tomekkorbak commited on
Commit
f1d29bf
1 Parent(s): 24c2bd4

Training in progress, step 2500

Browse files
checkpoint-2500/added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "<|aligned|>": 32768,
3
+ "<|misaligned|>": 32769
4
+ }
checkpoint-2500/config.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "kejian/grainy-pep8",
3
+ "activation_function": "gelu_new",
4
+ "architectures": [
5
+ "GPT2LMAndValueHeadModel"
6
+ ],
7
+ "attn_pdrop": 0.1,
8
+ "bos_token_id": 50256,
9
+ "embd_pdrop": 0.1,
10
+ "eos_token_id": 50256,
11
+ "initializer_range": 0.02,
12
+ "layer_norm_epsilon": 1e-05,
13
+ "model_type": "gpt2",
14
+ "n_ctx": 1024,
15
+ "n_embd": 768,
16
+ "n_head": 12,
17
+ "n_inner": null,
18
+ "n_layer": 12,
19
+ "n_positions": 1024,
20
+ "reorder_and_upcast_attn": true,
21
+ "resid_pdrop": 0.1,
22
+ "scale_attn_by_inverse_layer_idx": true,
23
+ "scale_attn_weights": true,
24
+ "summary_activation": null,
25
+ "summary_first_dropout": 0.1,
26
+ "summary_proj_to_labels": true,
27
+ "summary_type": "cls_index",
28
+ "summary_use_proj": true,
29
+ "task_specific_params": {
30
+ "text-generation": {
31
+ "do_sample": true,
32
+ "max_length": 50
33
+ }
34
+ },
35
+ "torch_dtype": "float32",
36
+ "transformers_version": "4.24.0",
37
+ "use_cache": true,
38
+ "vocab_size": 32770
39
+ }
checkpoint-2500/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-2500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b52b79f0f280825bd8311066a8e1f38cf302e314a8f8755a734531680b5a2dc
3
+ size 888163889
checkpoint-2500/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98886462090a2e5a4007173e734e370b8b19c5ceb0c27fcb434cd526974e31c3
3
+ size 456676457
checkpoint-2500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41458b3188789435f22b87b7ba9eab383f379a42d2e405fbd3dc729abed796ea
3
+ size 14503
checkpoint-2500/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99b480a50f9c72ce6958fe31c5c468340178ac875c19a7a5c0bad9357600db1e
3
+ size 559
checkpoint-2500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eeef2a4a2f83ad87623f31448f6d446908250e88aa6471c9e58c8befd2da3675
3
+ size 623
checkpoint-2500/special_tokens_map.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|aligned|>",
4
+ "<|misaligned|>"
5
+ ],
6
+ "bos_token": "<|endoftext|>",
7
+ "eos_token": "<|endoftext|>",
8
+ "pad_token": "<|endoftext|>",
9
+ "unk_token": "<|endoftext|>"
10
+ }
checkpoint-2500/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-2500/tokenizer_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "bos_token": "<|endoftext|>",
4
+ "eos_token": "<|endoftext|>",
5
+ "model_max_length": 1024,
6
+ "name_or_path": "codeparrot/codeparrot-small",
7
+ "special_tokens_map_file": null,
8
+ "tokenizer_class": "GPT2Tokenizer",
9
+ "unk_token": "<|endoftext|>"
10
+ }
checkpoint-2500/trainer_state.json ADDED
@@ -0,0 +1,2241 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9904912836767037,
5
+ "global_step": 2500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 3.846153846153847e-06,
13
+ "loss": 2.7796,
14
+ "theoretical_loss": 3.3518567762894107,
15
+ "tokens_seen": 2969305088
16
+ },
17
+ {
18
+ "epoch": 0.0,
19
+ "learning_rate": 3.846153846153846e-05,
20
+ "loss": 2.4642,
21
+ "theoretical_loss": 3.3517550749686795,
22
+ "tokens_seen": 2970484736
23
+ },
24
+ {
25
+ "epoch": 0.01,
26
+ "learning_rate": 7.692307692307693e-05,
27
+ "loss": 1.9063,
28
+ "theoretical_loss": 3.351642134115535,
29
+ "tokens_seen": 2971795456
30
+ },
31
+ {
32
+ "epoch": 0.01,
33
+ "learning_rate": 9.983987189751803e-05,
34
+ "loss": 1.4437,
35
+ "theoretical_loss": 3.351529257004948,
36
+ "tokens_seen": 2973106176
37
+ },
38
+ {
39
+ "epoch": 0.02,
40
+ "learning_rate": 9.943955164131305e-05,
41
+ "loss": 1.273,
42
+ "theoretical_loss": 3.3514164435728655,
43
+ "tokens_seen": 2974416896
44
+ },
45
+ {
46
+ "epoch": 0.02,
47
+ "learning_rate": 9.90392313851081e-05,
48
+ "loss": 1.2074,
49
+ "theoretical_loss": 3.3513036937553267,
50
+ "tokens_seen": 2975727616
51
+ },
52
+ {
53
+ "epoch": 0.02,
54
+ "learning_rate": 9.863891112890312e-05,
55
+ "loss": 1.2079,
56
+ "theoretical_loss": 3.3511910074884628,
57
+ "tokens_seen": 2977038336
58
+ },
59
+ {
60
+ "epoch": 0.02,
61
+ "objective/train/docs_used": 997783,
62
+ "objective/train/instantaneous_batch_size": 8,
63
+ "objective/train/instantaneous_microbatch_size": 8192,
64
+ "objective/train/original_loss": 0.9412798285484314,
65
+ "objective/train/theoretical_loss": 3.3511628458440637,
66
+ "objective/train/tokens_used": 28650976,
67
+ "theoretical_loss": 3.3511628458440637,
68
+ "tokens_seen": 2977366016
69
+ },
70
+ {
71
+ "epoch": 0.03,
72
+ "learning_rate": 9.823859087269817e-05,
73
+ "loss": 1.2032,
74
+ "theoretical_loss": 3.3510783847084977,
75
+ "tokens_seen": 2978349056
76
+ },
77
+ {
78
+ "epoch": 0.03,
79
+ "learning_rate": 9.78382706164932e-05,
80
+ "loss": 1.2178,
81
+ "theoretical_loss": 3.350965825351748,
82
+ "tokens_seen": 2979659776
83
+ },
84
+ {
85
+ "epoch": 0.04,
86
+ "learning_rate": 9.743795036028824e-05,
87
+ "loss": 1.189,
88
+ "theoretical_loss": 3.3508533293546208,
89
+ "tokens_seen": 2980970496
90
+ },
91
+ {
92
+ "epoch": 0.04,
93
+ "learning_rate": 9.703763010408326e-05,
94
+ "loss": 1.2086,
95
+ "theoretical_loss": 3.3507408966536154,
96
+ "tokens_seen": 2982281216
97
+ },
98
+ {
99
+ "epoch": 0.04,
100
+ "learning_rate": 9.663730984787832e-05,
101
+ "loss": 1.1797,
102
+ "theoretical_loss": 3.3506285271853233,
103
+ "tokens_seen": 2983591936
104
+ },
105
+ {
106
+ "epoch": 0.05,
107
+ "learning_rate": 9.623698959167334e-05,
108
+ "loss": 1.1709,
109
+ "theoretical_loss": 3.350516220886427,
110
+ "tokens_seen": 2984902656
111
+ },
112
+ {
113
+ "epoch": 0.05,
114
+ "objective/train/docs_used": 1000073,
115
+ "objective/train/instantaneous_batch_size": 8,
116
+ "objective/train/instantaneous_microbatch_size": 8192,
117
+ "objective/train/original_loss": 1.0221836566925049,
118
+ "objective/train/theoretical_loss": 3.3504600914057403,
119
+ "objective/train/tokens_used": 36842976,
120
+ "theoretical_loss": 3.3504600914057403,
121
+ "tokens_seen": 2985558016
122
+ },
123
+ {
124
+ "epoch": 0.05,
125
+ "learning_rate": 9.583666933546838e-05,
126
+ "loss": 1.1658,
127
+ "theoretical_loss": 3.3504039776936994,
128
+ "tokens_seen": 2986213376
129
+ },
130
+ {
131
+ "epoch": 0.06,
132
+ "learning_rate": 9.543634907926342e-05,
133
+ "loss": 1.1823,
134
+ "theoretical_loss": 3.350291797544005,
135
+ "tokens_seen": 2987524096
136
+ },
137
+ {
138
+ "epoch": 0.06,
139
+ "learning_rate": 9.503602882305846e-05,
140
+ "loss": 1.1859,
141
+ "theoretical_loss": 3.3501796803742994,
142
+ "tokens_seen": 2988834816
143
+ },
144
+ {
145
+ "epoch": 0.06,
146
+ "learning_rate": 9.463570856685348e-05,
147
+ "loss": 1.1906,
148
+ "theoretical_loss": 3.3500676261216285,
149
+ "tokens_seen": 2990145536
150
+ },
151
+ {
152
+ "epoch": 0.07,
153
+ "learning_rate": 9.423538831064852e-05,
154
+ "loss": 1.185,
155
+ "theoretical_loss": 3.3499556347231287,
156
+ "tokens_seen": 2991456256
157
+ },
158
+ {
159
+ "epoch": 0.07,
160
+ "learning_rate": 9.383506805444356e-05,
161
+ "loss": 1.2042,
162
+ "theoretical_loss": 3.349843706116027,
163
+ "tokens_seen": 2992766976
164
+ },
165
+ {
166
+ "epoch": 0.07,
167
+ "objective/train/docs_used": 1002769,
168
+ "objective/train/instantaneous_batch_size": 8,
169
+ "objective/train/instantaneous_microbatch_size": 8192,
170
+ "objective/train/original_loss": 1.227379560470581,
171
+ "objective/train/theoretical_loss": 3.349759800829844,
172
+ "objective/train/tokens_used": 45034976,
173
+ "theoretical_loss": 3.349759800829844,
174
+ "tokens_seen": 2993750016
175
+ },
176
+ {
177
+ "epoch": 0.08,
178
+ "learning_rate": 9.34347477982386e-05,
179
+ "loss": 1.1807,
180
+ "theoretical_loss": 3.3497318402376397,
181
+ "tokens_seen": 2994077696
182
+ },
183
+ {
184
+ "epoch": 0.08,
185
+ "learning_rate": 9.303442754203363e-05,
186
+ "loss": 1.1862,
187
+ "theoretical_loss": 3.349620037025374,
188
+ "tokens_seen": 2995388416
189
+ },
190
+ {
191
+ "epoch": 0.08,
192
+ "learning_rate": 9.263410728582867e-05,
193
+ "loss": 1.178,
194
+ "theoretical_loss": 3.349508296416727,
195
+ "tokens_seen": 2996699136
196
+ },
197
+ {
198
+ "epoch": 0.09,
199
+ "learning_rate": 9.22337870296237e-05,
200
+ "loss": 1.1656,
201
+ "theoretical_loss": 3.3493966183492847,
202
+ "tokens_seen": 2998009856
203
+ },
204
+ {
205
+ "epoch": 0.09,
206
+ "learning_rate": 9.183346677341874e-05,
207
+ "loss": 1.1769,
208
+ "theoretical_loss": 3.349285002760723,
209
+ "tokens_seen": 2999320576
210
+ },
211
+ {
212
+ "epoch": 0.1,
213
+ "learning_rate": 9.143314651721377e-05,
214
+ "loss": 1.1513,
215
+ "theoretical_loss": 3.3491734495888066,
216
+ "tokens_seen": 3000631296
217
+ },
218
+ {
219
+ "epoch": 0.1,
220
+ "objective/train/docs_used": 1005438,
221
+ "objective/train/instantaneous_batch_size": 8,
222
+ "objective/train/instantaneous_microbatch_size": 8192,
223
+ "objective/train/original_loss": 1.1515272855758667,
224
+ "objective/train/theoretical_loss": 3.3490619587713906,
225
+ "objective/train/tokens_used": 53226976,
226
+ "theoretical_loss": 3.3490619587713906,
227
+ "tokens_seen": 3001942016
228
+ },
229
+ {
230
+ "epoch": 0.1,
231
+ "learning_rate": 9.103282626100881e-05,
232
+ "loss": 1.1793,
233
+ "theoretical_loss": 3.3490619587713906,
234
+ "tokens_seen": 3001942016
235
+ },
236
+ {
237
+ "epoch": 0.1,
238
+ "learning_rate": 9.063250600480385e-05,
239
+ "loss": 1.1534,
240
+ "theoretical_loss": 3.348950530246418,
241
+ "tokens_seen": 3003252736
242
+ },
243
+ {
244
+ "epoch": 0.11,
245
+ "learning_rate": 9.023218574859889e-05,
246
+ "loss": 1.1928,
247
+ "theoretical_loss": 3.348839163951921,
248
+ "tokens_seen": 3004563456
249
+ },
250
+ {
251
+ "epoch": 0.11,
252
+ "learning_rate": 8.983186549239391e-05,
253
+ "loss": 1.1678,
254
+ "theoretical_loss": 3.3487278598260195,
255
+ "tokens_seen": 3005874176
256
+ },
257
+ {
258
+ "epoch": 0.11,
259
+ "learning_rate": 8.943154523618895e-05,
260
+ "loss": 1.1477,
261
+ "theoretical_loss": 3.348616617806923,
262
+ "tokens_seen": 3007184896
263
+ },
264
+ {
265
+ "epoch": 0.12,
266
+ "learning_rate": 8.903122497998399e-05,
267
+ "loss": 1.1773,
268
+ "theoretical_loss": 3.34850543783293,
269
+ "tokens_seen": 3008495616
270
+ },
271
+ {
272
+ "epoch": 0.12,
273
+ "learning_rate": 8.863090472377903e-05,
274
+ "loss": 1.185,
275
+ "theoretical_loss": 3.3483943198424253,
276
+ "tokens_seen": 3009806336
277
+ },
278
+ {
279
+ "epoch": 0.12,
280
+ "objective/train/docs_used": 1008334,
281
+ "objective/train/instantaneous_batch_size": 8,
282
+ "objective/train/instantaneous_microbatch_size": 8192,
283
+ "objective/train/original_loss": 1.3460177183151245,
284
+ "objective/train/theoretical_loss": 3.348366550022507,
285
+ "objective/train/tokens_used": 61418976,
286
+ "theoretical_loss": 3.348366550022507,
287
+ "tokens_seen": 3010134016
288
+ },
289
+ {
290
+ "epoch": 0.13,
291
+ "learning_rate": 8.823058446757406e-05,
292
+ "loss": 1.1798,
293
+ "theoretical_loss": 3.348283263773882,
294
+ "tokens_seen": 3011117056
295
+ },
296
+ {
297
+ "epoch": 0.13,
298
+ "learning_rate": 8.78302642113691e-05,
299
+ "loss": 1.1794,
300
+ "theoretical_loss": 3.3481722695658633,
301
+ "tokens_seen": 3012427776
302
+ },
303
+ {
304
+ "epoch": 0.13,
305
+ "learning_rate": 8.742994395516414e-05,
306
+ "loss": 1.1759,
307
+ "theoretical_loss": 3.3480613371570165,
308
+ "tokens_seen": 3013738496
309
+ },
310
+ {
311
+ "epoch": 0.14,
312
+ "learning_rate": 8.702962369895917e-05,
313
+ "loss": 1.1839,
314
+ "theoretical_loss": 3.347950466486079,
315
+ "tokens_seen": 3015049216
316
+ },
317
+ {
318
+ "epoch": 0.14,
319
+ "learning_rate": 8.66293034427542e-05,
320
+ "loss": 1.1996,
321
+ "theoretical_loss": 3.3478396574918747,
322
+ "tokens_seen": 3016359936
323
+ },
324
+ {
325
+ "epoch": 0.15,
326
+ "learning_rate": 8.622898318654924e-05,
327
+ "loss": 1.1897,
328
+ "theoretical_loss": 3.347728910113315,
329
+ "tokens_seen": 3017670656
330
+ },
331
+ {
332
+ "epoch": 0.15,
333
+ "objective/train/docs_used": 1011213,
334
+ "objective/train/instantaneous_batch_size": 8,
335
+ "objective/train/instantaneous_microbatch_size": 8192,
336
+ "objective/train/original_loss": 1.227089524269104,
337
+ "objective/train/theoretical_loss": 3.3476735595108353,
338
+ "objective/train/tokens_used": 69610976,
339
+ "theoretical_loss": 3.3476735595108353,
340
+ "tokens_seen": 3018326016
341
+ },
342
+ {
343
+ "epoch": 0.15,
344
+ "learning_rate": 8.582866293034428e-05,
345
+ "loss": 1.1741,
346
+ "theoretical_loss": 3.3476182242893975,
347
+ "tokens_seen": 3018981376
348
+ },
349
+ {
350
+ "epoch": 0.15,
351
+ "learning_rate": 8.542834267413932e-05,
352
+ "loss": 1.191,
353
+ "theoretical_loss": 3.3475075999592074,
354
+ "tokens_seen": 3020292096
355
+ },
356
+ {
357
+ "epoch": 0.16,
358
+ "learning_rate": 8.502802241793434e-05,
359
+ "loss": 1.1924,
360
+ "theoretical_loss": 3.347397037061916,
361
+ "tokens_seen": 3021602816
362
+ },
363
+ {
364
+ "epoch": 0.16,
365
+ "learning_rate": 8.462770216172938e-05,
366
+ "loss": 1.188,
367
+ "theoretical_loss": 3.3472865355367825,
368
+ "tokens_seen": 3022913536
369
+ },
370
+ {
371
+ "epoch": 0.17,
372
+ "learning_rate": 8.422738190552442e-05,
373
+ "loss": 1.1754,
374
+ "theoretical_loss": 3.3471760953231504,
375
+ "tokens_seen": 3024224256
376
+ },
377
+ {
378
+ "epoch": 0.17,
379
+ "learning_rate": 8.382706164931946e-05,
380
+ "loss": 1.2086,
381
+ "theoretical_loss": 3.3470657163604507,
382
+ "tokens_seen": 3025534976
383
+ },
384
+ {
385
+ "epoch": 0.17,
386
+ "objective/train/docs_used": 1013697,
387
+ "objective/train/instantaneous_batch_size": 8,
388
+ "objective/train/instantaneous_microbatch_size": 8192,
389
+ "objective/train/original_loss": 1.1135369539260864,
390
+ "objective/train/theoretical_loss": 3.3469829722979627,
391
+ "objective/train/tokens_used": 77802976,
392
+ "theoretical_loss": 3.3469829722979627,
393
+ "tokens_seen": 3026518016
394
+ },
395
+ {
396
+ "epoch": 0.17,
397
+ "learning_rate": 8.34267413931145e-05,
398
+ "loss": 1.1969,
399
+ "theoretical_loss": 3.3469553985882,
400
+ "tokens_seen": 3026845696
401
+ },
402
+ {
403
+ "epoch": 0.18,
404
+ "learning_rate": 8.302642113690953e-05,
405
+ "loss": 1.2176,
406
+ "theoretical_loss": 3.3468451419460017,
407
+ "tokens_seen": 3028156416
408
+ },
409
+ {
410
+ "epoch": 0.18,
411
+ "learning_rate": 8.262610088070457e-05,
412
+ "loss": 1.188,
413
+ "theoretical_loss": 3.3467349463735436,
414
+ "tokens_seen": 3029467136
415
+ },
416
+ {
417
+ "epoch": 0.19,
418
+ "learning_rate": 8.22257806244996e-05,
419
+ "loss": 1.1834,
420
+ "theoretical_loss": 3.3466248118105995,
421
+ "tokens_seen": 3030777856
422
+ },
423
+ {
424
+ "epoch": 0.19,
425
+ "learning_rate": 8.182546036829464e-05,
426
+ "loss": 1.1977,
427
+ "theoretical_loss": 3.3465147381970293,
428
+ "tokens_seen": 3032088576
429
+ },
430
+ {
431
+ "epoch": 0.19,
432
+ "learning_rate": 8.142514011208967e-05,
433
+ "loss": 1.1889,
434
+ "theoretical_loss": 3.346404725472777,
435
+ "tokens_seen": 3033399296
436
+ },
437
+ {
438
+ "epoch": 0.2,
439
+ "objective/train/docs_used": 1016482,
440
+ "objective/train/instantaneous_batch_size": 8,
441
+ "objective/train/instantaneous_microbatch_size": 8192,
442
+ "objective/train/original_loss": 1.4041112661361694,
443
+ "objective/train/theoretical_loss": 3.3462947735778727,
444
+ "objective/train/tokens_used": 85994976,
445
+ "theoretical_loss": 3.3462947735778727,
446
+ "tokens_seen": 3034710016
447
+ },
448
+ {
449
+ "epoch": 0.2,
450
+ "learning_rate": 8.102481985588471e-05,
451
+ "loss": 1.1983,
452
+ "theoretical_loss": 3.3462947735778727,
453
+ "tokens_seen": 3034710016
454
+ },
455
+ {
456
+ "epoch": 0.2,
457
+ "learning_rate": 8.062449959967975e-05,
458
+ "loss": 1.1694,
459
+ "theoretical_loss": 3.346184882452431,
460
+ "tokens_seen": 3036020736
461
+ },
462
+ {
463
+ "epoch": 0.21,
464
+ "learning_rate": 8.022417934347479e-05,
465
+ "loss": 1.1871,
466
+ "theoretical_loss": 3.3460750520366513,
467
+ "tokens_seen": 3037331456
468
+ },
469
+ {
470
+ "epoch": 0.21,
471
+ "learning_rate": 7.982385908726981e-05,
472
+ "loss": 1.1833,
473
+ "theoretical_loss": 3.3459652822708175,
474
+ "tokens_seen": 3038642176
475
+ },
476
+ {
477
+ "epoch": 0.21,
478
+ "learning_rate": 7.942353883106487e-05,
479
+ "loss": 1.1831,
480
+ "theoretical_loss": 3.345855573095298,
481
+ "tokens_seen": 3039952896
482
+ },
483
+ {
484
+ "epoch": 0.22,
485
+ "learning_rate": 7.902321857485989e-05,
486
+ "loss": 1.1877,
487
+ "theoretical_loss": 3.3457459244505463,
488
+ "tokens_seen": 3041263616
489
+ },
490
+ {
491
+ "epoch": 0.22,
492
+ "learning_rate": 7.862289831865493e-05,
493
+ "loss": 1.2048,
494
+ "theoretical_loss": 3.3456363362770984,
495
+ "tokens_seen": 3042574336
496
+ },
497
+ {
498
+ "epoch": 0.22,
499
+ "objective/train/docs_used": 1019410,
500
+ "objective/train/instantaneous_batch_size": 8,
501
+ "objective/train/instantaneous_microbatch_size": 8192,
502
+ "objective/train/original_loss": 1.1971296072006226,
503
+ "objective/train/theoretical_loss": 3.3456089486754172,
504
+ "objective/train/tokens_used": 94186976,
505
+ "theoretical_loss": 3.3456089486754172,
506
+ "tokens_seen": 3042902016
507
+ },
508
+ {
509
+ "epoch": 0.23,
510
+ "learning_rate": 7.822257806244996e-05,
511
+ "loss": 1.1685,
512
+ "theoretical_loss": 3.345526808515576,
513
+ "tokens_seen": 3043885056
514
+ },
515
+ {
516
+ "epoch": 0.23,
517
+ "learning_rate": 7.782225780624501e-05,
518
+ "loss": 1.1907,
519
+ "theoretical_loss": 3.3454173411066828,
520
+ "tokens_seen": 3045195776
521
+ },
522
+ {
523
+ "epoch": 0.23,
524
+ "learning_rate": 7.742193755004003e-05,
525
+ "loss": 1.1949,
526
+ "theoretical_loss": 3.345307933991209,
527
+ "tokens_seen": 3046506496
528
+ },
529
+ {
530
+ "epoch": 0.24,
531
+ "learning_rate": 7.702161729383507e-05,
532
+ "loss": 1.1959,
533
+ "theoretical_loss": 3.3451985871100254,
534
+ "tokens_seen": 3047817216
535
+ },
536
+ {
537
+ "epoch": 0.24,
538
+ "learning_rate": 7.662129703763011e-05,
539
+ "loss": 1.1728,
540
+ "theoretical_loss": 3.345089300404087,
541
+ "tokens_seen": 3049127936
542
+ },
543
+ {
544
+ "epoch": 0.25,
545
+ "learning_rate": 7.622097678142515e-05,
546
+ "loss": 1.1995,
547
+ "theoretical_loss": 3.344980073814434,
548
+ "tokens_seen": 3050438656
549
+ },
550
+ {
551
+ "debugging/Compilability": 0.0,
552
+ "debugging/distinct-1-grams": 0.7613636363636364,
553
+ "debugging/entropy-1-grams": 3.9967058025744,
554
+ "debugging/length": 568.0,
555
+ "debugging/num_segments": 1,
556
+ "debugging/score": 0.045774647887323945,
557
+ "debugging/score_std": 0.0,
558
+ "epoch": 0.25,
559
+ "objective/train/docs_used": 1022217,
560
+ "objective/train/instantaneous_batch_size": 8,
561
+ "objective/train/instantaneous_microbatch_size": 8192,
562
+ "objective/train/original_loss": 0.8801804780960083,
563
+ "objective/train/theoretical_loss": 3.3449254830448116,
564
+ "objective/train/tokens_used": 102378976,
565
+ "theoretical_loss": 3.3449254830448116,
566
+ "tokens_seen": 3051094016
567
+ },
568
+ {
569
+ "epoch": 0.25,
570
+ "learning_rate": 7.582065652522018e-05,
571
+ "loss": 1.1905,
572
+ "theoretical_loss": 3.3448709072821874,
573
+ "tokens_seen": 3051749376
574
+ },
575
+ {
576
+ "epoch": 0.25,
577
+ "learning_rate": 7.542033626901522e-05,
578
+ "loss": 1.1939,
579
+ "theoretical_loss": 3.344761800748551,
580
+ "tokens_seen": 3053060096
581
+ },
582
+ {
583
+ "epoch": 0.26,
584
+ "learning_rate": 7.502001601281026e-05,
585
+ "loss": 1.1901,
586
+ "theoretical_loss": 3.3446527541548132,
587
+ "tokens_seen": 3054370816
588
+ },
589
+ {
590
+ "epoch": 0.26,
591
+ "learning_rate": 7.46196957566053e-05,
592
+ "loss": 1.2002,
593
+ "theoretical_loss": 3.3445437674423437,
594
+ "tokens_seen": 3055681536
595
+ },
596
+ {
597
+ "epoch": 0.27,
598
+ "learning_rate": 7.421937550040032e-05,
599
+ "loss": 1.2223,
600
+ "theoretical_loss": 3.3444348405525948,
601
+ "tokens_seen": 3056992256
602
+ },
603
+ {
604
+ "epoch": 0.27,
605
+ "learning_rate": 7.381905524419536e-05,
606
+ "loss": 1.2028,
607
+ "theoretical_loss": 3.3443259734271016,
608
+ "tokens_seen": 3058302976
609
+ },
610
+ {
611
+ "epoch": 0.27,
612
+ "objective/train/docs_used": 1025165,
613
+ "objective/train/instantaneous_batch_size": 8,
614
+ "objective/train/instantaneous_microbatch_size": 8192,
615
+ "objective/train/original_loss": 1.2416154146194458,
616
+ "objective/train/theoretical_loss": 3.3442443622681504,
617
+ "objective/train/tokens_used": 110570976,
618
+ "theoretical_loss": 3.3442443622681504,
619
+ "tokens_seen": 3059286016
620
+ },
621
+ {
622
+ "epoch": 0.27,
623
+ "learning_rate": 7.34187349879904e-05,
624
+ "loss": 1.1917,
625
+ "theoretical_loss": 3.344217166007481,
626
+ "tokens_seen": 3059613696
627
+ },
628
+ {
629
+ "epoch": 0.28,
630
+ "learning_rate": 7.301841473178544e-05,
631
+ "loss": 1.1963,
632
+ "theoretical_loss": 3.344108418235432,
633
+ "tokens_seen": 3060924416
634
+ },
635
+ {
636
+ "epoch": 0.28,
637
+ "learning_rate": 7.261809447558046e-05,
638
+ "loss": 1.1737,
639
+ "theoretical_loss": 3.343999730052735,
640
+ "tokens_seen": 3062235136
641
+ },
642
+ {
643
+ "epoch": 0.29,
644
+ "learning_rate": 7.22177742193755e-05,
645
+ "loss": 1.1848,
646
+ "theoretical_loss": 3.343891101401253,
647
+ "tokens_seen": 3063545856
648
+ },
649
+ {
650
+ "epoch": 0.29,
651
+ "learning_rate": 7.181745396317054e-05,
652
+ "loss": 1.1799,
653
+ "theoretical_loss": 3.3437825322229298,
654
+ "tokens_seen": 3064856576
655
+ },
656
+ {
657
+ "epoch": 0.29,
658
+ "learning_rate": 7.141713370696558e-05,
659
+ "loss": 1.1818,
660
+ "theoretical_loss": 3.3436740224597914,
661
+ "tokens_seen": 3066167296
662
+ },
663
+ {
664
+ "epoch": 0.3,
665
+ "objective/train/docs_used": 1028068,
666
+ "objective/train/instantaneous_batch_size": 8,
667
+ "objective/train/instantaneous_microbatch_size": 8192,
668
+ "objective/train/original_loss": 1.3919180631637573,
669
+ "objective/train/theoretical_loss": 3.3435655720539446,
670
+ "objective/train/tokens_used": 118762976,
671
+ "theoretical_loss": 3.3435655720539446,
672
+ "tokens_seen": 3067478016
673
+ },
674
+ {
675
+ "epoch": 0.3,
676
+ "learning_rate": 7.101681345076061e-05,
677
+ "loss": 1.1999,
678
+ "theoretical_loss": 3.3435655720539446,
679
+ "tokens_seen": 3067478016
680
+ },
681
+ {
682
+ "epoch": 0.3,
683
+ "learning_rate": 7.061649319455565e-05,
684
+ "loss": 1.1951,
685
+ "theoretical_loss": 3.343457180947577,
686
+ "tokens_seen": 3068788736
687
+ },
688
+ {
689
+ "epoch": 0.31,
690
+ "learning_rate": 7.021617293835069e-05,
691
+ "loss": 1.1921,
692
+ "theoretical_loss": 3.343348849082957,
693
+ "tokens_seen": 3070099456
694
+ },
695
+ {
696
+ "epoch": 0.31,
697
+ "learning_rate": 6.981585268214572e-05,
698
+ "loss": 1.1899,
699
+ "theoretical_loss": 3.3432405764024353,
700
+ "tokens_seen": 3071410176
701
+ },
702
+ {
703
+ "epoch": 0.31,
704
+ "learning_rate": 6.941553242594075e-05,
705
+ "loss": 1.1887,
706
+ "theoretical_loss": 3.3431323628484417,
707
+ "tokens_seen": 3072720896
708
+ },
709
+ {
710
+ "epoch": 0.32,
711
+ "learning_rate": 6.901521216973579e-05,
712
+ "loss": 1.1913,
713
+ "theoretical_loss": 3.343024208363487,
714
+ "tokens_seen": 3074031616
715
+ },
716
+ {
717
+ "epoch": 0.32,
718
+ "learning_rate": 6.861489191353083e-05,
719
+ "loss": 1.1801,
720
+ "theoretical_loss": 3.3429161128901628,
721
+ "tokens_seen": 3075342336
722
+ },
723
+ {
724
+ "epoch": 0.32,
725
+ "objective/train/docs_used": 1031046,
726
+ "objective/train/instantaneous_batch_size": 8,
727
+ "objective/train/instantaneous_microbatch_size": 8192,
728
+ "objective/train/original_loss": 1.355250597000122,
729
+ "objective/train/theoretical_loss": 3.342889098235679,
730
+ "objective/train/tokens_used": 126954976,
731
+ "theoretical_loss": 3.342889098235679,
732
+ "tokens_seen": 3075670016
733
+ },
734
+ {
735
+ "epoch": 0.32,
736
+ "learning_rate": 6.821457165732587e-05,
737
+ "loss": 1.186,
738
+ "theoretical_loss": 3.3428080763711403,
739
+ "tokens_seen": 3076653056
740
+ },
741
+ {
742
+ "epoch": 0.33,
743
+ "learning_rate": 6.78142514011209e-05,
744
+ "loss": 1.1817,
745
+ "theoretical_loss": 3.3427000987491717,
746
+ "tokens_seen": 3077963776
747
+ },
748
+ {
749
+ "epoch": 0.33,
750
+ "learning_rate": 6.741393114491593e-05,
751
+ "loss": 1.1953,
752
+ "theoretical_loss": 3.342592179967088,
753
+ "tokens_seen": 3079274496
754
+ },
755
+ {
756
+ "epoch": 0.34,
757
+ "learning_rate": 6.701361088871097e-05,
758
+ "loss": 1.1954,
759
+ "theoretical_loss": 3.3424843199678005,
760
+ "tokens_seen": 3080585216
761
+ },
762
+ {
763
+ "epoch": 0.34,
764
+ "learning_rate": 6.661329063250601e-05,
765
+ "loss": 1.1849,
766
+ "theoretical_loss": 3.3423765186943006,
767
+ "tokens_seen": 3081895936
768
+ },
769
+ {
770
+ "epoch": 0.34,
771
+ "learning_rate": 6.621297037630104e-05,
772
+ "loss": 1.1971,
773
+ "theoretical_loss": 3.3422687760896586,
774
+ "tokens_seen": 3083206656
775
+ },
776
+ {
777
+ "epoch": 0.35,
778
+ "objective/train/docs_used": 1033433,
779
+ "objective/train/instantaneous_batch_size": 8,
780
+ "objective/train/instantaneous_microbatch_size": 8192,
781
+ "objective/train/original_loss": 1.3183742761611938,
782
+ "objective/train/theoretical_loss": 3.3422149267703904,
783
+ "objective/train/tokens_used": 135146976,
784
+ "theoretical_loss": 3.3422149267703904,
785
+ "tokens_seen": 3083862016
786
+ },
787
+ {
788
+ "epoch": 0.35,
789
+ "learning_rate": 6.581265012009608e-05,
790
+ "loss": 1.1814,
791
+ "theoretical_loss": 3.342161092097024,
792
+ "tokens_seen": 3084517376
793
+ },
794
+ {
795
+ "epoch": 0.35,
796
+ "learning_rate": 6.541232986389112e-05,
797
+ "loss": 1.1732,
798
+ "theoretical_loss": 3.342053466659626,
799
+ "tokens_seen": 3085828096
800
+ },
801
+ {
802
+ "epoch": 0.36,
803
+ "learning_rate": 6.501200960768615e-05,
804
+ "loss": 1.1698,
805
+ "theoretical_loss": 3.3419458997207734,
806
+ "tokens_seen": 3087138816
807
+ },
808
+ {
809
+ "epoch": 0.36,
810
+ "learning_rate": 6.461168935148118e-05,
811
+ "loss": 1.1775,
812
+ "theoretical_loss": 3.3418383912238525,
813
+ "tokens_seen": 3088449536
814
+ },
815
+ {
816
+ "epoch": 0.36,
817
+ "learning_rate": 6.421136909527622e-05,
818
+ "loss": 1.1784,
819
+ "theoretical_loss": 3.341730941112329,
820
+ "tokens_seen": 3089760256
821
+ },
822
+ {
823
+ "epoch": 0.37,
824
+ "learning_rate": 6.381104883907126e-05,
825
+ "loss": 1.1965,
826
+ "theoretical_loss": 3.341623549329748,
827
+ "tokens_seen": 3091070976
828
+ },
829
+ {
830
+ "epoch": 0.37,
831
+ "objective/train/docs_used": 1036380,
832
+ "objective/train/instantaneous_batch_size": 8,
833
+ "objective/train/instantaneous_microbatch_size": 8192,
834
+ "objective/train/original_loss": 1.2508410215377808,
835
+ "objective/train/theoretical_loss": 3.3415430437372637,
836
+ "objective/train/tokens_used": 143338976,
837
+ "theoretical_loss": 3.3415430437372637,
838
+ "tokens_seen": 3092054016
839
+ },
840
+ {
841
+ "epoch": 0.37,
842
+ "learning_rate": 6.34107285828663e-05,
843
+ "loss": 1.1741,
844
+ "theoretical_loss": 3.3415162158197327,
845
+ "tokens_seen": 3092381696
846
+ },
847
+ {
848
+ "epoch": 0.38,
849
+ "learning_rate": 6.301040832666132e-05,
850
+ "loss": 1.1813,
851
+ "theoretical_loss": 3.3414089405259837,
852
+ "tokens_seen": 3093692416
853
+ },
854
+ {
855
+ "epoch": 0.38,
856
+ "learning_rate": 6.261008807045636e-05,
857
+ "loss": 1.1763,
858
+ "theoretical_loss": 3.341301723392281,
859
+ "tokens_seen": 3095003136
860
+ },
861
+ {
862
+ "epoch": 0.38,
863
+ "learning_rate": 6.22097678142514e-05,
864
+ "loss": 1.1828,
865
+ "theoretical_loss": 3.3411945643624814,
866
+ "tokens_seen": 3096313856
867
+ },
868
+ {
869
+ "epoch": 0.39,
870
+ "learning_rate": 6.180944755804644e-05,
871
+ "loss": 1.1958,
872
+ "theoretical_loss": 3.3410874633805214,
873
+ "tokens_seen": 3097624576
874
+ },
875
+ {
876
+ "epoch": 0.39,
877
+ "learning_rate": 6.140912730184148e-05,
878
+ "loss": 1.1942,
879
+ "theoretical_loss": 3.3409804203904145,
880
+ "tokens_seen": 3098935296
881
+ },
882
+ {
883
+ "epoch": 0.4,
884
+ "objective/train/docs_used": 1039074,
885
+ "objective/train/instantaneous_batch_size": 8,
886
+ "objective/train/instantaneous_microbatch_size": 8192,
887
+ "objective/train/original_loss": 1.3202261924743652,
888
+ "objective/train/theoretical_loss": 3.3408734353362513,
889
+ "objective/train/tokens_used": 151530976,
890
+ "theoretical_loss": 3.3408734353362513,
891
+ "tokens_seen": 3100246016
892
+ },
893
+ {
894
+ "epoch": 0.4,
895
+ "learning_rate": 6.1008807045636506e-05,
896
+ "loss": 1.1996,
897
+ "theoretical_loss": 3.3408734353362513,
898
+ "tokens_seen": 3100246016
899
+ },
900
+ {
901
+ "epoch": 0.4,
902
+ "learning_rate": 6.060848678943155e-05,
903
+ "loss": 1.1876,
904
+ "theoretical_loss": 3.3407665081622,
905
+ "tokens_seen": 3101556736
906
+ },
907
+ {
908
+ "epoch": 0.4,
909
+ "learning_rate": 6.0208166533226584e-05,
910
+ "loss": 1.1839,
911
+ "theoretical_loss": 3.340659638812507,
912
+ "tokens_seen": 3102867456
913
+ },
914
+ {
915
+ "epoch": 0.41,
916
+ "learning_rate": 5.9807846277021617e-05,
917
+ "loss": 1.192,
918
+ "theoretical_loss": 3.340552827231495,
919
+ "tokens_seen": 3104178176
920
+ },
921
+ {
922
+ "epoch": 0.41,
923
+ "learning_rate": 5.940752602081665e-05,
924
+ "loss": 1.1838,
925
+ "theoretical_loss": 3.3404460733635646,
926
+ "tokens_seen": 3105488896
927
+ },
928
+ {
929
+ "epoch": 0.42,
930
+ "learning_rate": 5.9007205764611695e-05,
931
+ "loss": 1.1796,
932
+ "theoretical_loss": 3.340339377153193,
933
+ "tokens_seen": 3106799616
934
+ },
935
+ {
936
+ "epoch": 0.42,
937
+ "learning_rate": 5.860688550840673e-05,
938
+ "loss": 1.1824,
939
+ "theoretical_loss": 3.340232738544933,
940
+ "tokens_seen": 3108110336
941
+ },
942
+ {
943
+ "epoch": 0.42,
944
+ "objective/train/docs_used": 1042051,
945
+ "objective/train/instantaneous_batch_size": 8,
946
+ "objective/train/instantaneous_microbatch_size": 8192,
947
+ "objective/train/original_loss": 1.5605241060256958,
948
+ "objective/train/theoretical_loss": 3.340206087886708,
949
+ "objective/train/tokens_used": 159722976,
950
+ "theoretical_loss": 3.340206087886708,
951
+ "tokens_seen": 3108438016
952
+ },
953
+ {
954
+ "epoch": 0.42,
955
+ "learning_rate": 5.820656525220176e-05,
956
+ "loss": 1.1953,
957
+ "theoretical_loss": 3.340126157483417,
958
+ "tokens_seen": 3109421056
959
+ },
960
+ {
961
+ "epoch": 0.43,
962
+ "learning_rate": 5.7806244995996806e-05,
963
+ "loss": 1.1918,
964
+ "theoretical_loss": 3.340019633913351,
965
+ "tokens_seen": 3110731776
966
+ },
967
+ {
968
+ "epoch": 0.43,
969
+ "learning_rate": 5.740592473979184e-05,
970
+ "loss": 1.2149,
971
+ "theoretical_loss": 3.3399131677795184,
972
+ "tokens_seen": 3112042496
973
+ },
974
+ {
975
+ "epoch": 0.44,
976
+ "learning_rate": 5.700560448358687e-05,
977
+ "loss": 1.2105,
978
+ "theoretical_loss": 3.3398067590267804,
979
+ "tokens_seen": 3113353216
980
+ },
981
+ {
982
+ "epoch": 0.44,
983
+ "learning_rate": 5.66052842273819e-05,
984
+ "loss": 1.1876,
985
+ "theoretical_loss": 3.339700407600071,
986
+ "tokens_seen": 3114663936
987
+ },
988
+ {
989
+ "epoch": 0.44,
990
+ "learning_rate": 5.620496397117695e-05,
991
+ "loss": 1.1859,
992
+ "theoretical_loss": 3.339594113444404,
993
+ "tokens_seen": 3115974656
994
+ },
995
+ {
996
+ "epoch": 0.45,
997
+ "objective/train/docs_used": 1044567,
998
+ "objective/train/instantaneous_batch_size": 8,
999
+ "objective/train/instantaneous_microbatch_size": 8192,
1000
+ "objective/train/original_loss": 1.1618666648864746,
1001
+ "objective/train/theoretical_loss": 3.339540987826048,
1002
+ "objective/train/tokens_used": 167914976,
1003
+ "theoretical_loss": 3.339540987826048,
1004
+ "tokens_seen": 3116630016
1005
+ },
1006
+ {
1007
+ "epoch": 0.45,
1008
+ "learning_rate": 5.580464371497198e-05,
1009
+ "loss": 1.1922,
1010
+ "theoretical_loss": 3.3394878765048657,
1011
+ "tokens_seen": 3117285376
1012
+ },
1013
+ {
1014
+ "epoch": 0.45,
1015
+ "learning_rate": 5.5404323458767014e-05,
1016
+ "loss": 1.1874,
1017
+ "theoretical_loss": 3.3393816967266208,
1018
+ "tokens_seen": 3118596096
1019
+ },
1020
+ {
1021
+ "epoch": 0.46,
1022
+ "learning_rate": 5.5004003202562046e-05,
1023
+ "loss": 1.1985,
1024
+ "theoretical_loss": 3.3392755740549074,
1025
+ "tokens_seen": 3119906816
1026
+ },
1027
+ {
1028
+ "epoch": 0.46,
1029
+ "learning_rate": 5.460368294635709e-05,
1030
+ "loss": 1.1583,
1031
+ "theoretical_loss": 3.33916950843504,
1032
+ "tokens_seen": 3121217536
1033
+ },
1034
+ {
1035
+ "epoch": 0.46,
1036
+ "learning_rate": 5.4203362690152125e-05,
1037
+ "loss": 1.1743,
1038
+ "theoretical_loss": 3.3390634998124096,
1039
+ "tokens_seen": 3122528256
1040
+ },
1041
+ {
1042
+ "epoch": 0.47,
1043
+ "learning_rate": 5.380304243394716e-05,
1044
+ "loss": 1.1793,
1045
+ "theoretical_loss": 3.3389575481324796,
1046
+ "tokens_seen": 3123838976
1047
+ },
1048
+ {
1049
+ "epoch": 0.47,
1050
+ "objective/train/docs_used": 1047326,
1051
+ "objective/train/instantaneous_batch_size": 8,
1052
+ "objective/train/instantaneous_microbatch_size": 8192,
1053
+ "objective/train/original_loss": 1.2724859714508057,
1054
+ "objective/train/theoretical_loss": 3.338878121708416,
1055
+ "objective/train/tokens_used": 176106976,
1056
+ "theoretical_loss": 3.338878121708416,
1057
+ "tokens_seen": 3124822016
1058
+ },
1059
+ {
1060
+ "epoch": 0.47,
1061
+ "learning_rate": 5.340272217774219e-05,
1062
+ "loss": 1.209,
1063
+ "theoretical_loss": 3.338851653340791,
1064
+ "tokens_seen": 3125149696
1065
+ },
1066
+ {
1067
+ "epoch": 0.48,
1068
+ "learning_rate": 5.3002401921537236e-05,
1069
+ "loss": 1.1861,
1070
+ "theoretical_loss": 3.3387458153829583,
1071
+ "tokens_seen": 3126460416
1072
+ },
1073
+ {
1074
+ "epoch": 0.48,
1075
+ "learning_rate": 5.260208166533227e-05,
1076
+ "loss": 1.176,
1077
+ "theoretical_loss": 3.338640034204672,
1078
+ "tokens_seen": 3127771136
1079
+ },
1080
+ {
1081
+ "epoch": 0.48,
1082
+ "learning_rate": 5.22017614091273e-05,
1083
+ "loss": 1.1889,
1084
+ "theoretical_loss": 3.3385343097516946,
1085
+ "tokens_seen": 3129081856
1086
+ },
1087
+ {
1088
+ "epoch": 0.49,
1089
+ "learning_rate": 5.180144115292234e-05,
1090
+ "loss": 1.1717,
1091
+ "theoretical_loss": 3.338428641969867,
1092
+ "tokens_seen": 3130392576
1093
+ },
1094
+ {
1095
+ "epoch": 0.49,
1096
+ "learning_rate": 5.140112089671738e-05,
1097
+ "loss": 1.1977,
1098
+ "theoretical_loss": 3.3383230308051006,
1099
+ "tokens_seen": 3131703296
1100
+ },
1101
+ {
1102
+ "debugging/Compilability": 0.0,
1103
+ "debugging/distinct-1-grams": 0.6651816214290593,
1104
+ "debugging/entropy-1-grams": 4.2985685702692615,
1105
+ "debugging/length": 583.75,
1106
+ "debugging/num_segments": 4,
1107
+ "debugging/score": 0.06321276390574523,
1108
+ "debugging/score_std": 0.02740587124120667,
1109
+ "epoch": 0.49,
1110
+ "objective/train/docs_used": 1050300,
1111
+ "objective/train/instantaneous_batch_size": 8,
1112
+ "objective/train/instantaneous_microbatch_size": 8192,
1113
+ "objective/train/original_loss": 1.2621127367019653,
1114
+ "objective/train/theoretical_loss": 3.338217476203384,
1115
+ "objective/train/tokens_used": 184298976,
1116
+ "theoretical_loss": 3.338217476203384,
1117
+ "tokens_seen": 3133014016
1118
+ },
1119
+ {
1120
+ "epoch": 0.5,
1121
+ "learning_rate": 5.100080064051241e-05,
1122
+ "loss": 1.1834,
1123
+ "theoretical_loss": 3.338217476203384,
1124
+ "tokens_seen": 3133014016
1125
+ },
1126
+ {
1127
+ "epoch": 0.5,
1128
+ "learning_rate": 5.060048038430745e-05,
1129
+ "loss": 1.1796,
1130
+ "theoretical_loss": 3.3381119781107778,
1131
+ "tokens_seen": 3134324736
1132
+ },
1133
+ {
1134
+ "epoch": 0.5,
1135
+ "learning_rate": 5.020016012810248e-05,
1136
+ "loss": 1.1743,
1137
+ "theoretical_loss": 3.3380065364734186,
1138
+ "tokens_seen": 3135635456
1139
+ },
1140
+ {
1141
+ "epoch": 0.51,
1142
+ "learning_rate": 4.979983987189752e-05,
1143
+ "loss": 1.1641,
1144
+ "theoretical_loss": 3.3379011512375145,
1145
+ "tokens_seen": 3136946176
1146
+ },
1147
+ {
1148
+ "epoch": 0.51,
1149
+ "learning_rate": 4.9399519615692554e-05,
1150
+ "loss": 1.1767,
1151
+ "theoretical_loss": 3.3377958223493485,
1152
+ "tokens_seen": 3138256896
1153
+ },
1154
+ {
1155
+ "epoch": 0.52,
1156
+ "learning_rate": 4.8999199359487594e-05,
1157
+ "loss": 1.1809,
1158
+ "theoretical_loss": 3.337690549755278,
1159
+ "tokens_seen": 3139567616
1160
+ },
1161
+ {
1162
+ "epoch": 0.52,
1163
+ "learning_rate": 4.859887910328263e-05,
1164
+ "loss": 1.1658,
1165
+ "theoretical_loss": 3.3375853334017327,
1166
+ "tokens_seen": 3140878336
1167
+ },
1168
+ {
1169
+ "epoch": 0.52,
1170
+ "objective/train/docs_used": 1053223,
1171
+ "objective/train/instantaneous_batch_size": 8,
1172
+ "objective/train/instantaneous_microbatch_size": 8192,
1173
+ "objective/train/original_loss": 1.286133885383606,
1174
+ "objective/train/theoretical_loss": 3.337559038094658,
1175
+ "objective/train/tokens_used": 192490976,
1176
+ "theoretical_loss": 3.337559038094658,
1177
+ "tokens_seen": 3141206016
1178
+ },
1179
+ {
1180
+ "epoch": 0.52,
1181
+ "learning_rate": 4.8198558847077665e-05,
1182
+ "loss": 1.2072,
1183
+ "theoretical_loss": 3.3374801732352157,
1184
+ "tokens_seen": 3142189056
1185
+ },
1186
+ {
1187
+ "epoch": 0.53,
1188
+ "learning_rate": 4.7798238590872704e-05,
1189
+ "loss": 1.1917,
1190
+ "theoretical_loss": 3.337375069202303,
1191
+ "tokens_seen": 3143499776
1192
+ },
1193
+ {
1194
+ "epoch": 0.53,
1195
+ "learning_rate": 4.739791833466774e-05,
1196
+ "loss": 1.1742,
1197
+ "theoretical_loss": 3.337270021249646,
1198
+ "tokens_seen": 3144810496
1199
+ },
1200
+ {
1201
+ "epoch": 0.53,
1202
+ "learning_rate": 4.6997598078462776e-05,
1203
+ "loss": 1.1901,
1204
+ "theoretical_loss": 3.3371650293239643,
1205
+ "tokens_seen": 3146121216
1206
+ },
1207
+ {
1208
+ "epoch": 0.54,
1209
+ "learning_rate": 4.659727782225781e-05,
1210
+ "loss": 1.1848,
1211
+ "theoretical_loss": 3.337060093372055,
1212
+ "tokens_seen": 3147431936
1213
+ },
1214
+ {
1215
+ "epoch": 0.54,
1216
+ "learning_rate": 4.619695756605285e-05,
1217
+ "loss": 1.184,
1218
+ "theoretical_loss": 3.3369552133407856,
1219
+ "tokens_seen": 3148742656
1220
+ },
1221
+ {
1222
+ "epoch": 0.54,
1223
+ "objective/train/docs_used": 1056133,
1224
+ "objective/train/instantaneous_batch_size": 8,
1225
+ "objective/train/instantaneous_microbatch_size": 8192,
1226
+ "objective/train/original_loss": 1.0991315841674805,
1227
+ "objective/train/theoretical_loss": 3.336902794278807,
1228
+ "objective/train/tokens_used": 200682976,
1229
+ "theoretical_loss": 3.336902794278807,
1230
+ "tokens_seen": 3149398016
1231
+ },
1232
+ {
1233
+ "epoch": 0.55,
1234
+ "learning_rate": 4.579663730984788e-05,
1235
+ "loss": 1.1872,
1236
+ "theoretical_loss": 3.3368503891770964,
1237
+ "tokens_seen": 3150053376
1238
+ },
1239
+ {
1240
+ "epoch": 0.55,
1241
+ "learning_rate": 4.539631705364292e-05,
1242
+ "loss": 1.1918,
1243
+ "theoretical_loss": 3.336745620828,
1244
+ "tokens_seen": 3151364096
1245
+ },
1246
+ {
1247
+ "epoch": 0.55,
1248
+ "learning_rate": 4.499599679743795e-05,
1249
+ "loss": 1.2037,
1250
+ "theoretical_loss": 3.3366409082405823,
1251
+ "tokens_seen": 3152674816
1252
+ },
1253
+ {
1254
+ "epoch": 0.56,
1255
+ "learning_rate": 4.459567654123299e-05,
1256
+ "loss": 1.1804,
1257
+ "theoretical_loss": 3.3365362513619994,
1258
+ "tokens_seen": 3153985536
1259
+ },
1260
+ {
1261
+ "epoch": 0.56,
1262
+ "learning_rate": 4.419535628502802e-05,
1263
+ "loss": 1.178,
1264
+ "theoretical_loss": 3.336431650139481,
1265
+ "tokens_seen": 3155296256
1266
+ },
1267
+ {
1268
+ "epoch": 0.57,
1269
+ "learning_rate": 4.379503602882306e-05,
1270
+ "loss": 1.1858,
1271
+ "theoretical_loss": 3.336327104520328,
1272
+ "tokens_seen": 3156606976
1273
+ },
1274
+ {
1275
+ "epoch": 0.57,
1276
+ "objective/train/docs_used": 1058478,
1277
+ "objective/train/instantaneous_batch_size": 8,
1278
+ "objective/train/instantaneous_microbatch_size": 8192,
1279
+ "objective/train/original_loss": 1.1500177383422852,
1280
+ "objective/train/theoretical_loss": 3.336248731764011,
1281
+ "objective/train/tokens_used": 208874976,
1282
+ "theoretical_loss": 3.336248731764011,
1283
+ "tokens_seen": 3157590016
1284
+ },
1285
+ {
1286
+ "epoch": 0.57,
1287
+ "learning_rate": 4.3394715772618095e-05,
1288
+ "loss": 1.1836,
1289
+ "theoretical_loss": 3.3362226144519136,
1290
+ "tokens_seen": 3157917696
1291
+ },
1292
+ {
1293
+ "epoch": 0.57,
1294
+ "learning_rate": 4.2994395516413134e-05,
1295
+ "loss": 1.1917,
1296
+ "theoretical_loss": 3.336118179881682,
1297
+ "tokens_seen": 3159228416
1298
+ },
1299
+ {
1300
+ "epoch": 0.58,
1301
+ "learning_rate": 4.2594075260208167e-05,
1302
+ "loss": 1.1829,
1303
+ "theoretical_loss": 3.336013800757149,
1304
+ "tokens_seen": 3160539136
1305
+ },
1306
+ {
1307
+ "epoch": 0.58,
1308
+ "learning_rate": 4.2193755004003206e-05,
1309
+ "loss": 1.1744,
1310
+ "theoretical_loss": 3.3359094770259023,
1311
+ "tokens_seen": 3161849856
1312
+ },
1313
+ {
1314
+ "epoch": 0.59,
1315
+ "learning_rate": 4.179343474779824e-05,
1316
+ "loss": 1.1823,
1317
+ "theoretical_loss": 3.3358052086356,
1318
+ "tokens_seen": 3163160576
1319
+ },
1320
+ {
1321
+ "epoch": 0.59,
1322
+ "learning_rate": 4.139311449159328e-05,
1323
+ "loss": 1.1667,
1324
+ "theoretical_loss": 3.3357009955339727,
1325
+ "tokens_seen": 3164471296
1326
+ },
1327
+ {
1328
+ "epoch": 0.59,
1329
+ "objective/train/docs_used": 1061123,
1330
+ "objective/train/instantaneous_batch_size": 8,
1331
+ "objective/train/instantaneous_microbatch_size": 8192,
1332
+ "objective/train/original_loss": 0.7507131099700928,
1333
+ "objective/train/theoretical_loss": 3.33559683766882,
1334
+ "objective/train/tokens_used": 217066976,
1335
+ "theoretical_loss": 3.33559683766882,
1336
+ "tokens_seen": 3165782016
1337
+ },
1338
+ {
1339
+ "epoch": 0.59,
1340
+ "learning_rate": 4.099279423538831e-05,
1341
+ "loss": 1.1938,
1342
+ "theoretical_loss": 3.33559683766882,
1343
+ "tokens_seen": 3165782016
1344
+ },
1345
+ {
1346
+ "epoch": 0.6,
1347
+ "learning_rate": 4.059247397918335e-05,
1348
+ "loss": 1.1858,
1349
+ "theoretical_loss": 3.335492734988014,
1350
+ "tokens_seen": 3167092736
1351
+ },
1352
+ {
1353
+ "epoch": 0.6,
1354
+ "learning_rate": 4.019215372297839e-05,
1355
+ "loss": 1.1758,
1356
+ "theoretical_loss": 3.3353886874394973,
1357
+ "tokens_seen": 3168403456
1358
+ },
1359
+ {
1360
+ "epoch": 0.61,
1361
+ "learning_rate": 3.979183346677342e-05,
1362
+ "loss": 1.1935,
1363
+ "theoretical_loss": 3.3352846949712824,
1364
+ "tokens_seen": 3169714176
1365
+ },
1366
+ {
1367
+ "epoch": 0.61,
1368
+ "learning_rate": 3.939151321056846e-05,
1369
+ "loss": 1.1835,
1370
+ "theoretical_loss": 3.3351807575314525,
1371
+ "tokens_seen": 3171024896
1372
+ },
1373
+ {
1374
+ "epoch": 0.61,
1375
+ "learning_rate": 3.899119295436349e-05,
1376
+ "loss": 1.1861,
1377
+ "theoretical_loss": 3.3350768750681623,
1378
+ "tokens_seen": 3172335616
1379
+ },
1380
+ {
1381
+ "epoch": 0.62,
1382
+ "learning_rate": 3.859087269815853e-05,
1383
+ "loss": 1.2016,
1384
+ "theoretical_loss": 3.334973047529635,
1385
+ "tokens_seen": 3173646336
1386
+ },
1387
+ {
1388
+ "epoch": 0.62,
1389
+ "objective/train/docs_used": 1063909,
1390
+ "objective/train/instantaneous_batch_size": 8,
1391
+ "objective/train/instantaneous_microbatch_size": 8192,
1392
+ "objective/train/original_loss": 1.3667196035385132,
1393
+ "objective/train/theoretical_loss": 3.3349470992209365,
1394
+ "objective/train/tokens_used": 225258976,
1395
+ "theoretical_loss": 3.3349470992209365,
1396
+ "tokens_seen": 3173974016
1397
+ },
1398
+ {
1399
+ "epoch": 0.62,
1400
+ "learning_rate": 3.8190552441953564e-05,
1401
+ "loss": 1.208,
1402
+ "theoretical_loss": 3.3348692748641646,
1403
+ "tokens_seen": 3174957056
1404
+ },
1405
+ {
1406
+ "epoch": 0.63,
1407
+ "learning_rate": 3.77902321857486e-05,
1408
+ "loss": 1.1676,
1409
+ "theoretical_loss": 3.3347655570201162,
1410
+ "tokens_seen": 3176267776
1411
+ },
1412
+ {
1413
+ "epoch": 0.63,
1414
+ "learning_rate": 3.7389911929543636e-05,
1415
+ "loss": 1.176,
1416
+ "theoretical_loss": 3.334661893945923,
1417
+ "tokens_seen": 3177578496
1418
+ },
1419
+ {
1420
+ "epoch": 0.63,
1421
+ "learning_rate": 3.6989591673338675e-05,
1422
+ "loss": 1.1655,
1423
+ "theoretical_loss": 3.3345582855900893,
1424
+ "tokens_seen": 3178889216
1425
+ },
1426
+ {
1427
+ "epoch": 0.64,
1428
+ "learning_rate": 3.658927141713371e-05,
1429
+ "loss": 1.2031,
1430
+ "theoretical_loss": 3.334454731901188,
1431
+ "tokens_seen": 3180199936
1432
+ },
1433
+ {
1434
+ "epoch": 0.64,
1435
+ "learning_rate": 3.6188951160928746e-05,
1436
+ "loss": 1.1675,
1437
+ "theoretical_loss": 3.3343512328278626,
1438
+ "tokens_seen": 3181510656
1439
+ },
1440
+ {
1441
+ "epoch": 0.64,
1442
+ "objective/train/docs_used": 1066709,
1443
+ "objective/train/instantaneous_batch_size": 8,
1444
+ "objective/train/instantaneous_microbatch_size": 8192,
1445
+ "objective/train/original_loss": 1.2202755212783813,
1446
+ "objective/train/theoretical_loss": 3.3342995037560104,
1447
+ "objective/train/tokens_used": 233450976,
1448
+ "theoretical_loss": 3.3342995037560104,
1449
+ "tokens_seen": 3182166016
1450
+ },
1451
+ {
1452
+ "epoch": 0.65,
1453
+ "learning_rate": 3.578863090472378e-05,
1454
+ "loss": 1.1818,
1455
+ "theoretical_loss": 3.3342477883188244,
1456
+ "tokens_seen": 3182821376
1457
+ },
1458
+ {
1459
+ "epoch": 0.65,
1460
+ "learning_rate": 3.538831064851882e-05,
1461
+ "loss": 1.1759,
1462
+ "theoretical_loss": 3.334144398322856,
1463
+ "tokens_seen": 3184132096
1464
+ },
1465
+ {
1466
+ "epoch": 0.65,
1467
+ "learning_rate": 3.498799039231385e-05,
1468
+ "loss": 1.199,
1469
+ "theoretical_loss": 3.3340410627888075,
1470
+ "tokens_seen": 3185442816
1471
+ },
1472
+ {
1473
+ "epoch": 0.66,
1474
+ "learning_rate": 3.458767013610889e-05,
1475
+ "loss": 1.1651,
1476
+ "theoretical_loss": 3.3339377816655986,
1477
+ "tokens_seen": 3186753536
1478
+ },
1479
+ {
1480
+ "epoch": 0.66,
1481
+ "learning_rate": 3.418734987990392e-05,
1482
+ "loss": 1.1638,
1483
+ "theoretical_loss": 3.333834554902218,
1484
+ "tokens_seen": 3188064256
1485
+ },
1486
+ {
1487
+ "epoch": 0.67,
1488
+ "learning_rate": 3.378702962369896e-05,
1489
+ "loss": 1.1793,
1490
+ "theoretical_loss": 3.333731382447723,
1491
+ "tokens_seen": 3189374976
1492
+ },
1493
+ {
1494
+ "epoch": 0.67,
1495
+ "objective/train/docs_used": 1069727,
1496
+ "objective/train/instantaneous_batch_size": 8,
1497
+ "objective/train/instantaneous_microbatch_size": 8192,
1498
+ "objective/train/original_loss": 1.1584644317626953,
1499
+ "objective/train/theoretical_loss": 3.3336540387164515,
1500
+ "objective/train/tokens_used": 241642976,
1501
+ "theoretical_loss": 3.3336540387164515,
1502
+ "tokens_seen": 3190358016
1503
+ },
1504
+ {
1505
+ "epoch": 0.67,
1506
+ "learning_rate": 3.3386709367493994e-05,
1507
+ "loss": 1.1984,
1508
+ "theoretical_loss": 3.3336282642512396,
1509
+ "tokens_seen": 3190685696
1510
+ },
1511
+ {
1512
+ "epoch": 0.67,
1513
+ "learning_rate": 3.298638911128903e-05,
1514
+ "loss": 1.2295,
1515
+ "theoretical_loss": 3.3335252002619624,
1516
+ "tokens_seen": 3191996416
1517
+ },
1518
+ {
1519
+ "epoch": 0.68,
1520
+ "learning_rate": 3.2586068855084065e-05,
1521
+ "loss": 1.1866,
1522
+ "theoretical_loss": 3.3334221904291543,
1523
+ "tokens_seen": 3193307136
1524
+ },
1525
+ {
1526
+ "epoch": 0.68,
1527
+ "learning_rate": 3.2185748598879104e-05,
1528
+ "loss": 1.2311,
1529
+ "theoretical_loss": 3.3333192347021465,
1530
+ "tokens_seen": 3194617856
1531
+ },
1532
+ {
1533
+ "epoch": 0.69,
1534
+ "learning_rate": 3.178542834267414e-05,
1535
+ "loss": 1.1745,
1536
+ "theoretical_loss": 3.3332163330303386,
1537
+ "tokens_seen": 3195928576
1538
+ },
1539
+ {
1540
+ "epoch": 0.69,
1541
+ "learning_rate": 3.1385108086469176e-05,
1542
+ "loss": 1.1539,
1543
+ "theoretical_loss": 3.3331134853631976,
1544
+ "tokens_seen": 3197239296
1545
+ },
1546
+ {
1547
+ "epoch": 0.69,
1548
+ "objective/train/docs_used": 1072138,
1549
+ "objective/train/instantaneous_batch_size": 8,
1550
+ "objective/train/instantaneous_microbatch_size": 8192,
1551
+ "objective/train/original_loss": 1.353588342666626,
1552
+ "objective/train/theoretical_loss": 3.3330106916502595,
1553
+ "objective/train/tokens_used": 249834976,
1554
+ "theoretical_loss": 3.3330106916502595,
1555
+ "tokens_seen": 3198550016
1556
+ },
1557
+ {
1558
+ "epoch": 0.69,
1559
+ "learning_rate": 3.098478783026421e-05,
1560
+ "loss": 1.1742,
1561
+ "theoretical_loss": 3.3330106916502595,
1562
+ "tokens_seen": 3198550016
1563
+ },
1564
+ {
1565
+ "epoch": 0.7,
1566
+ "learning_rate": 3.058446757405925e-05,
1567
+ "loss": 1.1751,
1568
+ "theoretical_loss": 3.332907951841127,
1569
+ "tokens_seen": 3199860736
1570
+ },
1571
+ {
1572
+ "epoch": 0.7,
1573
+ "learning_rate": 3.0184147317854284e-05,
1574
+ "loss": 1.1712,
1575
+ "theoretical_loss": 3.332805265885471,
1576
+ "tokens_seen": 3201171456
1577
+ },
1578
+ {
1579
+ "epoch": 0.71,
1580
+ "learning_rate": 2.9783827061649323e-05,
1581
+ "loss": 1.1948,
1582
+ "theoretical_loss": 3.33270263373303,
1583
+ "tokens_seen": 3202482176
1584
+ },
1585
+ {
1586
+ "epoch": 0.71,
1587
+ "learning_rate": 2.9383506805444355e-05,
1588
+ "loss": 1.2016,
1589
+ "theoretical_loss": 3.33260005533361,
1590
+ "tokens_seen": 3203792896
1591
+ },
1592
+ {
1593
+ "epoch": 0.71,
1594
+ "learning_rate": 2.8983186549239394e-05,
1595
+ "loss": 1.2092,
1596
+ "theoretical_loss": 3.3324975306370845,
1597
+ "tokens_seen": 3205103616
1598
+ },
1599
+ {
1600
+ "epoch": 0.72,
1601
+ "learning_rate": 2.8582866293034427e-05,
1602
+ "loss": 1.2189,
1603
+ "theoretical_loss": 3.3323950595933933,
1604
+ "tokens_seen": 3206414336
1605
+ },
1606
+ {
1607
+ "epoch": 0.72,
1608
+ "objective/train/docs_used": 1074937,
1609
+ "objective/train/instantaneous_batch_size": 8,
1610
+ "objective/train/instantaneous_microbatch_size": 8192,
1611
+ "objective/train/original_loss": 1.3788819313049316,
1612
+ "objective/train/theoretical_loss": 3.332369450209866,
1613
+ "objective/train/tokens_used": 258026976,
1614
+ "theoretical_loss": 3.332369450209866,
1615
+ "tokens_seen": 3206742016
1616
+ },
1617
+ {
1618
+ "epoch": 0.72,
1619
+ "learning_rate": 2.8182546036829466e-05,
1620
+ "loss": 1.1972,
1621
+ "theoretical_loss": 3.3322926421525447,
1622
+ "tokens_seen": 3207725056
1623
+ },
1624
+ {
1625
+ "epoch": 0.73,
1626
+ "learning_rate": 2.77822257806245e-05,
1627
+ "loss": 1.2114,
1628
+ "theoretical_loss": 3.332190278264613,
1629
+ "tokens_seen": 3209035776
1630
+ },
1631
+ {
1632
+ "epoch": 0.73,
1633
+ "learning_rate": 2.7381905524419538e-05,
1634
+ "loss": 1.1867,
1635
+ "theoretical_loss": 3.3320879678797395,
1636
+ "tokens_seen": 3210346496
1637
+ },
1638
+ {
1639
+ "epoch": 0.73,
1640
+ "learning_rate": 2.698158526821457e-05,
1641
+ "loss": 1.1942,
1642
+ "theoretical_loss": 3.3319857109481323,
1643
+ "tokens_seen": 3211657216
1644
+ },
1645
+ {
1646
+ "epoch": 0.74,
1647
+ "learning_rate": 2.658126501200961e-05,
1648
+ "loss": 1.2116,
1649
+ "theoretical_loss": 3.331883507420067,
1650
+ "tokens_seen": 3212967936
1651
+ },
1652
+ {
1653
+ "epoch": 0.74,
1654
+ "learning_rate": 2.618094475580464e-05,
1655
+ "loss": 1.2044,
1656
+ "theoretical_loss": 3.3317813572458834,
1657
+ "tokens_seen": 3214278656
1658
+ },
1659
+ {
1660
+ "debugging/Compilability": 0.0,
1661
+ "debugging/distinct-1-grams": 0.7649685585732098,
1662
+ "debugging/entropy-1-grams": 4.842071493424633,
1663
+ "debugging/length": 542.8,
1664
+ "debugging/num_segments": 5,
1665
+ "debugging/score": 0.05065775917820918,
1666
+ "debugging/score_std": 0.023966644867425342,
1667
+ "epoch": 0.74,
1668
+ "objective/train/docs_used": 1077839,
1669
+ "objective/train/instantaneous_batch_size": 8,
1670
+ "objective/train/instantaneous_microbatch_size": 8192,
1671
+ "objective/train/original_loss": 1.1823174953460693,
1672
+ "objective/train/theoretical_loss": 3.331730302150998,
1673
+ "objective/train/tokens_used": 266218976,
1674
+ "theoretical_loss": 3.331730302150998,
1675
+ "tokens_seen": 3214934016
1676
+ },
1677
+ {
1678
+ "epoch": 0.74,
1679
+ "learning_rate": 2.578062449959968e-05,
1680
+ "loss": 1.1801,
1681
+ "theoretical_loss": 3.331679260375991,
1682
+ "tokens_seen": 3215589376
1683
+ },
1684
+ {
1685
+ "epoch": 0.75,
1686
+ "learning_rate": 2.5380304243394713e-05,
1687
+ "loss": 1.1625,
1688
+ "theoretical_loss": 3.331577216760862,
1689
+ "tokens_seen": 3216900096
1690
+ },
1691
+ {
1692
+ "epoch": 0.75,
1693
+ "learning_rate": 2.4979983987189752e-05,
1694
+ "loss": 1.2035,
1695
+ "theoretical_loss": 3.331475226351038,
1696
+ "tokens_seen": 3218210816
1697
+ },
1698
+ {
1699
+ "epoch": 0.76,
1700
+ "learning_rate": 2.4579663730984788e-05,
1701
+ "loss": 1.1689,
1702
+ "theoretical_loss": 3.331373289097125,
1703
+ "tokens_seen": 3219521536
1704
+ },
1705
+ {
1706
+ "epoch": 0.76,
1707
+ "learning_rate": 2.4179343474779824e-05,
1708
+ "loss": 1.1771,
1709
+ "theoretical_loss": 3.3312714049497947,
1710
+ "tokens_seen": 3220832256
1711
+ },
1712
+ {
1713
+ "epoch": 0.76,
1714
+ "learning_rate": 2.377902321857486e-05,
1715
+ "loss": 1.2113,
1716
+ "theoretical_loss": 3.3311695738597855,
1717
+ "tokens_seen": 3222142976
1718
+ },
1719
+ {
1720
+ "epoch": 0.77,
1721
+ "objective/train/docs_used": 1080652,
1722
+ "objective/train/instantaneous_batch_size": 8,
1723
+ "objective/train/instantaneous_microbatch_size": 8192,
1724
+ "objective/train/original_loss": 1.0484637022018433,
1725
+ "objective/train/theoretical_loss": 3.3310932353315486,
1726
+ "objective/train/tokens_used": 274410976,
1727
+ "theoretical_loss": 3.3310932353315486,
1728
+ "tokens_seen": 3223126016
1729
+ },
1730
+ {
1731
+ "epoch": 0.77,
1732
+ "learning_rate": 2.3378702962369896e-05,
1733
+ "loss": 1.1704,
1734
+ "theoretical_loss": 3.3310677957779014,
1735
+ "tokens_seen": 3223453696
1736
+ },
1737
+ {
1738
+ "epoch": 0.77,
1739
+ "learning_rate": 2.297838270616493e-05,
1740
+ "loss": 1.2005,
1741
+ "theoretical_loss": 3.330966070655011,
1742
+ "tokens_seen": 3224764416
1743
+ },
1744
+ {
1745
+ "epoch": 0.78,
1746
+ "learning_rate": 2.257806244995997e-05,
1747
+ "loss": 1.1869,
1748
+ "theoretical_loss": 3.330864398442049,
1749
+ "tokens_seen": 3226075136
1750
+ },
1751
+ {
1752
+ "epoch": 0.78,
1753
+ "learning_rate": 2.2177742193755007e-05,
1754
+ "loss": 1.1755,
1755
+ "theoretical_loss": 3.3307627790900165,
1756
+ "tokens_seen": 3227385856
1757
+ },
1758
+ {
1759
+ "epoch": 0.78,
1760
+ "learning_rate": 2.1777421937550042e-05,
1761
+ "loss": 1.1833,
1762
+ "theoretical_loss": 3.330661212549978,
1763
+ "tokens_seen": 3228696576
1764
+ },
1765
+ {
1766
+ "epoch": 0.79,
1767
+ "learning_rate": 2.1377101681345078e-05,
1768
+ "loss": 1.21,
1769
+ "theoretical_loss": 3.330559698773065,
1770
+ "tokens_seen": 3230007296
1771
+ },
1772
+ {
1773
+ "epoch": 0.79,
1774
+ "objective/train/docs_used": 1083207,
1775
+ "objective/train/instantaneous_batch_size": 8,
1776
+ "objective/train/instantaneous_microbatch_size": 8192,
1777
+ "objective/train/original_loss": 1.077799677848816,
1778
+ "objective/train/theoretical_loss": 3.330458237710472,
1779
+ "objective/train/tokens_used": 282602976,
1780
+ "theoretical_loss": 3.330458237710472,
1781
+ "tokens_seen": 3231318016
1782
+ },
1783
+ {
1784
+ "epoch": 0.79,
1785
+ "learning_rate": 2.0976781425140114e-05,
1786
+ "loss": 1.2286,
1787
+ "theoretical_loss": 3.330458237710472,
1788
+ "tokens_seen": 3231318016
1789
+ },
1790
+ {
1791
+ "epoch": 0.8,
1792
+ "learning_rate": 2.057646116893515e-05,
1793
+ "loss": 1.2068,
1794
+ "theoretical_loss": 3.3303568293134607,
1795
+ "tokens_seen": 3232628736
1796
+ },
1797
+ {
1798
+ "epoch": 0.8,
1799
+ "learning_rate": 2.0176140912730186e-05,
1800
+ "loss": 1.2013,
1801
+ "theoretical_loss": 3.3302554735333554,
1802
+ "tokens_seen": 3233939456
1803
+ },
1804
+ {
1805
+ "epoch": 0.8,
1806
+ "learning_rate": 1.977582065652522e-05,
1807
+ "loss": 1.1972,
1808
+ "theoretical_loss": 3.3301541703215465,
1809
+ "tokens_seen": 3235250176
1810
+ },
1811
+ {
1812
+ "epoch": 0.81,
1813
+ "learning_rate": 1.9375500400320257e-05,
1814
+ "loss": 1.1842,
1815
+ "theoretical_loss": 3.330052919629489,
1816
+ "tokens_seen": 3236560896
1817
+ },
1818
+ {
1819
+ "epoch": 0.81,
1820
+ "learning_rate": 1.8975180144115293e-05,
1821
+ "loss": 1.214,
1822
+ "theoretical_loss": 3.329951721408701,
1823
+ "tokens_seen": 3237871616
1824
+ },
1825
+ {
1826
+ "epoch": 0.82,
1827
+ "learning_rate": 1.857485988791033e-05,
1828
+ "loss": 1.1823,
1829
+ "theoretical_loss": 3.3298505756107666,
1830
+ "tokens_seen": 3239182336
1831
+ },
1832
+ {
1833
+ "epoch": 0.82,
1834
+ "objective/train/docs_used": 1085940,
1835
+ "objective/train/instantaneous_batch_size": 8,
1836
+ "objective/train/instantaneous_microbatch_size": 8192,
1837
+ "objective/train/original_loss": 0.996847927570343,
1838
+ "objective/train/theoretical_loss": 3.3298252973466864,
1839
+ "objective/train/tokens_used": 290794976,
1840
+ "theoretical_loss": 3.3298252973466864,
1841
+ "tokens_seen": 3239510016
1842
+ },
1843
+ {
1844
+ "epoch": 0.82,
1845
+ "learning_rate": 1.8174539631705365e-05,
1846
+ "loss": 1.1685,
1847
+ "theoretical_loss": 3.3297494821873337,
1848
+ "tokens_seen": 3240493056
1849
+ },
1850
+ {
1851
+ "epoch": 0.82,
1852
+ "learning_rate": 1.77742193755004e-05,
1853
+ "loss": 1.1904,
1854
+ "theoretical_loss": 3.3296484410901135,
1855
+ "tokens_seen": 3241803776
1856
+ },
1857
+ {
1858
+ "epoch": 0.83,
1859
+ "learning_rate": 1.7373899119295436e-05,
1860
+ "loss": 1.1923,
1861
+ "theoretical_loss": 3.329547452270882,
1862
+ "tokens_seen": 3243114496
1863
+ },
1864
+ {
1865
+ "epoch": 0.83,
1866
+ "learning_rate": 1.6973578863090472e-05,
1867
+ "loss": 1.1947,
1868
+ "theoretical_loss": 3.3294465156814788,
1869
+ "tokens_seen": 3244425216
1870
+ },
1871
+ {
1872
+ "epoch": 0.84,
1873
+ "learning_rate": 1.6573258606885508e-05,
1874
+ "loss": 1.198,
1875
+ "theoretical_loss": 3.3293456312738074,
1876
+ "tokens_seen": 3245735936
1877
+ },
1878
+ {
1879
+ "epoch": 0.84,
1880
+ "learning_rate": 1.6172938350680544e-05,
1881
+ "loss": 1.1954,
1882
+ "theoretical_loss": 3.3292447989998353,
1883
+ "tokens_seen": 3247046656
1884
+ },
1885
+ {
1886
+ "epoch": 0.84,
1887
+ "objective/train/docs_used": 1088712,
1888
+ "objective/train/instantaneous_batch_size": 8,
1889
+ "objective/train/instantaneous_microbatch_size": 8192,
1890
+ "objective/train/original_loss": 1.440091848373413,
1891
+ "objective/train/theoretical_loss": 3.3291944023979934,
1892
+ "objective/train/tokens_used": 298986976,
1893
+ "theoretical_loss": 3.3291944023979934,
1894
+ "tokens_seen": 3247702016
1895
+ },
1896
+ {
1897
+ "epoch": 0.84,
1898
+ "learning_rate": 1.577261809447558e-05,
1899
+ "loss": 1.1908,
1900
+ "theoretical_loss": 3.329144018811593,
1901
+ "tokens_seen": 3248357376
1902
+ },
1903
+ {
1904
+ "epoch": 0.85,
1905
+ "learning_rate": 1.5372297838270615e-05,
1906
+ "loss": 1.2103,
1907
+ "theoretical_loss": 3.3290432906611747,
1908
+ "tokens_seen": 3249668096
1909
+ },
1910
+ {
1911
+ "epoch": 0.85,
1912
+ "learning_rate": 1.4971977582065653e-05,
1913
+ "loss": 1.1834,
1914
+ "theoretical_loss": 3.328942614500739,
1915
+ "tokens_seen": 3250978816
1916
+ },
1917
+ {
1918
+ "epoch": 0.86,
1919
+ "learning_rate": 1.4571657325860689e-05,
1920
+ "loss": 1.187,
1921
+ "theoretical_loss": 3.328841990282505,
1922
+ "tokens_seen": 3252289536
1923
+ },
1924
+ {
1925
+ "epoch": 0.86,
1926
+ "learning_rate": 1.4171337069655724e-05,
1927
+ "loss": 1.1827,
1928
+ "theoretical_loss": 3.3287414179587578,
1929
+ "tokens_seen": 3253600256
1930
+ },
1931
+ {
1932
+ "epoch": 0.86,
1933
+ "learning_rate": 1.377101681345076e-05,
1934
+ "loss": 1.2075,
1935
+ "theoretical_loss": 3.3286408974818444,
1936
+ "tokens_seen": 3254910976
1937
+ },
1938
+ {
1939
+ "epoch": 0.87,
1940
+ "objective/train/docs_used": 1091411,
1941
+ "objective/train/instantaneous_batch_size": 8,
1942
+ "objective/train/instantaneous_microbatch_size": 8192,
1943
+ "objective/train/original_loss": 1.2068538665771484,
1944
+ "objective/train/theoretical_loss": 3.3285655411200135,
1945
+ "objective/train/tokens_used": 307178976,
1946
+ "theoretical_loss": 3.3285655411200135,
1947
+ "tokens_seen": 3255894016
1948
+ },
1949
+ {
1950
+ "epoch": 0.87,
1951
+ "learning_rate": 1.3370696557245796e-05,
1952
+ "loss": 1.1996,
1953
+ "theoretical_loss": 3.328540428804175,
1954
+ "tokens_seen": 3256221696
1955
+ },
1956
+ {
1957
+ "epoch": 0.87,
1958
+ "learning_rate": 1.2970376301040832e-05,
1959
+ "loss": 1.2049,
1960
+ "theoretical_loss": 3.3284400118782216,
1961
+ "tokens_seen": 3257532416
1962
+ },
1963
+ {
1964
+ "epoch": 0.88,
1965
+ "learning_rate": 1.2570056044835871e-05,
1966
+ "loss": 1.1909,
1967
+ "theoretical_loss": 3.3283396466565205,
1968
+ "tokens_seen": 3258843136
1969
+ },
1970
+ {
1971
+ "epoch": 0.88,
1972
+ "learning_rate": 1.2169735788630905e-05,
1973
+ "loss": 1.2063,
1974
+ "theoretical_loss": 3.3282393330916693,
1975
+ "tokens_seen": 3260153856
1976
+ },
1977
+ {
1978
+ "epoch": 0.88,
1979
+ "learning_rate": 1.1769415532425941e-05,
1980
+ "loss": 1.2061,
1981
+ "theoretical_loss": 3.3281390711363286,
1982
+ "tokens_seen": 3261464576
1983
+ },
1984
+ {
1985
+ "epoch": 0.89,
1986
+ "learning_rate": 1.1369095276220977e-05,
1987
+ "loss": 1.2009,
1988
+ "theoretical_loss": 3.328038860743222,
1989
+ "tokens_seen": 3262775296
1990
+ },
1991
+ {
1992
+ "epoch": 0.89,
1993
+ "objective/train/docs_used": 1094282,
1994
+ "objective/train/instantaneous_batch_size": 8,
1995
+ "objective/train/instantaneous_microbatch_size": 8192,
1996
+ "objective/train/original_loss": 1.0986521244049072,
1997
+ "objective/train/theoretical_loss": 3.3279387018651336,
1998
+ "objective/train/tokens_used": 315370976,
1999
+ "theoretical_loss": 3.3279387018651336,
2000
+ "tokens_seen": 3264086016
2001
+ },
2002
+ {
2003
+ "epoch": 0.89,
2004
+ "learning_rate": 1.0968775020016013e-05,
2005
+ "loss": 1.1938,
2006
+ "theoretical_loss": 3.3279387018651336,
2007
+ "tokens_seen": 3264086016
2008
+ },
2009
+ {
2010
+ "epoch": 0.9,
2011
+ "learning_rate": 1.0568454763811048e-05,
2012
+ "loss": 1.1947,
2013
+ "theoretical_loss": 3.3278385944549114,
2014
+ "tokens_seen": 3265396736
2015
+ },
2016
+ {
2017
+ "epoch": 0.9,
2018
+ "learning_rate": 1.0168134507606084e-05,
2019
+ "loss": 1.1832,
2020
+ "theoretical_loss": 3.327738538465465,
2021
+ "tokens_seen": 3266707456
2022
+ },
2023
+ {
2024
+ "epoch": 0.9,
2025
+ "learning_rate": 9.76781425140112e-06,
2026
+ "loss": 1.212,
2027
+ "theoretical_loss": 3.327638533849766,
2028
+ "tokens_seen": 3268018176
2029
+ },
2030
+ {
2031
+ "epoch": 0.91,
2032
+ "learning_rate": 9.367493995196158e-06,
2033
+ "loss": 1.1694,
2034
+ "theoretical_loss": 3.3275385805608475,
2035
+ "tokens_seen": 3269328896
2036
+ },
2037
+ {
2038
+ "epoch": 0.91,
2039
+ "learning_rate": 8.967173738991193e-06,
2040
+ "loss": 1.197,
2041
+ "theoretical_loss": 3.327438678551805,
2042
+ "tokens_seen": 3270639616
2043
+ },
2044
+ {
2045
+ "epoch": 0.92,
2046
+ "learning_rate": 8.56685348278623e-06,
2047
+ "loss": 1.183,
2048
+ "theoretical_loss": 3.327338827775794,
2049
+ "tokens_seen": 3271950336
2050
+ },
2051
+ {
2052
+ "epoch": 0.92,
2053
+ "objective/train/docs_used": 1096953,
2054
+ "objective/train/instantaneous_batch_size": 8,
2055
+ "objective/train/instantaneous_microbatch_size": 8192,
2056
+ "objective/train/original_loss": 1.1763838529586792,
2057
+ "objective/train/theoretical_loss": 3.3273138730814686,
2058
+ "objective/train/tokens_used": 323562976,
2059
+ "theoretical_loss": 3.3273138730814686,
2060
+ "tokens_seen": 3272278016
2061
+ },
2062
+ {
2063
+ "epoch": 0.92,
2064
+ "learning_rate": 8.166533226581267e-06,
2065
+ "loss": 1.1965,
2066
+ "theoretical_loss": 3.3272390281860336,
2067
+ "tokens_seen": 3273261056
2068
+ },
2069
+ {
2070
+ "epoch": 0.92,
2071
+ "learning_rate": 7.766212970376303e-06,
2072
+ "loss": 1.2072,
2073
+ "theoretical_loss": 3.3271392797358033,
2074
+ "tokens_seen": 3274571776
2075
+ },
2076
+ {
2077
+ "epoch": 0.93,
2078
+ "learning_rate": 7.365892714171338e-06,
2079
+ "loss": 1.1885,
2080
+ "theoretical_loss": 3.3270395823784438,
2081
+ "tokens_seen": 3275882496
2082
+ },
2083
+ {
2084
+ "epoch": 0.93,
2085
+ "learning_rate": 6.965572457966374e-06,
2086
+ "loss": 1.1976,
2087
+ "theoretical_loss": 3.3269399360673577,
2088
+ "tokens_seen": 3277193216
2089
+ },
2090
+ {
2091
+ "epoch": 0.94,
2092
+ "learning_rate": 6.56525220176141e-06,
2093
+ "loss": 1.1947,
2094
+ "theoretical_loss": 3.326840340756008,
2095
+ "tokens_seen": 3278503936
2096
+ },
2097
+ {
2098
+ "epoch": 0.94,
2099
+ "learning_rate": 6.164931945556445e-06,
2100
+ "loss": 1.1972,
2101
+ "theoretical_loss": 3.326740796397919,
2102
+ "tokens_seen": 3279814656
2103
+ },
2104
+ {
2105
+ "epoch": 0.94,
2106
+ "objective/train/docs_used": 1099304,
2107
+ "objective/train/instantaneous_batch_size": 8,
2108
+ "objective/train/instantaneous_microbatch_size": 8192,
2109
+ "objective/train/original_loss": 1.0297110080718994,
2110
+ "objective/train/theoretical_loss": 3.3266910433118406,
2111
+ "objective/train/tokens_used": 331754976,
2112
+ "theoretical_loss": 3.3266910433118406,
2113
+ "tokens_seen": 3280470016
2114
+ },
2115
+ {
2116
+ "epoch": 0.94,
2117
+ "learning_rate": 5.764611689351482e-06,
2118
+ "loss": 1.2128,
2119
+ "theoretical_loss": 3.326641302946676,
2120
+ "tokens_seen": 3281125376
2121
+ },
2122
+ {
2123
+ "epoch": 0.95,
2124
+ "learning_rate": 5.364291433146517e-06,
2125
+ "loss": 1.2015,
2126
+ "theoretical_loss": 3.3265418603559254,
2127
+ "tokens_seen": 3282436096
2128
+ },
2129
+ {
2130
+ "epoch": 0.95,
2131
+ "learning_rate": 4.963971176941554e-06,
2132
+ "loss": 1.2106,
2133
+ "theoretical_loss": 3.3264424685793736,
2134
+ "tokens_seen": 3283746816
2135
+ },
2136
+ {
2137
+ "epoch": 0.95,
2138
+ "learning_rate": 4.56365092073659e-06,
2139
+ "loss": 1.2275,
2140
+ "theoretical_loss": 3.3263431275707873,
2141
+ "tokens_seen": 3285057536
2142
+ },
2143
+ {
2144
+ "epoch": 0.96,
2145
+ "learning_rate": 4.163330664531626e-06,
2146
+ "loss": 1.2281,
2147
+ "theoretical_loss": 3.326243837283995,
2148
+ "tokens_seen": 3286368256
2149
+ },
2150
+ {
2151
+ "epoch": 0.96,
2152
+ "learning_rate": 3.7630104083266615e-06,
2153
+ "loss": 1.1995,
2154
+ "theoretical_loss": 3.3261445976728847,
2155
+ "tokens_seen": 3287678976
2156
+ },
2157
+ {
2158
+ "epoch": 0.97,
2159
+ "objective/train/docs_used": 1102162,
2160
+ "objective/train/instantaneous_batch_size": 8,
2161
+ "objective/train/instantaneous_microbatch_size": 8192,
2162
+ "objective/train/original_loss": 1.0746601819992065,
2163
+ "objective/train/theoretical_loss": 3.326070201192763,
2164
+ "objective/train/tokens_used": 339946976,
2165
+ "theoretical_loss": 3.326070201192763,
2166
+ "tokens_seen": 3288662016
2167
+ },
2168
+ {
2169
+ "epoch": 0.97,
2170
+ "learning_rate": 3.3626901521216973e-06,
2171
+ "loss": 1.2202,
2172
+ "theoretical_loss": 3.326045408691405,
2173
+ "tokens_seen": 3288989696
2174
+ },
2175
+ {
2176
+ "epoch": 0.97,
2177
+ "learning_rate": 2.9623698959167336e-06,
2178
+ "loss": 1.2118,
2179
+ "theoretical_loss": 3.325946270293564,
2180
+ "tokens_seen": 3290300416
2181
+ },
2182
+ {
2183
+ "epoch": 0.97,
2184
+ "learning_rate": 2.5620496397117694e-06,
2185
+ "loss": 1.224,
2186
+ "theoretical_loss": 3.325847182433431,
2187
+ "tokens_seen": 3291611136
2188
+ },
2189
+ {
2190
+ "epoch": 0.98,
2191
+ "learning_rate": 2.1617293835068056e-06,
2192
+ "loss": 1.2288,
2193
+ "theoretical_loss": 3.3257481450651345,
2194
+ "tokens_seen": 3292921856
2195
+ },
2196
+ {
2197
+ "epoch": 0.98,
2198
+ "learning_rate": 1.7614091273018416e-06,
2199
+ "loss": 1.2148,
2200
+ "theoretical_loss": 3.3256491581428627,
2201
+ "tokens_seen": 3294232576
2202
+ },
2203
+ {
2204
+ "epoch": 0.99,
2205
+ "learning_rate": 1.3610888710968776e-06,
2206
+ "loss": 1.2076,
2207
+ "theoretical_loss": 3.3255502216208646,
2208
+ "tokens_seen": 3295543296
2209
+ },
2210
+ {
2211
+ "debugging/Compilability": 0.0,
2212
+ "debugging/distinct-1-grams": 0.7111111111111111,
2213
+ "debugging/entropy-1-grams": 3.2405170087942627,
2214
+ "debugging/length": 581.0,
2215
+ "debugging/num_segments": 1,
2216
+ "debugging/score": 0.07228915662650602,
2217
+ "debugging/score_std": 0.0,
2218
+ "epoch": 0.99,
2219
+ "objective/train/docs_used": 1104945,
2220
+ "objective/train/instantaneous_batch_size": 8,
2221
+ "objective/train/instantaneous_microbatch_size": 8192,
2222
+ "objective/train/original_loss": 1.1754258871078491,
2223
+ "objective/train/theoretical_loss": 3.325451335453448,
2224
+ "objective/train/tokens_used": 348138976,
2225
+ "theoretical_loss": 3.325451335453448,
2226
+ "tokens_seen": 3296854016
2227
+ },
2228
+ {
2229
+ "epoch": 0.99,
2230
+ "learning_rate": 9.607686148919135e-07,
2231
+ "loss": 1.2225,
2232
+ "theoretical_loss": 3.325451335453448,
2233
+ "tokens_seen": 3296854016
2234
+ }
2235
+ ],
2236
+ "max_steps": 2524,
2237
+ "num_train_epochs": 9223372036854775807,
2238
+ "total_flos": 1.6722690048e+17,
2239
+ "trial_name": null,
2240
+ "trial_params": null
2241
+ }
checkpoint-2500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:62606ef86d425c76460a848b984d348a02cd2616fd10f4951e23cea6e71b3d5c
3
+ size 3375
checkpoint-2500/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:99d10e2348be5bf3094d6366f1766a617d2050c872fd06262b3c53a17240cb14
3
  size 456676457
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98886462090a2e5a4007173e734e370b8b19c5ceb0c27fcb434cd526974e31c3
3
  size 456676457