456
+ {
1457
+ "epoch": 0.35,
1458
+ "learning_rate": 3.547129695251595e-05,
1459
+ "loss": 0.3885,
1460
+ "step": 2140
1461
+ },
1462
+ {
1463
+ "epoch": 0.35,
1464
+ "learning_rate": 3.5382707299787384e-05,
1465
+ "loss": 0.3947,
1466
+ "step": 2150
1467
+ },
1468
+ {
1469
+ "epoch": 0.35,
1470
+ "learning_rate": 3.529411764705883e-05,
1471
+ "loss": 0.3855,
1472
+ "step": 2160
1473
+ },
1474
+ {
1475
+ "epoch": 0.35,
1476
+ "learning_rate": 3.520552799433026e-05,
1477
+ "loss": 0.4188,
1478
+ "step": 2170
1479
+ },
1480
+ {
1481
+ "epoch": 0.35,
1482
+ "learning_rate": 3.51169383416017e-05,
1483
+ "loss": 0.3786,
1484
+ "step": 2180
1485
+ },
1486
+ {
1487
+ "epoch": 0.36,
1488
+ "learning_rate": 3.502834868887314e-05,
1489
+ "loss": 0.3861,
1490
+ "step": 2190
1491
+ },
1492
+ {
1493
+ "epoch": 0.36,
1494
+ "learning_rate": 3.4939759036144585e-05,
1495
+ "loss": 0.3995,
1496
+ "step": 2200
1497
+ },
1498
+ {
1499
+ "epoch": 0.36,
1500
+ "eval_loss": 0.40360766649246216,
1501
+ "eval_runtime": 138.2299,
1502
+ "eval_samples_per_second": 203.523,
1503
+ "eval_steps_per_second": 6.366,
1504
+ "step": 2200
1505
+ },
1506
+ {
1507
+ "epoch": 0.36,
1508
+ "learning_rate": 3.485116938341602e-05,
1509
+ "loss": 0.3947,
1510
+ "step": 2210
1511
+ },
1512
+ {
1513
+ "epoch": 0.36,
1514
+ "learning_rate": 3.476257973068746e-05,
1515
+ "loss": 0.3798,
1516
+ "step": 2220
1517
+ },
1518
+ {
1519
+ "epoch": 0.36,
1520
+ "learning_rate": 3.467399007795889e-05,
1521
+ "loss": 0.3783,
1522
+ "step": 2230
1523
+ },
1524
+ {
1525
+ "epoch": 0.36,
1526
+ "learning_rate": 3.4585400425230336e-05,
1527
+ "loss": 0.3756,
1528
+ "step": 2240
1529
+ },
1530
+ {
1531
+ "epoch": 0.37,
1532
+ "learning_rate": 3.449681077250178e-05,
1533
+ "loss": 0.4065,
1534
+ "step": 2250
1535
+ },
1536
+ {
1537
+ "epoch": 0.37,
1538
+ "learning_rate": 3.440822111977321e-05,
1539
+ "loss": 0.3917,
1540
+ "step": 2260
1541
+ },
1542
+ {
1543
+ "epoch": 0.37,
1544
+ "learning_rate": 3.431963146704465e-05,
1545
+ "loss": 0.3934,
1546
+ "step": 2270
1547
+ },
1548
+ {
1549
+ "epoch": 0.37,
1550
+ "learning_rate": 3.4231041814316086e-05,
1551
+ "loss": 0.3628,
1552
+ "step": 2280
1553
+ },
1554
+ {
1555
+ "epoch": 0.37,
1556
+ "learning_rate": 3.414245216158753e-05,
1557
+ "loss": 0.3838,
1558
+ "step": 2290
1559
+ },
1560
+ {
1561
+ "epoch": 0.37,
1562
+ "learning_rate": 3.405386250885897e-05,
1563
+ "loss": 0.3773,
1564
+ "step": 2300
1565
+ },
1566
+ {
1567
+ "epoch": 0.37,
1568
+ "eval_loss": 0.40338262915611267,
1569
+ "eval_runtime": 138.1211,
1570
+ "eval_samples_per_second": 203.684,
1571
+ "eval_steps_per_second": 6.371,
1572
+ "step": 2300
1573
+ },
1574
+ {
1575
+ "epoch": 0.38,
1576
+ "learning_rate": 3.39652728561304e-05,
1577
+ "loss": 0.3693,
1578
+ "step": 2310
1579
+ },
1580
+ {
1581
+ "epoch": 0.38,
1582
+ "learning_rate": 3.3876683203401844e-05,
1583
+ "loss": 0.4011,
1584
+ "step": 2320
1585
+ },
1586
+ {
1587
+ "epoch": 0.38,
1588
+ "learning_rate": 3.378809355067328e-05,
1589
+ "loss": 0.3992,
1590
+ "step": 2330
1591
+ },
1592
+ {
1593
+ "epoch": 0.38,
1594
+ "learning_rate": 3.369950389794472e-05,
1595
+ "loss": 0.3754,
1596
+ "step": 2340
1597
+ },
1598
+ {
1599
+ "epoch": 0.38,
1600
+ "learning_rate": 3.3610914245216166e-05,
1601
+ "loss": 0.3694,
1602
+ "step": 2350
1603
+ },
1604
+ {
1605
+ "epoch": 0.38,
1606
+ "learning_rate": 3.3522324592487595e-05,
1607
+ "loss": 0.3937,
1608
+ "step": 2360
1609
+ },
1610
+ {
1611
+ "epoch": 0.39,
1612
+ "learning_rate": 3.343373493975904e-05,
1613
+ "loss": 0.377,
1614
+ "step": 2370
1615
+ },
1616
+ {
1617
+ "epoch": 0.39,
1618
+ "learning_rate": 3.3345145287030474e-05,
1619
+ "loss": 0.3939,
1620
+ "step": 2380
1621
+ },
1622
+ {
1623
+ "epoch": 0.39,
1624
+ "learning_rate": 3.3256555634301916e-05,
1625
+ "loss": 0.3998,
1626
+ "step": 2390
1627
+ },
1628
+ {
1629
+ "epoch": 0.39,
1630
+ "learning_rate": 3.316796598157335e-05,
1631
+ "loss": 0.3758,
1632
+ "step": 2400
1633
+ },
1634
+ {
1635
+ "epoch": 0.39,
1636
+ "eval_loss": 0.40262076258659363,
1637
+ "eval_runtime": 138.1252,
1638
+ "eval_samples_per_second": 203.678,
1639
+ "eval_steps_per_second": 6.371,
1640
+ "step": 2400
1641
+ },
1642
+ {
1643
+ "epoch": 0.39,
1644
+ "learning_rate": 3.307937632884479e-05,
1645
+ "loss": 0.3703,
1646
+ "step": 2410
1647
+ },
1648
+ {
1649
+ "epoch": 0.39,
1650
+ "learning_rate": 3.299078667611623e-05,
1651
+ "loss": 0.391,
1652
+ "step": 2420
1653
+ },
1654
+ {
1655
+ "epoch": 0.4,
1656
+ "learning_rate": 3.2902197023387674e-05,
1657
+ "loss": 0.3804,
1658
+ "step": 2430
1659
+ },
1660
+ {
1661
+ "epoch": 0.4,
1662
+ "learning_rate": 3.281360737065911e-05,
1663
+ "loss": 0.3909,
1664
+ "step": 2440
1665
+ },
1666
+ {
1667
+ "epoch": 0.4,
1668
+ "learning_rate": 3.2725017717930546e-05,
1669
+ "loss": 0.3904,
1670
+ "step": 2450
1671
+ },
1672
+ {
1673
+ "epoch": 0.4,
1674
+ "learning_rate": 3.263642806520198e-05,
1675
+ "loss": 0.3892,
1676
+ "step": 2460
1677
+ },
1678
+ {
1679
+ "epoch": 0.4,
1680
+ "learning_rate": 3.2547838412473425e-05,
1681
+ "loss": 0.3925,
1682
+ "step": 2470
1683
+ },
1684
+ {
1685
+ "epoch": 0.4,
1686
+ "learning_rate": 3.245924875974487e-05,
1687
+ "loss": 0.3827,
1688
+ "step": 2480
1689
+ },
1690
+ {
1691
+ "epoch": 0.41,
1692
+ "learning_rate": 3.2370659107016304e-05,
1693
+ "loss": 0.3816,
1694
+ "step": 2490
1695
+ },
1696
+ {
1697
+ "epoch": 0.41,
1698
+ "learning_rate": 3.228206945428774e-05,
1699
+ "loss": 0.3751,
1700
+ "step": 2500
1701
+ },
1702
+ {
1703
+ "epoch": 0.41,
1704
+ "eval_loss": 0.4022020995616913,
1705
+ "eval_runtime": 138.0981,
1706
+ "eval_samples_per_second": 203.718,
1707
+ "eval_steps_per_second": 6.372,
1708
+ "step": 2500
1709
+ },
1710
+ {
1711
+ "epoch": 0.41,
1712
+ "learning_rate": 3.2193479801559176e-05,
1713
+ "loss": 0.3725,
1714
+ "step": 2510
1715
+ },
1716
+ {
1717
+ "epoch": 0.41,
1718
+ "learning_rate": 3.210489014883062e-05,
1719
+ "loss": 0.3864,
1720
+ "step": 2520
1721
+ },
1722
+ {
1723
+ "epoch": 0.41,
1724
+ "learning_rate": 3.201630049610206e-05,
1725
+ "loss": 0.3553,
1726
+ "step": 2530
1727
+ },
1728
+ {
1729
+ "epoch": 0.41,
1730
+ "learning_rate": 3.192771084337349e-05,
1731
+ "loss": 0.357,
1732
+ "step": 2540
1733
+ },
1734
+ {
1735
+ "epoch": 0.42,
1736
+ "learning_rate": 3.183912119064493e-05,
1737
+ "loss": 0.4031,
1738
+ "step": 2550
1739
+ },
1740
+ {
1741
+ "epoch": 0.42,
1742
+ "learning_rate": 3.175053153791637e-05,
1743
+ "loss": 0.3691,
1744
+ "step": 2560
1745
+ },
1746
+ {
1747
+ "epoch": 0.42,
1748
+ "learning_rate": 3.166194188518781e-05,
1749
+ "loss": 0.4035,
1750
+ "step": 2570
1751
+ },
1752
+ {
1753
+ "epoch": 0.42,
1754
+ "learning_rate": 3.1573352232459255e-05,
1755
+ "loss": 0.3978,
1756
+ "step": 2580
1757
+ },
1758
+ {
1759
+ "epoch": 0.42,
1760
+ "learning_rate": 3.1484762579730684e-05,
1761
+ "loss": 0.3844,
1762
+ "step": 2590
1763
+ },
1764
+ {
1765
+ "epoch": 0.42,
1766
+ "learning_rate": 3.139617292700213e-05,
1767
+ "loss": 0.3833,
1768
+ "step": 2600
1769
+ },
1770
+ {
1771
+ "epoch": 0.42,
1772
+ "eval_loss": 0.39937448501586914,
1773
+ "eval_runtime": 138.8709,
1774
+ "eval_samples_per_second": 202.584,
1775
+ "eval_steps_per_second": 6.337,
1776
+ "step": 2600
1777
+ },
1778
+ {
1779
+ "epoch": 0.42,
1780
+ "learning_rate": 3.130758327427357e-05,
1781
+ "loss": 0.3883,
1782
+ "step": 2610
1783
+ },
1784
+ {
1785
+ "epoch": 0.43,
1786
+ "learning_rate": 3.1218993621545006e-05,
1787
+ "loss": 0.3797,
1788
+ "step": 2620
1789
+ },
1790
+ {
1791
+ "epoch": 0.43,
1792
+ "learning_rate": 3.113040396881645e-05,
1793
+ "loss": 0.3992,
1794
+ "step": 2630
1795
+ },
1796
+ {
1797
+ "epoch": 0.43,
1798
+ "learning_rate": 3.104181431608788e-05,
1799
+ "loss": 0.3818,
1800
+ "step": 2640
1801
+ },
1802
+ {
1803
+ "epoch": 0.43,
1804
+ "learning_rate": 3.095322466335932e-05,
1805
+ "loss": 0.3872,
1806
+ "step": 2650
1807
+ },
1808
+ {
1809
+ "epoch": 0.43,
1810
+ "learning_rate": 3.086463501063076e-05,
1811
+ "loss": 0.4047,
1812
+ "step": 2660
1813
+ },
1814
+ {
1815
+ "epoch": 0.43,
1816
+ "learning_rate": 3.07760453579022e-05,
1817
+ "loss": 0.4149,
1818
+ "step": 2670
1819
+ },
1820
+ {
1821
+ "epoch": 0.44,
1822
+ "learning_rate": 3.0687455705173635e-05,
1823
+ "loss": 0.3741,
1824
+ "step": 2680
1825
+ },
1826
+ {
1827
+ "epoch": 0.44,
1828
+ "learning_rate": 3.059886605244507e-05,
1829
+ "loss": 0.3898,
1830
+ "step": 2690
1831
+ },
1832
+ {
1833
+ "epoch": 0.44,
1834
+ "learning_rate": 3.0510276399716514e-05,
1835
+ "loss": 0.381,
1836
+ "step": 2700
1837
+ },
1838
+ {
1839
+ "epoch": 0.44,
1840
+ "eval_loss": 0.39839911460876465,
1841
+ "eval_runtime": 138.3419,
1842
+ "eval_samples_per_second": 203.359,
1843
+ "eval_steps_per_second": 6.361,
1844
+ "step": 2700
1845
+ },
1846
+ {
1847
+ "epoch": 0.44,
1848
+ "learning_rate": 3.0421686746987953e-05,
1849
+ "loss": 0.3634,
1850
+ "step": 2710
1851
+ },
1852
+ {
1853
+ "epoch": 0.44,
1854
+ "learning_rate": 3.0333097094259393e-05,
1855
+ "loss": 0.3783,
1856
+ "step": 2720
1857
+ },
1858
+ {
1859
+ "epoch": 0.44,
1860
+ "learning_rate": 3.024450744153083e-05,
1861
+ "loss": 0.3854,
1862
+ "step": 2730
1863
+ },
1864
+ {
1865
+ "epoch": 0.45,
1866
+ "learning_rate": 3.0155917788802268e-05,
1867
+ "loss": 0.3931,
1868
+ "step": 2740
1869
+ },
1870
+ {
1871
+ "epoch": 0.45,
1872
+ "learning_rate": 3.0067328136073708e-05,
1873
+ "loss": 0.371,
1874
+ "step": 2750
1875
+ },
1876
+ {
1877
+ "epoch": 0.45,
1878
+ "learning_rate": 2.9978738483345147e-05,
1879
+ "loss": 0.4089,
1880
+ "step": 2760
1881
+ },
1882
+ {
1883
+ "epoch": 0.45,
1884
+ "learning_rate": 2.9890148830616586e-05,
1885
+ "loss": 0.3854,
1886
+ "step": 2770
1887
+ },
1888
+ {
1889
+ "epoch": 0.45,
1890
+ "learning_rate": 2.9801559177888022e-05,
1891
+ "loss": 0.3921,
1892
+ "step": 2780
1893
+ },
1894
+ {
1895
+ "epoch": 0.45,
1896
+ "learning_rate": 2.9712969525159462e-05,
1897
+ "loss": 0.3753,
1898
+ "step": 2790
1899
+ },
1900
+ {
1901
+ "epoch": 0.46,
1902
+ "learning_rate": 2.96243798724309e-05,
1903
+ "loss": 0.377,
1904
+ "step": 2800
1905
+ },
1906
+ {
1907
+ "epoch": 0.46,
1908
+ "eval_loss": 0.3989295959472656,
1909
+ "eval_runtime": 138.1703,
1910
+ "eval_samples_per_second": 203.611,
1911
+ "eval_steps_per_second": 6.369,
1912
+ "step": 2800
1913
+ },
1914
+ {
1915
+ "epoch": 0.46,
1916
+ "learning_rate": 2.953579021970234e-05,
1917
+ "loss": 0.4242,
1918
+ "step": 2810
1919
+ },
1920
+ {
1921
+ "epoch": 0.46,
1922
+ "learning_rate": 2.944720056697378e-05,
1923
+ "loss": 0.3857,
1924
+ "step": 2820
1925
+ },
1926
+ {
1927
+ "epoch": 0.46,
1928
+ "learning_rate": 2.9358610914245216e-05,
1929
+ "loss": 0.3678,
1930
+ "step": 2830
1931
+ },
1932
+ {
1933
+ "epoch": 0.46,
1934
+ "learning_rate": 2.9270021261516655e-05,
1935
+ "loss": 0.3798,
1936
+ "step": 2840
1937
+ },
1938
+ {
1939
+ "epoch": 0.46,
1940
+ "learning_rate": 2.9181431608788095e-05,
1941
+ "loss": 0.3798,
1942
+ "step": 2850
1943
+ },
1944
+ {
1945
+ "epoch": 0.47,
1946
+ "learning_rate": 2.9092841956059534e-05,
1947
+ "loss": 0.3932,
1948
+ "step": 2860
1949
+ },
1950
+ {
1951
+ "epoch": 0.47,
1952
+ "learning_rate": 2.900425230333097e-05,
1953
+ "loss": 0.3635,
1954
+ "step": 2870
1955
+ },
1956
+ {
1957
+ "epoch": 0.47,
1958
+ "learning_rate": 2.891566265060241e-05,
1959
+ "loss": 0.3933,
1960
+ "step": 2880
1961
+ },
1962
+ {
1963
+ "epoch": 0.47,
1964
+ "learning_rate": 2.882707299787385e-05,
1965
+ "loss": 0.3749,
1966
+ "step": 2890
1967
+ },
1968
+ {
1969
+ "epoch": 0.47,
1970
+ "learning_rate": 2.873848334514529e-05,
1971
+ "loss": 0.3974,
1972
+ "step": 2900
1973
+ },
1974
+ {
1975
+ "epoch": 0.47,
1976
+ "eval_loss": 0.3973732888698578,
1977
+ "eval_runtime": 138.1232,
1978
+ "eval_samples_per_second": 203.681,
1979
+ "eval_steps_per_second": 6.371,
1980
+ "step": 2900
1981
+ },
1982
+ {
1983
+ "epoch": 0.47,
1984
+ "learning_rate": 2.8649893692416728e-05,
1985
+ "loss": 0.3573,
1986
+ "step": 2910
1987
+ },
1988
+ {
1989
+ "epoch": 0.48,
1990
+ "learning_rate": 2.8561304039688164e-05,
1991
+ "loss": 0.3608,
1992
+ "step": 2920
1993
+ },
1994
+ {
1995
+ "epoch": 0.48,
1996
+ "learning_rate": 2.8472714386959603e-05,
1997
+ "loss": 0.373,
1998
+ "step": 2930
1999
+ },
2000
+ {
2001
+ "epoch": 0.48,
2002
+ "learning_rate": 2.8384124734231043e-05,
2003
+ "loss": 0.357,
2004
+ "step": 2940
2005
+ },
2006
+ {
2007
+ "epoch": 0.48,
2008
+ "learning_rate": 2.8295535081502482e-05,
2009
+ "loss": 0.3894,
2010
+ "step": 2950
2011
+ },
2012
+ {
2013
+ "epoch": 0.48,
2014
+ "learning_rate": 2.8206945428773925e-05,
2015
+ "loss": 0.3674,
2016
+ "step": 2960
2017
+ },
2018
+ {
2019
+ "epoch": 0.48,
2020
+ "learning_rate": 2.8118355776045357e-05,
2021
+ "loss": 0.3855,
2022
+ "step": 2970
2023
+ },
2024
+ {
2025
+ "epoch": 0.49,
2026
+ "learning_rate": 2.8029766123316797e-05,
2027
+ "loss": 0.3838,
2028
+ "step": 2980
2029
+ },
2030
+ {
2031
+ "epoch": 0.49,
2032
+ "learning_rate": 2.7941176470588236e-05,
2033
+ "loss": 0.3675,
2034
+ "step": 2990
2035
+ },
2036
+ {
2037
+ "epoch": 0.49,
2038
+ "learning_rate": 2.7852586817859676e-05,
2039
+ "loss": 0.3687,
2040
+ "step": 3000
2041
+ },
2042
+ {
2043
+ "epoch": 0.49,
2044
+ "eval_loss": 0.3975074291229248,
2045
+ "eval_runtime": 138.0969,
2046
+ "eval_samples_per_second": 203.719,
2047
+ "eval_steps_per_second": 6.372,
2048
+ "step": 3000
2049
+ },
2050
+ {
2051
+ "epoch": 0.49,
2052
+ "learning_rate": 2.776399716513111e-05,
2053
+ "loss": 0.3982,
2054
+ "step": 3010
2055
+ },
2056
+ {
2057
+ "epoch": 0.49,
2058
+ "learning_rate": 2.767540751240255e-05,
2059
+ "loss": 0.3792,
2060
+ "step": 3020
2061
+ },
2062
+ {
2063
+ "epoch": 0.49,
2064
+ "learning_rate": 2.758681785967399e-05,
2065
+ "loss": 0.3771,
2066
+ "step": 3030
2067
+ },
2068
+ {
2069
+ "epoch": 0.49,
2070
+ "learning_rate": 2.749822820694543e-05,
2071
+ "loss": 0.3795,
2072
+ "step": 3040
2073
+ },
2074
+ {
2075
+ "epoch": 0.5,
2076
+ "learning_rate": 2.7409638554216873e-05,
2077
+ "loss": 0.3865,
2078
+ "step": 3050
2079
+ },
2080
+ {
2081
+ "epoch": 0.5,
2082
+ "learning_rate": 2.7321048901488305e-05,
2083
+ "loss": 0.3795,
2084
+ "step": 3060
2085
+ },
2086
+ {
2087
+ "epoch": 0.5,
2088
+ "learning_rate": 2.7232459248759745e-05,
2089
+ "loss": 0.3694,
2090
+ "step": 3070
2091
+ },
2092
+ {
2093
+ "epoch": 0.5,
2094
+ "learning_rate": 2.7143869596031184e-05,
2095
+ "loss": 0.3661,
2096
+ "step": 3080
2097
+ },
2098
+ {
2099
+ "epoch": 0.5,
2100
+ "learning_rate": 2.7055279943302623e-05,
2101
+ "loss": 0.3678,
2102
+ "step": 3090
2103
+ },
2104
+ {
2105
+ "epoch": 0.5,
2106
+ "learning_rate": 2.6966690290574066e-05,
2107
+ "loss": 0.3817,
2108
+ "step": 3100
2109
+ },
2110
+ {
2111
+ "epoch": 0.5,
2112
+ "eval_loss": 0.39725953340530396,
2113
+ "eval_runtime": 138.119,
2114
+ "eval_samples_per_second": 203.687,
2115
+ "eval_steps_per_second": 6.371,
2116
+ "step": 3100
2117
+ },
2118
+ {
2119
+ "epoch": 0.51,
2120
+ "learning_rate": 2.68781006378455e-05,
2121
+ "loss": 0.3815,
2122
+ "step": 3110
2123
+ },
2124
+ {
2125
+ "epoch": 0.51,
2126
+ "learning_rate": 2.6789510985116938e-05,
2127
+ "loss": 0.3784,
2128
+ "step": 3120
2129
+ },
2130
+ {
2131
+ "epoch": 0.51,
2132
+ "learning_rate": 2.6700921332388378e-05,
2133
+ "loss": 0.3681,
2134
+ "step": 3130
2135
+ },
2136
+ {
2137
+ "epoch": 0.51,
2138
+ "learning_rate": 2.661233167965982e-05,
2139
+ "loss": 0.4035,
2140
+ "step": 3140
2141
+ },
2142
+ {
2143
+ "epoch": 0.51,
2144
+ "learning_rate": 2.6523742026931253e-05,
2145
+ "loss": 0.3744,
2146
+ "step": 3150
2147
+ },
2148
+ {
2149
+ "epoch": 0.51,
2150
+ "learning_rate": 2.6435152374202692e-05,
2151
+ "loss": 0.3783,
2152
+ "step": 3160
2153
+ },
2154
+ {
2155
+ "epoch": 0.52,
2156
+ "learning_rate": 2.6346562721474132e-05,
2157
+ "loss": 0.404,
2158
+ "step": 3170
2159
+ },
2160
+ {
2161
+ "epoch": 0.52,
2162
+ "learning_rate": 2.625797306874557e-05,
2163
+ "loss": 0.3678,
2164
+ "step": 3180
2165
+ },
2166
+ {
2167
+ "epoch": 0.52,
2168
+ "learning_rate": 2.6169383416017014e-05,
2169
+ "loss": 0.3958,
2170
+ "step": 3190
2171
+ },
2172
+ {
2173
+ "epoch": 0.52,
2174
+ "learning_rate": 2.6080793763288447e-05,
2175
+ "loss": 0.3771,
2176
+ "step": 3200
2177
+ },
2178
+ {
2179
+ "epoch": 0.52,
2180
+ "eval_loss": 0.39562681317329407,
2181
+ "eval_runtime": 138.1817,
2182
+ "eval_samples_per_second": 203.594,
2183
+ "eval_steps_per_second": 6.368,
2184
+ "step": 3200
2185
+ },
2186
+ {
2187
+ "epoch": 0.52,
2188
+ "learning_rate": 2.5992204110559886e-05,
2189
+ "loss": 0.3917,
2190
+ "step": 3210
2191
+ },
2192
+ {
2193
+ "epoch": 0.52,
2194
+ "learning_rate": 2.5903614457831325e-05,
2195
+ "loss": 0.3648,
2196
+ "step": 3220
2197
+ },
2198
+ {
2199
+ "epoch": 0.53,
2200
+ "learning_rate": 2.5815024805102768e-05,
2201
+ "loss": 0.3944,
2202
+ "step": 3230
2203
+ },
2204
+ {
2205
+ "epoch": 0.53,
2206
+ "learning_rate": 2.5726435152374208e-05,
2207
+ "loss": 0.3715,
2208
+ "step": 3240
2209
+ },
2210
+ {
2211
+ "epoch": 0.53,
2212
+ "learning_rate": 2.563784549964564e-05,
2213
+ "loss": 0.3784,
2214
+ "step": 3250
2215
+ },
2216
+ {
2217
+ "epoch": 0.53,
2218
+ "learning_rate": 2.554925584691708e-05,
2219
+ "loss": 0.3666,
2220
+ "step": 3260
2221
+ },
2222
+ {
2223
+ "epoch": 0.53,
2224
+ "learning_rate": 2.546066619418852e-05,
2225
+ "loss": 0.3876,
2226
+ "step": 3270
2227
+ },
2228
+ {
2229
+ "epoch": 0.53,
2230
+ "learning_rate": 2.5372076541459962e-05,
2231
+ "loss": 0.395,
2232
+ "step": 3280
2233
+ },
2234
+ {
2235
+ "epoch": 0.54,
2236
+ "learning_rate": 2.5283486888731394e-05,
2237
+ "loss": 0.3936,
2238
+ "step": 3290
2239
+ },
2240
+ {
2241
+ "epoch": 0.54,
2242
+ "learning_rate": 2.5194897236002834e-05,
2243
+ "loss": 0.3756,
2244
+ "step": 3300
2245
+ },
2246
+ {
2247
+ "epoch": 0.54,
2248
+ "eval_loss": 0.39499565958976746,
2249
+ "eval_runtime": 138.1729,
2250
+ "eval_samples_per_second": 203.607,
2251
+ "eval_steps_per_second": 6.369,
2252
+ "step": 3300
2253
+ },
2254
+ {
2255
+ "epoch": 0.54,
2256
+ "learning_rate": 2.5106307583274273e-05,
2257
+ "loss": 0.3641,
2258
+ "step": 3310
2259
+ },
2260
+ {
2261
+ "epoch": 0.54,
2262
+ "learning_rate": 2.5017717930545713e-05,
2263
+ "loss": 0.3718,
2264
+ "step": 3320
2265
+ },
2266
+ {
2267
+ "epoch": 0.54,
2268
+ "learning_rate": 2.4929128277817152e-05,
2269
+ "loss": 0.3719,
2270
+ "step": 3330
2271
+ },
2272
+ {
2273
+ "epoch": 0.54,
2274
+ "learning_rate": 2.484053862508859e-05,
2275
+ "loss": 0.3883,
2276
+ "step": 3340
2277
+ },
2278
+ {
2279
+ "epoch": 0.55,
2280
+ "learning_rate": 2.4751948972360027e-05,
2281
+ "loss": 0.3869,
2282
+ "step": 3350
2283
+ },
2284
+ {
2285
+ "epoch": 0.55,
2286
+ "learning_rate": 2.4663359319631467e-05,
2287
+ "loss": 0.3566,
2288
+ "step": 3360
2289
+ },
2290
+ {
2291
+ "epoch": 0.55,
2292
+ "learning_rate": 2.457476966690291e-05,
2293
+ "loss": 0.3876,
2294
+ "step": 3370
2295
+ },
2296
+ {
2297
+ "epoch": 0.55,
2298
+ "learning_rate": 2.4486180014174346e-05,
2299
+ "loss": 0.3719,
2300
+ "step": 3380
2301
+ },
2302
+ {
2303
+ "epoch": 0.55,
2304
+ "learning_rate": 2.4397590361445785e-05,
2305
+ "loss": 0.3655,
2306
+ "step": 3390
2307
+ },
2308
+ {
2309
+ "epoch": 0.55,
2310
+ "learning_rate": 2.430900070871722e-05,
2311
+ "loss": 0.3809,
2312
+ "step": 3400
2313
+ },
2314
+ {
2315
+ "epoch": 0.55,
2316
+ "eval_loss": 0.39441797137260437,
2317
+ "eval_runtime": 138.0968,
2318
+ "eval_samples_per_second": 203.719,
2319
+ "eval_steps_per_second": 6.372,
2320
+ "step": 3400
2321
+ },
2322
+ {
2323
+ "epoch": 0.56,
2324
+ "learning_rate": 2.422041105598866e-05,
2325
+ "loss": 0.3613,
2326
+ "step": 3410
2327
+ },
2328
+ {
2329
+ "epoch": 0.56,
2330
+ "learning_rate": 2.41318214032601e-05,
2331
+ "loss": 0.389,
2332
+ "step": 3420
2333
+ },
2334
+ {
2335
+ "epoch": 0.56,
2336
+ "learning_rate": 2.404323175053154e-05,
2337
+ "loss": 0.3595,
2338
+ "step": 3430
2339
+ },
2340
+ {
2341
+ "epoch": 0.56,
2342
+ "learning_rate": 2.395464209780298e-05,
2343
+ "loss": 0.3718,
2344
+ "step": 3440
2345
+ },
2346
+ {
2347
+ "epoch": 0.56,
2348
+ "learning_rate": 2.3866052445074415e-05,
2349
+ "loss": 0.3757,
2350
+ "step": 3450
2351
+ },
2352
+ {
2353
+ "epoch": 0.56,
2354
+ "learning_rate": 2.3777462792345857e-05,
2355
+ "loss": 0.3724,
2356
+ "step": 3460
2357
+ },
2358
+ {
2359
+ "epoch": 0.56,
2360
+ "learning_rate": 2.3688873139617293e-05,
2361
+ "loss": 0.3796,
2362
+ "step": 3470
2363
+ },
2364
+ {
2365
+ "epoch": 0.57,
2366
+ "learning_rate": 2.3600283486888733e-05,
2367
+ "loss": 0.3801,
2368
+ "step": 3480
2369
+ },
2370
+ {
2371
+ "epoch": 0.57,
2372
+ "learning_rate": 2.351169383416017e-05,
2373
+ "loss": 0.3712,
2374
+ "step": 3490
2375
+ },
2376
+ {
2377
+ "epoch": 0.57,
2378
+ "learning_rate": 2.3423104181431608e-05,
2379
+ "loss": 0.3941,
2380
+ "step": 3500
2381
+ },
2382
+ {
2383
+ "epoch": 0.57,
2384
+ "eval_loss": 0.3938775360584259,
2385
+ "eval_runtime": 138.1358,
2386
+ "eval_samples_per_second": 203.662,
2387
+ "eval_steps_per_second": 6.371,
2388
+ "step": 3500
2389
+ },
2390
+ {
2391
+ "epoch": 0.57,
2392
+ "learning_rate": 2.333451452870305e-05,
2393
+ "loss": 0.3727,
2394
+ "step": 3510
2395
+ },
2396
+ {
2397
+ "epoch": 0.57,
2398
+ "learning_rate": 2.3245924875974487e-05,
2399
+ "loss": 0.3784,
2400
+ "step": 3520
2401
+ },
2402
+ {
2403
+ "epoch": 0.57,
2404
+ "learning_rate": 2.3157335223245926e-05,
2405
+ "loss": 0.379,
2406
+ "step": 3530
2407
+ },
2408
+ {
2409
+ "epoch": 0.58,
2410
+ "learning_rate": 2.3068745570517362e-05,
2411
+ "loss": 0.3657,
2412
+ "step": 3540
2413
+ },
2414
+ {
2415
+ "epoch": 0.58,
2416
+ "learning_rate": 2.2980155917788805e-05,
2417
+ "loss": 0.3863,
2418
+ "step": 3550
2419
+ },
2420
+ {
2421
+ "epoch": 0.58,
2422
+ "learning_rate": 2.289156626506024e-05,
2423
+ "loss": 0.3848,
2424
+ "step": 3560
2425
+ },
2426
+ {
2427
+ "epoch": 0.58,
2428
+ "learning_rate": 2.280297661233168e-05,
2429
+ "loss": 0.3764,
2430
+ "step": 3570
2431
+ },
2432
+ {
2433
+ "epoch": 0.58,
2434
+ "learning_rate": 2.271438695960312e-05,
2435
+ "loss": 0.3739,
2436
+ "step": 3580
2437
+ },
2438
+ {
2439
+ "epoch": 0.58,
2440
+ "learning_rate": 2.2625797306874556e-05,
2441
+ "loss": 0.3836,
2442
+ "step": 3590
2443
+ },
2444
+ {
2445
+ "epoch": 0.59,
2446
+ "learning_rate": 2.2537207654146e-05,
2447
+ "loss": 0.3751,
2448
+ "step": 3600
2449
+ },
2450
+ {
2451
+ "epoch": 0.59,
2452
+ "eval_loss": 0.3936668634414673,
2453
+ "eval_runtime": 138.1758,
2454
+ "eval_samples_per_second": 203.603,
2455
+ "eval_steps_per_second": 6.369,
2456
+ "step": 3600
2457
+ },
2458
+ {
2459
+ "epoch": 0.59,
2460
+ "learning_rate": 2.2448618001417435e-05,
2461
+ "loss": 0.3827,
2462
+ "step": 3610
2463
+ },
2464
+ {
2465
+ "epoch": 0.59,
2466
+ "learning_rate": 2.2360028348688874e-05,
2467
+ "loss": 0.3943,
2468
+ "step": 3620
2469
+ },
2470
+ {
2471
+ "epoch": 0.59,
2472
+ "learning_rate": 2.227143869596031e-05,
2473
+ "loss": 0.3678,
2474
+ "step": 3630
2475
+ },
2476
+ {
2477
+ "epoch": 0.59,
2478
+ "learning_rate": 2.2182849043231753e-05,
2479
+ "loss": 0.3613,
2480
+ "step": 3640
2481
+ },
2482
+ {
2483
+ "epoch": 0.59,
2484
+ "learning_rate": 2.2094259390503192e-05,
2485
+ "loss": 0.3802,
2486
+ "step": 3650
2487
+ },
2488
+ {
2489
+ "epoch": 0.6,
2490
+ "learning_rate": 2.200566973777463e-05,
2491
+ "loss": 0.3937,
2492
+ "step": 3660
2493
+ },
2494
+ {
2495
+ "epoch": 0.6,
2496
+ "learning_rate": 2.1917080085046068e-05,
2497
+ "loss": 0.3737,
2498
+ "step": 3670
2499
+ },
2500
+ {
2501
+ "epoch": 0.6,
2502
+ "learning_rate": 2.1828490432317504e-05,
2503
+ "loss": 0.3755,
2504
+ "step": 3680
2505
+ },
2506
+ {
2507
+ "epoch": 0.6,
2508
+ "learning_rate": 2.1739900779588947e-05,
2509
+ "loss": 0.3602,
2510
+ "step": 3690
2511
+ },
2512
+ {
2513
+ "epoch": 0.6,
2514
+ "learning_rate": 2.1651311126860383e-05,
2515
+ "loss": 0.3771,
2516
+ "step": 3700
2517
+ },
2518
+ {
2519
+ "epoch": 0.6,
2520
+ "eval_loss": 0.39319226145744324,
2521
+ "eval_runtime": 138.1244,
2522
+ "eval_samples_per_second": 203.679,
2523
+ "eval_steps_per_second": 6.371,
2524
+ "step": 3700
2525
+ },
2526
+ {
2527
+ "epoch": 0.6,
2528
+ "learning_rate": 2.1562721474131822e-05,
2529
+ "loss": 0.3838,
2530
+ "step": 3710
2531
+ },
2532
+ {
2533
+ "epoch": 0.61,
2534
+ "learning_rate": 2.147413182140326e-05,
2535
+ "loss": 0.3674,
2536
+ "step": 3720
2537
+ },
2538
+ {
2539
+ "epoch": 0.61,
2540
+ "learning_rate": 2.13855421686747e-05,
2541
+ "loss": 0.3547,
2542
+ "step": 3730
2543
+ },
2544
+ {
2545
+ "epoch": 0.61,
2546
+ "learning_rate": 2.129695251594614e-05,
2547
+ "loss": 0.3734,
2548
+ "step": 3740
2549
+ },
2550
+ {
2551
+ "epoch": 0.61,
2552
+ "learning_rate": 2.1208362863217576e-05,
2553
+ "loss": 0.3758,
2554
+ "step": 3750
2555
+ },
2556
+ {
2557
+ "epoch": 0.61,
2558
+ "learning_rate": 2.1119773210489016e-05,
2559
+ "loss": 0.3705,
2560
+ "step": 3760
2561
+ },
2562
+ {
2563
+ "epoch": 0.61,
2564
+ "learning_rate": 2.103118355776045e-05,
2565
+ "loss": 0.3827,
2566
+ "step": 3770
2567
+ },
2568
+ {
2569
+ "epoch": 0.62,
2570
+ "learning_rate": 2.0942593905031894e-05,
2571
+ "loss": 0.375,
2572
+ "step": 3780
2573
+ },
2574
+ {
2575
+ "epoch": 0.62,
2576
+ "learning_rate": 2.0854004252303334e-05,
2577
+ "loss": 0.3836,
2578
+ "step": 3790
2579
+ },
2580
+ {
2581
+ "epoch": 0.62,
2582
+ "learning_rate": 2.076541459957477e-05,
2583
+ "loss": 0.3624,
2584
+ "step": 3800
2585
+ },
2586
+ {
2587
+ "epoch": 0.62,
2588
+ "eval_loss": 0.3926844894886017,
2589
+ "eval_runtime": 138.1212,
2590
+ "eval_samples_per_second": 203.683,
2591
+ "eval_steps_per_second": 6.371,
2592
+ "step": 3800
2593
+ },
2594
+ {
2595
+ "epoch": 0.62,
2596
+ "learning_rate": 2.067682494684621e-05,
2597
+ "loss": 0.394,
2598
+ "step": 3810
2599
+ },
2600
+ {
2601
+ "epoch": 0.62,
2602
+ "learning_rate": 2.058823529411765e-05,
2603
+ "loss": 0.3529,
2604
+ "step": 3820
2605
+ },
2606
+ {
2607
+ "epoch": 0.62,
2608
+ "learning_rate": 2.0499645641389088e-05,
2609
+ "loss": 0.3796,
2610
+ "step": 3830
2611
+ },
2612
+ {
2613
+ "epoch": 0.62,
2614
+ "learning_rate": 2.0411055988660524e-05,
2615
+ "loss": 0.3557,
2616
+ "step": 3840
2617
+ },
2618
+ {
2619
+ "epoch": 0.63,
2620
+ "learning_rate": 2.0322466335931963e-05,
2621
+ "loss": 0.3607,
2622
+ "step": 3850
2623
+ },
2624
+ {
2625
+ "epoch": 0.63,
2626
+ "learning_rate": 2.0233876683203403e-05,
2627
+ "loss": 0.3765,
2628
+ "step": 3860
2629
+ },
2630
+ {
2631
+ "epoch": 0.63,
2632
+ "learning_rate": 2.0145287030474842e-05,
2633
+ "loss": 0.3682,
2634
+ "step": 3870
2635
+ },
2636
+ {
2637
+ "epoch": 0.63,
2638
+ "learning_rate": 2.005669737774628e-05,
2639
+ "loss": 0.3771,
2640
+ "step": 3880
2641
+ },
2642
+ {
2643
+ "epoch": 0.63,
2644
+ "learning_rate": 1.9968107725017718e-05,
2645
+ "loss": 0.3793,
2646
+ "step": 3890
2647
+ },
2648
+ {
2649
+ "epoch": 0.63,
2650
+ "learning_rate": 1.9879518072289157e-05,
2651
+ "loss": 0.3504,
2652
+ "step": 3900
2653
+ },
2654
+ {
2655
+ "epoch": 0.63,
2656
+ "eval_loss": 0.3924194574356079,
2657
+ "eval_runtime": 138.059,
2658
+ "eval_samples_per_second": 203.775,
2659
+ "eval_steps_per_second": 6.374,
2660
+ "step": 3900
2661
+ },
2662
+ {
2663
+ "epoch": 0.64,
2664
+ "learning_rate": 1.9790928419560596e-05,
2665
+ "loss": 0.3469,
2666
+ "step": 3910
2667
+ },
2668
+ {
2669
+ "epoch": 0.64,
2670
+ "learning_rate": 1.9702338766832036e-05,
2671
+ "loss": 0.3801,
2672
+ "step": 3920
2673
+ },
2674
+ {
2675
+ "epoch": 0.64,
2676
+ "learning_rate": 1.9613749114103475e-05,
2677
+ "loss": 0.3841,
2678
+ "step": 3930
2679
+ },
2680
+ {
2681
+ "epoch": 0.64,
2682
+ "learning_rate": 1.952515946137491e-05,
2683
+ "loss": 0.3513,
2684
+ "step": 3940
2685
+ },
2686
+ {
2687
+ "epoch": 0.64,
2688
+ "learning_rate": 1.943656980864635e-05,
2689
+ "loss": 0.3553,
2690
+ "step": 3950
2691
+ },
2692
+ {
2693
+ "epoch": 0.64,
2694
+ "learning_rate": 1.934798015591779e-05,
2695
+ "loss": 0.3862,
2696
+ "step": 3960
2697
+ },
2698
+ {
2699
+ "epoch": 0.65,
2700
+ "learning_rate": 1.925939050318923e-05,
2701
+ "loss": 0.3854,
2702
+ "step": 3970
2703
+ },
2704
+ {
2705
+ "epoch": 0.65,
2706
+ "learning_rate": 1.917080085046067e-05,
2707
+ "loss": 0.3646,
2708
+ "step": 3980
2709
+ },
2710
+ {
2711
+ "epoch": 0.65,
2712
+ "learning_rate": 1.9082211197732105e-05,
2713
+ "loss": 0.3793,
2714
+ "step": 3990
2715
+ },
2716
+ {
2717
+ "epoch": 0.65,
2718
+ "learning_rate": 1.8993621545003544e-05,
2719
+ "loss": 0.3738,
2720
+ "step": 4000
2721
+ },
2722
+ {
2723
+ "epoch": 0.65,
2724
+ "eval_loss": 0.39233776926994324,
2725
+ "eval_runtime": 138.3177,
2726
+ "eval_samples_per_second": 203.394,
2727
+ "eval_steps_per_second": 6.362,
2728
+ "step": 4000
2729
+ },
2730
+ {
2731
+ "epoch": 0.65,
2732
+ "learning_rate": 1.8905031892274984e-05,
2733
+ "loss": 0.3729,
2734
+ "step": 4010
2735
+ },
2736
+ {
2737
+ "epoch": 0.65,
2738
+ "learning_rate": 1.8816442239546423e-05,
2739
+ "loss": 0.3629,
2740
+ "step": 4020
2741
+ },
2742
+ {
2743
+ "epoch": 0.66,
2744
+ "learning_rate": 1.872785258681786e-05,
2745
+ "loss": 0.3743,
2746
+ "step": 4030
2747
+ },
2748
+ {
2749
+ "epoch": 0.66,
2750
+ "learning_rate": 1.86392629340893e-05,
2751
+ "loss": 0.3722,
2752
+ "step": 4040
2753
+ },
2754
+ {
2755
+ "epoch": 0.66,
2756
+ "learning_rate": 1.8550673281360738e-05,
2757
+ "loss": 0.3612,
2758
+ "step": 4050
2759
+ },
2760
+ {
2761
+ "epoch": 0.66,
2762
+ "learning_rate": 1.8462083628632177e-05,
2763
+ "loss": 0.3705,
2764
+ "step": 4060
2765
+ },
2766
+ {
2767
+ "epoch": 0.66,
2768
+ "learning_rate": 1.8373493975903617e-05,
2769
+ "loss": 0.3557,
2770
+ "step": 4070
2771
+ },
2772
+ {
2773
+ "epoch": 0.66,
2774
+ "learning_rate": 1.8284904323175053e-05,
2775
+ "loss": 0.3723,
2776
+ "step": 4080
2777
+ },
2778
+ {
2779
+ "epoch": 0.67,
2780
+ "learning_rate": 1.8196314670446492e-05,
2781
+ "loss": 0.3726,
2782
+ "step": 4090
2783
+ },
2784
+ {
2785
+ "epoch": 0.67,
2786
+ "learning_rate": 1.810772501771793e-05,
2787
+ "loss": 0.3642,
2788
+ "step": 4100
2789
+ },
2790
+ {
2791
+ "epoch": 0.67,
2792
+ "eval_loss": 0.39176103472709656,
2793
+ "eval_runtime": 139.016,
2794
+ "eval_samples_per_second": 202.372,
2795
+ "eval_steps_per_second": 6.33,
2796
+ "step": 4100
2797
+ },
2798
+ {
2799
+ "epoch": 0.67,
2800
+ "learning_rate": 1.801913536498937e-05,
2801
+ "loss": 0.3592,
2802
+ "step": 4110
2803
+ },
2804
+ {
2805
+ "epoch": 0.67,
2806
+ "learning_rate": 1.793054571226081e-05,
2807
+ "loss": 0.3506,
2808
+ "step": 4120
2809
+ },
2810
+ {
2811
+ "epoch": 0.67,
2812
+ "learning_rate": 1.7841956059532246e-05,
2813
+ "loss": 0.3834,
2814
+ "step": 4130
2815
+ },
2816
+ {
2817
+ "epoch": 0.67,
2818
+ "learning_rate": 1.7753366406803686e-05,
2819
+ "loss": 0.355,
2820
+ "step": 4140
2821
+ },
2822
+ {
2823
+ "epoch": 0.68,
2824
+ "learning_rate": 1.7664776754075125e-05,
2825
+ "loss": 0.3745,
2826
+ "step": 4150
2827
+ },
2828
+ {
2829
+ "epoch": 0.68,
2830
+ "learning_rate": 1.7576187101346564e-05,
2831
+ "loss": 0.383,
2832
+ "step": 4160
2833
+ },
2834
+ {
2835
+ "epoch": 0.68,
2836
+ "learning_rate": 1.7487597448618e-05,
2837
+ "loss": 0.3733,
2838
+ "step": 4170
2839
+ },
2840
+ {
2841
+ "epoch": 0.68,
2842
+ "learning_rate": 1.739900779588944e-05,
2843
+ "loss": 0.3686,
2844
+ "step": 4180
2845
+ },
2846
+ {
2847
+ "epoch": 0.68,
2848
+ "learning_rate": 1.731041814316088e-05,
2849
+ "loss": 0.3766,
2850
+ "step": 4190
2851
+ },
2852
+ {
2853
+ "epoch": 0.68,
2854
+ "learning_rate": 1.722182849043232e-05,
2855
+ "loss": 0.3681,
2856
+ "step": 4200
2857
+ },
2858
+ {
2859
+ "epoch": 0.68,
2860
+ "eval_loss": 0.39138132333755493,
2861
+ "eval_runtime": 138.2113,
2862
+ "eval_samples_per_second": 203.551,
2863
+ "eval_steps_per_second": 6.367,
2864
+ "step": 4200
2865
+ },
2866
+ {
2867
+ "epoch": 0.69,
2868
+ "learning_rate": 1.7133238837703758e-05,
2869
+ "loss": 0.3727,
2870
+ "step": 4210
2871
+ },
2872
+ {
2873
+ "epoch": 0.69,
2874
+ "learning_rate": 1.7044649184975194e-05,
2875
+ "loss": 0.3811,
2876
+ "step": 4220
2877
+ },
2878
+ {
2879
+ "epoch": 0.69,
2880
+ "learning_rate": 1.6956059532246633e-05,
2881
+ "loss": 0.3719,
2882
+ "step": 4230
2883
+ },
2884
+ {
2885
+ "epoch": 0.69,
2886
+ "learning_rate": 1.6867469879518073e-05,
2887
+ "loss": 0.3538,
2888
+ "step": 4240
2889
+ },
2890
+ {
2891
+ "epoch": 0.69,
2892
+ "learning_rate": 1.6778880226789512e-05,
2893
+ "loss": 0.3778,
2894
+ "step": 4250
2895
+ },
2896
+ {
2897
+ "epoch": 0.69,
2898
+ "learning_rate": 1.669029057406095e-05,
2899
+ "loss": 0.3711,
2900
+ "step": 4260
2901
+ },
2902
+ {
2903
+ "epoch": 0.69,
2904
+ "learning_rate": 1.6601700921332388e-05,
2905
+ "loss": 0.3785,
2906
+ "step": 4270
2907
+ },
2908
+ {
2909
+ "epoch": 0.7,
2910
+ "learning_rate": 1.6513111268603827e-05,
2911
+ "loss": 0.3657,
2912
+ "step": 4280
2913
+ },
2914
+ {
2915
+ "epoch": 0.7,
2916
+ "learning_rate": 1.6424521615875266e-05,
2917
+ "loss": 0.3825,
2918
+ "step": 4290
2919
+ },
2920
+ {
2921
+ "epoch": 0.7,
2922
+ "learning_rate": 1.6335931963146706e-05,
2923
+ "loss": 0.3602,
2924
+ "step": 4300
2925
+ },
2926
+ {
2927
+ "epoch": 0.7,
2928
+ "eval_loss": 0.3910753130912781,
2929
+ "eval_runtime": 138.112,
2930
+ "eval_samples_per_second": 203.697,
2931
+ "eval_steps_per_second": 6.372,
2932
+ "step": 4300
2933
+ },
2934
+ {
2935
+ "epoch": 0.7,
2936
+ "learning_rate": 1.6247342310418142e-05,
2937
+ "loss": 0.3474,
2938
+ "step": 4310
2939
+ },
2940
+ {
2941
+ "epoch": 0.7,
2942
+ "learning_rate": 1.615875265768958e-05,
2943
+ "loss": 0.3464,
2944
+ "step": 4320
2945
+ },
2946
+ {
2947
+ "epoch": 0.7,
2948
+ "learning_rate": 1.6070163004961024e-05,
2949
+ "loss": 0.3618,
2950
+ "step": 4330
2951
+ },
2952
+ {
2953
+ "epoch": 0.71,
2954
+ "learning_rate": 1.598157335223246e-05,
2955
+ "loss": 0.4017,
2956
+ "step": 4340
2957
+ },
2958
+ {
2959
+ "epoch": 0.71,
2960
+ "learning_rate": 1.58929836995039e-05,
2961
+ "loss": 0.3742,
2962
+ "step": 4350
2963
+ },
2964
+ {
2965
+ "epoch": 0.71,
2966
+ "learning_rate": 1.5804394046775335e-05,
2967
+ "loss": 0.3679,
2968
+ "step": 4360
2969
+ },
2970
+ {
2971
+ "epoch": 0.71,
2972
+ "learning_rate": 1.5715804394046775e-05,
2973
+ "loss": 0.3797,
2974
+ "step": 4370
2975
+ },
2976
+ {
2977
+ "epoch": 0.71,
2978
+ "learning_rate": 1.5627214741318214e-05,
2979
+ "loss": 0.3547,
2980
+ "step": 4380
2981
+ },
2982
+ {
2983
+ "epoch": 0.71,
2984
+ "learning_rate": 1.5538625088589654e-05,
2985
+ "loss": 0.383,
2986
+ "step": 4390
2987
+ },
2988
+ {
2989
+ "epoch": 0.72,
2990
+ "learning_rate": 1.5450035435861093e-05,
2991
+ "loss": 0.3667,
2992
+ "step": 4400
2993
+ },
2994
+ {
2995
+ "epoch": 0.72,
2996
+ "eval_loss": 0.39077791571617126,
2997
+ "eval_runtime": 138.1368,
2998
+ "eval_samples_per_second": 203.66,
2999
+ "eval_steps_per_second": 6.37,
3000
+ "step": 4400
3001
+ },
3002
+ {
3003
+ "epoch": 0.72,
3004
+ "learning_rate": 1.536144578313253e-05,
3005
+ "loss": 0.3816,
3006
+ "step": 4410
3007
+ },
3008
+ {
3009
+ "epoch": 0.72,
3010
+ "learning_rate": 1.5272856130403972e-05,
3011
+ "loss": 0.3754,
3012
+ "step": 4420
3013
+ },
3014
+ {
3015
+ "epoch": 0.72,
3016
+ "learning_rate": 1.5184266477675408e-05,
3017
+ "loss": 0.3622,
3018
+ "step": 4430
3019
+ },
3020
+ {
3021
+ "epoch": 0.72,
3022
+ "learning_rate": 1.5095676824946847e-05,
3023
+ "loss": 0.3861,
3024
+ "step": 4440
3025
+ },
3026
+ {
3027
+ "epoch": 0.72,
3028
+ "learning_rate": 1.5007087172218285e-05,
3029
+ "loss": 0.3876,
3030
+ "step": 4450
3031
+ },
3032
+ {
3033
+ "epoch": 0.73,
3034
+ "learning_rate": 1.4918497519489724e-05,
3035
+ "loss": 0.3609,
3036
+ "step": 4460
3037
+ },
3038
+ {
3039
+ "epoch": 0.73,
3040
+ "learning_rate": 1.4829907866761164e-05,
3041
+ "loss": 0.3666,
3042
+ "step": 4470
3043
+ },
3044
+ {
3045
+ "epoch": 0.73,
3046
+ "learning_rate": 1.4741318214032601e-05,
3047
+ "loss": 0.3468,
3048
+ "step": 4480
3049
+ },
3050
+ {
3051
+ "epoch": 0.73,
3052
+ "learning_rate": 1.4652728561304041e-05,
3053
+ "loss": 0.3719,
3054
+ "step": 4490
3055
+ },
3056
+ {
3057
+ "epoch": 0.73,
3058
+ "learning_rate": 1.4564138908575479e-05,
3059
+ "loss": 0.3828,
3060
+ "step": 4500
3061
+ },
3062
+ {
3063
+ "epoch": 0.73,
3064
+ "eval_loss": 0.39073827862739563,
3065
+ "eval_runtime": 138.1539,
3066
+ "eval_samples_per_second": 203.635,
3067
+ "eval_steps_per_second": 6.37,
3068
+ "step": 4500
3069
+ },
3070
+ {
3071
+ "epoch": 0.73,
3072
+ "learning_rate": 1.4475549255846918e-05,
3073
+ "loss": 0.3625,
3074
+ "step": 4510
3075
+ },
3076
+ {
3077
+ "epoch": 0.74,
3078
+ "learning_rate": 1.4386959603118356e-05,
3079
+ "loss": 0.3575,
3080
+ "step": 4520
3081
+ },
3082
+ {
3083
+ "epoch": 0.74,
3084
+ "learning_rate": 1.4298369950389795e-05,
3085
+ "loss": 0.3582,
3086
+ "step": 4530
3087
+ },
3088
+ {
3089
+ "epoch": 0.74,
3090
+ "learning_rate": 1.4209780297661234e-05,
3091
+ "loss": 0.384,
3092
+ "step": 4540
3093
+ },
3094
+ {
3095
+ "epoch": 0.74,
3096
+ "learning_rate": 1.4121190644932672e-05,
3097
+ "loss": 0.3652,
3098
+ "step": 4550
3099
+ },
3100
+ {
3101
+ "epoch": 0.74,
3102
+ "learning_rate": 1.4032600992204112e-05,
3103
+ "loss": 0.3643,
3104
+ "step": 4560
3105
+ },
3106
+ {
3107
+ "epoch": 0.74,
3108
+ "learning_rate": 1.394401133947555e-05,
3109
+ "loss": 0.3648,
3110
+ "step": 4570
3111
+ },
3112
+ {
3113
+ "epoch": 0.75,
3114
+ "learning_rate": 1.3855421686746989e-05,
3115
+ "loss": 0.3733,
3116
+ "step": 4580
3117
+ },
3118
+ {
3119
+ "epoch": 0.75,
3120
+ "learning_rate": 1.3766832034018428e-05,
3121
+ "loss": 0.3619,
3122
+ "step": 4590
3123
+ },
3124
+ {
3125
+ "epoch": 0.75,
3126
+ "learning_rate": 1.3678242381289866e-05,
3127
+ "loss": 0.3862,
3128
+ "step": 4600
3129
+ },
3130
+ {
3131
+ "epoch": 0.75,
3132
+ "eval_loss": 0.38958612084388733,
3133
+ "eval_runtime": 138.6074,
3134
+ "eval_samples_per_second": 202.969,
3135
+ "eval_steps_per_second": 6.349,
3136
+ "step": 4600
3137
+ },
3138
+ {
3139
+ "epoch": 0.75,
3140
+ "learning_rate": 1.3589652728561305e-05,
3141
+ "loss": 0.3763,
3142
+ "step": 4610
3143
+ },
3144
+ {
3145
+ "epoch": 0.75,
3146
+ "learning_rate": 1.3501063075832743e-05,
3147
+ "loss": 0.3628,
3148
+ "step": 4620
3149
+ },
3150
+ {
3151
+ "epoch": 0.75,
3152
+ "learning_rate": 1.3412473423104182e-05,
3153
+ "loss": 0.3716,
3154
+ "step": 4630
3155
+ },
3156
+ {
3157
+ "epoch": 0.76,
3158
+ "learning_rate": 1.332388377037562e-05,
3159
+ "loss": 0.3409,
3160
+ "step": 4640
3161
+ },
3162
+ {
3163
+ "epoch": 0.76,
3164
+ "learning_rate": 1.323529411764706e-05,
3165
+ "loss": 0.3749,
3166
+ "step": 4650
3167
+ },
3168
+ {
3169
+ "epoch": 0.76,
3170
+ "learning_rate": 1.3146704464918499e-05,
3171
+ "loss": 0.3732,
3172
+ "step": 4660
3173
+ },
3174
+ {
3175
+ "epoch": 0.76,
3176
+ "learning_rate": 1.3058114812189936e-05,
3177
+ "loss": 0.3566,
3178
+ "step": 4670
3179
+ },
3180
+ {
3181
+ "epoch": 0.76,
3182
+ "learning_rate": 1.2969525159461376e-05,
3183
+ "loss": 0.3507,
3184
+ "step": 4680
3185
+ },
3186
+ {
3187
+ "epoch": 0.76,
3188
+ "learning_rate": 1.2880935506732814e-05,
3189
+ "loss": 0.3546,
3190
+ "step": 4690
3191
+ },
3192
+ {
3193
+ "epoch": 0.76,
3194
+ "learning_rate": 1.2792345854004253e-05,
3195
+ "loss": 0.3929,
3196
+ "step": 4700
3197
+ },
3198
+ {
3199
+ "epoch": 0.76,
3200
+ "eval_loss": 0.3902750611305237,
3201
+ "eval_runtime": 138.1169,
3202
+ "eval_samples_per_second": 203.69,
3203
+ "eval_steps_per_second": 6.371,
3204
+ "step": 4700
3205
+ },
3206
+ {
3207
+ "epoch": 0.77,
3208
+ "learning_rate": 1.270375620127569e-05,
3209
+ "loss": 0.3738,
3210
+ "step": 4710
3211
+ },
3212
+ {
3213
+ "epoch": 0.77,
3214
+ "learning_rate": 1.261516654854713e-05,
3215
+ "loss": 0.379,
3216
+ "step": 4720
3217
+ },
3218
+ {
3219
+ "epoch": 0.77,
3220
+ "learning_rate": 1.252657689581857e-05,
3221
+ "loss": 0.3727,
3222
+ "step": 4730
3223
+ },
3224
+ {
3225
+ "epoch": 0.77,
3226
+ "learning_rate": 1.2437987243090007e-05,
3227
+ "loss": 0.3525,
3228
+ "step": 4740
3229
+ },
3230
+ {
3231
+ "epoch": 0.77,
3232
+ "learning_rate": 1.2349397590361447e-05,
3233
+ "loss": 0.3796,
3234
+ "step": 4750
3235
+ },
3236
+ {
3237
+ "epoch": 0.77,
3238
+ "learning_rate": 1.2260807937632884e-05,
3239
+ "loss": 0.3811,
3240
+ "step": 4760
3241
+ },
3242
+ {
3243
+ "epoch": 0.78,
3244
+ "learning_rate": 1.2172218284904324e-05,
3245
+ "loss": 0.3569,
3246
+ "step": 4770
3247
+ },
3248
+ {
3249
+ "epoch": 0.78,
3250
+ "learning_rate": 1.2083628632175763e-05,
3251
+ "loss": 0.3749,
3252
+ "step": 4780
3253
+ },
3254
+ {
3255
+ "epoch": 0.78,
3256
+ "learning_rate": 1.19950389794472e-05,
3257
+ "loss": 0.3537,
3258
+ "step": 4790
3259
+ },
3260
+ {
3261
+ "epoch": 0.78,
3262
+ "learning_rate": 1.190644932671864e-05,
3263
+ "loss": 0.3749,
3264
+ "step": 4800
3265
+ },
3266
+ {
3267
+ "epoch": 0.78,
3268
+ "eval_loss": 0.3902783989906311,
3269
+ "eval_runtime": 138.1622,
3270
+ "eval_samples_per_second": 203.623,
3271
+ "eval_steps_per_second": 6.369,
3272
+ "step": 4800
3273
+ },
3274
+ {
3275
+ "epoch": 0.78,
3276
+ "learning_rate": 1.1817859673990078e-05,
3277
+ "loss": 0.3788,
3278
+ "step": 4810
3279
+ },
3280
+ {
3281
+ "epoch": 0.78,
3282
+ "learning_rate": 1.1729270021261517e-05,
3283
+ "loss": 0.3719,
3284
+ "step": 4820
3285
+ },
3286
+ {
3287
+ "epoch": 0.79,
3288
+ "learning_rate": 1.1640680368532957e-05,
3289
+ "loss": 0.3612,
3290
+ "step": 4830
3291
+ },
3292
+ {
3293
+ "epoch": 0.79,
3294
+ "learning_rate": 1.1552090715804394e-05,
3295
+ "loss": 0.3716,
3296
+ "step": 4840
3297
+ },
3298
+ {
3299
+ "epoch": 0.79,
3300
+ "learning_rate": 1.1463501063075834e-05,
3301
+ "loss": 0.3788,
3302
+ "step": 4850
3303
+ },
3304
+ {
3305
+ "epoch": 0.79,
3306
+ "learning_rate": 1.1374911410347271e-05,
3307
+ "loss": 0.3742,
3308
+ "step": 4860
3309
+ },
3310
+ {
3311
+ "epoch": 0.79,
3312
+ "learning_rate": 1.1286321757618711e-05,
3313
+ "loss": 0.3616,
3314
+ "step": 4870
3315
+ },
3316
+ {
3317
+ "epoch": 0.79,
3318
+ "learning_rate": 1.1197732104890149e-05,
3319
+ "loss": 0.3746,
3320
+ "step": 4880
3321
+ },
3322
+ {
3323
+ "epoch": 0.8,
3324
+ "learning_rate": 1.1109142452161588e-05,
3325
+ "loss": 0.3922,
3326
+ "step": 4890
3327
+ },
3328
+ {
3329
+ "epoch": 0.8,
3330
+ "learning_rate": 1.1020552799433027e-05,
3331
+ "loss": 0.37,
3332
+ "step": 4900
3333
+ },
3334
+ {
3335
+ "epoch": 0.8,
3336
+ "eval_loss": 0.3897284269332886,
3337
+ "eval_runtime": 138.1234,
3338
+ "eval_samples_per_second": 203.68,
3339
+ "eval_steps_per_second": 6.371,
3340
+ "step": 4900
3341
+ },
3342
+ {
3343
+ "epoch": 0.8,
3344
+ "learning_rate": 1.0931963146704465e-05,
3345
+ "loss": 0.3741,
3346
+ "step": 4910
3347
+ },
3348
+ {
3349
+ "epoch": 0.8,
3350
+ "learning_rate": 1.0843373493975904e-05,
3351
+ "loss": 0.3902,
3352
+ "step": 4920
3353
+ },
3354
+ {
3355
+ "epoch": 0.8,
3356
+ "learning_rate": 1.0754783841247342e-05,
3357
+ "loss": 0.3426,
3358
+ "step": 4930
3359
+ },
3360
+ {
3361
+ "epoch": 0.8,
3362
+ "learning_rate": 1.0666194188518782e-05,
3363
+ "loss": 0.3825,
3364
+ "step": 4940
3365
+ },
3366
+ {
3367
+ "epoch": 0.81,
3368
+ "learning_rate": 1.057760453579022e-05,
3369
+ "loss": 0.3594,
3370
+ "step": 4950
3371
+ },
3372
+ {
3373
+ "epoch": 0.81,
3374
+ "learning_rate": 1.0489014883061659e-05,
3375
+ "loss": 0.3404,
3376
+ "step": 4960
3377
+ },
3378
+ {
3379
+ "epoch": 0.81,
3380
+ "learning_rate": 1.0400425230333098e-05,
3381
+ "loss": 0.3725,
3382
+ "step": 4970
3383
+ },
3384
+ {
3385
+ "epoch": 0.81,
3386
+ "learning_rate": 1.0311835577604537e-05,
3387
+ "loss": 0.3662,
3388
+ "step": 4980
3389
+ },
3390
+ {
3391
+ "epoch": 0.81,
3392
+ "learning_rate": 1.0223245924875975e-05,
3393
+ "loss": 0.3688,
3394
+ "step": 4990
3395
+ },
3396
+ {
3397
+ "epoch": 0.81,
3398
+ "learning_rate": 1.0134656272147413e-05,
3399
+ "loss": 0.374,
3400
+ "step": 5000
3401
+ },
3402
+ {
3403
+ "epoch": 0.81,
3404
+ "eval_loss": 0.389631450176239,
3405
+ "eval_runtime": 138.1297,
3406
+ "eval_samples_per_second": 203.671,
3407
+ "eval_steps_per_second": 6.371,
3408
+ "step": 5000
3409
+ },
3410
+ {
3411
+ "epoch": 0.82,
3412
+ "learning_rate": 1.0046066619418852e-05,
3413
+ "loss": 0.3866,
3414
+ "step": 5010
3415
+ },
3416
+ {
3417
+ "epoch": 0.82,
3418
+ "learning_rate": 9.95747696669029e-06,
3419
+ "loss": 0.3423,
3420
+ "step": 5020
3421
+ },
3422
+ {
3423
+ "epoch": 0.82,
3424
+ "learning_rate": 9.86888731396173e-06,
3425
+ "loss": 0.3497,
3426
+ "step": 5030
3427
+ },
3428
+ {
3429
+ "epoch": 0.82,
3430
+ "learning_rate": 9.780297661233169e-06,
3431
+ "loss": 0.3625,
3432
+ "step": 5040
3433
+ },
3434
+ {
3435
+ "epoch": 0.82,
3436
+ "learning_rate": 9.691708008504608e-06,
3437
+ "loss": 0.3752,
3438
+ "step": 5050
3439
+ },
3440
+ {
3441
+ "epoch": 0.82,
3442
+ "learning_rate": 9.603118355776046e-06,
3443
+ "loss": 0.3732,
3444
+ "step": 5060
3445
+ },
3446
+ {
3447
+ "epoch": 0.83,
3448
+ "learning_rate": 9.514528703047485e-06,
3449
+ "loss": 0.3546,
3450
+ "step": 5070
3451
+ },
3452
+ {
3453
+ "epoch": 0.83,
3454
+ "learning_rate": 9.425939050318923e-06,
3455
+ "loss": 0.366,
3456
+ "step": 5080
3457
+ },
3458
+ {
3459
+ "epoch": 0.83,
3460
+ "learning_rate": 9.33734939759036e-06,
3461
+ "loss": 0.3713,
3462
+ "step": 5090
3463
+ },
3464
+ {
3465
+ "epoch": 0.83,
3466
+ "learning_rate": 9.2487597448618e-06,
3467
+ "loss": 0.3495,
3468
+ "step": 5100
3469
+ },
3470
+ {
3471
+ "epoch": 0.83,
3472
+ "eval_loss": 0.38959911465644836,
3473
+ "eval_runtime": 139.3255,
3474
+ "eval_samples_per_second": 201.923,
3475
+ "eval_steps_per_second": 6.316,
3476
+ "step": 5100
3477
+ },
3478
+ {
3479
+ "epoch": 0.83,
3480
+ "learning_rate": 9.16017009213324e-06,
3481
+ "loss": 0.3578,
3482
+ "step": 5110
3483
+ },
3484
+ {
3485
+ "epoch": 0.83,
3486
+ "learning_rate": 9.071580439404679e-06,
3487
+ "loss": 0.351,
3488
+ "step": 5120
3489
+ },
3490
+ {
3491
+ "epoch": 0.83,
3492
+ "learning_rate": 8.982990786676117e-06,
3493
+ "loss": 0.4097,
3494
+ "step": 5130
3495
+ },
3496
+ {
3497
+ "epoch": 0.84,
3498
+ "learning_rate": 8.894401133947556e-06,
3499
+ "loss": 0.3707,
3500
+ "step": 5140
3501
+ },
3502
+ {
3503
+ "epoch": 0.84,
3504
+ "learning_rate": 8.805811481218994e-06,
3505
+ "loss": 0.3492,
3506
+ "step": 5150
3507
+ },
3508
+ {
3509
+ "epoch": 0.84,
3510
+ "learning_rate": 8.717221828490433e-06,
3511
+ "loss": 0.3668,
3512
+ "step": 5160
3513
+ },
3514
+ {
3515
+ "epoch": 0.84,
3516
+ "learning_rate": 8.628632175761872e-06,
3517
+ "loss": 0.3609,
3518
+ "step": 5170
3519
+ },
3520
+ {
3521
+ "epoch": 0.84,
3522
+ "learning_rate": 8.54004252303331e-06,
3523
+ "loss": 0.3528,
3524
+ "step": 5180
3525
+ },
3526
+ {
3527
+ "epoch": 0.84,
3528
+ "learning_rate": 8.45145287030475e-06,
3529
+ "loss": 0.3693,
3530
+ "step": 5190
3531
+ },
3532
+ {
3533
+ "epoch": 0.85,
3534
+ "learning_rate": 8.362863217576187e-06,
3535
+ "loss": 0.3768,
3536
+ "step": 5200
3537
+ },
3538
+ {
3539
+ "epoch": 0.85,
3540
+ "eval_loss": 0.3894375264644623,
3541
+ "eval_runtime": 138.7537,
3542
+ "eval_samples_per_second": 202.755,
3543
+ "eval_steps_per_second": 6.342,
3544
+ "step": 5200
3545
+ },
3546
+ {
3547
+ "epoch": 0.85,
3548
+ "learning_rate": 8.274273564847627e-06,
3549
+ "loss": 0.3619,
3550
+ "step": 5210
3551
+ },
3552
+ {
3553
+ "epoch": 0.85,
3554
+ "learning_rate": 8.185683912119064e-06,
3555
+ "loss": 0.3742,
3556
+ "step": 5220
3557
+ },
3558
+ {
3559
+ "epoch": 0.85,
3560
+ "learning_rate": 8.097094259390504e-06,
3561
+ "loss": 0.3461,
3562
+ "step": 5230
3563
+ },
3564
+ {
3565
+ "epoch": 0.85,
3566
+ "learning_rate": 8.008504606661943e-06,
3567
+ "loss": 0.3743,
3568
+ "step": 5240
3569
+ },
3570
+ {
3571
+ "epoch": 0.85,
3572
+ "learning_rate": 7.919914953933381e-06,
3573
+ "loss": 0.3434,
3574
+ "step": 5250
3575
+ },
3576
+ {
3577
+ "epoch": 0.86,
3578
+ "learning_rate": 7.83132530120482e-06,
3579
+ "loss": 0.3784,
3580
+ "step": 5260
3581
+ },
3582
+ {
3583
+ "epoch": 0.86,
3584
+ "learning_rate": 7.742735648476258e-06,
3585
+ "loss": 0.3671,
3586
+ "step": 5270
3587
+ },
3588
+ {
3589
+ "epoch": 0.86,
3590
+ "learning_rate": 7.654145995747697e-06,
3591
+ "loss": 0.367,
3592
+ "step": 5280
3593
+ },
3594
+ {
3595
+ "epoch": 0.86,
3596
+ "learning_rate": 7.565556343019135e-06,
3597
+ "loss": 0.3714,
3598
+ "step": 5290
3599
+ },
3600
+ {
3601
+ "epoch": 0.86,
3602
+ "learning_rate": 7.476966690290574e-06,
3603
+ "loss": 0.348,
3604
+ "step": 5300
3605
+ },
3606
+ {
3607
+ "epoch": 0.86,
3608
+ "eval_loss": 0.38949331641197205,
3609
+ "eval_runtime": 138.1986,
3610
+ "eval_samples_per_second": 203.569,
3611
+ "eval_steps_per_second": 6.368,
3612
+ "step": 5300
3613
+ },
3614
+ {
3615
+ "epoch": 0.86,
3616
+ "learning_rate": 7.388377037562014e-06,
3617
+ "loss": 0.3743,
3618
+ "step": 5310
3619
+ },
3620
+ {
3621
+ "epoch": 0.87,
3622
+ "learning_rate": 7.299787384833452e-06,
3623
+ "loss": 0.3569,
3624
+ "step": 5320
3625
+ },
3626
+ {
3627
+ "epoch": 0.87,
3628
+ "learning_rate": 7.211197732104891e-06,
3629
+ "loss": 0.3434,
3630
+ "step": 5330
3631
+ },
3632
+ {
3633
+ "epoch": 0.87,
3634
+ "learning_rate": 7.1226080793763295e-06,
3635
+ "loss": 0.3783,
3636
+ "step": 5340
3637
+ },
3638
+ {
3639
+ "epoch": 0.87,
3640
+ "learning_rate": 7.034018426647768e-06,
3641
+ "loss": 0.3399,
3642
+ "step": 5350
3643
+ },
3644
+ {
3645
+ "epoch": 0.87,
3646
+ "learning_rate": 6.945428773919206e-06,
3647
+ "loss": 0.3433,
3648
+ "step": 5360
3649
+ },
3650
+ {
3651
+ "epoch": 0.87,
3652
+ "learning_rate": 6.856839121190644e-06,
3653
+ "loss": 0.3667,
3654
+ "step": 5370
3655
+ },
3656
+ {
3657
+ "epoch": 0.88,
3658
+ "learning_rate": 6.7682494684620846e-06,
3659
+ "loss": 0.3848,
3660
+ "step": 5380
3661
+ },
3662
+ {
3663
+ "epoch": 0.88,
3664
+ "learning_rate": 6.679659815733523e-06,
3665
+ "loss": 0.3624,
3666
+ "step": 5390
3667
+ },
3668
+ {
3669
+ "epoch": 0.88,
3670
+ "learning_rate": 6.591070163004962e-06,
3671
+ "loss": 0.3771,
3672
+ "step": 5400
3673
+ },
3674
+ {
3675
+ "epoch": 0.88,
3676
+ "eval_loss": 0.38931286334991455,
3677
+ "eval_runtime": 138.1336,
3678
+ "eval_samples_per_second": 203.665,
3679
+ "eval_steps_per_second": 6.371,
3680
+ "step": 5400
3681
+ },
3682
+ {
3683
+ "epoch": 0.88,
3684
+ "learning_rate": 6.5024805102764e-06,
3685
+ "loss": 0.3726,
3686
+ "step": 5410
3687
+ },
3688
+ {
3689
+ "epoch": 0.88,
3690
+ "learning_rate": 6.413890857547839e-06,
3691
+ "loss": 0.3551,
3692
+ "step": 5420
3693
+ },
3694
+ {
3695
+ "epoch": 0.88,
3696
+ "learning_rate": 6.325301204819277e-06,
3697
+ "loss": 0.3743,
3698
+ "step": 5430
3699
+ },
3700
+ {
3701
+ "epoch": 0.89,
3702
+ "learning_rate": 6.236711552090716e-06,
3703
+ "loss": 0.3744,
3704
+ "step": 5440
3705
+ },
3706
+ {
3707
+ "epoch": 0.89,
3708
+ "learning_rate": 6.148121899362154e-06,
3709
+ "loss": 0.3712,
3710
+ "step": 5450
3711
+ },
3712
+ {
3713
+ "epoch": 0.89,
3714
+ "learning_rate": 6.059532246633594e-06,
3715
+ "loss": 0.3535,
3716
+ "step": 5460
3717
+ },
3718
+ {
3719
+ "epoch": 0.89,
3720
+ "learning_rate": 5.970942593905032e-06,
3721
+ "loss": 0.3534,
3722
+ "step": 5470
3723
+ },
3724
+ {
3725
+ "epoch": 0.89,
3726
+ "learning_rate": 5.882352941176471e-06,
3727
+ "loss": 0.3498,
3728
+ "step": 5480
3729
+ },
3730
+ {
3731
+ "epoch": 0.89,
3732
+ "learning_rate": 5.7937632884479095e-06,
3733
+ "loss": 0.3767,
3734
+ "step": 5490
3735
+ },
3736
+ {
3737
+ "epoch": 0.9,
3738
+ "learning_rate": 5.705173635719349e-06,
3739
+ "loss": 0.3584,
3740
+ "step": 5500
3741
+ },
3742
+ {
3743
+ "epoch": 0.9,
3744
+ "eval_loss": 0.3893459141254425,
3745
+ "eval_runtime": 138.1479,
3746
+ "eval_samples_per_second": 203.644,
3747
+ "eval_steps_per_second": 6.37,
3748
+ "step": 5500
3749
+ },
3750
+ {
3751
+ "epoch": 0.9,
3752
+ "learning_rate": 5.6165839829907866e-06,
3753
+ "loss": 0.3928,
3754
+ "step": 5510
3755
+ },
3756
+ {
3757
+ "epoch": 0.9,
3758
+ "learning_rate": 5.527994330262225e-06,
3759
+ "loss": 0.339,
3760
+ "step": 5520
3761
+ },
3762
+ {
3763
+ "epoch": 0.9,
3764
+ "learning_rate": 5.4394046775336645e-06,
3765
+ "loss": 0.3619,
3766
+ "step": 5530
3767
+ },
3768
+ {
3769
+ "epoch": 0.9,
3770
+ "learning_rate": 5.350815024805103e-06,
3771
+ "loss": 0.3679,
3772
+ "step": 5540
3773
+ },
3774
+ {
3775
+ "epoch": 0.9,
3776
+ "learning_rate": 5.262225372076542e-06,
3777
+ "loss": 0.3678,
3778
+ "step": 5550
3779
+ },
3780
+ {
3781
+ "epoch": 0.9,
3782
+ "learning_rate": 5.17363571934798e-06,
3783
+ "loss": 0.3864,
3784
+ "step": 5560
3785
+ },
3786
+ {
3787
+ "epoch": 0.91,
3788
+ "learning_rate": 5.0850460666194196e-06,
3789
+ "loss": 0.3724,
3790
+ "step": 5570
3791
+ },
3792
+ {
3793
+ "epoch": 0.91,
3794
+ "learning_rate": 4.996456413890858e-06,
3795
+ "loss": 0.3632,
3796
+ "step": 5580
3797
+ },
3798
+ {
3799
+ "epoch": 0.91,
3800
+ "learning_rate": 4.907866761162296e-06,
3801
+ "loss": 0.3617,
3802
+ "step": 5590
3803
+ },
3804
+ {
3805
+ "epoch": 0.91,
3806
+ "learning_rate": 4.819277108433735e-06,
3807
+ "loss": 0.3609,
3808
+ "step": 5600
3809
+ },
3810
+ {
3811
+ "epoch": 0.91,
3812
+ "eval_loss": 0.3893371820449829,
3813
+ "eval_runtime": 138.139,
3814
+ "eval_samples_per_second": 203.657,
3815
+ "eval_steps_per_second": 6.37,
3816
+ "step": 5600
3817
+ },
3818
+ {
3819
+ "epoch": 0.91,
3820
+ "learning_rate": 4.730687455705174e-06,
3821
+ "loss": 0.3732,
3822
+ "step": 5610
3823
+ },
3824
+ {
3825
+ "epoch": 0.91,
3826
+ "learning_rate": 4.642097802976612e-06,
3827
+ "loss": 0.354,
3828
+ "step": 5620
3829
+ },
3830
+ {
3831
+ "epoch": 0.92,
3832
+ "learning_rate": 4.553508150248052e-06,
3833
+ "loss": 0.3518,
3834
+ "step": 5630
3835
+ },
3836
+ {
3837
+ "epoch": 0.92,
3838
+ "learning_rate": 4.46491849751949e-06,
3839
+ "loss": 0.3579,
3840
+ "step": 5640
3841
+ },
3842
+ {
3843
+ "epoch": 0.92,
3844
+ "learning_rate": 4.376328844790929e-06,
3845
+ "loss": 0.3636,
3846
+ "step": 5650
3847
+ },
3848
+ {
3849
+ "epoch": 0.92,
3850
+ "learning_rate": 4.287739192062367e-06,
3851
+ "loss": 0.3509,
3852
+ "step": 5660
3853
+ },
3854
+ {
3855
+ "epoch": 0.92,
3856
+ "learning_rate": 4.199149539333806e-06,
3857
+ "loss": 0.3624,
3858
+ "step": 5670
3859
+ },
3860
+ {
3861
+ "epoch": 0.92,
3862
+ "learning_rate": 4.1105598866052445e-06,
3863
+ "loss": 0.3735,
3864
+ "step": 5680
3865
+ },
3866
+ {
3867
+ "epoch": 0.93,
3868
+ "learning_rate": 4.021970233876683e-06,
3869
+ "loss": 0.3573,
3870
+ "step": 5690
3871
+ },
3872
+ {
3873
+ "epoch": 0.93,
3874
+ "learning_rate": 3.933380581148122e-06,
3875
+ "loss": 0.3707,
3876
+ "step": 5700
3877
+ },
3878
+ {
3879
+ "epoch": 0.93,
3880
+ "eval_loss": 0.3889307379722595,
3881
+ "eval_runtime": 138.1382,
3882
+ "eval_samples_per_second": 203.658,
3883
+ "eval_steps_per_second": 6.37,
3884
+ "step": 5700
3885
+ },
3886
+ {
3887
+ "epoch": 0.93,
3888
+ "learning_rate": 3.844790928419561e-06,
3889
+ "loss": 0.3446,
3890
+ "step": 5710
3891
+ },
3892
+ {
3893
+ "epoch": 0.93,
3894
+ "learning_rate": 3.7562012756909995e-06,
3895
+ "loss": 0.366,
3896
+ "step": 5720
3897
+ },
3898
+ {
3899
+ "epoch": 0.93,
3900
+ "learning_rate": 3.6676116229624377e-06,
3901
+ "loss": 0.3686,
3902
+ "step": 5730
3903
+ },
3904
+ {
3905
+ "epoch": 0.93,
3906
+ "learning_rate": 3.579021970233877e-06,
3907
+ "loss": 0.3696,
3908
+ "step": 5740
3909
+ },
3910
+ {
3911
+ "epoch": 0.94,
3912
+ "learning_rate": 3.4904323175053156e-06,
3913
+ "loss": 0.388,
3914
+ "step": 5750
3915
+ },
3916
+ {
3917
+ "epoch": 0.94,
3918
+ "learning_rate": 3.401842664776754e-06,
3919
+ "loss": 0.3788,
3920
+ "step": 5760
3921
+ },
3922
+ {
3923
+ "epoch": 0.94,
3924
+ "learning_rate": 3.313253012048193e-06,
3925
+ "loss": 0.3843,
3926
+ "step": 5770
3927
+ },
3928
+ {
3929
+ "epoch": 0.94,
3930
+ "learning_rate": 3.2246633593196317e-06,
3931
+ "loss": 0.3731,
3932
+ "step": 5780
3933
+ },
3934
+ {
3935
+ "epoch": 0.94,
3936
+ "learning_rate": 3.1360737065910702e-06,
3937
+ "loss": 0.3626,
3938
+ "step": 5790
3939
+ },
3940
+ {
3941
+ "epoch": 0.94,
3942
+ "learning_rate": 3.047484053862509e-06,
3943
+ "loss": 0.3787,
3944
+ "step": 5800
3945
+ },
3946
+ {
3947
+ "epoch": 0.94,
3948
+ "eval_loss": 0.3886876702308655,
3949
+ "eval_runtime": 138.1679,
3950
+ "eval_samples_per_second": 203.615,
3951
+ "eval_steps_per_second": 6.369,
3952
+ "step": 5800
3953
+ },
3954
+ {
3955
+ "epoch": 0.95,
3956
+ "learning_rate": 2.9588944011339477e-06,
3957
+ "loss": 0.3721,
3958
+ "step": 5810
3959
+ },
3960
+ {
3961
+ "epoch": 0.95,
3962
+ "learning_rate": 2.8703047484053863e-06,
3963
+ "loss": 0.3575,
3964
+ "step": 5820
3965
+ },
3966
+ {
3967
+ "epoch": 0.95,
3968
+ "learning_rate": 2.7817150956768253e-06,
3969
+ "loss": 0.3607,
3970
+ "step": 5830
3971
+ },
3972
+ {
3973
+ "epoch": 0.95,
3974
+ "learning_rate": 2.693125442948264e-06,
3975
+ "loss": 0.3633,
3976
+ "step": 5840
3977
+ },
3978
+ {
3979
+ "epoch": 0.95,
3980
+ "learning_rate": 2.6045357902197024e-06,
3981
+ "loss": 0.3506,
3982
+ "step": 5850
3983
+ },
3984
+ {
3985
+ "epoch": 0.95,
3986
+ "learning_rate": 2.5159461374911414e-06,
3987
+ "loss": 0.3471,
3988
+ "step": 5860
3989
+ },
3990
+ {
3991
+ "epoch": 0.96,
3992
+ "learning_rate": 2.42735648476258e-06,
3993
+ "loss": 0.3693,
3994
+ "step": 5870
3995
+ },
3996
+ {
3997
+ "epoch": 0.96,
3998
+ "learning_rate": 2.3387668320340185e-06,
3999
+ "loss": 0.3577,
4000
+ "step": 5880
4001
+ },
4002
+ {
4003
+ "epoch": 0.96,
4004
+ "learning_rate": 2.250177179305457e-06,
4005
+ "loss": 0.3685,
4006
+ "step": 5890
4007
+ },
4008
+ {
4009
+ "epoch": 0.96,
4010
+ "learning_rate": 2.161587526576896e-06,
4011
+ "loss": 0.3694,
4012
+ "step": 5900
4013
+ },
4014
+ {
4015
+ "epoch": 0.96,
4016
+ "eval_loss": 0.3885899484157562,
4017
+ "eval_runtime": 138.158,
4018
+ "eval_samples_per_second": 203.629,
4019
+ "eval_steps_per_second": 6.37,
4020
+ "step": 5900
4021
+ },
4022
+ {
4023
+ "epoch": 0.96,
4024
+ "learning_rate": 2.0729978738483345e-06,
4025
+ "loss": 0.3645,
4026
+ "step": 5910
4027
+ },
4028
+ {
4029
+ "epoch": 0.96,
4030
+ "learning_rate": 1.984408221119773e-06,
4031
+ "loss": 0.3786,
4032
+ "step": 5920
4033
+ },
4034
+ {
4035
+ "epoch": 0.97,
4036
+ "learning_rate": 1.895818568391212e-06,
4037
+ "loss": 0.368,
4038
+ "step": 5930
4039
+ },
4040
+ {
4041
+ "epoch": 0.97,
4042
+ "learning_rate": 1.8072289156626506e-06,
4043
+ "loss": 0.3771,
4044
+ "step": 5940
4045
+ },
4046
+ {
4047
+ "epoch": 0.97,
4048
+ "learning_rate": 1.7186392629340894e-06,
4049
+ "loss": 0.3614,
4050
+ "step": 5950
4051
+ },
4052
+ {
4053
+ "epoch": 0.97,
4054
+ "learning_rate": 1.630049610205528e-06,
4055
+ "loss": 0.3739,
4056
+ "step": 5960
4057
+ },
4058
+ {
4059
+ "epoch": 0.97,
4060
+ "learning_rate": 1.5414599574769667e-06,
4061
+ "loss": 0.3649,
4062
+ "step": 5970
4063
+ },
4064
+ {
4065
+ "epoch": 0.97,
4066
+ "learning_rate": 1.4528703047484054e-06,
4067
+ "loss": 0.353,
4068
+ "step": 5980
4069
+ },
4070
+ {
4071
+ "epoch": 0.97,
4072
+ "learning_rate": 1.364280652019844e-06,
4073
+ "loss": 0.3623,
4074
+ "step": 5990
4075
+ },
4076
+ {
4077
+ "epoch": 0.98,
4078
+ "learning_rate": 1.2756909992912828e-06,
4079
+ "loss": 0.3832,
4080
+ "step": 6000
4081
+ },
4082
+ {
4083
+ "epoch": 0.98,
4084
+ "eval_loss": 0.3886074423789978,
4085
+ "eval_runtime": 138.1707,
4086
+ "eval_samples_per_second": 203.61,
4087
+ "eval_steps_per_second": 6.369,
4088
+ "step": 6000
4089
+ }
4090
+ ],
4091
+ "max_steps": 6144,
4092
+ "num_train_epochs": 1,
4093
+ "total_flos": 3.250647465984e+16,
4094
+ "trial_name": null,
4095
+ "trial_params": null
4096
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7231bbae7f083b4848e6cd8c302aa31555eeb42917ca0c4224f74fc22794978
3
+ size 3835