ddhruvkr commited on
Commit
f522bae
1 Parent(s): 0ed2393

Upload 9 files

Browse files

Adding the 770M large model

config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "google/flan-t5-large",
3
+ "architectures": [
4
+ "T5ForConditionalGeneration"
5
+ ],
6
+ "d_ff": 2816,
7
+ "d_kv": 64,
8
+ "d_model": 1024,
9
+ "decoder_start_token_id": 0,
10
+ "dense_act_fn": "gelu_new",
11
+ "dropout_rate": 0.1,
12
+ "eos_token_id": 1,
13
+ "feed_forward_proj": "gated-gelu",
14
+ "initializer_factor": 1.0,
15
+ "is_encoder_decoder": true,
16
+ "is_gated_act": true,
17
+ "layer_norm_epsilon": 1e-06,
18
+ "model_type": "t5",
19
+ "n_positions": 512,
20
+ "num_decoder_layers": 24,
21
+ "num_heads": 16,
22
+ "num_layers": 24,
23
+ "output_past": true,
24
+ "pad_token_id": 0,
25
+ "relative_attention_max_distance": 128,
26
+ "relative_attention_num_buckets": 32,
27
+ "tie_word_embeddings": false,
28
+ "torch_dtype": "float32",
29
+ "transformers_version": "4.27.4",
30
+ "use_cache": true,
31
+ "vocab_size": 32100
32
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "decoder_start_token_id": 0,
4
+ "eos_token_id": 1,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.27.4"
7
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a65c121b7a284783286d1b502d76b05ef886f7baf8b7169af9a2f11d2ac4d74
3
+ size 3132556293
special_tokens_map.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<extra_id_0>",
4
+ "<extra_id_1>",
5
+ "<extra_id_2>",
6
+ "<extra_id_3>",
7
+ "<extra_id_4>",
8
+ "<extra_id_5>",
9
+ "<extra_id_6>",
10
+ "<extra_id_7>",
11
+ "<extra_id_8>",
12
+ "<extra_id_9>",
13
+ "<extra_id_10>",
14
+ "<extra_id_11>",
15
+ "<extra_id_12>",
16
+ "<extra_id_13>",
17
+ "<extra_id_14>",
18
+ "<extra_id_15>",
19
+ "<extra_id_16>",
20
+ "<extra_id_17>",
21
+ "<extra_id_18>",
22
+ "<extra_id_19>",
23
+ "<extra_id_20>",
24
+ "<extra_id_21>",
25
+ "<extra_id_22>",
26
+ "<extra_id_23>",
27
+ "<extra_id_24>",
28
+ "<extra_id_25>",
29
+ "<extra_id_26>",
30
+ "<extra_id_27>",
31
+ "<extra_id_28>",
32
+ "<extra_id_29>",
33
+ "<extra_id_30>",
34
+ "<extra_id_31>",
35
+ "<extra_id_32>",
36
+ "<extra_id_33>",
37
+ "<extra_id_34>",
38
+ "<extra_id_35>",
39
+ "<extra_id_36>",
40
+ "<extra_id_37>",
41
+ "<extra_id_38>",
42
+ "<extra_id_39>",
43
+ "<extra_id_40>",
44
+ "<extra_id_41>",
45
+ "<extra_id_42>",
46
+ "<extra_id_43>",
47
+ "<extra_id_44>",
48
+ "<extra_id_45>",
49
+ "<extra_id_46>",
50
+ "<extra_id_47>",
51
+ "<extra_id_48>",
52
+ "<extra_id_49>",
53
+ "<extra_id_50>",
54
+ "<extra_id_51>",
55
+ "<extra_id_52>",
56
+ "<extra_id_53>",
57
+ "<extra_id_54>",
58
+ "<extra_id_55>",
59
+ "<extra_id_56>",
60
+ "<extra_id_57>",
61
+ "<extra_id_58>",
62
+ "<extra_id_59>",
63
+ "<extra_id_60>",
64
+ "<extra_id_61>",
65
+ "<extra_id_62>",
66
+ "<extra_id_63>",
67
+ "<extra_id_64>",
68
+ "<extra_id_65>",
69
+ "<extra_id_66>",
70
+ "<extra_id_67>",
71
+ "<extra_id_68>",
72
+ "<extra_id_69>",
73
+ "<extra_id_70>",
74
+ "<extra_id_71>",
75
+ "<extra_id_72>",
76
+ "<extra_id_73>",
77
+ "<extra_id_74>",
78
+ "<extra_id_75>",
79
+ "<extra_id_76>",
80
+ "<extra_id_77>",
81
+ "<extra_id_78>",
82
+ "<extra_id_79>",
83
+ "<extra_id_80>",
84
+ "<extra_id_81>",
85
+ "<extra_id_82>",
86
+ "<extra_id_83>",
87
+ "<extra_id_84>",
88
+ "<extra_id_85>",
89
+ "<extra_id_86>",
90
+ "<extra_id_87>",
91
+ "<extra_id_88>",
92
+ "<extra_id_89>",
93
+ "<extra_id_90>",
94
+ "<extra_id_91>",
95
+ "<extra_id_92>",
96
+ "<extra_id_93>",
97
+ "<extra_id_94>",
98
+ "<extra_id_95>",
99
+ "<extra_id_96>",
100
+ "<extra_id_97>",
101
+ "<extra_id_98>",
102
+ "<extra_id_99>"
103
+ ],
104
+ "eos_token": "</s>",
105
+ "pad_token": "<pad>",
106
+ "unk_token": "<unk>"
107
+ }
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86
3
+ size 791656
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,112 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<extra_id_0>",
4
+ "<extra_id_1>",
5
+ "<extra_id_2>",
6
+ "<extra_id_3>",
7
+ "<extra_id_4>",
8
+ "<extra_id_5>",
9
+ "<extra_id_6>",
10
+ "<extra_id_7>",
11
+ "<extra_id_8>",
12
+ "<extra_id_9>",
13
+ "<extra_id_10>",
14
+ "<extra_id_11>",
15
+ "<extra_id_12>",
16
+ "<extra_id_13>",
17
+ "<extra_id_14>",
18
+ "<extra_id_15>",
19
+ "<extra_id_16>",
20
+ "<extra_id_17>",
21
+ "<extra_id_18>",
22
+ "<extra_id_19>",
23
+ "<extra_id_20>",
24
+ "<extra_id_21>",
25
+ "<extra_id_22>",
26
+ "<extra_id_23>",
27
+ "<extra_id_24>",
28
+ "<extra_id_25>",
29
+ "<extra_id_26>",
30
+ "<extra_id_27>",
31
+ "<extra_id_28>",
32
+ "<extra_id_29>",
33
+ "<extra_id_30>",
34
+ "<extra_id_31>",
35
+ "<extra_id_32>",
36
+ "<extra_id_33>",
37
+ "<extra_id_34>",
38
+ "<extra_id_35>",
39
+ "<extra_id_36>",
40
+ "<extra_id_37>",
41
+ "<extra_id_38>",
42
+ "<extra_id_39>",
43
+ "<extra_id_40>",
44
+ "<extra_id_41>",
45
+ "<extra_id_42>",
46
+ "<extra_id_43>",
47
+ "<extra_id_44>",
48
+ "<extra_id_45>",
49
+ "<extra_id_46>",
50
+ "<extra_id_47>",
51
+ "<extra_id_48>",
52
+ "<extra_id_49>",
53
+ "<extra_id_50>",
54
+ "<extra_id_51>",
55
+ "<extra_id_52>",
56
+ "<extra_id_53>",
57
+ "<extra_id_54>",
58
+ "<extra_id_55>",
59
+ "<extra_id_56>",
60
+ "<extra_id_57>",
61
+ "<extra_id_58>",
62
+ "<extra_id_59>",
63
+ "<extra_id_60>",
64
+ "<extra_id_61>",
65
+ "<extra_id_62>",
66
+ "<extra_id_63>",
67
+ "<extra_id_64>",
68
+ "<extra_id_65>",
69
+ "<extra_id_66>",
70
+ "<extra_id_67>",
71
+ "<extra_id_68>",
72
+ "<extra_id_69>",
73
+ "<extra_id_70>",
74
+ "<extra_id_71>",
75
+ "<extra_id_72>",
76
+ "<extra_id_73>",
77
+ "<extra_id_74>",
78
+ "<extra_id_75>",
79
+ "<extra_id_76>",
80
+ "<extra_id_77>",
81
+ "<extra_id_78>",
82
+ "<extra_id_79>",
83
+ "<extra_id_80>",
84
+ "<extra_id_81>",
85
+ "<extra_id_82>",
86
+ "<extra_id_83>",
87
+ "<extra_id_84>",
88
+ "<extra_id_85>",
89
+ "<extra_id_86>",
90
+ "<extra_id_87>",
91
+ "<extra_id_88>",
92
+ "<extra_id_89>",
93
+ "<extra_id_90>",
94
+ "<extra_id_91>",
95
+ "<extra_id_92>",
96
+ "<extra_id_93>",
97
+ "<extra_id_94>",
98
+ "<extra_id_95>",
99
+ "<extra_id_96>",
100
+ "<extra_id_97>",
101
+ "<extra_id_98>",
102
+ "<extra_id_99>"
103
+ ],
104
+ "eos_token": "</s>",
105
+ "extra_ids": 100,
106
+ "model_max_length": 512,
107
+ "pad_token": "<pad>",
108
+ "sp_model_kwargs": {},
109
+ "special_tokens_map_file": "/home/younes_huggingface_co/.cache/huggingface/hub/models--google--t5-v1_1-large/snapshots/314bc112b191ec17b625ba81438dc73d6c23659d/special_tokens_map.json",
110
+ "tokenizer_class": "T5Tokenizer",
111
+ "unk_token": "<unk>"
112
+ }
trainer_state.json ADDED
@@ -0,0 +1,3392 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 4.999775432292836,
5
+ "global_step": 27830,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 0.0001,
13
+ "loss": 0.9801,
14
+ "step": 50
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 0.0001,
19
+ "loss": 0.9375,
20
+ "step": 100
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 0.0001,
25
+ "loss": 0.8924,
26
+ "step": 150
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 0.0001,
31
+ "loss": 0.8901,
32
+ "step": 200
33
+ },
34
+ {
35
+ "epoch": 0.04,
36
+ "learning_rate": 0.0001,
37
+ "loss": 0.8608,
38
+ "step": 250
39
+ },
40
+ {
41
+ "epoch": 0.05,
42
+ "learning_rate": 0.0001,
43
+ "loss": 0.8033,
44
+ "step": 300
45
+ },
46
+ {
47
+ "epoch": 0.06,
48
+ "learning_rate": 0.0001,
49
+ "loss": 0.826,
50
+ "step": 350
51
+ },
52
+ {
53
+ "epoch": 0.07,
54
+ "learning_rate": 0.0001,
55
+ "loss": 0.8897,
56
+ "step": 400
57
+ },
58
+ {
59
+ "epoch": 0.08,
60
+ "learning_rate": 0.0001,
61
+ "loss": 0.8379,
62
+ "step": 450
63
+ },
64
+ {
65
+ "epoch": 0.09,
66
+ "learning_rate": 0.0001,
67
+ "loss": 0.8547,
68
+ "step": 500
69
+ },
70
+ {
71
+ "epoch": 0.1,
72
+ "learning_rate": 0.0001,
73
+ "loss": 0.8092,
74
+ "step": 550
75
+ },
76
+ {
77
+ "epoch": 0.11,
78
+ "learning_rate": 0.0001,
79
+ "loss": 0.8155,
80
+ "step": 600
81
+ },
82
+ {
83
+ "epoch": 0.12,
84
+ "learning_rate": 0.0001,
85
+ "loss": 0.7923,
86
+ "step": 650
87
+ },
88
+ {
89
+ "epoch": 0.13,
90
+ "learning_rate": 0.0001,
91
+ "loss": 0.8267,
92
+ "step": 700
93
+ },
94
+ {
95
+ "epoch": 0.13,
96
+ "learning_rate": 0.0001,
97
+ "loss": 0.8055,
98
+ "step": 750
99
+ },
100
+ {
101
+ "epoch": 0.14,
102
+ "learning_rate": 0.0001,
103
+ "loss": 0.8084,
104
+ "step": 800
105
+ },
106
+ {
107
+ "epoch": 0.15,
108
+ "learning_rate": 0.0001,
109
+ "loss": 0.8232,
110
+ "step": 850
111
+ },
112
+ {
113
+ "epoch": 0.16,
114
+ "learning_rate": 0.0001,
115
+ "loss": 0.7805,
116
+ "step": 900
117
+ },
118
+ {
119
+ "epoch": 0.17,
120
+ "learning_rate": 0.0001,
121
+ "loss": 0.8009,
122
+ "step": 950
123
+ },
124
+ {
125
+ "epoch": 0.18,
126
+ "learning_rate": 0.0001,
127
+ "loss": 0.7566,
128
+ "step": 1000
129
+ },
130
+ {
131
+ "epoch": 0.19,
132
+ "learning_rate": 0.0001,
133
+ "loss": 0.7745,
134
+ "step": 1050
135
+ },
136
+ {
137
+ "epoch": 0.2,
138
+ "learning_rate": 0.0001,
139
+ "loss": 0.7858,
140
+ "step": 1100
141
+ },
142
+ {
143
+ "epoch": 0.21,
144
+ "learning_rate": 0.0001,
145
+ "loss": 0.7508,
146
+ "step": 1150
147
+ },
148
+ {
149
+ "epoch": 0.22,
150
+ "learning_rate": 0.0001,
151
+ "loss": 0.8148,
152
+ "step": 1200
153
+ },
154
+ {
155
+ "epoch": 0.22,
156
+ "learning_rate": 0.0001,
157
+ "loss": 0.8054,
158
+ "step": 1250
159
+ },
160
+ {
161
+ "epoch": 0.23,
162
+ "learning_rate": 0.0001,
163
+ "loss": 0.8149,
164
+ "step": 1300
165
+ },
166
+ {
167
+ "epoch": 0.24,
168
+ "learning_rate": 0.0001,
169
+ "loss": 0.7903,
170
+ "step": 1350
171
+ },
172
+ {
173
+ "epoch": 0.25,
174
+ "learning_rate": 0.0001,
175
+ "loss": 0.7937,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 0.26,
180
+ "learning_rate": 0.0001,
181
+ "loss": 0.7489,
182
+ "step": 1450
183
+ },
184
+ {
185
+ "epoch": 0.27,
186
+ "learning_rate": 0.0001,
187
+ "loss": 0.7783,
188
+ "step": 1500
189
+ },
190
+ {
191
+ "epoch": 0.28,
192
+ "learning_rate": 0.0001,
193
+ "loss": 0.7757,
194
+ "step": 1550
195
+ },
196
+ {
197
+ "epoch": 0.29,
198
+ "learning_rate": 0.0001,
199
+ "loss": 0.7746,
200
+ "step": 1600
201
+ },
202
+ {
203
+ "epoch": 0.3,
204
+ "learning_rate": 0.0001,
205
+ "loss": 0.778,
206
+ "step": 1650
207
+ },
208
+ {
209
+ "epoch": 0.31,
210
+ "learning_rate": 0.0001,
211
+ "loss": 0.7951,
212
+ "step": 1700
213
+ },
214
+ {
215
+ "epoch": 0.31,
216
+ "learning_rate": 0.0001,
217
+ "loss": 0.7764,
218
+ "step": 1750
219
+ },
220
+ {
221
+ "epoch": 0.32,
222
+ "learning_rate": 0.0001,
223
+ "loss": 0.7385,
224
+ "step": 1800
225
+ },
226
+ {
227
+ "epoch": 0.33,
228
+ "learning_rate": 0.0001,
229
+ "loss": 0.7503,
230
+ "step": 1850
231
+ },
232
+ {
233
+ "epoch": 0.34,
234
+ "learning_rate": 0.0001,
235
+ "loss": 0.7673,
236
+ "step": 1900
237
+ },
238
+ {
239
+ "epoch": 0.35,
240
+ "learning_rate": 0.0001,
241
+ "loss": 0.7718,
242
+ "step": 1950
243
+ },
244
+ {
245
+ "epoch": 0.36,
246
+ "learning_rate": 0.0001,
247
+ "loss": 0.7988,
248
+ "step": 2000
249
+ },
250
+ {
251
+ "epoch": 0.37,
252
+ "learning_rate": 0.0001,
253
+ "loss": 0.7394,
254
+ "step": 2050
255
+ },
256
+ {
257
+ "epoch": 0.38,
258
+ "learning_rate": 0.0001,
259
+ "loss": 0.7757,
260
+ "step": 2100
261
+ },
262
+ {
263
+ "epoch": 0.39,
264
+ "learning_rate": 0.0001,
265
+ "loss": 0.7381,
266
+ "step": 2150
267
+ },
268
+ {
269
+ "epoch": 0.4,
270
+ "learning_rate": 0.0001,
271
+ "loss": 0.7857,
272
+ "step": 2200
273
+ },
274
+ {
275
+ "epoch": 0.4,
276
+ "learning_rate": 0.0001,
277
+ "loss": 0.798,
278
+ "step": 2250
279
+ },
280
+ {
281
+ "epoch": 0.41,
282
+ "learning_rate": 0.0001,
283
+ "loss": 0.7987,
284
+ "step": 2300
285
+ },
286
+ {
287
+ "epoch": 0.42,
288
+ "learning_rate": 0.0001,
289
+ "loss": 0.74,
290
+ "step": 2350
291
+ },
292
+ {
293
+ "epoch": 0.43,
294
+ "learning_rate": 0.0001,
295
+ "loss": 0.7832,
296
+ "step": 2400
297
+ },
298
+ {
299
+ "epoch": 0.44,
300
+ "learning_rate": 0.0001,
301
+ "loss": 0.7566,
302
+ "step": 2450
303
+ },
304
+ {
305
+ "epoch": 0.45,
306
+ "learning_rate": 0.0001,
307
+ "loss": 0.7863,
308
+ "step": 2500
309
+ },
310
+ {
311
+ "epoch": 0.46,
312
+ "learning_rate": 0.0001,
313
+ "loss": 0.7419,
314
+ "step": 2550
315
+ },
316
+ {
317
+ "epoch": 0.47,
318
+ "learning_rate": 0.0001,
319
+ "loss": 0.7553,
320
+ "step": 2600
321
+ },
322
+ {
323
+ "epoch": 0.48,
324
+ "learning_rate": 0.0001,
325
+ "loss": 0.7749,
326
+ "step": 2650
327
+ },
328
+ {
329
+ "epoch": 0.49,
330
+ "learning_rate": 0.0001,
331
+ "loss": 0.8235,
332
+ "step": 2700
333
+ },
334
+ {
335
+ "epoch": 0.49,
336
+ "learning_rate": 0.0001,
337
+ "loss": 0.7366,
338
+ "step": 2750
339
+ },
340
+ {
341
+ "epoch": 0.5,
342
+ "learning_rate": 0.0001,
343
+ "loss": 0.802,
344
+ "step": 2800
345
+ },
346
+ {
347
+ "epoch": 0.51,
348
+ "learning_rate": 0.0001,
349
+ "loss": 0.7669,
350
+ "step": 2850
351
+ },
352
+ {
353
+ "epoch": 0.52,
354
+ "learning_rate": 0.0001,
355
+ "loss": 0.7899,
356
+ "step": 2900
357
+ },
358
+ {
359
+ "epoch": 0.53,
360
+ "learning_rate": 0.0001,
361
+ "loss": 0.7651,
362
+ "step": 2950
363
+ },
364
+ {
365
+ "epoch": 0.54,
366
+ "learning_rate": 0.0001,
367
+ "loss": 0.7202,
368
+ "step": 3000
369
+ },
370
+ {
371
+ "epoch": 0.55,
372
+ "learning_rate": 0.0001,
373
+ "loss": 0.7423,
374
+ "step": 3050
375
+ },
376
+ {
377
+ "epoch": 0.56,
378
+ "learning_rate": 0.0001,
379
+ "loss": 0.7732,
380
+ "step": 3100
381
+ },
382
+ {
383
+ "epoch": 0.57,
384
+ "learning_rate": 0.0001,
385
+ "loss": 0.7635,
386
+ "step": 3150
387
+ },
388
+ {
389
+ "epoch": 0.57,
390
+ "learning_rate": 0.0001,
391
+ "loss": 0.7664,
392
+ "step": 3200
393
+ },
394
+ {
395
+ "epoch": 0.58,
396
+ "learning_rate": 0.0001,
397
+ "loss": 0.7623,
398
+ "step": 3250
399
+ },
400
+ {
401
+ "epoch": 0.59,
402
+ "learning_rate": 0.0001,
403
+ "loss": 0.7787,
404
+ "step": 3300
405
+ },
406
+ {
407
+ "epoch": 0.6,
408
+ "learning_rate": 0.0001,
409
+ "loss": 0.7563,
410
+ "step": 3350
411
+ },
412
+ {
413
+ "epoch": 0.61,
414
+ "learning_rate": 0.0001,
415
+ "loss": 0.7318,
416
+ "step": 3400
417
+ },
418
+ {
419
+ "epoch": 0.62,
420
+ "learning_rate": 0.0001,
421
+ "loss": 0.762,
422
+ "step": 3450
423
+ },
424
+ {
425
+ "epoch": 0.63,
426
+ "learning_rate": 0.0001,
427
+ "loss": 0.782,
428
+ "step": 3500
429
+ },
430
+ {
431
+ "epoch": 0.64,
432
+ "learning_rate": 0.0001,
433
+ "loss": 0.8114,
434
+ "step": 3550
435
+ },
436
+ {
437
+ "epoch": 0.65,
438
+ "learning_rate": 0.0001,
439
+ "loss": 0.7182,
440
+ "step": 3600
441
+ },
442
+ {
443
+ "epoch": 0.66,
444
+ "learning_rate": 0.0001,
445
+ "loss": 0.7546,
446
+ "step": 3650
447
+ },
448
+ {
449
+ "epoch": 0.66,
450
+ "learning_rate": 0.0001,
451
+ "loss": 0.7953,
452
+ "step": 3700
453
+ },
454
+ {
455
+ "epoch": 0.67,
456
+ "learning_rate": 0.0001,
457
+ "loss": 0.7772,
458
+ "step": 3750
459
+ },
460
+ {
461
+ "epoch": 0.68,
462
+ "learning_rate": 0.0001,
463
+ "loss": 0.7737,
464
+ "step": 3800
465
+ },
466
+ {
467
+ "epoch": 0.69,
468
+ "learning_rate": 0.0001,
469
+ "loss": 0.7334,
470
+ "step": 3850
471
+ },
472
+ {
473
+ "epoch": 0.7,
474
+ "learning_rate": 0.0001,
475
+ "loss": 0.7713,
476
+ "step": 3900
477
+ },
478
+ {
479
+ "epoch": 0.71,
480
+ "learning_rate": 0.0001,
481
+ "loss": 0.772,
482
+ "step": 3950
483
+ },
484
+ {
485
+ "epoch": 0.72,
486
+ "learning_rate": 0.0001,
487
+ "loss": 0.7964,
488
+ "step": 4000
489
+ },
490
+ {
491
+ "epoch": 0.73,
492
+ "learning_rate": 0.0001,
493
+ "loss": 0.7997,
494
+ "step": 4050
495
+ },
496
+ {
497
+ "epoch": 0.74,
498
+ "learning_rate": 0.0001,
499
+ "loss": 0.7341,
500
+ "step": 4100
501
+ },
502
+ {
503
+ "epoch": 0.75,
504
+ "learning_rate": 0.0001,
505
+ "loss": 0.7837,
506
+ "step": 4150
507
+ },
508
+ {
509
+ "epoch": 0.75,
510
+ "learning_rate": 0.0001,
511
+ "loss": 0.7973,
512
+ "step": 4200
513
+ },
514
+ {
515
+ "epoch": 0.76,
516
+ "learning_rate": 0.0001,
517
+ "loss": 0.775,
518
+ "step": 4250
519
+ },
520
+ {
521
+ "epoch": 0.77,
522
+ "learning_rate": 0.0001,
523
+ "loss": 0.7599,
524
+ "step": 4300
525
+ },
526
+ {
527
+ "epoch": 0.78,
528
+ "learning_rate": 0.0001,
529
+ "loss": 0.7406,
530
+ "step": 4350
531
+ },
532
+ {
533
+ "epoch": 0.79,
534
+ "learning_rate": 0.0001,
535
+ "loss": 0.7875,
536
+ "step": 4400
537
+ },
538
+ {
539
+ "epoch": 0.8,
540
+ "learning_rate": 0.0001,
541
+ "loss": 0.7264,
542
+ "step": 4450
543
+ },
544
+ {
545
+ "epoch": 0.81,
546
+ "learning_rate": 0.0001,
547
+ "loss": 0.7318,
548
+ "step": 4500
549
+ },
550
+ {
551
+ "epoch": 0.82,
552
+ "learning_rate": 0.0001,
553
+ "loss": 0.7532,
554
+ "step": 4550
555
+ },
556
+ {
557
+ "epoch": 0.83,
558
+ "learning_rate": 0.0001,
559
+ "loss": 0.7386,
560
+ "step": 4600
561
+ },
562
+ {
563
+ "epoch": 0.84,
564
+ "learning_rate": 0.0001,
565
+ "loss": 0.7729,
566
+ "step": 4650
567
+ },
568
+ {
569
+ "epoch": 0.84,
570
+ "learning_rate": 0.0001,
571
+ "loss": 0.7428,
572
+ "step": 4700
573
+ },
574
+ {
575
+ "epoch": 0.85,
576
+ "learning_rate": 0.0001,
577
+ "loss": 0.7561,
578
+ "step": 4750
579
+ },
580
+ {
581
+ "epoch": 0.86,
582
+ "learning_rate": 0.0001,
583
+ "loss": 0.7798,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 0.87,
588
+ "learning_rate": 0.0001,
589
+ "loss": 0.7675,
590
+ "step": 4850
591
+ },
592
+ {
593
+ "epoch": 0.88,
594
+ "learning_rate": 0.0001,
595
+ "loss": 0.7443,
596
+ "step": 4900
597
+ },
598
+ {
599
+ "epoch": 0.89,
600
+ "learning_rate": 0.0001,
601
+ "loss": 0.7992,
602
+ "step": 4950
603
+ },
604
+ {
605
+ "epoch": 0.9,
606
+ "learning_rate": 0.0001,
607
+ "loss": 0.7841,
608
+ "step": 5000
609
+ },
610
+ {
611
+ "epoch": 0.91,
612
+ "learning_rate": 0.0001,
613
+ "loss": 0.7626,
614
+ "step": 5050
615
+ },
616
+ {
617
+ "epoch": 0.92,
618
+ "learning_rate": 0.0001,
619
+ "loss": 0.7272,
620
+ "step": 5100
621
+ },
622
+ {
623
+ "epoch": 0.93,
624
+ "learning_rate": 0.0001,
625
+ "loss": 0.7563,
626
+ "step": 5150
627
+ },
628
+ {
629
+ "epoch": 0.93,
630
+ "learning_rate": 0.0001,
631
+ "loss": 0.732,
632
+ "step": 5200
633
+ },
634
+ {
635
+ "epoch": 0.94,
636
+ "learning_rate": 0.0001,
637
+ "loss": 0.7397,
638
+ "step": 5250
639
+ },
640
+ {
641
+ "epoch": 0.95,
642
+ "learning_rate": 0.0001,
643
+ "loss": 0.723,
644
+ "step": 5300
645
+ },
646
+ {
647
+ "epoch": 0.96,
648
+ "learning_rate": 0.0001,
649
+ "loss": 0.7764,
650
+ "step": 5350
651
+ },
652
+ {
653
+ "epoch": 0.97,
654
+ "learning_rate": 0.0001,
655
+ "loss": 0.7605,
656
+ "step": 5400
657
+ },
658
+ {
659
+ "epoch": 0.98,
660
+ "learning_rate": 0.0001,
661
+ "loss": 0.7118,
662
+ "step": 5450
663
+ },
664
+ {
665
+ "epoch": 0.99,
666
+ "learning_rate": 0.0001,
667
+ "loss": 0.7386,
668
+ "step": 5500
669
+ },
670
+ {
671
+ "epoch": 1.0,
672
+ "learning_rate": 0.0001,
673
+ "loss": 0.7196,
674
+ "step": 5550
675
+ },
676
+ {
677
+ "epoch": 1.0,
678
+ "eval_loss": 0.7622227668762207,
679
+ "eval_runtime": 209.9187,
680
+ "eval_samples_per_second": 15.306,
681
+ "eval_steps_per_second": 0.958,
682
+ "step": 5566
683
+ },
684
+ {
685
+ "epoch": 1.01,
686
+ "learning_rate": 0.0001,
687
+ "loss": 0.6938,
688
+ "step": 5600
689
+ },
690
+ {
691
+ "epoch": 1.02,
692
+ "learning_rate": 0.0001,
693
+ "loss": 0.6087,
694
+ "step": 5650
695
+ },
696
+ {
697
+ "epoch": 1.02,
698
+ "learning_rate": 0.0001,
699
+ "loss": 0.6136,
700
+ "step": 5700
701
+ },
702
+ {
703
+ "epoch": 1.03,
704
+ "learning_rate": 0.0001,
705
+ "loss": 0.6061,
706
+ "step": 5750
707
+ },
708
+ {
709
+ "epoch": 1.04,
710
+ "learning_rate": 0.0001,
711
+ "loss": 0.6124,
712
+ "step": 5800
713
+ },
714
+ {
715
+ "epoch": 1.05,
716
+ "learning_rate": 0.0001,
717
+ "loss": 0.5955,
718
+ "step": 5850
719
+ },
720
+ {
721
+ "epoch": 1.06,
722
+ "learning_rate": 0.0001,
723
+ "loss": 0.6269,
724
+ "step": 5900
725
+ },
726
+ {
727
+ "epoch": 1.07,
728
+ "learning_rate": 0.0001,
729
+ "loss": 0.6517,
730
+ "step": 5950
731
+ },
732
+ {
733
+ "epoch": 1.08,
734
+ "learning_rate": 0.0001,
735
+ "loss": 0.6251,
736
+ "step": 6000
737
+ },
738
+ {
739
+ "epoch": 1.09,
740
+ "learning_rate": 0.0001,
741
+ "loss": 0.6702,
742
+ "step": 6050
743
+ },
744
+ {
745
+ "epoch": 1.1,
746
+ "learning_rate": 0.0001,
747
+ "loss": 0.6072,
748
+ "step": 6100
749
+ },
750
+ {
751
+ "epoch": 1.1,
752
+ "learning_rate": 0.0001,
753
+ "loss": 0.6186,
754
+ "step": 6150
755
+ },
756
+ {
757
+ "epoch": 1.11,
758
+ "learning_rate": 0.0001,
759
+ "loss": 0.6037,
760
+ "step": 6200
761
+ },
762
+ {
763
+ "epoch": 1.12,
764
+ "learning_rate": 0.0001,
765
+ "loss": 0.5849,
766
+ "step": 6250
767
+ },
768
+ {
769
+ "epoch": 1.13,
770
+ "learning_rate": 0.0001,
771
+ "loss": 0.5903,
772
+ "step": 6300
773
+ },
774
+ {
775
+ "epoch": 1.14,
776
+ "learning_rate": 0.0001,
777
+ "loss": 0.6206,
778
+ "step": 6350
779
+ },
780
+ {
781
+ "epoch": 1.15,
782
+ "learning_rate": 0.0001,
783
+ "loss": 0.6017,
784
+ "step": 6400
785
+ },
786
+ {
787
+ "epoch": 1.16,
788
+ "learning_rate": 0.0001,
789
+ "loss": 0.5683,
790
+ "step": 6450
791
+ },
792
+ {
793
+ "epoch": 1.17,
794
+ "learning_rate": 0.0001,
795
+ "loss": 0.6193,
796
+ "step": 6500
797
+ },
798
+ {
799
+ "epoch": 1.18,
800
+ "learning_rate": 0.0001,
801
+ "loss": 0.619,
802
+ "step": 6550
803
+ },
804
+ {
805
+ "epoch": 1.19,
806
+ "learning_rate": 0.0001,
807
+ "loss": 0.5784,
808
+ "step": 6600
809
+ },
810
+ {
811
+ "epoch": 1.19,
812
+ "learning_rate": 0.0001,
813
+ "loss": 0.6097,
814
+ "step": 6650
815
+ },
816
+ {
817
+ "epoch": 1.2,
818
+ "learning_rate": 0.0001,
819
+ "loss": 0.598,
820
+ "step": 6700
821
+ },
822
+ {
823
+ "epoch": 1.21,
824
+ "learning_rate": 0.0001,
825
+ "loss": 0.5999,
826
+ "step": 6750
827
+ },
828
+ {
829
+ "epoch": 1.22,
830
+ "learning_rate": 0.0001,
831
+ "loss": 0.6172,
832
+ "step": 6800
833
+ },
834
+ {
835
+ "epoch": 1.23,
836
+ "learning_rate": 0.0001,
837
+ "loss": 0.677,
838
+ "step": 6850
839
+ },
840
+ {
841
+ "epoch": 1.24,
842
+ "learning_rate": 0.0001,
843
+ "loss": 0.6057,
844
+ "step": 6900
845
+ },
846
+ {
847
+ "epoch": 1.25,
848
+ "learning_rate": 0.0001,
849
+ "loss": 0.6055,
850
+ "step": 6950
851
+ },
852
+ {
853
+ "epoch": 1.26,
854
+ "learning_rate": 0.0001,
855
+ "loss": 0.6023,
856
+ "step": 7000
857
+ },
858
+ {
859
+ "epoch": 1.27,
860
+ "learning_rate": 0.0001,
861
+ "loss": 0.6063,
862
+ "step": 7050
863
+ },
864
+ {
865
+ "epoch": 1.28,
866
+ "learning_rate": 0.0001,
867
+ "loss": 0.6092,
868
+ "step": 7100
869
+ },
870
+ {
871
+ "epoch": 1.28,
872
+ "learning_rate": 0.0001,
873
+ "loss": 0.622,
874
+ "step": 7150
875
+ },
876
+ {
877
+ "epoch": 1.29,
878
+ "learning_rate": 0.0001,
879
+ "loss": 0.5788,
880
+ "step": 7200
881
+ },
882
+ {
883
+ "epoch": 1.3,
884
+ "learning_rate": 0.0001,
885
+ "loss": 0.6234,
886
+ "step": 7250
887
+ },
888
+ {
889
+ "epoch": 1.31,
890
+ "learning_rate": 0.0001,
891
+ "loss": 0.566,
892
+ "step": 7300
893
+ },
894
+ {
895
+ "epoch": 1.32,
896
+ "learning_rate": 0.0001,
897
+ "loss": 0.6285,
898
+ "step": 7350
899
+ },
900
+ {
901
+ "epoch": 1.33,
902
+ "learning_rate": 0.0001,
903
+ "loss": 0.5961,
904
+ "step": 7400
905
+ },
906
+ {
907
+ "epoch": 1.34,
908
+ "learning_rate": 0.0001,
909
+ "loss": 0.5837,
910
+ "step": 7450
911
+ },
912
+ {
913
+ "epoch": 1.35,
914
+ "learning_rate": 0.0001,
915
+ "loss": 0.6118,
916
+ "step": 7500
917
+ },
918
+ {
919
+ "epoch": 1.36,
920
+ "learning_rate": 0.0001,
921
+ "loss": 0.6089,
922
+ "step": 7550
923
+ },
924
+ {
925
+ "epoch": 1.37,
926
+ "learning_rate": 0.0001,
927
+ "loss": 0.6041,
928
+ "step": 7600
929
+ },
930
+ {
931
+ "epoch": 1.37,
932
+ "learning_rate": 0.0001,
933
+ "loss": 0.607,
934
+ "step": 7650
935
+ },
936
+ {
937
+ "epoch": 1.38,
938
+ "learning_rate": 0.0001,
939
+ "loss": 0.6613,
940
+ "step": 7700
941
+ },
942
+ {
943
+ "epoch": 1.39,
944
+ "learning_rate": 0.0001,
945
+ "loss": 0.6176,
946
+ "step": 7750
947
+ },
948
+ {
949
+ "epoch": 1.4,
950
+ "learning_rate": 0.0001,
951
+ "loss": 0.5791,
952
+ "step": 7800
953
+ },
954
+ {
955
+ "epoch": 1.41,
956
+ "learning_rate": 0.0001,
957
+ "loss": 0.6092,
958
+ "step": 7850
959
+ },
960
+ {
961
+ "epoch": 1.42,
962
+ "learning_rate": 0.0001,
963
+ "loss": 0.6262,
964
+ "step": 7900
965
+ },
966
+ {
967
+ "epoch": 1.43,
968
+ "learning_rate": 0.0001,
969
+ "loss": 0.6086,
970
+ "step": 7950
971
+ },
972
+ {
973
+ "epoch": 1.44,
974
+ "learning_rate": 0.0001,
975
+ "loss": 0.5735,
976
+ "step": 8000
977
+ },
978
+ {
979
+ "epoch": 1.45,
980
+ "learning_rate": 0.0001,
981
+ "loss": 0.6109,
982
+ "step": 8050
983
+ },
984
+ {
985
+ "epoch": 1.46,
986
+ "learning_rate": 0.0001,
987
+ "loss": 0.6125,
988
+ "step": 8100
989
+ },
990
+ {
991
+ "epoch": 1.46,
992
+ "learning_rate": 0.0001,
993
+ "loss": 0.6218,
994
+ "step": 8150
995
+ },
996
+ {
997
+ "epoch": 1.47,
998
+ "learning_rate": 0.0001,
999
+ "loss": 0.6117,
1000
+ "step": 8200
1001
+ },
1002
+ {
1003
+ "epoch": 1.48,
1004
+ "learning_rate": 0.0001,
1005
+ "loss": 0.5824,
1006
+ "step": 8250
1007
+ },
1008
+ {
1009
+ "epoch": 1.49,
1010
+ "learning_rate": 0.0001,
1011
+ "loss": 0.6448,
1012
+ "step": 8300
1013
+ },
1014
+ {
1015
+ "epoch": 1.5,
1016
+ "learning_rate": 0.0001,
1017
+ "loss": 0.6563,
1018
+ "step": 8350
1019
+ },
1020
+ {
1021
+ "epoch": 1.51,
1022
+ "learning_rate": 0.0001,
1023
+ "loss": 0.6707,
1024
+ "step": 8400
1025
+ },
1026
+ {
1027
+ "epoch": 1.52,
1028
+ "learning_rate": 0.0001,
1029
+ "loss": 0.6097,
1030
+ "step": 8450
1031
+ },
1032
+ {
1033
+ "epoch": 1.53,
1034
+ "learning_rate": 0.0001,
1035
+ "loss": 0.5972,
1036
+ "step": 8500
1037
+ },
1038
+ {
1039
+ "epoch": 1.54,
1040
+ "learning_rate": 0.0001,
1041
+ "loss": 0.6104,
1042
+ "step": 8550
1043
+ },
1044
+ {
1045
+ "epoch": 1.55,
1046
+ "learning_rate": 0.0001,
1047
+ "loss": 0.601,
1048
+ "step": 8600
1049
+ },
1050
+ {
1051
+ "epoch": 1.55,
1052
+ "learning_rate": 0.0001,
1053
+ "loss": 0.581,
1054
+ "step": 8650
1055
+ },
1056
+ {
1057
+ "epoch": 1.56,
1058
+ "learning_rate": 0.0001,
1059
+ "loss": 0.6065,
1060
+ "step": 8700
1061
+ },
1062
+ {
1063
+ "epoch": 1.57,
1064
+ "learning_rate": 0.0001,
1065
+ "loss": 0.5787,
1066
+ "step": 8750
1067
+ },
1068
+ {
1069
+ "epoch": 1.58,
1070
+ "learning_rate": 0.0001,
1071
+ "loss": 0.6347,
1072
+ "step": 8800
1073
+ },
1074
+ {
1075
+ "epoch": 1.59,
1076
+ "learning_rate": 0.0001,
1077
+ "loss": 0.6461,
1078
+ "step": 8850
1079
+ },
1080
+ {
1081
+ "epoch": 1.6,
1082
+ "learning_rate": 0.0001,
1083
+ "loss": 0.6046,
1084
+ "step": 8900
1085
+ },
1086
+ {
1087
+ "epoch": 1.61,
1088
+ "learning_rate": 0.0001,
1089
+ "loss": 0.6498,
1090
+ "step": 8950
1091
+ },
1092
+ {
1093
+ "epoch": 1.62,
1094
+ "learning_rate": 0.0001,
1095
+ "loss": 0.6283,
1096
+ "step": 9000
1097
+ },
1098
+ {
1099
+ "epoch": 1.63,
1100
+ "learning_rate": 0.0001,
1101
+ "loss": 0.6221,
1102
+ "step": 9050
1103
+ },
1104
+ {
1105
+ "epoch": 1.63,
1106
+ "learning_rate": 0.0001,
1107
+ "loss": 0.6324,
1108
+ "step": 9100
1109
+ },
1110
+ {
1111
+ "epoch": 1.64,
1112
+ "learning_rate": 0.0001,
1113
+ "loss": 0.6142,
1114
+ "step": 9150
1115
+ },
1116
+ {
1117
+ "epoch": 1.65,
1118
+ "learning_rate": 0.0001,
1119
+ "loss": 0.5834,
1120
+ "step": 9200
1121
+ },
1122
+ {
1123
+ "epoch": 1.66,
1124
+ "learning_rate": 0.0001,
1125
+ "loss": 0.6326,
1126
+ "step": 9250
1127
+ },
1128
+ {
1129
+ "epoch": 1.67,
1130
+ "learning_rate": 0.0001,
1131
+ "loss": 0.5982,
1132
+ "step": 9300
1133
+ },
1134
+ {
1135
+ "epoch": 1.68,
1136
+ "learning_rate": 0.0001,
1137
+ "loss": 0.5965,
1138
+ "step": 9350
1139
+ },
1140
+ {
1141
+ "epoch": 1.69,
1142
+ "learning_rate": 0.0001,
1143
+ "loss": 0.6052,
1144
+ "step": 9400
1145
+ },
1146
+ {
1147
+ "epoch": 1.7,
1148
+ "learning_rate": 0.0001,
1149
+ "loss": 0.6658,
1150
+ "step": 9450
1151
+ },
1152
+ {
1153
+ "epoch": 1.71,
1154
+ "learning_rate": 0.0001,
1155
+ "loss": 0.6163,
1156
+ "step": 9500
1157
+ },
1158
+ {
1159
+ "epoch": 1.72,
1160
+ "learning_rate": 0.0001,
1161
+ "loss": 0.6627,
1162
+ "step": 9550
1163
+ },
1164
+ {
1165
+ "epoch": 1.72,
1166
+ "learning_rate": 0.0001,
1167
+ "loss": 0.6203,
1168
+ "step": 9600
1169
+ },
1170
+ {
1171
+ "epoch": 1.73,
1172
+ "learning_rate": 0.0001,
1173
+ "loss": 0.633,
1174
+ "step": 9650
1175
+ },
1176
+ {
1177
+ "epoch": 1.74,
1178
+ "learning_rate": 0.0001,
1179
+ "loss": 0.651,
1180
+ "step": 9700
1181
+ },
1182
+ {
1183
+ "epoch": 1.75,
1184
+ "learning_rate": 0.0001,
1185
+ "loss": 0.5837,
1186
+ "step": 9750
1187
+ },
1188
+ {
1189
+ "epoch": 1.76,
1190
+ "learning_rate": 0.0001,
1191
+ "loss": 0.6166,
1192
+ "step": 9800
1193
+ },
1194
+ {
1195
+ "epoch": 1.77,
1196
+ "learning_rate": 0.0001,
1197
+ "loss": 0.5995,
1198
+ "step": 9850
1199
+ },
1200
+ {
1201
+ "epoch": 1.78,
1202
+ "learning_rate": 0.0001,
1203
+ "loss": 0.5883,
1204
+ "step": 9900
1205
+ },
1206
+ {
1207
+ "epoch": 1.79,
1208
+ "learning_rate": 0.0001,
1209
+ "loss": 0.5918,
1210
+ "step": 9950
1211
+ },
1212
+ {
1213
+ "epoch": 1.8,
1214
+ "learning_rate": 0.0001,
1215
+ "loss": 0.5785,
1216
+ "step": 10000
1217
+ },
1218
+ {
1219
+ "epoch": 1.81,
1220
+ "learning_rate": 0.0001,
1221
+ "loss": 0.6557,
1222
+ "step": 10050
1223
+ },
1224
+ {
1225
+ "epoch": 1.81,
1226
+ "learning_rate": 0.0001,
1227
+ "loss": 0.5912,
1228
+ "step": 10100
1229
+ },
1230
+ {
1231
+ "epoch": 1.82,
1232
+ "learning_rate": 0.0001,
1233
+ "loss": 0.6109,
1234
+ "step": 10150
1235
+ },
1236
+ {
1237
+ "epoch": 1.83,
1238
+ "learning_rate": 0.0001,
1239
+ "loss": 0.599,
1240
+ "step": 10200
1241
+ },
1242
+ {
1243
+ "epoch": 1.84,
1244
+ "learning_rate": 0.0001,
1245
+ "loss": 0.6325,
1246
+ "step": 10250
1247
+ },
1248
+ {
1249
+ "epoch": 1.85,
1250
+ "learning_rate": 0.0001,
1251
+ "loss": 0.6339,
1252
+ "step": 10300
1253
+ },
1254
+ {
1255
+ "epoch": 1.86,
1256
+ "learning_rate": 0.0001,
1257
+ "loss": 0.6074,
1258
+ "step": 10350
1259
+ },
1260
+ {
1261
+ "epoch": 1.87,
1262
+ "learning_rate": 0.0001,
1263
+ "loss": 0.5943,
1264
+ "step": 10400
1265
+ },
1266
+ {
1267
+ "epoch": 1.88,
1268
+ "learning_rate": 0.0001,
1269
+ "loss": 0.6133,
1270
+ "step": 10450
1271
+ },
1272
+ {
1273
+ "epoch": 1.89,
1274
+ "learning_rate": 0.0001,
1275
+ "loss": 0.5984,
1276
+ "step": 10500
1277
+ },
1278
+ {
1279
+ "epoch": 1.9,
1280
+ "learning_rate": 0.0001,
1281
+ "loss": 0.6695,
1282
+ "step": 10550
1283
+ },
1284
+ {
1285
+ "epoch": 1.9,
1286
+ "learning_rate": 0.0001,
1287
+ "loss": 0.591,
1288
+ "step": 10600
1289
+ },
1290
+ {
1291
+ "epoch": 1.91,
1292
+ "learning_rate": 0.0001,
1293
+ "loss": 0.6208,
1294
+ "step": 10650
1295
+ },
1296
+ {
1297
+ "epoch": 1.92,
1298
+ "learning_rate": 0.0001,
1299
+ "loss": 0.5744,
1300
+ "step": 10700
1301
+ },
1302
+ {
1303
+ "epoch": 1.93,
1304
+ "learning_rate": 0.0001,
1305
+ "loss": 0.5873,
1306
+ "step": 10750
1307
+ },
1308
+ {
1309
+ "epoch": 1.94,
1310
+ "learning_rate": 0.0001,
1311
+ "loss": 0.6261,
1312
+ "step": 10800
1313
+ },
1314
+ {
1315
+ "epoch": 1.95,
1316
+ "learning_rate": 0.0001,
1317
+ "loss": 0.617,
1318
+ "step": 10850
1319
+ },
1320
+ {
1321
+ "epoch": 1.96,
1322
+ "learning_rate": 0.0001,
1323
+ "loss": 0.6151,
1324
+ "step": 10900
1325
+ },
1326
+ {
1327
+ "epoch": 1.97,
1328
+ "learning_rate": 0.0001,
1329
+ "loss": 0.5936,
1330
+ "step": 10950
1331
+ },
1332
+ {
1333
+ "epoch": 1.98,
1334
+ "learning_rate": 0.0001,
1335
+ "loss": 0.6366,
1336
+ "step": 11000
1337
+ },
1338
+ {
1339
+ "epoch": 1.99,
1340
+ "learning_rate": 0.0001,
1341
+ "loss": 0.644,
1342
+ "step": 11050
1343
+ },
1344
+ {
1345
+ "epoch": 1.99,
1346
+ "learning_rate": 0.0001,
1347
+ "loss": 0.6302,
1348
+ "step": 11100
1349
+ },
1350
+ {
1351
+ "epoch": 2.0,
1352
+ "eval_loss": 0.7252166271209717,
1353
+ "eval_runtime": 201.6817,
1354
+ "eval_samples_per_second": 15.931,
1355
+ "eval_steps_per_second": 0.997,
1356
+ "step": 11132
1357
+ },
1358
+ {
1359
+ "epoch": 2.0,
1360
+ "learning_rate": 0.0001,
1361
+ "loss": 0.5489,
1362
+ "step": 11150
1363
+ },
1364
+ {
1365
+ "epoch": 2.01,
1366
+ "learning_rate": 0.0001,
1367
+ "loss": 0.5112,
1368
+ "step": 11200
1369
+ },
1370
+ {
1371
+ "epoch": 2.02,
1372
+ "learning_rate": 0.0001,
1373
+ "loss": 0.4868,
1374
+ "step": 11250
1375
+ },
1376
+ {
1377
+ "epoch": 2.03,
1378
+ "learning_rate": 0.0001,
1379
+ "loss": 0.483,
1380
+ "step": 11300
1381
+ },
1382
+ {
1383
+ "epoch": 2.04,
1384
+ "learning_rate": 0.0001,
1385
+ "loss": 0.4569,
1386
+ "step": 11350
1387
+ },
1388
+ {
1389
+ "epoch": 2.05,
1390
+ "learning_rate": 0.0001,
1391
+ "loss": 0.4638,
1392
+ "step": 11400
1393
+ },
1394
+ {
1395
+ "epoch": 2.06,
1396
+ "learning_rate": 0.0001,
1397
+ "loss": 0.4764,
1398
+ "step": 11450
1399
+ },
1400
+ {
1401
+ "epoch": 2.07,
1402
+ "learning_rate": 0.0001,
1403
+ "loss": 0.4984,
1404
+ "step": 11500
1405
+ },
1406
+ {
1407
+ "epoch": 2.08,
1408
+ "learning_rate": 0.0001,
1409
+ "loss": 0.4797,
1410
+ "step": 11550
1411
+ },
1412
+ {
1413
+ "epoch": 2.08,
1414
+ "learning_rate": 0.0001,
1415
+ "loss": 0.4924,
1416
+ "step": 11600
1417
+ },
1418
+ {
1419
+ "epoch": 2.09,
1420
+ "learning_rate": 0.0001,
1421
+ "loss": 0.4921,
1422
+ "step": 11650
1423
+ },
1424
+ {
1425
+ "epoch": 2.1,
1426
+ "learning_rate": 0.0001,
1427
+ "loss": 0.4652,
1428
+ "step": 11700
1429
+ },
1430
+ {
1431
+ "epoch": 2.11,
1432
+ "learning_rate": 0.0001,
1433
+ "loss": 0.4842,
1434
+ "step": 11750
1435
+ },
1436
+ {
1437
+ "epoch": 2.12,
1438
+ "learning_rate": 0.0001,
1439
+ "loss": 0.4982,
1440
+ "step": 11800
1441
+ },
1442
+ {
1443
+ "epoch": 2.13,
1444
+ "learning_rate": 0.0001,
1445
+ "loss": 0.4682,
1446
+ "step": 11850
1447
+ },
1448
+ {
1449
+ "epoch": 2.14,
1450
+ "learning_rate": 0.0001,
1451
+ "loss": 0.4642,
1452
+ "step": 11900
1453
+ },
1454
+ {
1455
+ "epoch": 2.15,
1456
+ "learning_rate": 0.0001,
1457
+ "loss": 0.4982,
1458
+ "step": 11950
1459
+ },
1460
+ {
1461
+ "epoch": 2.16,
1462
+ "learning_rate": 0.0001,
1463
+ "loss": 0.5336,
1464
+ "step": 12000
1465
+ },
1466
+ {
1467
+ "epoch": 2.16,
1468
+ "learning_rate": 0.0001,
1469
+ "loss": 0.4739,
1470
+ "step": 12050
1471
+ },
1472
+ {
1473
+ "epoch": 2.17,
1474
+ "learning_rate": 0.0001,
1475
+ "loss": 0.4938,
1476
+ "step": 12100
1477
+ },
1478
+ {
1479
+ "epoch": 2.18,
1480
+ "learning_rate": 0.0001,
1481
+ "loss": 0.5114,
1482
+ "step": 12150
1483
+ },
1484
+ {
1485
+ "epoch": 2.19,
1486
+ "learning_rate": 0.0001,
1487
+ "loss": 0.4893,
1488
+ "step": 12200
1489
+ },
1490
+ {
1491
+ "epoch": 2.2,
1492
+ "learning_rate": 0.0001,
1493
+ "loss": 0.4924,
1494
+ "step": 12250
1495
+ },
1496
+ {
1497
+ "epoch": 2.21,
1498
+ "learning_rate": 0.0001,
1499
+ "loss": 0.4826,
1500
+ "step": 12300
1501
+ },
1502
+ {
1503
+ "epoch": 2.22,
1504
+ "learning_rate": 0.0001,
1505
+ "loss": 0.4651,
1506
+ "step": 12350
1507
+ },
1508
+ {
1509
+ "epoch": 2.23,
1510
+ "learning_rate": 0.0001,
1511
+ "loss": 0.5192,
1512
+ "step": 12400
1513
+ },
1514
+ {
1515
+ "epoch": 2.24,
1516
+ "learning_rate": 0.0001,
1517
+ "loss": 0.5012,
1518
+ "step": 12450
1519
+ },
1520
+ {
1521
+ "epoch": 2.25,
1522
+ "learning_rate": 0.0001,
1523
+ "loss": 0.4913,
1524
+ "step": 12500
1525
+ },
1526
+ {
1527
+ "epoch": 2.25,
1528
+ "learning_rate": 0.0001,
1529
+ "loss": 0.4999,
1530
+ "step": 12550
1531
+ },
1532
+ {
1533
+ "epoch": 2.26,
1534
+ "learning_rate": 0.0001,
1535
+ "loss": 0.5245,
1536
+ "step": 12600
1537
+ },
1538
+ {
1539
+ "epoch": 2.27,
1540
+ "learning_rate": 0.0001,
1541
+ "loss": 0.5264,
1542
+ "step": 12650
1543
+ },
1544
+ {
1545
+ "epoch": 2.28,
1546
+ "learning_rate": 0.0001,
1547
+ "loss": 0.5132,
1548
+ "step": 12700
1549
+ },
1550
+ {
1551
+ "epoch": 2.29,
1552
+ "learning_rate": 0.0001,
1553
+ "loss": 0.4598,
1554
+ "step": 12750
1555
+ },
1556
+ {
1557
+ "epoch": 2.3,
1558
+ "learning_rate": 0.0001,
1559
+ "loss": 0.4883,
1560
+ "step": 12800
1561
+ },
1562
+ {
1563
+ "epoch": 2.31,
1564
+ "learning_rate": 0.0001,
1565
+ "loss": 0.5397,
1566
+ "step": 12850
1567
+ },
1568
+ {
1569
+ "epoch": 2.32,
1570
+ "learning_rate": 0.0001,
1571
+ "loss": 0.5253,
1572
+ "step": 12900
1573
+ },
1574
+ {
1575
+ "epoch": 2.33,
1576
+ "learning_rate": 0.0001,
1577
+ "loss": 0.5067,
1578
+ "step": 12950
1579
+ },
1580
+ {
1581
+ "epoch": 2.34,
1582
+ "learning_rate": 0.0001,
1583
+ "loss": 0.4831,
1584
+ "step": 13000
1585
+ },
1586
+ {
1587
+ "epoch": 2.34,
1588
+ "learning_rate": 0.0001,
1589
+ "loss": 0.482,
1590
+ "step": 13050
1591
+ },
1592
+ {
1593
+ "epoch": 2.35,
1594
+ "learning_rate": 0.0001,
1595
+ "loss": 0.5039,
1596
+ "step": 13100
1597
+ },
1598
+ {
1599
+ "epoch": 2.36,
1600
+ "learning_rate": 0.0001,
1601
+ "loss": 0.473,
1602
+ "step": 13150
1603
+ },
1604
+ {
1605
+ "epoch": 2.37,
1606
+ "learning_rate": 0.0001,
1607
+ "loss": 0.5094,
1608
+ "step": 13200
1609
+ },
1610
+ {
1611
+ "epoch": 2.38,
1612
+ "learning_rate": 0.0001,
1613
+ "loss": 0.4866,
1614
+ "step": 13250
1615
+ },
1616
+ {
1617
+ "epoch": 2.39,
1618
+ "learning_rate": 0.0001,
1619
+ "loss": 0.4881,
1620
+ "step": 13300
1621
+ },
1622
+ {
1623
+ "epoch": 2.4,
1624
+ "learning_rate": 0.0001,
1625
+ "loss": 0.5268,
1626
+ "step": 13350
1627
+ },
1628
+ {
1629
+ "epoch": 2.41,
1630
+ "learning_rate": 0.0001,
1631
+ "loss": 0.5188,
1632
+ "step": 13400
1633
+ },
1634
+ {
1635
+ "epoch": 2.42,
1636
+ "learning_rate": 0.0001,
1637
+ "loss": 0.501,
1638
+ "step": 13450
1639
+ },
1640
+ {
1641
+ "epoch": 2.43,
1642
+ "learning_rate": 0.0001,
1643
+ "loss": 0.5135,
1644
+ "step": 13500
1645
+ },
1646
+ {
1647
+ "epoch": 2.43,
1648
+ "learning_rate": 0.0001,
1649
+ "loss": 0.5056,
1650
+ "step": 13550
1651
+ },
1652
+ {
1653
+ "epoch": 2.44,
1654
+ "learning_rate": 0.0001,
1655
+ "loss": 0.5062,
1656
+ "step": 13600
1657
+ },
1658
+ {
1659
+ "epoch": 2.45,
1660
+ "learning_rate": 0.0001,
1661
+ "loss": 0.4823,
1662
+ "step": 13650
1663
+ },
1664
+ {
1665
+ "epoch": 2.46,
1666
+ "learning_rate": 0.0001,
1667
+ "loss": 0.4636,
1668
+ "step": 13700
1669
+ },
1670
+ {
1671
+ "epoch": 2.47,
1672
+ "learning_rate": 0.0001,
1673
+ "loss": 0.507,
1674
+ "step": 13750
1675
+ },
1676
+ {
1677
+ "epoch": 2.48,
1678
+ "learning_rate": 0.0001,
1679
+ "loss": 0.4648,
1680
+ "step": 13800
1681
+ },
1682
+ {
1683
+ "epoch": 2.49,
1684
+ "learning_rate": 0.0001,
1685
+ "loss": 0.5061,
1686
+ "step": 13850
1687
+ },
1688
+ {
1689
+ "epoch": 2.5,
1690
+ "learning_rate": 0.0001,
1691
+ "loss": 0.4872,
1692
+ "step": 13900
1693
+ },
1694
+ {
1695
+ "epoch": 2.51,
1696
+ "learning_rate": 0.0001,
1697
+ "loss": 0.4897,
1698
+ "step": 13950
1699
+ },
1700
+ {
1701
+ "epoch": 2.52,
1702
+ "learning_rate": 0.0001,
1703
+ "loss": 0.5166,
1704
+ "step": 14000
1705
+ },
1706
+ {
1707
+ "epoch": 2.52,
1708
+ "learning_rate": 0.0001,
1709
+ "loss": 0.4764,
1710
+ "step": 14050
1711
+ },
1712
+ {
1713
+ "epoch": 2.53,
1714
+ "learning_rate": 0.0001,
1715
+ "loss": 0.5094,
1716
+ "step": 14100
1717
+ },
1718
+ {
1719
+ "epoch": 2.54,
1720
+ "learning_rate": 0.0001,
1721
+ "loss": 0.4821,
1722
+ "step": 14150
1723
+ },
1724
+ {
1725
+ "epoch": 2.55,
1726
+ "learning_rate": 0.0001,
1727
+ "loss": 0.5222,
1728
+ "step": 14200
1729
+ },
1730
+ {
1731
+ "epoch": 2.56,
1732
+ "learning_rate": 0.0001,
1733
+ "loss": 0.4904,
1734
+ "step": 14250
1735
+ },
1736
+ {
1737
+ "epoch": 2.57,
1738
+ "learning_rate": 0.0001,
1739
+ "loss": 0.5085,
1740
+ "step": 14300
1741
+ },
1742
+ {
1743
+ "epoch": 2.58,
1744
+ "learning_rate": 0.0001,
1745
+ "loss": 0.4719,
1746
+ "step": 14350
1747
+ },
1748
+ {
1749
+ "epoch": 2.59,
1750
+ "learning_rate": 0.0001,
1751
+ "loss": 0.5112,
1752
+ "step": 14400
1753
+ },
1754
+ {
1755
+ "epoch": 2.6,
1756
+ "learning_rate": 0.0001,
1757
+ "loss": 0.5052,
1758
+ "step": 14450
1759
+ },
1760
+ {
1761
+ "epoch": 2.6,
1762
+ "learning_rate": 0.0001,
1763
+ "loss": 0.5457,
1764
+ "step": 14500
1765
+ },
1766
+ {
1767
+ "epoch": 2.61,
1768
+ "learning_rate": 0.0001,
1769
+ "loss": 0.5034,
1770
+ "step": 14550
1771
+ },
1772
+ {
1773
+ "epoch": 2.62,
1774
+ "learning_rate": 0.0001,
1775
+ "loss": 0.5023,
1776
+ "step": 14600
1777
+ },
1778
+ {
1779
+ "epoch": 2.63,
1780
+ "learning_rate": 0.0001,
1781
+ "loss": 0.4967,
1782
+ "step": 14650
1783
+ },
1784
+ {
1785
+ "epoch": 2.64,
1786
+ "learning_rate": 0.0001,
1787
+ "loss": 0.5124,
1788
+ "step": 14700
1789
+ },
1790
+ {
1791
+ "epoch": 2.65,
1792
+ "learning_rate": 0.0001,
1793
+ "loss": 0.4987,
1794
+ "step": 14750
1795
+ },
1796
+ {
1797
+ "epoch": 2.66,
1798
+ "learning_rate": 0.0001,
1799
+ "loss": 0.5234,
1800
+ "step": 14800
1801
+ },
1802
+ {
1803
+ "epoch": 2.67,
1804
+ "learning_rate": 0.0001,
1805
+ "loss": 0.4957,
1806
+ "step": 14850
1807
+ },
1808
+ {
1809
+ "epoch": 2.68,
1810
+ "learning_rate": 0.0001,
1811
+ "loss": 0.516,
1812
+ "step": 14900
1813
+ },
1814
+ {
1815
+ "epoch": 2.69,
1816
+ "learning_rate": 0.0001,
1817
+ "loss": 0.479,
1818
+ "step": 14950
1819
+ },
1820
+ {
1821
+ "epoch": 2.69,
1822
+ "learning_rate": 0.0001,
1823
+ "loss": 0.5057,
1824
+ "step": 15000
1825
+ },
1826
+ {
1827
+ "epoch": 2.7,
1828
+ "learning_rate": 0.0001,
1829
+ "loss": 0.5224,
1830
+ "step": 15050
1831
+ },
1832
+ {
1833
+ "epoch": 2.71,
1834
+ "learning_rate": 0.0001,
1835
+ "loss": 0.5428,
1836
+ "step": 15100
1837
+ },
1838
+ {
1839
+ "epoch": 2.72,
1840
+ "learning_rate": 0.0001,
1841
+ "loss": 0.4645,
1842
+ "step": 15150
1843
+ },
1844
+ {
1845
+ "epoch": 2.73,
1846
+ "learning_rate": 0.0001,
1847
+ "loss": 0.5114,
1848
+ "step": 15200
1849
+ },
1850
+ {
1851
+ "epoch": 2.74,
1852
+ "learning_rate": 0.0001,
1853
+ "loss": 0.5074,
1854
+ "step": 15250
1855
+ },
1856
+ {
1857
+ "epoch": 2.75,
1858
+ "learning_rate": 0.0001,
1859
+ "loss": 0.5103,
1860
+ "step": 15300
1861
+ },
1862
+ {
1863
+ "epoch": 2.76,
1864
+ "learning_rate": 0.0001,
1865
+ "loss": 0.5326,
1866
+ "step": 15350
1867
+ },
1868
+ {
1869
+ "epoch": 2.77,
1870
+ "learning_rate": 0.0001,
1871
+ "loss": 0.5023,
1872
+ "step": 15400
1873
+ },
1874
+ {
1875
+ "epoch": 2.78,
1876
+ "learning_rate": 0.0001,
1877
+ "loss": 0.467,
1878
+ "step": 15450
1879
+ },
1880
+ {
1881
+ "epoch": 2.78,
1882
+ "learning_rate": 0.0001,
1883
+ "loss": 0.4833,
1884
+ "step": 15500
1885
+ },
1886
+ {
1887
+ "epoch": 2.79,
1888
+ "learning_rate": 0.0001,
1889
+ "loss": 0.5119,
1890
+ "step": 15550
1891
+ },
1892
+ {
1893
+ "epoch": 2.8,
1894
+ "learning_rate": 0.0001,
1895
+ "loss": 0.5181,
1896
+ "step": 15600
1897
+ },
1898
+ {
1899
+ "epoch": 2.81,
1900
+ "learning_rate": 0.0001,
1901
+ "loss": 0.4917,
1902
+ "step": 15650
1903
+ },
1904
+ {
1905
+ "epoch": 2.82,
1906
+ "learning_rate": 0.0001,
1907
+ "loss": 0.5281,
1908
+ "step": 15700
1909
+ },
1910
+ {
1911
+ "epoch": 2.83,
1912
+ "learning_rate": 0.0001,
1913
+ "loss": 0.4971,
1914
+ "step": 15750
1915
+ },
1916
+ {
1917
+ "epoch": 2.84,
1918
+ "learning_rate": 0.0001,
1919
+ "loss": 0.483,
1920
+ "step": 15800
1921
+ },
1922
+ {
1923
+ "epoch": 2.85,
1924
+ "learning_rate": 0.0001,
1925
+ "loss": 0.5254,
1926
+ "step": 15850
1927
+ },
1928
+ {
1929
+ "epoch": 2.86,
1930
+ "learning_rate": 0.0001,
1931
+ "loss": 0.5838,
1932
+ "step": 15900
1933
+ },
1934
+ {
1935
+ "epoch": 2.87,
1936
+ "learning_rate": 0.0001,
1937
+ "loss": 0.5201,
1938
+ "step": 15950
1939
+ },
1940
+ {
1941
+ "epoch": 2.87,
1942
+ "learning_rate": 0.0001,
1943
+ "loss": 0.5232,
1944
+ "step": 16000
1945
+ },
1946
+ {
1947
+ "epoch": 2.88,
1948
+ "learning_rate": 0.0001,
1949
+ "loss": 0.5249,
1950
+ "step": 16050
1951
+ },
1952
+ {
1953
+ "epoch": 2.89,
1954
+ "learning_rate": 0.0001,
1955
+ "loss": 0.4707,
1956
+ "step": 16100
1957
+ },
1958
+ {
1959
+ "epoch": 2.9,
1960
+ "learning_rate": 0.0001,
1961
+ "loss": 0.5058,
1962
+ "step": 16150
1963
+ },
1964
+ {
1965
+ "epoch": 2.91,
1966
+ "learning_rate": 0.0001,
1967
+ "loss": 0.4838,
1968
+ "step": 16200
1969
+ },
1970
+ {
1971
+ "epoch": 2.92,
1972
+ "learning_rate": 0.0001,
1973
+ "loss": 0.5012,
1974
+ "step": 16250
1975
+ },
1976
+ {
1977
+ "epoch": 2.93,
1978
+ "learning_rate": 0.0001,
1979
+ "loss": 0.5369,
1980
+ "step": 16300
1981
+ },
1982
+ {
1983
+ "epoch": 2.94,
1984
+ "learning_rate": 0.0001,
1985
+ "loss": 0.5372,
1986
+ "step": 16350
1987
+ },
1988
+ {
1989
+ "epoch": 2.95,
1990
+ "learning_rate": 0.0001,
1991
+ "loss": 0.5441,
1992
+ "step": 16400
1993
+ },
1994
+ {
1995
+ "epoch": 2.96,
1996
+ "learning_rate": 0.0001,
1997
+ "loss": 0.5506,
1998
+ "step": 16450
1999
+ },
2000
+ {
2001
+ "epoch": 2.96,
2002
+ "learning_rate": 0.0001,
2003
+ "loss": 0.5412,
2004
+ "step": 16500
2005
+ },
2006
+ {
2007
+ "epoch": 2.97,
2008
+ "learning_rate": 0.0001,
2009
+ "loss": 0.4903,
2010
+ "step": 16550
2011
+ },
2012
+ {
2013
+ "epoch": 2.98,
2014
+ "learning_rate": 0.0001,
2015
+ "loss": 0.5145,
2016
+ "step": 16600
2017
+ },
2018
+ {
2019
+ "epoch": 2.99,
2020
+ "learning_rate": 0.0001,
2021
+ "loss": 0.5237,
2022
+ "step": 16650
2023
+ },
2024
+ {
2025
+ "epoch": 3.0,
2026
+ "eval_loss": 0.7354088425636292,
2027
+ "eval_runtime": 207.8886,
2028
+ "eval_samples_per_second": 15.455,
2029
+ "eval_steps_per_second": 0.967,
2030
+ "step": 16698
2031
+ },
2032
+ {
2033
+ "epoch": 3.0,
2034
+ "learning_rate": 0.0001,
2035
+ "loss": 0.5086,
2036
+ "step": 16700
2037
+ },
2038
+ {
2039
+ "epoch": 3.01,
2040
+ "learning_rate": 0.0001,
2041
+ "loss": 0.4071,
2042
+ "step": 16750
2043
+ },
2044
+ {
2045
+ "epoch": 3.02,
2046
+ "learning_rate": 0.0001,
2047
+ "loss": 0.4299,
2048
+ "step": 16800
2049
+ },
2050
+ {
2051
+ "epoch": 3.03,
2052
+ "learning_rate": 0.0001,
2053
+ "loss": 0.3996,
2054
+ "step": 16850
2055
+ },
2056
+ {
2057
+ "epoch": 3.04,
2058
+ "learning_rate": 0.0001,
2059
+ "loss": 0.412,
2060
+ "step": 16900
2061
+ },
2062
+ {
2063
+ "epoch": 3.05,
2064
+ "learning_rate": 0.0001,
2065
+ "loss": 0.4214,
2066
+ "step": 16950
2067
+ },
2068
+ {
2069
+ "epoch": 3.05,
2070
+ "learning_rate": 0.0001,
2071
+ "loss": 0.3986,
2072
+ "step": 17000
2073
+ },
2074
+ {
2075
+ "epoch": 3.06,
2076
+ "learning_rate": 0.0001,
2077
+ "loss": 0.4177,
2078
+ "step": 17050
2079
+ },
2080
+ {
2081
+ "epoch": 3.07,
2082
+ "learning_rate": 0.0001,
2083
+ "loss": 0.3833,
2084
+ "step": 17100
2085
+ },
2086
+ {
2087
+ "epoch": 3.08,
2088
+ "learning_rate": 0.0001,
2089
+ "loss": 0.4054,
2090
+ "step": 17150
2091
+ },
2092
+ {
2093
+ "epoch": 3.09,
2094
+ "learning_rate": 0.0001,
2095
+ "loss": 0.3802,
2096
+ "step": 17200
2097
+ },
2098
+ {
2099
+ "epoch": 3.1,
2100
+ "learning_rate": 0.0001,
2101
+ "loss": 0.4142,
2102
+ "step": 17250
2103
+ },
2104
+ {
2105
+ "epoch": 3.11,
2106
+ "learning_rate": 0.0001,
2107
+ "loss": 0.3996,
2108
+ "step": 17300
2109
+ },
2110
+ {
2111
+ "epoch": 3.12,
2112
+ "learning_rate": 0.0001,
2113
+ "loss": 0.4025,
2114
+ "step": 17350
2115
+ },
2116
+ {
2117
+ "epoch": 3.13,
2118
+ "learning_rate": 0.0001,
2119
+ "loss": 0.3554,
2120
+ "step": 17400
2121
+ },
2122
+ {
2123
+ "epoch": 3.13,
2124
+ "learning_rate": 0.0001,
2125
+ "loss": 0.3842,
2126
+ "step": 17450
2127
+ },
2128
+ {
2129
+ "epoch": 3.14,
2130
+ "learning_rate": 0.0001,
2131
+ "loss": 0.3796,
2132
+ "step": 17500
2133
+ },
2134
+ {
2135
+ "epoch": 3.15,
2136
+ "learning_rate": 0.0001,
2137
+ "loss": 0.3765,
2138
+ "step": 17550
2139
+ },
2140
+ {
2141
+ "epoch": 3.16,
2142
+ "learning_rate": 0.0001,
2143
+ "loss": 0.4065,
2144
+ "step": 17600
2145
+ },
2146
+ {
2147
+ "epoch": 3.17,
2148
+ "learning_rate": 0.0001,
2149
+ "loss": 0.3981,
2150
+ "step": 17650
2151
+ },
2152
+ {
2153
+ "epoch": 3.18,
2154
+ "learning_rate": 0.0001,
2155
+ "loss": 0.4013,
2156
+ "step": 17700
2157
+ },
2158
+ {
2159
+ "epoch": 3.19,
2160
+ "learning_rate": 0.0001,
2161
+ "loss": 0.4043,
2162
+ "step": 17750
2163
+ },
2164
+ {
2165
+ "epoch": 3.2,
2166
+ "learning_rate": 0.0001,
2167
+ "loss": 0.422,
2168
+ "step": 17800
2169
+ },
2170
+ {
2171
+ "epoch": 3.21,
2172
+ "learning_rate": 0.0001,
2173
+ "loss": 0.4058,
2174
+ "step": 17850
2175
+ },
2176
+ {
2177
+ "epoch": 3.22,
2178
+ "learning_rate": 0.0001,
2179
+ "loss": 0.3772,
2180
+ "step": 17900
2181
+ },
2182
+ {
2183
+ "epoch": 3.22,
2184
+ "learning_rate": 0.0001,
2185
+ "loss": 0.3926,
2186
+ "step": 17950
2187
+ },
2188
+ {
2189
+ "epoch": 3.23,
2190
+ "learning_rate": 0.0001,
2191
+ "loss": 0.3846,
2192
+ "step": 18000
2193
+ },
2194
+ {
2195
+ "epoch": 3.24,
2196
+ "learning_rate": 0.0001,
2197
+ "loss": 0.4259,
2198
+ "step": 18050
2199
+ },
2200
+ {
2201
+ "epoch": 3.25,
2202
+ "learning_rate": 0.0001,
2203
+ "loss": 0.3775,
2204
+ "step": 18100
2205
+ },
2206
+ {
2207
+ "epoch": 3.26,
2208
+ "learning_rate": 0.0001,
2209
+ "loss": 0.4124,
2210
+ "step": 18150
2211
+ },
2212
+ {
2213
+ "epoch": 3.27,
2214
+ "learning_rate": 0.0001,
2215
+ "loss": 0.404,
2216
+ "step": 18200
2217
+ },
2218
+ {
2219
+ "epoch": 3.28,
2220
+ "learning_rate": 0.0001,
2221
+ "loss": 0.4017,
2222
+ "step": 18250
2223
+ },
2224
+ {
2225
+ "epoch": 3.29,
2226
+ "learning_rate": 0.0001,
2227
+ "loss": 0.4419,
2228
+ "step": 18300
2229
+ },
2230
+ {
2231
+ "epoch": 3.3,
2232
+ "learning_rate": 0.0001,
2233
+ "loss": 0.4284,
2234
+ "step": 18350
2235
+ },
2236
+ {
2237
+ "epoch": 3.31,
2238
+ "learning_rate": 0.0001,
2239
+ "loss": 0.4295,
2240
+ "step": 18400
2241
+ },
2242
+ {
2243
+ "epoch": 3.31,
2244
+ "learning_rate": 0.0001,
2245
+ "loss": 0.4021,
2246
+ "step": 18450
2247
+ },
2248
+ {
2249
+ "epoch": 3.32,
2250
+ "learning_rate": 0.0001,
2251
+ "loss": 0.4314,
2252
+ "step": 18500
2253
+ },
2254
+ {
2255
+ "epoch": 3.33,
2256
+ "learning_rate": 0.0001,
2257
+ "loss": 0.4332,
2258
+ "step": 18550
2259
+ },
2260
+ {
2261
+ "epoch": 3.34,
2262
+ "learning_rate": 0.0001,
2263
+ "loss": 0.3919,
2264
+ "step": 18600
2265
+ },
2266
+ {
2267
+ "epoch": 3.35,
2268
+ "learning_rate": 0.0001,
2269
+ "loss": 0.4467,
2270
+ "step": 18650
2271
+ },
2272
+ {
2273
+ "epoch": 3.36,
2274
+ "learning_rate": 0.0001,
2275
+ "loss": 0.3978,
2276
+ "step": 18700
2277
+ },
2278
+ {
2279
+ "epoch": 3.37,
2280
+ "learning_rate": 0.0001,
2281
+ "loss": 0.4264,
2282
+ "step": 18750
2283
+ },
2284
+ {
2285
+ "epoch": 3.38,
2286
+ "learning_rate": 0.0001,
2287
+ "loss": 0.421,
2288
+ "step": 18800
2289
+ },
2290
+ {
2291
+ "epoch": 3.39,
2292
+ "learning_rate": 0.0001,
2293
+ "loss": 0.4217,
2294
+ "step": 18850
2295
+ },
2296
+ {
2297
+ "epoch": 3.4,
2298
+ "learning_rate": 0.0001,
2299
+ "loss": 0.4244,
2300
+ "step": 18900
2301
+ },
2302
+ {
2303
+ "epoch": 3.4,
2304
+ "learning_rate": 0.0001,
2305
+ "loss": 0.3918,
2306
+ "step": 18950
2307
+ },
2308
+ {
2309
+ "epoch": 3.41,
2310
+ "learning_rate": 0.0001,
2311
+ "loss": 0.4048,
2312
+ "step": 19000
2313
+ },
2314
+ {
2315
+ "epoch": 3.42,
2316
+ "learning_rate": 0.0001,
2317
+ "loss": 0.4142,
2318
+ "step": 19050
2319
+ },
2320
+ {
2321
+ "epoch": 3.43,
2322
+ "learning_rate": 0.0001,
2323
+ "loss": 0.4071,
2324
+ "step": 19100
2325
+ },
2326
+ {
2327
+ "epoch": 3.44,
2328
+ "learning_rate": 0.0001,
2329
+ "loss": 0.39,
2330
+ "step": 19150
2331
+ },
2332
+ {
2333
+ "epoch": 3.45,
2334
+ "learning_rate": 0.0001,
2335
+ "loss": 0.3983,
2336
+ "step": 19200
2337
+ },
2338
+ {
2339
+ "epoch": 3.46,
2340
+ "learning_rate": 0.0001,
2341
+ "loss": 0.4342,
2342
+ "step": 19250
2343
+ },
2344
+ {
2345
+ "epoch": 3.47,
2346
+ "learning_rate": 0.0001,
2347
+ "loss": 0.4071,
2348
+ "step": 19300
2349
+ },
2350
+ {
2351
+ "epoch": 3.48,
2352
+ "learning_rate": 0.0001,
2353
+ "loss": 0.4294,
2354
+ "step": 19350
2355
+ },
2356
+ {
2357
+ "epoch": 3.49,
2358
+ "learning_rate": 0.0001,
2359
+ "loss": 0.3989,
2360
+ "step": 19400
2361
+ },
2362
+ {
2363
+ "epoch": 3.49,
2364
+ "learning_rate": 0.0001,
2365
+ "loss": 0.4115,
2366
+ "step": 19450
2367
+ },
2368
+ {
2369
+ "epoch": 3.5,
2370
+ "learning_rate": 0.0001,
2371
+ "loss": 0.4238,
2372
+ "step": 19500
2373
+ },
2374
+ {
2375
+ "epoch": 3.51,
2376
+ "learning_rate": 0.0001,
2377
+ "loss": 0.3964,
2378
+ "step": 19550
2379
+ },
2380
+ {
2381
+ "epoch": 3.52,
2382
+ "learning_rate": 0.0001,
2383
+ "loss": 0.4103,
2384
+ "step": 19600
2385
+ },
2386
+ {
2387
+ "epoch": 3.53,
2388
+ "learning_rate": 0.0001,
2389
+ "loss": 0.416,
2390
+ "step": 19650
2391
+ },
2392
+ {
2393
+ "epoch": 3.54,
2394
+ "learning_rate": 0.0001,
2395
+ "loss": 0.4362,
2396
+ "step": 19700
2397
+ },
2398
+ {
2399
+ "epoch": 3.55,
2400
+ "learning_rate": 0.0001,
2401
+ "loss": 0.4094,
2402
+ "step": 19750
2403
+ },
2404
+ {
2405
+ "epoch": 3.56,
2406
+ "learning_rate": 0.0001,
2407
+ "loss": 0.4167,
2408
+ "step": 19800
2409
+ },
2410
+ {
2411
+ "epoch": 3.57,
2412
+ "learning_rate": 0.0001,
2413
+ "loss": 0.4025,
2414
+ "step": 19850
2415
+ },
2416
+ {
2417
+ "epoch": 3.58,
2418
+ "learning_rate": 0.0001,
2419
+ "loss": 0.3804,
2420
+ "step": 19900
2421
+ },
2422
+ {
2423
+ "epoch": 3.58,
2424
+ "learning_rate": 0.0001,
2425
+ "loss": 0.4188,
2426
+ "step": 19950
2427
+ },
2428
+ {
2429
+ "epoch": 3.59,
2430
+ "learning_rate": 0.0001,
2431
+ "loss": 0.4333,
2432
+ "step": 20000
2433
+ },
2434
+ {
2435
+ "epoch": 3.6,
2436
+ "learning_rate": 0.0001,
2437
+ "loss": 0.4001,
2438
+ "step": 20050
2439
+ },
2440
+ {
2441
+ "epoch": 3.61,
2442
+ "learning_rate": 0.0001,
2443
+ "loss": 0.3774,
2444
+ "step": 20100
2445
+ },
2446
+ {
2447
+ "epoch": 3.62,
2448
+ "learning_rate": 0.0001,
2449
+ "loss": 0.4332,
2450
+ "step": 20150
2451
+ },
2452
+ {
2453
+ "epoch": 3.63,
2454
+ "learning_rate": 0.0001,
2455
+ "loss": 0.4342,
2456
+ "step": 20200
2457
+ },
2458
+ {
2459
+ "epoch": 3.64,
2460
+ "learning_rate": 0.0001,
2461
+ "loss": 0.4147,
2462
+ "step": 20250
2463
+ },
2464
+ {
2465
+ "epoch": 3.65,
2466
+ "learning_rate": 0.0001,
2467
+ "loss": 0.4372,
2468
+ "step": 20300
2469
+ },
2470
+ {
2471
+ "epoch": 3.66,
2472
+ "learning_rate": 0.0001,
2473
+ "loss": 0.3962,
2474
+ "step": 20350
2475
+ },
2476
+ {
2477
+ "epoch": 3.66,
2478
+ "learning_rate": 0.0001,
2479
+ "loss": 0.4262,
2480
+ "step": 20400
2481
+ },
2482
+ {
2483
+ "epoch": 3.67,
2484
+ "learning_rate": 0.0001,
2485
+ "loss": 0.4362,
2486
+ "step": 20450
2487
+ },
2488
+ {
2489
+ "epoch": 3.68,
2490
+ "learning_rate": 0.0001,
2491
+ "loss": 0.4112,
2492
+ "step": 20500
2493
+ },
2494
+ {
2495
+ "epoch": 3.69,
2496
+ "learning_rate": 0.0001,
2497
+ "loss": 0.4011,
2498
+ "step": 20550
2499
+ },
2500
+ {
2501
+ "epoch": 3.7,
2502
+ "learning_rate": 0.0001,
2503
+ "loss": 0.3906,
2504
+ "step": 20600
2505
+ },
2506
+ {
2507
+ "epoch": 3.71,
2508
+ "learning_rate": 0.0001,
2509
+ "loss": 0.4094,
2510
+ "step": 20650
2511
+ },
2512
+ {
2513
+ "epoch": 3.72,
2514
+ "learning_rate": 0.0001,
2515
+ "loss": 0.4302,
2516
+ "step": 20700
2517
+ },
2518
+ {
2519
+ "epoch": 3.73,
2520
+ "learning_rate": 0.0001,
2521
+ "loss": 0.4143,
2522
+ "step": 20750
2523
+ },
2524
+ {
2525
+ "epoch": 3.74,
2526
+ "learning_rate": 0.0001,
2527
+ "loss": 0.398,
2528
+ "step": 20800
2529
+ },
2530
+ {
2531
+ "epoch": 3.75,
2532
+ "learning_rate": 0.0001,
2533
+ "loss": 0.43,
2534
+ "step": 20850
2535
+ },
2536
+ {
2537
+ "epoch": 3.75,
2538
+ "learning_rate": 0.0001,
2539
+ "loss": 0.4124,
2540
+ "step": 20900
2541
+ },
2542
+ {
2543
+ "epoch": 3.76,
2544
+ "learning_rate": 0.0001,
2545
+ "loss": 0.4167,
2546
+ "step": 20950
2547
+ },
2548
+ {
2549
+ "epoch": 3.77,
2550
+ "learning_rate": 0.0001,
2551
+ "loss": 0.4219,
2552
+ "step": 21000
2553
+ },
2554
+ {
2555
+ "epoch": 3.78,
2556
+ "learning_rate": 0.0001,
2557
+ "loss": 0.4264,
2558
+ "step": 21050
2559
+ },
2560
+ {
2561
+ "epoch": 3.79,
2562
+ "learning_rate": 0.0001,
2563
+ "loss": 0.435,
2564
+ "step": 21100
2565
+ },
2566
+ {
2567
+ "epoch": 3.8,
2568
+ "learning_rate": 0.0001,
2569
+ "loss": 0.4449,
2570
+ "step": 21150
2571
+ },
2572
+ {
2573
+ "epoch": 3.81,
2574
+ "learning_rate": 0.0001,
2575
+ "loss": 0.4595,
2576
+ "step": 21200
2577
+ },
2578
+ {
2579
+ "epoch": 3.82,
2580
+ "learning_rate": 0.0001,
2581
+ "loss": 0.4202,
2582
+ "step": 21250
2583
+ },
2584
+ {
2585
+ "epoch": 3.83,
2586
+ "learning_rate": 0.0001,
2587
+ "loss": 0.3981,
2588
+ "step": 21300
2589
+ },
2590
+ {
2591
+ "epoch": 3.84,
2592
+ "learning_rate": 0.0001,
2593
+ "loss": 0.4261,
2594
+ "step": 21350
2595
+ },
2596
+ {
2597
+ "epoch": 3.84,
2598
+ "learning_rate": 0.0001,
2599
+ "loss": 0.4412,
2600
+ "step": 21400
2601
+ },
2602
+ {
2603
+ "epoch": 3.85,
2604
+ "learning_rate": 0.0001,
2605
+ "loss": 0.3918,
2606
+ "step": 21450
2607
+ },
2608
+ {
2609
+ "epoch": 3.86,
2610
+ "learning_rate": 0.0001,
2611
+ "loss": 0.4354,
2612
+ "step": 21500
2613
+ },
2614
+ {
2615
+ "epoch": 3.87,
2616
+ "learning_rate": 0.0001,
2617
+ "loss": 0.4373,
2618
+ "step": 21550
2619
+ },
2620
+ {
2621
+ "epoch": 3.88,
2622
+ "learning_rate": 0.0001,
2623
+ "loss": 0.4302,
2624
+ "step": 21600
2625
+ },
2626
+ {
2627
+ "epoch": 3.89,
2628
+ "learning_rate": 0.0001,
2629
+ "loss": 0.4113,
2630
+ "step": 21650
2631
+ },
2632
+ {
2633
+ "epoch": 3.9,
2634
+ "learning_rate": 0.0001,
2635
+ "loss": 0.392,
2636
+ "step": 21700
2637
+ },
2638
+ {
2639
+ "epoch": 3.91,
2640
+ "learning_rate": 0.0001,
2641
+ "loss": 0.4335,
2642
+ "step": 21750
2643
+ },
2644
+ {
2645
+ "epoch": 3.92,
2646
+ "learning_rate": 0.0001,
2647
+ "loss": 0.4211,
2648
+ "step": 21800
2649
+ },
2650
+ {
2651
+ "epoch": 3.93,
2652
+ "learning_rate": 0.0001,
2653
+ "loss": 0.4178,
2654
+ "step": 21850
2655
+ },
2656
+ {
2657
+ "epoch": 3.93,
2658
+ "learning_rate": 0.0001,
2659
+ "loss": 0.4333,
2660
+ "step": 21900
2661
+ },
2662
+ {
2663
+ "epoch": 3.94,
2664
+ "learning_rate": 0.0001,
2665
+ "loss": 0.4438,
2666
+ "step": 21950
2667
+ },
2668
+ {
2669
+ "epoch": 3.95,
2670
+ "learning_rate": 0.0001,
2671
+ "loss": 0.4118,
2672
+ "step": 22000
2673
+ },
2674
+ {
2675
+ "epoch": 3.96,
2676
+ "learning_rate": 0.0001,
2677
+ "loss": 0.4256,
2678
+ "step": 22050
2679
+ },
2680
+ {
2681
+ "epoch": 3.97,
2682
+ "learning_rate": 0.0001,
2683
+ "loss": 0.4394,
2684
+ "step": 22100
2685
+ },
2686
+ {
2687
+ "epoch": 3.98,
2688
+ "learning_rate": 0.0001,
2689
+ "loss": 0.4329,
2690
+ "step": 22150
2691
+ },
2692
+ {
2693
+ "epoch": 3.99,
2694
+ "learning_rate": 0.0001,
2695
+ "loss": 0.4481,
2696
+ "step": 22200
2697
+ },
2698
+ {
2699
+ "epoch": 4.0,
2700
+ "learning_rate": 0.0001,
2701
+ "loss": 0.4215,
2702
+ "step": 22250
2703
+ },
2704
+ {
2705
+ "epoch": 4.0,
2706
+ "eval_loss": 0.7574929594993591,
2707
+ "eval_runtime": 207.1872,
2708
+ "eval_samples_per_second": 15.508,
2709
+ "eval_steps_per_second": 0.97,
2710
+ "step": 22265
2711
+ },
2712
+ {
2713
+ "epoch": 4.01,
2714
+ "learning_rate": 0.0001,
2715
+ "loss": 0.3323,
2716
+ "step": 22300
2717
+ },
2718
+ {
2719
+ "epoch": 4.02,
2720
+ "learning_rate": 0.0001,
2721
+ "loss": 0.3289,
2722
+ "step": 22350
2723
+ },
2724
+ {
2725
+ "epoch": 4.02,
2726
+ "learning_rate": 0.0001,
2727
+ "loss": 0.3332,
2728
+ "step": 22400
2729
+ },
2730
+ {
2731
+ "epoch": 4.03,
2732
+ "learning_rate": 0.0001,
2733
+ "loss": 0.3404,
2734
+ "step": 22450
2735
+ },
2736
+ {
2737
+ "epoch": 4.04,
2738
+ "learning_rate": 0.0001,
2739
+ "loss": 0.3108,
2740
+ "step": 22500
2741
+ },
2742
+ {
2743
+ "epoch": 4.05,
2744
+ "learning_rate": 0.0001,
2745
+ "loss": 0.3362,
2746
+ "step": 22550
2747
+ },
2748
+ {
2749
+ "epoch": 4.06,
2750
+ "learning_rate": 0.0001,
2751
+ "loss": 0.3105,
2752
+ "step": 22600
2753
+ },
2754
+ {
2755
+ "epoch": 4.07,
2756
+ "learning_rate": 0.0001,
2757
+ "loss": 0.3056,
2758
+ "step": 22650
2759
+ },
2760
+ {
2761
+ "epoch": 4.08,
2762
+ "learning_rate": 0.0001,
2763
+ "loss": 0.3253,
2764
+ "step": 22700
2765
+ },
2766
+ {
2767
+ "epoch": 4.09,
2768
+ "learning_rate": 0.0001,
2769
+ "loss": 0.3151,
2770
+ "step": 22750
2771
+ },
2772
+ {
2773
+ "epoch": 4.1,
2774
+ "learning_rate": 0.0001,
2775
+ "loss": 0.3482,
2776
+ "step": 22800
2777
+ },
2778
+ {
2779
+ "epoch": 4.11,
2780
+ "learning_rate": 0.0001,
2781
+ "loss": 0.3216,
2782
+ "step": 22850
2783
+ },
2784
+ {
2785
+ "epoch": 4.11,
2786
+ "learning_rate": 0.0001,
2787
+ "loss": 0.3492,
2788
+ "step": 22900
2789
+ },
2790
+ {
2791
+ "epoch": 4.12,
2792
+ "learning_rate": 0.0001,
2793
+ "loss": 0.3172,
2794
+ "step": 22950
2795
+ },
2796
+ {
2797
+ "epoch": 4.13,
2798
+ "learning_rate": 0.0001,
2799
+ "loss": 0.3397,
2800
+ "step": 23000
2801
+ },
2802
+ {
2803
+ "epoch": 4.14,
2804
+ "learning_rate": 0.0001,
2805
+ "loss": 0.3569,
2806
+ "step": 23050
2807
+ },
2808
+ {
2809
+ "epoch": 4.15,
2810
+ "learning_rate": 0.0001,
2811
+ "loss": 0.3143,
2812
+ "step": 23100
2813
+ },
2814
+ {
2815
+ "epoch": 4.16,
2816
+ "learning_rate": 0.0001,
2817
+ "loss": 0.313,
2818
+ "step": 23150
2819
+ },
2820
+ {
2821
+ "epoch": 4.17,
2822
+ "learning_rate": 0.0001,
2823
+ "loss": 0.3394,
2824
+ "step": 23200
2825
+ },
2826
+ {
2827
+ "epoch": 4.18,
2828
+ "learning_rate": 0.0001,
2829
+ "loss": 0.3434,
2830
+ "step": 23250
2831
+ },
2832
+ {
2833
+ "epoch": 4.19,
2834
+ "learning_rate": 0.0001,
2835
+ "loss": 0.3217,
2836
+ "step": 23300
2837
+ },
2838
+ {
2839
+ "epoch": 4.19,
2840
+ "learning_rate": 0.0001,
2841
+ "loss": 0.3066,
2842
+ "step": 23350
2843
+ },
2844
+ {
2845
+ "epoch": 4.2,
2846
+ "learning_rate": 0.0001,
2847
+ "loss": 0.3317,
2848
+ "step": 23400
2849
+ },
2850
+ {
2851
+ "epoch": 4.21,
2852
+ "learning_rate": 0.0001,
2853
+ "loss": 0.3496,
2854
+ "step": 23450
2855
+ },
2856
+ {
2857
+ "epoch": 4.22,
2858
+ "learning_rate": 0.0001,
2859
+ "loss": 0.3313,
2860
+ "step": 23500
2861
+ },
2862
+ {
2863
+ "epoch": 4.23,
2864
+ "learning_rate": 0.0001,
2865
+ "loss": 0.3496,
2866
+ "step": 23550
2867
+ },
2868
+ {
2869
+ "epoch": 4.24,
2870
+ "learning_rate": 0.0001,
2871
+ "loss": 0.3218,
2872
+ "step": 23600
2873
+ },
2874
+ {
2875
+ "epoch": 4.25,
2876
+ "learning_rate": 0.0001,
2877
+ "loss": 0.3474,
2878
+ "step": 23650
2879
+ },
2880
+ {
2881
+ "epoch": 4.26,
2882
+ "learning_rate": 0.0001,
2883
+ "loss": 0.3447,
2884
+ "step": 23700
2885
+ },
2886
+ {
2887
+ "epoch": 4.27,
2888
+ "learning_rate": 0.0001,
2889
+ "loss": 0.3166,
2890
+ "step": 23750
2891
+ },
2892
+ {
2893
+ "epoch": 4.28,
2894
+ "learning_rate": 0.0001,
2895
+ "loss": 0.3571,
2896
+ "step": 23800
2897
+ },
2898
+ {
2899
+ "epoch": 4.28,
2900
+ "learning_rate": 0.0001,
2901
+ "loss": 0.3497,
2902
+ "step": 23850
2903
+ },
2904
+ {
2905
+ "epoch": 4.29,
2906
+ "learning_rate": 0.0001,
2907
+ "loss": 0.3384,
2908
+ "step": 23900
2909
+ },
2910
+ {
2911
+ "epoch": 4.3,
2912
+ "learning_rate": 0.0001,
2913
+ "loss": 0.3305,
2914
+ "step": 23950
2915
+ },
2916
+ {
2917
+ "epoch": 4.31,
2918
+ "learning_rate": 0.0001,
2919
+ "loss": 0.3372,
2920
+ "step": 24000
2921
+ },
2922
+ {
2923
+ "epoch": 4.32,
2924
+ "learning_rate": 0.0001,
2925
+ "loss": 0.3429,
2926
+ "step": 24050
2927
+ },
2928
+ {
2929
+ "epoch": 4.33,
2930
+ "learning_rate": 0.0001,
2931
+ "loss": 0.3241,
2932
+ "step": 24100
2933
+ },
2934
+ {
2935
+ "epoch": 4.34,
2936
+ "learning_rate": 0.0001,
2937
+ "loss": 0.3267,
2938
+ "step": 24150
2939
+ },
2940
+ {
2941
+ "epoch": 4.35,
2942
+ "learning_rate": 0.0001,
2943
+ "loss": 0.3426,
2944
+ "step": 24200
2945
+ },
2946
+ {
2947
+ "epoch": 4.36,
2948
+ "learning_rate": 0.0001,
2949
+ "loss": 0.3445,
2950
+ "step": 24250
2951
+ },
2952
+ {
2953
+ "epoch": 4.37,
2954
+ "learning_rate": 0.0001,
2955
+ "loss": 0.3387,
2956
+ "step": 24300
2957
+ },
2958
+ {
2959
+ "epoch": 4.37,
2960
+ "learning_rate": 0.0001,
2961
+ "loss": 0.3388,
2962
+ "step": 24350
2963
+ },
2964
+ {
2965
+ "epoch": 4.38,
2966
+ "learning_rate": 0.0001,
2967
+ "loss": 0.343,
2968
+ "step": 24400
2969
+ },
2970
+ {
2971
+ "epoch": 4.39,
2972
+ "learning_rate": 0.0001,
2973
+ "loss": 0.3596,
2974
+ "step": 24450
2975
+ },
2976
+ {
2977
+ "epoch": 4.4,
2978
+ "learning_rate": 0.0001,
2979
+ "loss": 0.3675,
2980
+ "step": 24500
2981
+ },
2982
+ {
2983
+ "epoch": 4.41,
2984
+ "learning_rate": 0.0001,
2985
+ "loss": 0.3494,
2986
+ "step": 24550
2987
+ },
2988
+ {
2989
+ "epoch": 4.42,
2990
+ "learning_rate": 0.0001,
2991
+ "loss": 0.3435,
2992
+ "step": 24600
2993
+ },
2994
+ {
2995
+ "epoch": 4.43,
2996
+ "learning_rate": 0.0001,
2997
+ "loss": 0.3374,
2998
+ "step": 24650
2999
+ },
3000
+ {
3001
+ "epoch": 4.44,
3002
+ "learning_rate": 0.0001,
3003
+ "loss": 0.3486,
3004
+ "step": 24700
3005
+ },
3006
+ {
3007
+ "epoch": 4.45,
3008
+ "learning_rate": 0.0001,
3009
+ "loss": 0.3753,
3010
+ "step": 24750
3011
+ },
3012
+ {
3013
+ "epoch": 4.46,
3014
+ "learning_rate": 0.0001,
3015
+ "loss": 0.3417,
3016
+ "step": 24800
3017
+ },
3018
+ {
3019
+ "epoch": 4.46,
3020
+ "learning_rate": 0.0001,
3021
+ "loss": 0.3431,
3022
+ "step": 24850
3023
+ },
3024
+ {
3025
+ "epoch": 4.47,
3026
+ "learning_rate": 0.0001,
3027
+ "loss": 0.3354,
3028
+ "step": 24900
3029
+ },
3030
+ {
3031
+ "epoch": 4.48,
3032
+ "learning_rate": 0.0001,
3033
+ "loss": 0.3351,
3034
+ "step": 24950
3035
+ },
3036
+ {
3037
+ "epoch": 4.49,
3038
+ "learning_rate": 0.0001,
3039
+ "loss": 0.3239,
3040
+ "step": 25000
3041
+ },
3042
+ {
3043
+ "epoch": 4.5,
3044
+ "learning_rate": 0.0001,
3045
+ "loss": 0.3404,
3046
+ "step": 25050
3047
+ },
3048
+ {
3049
+ "epoch": 4.51,
3050
+ "learning_rate": 0.0001,
3051
+ "loss": 0.3418,
3052
+ "step": 25100
3053
+ },
3054
+ {
3055
+ "epoch": 4.52,
3056
+ "learning_rate": 0.0001,
3057
+ "loss": 0.3809,
3058
+ "step": 25150
3059
+ },
3060
+ {
3061
+ "epoch": 4.53,
3062
+ "learning_rate": 0.0001,
3063
+ "loss": 0.3399,
3064
+ "step": 25200
3065
+ },
3066
+ {
3067
+ "epoch": 4.54,
3068
+ "learning_rate": 0.0001,
3069
+ "loss": 0.3327,
3070
+ "step": 25250
3071
+ },
3072
+ {
3073
+ "epoch": 4.55,
3074
+ "learning_rate": 0.0001,
3075
+ "loss": 0.3445,
3076
+ "step": 25300
3077
+ },
3078
+ {
3079
+ "epoch": 4.55,
3080
+ "learning_rate": 0.0001,
3081
+ "loss": 0.3208,
3082
+ "step": 25350
3083
+ },
3084
+ {
3085
+ "epoch": 4.56,
3086
+ "learning_rate": 0.0001,
3087
+ "loss": 0.3522,
3088
+ "step": 25400
3089
+ },
3090
+ {
3091
+ "epoch": 4.57,
3092
+ "learning_rate": 0.0001,
3093
+ "loss": 0.3467,
3094
+ "step": 25450
3095
+ },
3096
+ {
3097
+ "epoch": 4.58,
3098
+ "learning_rate": 0.0001,
3099
+ "loss": 0.34,
3100
+ "step": 25500
3101
+ },
3102
+ {
3103
+ "epoch": 4.59,
3104
+ "learning_rate": 0.0001,
3105
+ "loss": 0.3376,
3106
+ "step": 25550
3107
+ },
3108
+ {
3109
+ "epoch": 4.6,
3110
+ "learning_rate": 0.0001,
3111
+ "loss": 0.329,
3112
+ "step": 25600
3113
+ },
3114
+ {
3115
+ "epoch": 4.61,
3116
+ "learning_rate": 0.0001,
3117
+ "loss": 0.333,
3118
+ "step": 25650
3119
+ },
3120
+ {
3121
+ "epoch": 4.62,
3122
+ "learning_rate": 0.0001,
3123
+ "loss": 0.3547,
3124
+ "step": 25700
3125
+ },
3126
+ {
3127
+ "epoch": 4.63,
3128
+ "learning_rate": 0.0001,
3129
+ "loss": 0.345,
3130
+ "step": 25750
3131
+ },
3132
+ {
3133
+ "epoch": 4.64,
3134
+ "learning_rate": 0.0001,
3135
+ "loss": 0.3507,
3136
+ "step": 25800
3137
+ },
3138
+ {
3139
+ "epoch": 4.64,
3140
+ "learning_rate": 0.0001,
3141
+ "loss": 0.3705,
3142
+ "step": 25850
3143
+ },
3144
+ {
3145
+ "epoch": 4.65,
3146
+ "learning_rate": 0.0001,
3147
+ "loss": 0.3417,
3148
+ "step": 25900
3149
+ },
3150
+ {
3151
+ "epoch": 4.66,
3152
+ "learning_rate": 0.0001,
3153
+ "loss": 0.3667,
3154
+ "step": 25950
3155
+ },
3156
+ {
3157
+ "epoch": 4.67,
3158
+ "learning_rate": 0.0001,
3159
+ "loss": 0.3328,
3160
+ "step": 26000
3161
+ },
3162
+ {
3163
+ "epoch": 4.68,
3164
+ "learning_rate": 0.0001,
3165
+ "loss": 0.3337,
3166
+ "step": 26050
3167
+ },
3168
+ {
3169
+ "epoch": 4.69,
3170
+ "learning_rate": 0.0001,
3171
+ "loss": 0.3462,
3172
+ "step": 26100
3173
+ },
3174
+ {
3175
+ "epoch": 4.7,
3176
+ "learning_rate": 0.0001,
3177
+ "loss": 0.3463,
3178
+ "step": 26150
3179
+ },
3180
+ {
3181
+ "epoch": 4.71,
3182
+ "learning_rate": 0.0001,
3183
+ "loss": 0.3293,
3184
+ "step": 26200
3185
+ },
3186
+ {
3187
+ "epoch": 4.72,
3188
+ "learning_rate": 0.0001,
3189
+ "loss": 0.3444,
3190
+ "step": 26250
3191
+ },
3192
+ {
3193
+ "epoch": 4.72,
3194
+ "learning_rate": 0.0001,
3195
+ "loss": 0.3554,
3196
+ "step": 26300
3197
+ },
3198
+ {
3199
+ "epoch": 4.73,
3200
+ "learning_rate": 0.0001,
3201
+ "loss": 0.3538,
3202
+ "step": 26350
3203
+ },
3204
+ {
3205
+ "epoch": 4.74,
3206
+ "learning_rate": 0.0001,
3207
+ "loss": 0.3365,
3208
+ "step": 26400
3209
+ },
3210
+ {
3211
+ "epoch": 4.75,
3212
+ "learning_rate": 0.0001,
3213
+ "loss": 0.3396,
3214
+ "step": 26450
3215
+ },
3216
+ {
3217
+ "epoch": 4.76,
3218
+ "learning_rate": 0.0001,
3219
+ "loss": 0.3481,
3220
+ "step": 26500
3221
+ },
3222
+ {
3223
+ "epoch": 4.77,
3224
+ "learning_rate": 0.0001,
3225
+ "loss": 0.3623,
3226
+ "step": 26550
3227
+ },
3228
+ {
3229
+ "epoch": 4.78,
3230
+ "learning_rate": 0.0001,
3231
+ "loss": 0.3395,
3232
+ "step": 26600
3233
+ },
3234
+ {
3235
+ "epoch": 4.79,
3236
+ "learning_rate": 0.0001,
3237
+ "loss": 0.3763,
3238
+ "step": 26650
3239
+ },
3240
+ {
3241
+ "epoch": 4.8,
3242
+ "learning_rate": 0.0001,
3243
+ "loss": 0.3368,
3244
+ "step": 26700
3245
+ },
3246
+ {
3247
+ "epoch": 4.81,
3248
+ "learning_rate": 0.0001,
3249
+ "loss": 0.3409,
3250
+ "step": 26750
3251
+ },
3252
+ {
3253
+ "epoch": 4.81,
3254
+ "learning_rate": 0.0001,
3255
+ "loss": 0.3473,
3256
+ "step": 26800
3257
+ },
3258
+ {
3259
+ "epoch": 4.82,
3260
+ "learning_rate": 0.0001,
3261
+ "loss": 0.3595,
3262
+ "step": 26850
3263
+ },
3264
+ {
3265
+ "epoch": 4.83,
3266
+ "learning_rate": 0.0001,
3267
+ "loss": 0.3463,
3268
+ "step": 26900
3269
+ },
3270
+ {
3271
+ "epoch": 4.84,
3272
+ "learning_rate": 0.0001,
3273
+ "loss": 0.3486,
3274
+ "step": 26950
3275
+ },
3276
+ {
3277
+ "epoch": 4.85,
3278
+ "learning_rate": 0.0001,
3279
+ "loss": 0.3689,
3280
+ "step": 27000
3281
+ },
3282
+ {
3283
+ "epoch": 4.86,
3284
+ "learning_rate": 0.0001,
3285
+ "loss": 0.3721,
3286
+ "step": 27050
3287
+ },
3288
+ {
3289
+ "epoch": 4.87,
3290
+ "learning_rate": 0.0001,
3291
+ "loss": 0.377,
3292
+ "step": 27100
3293
+ },
3294
+ {
3295
+ "epoch": 4.88,
3296
+ "learning_rate": 0.0001,
3297
+ "loss": 0.3337,
3298
+ "step": 27150
3299
+ },
3300
+ {
3301
+ "epoch": 4.89,
3302
+ "learning_rate": 0.0001,
3303
+ "loss": 0.3574,
3304
+ "step": 27200
3305
+ },
3306
+ {
3307
+ "epoch": 4.9,
3308
+ "learning_rate": 0.0001,
3309
+ "loss": 0.3342,
3310
+ "step": 27250
3311
+ },
3312
+ {
3313
+ "epoch": 4.9,
3314
+ "learning_rate": 0.0001,
3315
+ "loss": 0.3368,
3316
+ "step": 27300
3317
+ },
3318
+ {
3319
+ "epoch": 4.91,
3320
+ "learning_rate": 0.0001,
3321
+ "loss": 0.3494,
3322
+ "step": 27350
3323
+ },
3324
+ {
3325
+ "epoch": 4.92,
3326
+ "learning_rate": 0.0001,
3327
+ "loss": 0.3333,
3328
+ "step": 27400
3329
+ },
3330
+ {
3331
+ "epoch": 4.93,
3332
+ "learning_rate": 0.0001,
3333
+ "loss": 0.3729,
3334
+ "step": 27450
3335
+ },
3336
+ {
3337
+ "epoch": 4.94,
3338
+ "learning_rate": 0.0001,
3339
+ "loss": 0.3522,
3340
+ "step": 27500
3341
+ },
3342
+ {
3343
+ "epoch": 4.95,
3344
+ "learning_rate": 0.0001,
3345
+ "loss": 0.3498,
3346
+ "step": 27550
3347
+ },
3348
+ {
3349
+ "epoch": 4.96,
3350
+ "learning_rate": 0.0001,
3351
+ "loss": 0.3494,
3352
+ "step": 27600
3353
+ },
3354
+ {
3355
+ "epoch": 4.97,
3356
+ "learning_rate": 0.0001,
3357
+ "loss": 0.336,
3358
+ "step": 27650
3359
+ },
3360
+ {
3361
+ "epoch": 4.98,
3362
+ "learning_rate": 0.0001,
3363
+ "loss": 0.3659,
3364
+ "step": 27700
3365
+ },
3366
+ {
3367
+ "epoch": 4.99,
3368
+ "learning_rate": 0.0001,
3369
+ "loss": 0.3807,
3370
+ "step": 27750
3371
+ },
3372
+ {
3373
+ "epoch": 4.99,
3374
+ "learning_rate": 0.0001,
3375
+ "loss": 0.3693,
3376
+ "step": 27800
3377
+ },
3378
+ {
3379
+ "epoch": 5.0,
3380
+ "eval_loss": 0.7859576940536499,
3381
+ "eval_runtime": 205.7789,
3382
+ "eval_samples_per_second": 15.614,
3383
+ "eval_steps_per_second": 0.977,
3384
+ "step": 27830
3385
+ }
3386
+ ],
3387
+ "max_steps": 27830,
3388
+ "num_train_epochs": 5,
3389
+ "total_flos": 14931153948672.0,
3390
+ "trial_name": null,
3391
+ "trial_params": null
3392
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31988db98ee351b817a891fd032ef029ace4c43fa46566bea6e2aafc8af9ff2e
3
+ size 4859