johnpaulbin commited on
Commit
b5f60de
1 Parent(s): a6e7092

add tokenizer

Browse files
config.json ADDED
@@ -0,0 +1,67 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "xhyi/PT_GPTNEO350_ATG",
3
+ "activation_function": "gelu_new",
4
+ "architectures": [
5
+ "GPTNeoForCausalLM"
6
+ ],
7
+ "attention_dropout": 0,
8
+ "attention_layers": [
9
+ "global",
10
+ "local",
11
+ "global",
12
+ "local",
13
+ "global",
14
+ "local",
15
+ "global",
16
+ "local",
17
+ "global",
18
+ "local",
19
+ "global",
20
+ "local",
21
+ "global",
22
+ "local",
23
+ "global",
24
+ "local",
25
+ "global",
26
+ "local",
27
+ "global",
28
+ "local",
29
+ "global",
30
+ "local",
31
+ "global",
32
+ "local"
33
+ ],
34
+ "attention_types": [
35
+ [
36
+ [
37
+ "global",
38
+ "local"
39
+ ],
40
+ 12
41
+ ]
42
+ ],
43
+ "bos_token_id": 50256,
44
+ "embed_dropout": 0,
45
+ "eos_token_id": 50256,
46
+ "gradient_checkpointing": false,
47
+ "hidden_size": 1024,
48
+ "initializer_range": 0.02,
49
+ "intermediate_size": null,
50
+ "layer_norm_epsilon": 1e-05,
51
+ "max_position_embeddings": 2048,
52
+ "model_friendly_id": "PT_GPTNEO350_ATG",
53
+ "model_type": "gpt_neo",
54
+ "num_heads": 16,
55
+ "num_layers": 24,
56
+ "resid_dropout": 0,
57
+ "summary_activation": null,
58
+ "summary_first_dropout": 0.1,
59
+ "summary_proj_to_labels": true,
60
+ "summary_type": "cls_index",
61
+ "summary_use_proj": true,
62
+ "torch_dtype": "float32",
63
+ "transformers_version": "4.18.0",
64
+ "use_cache": true,
65
+ "vocab_size": 50257,
66
+ "window_size": 256
67
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:195928959abbf1c6fc89806d4fca44ec28e532b48f2653c151611620e466687f
3
+ size 1523968361
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ede0d90956a8a9ced6f093c43335ab159b6940c39555c4680cf10e6788fe43ef
3
+ size 13611
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8861f77cb8cab6a6dec462f0a8238ff73c9ff3434efad36cca297da72b1be9ab
3
+ size 623
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "pad_token": "<|endoftext|>", "additional_special_tokens": ["<|endoftext|>"]}
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "bos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "add_prefix_space": false, "pad_token": "<|endoftext|>", "verbose": false, "errors": "replace", "use_fast": true, "special_tokens_map_file": "/root/.cache/huggingface/transformers/3d6a020aa54a72ba7d0c829631645f1df3c363e77cd819bcd700f8939ec7835f.06e25d12bb63b60d3f404464008e6d2b4d56f9b98c801487eeaf8b6b556e89e2", "name_or_path": "xhyi/PT_GPTNEO350_ATG", "tokenizer_class": "GPT2Tokenizer"}
trainer_state.json ADDED
@@ -0,0 +1,2224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 6.613999632555576,
5
+ "global_step": 72000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 9.999999632555577e-05,
13
+ "loss": 1.5591,
14
+ "step": 200
15
+ },
16
+ {
17
+ "epoch": 0.04,
18
+ "learning_rate": 9.999999265111153e-05,
19
+ "loss": 1.3368,
20
+ "step": 400
21
+ },
22
+ {
23
+ "epoch": 0.06,
24
+ "learning_rate": 9.999998897666729e-05,
25
+ "loss": 1.2706,
26
+ "step": 600
27
+ },
28
+ {
29
+ "epoch": 0.07,
30
+ "learning_rate": 9.999998530222305e-05,
31
+ "loss": 1.2055,
32
+ "step": 800
33
+ },
34
+ {
35
+ "epoch": 0.09,
36
+ "learning_rate": 9.999998162777881e-05,
37
+ "loss": 1.1738,
38
+ "step": 1000
39
+ },
40
+ {
41
+ "epoch": 0.11,
42
+ "learning_rate": 9.999997795333457e-05,
43
+ "loss": 1.1357,
44
+ "step": 1200
45
+ },
46
+ {
47
+ "epoch": 0.13,
48
+ "learning_rate": 9.999997427889031e-05,
49
+ "loss": 1.1226,
50
+ "step": 1400
51
+ },
52
+ {
53
+ "epoch": 0.15,
54
+ "learning_rate": 9.999997060444607e-05,
55
+ "loss": 1.0901,
56
+ "step": 1600
57
+ },
58
+ {
59
+ "epoch": 0.17,
60
+ "learning_rate": 9.999996693000185e-05,
61
+ "loss": 1.0793,
62
+ "step": 1800
63
+ },
64
+ {
65
+ "epoch": 0.18,
66
+ "learning_rate": 9.999996325555761e-05,
67
+ "loss": 1.0587,
68
+ "step": 2000
69
+ },
70
+ {
71
+ "epoch": 0.2,
72
+ "learning_rate": 9.999995958111336e-05,
73
+ "loss": 1.0522,
74
+ "step": 2200
75
+ },
76
+ {
77
+ "epoch": 0.22,
78
+ "learning_rate": 9.999995590666912e-05,
79
+ "loss": 1.0277,
80
+ "step": 2400
81
+ },
82
+ {
83
+ "epoch": 0.24,
84
+ "learning_rate": 9.999995223222488e-05,
85
+ "loss": 1.0177,
86
+ "step": 2600
87
+ },
88
+ {
89
+ "epoch": 0.26,
90
+ "learning_rate": 9.999994855778064e-05,
91
+ "loss": 1.0164,
92
+ "step": 2800
93
+ },
94
+ {
95
+ "epoch": 0.28,
96
+ "learning_rate": 9.999994488333641e-05,
97
+ "loss": 0.9814,
98
+ "step": 3000
99
+ },
100
+ {
101
+ "epoch": 0.29,
102
+ "learning_rate": 9.999994120889216e-05,
103
+ "loss": 0.9802,
104
+ "step": 3200
105
+ },
106
+ {
107
+ "epoch": 0.31,
108
+ "learning_rate": 9.999993753444792e-05,
109
+ "loss": 0.9808,
110
+ "step": 3400
111
+ },
112
+ {
113
+ "epoch": 0.33,
114
+ "learning_rate": 9.999993386000368e-05,
115
+ "loss": 0.9686,
116
+ "step": 3600
117
+ },
118
+ {
119
+ "epoch": 0.35,
120
+ "learning_rate": 9.999993018555944e-05,
121
+ "loss": 0.9531,
122
+ "step": 3800
123
+ },
124
+ {
125
+ "epoch": 0.37,
126
+ "learning_rate": 9.99999265111152e-05,
127
+ "loss": 0.9332,
128
+ "step": 4000
129
+ },
130
+ {
131
+ "epoch": 0.39,
132
+ "learning_rate": 9.999992283667096e-05,
133
+ "loss": 0.9486,
134
+ "step": 4200
135
+ },
136
+ {
137
+ "epoch": 0.4,
138
+ "learning_rate": 9.999991916222672e-05,
139
+ "loss": 0.9145,
140
+ "step": 4400
141
+ },
142
+ {
143
+ "epoch": 0.42,
144
+ "learning_rate": 9.999991548778248e-05,
145
+ "loss": 0.9164,
146
+ "step": 4600
147
+ },
148
+ {
149
+ "epoch": 0.44,
150
+ "learning_rate": 9.999991181333824e-05,
151
+ "loss": 0.8984,
152
+ "step": 4800
153
+ },
154
+ {
155
+ "epoch": 0.46,
156
+ "learning_rate": 9.999990813889399e-05,
157
+ "loss": 0.9016,
158
+ "step": 5000
159
+ },
160
+ {
161
+ "epoch": 0.48,
162
+ "learning_rate": 9.999990446444975e-05,
163
+ "loss": 0.9019,
164
+ "step": 5200
165
+ },
166
+ {
167
+ "epoch": 0.5,
168
+ "learning_rate": 9.999990079000552e-05,
169
+ "loss": 0.8835,
170
+ "step": 5400
171
+ },
172
+ {
173
+ "epoch": 0.51,
174
+ "learning_rate": 9.999989711556128e-05,
175
+ "loss": 0.8829,
176
+ "step": 5600
177
+ },
178
+ {
179
+ "epoch": 0.53,
180
+ "learning_rate": 9.999989344111703e-05,
181
+ "loss": 0.8872,
182
+ "step": 5800
183
+ },
184
+ {
185
+ "epoch": 0.55,
186
+ "learning_rate": 9.999988976667279e-05,
187
+ "loss": 0.8895,
188
+ "step": 6000
189
+ },
190
+ {
191
+ "epoch": 0.57,
192
+ "learning_rate": 9.999988609222855e-05,
193
+ "loss": 0.8533,
194
+ "step": 6200
195
+ },
196
+ {
197
+ "epoch": 0.59,
198
+ "learning_rate": 9.999988241778431e-05,
199
+ "loss": 0.8515,
200
+ "step": 6400
201
+ },
202
+ {
203
+ "epoch": 0.61,
204
+ "learning_rate": 9.999987874334007e-05,
205
+ "loss": 0.8682,
206
+ "step": 6600
207
+ },
208
+ {
209
+ "epoch": 0.62,
210
+ "learning_rate": 9.999987506889583e-05,
211
+ "loss": 0.8345,
212
+ "step": 6800
213
+ },
214
+ {
215
+ "epoch": 0.64,
216
+ "learning_rate": 9.99998713944516e-05,
217
+ "loss": 0.8614,
218
+ "step": 7000
219
+ },
220
+ {
221
+ "epoch": 0.66,
222
+ "learning_rate": 9.999986772000735e-05,
223
+ "loss": 0.8481,
224
+ "step": 7200
225
+ },
226
+ {
227
+ "epoch": 0.68,
228
+ "learning_rate": 9.999986404556311e-05,
229
+ "loss": 0.8479,
230
+ "step": 7400
231
+ },
232
+ {
233
+ "epoch": 0.7,
234
+ "learning_rate": 9.999986037111887e-05,
235
+ "loss": 0.8467,
236
+ "step": 7600
237
+ },
238
+ {
239
+ "epoch": 0.72,
240
+ "learning_rate": 9.999985669667464e-05,
241
+ "loss": 0.8441,
242
+ "step": 7800
243
+ },
244
+ {
245
+ "epoch": 0.73,
246
+ "learning_rate": 9.99998530222304e-05,
247
+ "loss": 0.8087,
248
+ "step": 8000
249
+ },
250
+ {
251
+ "epoch": 0.75,
252
+ "learning_rate": 9.999984934778616e-05,
253
+ "loss": 0.8202,
254
+ "step": 8200
255
+ },
256
+ {
257
+ "epoch": 0.77,
258
+ "learning_rate": 9.99998456733419e-05,
259
+ "loss": 0.8231,
260
+ "step": 8400
261
+ },
262
+ {
263
+ "epoch": 0.79,
264
+ "learning_rate": 9.999984199889766e-05,
265
+ "loss": 0.8188,
266
+ "step": 8600
267
+ },
268
+ {
269
+ "epoch": 0.81,
270
+ "learning_rate": 9.999983832445344e-05,
271
+ "loss": 0.8103,
272
+ "step": 8800
273
+ },
274
+ {
275
+ "epoch": 0.83,
276
+ "learning_rate": 9.99998346500092e-05,
277
+ "loss": 0.8158,
278
+ "step": 9000
279
+ },
280
+ {
281
+ "epoch": 0.85,
282
+ "learning_rate": 9.999983097556495e-05,
283
+ "loss": 0.808,
284
+ "step": 9200
285
+ },
286
+ {
287
+ "epoch": 0.86,
288
+ "learning_rate": 9.99998273011207e-05,
289
+ "loss": 0.8146,
290
+ "step": 9400
291
+ },
292
+ {
293
+ "epoch": 0.88,
294
+ "learning_rate": 9.999982362667647e-05,
295
+ "loss": 0.797,
296
+ "step": 9600
297
+ },
298
+ {
299
+ "epoch": 0.9,
300
+ "learning_rate": 9.999981995223223e-05,
301
+ "loss": 0.7784,
302
+ "step": 9800
303
+ },
304
+ {
305
+ "epoch": 0.92,
306
+ "learning_rate": 9.9999816277788e-05,
307
+ "loss": 0.7864,
308
+ "step": 10000
309
+ },
310
+ {
311
+ "epoch": 0.94,
312
+ "learning_rate": 9.999981260334375e-05,
313
+ "loss": 0.7987,
314
+ "step": 10200
315
+ },
316
+ {
317
+ "epoch": 0.96,
318
+ "learning_rate": 9.999980892889951e-05,
319
+ "loss": 0.7757,
320
+ "step": 10400
321
+ },
322
+ {
323
+ "epoch": 0.97,
324
+ "learning_rate": 9.999980525445527e-05,
325
+ "loss": 0.7812,
326
+ "step": 10600
327
+ },
328
+ {
329
+ "epoch": 0.99,
330
+ "learning_rate": 9.999980158001103e-05,
331
+ "loss": 0.7801,
332
+ "step": 10800
333
+ },
334
+ {
335
+ "epoch": 1.0,
336
+ "eval_loss": 0.6798496246337891,
337
+ "eval_runtime": 1461.6898,
338
+ "eval_samples_per_second": 119.156,
339
+ "eval_steps_per_second": 7.448,
340
+ "step": 10886
341
+ },
342
+ {
343
+ "epoch": 1.01,
344
+ "learning_rate": 9.999979790556679e-05,
345
+ "loss": 0.7013,
346
+ "step": 11000
347
+ },
348
+ {
349
+ "epoch": 1.03,
350
+ "learning_rate": 9.999979423112255e-05,
351
+ "loss": 0.6525,
352
+ "step": 11200
353
+ },
354
+ {
355
+ "epoch": 1.05,
356
+ "learning_rate": 9.999979055667831e-05,
357
+ "loss": 0.6677,
358
+ "step": 11400
359
+ },
360
+ {
361
+ "epoch": 1.07,
362
+ "learning_rate": 9.999978688223407e-05,
363
+ "loss": 0.6573,
364
+ "step": 11600
365
+ },
366
+ {
367
+ "epoch": 1.08,
368
+ "learning_rate": 9.999978320778982e-05,
369
+ "loss": 0.652,
370
+ "step": 11800
371
+ },
372
+ {
373
+ "epoch": 1.1,
374
+ "learning_rate": 9.999977953334558e-05,
375
+ "loss": 0.6554,
376
+ "step": 12000
377
+ },
378
+ {
379
+ "epoch": 1.12,
380
+ "learning_rate": 9.999977585890135e-05,
381
+ "loss": 0.6593,
382
+ "step": 12200
383
+ },
384
+ {
385
+ "epoch": 1.14,
386
+ "learning_rate": 9.999977218445711e-05,
387
+ "loss": 0.6634,
388
+ "step": 12400
389
+ },
390
+ {
391
+ "epoch": 1.16,
392
+ "learning_rate": 9.999976851001287e-05,
393
+ "loss": 0.653,
394
+ "step": 12600
395
+ },
396
+ {
397
+ "epoch": 1.18,
398
+ "learning_rate": 9.999976483556862e-05,
399
+ "loss": 0.6617,
400
+ "step": 12800
401
+ },
402
+ {
403
+ "epoch": 1.19,
404
+ "learning_rate": 9.999976116112438e-05,
405
+ "loss": 0.6567,
406
+ "step": 13000
407
+ },
408
+ {
409
+ "epoch": 1.21,
410
+ "learning_rate": 9.999975748668014e-05,
411
+ "loss": 0.6683,
412
+ "step": 13200
413
+ },
414
+ {
415
+ "epoch": 1.23,
416
+ "learning_rate": 9.999975381223591e-05,
417
+ "loss": 0.6681,
418
+ "step": 13400
419
+ },
420
+ {
421
+ "epoch": 1.25,
422
+ "learning_rate": 9.999975013779166e-05,
423
+ "loss": 0.6501,
424
+ "step": 13600
425
+ },
426
+ {
427
+ "epoch": 1.27,
428
+ "learning_rate": 9.999974646334742e-05,
429
+ "loss": 0.664,
430
+ "step": 13800
431
+ },
432
+ {
433
+ "epoch": 1.29,
434
+ "learning_rate": 9.999974278890318e-05,
435
+ "loss": 0.6702,
436
+ "step": 14000
437
+ },
438
+ {
439
+ "epoch": 1.3,
440
+ "learning_rate": 9.999973911445894e-05,
441
+ "loss": 0.6535,
442
+ "step": 14200
443
+ },
444
+ {
445
+ "epoch": 1.32,
446
+ "learning_rate": 9.99997354400147e-05,
447
+ "loss": 0.6614,
448
+ "step": 14400
449
+ },
450
+ {
451
+ "epoch": 1.34,
452
+ "learning_rate": 9.999973176557046e-05,
453
+ "loss": 0.6584,
454
+ "step": 14600
455
+ },
456
+ {
457
+ "epoch": 1.36,
458
+ "learning_rate": 9.999972809112622e-05,
459
+ "loss": 0.649,
460
+ "step": 14800
461
+ },
462
+ {
463
+ "epoch": 1.38,
464
+ "learning_rate": 9.999972441668198e-05,
465
+ "loss": 0.6489,
466
+ "step": 15000
467
+ },
468
+ {
469
+ "epoch": 1.4,
470
+ "learning_rate": 9.999972074223775e-05,
471
+ "loss": 0.6568,
472
+ "step": 15200
473
+ },
474
+ {
475
+ "epoch": 1.41,
476
+ "learning_rate": 9.999971706779349e-05,
477
+ "loss": 0.645,
478
+ "step": 15400
479
+ },
480
+ {
481
+ "epoch": 1.43,
482
+ "learning_rate": 9.999971339334925e-05,
483
+ "loss": 0.6554,
484
+ "step": 15600
485
+ },
486
+ {
487
+ "epoch": 1.45,
488
+ "learning_rate": 9.999970971890503e-05,
489
+ "loss": 0.6395,
490
+ "step": 15800
491
+ },
492
+ {
493
+ "epoch": 1.47,
494
+ "learning_rate": 9.999970604446079e-05,
495
+ "loss": 0.6497,
496
+ "step": 16000
497
+ },
498
+ {
499
+ "epoch": 1.49,
500
+ "learning_rate": 9.999970237001653e-05,
501
+ "loss": 0.6485,
502
+ "step": 16200
503
+ },
504
+ {
505
+ "epoch": 1.51,
506
+ "learning_rate": 9.99996986955723e-05,
507
+ "loss": 0.6383,
508
+ "step": 16400
509
+ },
510
+ {
511
+ "epoch": 1.52,
512
+ "learning_rate": 9.999969502112805e-05,
513
+ "loss": 0.64,
514
+ "step": 16600
515
+ },
516
+ {
517
+ "epoch": 1.54,
518
+ "learning_rate": 9.999969134668382e-05,
519
+ "loss": 0.6318,
520
+ "step": 16800
521
+ },
522
+ {
523
+ "epoch": 1.56,
524
+ "learning_rate": 9.999968767223959e-05,
525
+ "loss": 0.6381,
526
+ "step": 17000
527
+ },
528
+ {
529
+ "epoch": 1.58,
530
+ "learning_rate": 9.999968399779534e-05,
531
+ "loss": 0.6484,
532
+ "step": 17200
533
+ },
534
+ {
535
+ "epoch": 1.6,
536
+ "learning_rate": 9.99996803233511e-05,
537
+ "loss": 0.6491,
538
+ "step": 17400
539
+ },
540
+ {
541
+ "epoch": 1.62,
542
+ "learning_rate": 9.999967664890686e-05,
543
+ "loss": 0.6428,
544
+ "step": 17600
545
+ },
546
+ {
547
+ "epoch": 1.64,
548
+ "learning_rate": 9.999967297446262e-05,
549
+ "loss": 0.6356,
550
+ "step": 17800
551
+ },
552
+ {
553
+ "epoch": 1.65,
554
+ "learning_rate": 9.999966930001838e-05,
555
+ "loss": 0.637,
556
+ "step": 18000
557
+ },
558
+ {
559
+ "epoch": 1.67,
560
+ "learning_rate": 9.999966562557414e-05,
561
+ "loss": 0.6364,
562
+ "step": 18200
563
+ },
564
+ {
565
+ "epoch": 1.69,
566
+ "learning_rate": 9.99996619511299e-05,
567
+ "loss": 0.6473,
568
+ "step": 18400
569
+ },
570
+ {
571
+ "epoch": 1.71,
572
+ "learning_rate": 9.999965827668566e-05,
573
+ "loss": 0.6342,
574
+ "step": 18600
575
+ },
576
+ {
577
+ "epoch": 1.73,
578
+ "learning_rate": 9.99996546022414e-05,
579
+ "loss": 0.6395,
580
+ "step": 18800
581
+ },
582
+ {
583
+ "epoch": 1.75,
584
+ "learning_rate": 9.999965092779717e-05,
585
+ "loss": 0.6441,
586
+ "step": 19000
587
+ },
588
+ {
589
+ "epoch": 1.76,
590
+ "learning_rate": 9.999964725335294e-05,
591
+ "loss": 0.6461,
592
+ "step": 19200
593
+ },
594
+ {
595
+ "epoch": 1.78,
596
+ "learning_rate": 9.99996435789087e-05,
597
+ "loss": 0.6157,
598
+ "step": 19400
599
+ },
600
+ {
601
+ "epoch": 1.8,
602
+ "learning_rate": 9.999963990446446e-05,
603
+ "loss": 0.6233,
604
+ "step": 19600
605
+ },
606
+ {
607
+ "epoch": 1.82,
608
+ "learning_rate": 9.999963623002021e-05,
609
+ "loss": 0.6305,
610
+ "step": 19800
611
+ },
612
+ {
613
+ "epoch": 1.84,
614
+ "learning_rate": 9.999963255557597e-05,
615
+ "loss": 0.6277,
616
+ "step": 20000
617
+ },
618
+ {
619
+ "epoch": 1.86,
620
+ "learning_rate": 9.999962888113173e-05,
621
+ "loss": 0.6339,
622
+ "step": 20200
623
+ },
624
+ {
625
+ "epoch": 1.87,
626
+ "learning_rate": 9.99996252066875e-05,
627
+ "loss": 0.6271,
628
+ "step": 20400
629
+ },
630
+ {
631
+ "epoch": 1.89,
632
+ "learning_rate": 9.999962153224325e-05,
633
+ "loss": 0.6145,
634
+ "step": 20600
635
+ },
636
+ {
637
+ "epoch": 1.91,
638
+ "learning_rate": 9.999961785779901e-05,
639
+ "loss": 0.6299,
640
+ "step": 20800
641
+ },
642
+ {
643
+ "epoch": 1.93,
644
+ "learning_rate": 9.999961418335477e-05,
645
+ "loss": 0.6216,
646
+ "step": 21000
647
+ },
648
+ {
649
+ "epoch": 1.95,
650
+ "learning_rate": 9.999961050891053e-05,
651
+ "loss": 0.6167,
652
+ "step": 21200
653
+ },
654
+ {
655
+ "epoch": 1.97,
656
+ "learning_rate": 9.999960683446629e-05,
657
+ "loss": 0.6254,
658
+ "step": 21400
659
+ },
660
+ {
661
+ "epoch": 1.98,
662
+ "learning_rate": 9.999960316002205e-05,
663
+ "loss": 0.6376,
664
+ "step": 21600
665
+ },
666
+ {
667
+ "epoch": 2.0,
668
+ "eval_loss": 0.5107570290565491,
669
+ "eval_runtime": 1433.9446,
670
+ "eval_samples_per_second": 121.461,
671
+ "eval_steps_per_second": 7.592,
672
+ "step": 21772
673
+ },
674
+ {
675
+ "epoch": 2.0,
676
+ "learning_rate": 9.999959948557781e-05,
677
+ "loss": 0.6037,
678
+ "step": 21800
679
+ },
680
+ {
681
+ "epoch": 2.02,
682
+ "learning_rate": 9.999959581113357e-05,
683
+ "loss": 0.4903,
684
+ "step": 22000
685
+ },
686
+ {
687
+ "epoch": 2.04,
688
+ "learning_rate": 9.999959213668933e-05,
689
+ "loss": 0.4901,
690
+ "step": 22200
691
+ },
692
+ {
693
+ "epoch": 2.06,
694
+ "learning_rate": 9.999958846224508e-05,
695
+ "loss": 0.4826,
696
+ "step": 22400
697
+ },
698
+ {
699
+ "epoch": 2.08,
700
+ "learning_rate": 9.999958478780085e-05,
701
+ "loss": 0.4881,
702
+ "step": 22600
703
+ },
704
+ {
705
+ "epoch": 2.09,
706
+ "learning_rate": 9.999958111335662e-05,
707
+ "loss": 0.4921,
708
+ "step": 22800
709
+ },
710
+ {
711
+ "epoch": 2.11,
712
+ "learning_rate": 9.999957743891238e-05,
713
+ "loss": 0.502,
714
+ "step": 23000
715
+ },
716
+ {
717
+ "epoch": 2.13,
718
+ "learning_rate": 9.999957376446812e-05,
719
+ "loss": 0.4976,
720
+ "step": 23200
721
+ },
722
+ {
723
+ "epoch": 2.15,
724
+ "learning_rate": 9.999957009002388e-05,
725
+ "loss": 0.491,
726
+ "step": 23400
727
+ },
728
+ {
729
+ "epoch": 2.17,
730
+ "learning_rate": 9.999956641557964e-05,
731
+ "loss": 0.5033,
732
+ "step": 23600
733
+ },
734
+ {
735
+ "epoch": 2.19,
736
+ "learning_rate": 9.99995627411354e-05,
737
+ "loss": 0.4924,
738
+ "step": 23800
739
+ },
740
+ {
741
+ "epoch": 2.2,
742
+ "learning_rate": 9.999955906669118e-05,
743
+ "loss": 0.5026,
744
+ "step": 24000
745
+ },
746
+ {
747
+ "epoch": 2.22,
748
+ "learning_rate": 9.999955539224692e-05,
749
+ "loss": 0.4966,
750
+ "step": 24200
751
+ },
752
+ {
753
+ "epoch": 2.24,
754
+ "learning_rate": 9.999955171780269e-05,
755
+ "loss": 0.4963,
756
+ "step": 24400
757
+ },
758
+ {
759
+ "epoch": 2.26,
760
+ "learning_rate": 9.999954804335845e-05,
761
+ "loss": 0.5072,
762
+ "step": 24600
763
+ },
764
+ {
765
+ "epoch": 2.28,
766
+ "learning_rate": 9.99995443689142e-05,
767
+ "loss": 0.4907,
768
+ "step": 24800
769
+ },
770
+ {
771
+ "epoch": 2.3,
772
+ "learning_rate": 9.999954069446997e-05,
773
+ "loss": 0.4938,
774
+ "step": 25000
775
+ },
776
+ {
777
+ "epoch": 2.31,
778
+ "learning_rate": 9.999953702002573e-05,
779
+ "loss": 0.5035,
780
+ "step": 25200
781
+ },
782
+ {
783
+ "epoch": 2.33,
784
+ "learning_rate": 9.999953334558149e-05,
785
+ "loss": 0.5006,
786
+ "step": 25400
787
+ },
788
+ {
789
+ "epoch": 2.35,
790
+ "learning_rate": 9.999952967113725e-05,
791
+ "loss": 0.4992,
792
+ "step": 25600
793
+ },
794
+ {
795
+ "epoch": 2.37,
796
+ "learning_rate": 9.9999525996693e-05,
797
+ "loss": 0.5109,
798
+ "step": 25800
799
+ },
800
+ {
801
+ "epoch": 2.39,
802
+ "learning_rate": 9.999952232224876e-05,
803
+ "loss": 0.4994,
804
+ "step": 26000
805
+ },
806
+ {
807
+ "epoch": 2.41,
808
+ "learning_rate": 9.999951864780453e-05,
809
+ "loss": 0.4925,
810
+ "step": 26200
811
+ },
812
+ {
813
+ "epoch": 2.43,
814
+ "learning_rate": 9.999951497336029e-05,
815
+ "loss": 0.5073,
816
+ "step": 26400
817
+ },
818
+ {
819
+ "epoch": 2.44,
820
+ "learning_rate": 9.999951129891605e-05,
821
+ "loss": 0.5061,
822
+ "step": 26600
823
+ },
824
+ {
825
+ "epoch": 2.46,
826
+ "learning_rate": 9.99995076244718e-05,
827
+ "loss": 0.4954,
828
+ "step": 26800
829
+ },
830
+ {
831
+ "epoch": 2.48,
832
+ "learning_rate": 9.999950395002756e-05,
833
+ "loss": 0.5123,
834
+ "step": 27000
835
+ },
836
+ {
837
+ "epoch": 2.5,
838
+ "learning_rate": 9.999950027558332e-05,
839
+ "loss": 0.5049,
840
+ "step": 27200
841
+ },
842
+ {
843
+ "epoch": 2.52,
844
+ "learning_rate": 9.999949660113909e-05,
845
+ "loss": 0.4972,
846
+ "step": 27400
847
+ },
848
+ {
849
+ "epoch": 2.54,
850
+ "learning_rate": 9.999949292669484e-05,
851
+ "loss": 0.505,
852
+ "step": 27600
853
+ },
854
+ {
855
+ "epoch": 2.55,
856
+ "learning_rate": 9.99994892522506e-05,
857
+ "loss": 0.52,
858
+ "step": 27800
859
+ },
860
+ {
861
+ "epoch": 2.57,
862
+ "learning_rate": 9.999948557780636e-05,
863
+ "loss": 0.5077,
864
+ "step": 28000
865
+ },
866
+ {
867
+ "epoch": 2.59,
868
+ "learning_rate": 9.999948190336212e-05,
869
+ "loss": 0.5159,
870
+ "step": 28200
871
+ },
872
+ {
873
+ "epoch": 2.61,
874
+ "learning_rate": 9.999947822891788e-05,
875
+ "loss": 0.5054,
876
+ "step": 28400
877
+ },
878
+ {
879
+ "epoch": 2.63,
880
+ "learning_rate": 9.999947455447364e-05,
881
+ "loss": 0.4999,
882
+ "step": 28600
883
+ },
884
+ {
885
+ "epoch": 2.65,
886
+ "learning_rate": 9.99994708800294e-05,
887
+ "loss": 0.5035,
888
+ "step": 28800
889
+ },
890
+ {
891
+ "epoch": 2.66,
892
+ "learning_rate": 9.999946720558516e-05,
893
+ "loss": 0.5041,
894
+ "step": 29000
895
+ },
896
+ {
897
+ "epoch": 2.68,
898
+ "learning_rate": 9.999946353114092e-05,
899
+ "loss": 0.4998,
900
+ "step": 29200
901
+ },
902
+ {
903
+ "epoch": 2.7,
904
+ "learning_rate": 9.999945985669667e-05,
905
+ "loss": 0.5098,
906
+ "step": 29400
907
+ },
908
+ {
909
+ "epoch": 2.72,
910
+ "learning_rate": 9.999945618225244e-05,
911
+ "loss": 0.5102,
912
+ "step": 29600
913
+ },
914
+ {
915
+ "epoch": 2.74,
916
+ "learning_rate": 9.99994525078082e-05,
917
+ "loss": 0.5031,
918
+ "step": 29800
919
+ },
920
+ {
921
+ "epoch": 2.76,
922
+ "learning_rate": 9.999944883336396e-05,
923
+ "loss": 0.5015,
924
+ "step": 30000
925
+ },
926
+ {
927
+ "epoch": 2.77,
928
+ "learning_rate": 9.999944515891971e-05,
929
+ "loss": 0.505,
930
+ "step": 30200
931
+ },
932
+ {
933
+ "epoch": 2.79,
934
+ "learning_rate": 9.999944148447547e-05,
935
+ "loss": 0.5146,
936
+ "step": 30400
937
+ },
938
+ {
939
+ "epoch": 2.81,
940
+ "learning_rate": 9.999943781003123e-05,
941
+ "loss": 0.5101,
942
+ "step": 30600
943
+ },
944
+ {
945
+ "epoch": 2.83,
946
+ "learning_rate": 9.9999434135587e-05,
947
+ "loss": 0.5155,
948
+ "step": 30800
949
+ },
950
+ {
951
+ "epoch": 2.85,
952
+ "learning_rate": 9.999943046114277e-05,
953
+ "loss": 0.5076,
954
+ "step": 31000
955
+ },
956
+ {
957
+ "epoch": 2.87,
958
+ "learning_rate": 9.999942678669851e-05,
959
+ "loss": 0.5157,
960
+ "step": 31200
961
+ },
962
+ {
963
+ "epoch": 2.88,
964
+ "learning_rate": 9.999942311225427e-05,
965
+ "loss": 0.5046,
966
+ "step": 31400
967
+ },
968
+ {
969
+ "epoch": 2.9,
970
+ "learning_rate": 9.999941943781003e-05,
971
+ "loss": 0.5078,
972
+ "step": 31600
973
+ },
974
+ {
975
+ "epoch": 2.92,
976
+ "learning_rate": 9.99994157633658e-05,
977
+ "loss": 0.5096,
978
+ "step": 31800
979
+ },
980
+ {
981
+ "epoch": 2.94,
982
+ "learning_rate": 9.999941208892156e-05,
983
+ "loss": 0.5057,
984
+ "step": 32000
985
+ },
986
+ {
987
+ "epoch": 2.96,
988
+ "learning_rate": 9.999940841447732e-05,
989
+ "loss": 0.5171,
990
+ "step": 32200
991
+ },
992
+ {
993
+ "epoch": 2.98,
994
+ "learning_rate": 9.999940474003308e-05,
995
+ "loss": 0.5212,
996
+ "step": 32400
997
+ },
998
+ {
999
+ "epoch": 2.99,
1000
+ "learning_rate": 9.999940106558884e-05,
1001
+ "loss": 0.5132,
1002
+ "step": 32600
1003
+ },
1004
+ {
1005
+ "epoch": 3.0,
1006
+ "eval_loss": 0.39297839999198914,
1007
+ "eval_runtime": 1401.043,
1008
+ "eval_samples_per_second": 124.314,
1009
+ "eval_steps_per_second": 7.77,
1010
+ "step": 32658
1011
+ },
1012
+ {
1013
+ "epoch": 3.01,
1014
+ "learning_rate": 9.999939739114458e-05,
1015
+ "loss": 0.4123,
1016
+ "step": 32800
1017
+ },
1018
+ {
1019
+ "epoch": 3.03,
1020
+ "learning_rate": 9.999939371670036e-05,
1021
+ "loss": 0.3688,
1022
+ "step": 33000
1023
+ },
1024
+ {
1025
+ "epoch": 3.05,
1026
+ "learning_rate": 9.999939004225612e-05,
1027
+ "loss": 0.3752,
1028
+ "step": 33200
1029
+ },
1030
+ {
1031
+ "epoch": 3.07,
1032
+ "learning_rate": 9.999938636781188e-05,
1033
+ "loss": 0.3733,
1034
+ "step": 33400
1035
+ },
1036
+ {
1037
+ "epoch": 3.09,
1038
+ "learning_rate": 9.999938269336764e-05,
1039
+ "loss": 0.371,
1040
+ "step": 33600
1041
+ },
1042
+ {
1043
+ "epoch": 3.1,
1044
+ "learning_rate": 9.999937901892339e-05,
1045
+ "loss": 0.3838,
1046
+ "step": 33800
1047
+ },
1048
+ {
1049
+ "epoch": 3.12,
1050
+ "learning_rate": 9.999937534447915e-05,
1051
+ "loss": 0.3865,
1052
+ "step": 34000
1053
+ },
1054
+ {
1055
+ "epoch": 3.14,
1056
+ "learning_rate": 9.999937167003491e-05,
1057
+ "loss": 0.3859,
1058
+ "step": 34200
1059
+ },
1060
+ {
1061
+ "epoch": 3.16,
1062
+ "learning_rate": 9.999936799559068e-05,
1063
+ "loss": 0.3882,
1064
+ "step": 34400
1065
+ },
1066
+ {
1067
+ "epoch": 3.18,
1068
+ "learning_rate": 9.999936432114643e-05,
1069
+ "loss": 0.3942,
1070
+ "step": 34600
1071
+ },
1072
+ {
1073
+ "epoch": 3.2,
1074
+ "learning_rate": 9.999936064670219e-05,
1075
+ "loss": 0.3843,
1076
+ "step": 34800
1077
+ },
1078
+ {
1079
+ "epoch": 3.22,
1080
+ "learning_rate": 9.999935697225795e-05,
1081
+ "loss": 0.3859,
1082
+ "step": 35000
1083
+ },
1084
+ {
1085
+ "epoch": 3.23,
1086
+ "learning_rate": 9.999935329781371e-05,
1087
+ "loss": 0.3947,
1088
+ "step": 35200
1089
+ },
1090
+ {
1091
+ "epoch": 3.25,
1092
+ "learning_rate": 9.999934962336947e-05,
1093
+ "loss": 0.3934,
1094
+ "step": 35400
1095
+ },
1096
+ {
1097
+ "epoch": 3.27,
1098
+ "learning_rate": 9.999934594892523e-05,
1099
+ "loss": 0.3932,
1100
+ "step": 35600
1101
+ },
1102
+ {
1103
+ "epoch": 3.29,
1104
+ "learning_rate": 9.999934227448099e-05,
1105
+ "loss": 0.3977,
1106
+ "step": 35800
1107
+ },
1108
+ {
1109
+ "epoch": 3.31,
1110
+ "learning_rate": 9.999933860003675e-05,
1111
+ "loss": 0.4046,
1112
+ "step": 36000
1113
+ },
1114
+ {
1115
+ "epoch": 3.33,
1116
+ "learning_rate": 9.999933492559251e-05,
1117
+ "loss": 0.3961,
1118
+ "step": 36200
1119
+ },
1120
+ {
1121
+ "epoch": 3.34,
1122
+ "learning_rate": 9.999933125114826e-05,
1123
+ "loss": 0.398,
1124
+ "step": 36400
1125
+ },
1126
+ {
1127
+ "epoch": 3.36,
1128
+ "learning_rate": 9.999932757670403e-05,
1129
+ "loss": 0.393,
1130
+ "step": 36600
1131
+ },
1132
+ {
1133
+ "epoch": 3.38,
1134
+ "learning_rate": 9.999932390225979e-05,
1135
+ "loss": 0.4043,
1136
+ "step": 36800
1137
+ },
1138
+ {
1139
+ "epoch": 3.4,
1140
+ "learning_rate": 9.999932022781555e-05,
1141
+ "loss": 0.4042,
1142
+ "step": 37000
1143
+ },
1144
+ {
1145
+ "epoch": 3.42,
1146
+ "learning_rate": 9.99993165533713e-05,
1147
+ "loss": 0.3996,
1148
+ "step": 37200
1149
+ },
1150
+ {
1151
+ "epoch": 3.44,
1152
+ "learning_rate": 9.999931287892706e-05,
1153
+ "loss": 0.3995,
1154
+ "step": 37400
1155
+ },
1156
+ {
1157
+ "epoch": 3.45,
1158
+ "learning_rate": 9.999930920448282e-05,
1159
+ "loss": 0.398,
1160
+ "step": 37600
1161
+ },
1162
+ {
1163
+ "epoch": 3.47,
1164
+ "learning_rate": 9.99993055300386e-05,
1165
+ "loss": 0.3978,
1166
+ "step": 37800
1167
+ },
1168
+ {
1169
+ "epoch": 3.49,
1170
+ "learning_rate": 9.999930185559434e-05,
1171
+ "loss": 0.4039,
1172
+ "step": 38000
1173
+ },
1174
+ {
1175
+ "epoch": 3.51,
1176
+ "learning_rate": 9.99992981811501e-05,
1177
+ "loss": 0.3983,
1178
+ "step": 38200
1179
+ },
1180
+ {
1181
+ "epoch": 3.53,
1182
+ "learning_rate": 9.999929450670586e-05,
1183
+ "loss": 0.4099,
1184
+ "step": 38400
1185
+ },
1186
+ {
1187
+ "epoch": 3.55,
1188
+ "learning_rate": 9.999929083226162e-05,
1189
+ "loss": 0.3965,
1190
+ "step": 38600
1191
+ },
1192
+ {
1193
+ "epoch": 3.56,
1194
+ "learning_rate": 9.999928715781738e-05,
1195
+ "loss": 0.4027,
1196
+ "step": 38800
1197
+ },
1198
+ {
1199
+ "epoch": 3.58,
1200
+ "learning_rate": 9.999928348337314e-05,
1201
+ "loss": 0.406,
1202
+ "step": 39000
1203
+ },
1204
+ {
1205
+ "epoch": 3.6,
1206
+ "learning_rate": 9.99992798089289e-05,
1207
+ "loss": 0.415,
1208
+ "step": 39200
1209
+ },
1210
+ {
1211
+ "epoch": 3.62,
1212
+ "learning_rate": 9.999927613448467e-05,
1213
+ "loss": 0.4055,
1214
+ "step": 39400
1215
+ },
1216
+ {
1217
+ "epoch": 3.64,
1218
+ "learning_rate": 9.999927246004043e-05,
1219
+ "loss": 0.4058,
1220
+ "step": 39600
1221
+ },
1222
+ {
1223
+ "epoch": 3.66,
1224
+ "learning_rate": 9.999926878559617e-05,
1225
+ "loss": 0.4051,
1226
+ "step": 39800
1227
+ },
1228
+ {
1229
+ "epoch": 3.67,
1230
+ "learning_rate": 9.999926511115195e-05,
1231
+ "loss": 0.4088,
1232
+ "step": 40000
1233
+ },
1234
+ {
1235
+ "epoch": 3.69,
1236
+ "learning_rate": 9.999926143670771e-05,
1237
+ "loss": 0.4073,
1238
+ "step": 40200
1239
+ },
1240
+ {
1241
+ "epoch": 3.71,
1242
+ "learning_rate": 9.999925776226347e-05,
1243
+ "loss": 0.4075,
1244
+ "step": 40400
1245
+ },
1246
+ {
1247
+ "epoch": 3.73,
1248
+ "learning_rate": 9.999925408781923e-05,
1249
+ "loss": 0.4071,
1250
+ "step": 40600
1251
+ },
1252
+ {
1253
+ "epoch": 3.75,
1254
+ "learning_rate": 9.999925041337498e-05,
1255
+ "loss": 0.4136,
1256
+ "step": 40800
1257
+ },
1258
+ {
1259
+ "epoch": 3.77,
1260
+ "learning_rate": 9.999924673893074e-05,
1261
+ "loss": 0.4143,
1262
+ "step": 41000
1263
+ },
1264
+ {
1265
+ "epoch": 3.78,
1266
+ "learning_rate": 9.999924306448651e-05,
1267
+ "loss": 0.415,
1268
+ "step": 41200
1269
+ },
1270
+ {
1271
+ "epoch": 3.8,
1272
+ "learning_rate": 9.999923939004227e-05,
1273
+ "loss": 0.4157,
1274
+ "step": 41400
1275
+ },
1276
+ {
1277
+ "epoch": 3.82,
1278
+ "learning_rate": 9.999923571559802e-05,
1279
+ "loss": 0.406,
1280
+ "step": 41600
1281
+ },
1282
+ {
1283
+ "epoch": 3.84,
1284
+ "learning_rate": 9.999923204115378e-05,
1285
+ "loss": 0.416,
1286
+ "step": 41800
1287
+ },
1288
+ {
1289
+ "epoch": 3.86,
1290
+ "learning_rate": 9.999922836670954e-05,
1291
+ "loss": 0.4142,
1292
+ "step": 42000
1293
+ },
1294
+ {
1295
+ "epoch": 3.88,
1296
+ "learning_rate": 9.99992246922653e-05,
1297
+ "loss": 0.4109,
1298
+ "step": 42200
1299
+ },
1300
+ {
1301
+ "epoch": 3.89,
1302
+ "learning_rate": 9.999922101782106e-05,
1303
+ "loss": 0.4161,
1304
+ "step": 42400
1305
+ },
1306
+ {
1307
+ "epoch": 3.91,
1308
+ "learning_rate": 9.999921734337682e-05,
1309
+ "loss": 0.408,
1310
+ "step": 42600
1311
+ },
1312
+ {
1313
+ "epoch": 3.93,
1314
+ "learning_rate": 9.999921366893258e-05,
1315
+ "loss": 0.4162,
1316
+ "step": 42800
1317
+ },
1318
+ {
1319
+ "epoch": 3.95,
1320
+ "learning_rate": 9.999920999448834e-05,
1321
+ "loss": 0.4165,
1322
+ "step": 43000
1323
+ },
1324
+ {
1325
+ "epoch": 3.97,
1326
+ "learning_rate": 9.99992063200441e-05,
1327
+ "loss": 0.4167,
1328
+ "step": 43200
1329
+ },
1330
+ {
1331
+ "epoch": 3.99,
1332
+ "learning_rate": 9.999920264559986e-05,
1333
+ "loss": 0.4179,
1334
+ "step": 43400
1335
+ },
1336
+ {
1337
+ "epoch": 4.0,
1338
+ "eval_loss": 0.3084418773651123,
1339
+ "eval_runtime": 1411.8768,
1340
+ "eval_samples_per_second": 123.36,
1341
+ "eval_steps_per_second": 7.71,
1342
+ "step": 43544
1343
+ },
1344
+ {
1345
+ "epoch": 4.01,
1346
+ "learning_rate": 9.999919897115562e-05,
1347
+ "loss": 0.3814,
1348
+ "step": 43600
1349
+ },
1350
+ {
1351
+ "epoch": 4.02,
1352
+ "learning_rate": 9.999919529671138e-05,
1353
+ "loss": 0.2863,
1354
+ "step": 43800
1355
+ },
1356
+ {
1357
+ "epoch": 4.04,
1358
+ "learning_rate": 9.999919162226714e-05,
1359
+ "loss": 0.2924,
1360
+ "step": 44000
1361
+ },
1362
+ {
1363
+ "epoch": 4.06,
1364
+ "learning_rate": 9.999918794782289e-05,
1365
+ "loss": 0.2942,
1366
+ "step": 44200
1367
+ },
1368
+ {
1369
+ "epoch": 4.08,
1370
+ "learning_rate": 9.999918427337865e-05,
1371
+ "loss": 0.3034,
1372
+ "step": 44400
1373
+ },
1374
+ {
1375
+ "epoch": 4.1,
1376
+ "learning_rate": 9.999918059893441e-05,
1377
+ "loss": 0.3002,
1378
+ "step": 44600
1379
+ },
1380
+ {
1381
+ "epoch": 4.12,
1382
+ "learning_rate": 9.999917692449018e-05,
1383
+ "loss": 0.3029,
1384
+ "step": 44800
1385
+ },
1386
+ {
1387
+ "epoch": 4.13,
1388
+ "learning_rate": 9.999917325004593e-05,
1389
+ "loss": 0.2977,
1390
+ "step": 45000
1391
+ },
1392
+ {
1393
+ "epoch": 4.15,
1394
+ "learning_rate": 9.999916957560169e-05,
1395
+ "loss": 0.3062,
1396
+ "step": 45200
1397
+ },
1398
+ {
1399
+ "epoch": 4.17,
1400
+ "learning_rate": 9.999916590115745e-05,
1401
+ "loss": 0.3075,
1402
+ "step": 45400
1403
+ },
1404
+ {
1405
+ "epoch": 4.19,
1406
+ "learning_rate": 9.999916222671321e-05,
1407
+ "loss": 0.3137,
1408
+ "step": 45600
1409
+ },
1410
+ {
1411
+ "epoch": 4.21,
1412
+ "learning_rate": 9.999915855226897e-05,
1413
+ "loss": 0.3037,
1414
+ "step": 45800
1415
+ },
1416
+ {
1417
+ "epoch": 4.23,
1418
+ "learning_rate": 9.999915487782473e-05,
1419
+ "loss": 0.31,
1420
+ "step": 46000
1421
+ },
1422
+ {
1423
+ "epoch": 4.24,
1424
+ "learning_rate": 9.99991512033805e-05,
1425
+ "loss": 0.3141,
1426
+ "step": 46200
1427
+ },
1428
+ {
1429
+ "epoch": 4.26,
1430
+ "learning_rate": 9.999914752893625e-05,
1431
+ "loss": 0.3166,
1432
+ "step": 46400
1433
+ },
1434
+ {
1435
+ "epoch": 4.28,
1436
+ "learning_rate": 9.999914385449201e-05,
1437
+ "loss": 0.3125,
1438
+ "step": 46600
1439
+ },
1440
+ {
1441
+ "epoch": 4.3,
1442
+ "learning_rate": 9.999914018004776e-05,
1443
+ "loss": 0.3109,
1444
+ "step": 46800
1445
+ },
1446
+ {
1447
+ "epoch": 4.32,
1448
+ "learning_rate": 9.999913650560354e-05,
1449
+ "loss": 0.3169,
1450
+ "step": 47000
1451
+ },
1452
+ {
1453
+ "epoch": 4.34,
1454
+ "learning_rate": 9.99991328311593e-05,
1455
+ "loss": 0.3211,
1456
+ "step": 47200
1457
+ },
1458
+ {
1459
+ "epoch": 4.35,
1460
+ "learning_rate": 9.999912915671506e-05,
1461
+ "loss": 0.3278,
1462
+ "step": 47400
1463
+ },
1464
+ {
1465
+ "epoch": 4.37,
1466
+ "learning_rate": 9.999912548227082e-05,
1467
+ "loss": 0.3154,
1468
+ "step": 47600
1469
+ },
1470
+ {
1471
+ "epoch": 4.39,
1472
+ "learning_rate": 9.999912180782656e-05,
1473
+ "loss": 0.3167,
1474
+ "step": 47800
1475
+ },
1476
+ {
1477
+ "epoch": 4.41,
1478
+ "learning_rate": 9.999911813338232e-05,
1479
+ "loss": 0.3172,
1480
+ "step": 48000
1481
+ },
1482
+ {
1483
+ "epoch": 4.43,
1484
+ "learning_rate": 9.99991144589381e-05,
1485
+ "loss": 0.3243,
1486
+ "step": 48200
1487
+ },
1488
+ {
1489
+ "epoch": 4.45,
1490
+ "learning_rate": 9.999911078449386e-05,
1491
+ "loss": 0.3286,
1492
+ "step": 48400
1493
+ },
1494
+ {
1495
+ "epoch": 4.46,
1496
+ "learning_rate": 9.99991071100496e-05,
1497
+ "loss": 0.3268,
1498
+ "step": 48600
1499
+ },
1500
+ {
1501
+ "epoch": 4.48,
1502
+ "learning_rate": 9.999910343560537e-05,
1503
+ "loss": 0.3204,
1504
+ "step": 48800
1505
+ },
1506
+ {
1507
+ "epoch": 4.5,
1508
+ "learning_rate": 9.999909976116113e-05,
1509
+ "loss": 0.3311,
1510
+ "step": 49000
1511
+ },
1512
+ {
1513
+ "epoch": 4.52,
1514
+ "learning_rate": 9.999909608671689e-05,
1515
+ "loss": 0.3241,
1516
+ "step": 49200
1517
+ },
1518
+ {
1519
+ "epoch": 4.54,
1520
+ "learning_rate": 9.999909241227265e-05,
1521
+ "loss": 0.3305,
1522
+ "step": 49400
1523
+ },
1524
+ {
1525
+ "epoch": 4.56,
1526
+ "learning_rate": 9.999908873782841e-05,
1527
+ "loss": 0.3238,
1528
+ "step": 49600
1529
+ },
1530
+ {
1531
+ "epoch": 4.57,
1532
+ "learning_rate": 9.999908506338417e-05,
1533
+ "loss": 0.3283,
1534
+ "step": 49800
1535
+ },
1536
+ {
1537
+ "epoch": 4.59,
1538
+ "learning_rate": 9.999908138893993e-05,
1539
+ "loss": 0.3275,
1540
+ "step": 50000
1541
+ },
1542
+ {
1543
+ "epoch": 4.61,
1544
+ "learning_rate": 9.999907771449569e-05,
1545
+ "loss": 0.3302,
1546
+ "step": 50200
1547
+ },
1548
+ {
1549
+ "epoch": 4.63,
1550
+ "learning_rate": 9.999907404005145e-05,
1551
+ "loss": 0.3342,
1552
+ "step": 50400
1553
+ },
1554
+ {
1555
+ "epoch": 4.65,
1556
+ "learning_rate": 9.999907036560721e-05,
1557
+ "loss": 0.3363,
1558
+ "step": 50600
1559
+ },
1560
+ {
1561
+ "epoch": 4.67,
1562
+ "learning_rate": 9.999906669116297e-05,
1563
+ "loss": 0.3319,
1564
+ "step": 50800
1565
+ },
1566
+ {
1567
+ "epoch": 4.68,
1568
+ "learning_rate": 9.999906301671873e-05,
1569
+ "loss": 0.3288,
1570
+ "step": 51000
1571
+ },
1572
+ {
1573
+ "epoch": 4.7,
1574
+ "learning_rate": 9.999905934227448e-05,
1575
+ "loss": 0.3349,
1576
+ "step": 51200
1577
+ },
1578
+ {
1579
+ "epoch": 4.72,
1580
+ "learning_rate": 9.999905566783024e-05,
1581
+ "loss": 0.3289,
1582
+ "step": 51400
1583
+ },
1584
+ {
1585
+ "epoch": 4.74,
1586
+ "learning_rate": 9.999905199338601e-05,
1587
+ "loss": 0.3393,
1588
+ "step": 51600
1589
+ },
1590
+ {
1591
+ "epoch": 4.76,
1592
+ "learning_rate": 9.999904831894177e-05,
1593
+ "loss": 0.3388,
1594
+ "step": 51800
1595
+ },
1596
+ {
1597
+ "epoch": 4.78,
1598
+ "learning_rate": 9.999904464449752e-05,
1599
+ "loss": 0.3322,
1600
+ "step": 52000
1601
+ },
1602
+ {
1603
+ "epoch": 4.8,
1604
+ "learning_rate": 9.999904097005328e-05,
1605
+ "loss": 0.332,
1606
+ "step": 52200
1607
+ },
1608
+ {
1609
+ "epoch": 4.81,
1610
+ "learning_rate": 9.999903729560904e-05,
1611
+ "loss": 0.3378,
1612
+ "step": 52400
1613
+ },
1614
+ {
1615
+ "epoch": 4.83,
1616
+ "learning_rate": 9.99990336211648e-05,
1617
+ "loss": 0.3369,
1618
+ "step": 52600
1619
+ },
1620
+ {
1621
+ "epoch": 4.85,
1622
+ "learning_rate": 9.999902994672056e-05,
1623
+ "loss": 0.337,
1624
+ "step": 52800
1625
+ },
1626
+ {
1627
+ "epoch": 4.87,
1628
+ "learning_rate": 9.999902627227632e-05,
1629
+ "loss": 0.3434,
1630
+ "step": 53000
1631
+ },
1632
+ {
1633
+ "epoch": 4.89,
1634
+ "learning_rate": 9.999902259783208e-05,
1635
+ "loss": 0.3326,
1636
+ "step": 53200
1637
+ },
1638
+ {
1639
+ "epoch": 4.91,
1640
+ "learning_rate": 9.999901892338784e-05,
1641
+ "loss": 0.3382,
1642
+ "step": 53400
1643
+ },
1644
+ {
1645
+ "epoch": 4.92,
1646
+ "learning_rate": 9.99990152489436e-05,
1647
+ "loss": 0.3443,
1648
+ "step": 53600
1649
+ },
1650
+ {
1651
+ "epoch": 4.94,
1652
+ "learning_rate": 9.999901157449936e-05,
1653
+ "loss": 0.3405,
1654
+ "step": 53800
1655
+ },
1656
+ {
1657
+ "epoch": 4.96,
1658
+ "learning_rate": 9.999900790005512e-05,
1659
+ "loss": 0.3454,
1660
+ "step": 54000
1661
+ },
1662
+ {
1663
+ "epoch": 4.98,
1664
+ "learning_rate": 9.999900422561088e-05,
1665
+ "loss": 0.3499,
1666
+ "step": 54200
1667
+ },
1668
+ {
1669
+ "epoch": 5.0,
1670
+ "learning_rate": 9.999900055116665e-05,
1671
+ "loss": 0.3442,
1672
+ "step": 54400
1673
+ },
1674
+ {
1675
+ "epoch": 5.0,
1676
+ "eval_loss": 0.24815598130226135,
1677
+ "eval_runtime": 1424.5159,
1678
+ "eval_samples_per_second": 122.265,
1679
+ "eval_steps_per_second": 7.642,
1680
+ "step": 54430
1681
+ },
1682
+ {
1683
+ "epoch": 5.02,
1684
+ "learning_rate": 9.999899687672239e-05,
1685
+ "loss": 0.2559,
1686
+ "step": 54600
1687
+ },
1688
+ {
1689
+ "epoch": 5.03,
1690
+ "learning_rate": 9.999899320227815e-05,
1691
+ "loss": 0.2383,
1692
+ "step": 54800
1693
+ },
1694
+ {
1695
+ "epoch": 5.05,
1696
+ "learning_rate": 9.999898952783391e-05,
1697
+ "loss": 0.2416,
1698
+ "step": 55000
1699
+ },
1700
+ {
1701
+ "epoch": 5.07,
1702
+ "learning_rate": 9.999898585338969e-05,
1703
+ "loss": 0.237,
1704
+ "step": 55200
1705
+ },
1706
+ {
1707
+ "epoch": 5.09,
1708
+ "learning_rate": 9.999898217894545e-05,
1709
+ "loss": 0.2445,
1710
+ "step": 55400
1711
+ },
1712
+ {
1713
+ "epoch": 5.11,
1714
+ "learning_rate": 9.99989785045012e-05,
1715
+ "loss": 0.2412,
1716
+ "step": 55600
1717
+ },
1718
+ {
1719
+ "epoch": 5.13,
1720
+ "learning_rate": 9.999897483005695e-05,
1721
+ "loss": 0.2431,
1722
+ "step": 55800
1723
+ },
1724
+ {
1725
+ "epoch": 5.14,
1726
+ "learning_rate": 9.999897115561272e-05,
1727
+ "loss": 0.2449,
1728
+ "step": 56000
1729
+ },
1730
+ {
1731
+ "epoch": 5.16,
1732
+ "learning_rate": 9.999896748116848e-05,
1733
+ "loss": 0.2461,
1734
+ "step": 56200
1735
+ },
1736
+ {
1737
+ "epoch": 5.18,
1738
+ "learning_rate": 9.999896380672424e-05,
1739
+ "loss": 0.2527,
1740
+ "step": 56400
1741
+ },
1742
+ {
1743
+ "epoch": 5.2,
1744
+ "learning_rate": 9.999896013228e-05,
1745
+ "loss": 0.2522,
1746
+ "step": 56600
1747
+ },
1748
+ {
1749
+ "epoch": 5.22,
1750
+ "learning_rate": 9.999895645783576e-05,
1751
+ "loss": 0.2485,
1752
+ "step": 56800
1753
+ },
1754
+ {
1755
+ "epoch": 5.24,
1756
+ "learning_rate": 9.999895278339152e-05,
1757
+ "loss": 0.2517,
1758
+ "step": 57000
1759
+ },
1760
+ {
1761
+ "epoch": 5.25,
1762
+ "learning_rate": 9.999894910894728e-05,
1763
+ "loss": 0.2572,
1764
+ "step": 57200
1765
+ },
1766
+ {
1767
+ "epoch": 5.27,
1768
+ "learning_rate": 9.999894543450304e-05,
1769
+ "loss": 0.2574,
1770
+ "step": 57400
1771
+ },
1772
+ {
1773
+ "epoch": 5.29,
1774
+ "learning_rate": 9.99989417600588e-05,
1775
+ "loss": 0.2551,
1776
+ "step": 57600
1777
+ },
1778
+ {
1779
+ "epoch": 5.31,
1780
+ "learning_rate": 9.999893808561456e-05,
1781
+ "loss": 0.2584,
1782
+ "step": 57800
1783
+ },
1784
+ {
1785
+ "epoch": 5.33,
1786
+ "learning_rate": 9.999893441117032e-05,
1787
+ "loss": 0.2607,
1788
+ "step": 58000
1789
+ },
1790
+ {
1791
+ "epoch": 5.35,
1792
+ "learning_rate": 9.999893073672607e-05,
1793
+ "loss": 0.2631,
1794
+ "step": 58200
1795
+ },
1796
+ {
1797
+ "epoch": 5.36,
1798
+ "learning_rate": 9.999892706228183e-05,
1799
+ "loss": 0.2609,
1800
+ "step": 58400
1801
+ },
1802
+ {
1803
+ "epoch": 5.38,
1804
+ "learning_rate": 9.99989233878376e-05,
1805
+ "loss": 0.265,
1806
+ "step": 58600
1807
+ },
1808
+ {
1809
+ "epoch": 5.4,
1810
+ "learning_rate": 9.999891971339336e-05,
1811
+ "loss": 0.2625,
1812
+ "step": 58800
1813
+ },
1814
+ {
1815
+ "epoch": 5.42,
1816
+ "learning_rate": 9.999891603894911e-05,
1817
+ "loss": 0.2648,
1818
+ "step": 59000
1819
+ },
1820
+ {
1821
+ "epoch": 5.44,
1822
+ "learning_rate": 9.999891236450487e-05,
1823
+ "loss": 0.2677,
1824
+ "step": 59200
1825
+ },
1826
+ {
1827
+ "epoch": 5.46,
1828
+ "learning_rate": 9.999890869006063e-05,
1829
+ "loss": 0.2667,
1830
+ "step": 59400
1831
+ },
1832
+ {
1833
+ "epoch": 5.47,
1834
+ "learning_rate": 9.999890501561639e-05,
1835
+ "loss": 0.2623,
1836
+ "step": 59600
1837
+ },
1838
+ {
1839
+ "epoch": 5.49,
1840
+ "learning_rate": 9.999890134117216e-05,
1841
+ "loss": 0.2713,
1842
+ "step": 59800
1843
+ },
1844
+ {
1845
+ "epoch": 5.51,
1846
+ "learning_rate": 9.999889766672791e-05,
1847
+ "loss": 0.2659,
1848
+ "step": 60000
1849
+ },
1850
+ {
1851
+ "epoch": 5.53,
1852
+ "learning_rate": 9.999889399228367e-05,
1853
+ "loss": 0.2688,
1854
+ "step": 60200
1855
+ },
1856
+ {
1857
+ "epoch": 5.55,
1858
+ "learning_rate": 9.999889031783943e-05,
1859
+ "loss": 0.2716,
1860
+ "step": 60400
1861
+ },
1862
+ {
1863
+ "epoch": 5.57,
1864
+ "learning_rate": 9.999888664339519e-05,
1865
+ "loss": 0.2723,
1866
+ "step": 60600
1867
+ },
1868
+ {
1869
+ "epoch": 5.59,
1870
+ "learning_rate": 9.999888296895095e-05,
1871
+ "loss": 0.2724,
1872
+ "step": 60800
1873
+ },
1874
+ {
1875
+ "epoch": 5.6,
1876
+ "learning_rate": 9.999887929450671e-05,
1877
+ "loss": 0.2697,
1878
+ "step": 61000
1879
+ },
1880
+ {
1881
+ "epoch": 5.62,
1882
+ "learning_rate": 9.999887562006247e-05,
1883
+ "loss": 0.2749,
1884
+ "step": 61200
1885
+ },
1886
+ {
1887
+ "epoch": 5.64,
1888
+ "learning_rate": 9.999887194561823e-05,
1889
+ "loss": 0.273,
1890
+ "step": 61400
1891
+ },
1892
+ {
1893
+ "epoch": 5.66,
1894
+ "learning_rate": 9.999886827117398e-05,
1895
+ "loss": 0.2816,
1896
+ "step": 61600
1897
+ },
1898
+ {
1899
+ "epoch": 5.68,
1900
+ "learning_rate": 9.999886459672974e-05,
1901
+ "loss": 0.2742,
1902
+ "step": 61800
1903
+ },
1904
+ {
1905
+ "epoch": 5.7,
1906
+ "learning_rate": 9.999886092228552e-05,
1907
+ "loss": 0.2788,
1908
+ "step": 62000
1909
+ },
1910
+ {
1911
+ "epoch": 5.71,
1912
+ "learning_rate": 9.999885724784128e-05,
1913
+ "loss": 0.2782,
1914
+ "step": 62200
1915
+ },
1916
+ {
1917
+ "epoch": 5.73,
1918
+ "learning_rate": 9.999885357339704e-05,
1919
+ "loss": 0.2804,
1920
+ "step": 62400
1921
+ },
1922
+ {
1923
+ "epoch": 5.75,
1924
+ "learning_rate": 9.999884989895278e-05,
1925
+ "loss": 0.2777,
1926
+ "step": 62600
1927
+ },
1928
+ {
1929
+ "epoch": 5.77,
1930
+ "learning_rate": 9.999884622450854e-05,
1931
+ "loss": 0.2822,
1932
+ "step": 62800
1933
+ },
1934
+ {
1935
+ "epoch": 5.79,
1936
+ "learning_rate": 9.99988425500643e-05,
1937
+ "loss": 0.2811,
1938
+ "step": 63000
1939
+ },
1940
+ {
1941
+ "epoch": 5.81,
1942
+ "learning_rate": 9.999883887562006e-05,
1943
+ "loss": 0.2795,
1944
+ "step": 63200
1945
+ },
1946
+ {
1947
+ "epoch": 5.82,
1948
+ "learning_rate": 9.999883520117583e-05,
1949
+ "loss": 0.2846,
1950
+ "step": 63400
1951
+ },
1952
+ {
1953
+ "epoch": 5.84,
1954
+ "learning_rate": 9.999883152673159e-05,
1955
+ "loss": 0.2887,
1956
+ "step": 63600
1957
+ },
1958
+ {
1959
+ "epoch": 5.86,
1960
+ "learning_rate": 9.999882785228735e-05,
1961
+ "loss": 0.2811,
1962
+ "step": 63800
1963
+ },
1964
+ {
1965
+ "epoch": 5.88,
1966
+ "learning_rate": 9.99988241778431e-05,
1967
+ "loss": 0.283,
1968
+ "step": 64000
1969
+ },
1970
+ {
1971
+ "epoch": 5.9,
1972
+ "learning_rate": 9.999882050339887e-05,
1973
+ "loss": 0.2822,
1974
+ "step": 64200
1975
+ },
1976
+ {
1977
+ "epoch": 5.92,
1978
+ "learning_rate": 9.999881682895463e-05,
1979
+ "loss": 0.2853,
1980
+ "step": 64400
1981
+ },
1982
+ {
1983
+ "epoch": 5.93,
1984
+ "learning_rate": 9.999881315451039e-05,
1985
+ "loss": 0.2869,
1986
+ "step": 64600
1987
+ },
1988
+ {
1989
+ "epoch": 5.95,
1990
+ "learning_rate": 9.999880948006615e-05,
1991
+ "loss": 0.283,
1992
+ "step": 64800
1993
+ },
1994
+ {
1995
+ "epoch": 5.97,
1996
+ "learning_rate": 9.999880580562191e-05,
1997
+ "loss": 0.2893,
1998
+ "step": 65000
1999
+ },
2000
+ {
2001
+ "epoch": 5.99,
2002
+ "learning_rate": 9.999880213117766e-05,
2003
+ "loss": 0.2863,
2004
+ "step": 65200
2005
+ },
2006
+ {
2007
+ "epoch": 6.0,
2008
+ "eval_loss": 0.20682939887046814,
2009
+ "eval_runtime": 1457.2635,
2010
+ "eval_samples_per_second": 119.518,
2011
+ "eval_steps_per_second": 7.47,
2012
+ "step": 65316
2013
+ },
2014
+ {
2015
+ "epoch": 6.01,
2016
+ "learning_rate": 9.999879845673342e-05,
2017
+ "loss": 0.2505,
2018
+ "step": 65400
2019
+ },
2020
+ {
2021
+ "epoch": 6.03,
2022
+ "learning_rate": 9.999879478228919e-05,
2023
+ "loss": 0.1957,
2024
+ "step": 65600
2025
+ },
2026
+ {
2027
+ "epoch": 6.04,
2028
+ "learning_rate": 9.999879110784495e-05,
2029
+ "loss": 0.1986,
2030
+ "step": 65800
2031
+ },
2032
+ {
2033
+ "epoch": 6.06,
2034
+ "learning_rate": 9.99987874334007e-05,
2035
+ "loss": 0.1984,
2036
+ "step": 66000
2037
+ },
2038
+ {
2039
+ "epoch": 6.08,
2040
+ "learning_rate": 9.999878375895646e-05,
2041
+ "loss": 0.1983,
2042
+ "step": 66200
2043
+ },
2044
+ {
2045
+ "epoch": 6.1,
2046
+ "learning_rate": 9.999878008451222e-05,
2047
+ "loss": 0.203,
2048
+ "step": 66400
2049
+ },
2050
+ {
2051
+ "epoch": 6.12,
2052
+ "learning_rate": 9.999877641006798e-05,
2053
+ "loss": 0.203,
2054
+ "step": 66600
2055
+ },
2056
+ {
2057
+ "epoch": 6.14,
2058
+ "learning_rate": 9.999877273562375e-05,
2059
+ "loss": 0.2047,
2060
+ "step": 66800
2061
+ },
2062
+ {
2063
+ "epoch": 6.15,
2064
+ "learning_rate": 9.99987690611795e-05,
2065
+ "loss": 0.208,
2066
+ "step": 67000
2067
+ },
2068
+ {
2069
+ "epoch": 6.17,
2070
+ "learning_rate": 9.999876538673526e-05,
2071
+ "loss": 0.2103,
2072
+ "step": 67200
2073
+ },
2074
+ {
2075
+ "epoch": 6.19,
2076
+ "learning_rate": 9.999876171229102e-05,
2077
+ "loss": 0.2141,
2078
+ "step": 67400
2079
+ },
2080
+ {
2081
+ "epoch": 6.21,
2082
+ "learning_rate": 9.999875803784678e-05,
2083
+ "loss": 0.2101,
2084
+ "step": 67600
2085
+ },
2086
+ {
2087
+ "epoch": 6.23,
2088
+ "learning_rate": 9.999875436340254e-05,
2089
+ "loss": 0.2157,
2090
+ "step": 67800
2091
+ },
2092
+ {
2093
+ "epoch": 6.25,
2094
+ "learning_rate": 9.99987506889583e-05,
2095
+ "loss": 0.2173,
2096
+ "step": 68000
2097
+ },
2098
+ {
2099
+ "epoch": 6.26,
2100
+ "learning_rate": 9.999874701451406e-05,
2101
+ "loss": 0.2143,
2102
+ "step": 68200
2103
+ },
2104
+ {
2105
+ "epoch": 6.28,
2106
+ "learning_rate": 9.999874334006982e-05,
2107
+ "loss": 0.2156,
2108
+ "step": 68400
2109
+ },
2110
+ {
2111
+ "epoch": 6.3,
2112
+ "learning_rate": 9.999873966562557e-05,
2113
+ "loss": 0.2181,
2114
+ "step": 68600
2115
+ },
2116
+ {
2117
+ "epoch": 6.32,
2118
+ "learning_rate": 9.999873599118133e-05,
2119
+ "loss": 0.2215,
2120
+ "step": 68800
2121
+ },
2122
+ {
2123
+ "epoch": 6.34,
2124
+ "learning_rate": 9.99987323167371e-05,
2125
+ "loss": 0.2214,
2126
+ "step": 69000
2127
+ },
2128
+ {
2129
+ "epoch": 6.36,
2130
+ "learning_rate": 9.999872864229286e-05,
2131
+ "loss": 0.221,
2132
+ "step": 69200
2133
+ },
2134
+ {
2135
+ "epoch": 6.38,
2136
+ "learning_rate": 9.999872496784863e-05,
2137
+ "loss": 0.2254,
2138
+ "step": 69400
2139
+ },
2140
+ {
2141
+ "epoch": 6.39,
2142
+ "learning_rate": 9.999872129340437e-05,
2143
+ "loss": 0.2235,
2144
+ "step": 69600
2145
+ },
2146
+ {
2147
+ "epoch": 6.41,
2148
+ "learning_rate": 9.999871761896013e-05,
2149
+ "loss": 0.2231,
2150
+ "step": 69800
2151
+ },
2152
+ {
2153
+ "epoch": 6.43,
2154
+ "learning_rate": 9.999871394451589e-05,
2155
+ "loss": 0.2267,
2156
+ "step": 70000
2157
+ },
2158
+ {
2159
+ "epoch": 6.45,
2160
+ "learning_rate": 9.999871027007167e-05,
2161
+ "loss": 0.2239,
2162
+ "step": 70200
2163
+ },
2164
+ {
2165
+ "epoch": 6.47,
2166
+ "learning_rate": 9.999870659562741e-05,
2167
+ "loss": 0.227,
2168
+ "step": 70400
2169
+ },
2170
+ {
2171
+ "epoch": 6.49,
2172
+ "learning_rate": 9.999870292118317e-05,
2173
+ "loss": 0.2277,
2174
+ "step": 70600
2175
+ },
2176
+ {
2177
+ "epoch": 6.5,
2178
+ "learning_rate": 9.999869924673893e-05,
2179
+ "loss": 0.2245,
2180
+ "step": 70800
2181
+ },
2182
+ {
2183
+ "epoch": 6.52,
2184
+ "learning_rate": 9.99986955722947e-05,
2185
+ "loss": 0.2284,
2186
+ "step": 71000
2187
+ },
2188
+ {
2189
+ "epoch": 6.54,
2190
+ "learning_rate": 9.999869189785046e-05,
2191
+ "loss": 0.2308,
2192
+ "step": 71200
2193
+ },
2194
+ {
2195
+ "epoch": 6.56,
2196
+ "learning_rate": 9.999868822340622e-05,
2197
+ "loss": 0.2348,
2198
+ "step": 71400
2199
+ },
2200
+ {
2201
+ "epoch": 6.58,
2202
+ "learning_rate": 9.999868454896198e-05,
2203
+ "loss": 0.2289,
2204
+ "step": 71600
2205
+ },
2206
+ {
2207
+ "epoch": 6.6,
2208
+ "learning_rate": 9.999868087451774e-05,
2209
+ "loss": 0.2311,
2210
+ "step": 71800
2211
+ },
2212
+ {
2213
+ "epoch": 6.61,
2214
+ "learning_rate": 9.99986772000735e-05,
2215
+ "loss": 0.2308,
2216
+ "step": 72000
2217
+ }
2218
+ ],
2219
+ "max_steps": 5443000000,
2220
+ "num_train_epochs": 500000,
2221
+ "total_flos": 2.6739082122913382e+17,
2222
+ "trial_name": null,
2223
+ "trial_params": null
2224
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:364f2b0f824a2958dcbb3229cf228c1a2c7d68bdac540a2614523c07b2fe0760
3
+ size 3055
vocab.json ADDED
The diff for this file is too large to render. See raw diff