kanishka commited on
Commit
bbe342b
·
verified ·
1 Parent(s): 6556940

End of training

Browse files
Files changed (5) hide show
  1. README.md +14 -2
  2. all_results.json +16 -0
  3. eval_results.json +10 -0
  4. train_results.json +9 -0
  5. trainer_state.json +2399 -0
README.md CHANGED
@@ -2,11 +2,23 @@
2
  library_name: transformers
3
  tags:
4
  - generated_from_trainer
 
 
5
  metrics:
6
  - accuracy
7
  model-index:
8
  - name: opt-babylm2-clean-spacy-32k_seed-42_1e-3
9
- results: []
 
 
 
 
 
 
 
 
 
 
10
  ---
11
 
12
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -14,7 +26,7 @@ should probably proofread and complete it, then remove this comment. -->
14
 
15
  # opt-babylm2-clean-spacy-32k_seed-42_1e-3
16
 
17
- This model was trained from scratch on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
  - Loss: 3.0357
20
  - Accuracy: 0.4240
 
2
  library_name: transformers
3
  tags:
4
  - generated_from_trainer
5
+ datasets:
6
+ - kanishka/babylm2-clean-spacy
7
  metrics:
8
  - accuracy
9
  model-index:
10
  - name: opt-babylm2-clean-spacy-32k_seed-42_1e-3
11
+ results:
12
+ - task:
13
+ name: Causal Language Modeling
14
+ type: text-generation
15
+ dataset:
16
+ name: kanishka/babylm2-clean-spacy
17
+ type: kanishka/babylm2-clean-spacy
18
+ metrics:
19
+ - name: Accuracy
20
+ type: accuracy
21
+ value: 0.4239814649263961
22
  ---
23
 
24
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
26
 
27
  # opt-babylm2-clean-spacy-32k_seed-42_1e-3
28
 
29
+ This model was trained from scratch on the kanishka/babylm2-clean-spacy dataset.
30
  It achieves the following results on the evaluation set:
31
  - Loss: 3.0357
32
  - Accuracy: 0.4240
all_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "eval_accuracy": 0.4239814649263961,
4
+ "eval_loss": 3.035736322402954,
5
+ "eval_runtime": 112.5567,
6
+ "eval_samples": 52640,
7
+ "eval_samples_per_second": 467.675,
8
+ "eval_steps_per_second": 7.312,
9
+ "perplexity": 20.816299765730864,
10
+ "total_flos": 1.30265052217344e+18,
11
+ "train_loss": 2.6546336687942524,
12
+ "train_runtime": 44210.8645,
13
+ "train_samples": 498542,
14
+ "train_samples_per_second": 225.529,
15
+ "train_steps_per_second": 7.048
16
+ }
eval_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "eval_accuracy": 0.4239814649263961,
4
+ "eval_loss": 3.035736322402954,
5
+ "eval_runtime": 112.5567,
6
+ "eval_samples": 52640,
7
+ "eval_samples_per_second": 467.675,
8
+ "eval_steps_per_second": 7.312,
9
+ "perplexity": 20.816299765730864
10
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "total_flos": 1.30265052217344e+18,
4
+ "train_loss": 2.6546336687942524,
5
+ "train_runtime": 44210.8645,
6
+ "train_samples": 498542,
7
+ "train_samples_per_second": 225.529,
8
+ "train_steps_per_second": 7.048
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,2399 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 20.0,
5
+ "eval_steps": 500,
6
+ "global_step": 311600,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.06418485237483953,
13
+ "grad_norm": 1.043227195739746,
14
+ "learning_rate": 3.125e-05,
15
+ "loss": 6.1267,
16
+ "step": 1000
17
+ },
18
+ {
19
+ "epoch": 0.12836970474967907,
20
+ "grad_norm": 1.0838656425476074,
21
+ "learning_rate": 6.25e-05,
22
+ "loss": 4.5193,
23
+ "step": 2000
24
+ },
25
+ {
26
+ "epoch": 0.1925545571245186,
27
+ "grad_norm": 0.9696599841117859,
28
+ "learning_rate": 9.375e-05,
29
+ "loss": 4.1718,
30
+ "step": 3000
31
+ },
32
+ {
33
+ "epoch": 0.25673940949935814,
34
+ "grad_norm": 0.8352647423744202,
35
+ "learning_rate": 0.000125,
36
+ "loss": 3.9406,
37
+ "step": 4000
38
+ },
39
+ {
40
+ "epoch": 0.3209242618741977,
41
+ "grad_norm": 0.7503564953804016,
42
+ "learning_rate": 0.00015625,
43
+ "loss": 3.779,
44
+ "step": 5000
45
+ },
46
+ {
47
+ "epoch": 0.3851091142490372,
48
+ "grad_norm": 0.7538415193557739,
49
+ "learning_rate": 0.0001875,
50
+ "loss": 3.6577,
51
+ "step": 6000
52
+ },
53
+ {
54
+ "epoch": 0.4492939666238768,
55
+ "grad_norm": 0.6559199690818787,
56
+ "learning_rate": 0.00021875,
57
+ "loss": 3.5747,
58
+ "step": 7000
59
+ },
60
+ {
61
+ "epoch": 0.5134788189987163,
62
+ "grad_norm": 0.7036673426628113,
63
+ "learning_rate": 0.00025,
64
+ "loss": 3.5069,
65
+ "step": 8000
66
+ },
67
+ {
68
+ "epoch": 0.5776636713735558,
69
+ "grad_norm": 0.6067727208137512,
70
+ "learning_rate": 0.00028125000000000003,
71
+ "loss": 3.4443,
72
+ "step": 9000
73
+ },
74
+ {
75
+ "epoch": 0.6418485237483954,
76
+ "grad_norm": 0.5849200487136841,
77
+ "learning_rate": 0.0003125,
78
+ "loss": 3.4028,
79
+ "step": 10000
80
+ },
81
+ {
82
+ "epoch": 0.7060333761232349,
83
+ "grad_norm": 0.5407277345657349,
84
+ "learning_rate": 0.00034371875,
85
+ "loss": 3.3687,
86
+ "step": 11000
87
+ },
88
+ {
89
+ "epoch": 0.7702182284980744,
90
+ "grad_norm": 0.5096879601478577,
91
+ "learning_rate": 0.00037496875000000003,
92
+ "loss": 3.3356,
93
+ "step": 12000
94
+ },
95
+ {
96
+ "epoch": 0.834403080872914,
97
+ "grad_norm": 0.4796447455883026,
98
+ "learning_rate": 0.0004061875,
99
+ "loss": 3.2971,
100
+ "step": 13000
101
+ },
102
+ {
103
+ "epoch": 0.8985879332477535,
104
+ "grad_norm": 0.4423964321613312,
105
+ "learning_rate": 0.00043740625,
106
+ "loss": 3.2819,
107
+ "step": 14000
108
+ },
109
+ {
110
+ "epoch": 0.962772785622593,
111
+ "grad_norm": 0.3888336420059204,
112
+ "learning_rate": 0.00046865625,
113
+ "loss": 3.2533,
114
+ "step": 15000
115
+ },
116
+ {
117
+ "epoch": 1.0,
118
+ "eval_accuracy": 0.3827605191012575,
119
+ "eval_loss": 3.314154863357544,
120
+ "eval_runtime": 112.692,
121
+ "eval_samples_per_second": 467.114,
122
+ "eval_steps_per_second": 7.303,
123
+ "step": 15580
124
+ },
125
+ {
126
+ "epoch": 1.0269576379974326,
127
+ "grad_norm": 0.4023679196834564,
128
+ "learning_rate": 0.00049990625,
129
+ "loss": 3.2279,
130
+ "step": 16000
131
+ },
132
+ {
133
+ "epoch": 1.0911424903722722,
134
+ "grad_norm": 0.3901638686656952,
135
+ "learning_rate": 0.00053115625,
136
+ "loss": 3.1885,
137
+ "step": 17000
138
+ },
139
+ {
140
+ "epoch": 1.1553273427471118,
141
+ "grad_norm": 0.35213154554367065,
142
+ "learning_rate": 0.00056240625,
143
+ "loss": 3.1852,
144
+ "step": 18000
145
+ },
146
+ {
147
+ "epoch": 1.2195121951219512,
148
+ "grad_norm": 0.3163226246833801,
149
+ "learning_rate": 0.00059359375,
150
+ "loss": 3.1754,
151
+ "step": 19000
152
+ },
153
+ {
154
+ "epoch": 1.2836970474967908,
155
+ "grad_norm": 0.3097131848335266,
156
+ "learning_rate": 0.0006248437500000001,
157
+ "loss": 3.1646,
158
+ "step": 20000
159
+ },
160
+ {
161
+ "epoch": 1.3478818998716302,
162
+ "grad_norm": 0.30222874879837036,
163
+ "learning_rate": 0.00065609375,
164
+ "loss": 3.1492,
165
+ "step": 21000
166
+ },
167
+ {
168
+ "epoch": 1.4120667522464698,
169
+ "grad_norm": 0.27634382247924805,
170
+ "learning_rate": 0.0006873125,
171
+ "loss": 3.1411,
172
+ "step": 22000
173
+ },
174
+ {
175
+ "epoch": 1.4762516046213094,
176
+ "grad_norm": 0.27012380957603455,
177
+ "learning_rate": 0.0007185625,
178
+ "loss": 3.1381,
179
+ "step": 23000
180
+ },
181
+ {
182
+ "epoch": 1.540436456996149,
183
+ "grad_norm": 0.2659747004508972,
184
+ "learning_rate": 0.00074978125,
185
+ "loss": 3.1202,
186
+ "step": 24000
187
+ },
188
+ {
189
+ "epoch": 1.6046213093709885,
190
+ "grad_norm": 0.24373669922351837,
191
+ "learning_rate": 0.0007810312499999999,
192
+ "loss": 3.117,
193
+ "step": 25000
194
+ },
195
+ {
196
+ "epoch": 1.6688061617458279,
197
+ "grad_norm": 0.23827452957630157,
198
+ "learning_rate": 0.00081228125,
199
+ "loss": 3.104,
200
+ "step": 26000
201
+ },
202
+ {
203
+ "epoch": 1.7329910141206675,
204
+ "grad_norm": 0.2538315951824188,
205
+ "learning_rate": 0.00084353125,
206
+ "loss": 3.0942,
207
+ "step": 27000
208
+ },
209
+ {
210
+ "epoch": 1.797175866495507,
211
+ "grad_norm": 0.21684885025024414,
212
+ "learning_rate": 0.00087478125,
213
+ "loss": 3.0855,
214
+ "step": 28000
215
+ },
216
+ {
217
+ "epoch": 1.8613607188703467,
218
+ "grad_norm": 0.2181146740913391,
219
+ "learning_rate": 0.000906,
220
+ "loss": 3.0828,
221
+ "step": 29000
222
+ },
223
+ {
224
+ "epoch": 1.925545571245186,
225
+ "grad_norm": 0.19399482011795044,
226
+ "learning_rate": 0.00093725,
227
+ "loss": 3.0704,
228
+ "step": 30000
229
+ },
230
+ {
231
+ "epoch": 1.9897304236200257,
232
+ "grad_norm": 0.21062089502811432,
233
+ "learning_rate": 0.00096846875,
234
+ "loss": 3.064,
235
+ "step": 31000
236
+ },
237
+ {
238
+ "epoch": 2.0,
239
+ "eval_accuracy": 0.4002464390011324,
240
+ "eval_loss": 3.147815465927124,
241
+ "eval_runtime": 113.1564,
242
+ "eval_samples_per_second": 465.197,
243
+ "eval_steps_per_second": 7.273,
244
+ "step": 31160
245
+ },
246
+ {
247
+ "epoch": 2.053915275994865,
248
+ "grad_norm": 0.18824858963489532,
249
+ "learning_rate": 0.00099971875,
250
+ "loss": 3.014,
251
+ "step": 32000
252
+ },
253
+ {
254
+ "epoch": 2.1181001283697047,
255
+ "grad_norm": 0.1937263160943985,
256
+ "learning_rate": 0.0009964556509298999,
257
+ "loss": 3.003,
258
+ "step": 33000
259
+ },
260
+ {
261
+ "epoch": 2.1822849807445444,
262
+ "grad_norm": 0.18498077988624573,
263
+ "learning_rate": 0.0009928826895565094,
264
+ "loss": 3.0099,
265
+ "step": 34000
266
+ },
267
+ {
268
+ "epoch": 2.246469833119384,
269
+ "grad_norm": 0.17105349898338318,
270
+ "learning_rate": 0.0009893061516452074,
271
+ "loss": 2.9979,
272
+ "step": 35000
273
+ },
274
+ {
275
+ "epoch": 2.3106546854942236,
276
+ "grad_norm": 0.17657169699668884,
277
+ "learning_rate": 0.0009857331902718169,
278
+ "loss": 2.994,
279
+ "step": 36000
280
+ },
281
+ {
282
+ "epoch": 2.3748395378690628,
283
+ "grad_norm": 0.2064744532108307,
284
+ "learning_rate": 0.000982156652360515,
285
+ "loss": 2.9866,
286
+ "step": 37000
287
+ },
288
+ {
289
+ "epoch": 2.4390243902439024,
290
+ "grad_norm": 0.1920684576034546,
291
+ "learning_rate": 0.000978580114449213,
292
+ "loss": 2.9776,
293
+ "step": 38000
294
+ },
295
+ {
296
+ "epoch": 2.503209242618742,
297
+ "grad_norm": 0.16097630560398102,
298
+ "learning_rate": 0.0009750035765379113,
299
+ "loss": 2.9774,
300
+ "step": 39000
301
+ },
302
+ {
303
+ "epoch": 2.5673940949935816,
304
+ "grad_norm": 0.18222209811210632,
305
+ "learning_rate": 0.0009714306151645208,
306
+ "loss": 2.9705,
307
+ "step": 40000
308
+ },
309
+ {
310
+ "epoch": 2.6315789473684212,
311
+ "grad_norm": 0.17541208863258362,
312
+ "learning_rate": 0.000967854077253219,
313
+ "loss": 2.9642,
314
+ "step": 41000
315
+ },
316
+ {
317
+ "epoch": 2.6957637997432604,
318
+ "grad_norm": 0.15682058036327362,
319
+ "learning_rate": 0.0009642775393419171,
320
+ "loss": 2.9563,
321
+ "step": 42000
322
+ },
323
+ {
324
+ "epoch": 2.7599486521181,
325
+ "grad_norm": 0.1775752454996109,
326
+ "learning_rate": 0.0009607045779685265,
327
+ "loss": 2.9493,
328
+ "step": 43000
329
+ },
330
+ {
331
+ "epoch": 2.8241335044929397,
332
+ "grad_norm": 0.19568467140197754,
333
+ "learning_rate": 0.0009571280400572247,
334
+ "loss": 2.9404,
335
+ "step": 44000
336
+ },
337
+ {
338
+ "epoch": 2.8883183568677793,
339
+ "grad_norm": 0.1565084159374237,
340
+ "learning_rate": 0.0009535550786838341,
341
+ "loss": 2.9451,
342
+ "step": 45000
343
+ },
344
+ {
345
+ "epoch": 2.952503209242619,
346
+ "grad_norm": 0.1869024932384491,
347
+ "learning_rate": 0.0009499785407725322,
348
+ "loss": 2.9372,
349
+ "step": 46000
350
+ },
351
+ {
352
+ "epoch": 3.0,
353
+ "eval_accuracy": 0.41228246617796055,
354
+ "eval_loss": 3.0373053550720215,
355
+ "eval_runtime": 113.4953,
356
+ "eval_samples_per_second": 463.808,
357
+ "eval_steps_per_second": 7.251,
358
+ "step": 46740
359
+ },
360
+ {
361
+ "epoch": 3.016688061617458,
362
+ "grad_norm": 0.16253632307052612,
363
+ "learning_rate": 0.0009464055793991416,
364
+ "loss": 2.9111,
365
+ "step": 47000
366
+ },
367
+ {
368
+ "epoch": 3.0808729139922977,
369
+ "grad_norm": 0.15888482332229614,
370
+ "learning_rate": 0.0009428290414878398,
371
+ "loss": 2.8581,
372
+ "step": 48000
373
+ },
374
+ {
375
+ "epoch": 3.1450577663671373,
376
+ "grad_norm": 0.1883237361907959,
377
+ "learning_rate": 0.0009392525035765379,
378
+ "loss": 2.8599,
379
+ "step": 49000
380
+ },
381
+ {
382
+ "epoch": 3.209242618741977,
383
+ "grad_norm": 0.1678224354982376,
384
+ "learning_rate": 0.0009356831187410587,
385
+ "loss": 2.8651,
386
+ "step": 50000
387
+ },
388
+ {
389
+ "epoch": 3.2734274711168165,
390
+ "grad_norm": 0.1709676831960678,
391
+ "learning_rate": 0.0009321065808297568,
392
+ "loss": 2.8641,
393
+ "step": 51000
394
+ },
395
+ {
396
+ "epoch": 3.337612323491656,
397
+ "grad_norm": 0.1678624153137207,
398
+ "learning_rate": 0.000928530042918455,
399
+ "loss": 2.8589,
400
+ "step": 52000
401
+ },
402
+ {
403
+ "epoch": 3.4017971758664953,
404
+ "grad_norm": 0.19356407225131989,
405
+ "learning_rate": 0.0009249535050071531,
406
+ "loss": 2.866,
407
+ "step": 53000
408
+ },
409
+ {
410
+ "epoch": 3.465982028241335,
411
+ "grad_norm": NaN,
412
+ "learning_rate": 0.0009213805436337625,
413
+ "loss": 2.8603,
414
+ "step": 54000
415
+ },
416
+ {
417
+ "epoch": 3.5301668806161746,
418
+ "grad_norm": 0.19353008270263672,
419
+ "learning_rate": 0.0009178040057224607,
420
+ "loss": 2.8639,
421
+ "step": 55000
422
+ },
423
+ {
424
+ "epoch": 3.594351732991014,
425
+ "grad_norm": 0.16915424168109894,
426
+ "learning_rate": 0.0009142274678111588,
427
+ "loss": 2.8607,
428
+ "step": 56000
429
+ },
430
+ {
431
+ "epoch": 3.658536585365854,
432
+ "grad_norm": 0.16573752462863922,
433
+ "learning_rate": 0.0009106545064377682,
434
+ "loss": 2.8589,
435
+ "step": 57000
436
+ },
437
+ {
438
+ "epoch": 3.7227214377406934,
439
+ "grad_norm": 0.1988651305437088,
440
+ "learning_rate": 0.0009070779685264664,
441
+ "loss": 2.8545,
442
+ "step": 58000
443
+ },
444
+ {
445
+ "epoch": 3.7869062901155326,
446
+ "grad_norm": 0.16490334272384644,
447
+ "learning_rate": 0.0009035050071530759,
448
+ "loss": 2.8589,
449
+ "step": 59000
450
+ },
451
+ {
452
+ "epoch": 3.851091142490372,
453
+ "grad_norm": 0.1708660125732422,
454
+ "learning_rate": 0.0008999284692417739,
455
+ "loss": 2.8569,
456
+ "step": 60000
457
+ },
458
+ {
459
+ "epoch": 3.915275994865212,
460
+ "grad_norm": 0.20747113227844238,
461
+ "learning_rate": 0.0008963555078683834,
462
+ "loss": 2.8523,
463
+ "step": 61000
464
+ },
465
+ {
466
+ "epoch": 3.9794608472400514,
467
+ "grad_norm": 0.1617908626794815,
468
+ "learning_rate": 0.0008927789699570816,
469
+ "loss": 2.8458,
470
+ "step": 62000
471
+ },
472
+ {
473
+ "epoch": 4.0,
474
+ "eval_accuracy": 0.41800278622087134,
475
+ "eval_loss": 2.9891786575317383,
476
+ "eval_runtime": 113.4111,
477
+ "eval_samples_per_second": 464.152,
478
+ "eval_steps_per_second": 7.257,
479
+ "step": 62320
480
+ },
481
+ {
482
+ "epoch": 4.043645699614891,
483
+ "grad_norm": 0.20924492180347443,
484
+ "learning_rate": 0.0008892060085836911,
485
+ "loss": 2.7891,
486
+ "step": 63000
487
+ },
488
+ {
489
+ "epoch": 4.10783055198973,
490
+ "grad_norm": 0.18781903386116028,
491
+ "learning_rate": 0.0008856294706723892,
492
+ "loss": 2.7725,
493
+ "step": 64000
494
+ },
495
+ {
496
+ "epoch": 4.17201540436457,
497
+ "grad_norm": 0.23733529448509216,
498
+ "learning_rate": 0.0008820529327610874,
499
+ "loss": 2.778,
500
+ "step": 65000
501
+ },
502
+ {
503
+ "epoch": 4.2362002567394095,
504
+ "grad_norm": 0.19477002322673798,
505
+ "learning_rate": 0.0008784763948497855,
506
+ "loss": 2.7777,
507
+ "step": 66000
508
+ },
509
+ {
510
+ "epoch": 4.300385109114249,
511
+ "grad_norm": 0.17136947810649872,
512
+ "learning_rate": 0.0008749034334763949,
513
+ "loss": 2.7804,
514
+ "step": 67000
515
+ },
516
+ {
517
+ "epoch": 4.364569961489089,
518
+ "grad_norm": 0.25313475728034973,
519
+ "learning_rate": 0.0008713268955650931,
520
+ "loss": 2.7886,
521
+ "step": 68000
522
+ },
523
+ {
524
+ "epoch": 4.428754813863928,
525
+ "grad_norm": 0.1842871755361557,
526
+ "learning_rate": 0.0008677503576537912,
527
+ "loss": 2.7803,
528
+ "step": 69000
529
+ },
530
+ {
531
+ "epoch": 4.492939666238768,
532
+ "grad_norm": 0.2282925546169281,
533
+ "learning_rate": 0.0008641773962804006,
534
+ "loss": 2.7881,
535
+ "step": 70000
536
+ },
537
+ {
538
+ "epoch": 4.557124518613607,
539
+ "grad_norm": 0.17763297259807587,
540
+ "learning_rate": 0.0008606008583690988,
541
+ "loss": 2.7897,
542
+ "step": 71000
543
+ },
544
+ {
545
+ "epoch": 4.621309370988447,
546
+ "grad_norm": 0.17076101899147034,
547
+ "learning_rate": 0.0008570278969957082,
548
+ "loss": 2.7892,
549
+ "step": 72000
550
+ },
551
+ {
552
+ "epoch": 4.685494223363286,
553
+ "grad_norm": 0.20617830753326416,
554
+ "learning_rate": 0.0008534513590844063,
555
+ "loss": 2.7934,
556
+ "step": 73000
557
+ },
558
+ {
559
+ "epoch": 4.7496790757381255,
560
+ "grad_norm": 0.17589814960956573,
561
+ "learning_rate": 0.0008498783977110157,
562
+ "loss": 2.7888,
563
+ "step": 74000
564
+ },
565
+ {
566
+ "epoch": 4.813863928112966,
567
+ "grad_norm": 0.17282748222351074,
568
+ "learning_rate": 0.0008463018597997139,
569
+ "loss": 2.7928,
570
+ "step": 75000
571
+ },
572
+ {
573
+ "epoch": 4.878048780487805,
574
+ "grad_norm": 0.1879618763923645,
575
+ "learning_rate": 0.000842725321888412,
576
+ "loss": 2.7842,
577
+ "step": 76000
578
+ },
579
+ {
580
+ "epoch": 4.942233632862644,
581
+ "grad_norm": 0.27251970767974854,
582
+ "learning_rate": 0.0008391523605150215,
583
+ "loss": 2.7931,
584
+ "step": 77000
585
+ },
586
+ {
587
+ "epoch": 5.0,
588
+ "eval_accuracy": 0.4212023958519578,
589
+ "eval_loss": 2.9617300033569336,
590
+ "eval_runtime": 113.1396,
591
+ "eval_samples_per_second": 465.266,
592
+ "eval_steps_per_second": 7.274,
593
+ "step": 77900
594
+ },
595
+ {
596
+ "epoch": 5.006418485237484,
597
+ "grad_norm": 0.17574380338191986,
598
+ "learning_rate": 0.0008355793991416309,
599
+ "loss": 2.7806,
600
+ "step": 78000
601
+ },
602
+ {
603
+ "epoch": 5.070603337612323,
604
+ "grad_norm": 0.1976320743560791,
605
+ "learning_rate": 0.0008320028612303291,
606
+ "loss": 2.6979,
607
+ "step": 79000
608
+ },
609
+ {
610
+ "epoch": 5.134788189987163,
611
+ "grad_norm": 0.17811007797718048,
612
+ "learning_rate": 0.0008284263233190272,
613
+ "loss": 2.7114,
614
+ "step": 80000
615
+ },
616
+ {
617
+ "epoch": 5.198973042362002,
618
+ "grad_norm": 0.19787943363189697,
619
+ "learning_rate": 0.0008248497854077254,
620
+ "loss": 2.7157,
621
+ "step": 81000
622
+ },
623
+ {
624
+ "epoch": 5.2631578947368425,
625
+ "grad_norm": 0.231247678399086,
626
+ "learning_rate": 0.0008212732474964235,
627
+ "loss": 2.7172,
628
+ "step": 82000
629
+ },
630
+ {
631
+ "epoch": 5.327342747111682,
632
+ "grad_norm": 0.17853890359401703,
633
+ "learning_rate": 0.0008177002861230329,
634
+ "loss": 2.7299,
635
+ "step": 83000
636
+ },
637
+ {
638
+ "epoch": 5.391527599486521,
639
+ "grad_norm": 0.18848580121994019,
640
+ "learning_rate": 0.0008141273247496423,
641
+ "loss": 2.7296,
642
+ "step": 84000
643
+ },
644
+ {
645
+ "epoch": 5.455712451861361,
646
+ "grad_norm": 0.1779264509677887,
647
+ "learning_rate": 0.0008105507868383405,
648
+ "loss": 2.7282,
649
+ "step": 85000
650
+ },
651
+ {
652
+ "epoch": 5.5198973042362,
653
+ "grad_norm": 0.2169097363948822,
654
+ "learning_rate": 0.0008069742489270386,
655
+ "loss": 2.7305,
656
+ "step": 86000
657
+ },
658
+ {
659
+ "epoch": 5.58408215661104,
660
+ "grad_norm": 0.1910111904144287,
661
+ "learning_rate": 0.0008033977110157368,
662
+ "loss": 2.7319,
663
+ "step": 87000
664
+ },
665
+ {
666
+ "epoch": 5.648267008985879,
667
+ "grad_norm": 0.1949514001607895,
668
+ "learning_rate": 0.0007998247496423462,
669
+ "loss": 2.7392,
670
+ "step": 88000
671
+ },
672
+ {
673
+ "epoch": 5.712451861360719,
674
+ "grad_norm": 0.18040545284748077,
675
+ "learning_rate": 0.0007962482117310443,
676
+ "loss": 2.7338,
677
+ "step": 89000
678
+ },
679
+ {
680
+ "epoch": 5.7766367137355585,
681
+ "grad_norm": 0.2376387119293213,
682
+ "learning_rate": 0.0007926716738197425,
683
+ "loss": 2.7409,
684
+ "step": 90000
685
+ },
686
+ {
687
+ "epoch": 5.840821566110398,
688
+ "grad_norm": 0.23671500384807587,
689
+ "learning_rate": 0.0007890951359084406,
690
+ "loss": 2.7415,
691
+ "step": 91000
692
+ },
693
+ {
694
+ "epoch": 5.905006418485238,
695
+ "grad_norm": 0.20994797348976135,
696
+ "learning_rate": 0.00078552217453505,
697
+ "loss": 2.7396,
698
+ "step": 92000
699
+ },
700
+ {
701
+ "epoch": 5.969191270860077,
702
+ "grad_norm": 0.18293611705303192,
703
+ "learning_rate": 0.0007819456366237482,
704
+ "loss": 2.7418,
705
+ "step": 93000
706
+ },
707
+ {
708
+ "epoch": 6.0,
709
+ "eval_accuracy": 0.42353231718219203,
710
+ "eval_loss": 2.9472451210021973,
711
+ "eval_runtime": 112.282,
712
+ "eval_samples_per_second": 468.82,
713
+ "eval_steps_per_second": 7.33,
714
+ "step": 93480
715
+ },
716
+ {
717
+ "epoch": 6.033376123234916,
718
+ "grad_norm": 0.23322834074497223,
719
+ "learning_rate": 0.0007783726752503577,
720
+ "loss": 2.6918,
721
+ "step": 94000
722
+ },
723
+ {
724
+ "epoch": 6.097560975609756,
725
+ "grad_norm": 0.2009533792734146,
726
+ "learning_rate": 0.0007747997138769671,
727
+ "loss": 2.6555,
728
+ "step": 95000
729
+ },
730
+ {
731
+ "epoch": 6.161745827984595,
732
+ "grad_norm": 0.2933996319770813,
733
+ "learning_rate": 0.0007712231759656652,
734
+ "loss": 2.6563,
735
+ "step": 96000
736
+ },
737
+ {
738
+ "epoch": 6.225930680359435,
739
+ "grad_norm": 0.20901496708393097,
740
+ "learning_rate": 0.0007676466380543634,
741
+ "loss": 2.6716,
742
+ "step": 97000
743
+ },
744
+ {
745
+ "epoch": 6.290115532734275,
746
+ "grad_norm": 0.19593417644500732,
747
+ "learning_rate": 0.0007640701001430615,
748
+ "loss": 2.6735,
749
+ "step": 98000
750
+ },
751
+ {
752
+ "epoch": 6.354300385109115,
753
+ "grad_norm": 0.22803381085395813,
754
+ "learning_rate": 0.0007604935622317597,
755
+ "loss": 2.6748,
756
+ "step": 99000
757
+ },
758
+ {
759
+ "epoch": 6.418485237483954,
760
+ "grad_norm": 0.22151826322078705,
761
+ "learning_rate": 0.0007569206008583691,
762
+ "loss": 2.6874,
763
+ "step": 100000
764
+ },
765
+ {
766
+ "epoch": 6.482670089858793,
767
+ "grad_norm": 0.21168138086795807,
768
+ "learning_rate": 0.0007533476394849786,
769
+ "loss": 2.6896,
770
+ "step": 101000
771
+ },
772
+ {
773
+ "epoch": 6.546854942233633,
774
+ "grad_norm": 0.20149804651737213,
775
+ "learning_rate": 0.0007497711015736766,
776
+ "loss": 2.6884,
777
+ "step": 102000
778
+ },
779
+ {
780
+ "epoch": 6.611039794608472,
781
+ "grad_norm": 0.1844107061624527,
782
+ "learning_rate": 0.0007461945636623749,
783
+ "loss": 2.6858,
784
+ "step": 103000
785
+ },
786
+ {
787
+ "epoch": 6.675224646983312,
788
+ "grad_norm": 0.19826510548591614,
789
+ "learning_rate": 0.0007426180257510729,
790
+ "loss": 2.6913,
791
+ "step": 104000
792
+ },
793
+ {
794
+ "epoch": 6.7394094993581515,
795
+ "grad_norm": 0.19884894788265228,
796
+ "learning_rate": 0.0007390414878397711,
797
+ "loss": 2.6914,
798
+ "step": 105000
799
+ },
800
+ {
801
+ "epoch": 6.803594351732991,
802
+ "grad_norm": 0.18995630741119385,
803
+ "learning_rate": 0.0007354649499284692,
804
+ "loss": 2.6991,
805
+ "step": 106000
806
+ },
807
+ {
808
+ "epoch": 6.867779204107831,
809
+ "grad_norm": 0.18595723807811737,
810
+ "learning_rate": 0.0007318919885550786,
811
+ "loss": 2.6989,
812
+ "step": 107000
813
+ },
814
+ {
815
+ "epoch": 6.93196405648267,
816
+ "grad_norm": 0.22208259999752045,
817
+ "learning_rate": 0.0007283154506437768,
818
+ "loss": 2.6995,
819
+ "step": 108000
820
+ },
821
+ {
822
+ "epoch": 6.99614890885751,
823
+ "grad_norm": 0.2557336390018463,
824
+ "learning_rate": 0.0007247424892703864,
825
+ "loss": 2.7019,
826
+ "step": 109000
827
+ },
828
+ {
829
+ "epoch": 7.0,
830
+ "eval_accuracy": 0.4251007956374039,
831
+ "eval_loss": 2.938654661178589,
832
+ "eval_runtime": 113.1288,
833
+ "eval_samples_per_second": 465.31,
834
+ "eval_steps_per_second": 7.275,
835
+ "step": 109060
836
+ },
837
+ {
838
+ "epoch": 7.060333761232349,
839
+ "grad_norm": 0.2100907266139984,
840
+ "learning_rate": 0.0007211659513590845,
841
+ "loss": 2.6144,
842
+ "step": 110000
843
+ },
844
+ {
845
+ "epoch": 7.124518613607188,
846
+ "grad_norm": 0.20888903737068176,
847
+ "learning_rate": 0.0007175929899856939,
848
+ "loss": 2.6195,
849
+ "step": 111000
850
+ },
851
+ {
852
+ "epoch": 7.188703465982028,
853
+ "grad_norm": 0.2282038778066635,
854
+ "learning_rate": 0.0007140164520743921,
855
+ "loss": 2.6274,
856
+ "step": 112000
857
+ },
858
+ {
859
+ "epoch": 7.2528883183568675,
860
+ "grad_norm": 0.20698107779026031,
861
+ "learning_rate": 0.0007104434907010015,
862
+ "loss": 2.6321,
863
+ "step": 113000
864
+ },
865
+ {
866
+ "epoch": 7.317073170731708,
867
+ "grad_norm": 0.2202378660440445,
868
+ "learning_rate": 0.0007068669527896996,
869
+ "loss": 2.6296,
870
+ "step": 114000
871
+ },
872
+ {
873
+ "epoch": 7.381258023106547,
874
+ "grad_norm": 0.30359992384910583,
875
+ "learning_rate": 0.0007032904148783978,
876
+ "loss": 2.6413,
877
+ "step": 115000
878
+ },
879
+ {
880
+ "epoch": 7.445442875481387,
881
+ "grad_norm": 0.18242676556110382,
882
+ "learning_rate": 0.0006997210300429184,
883
+ "loss": 2.6454,
884
+ "step": 116000
885
+ },
886
+ {
887
+ "epoch": 7.509627727856226,
888
+ "grad_norm": 0.28255388140678406,
889
+ "learning_rate": 0.0006961444921316166,
890
+ "loss": 2.6507,
891
+ "step": 117000
892
+ },
893
+ {
894
+ "epoch": 7.573812580231065,
895
+ "grad_norm": 0.21654346585273743,
896
+ "learning_rate": 0.0006925679542203147,
897
+ "loss": 2.6457,
898
+ "step": 118000
899
+ },
900
+ {
901
+ "epoch": 7.637997432605905,
902
+ "grad_norm": 0.21898731589317322,
903
+ "learning_rate": 0.0006889914163090129,
904
+ "loss": 2.6465,
905
+ "step": 119000
906
+ },
907
+ {
908
+ "epoch": 7.702182284980744,
909
+ "grad_norm": 0.3429543972015381,
910
+ "learning_rate": 0.000685414878397711,
911
+ "loss": 2.6531,
912
+ "step": 120000
913
+ },
914
+ {
915
+ "epoch": 7.766367137355584,
916
+ "grad_norm": 0.20613309741020203,
917
+ "learning_rate": 0.0006818454935622318,
918
+ "loss": 2.6509,
919
+ "step": 121000
920
+ },
921
+ {
922
+ "epoch": 7.830551989730424,
923
+ "grad_norm": 0.2630804777145386,
924
+ "learning_rate": 0.0006782689556509299,
925
+ "loss": 2.6599,
926
+ "step": 122000
927
+ },
928
+ {
929
+ "epoch": 7.894736842105263,
930
+ "grad_norm": 0.22952473163604736,
931
+ "learning_rate": 0.0006746924177396281,
932
+ "loss": 2.6688,
933
+ "step": 123000
934
+ },
935
+ {
936
+ "epoch": 7.958921694480103,
937
+ "grad_norm": 0.21582633256912231,
938
+ "learning_rate": 0.0006711158798283262,
939
+ "loss": 2.6555,
940
+ "step": 124000
941
+ },
942
+ {
943
+ "epoch": 8.0,
944
+ "eval_accuracy": 0.42585136182132427,
945
+ "eval_loss": 2.938795566558838,
946
+ "eval_runtime": 112.5269,
947
+ "eval_samples_per_second": 467.799,
948
+ "eval_steps_per_second": 7.314,
949
+ "step": 124640
950
+ },
951
+ {
952
+ "epoch": 8.023106546854942,
953
+ "grad_norm": 0.21312978863716125,
954
+ "learning_rate": 0.0006675429184549356,
955
+ "loss": 2.6305,
956
+ "step": 125000
957
+ },
958
+ {
959
+ "epoch": 8.087291399229782,
960
+ "grad_norm": 0.20984013378620148,
961
+ "learning_rate": 0.0006639663805436338,
962
+ "loss": 2.579,
963
+ "step": 126000
964
+ },
965
+ {
966
+ "epoch": 8.15147625160462,
967
+ "grad_norm": 0.20877642929553986,
968
+ "learning_rate": 0.0006603934191702432,
969
+ "loss": 2.581,
970
+ "step": 127000
971
+ },
972
+ {
973
+ "epoch": 8.21566110397946,
974
+ "grad_norm": 0.22023415565490723,
975
+ "learning_rate": 0.0006568168812589413,
976
+ "loss": 2.5914,
977
+ "step": 128000
978
+ },
979
+ {
980
+ "epoch": 8.2798459563543,
981
+ "grad_norm": 0.36641648411750793,
982
+ "learning_rate": 0.0006532439198855507,
983
+ "loss": 2.6003,
984
+ "step": 129000
985
+ },
986
+ {
987
+ "epoch": 8.34403080872914,
988
+ "grad_norm": 0.21248340606689453,
989
+ "learning_rate": 0.0006496673819742489,
990
+ "loss": 2.6038,
991
+ "step": 130000
992
+ },
993
+ {
994
+ "epoch": 8.408215661103979,
995
+ "grad_norm": 0.2076880782842636,
996
+ "learning_rate": 0.000646090844062947,
997
+ "loss": 2.6018,
998
+ "step": 131000
999
+ },
1000
+ {
1001
+ "epoch": 8.472400513478819,
1002
+ "grad_norm": 0.22432173788547516,
1003
+ "learning_rate": 0.0006425143061516452,
1004
+ "loss": 2.6112,
1005
+ "step": 132000
1006
+ },
1007
+ {
1008
+ "epoch": 8.536585365853659,
1009
+ "grad_norm": 0.21105310320854187,
1010
+ "learning_rate": 0.0006389377682403433,
1011
+ "loss": 2.6137,
1012
+ "step": 133000
1013
+ },
1014
+ {
1015
+ "epoch": 8.600770218228497,
1016
+ "grad_norm": 0.1929914504289627,
1017
+ "learning_rate": 0.0006353683834048642,
1018
+ "loss": 2.6134,
1019
+ "step": 134000
1020
+ },
1021
+ {
1022
+ "epoch": 8.664955070603337,
1023
+ "grad_norm": 0.2183075249195099,
1024
+ "learning_rate": 0.0006317918454935622,
1025
+ "loss": 2.62,
1026
+ "step": 135000
1027
+ },
1028
+ {
1029
+ "epoch": 8.729139922978177,
1030
+ "grad_norm": 0.2047470211982727,
1031
+ "learning_rate": 0.0006282153075822604,
1032
+ "loss": 2.619,
1033
+ "step": 136000
1034
+ },
1035
+ {
1036
+ "epoch": 8.793324775353017,
1037
+ "grad_norm": 0.1994733363389969,
1038
+ "learning_rate": 0.0006246387696709585,
1039
+ "loss": 2.6245,
1040
+ "step": 137000
1041
+ },
1042
+ {
1043
+ "epoch": 8.857509627727856,
1044
+ "grad_norm": 0.28145477175712585,
1045
+ "learning_rate": 0.000621065808297568,
1046
+ "loss": 2.6199,
1047
+ "step": 138000
1048
+ },
1049
+ {
1050
+ "epoch": 8.921694480102696,
1051
+ "grad_norm": 0.20502200722694397,
1052
+ "learning_rate": 0.0006174928469241774,
1053
+ "loss": 2.624,
1054
+ "step": 139000
1055
+ },
1056
+ {
1057
+ "epoch": 8.985879332477536,
1058
+ "grad_norm": 0.21982279419898987,
1059
+ "learning_rate": 0.0006139163090128756,
1060
+ "loss": 2.6321,
1061
+ "step": 140000
1062
+ },
1063
+ {
1064
+ "epoch": 9.0,
1065
+ "eval_accuracy": 0.4262641545980094,
1066
+ "eval_loss": 2.9394068717956543,
1067
+ "eval_runtime": 112.5515,
1068
+ "eval_samples_per_second": 467.697,
1069
+ "eval_steps_per_second": 7.312,
1070
+ "step": 140220
1071
+ },
1072
+ {
1073
+ "epoch": 9.050064184852374,
1074
+ "grad_norm": 0.24569900333881378,
1075
+ "learning_rate": 0.000610343347639485,
1076
+ "loss": 2.5553,
1077
+ "step": 141000
1078
+ },
1079
+ {
1080
+ "epoch": 9.114249037227214,
1081
+ "grad_norm": 0.22856919467449188,
1082
+ "learning_rate": 0.0006067668097281831,
1083
+ "loss": 2.5497,
1084
+ "step": 142000
1085
+ },
1086
+ {
1087
+ "epoch": 9.178433889602054,
1088
+ "grad_norm": 0.23001812398433685,
1089
+ "learning_rate": 0.0006031902718168813,
1090
+ "loss": 2.5568,
1091
+ "step": 143000
1092
+ },
1093
+ {
1094
+ "epoch": 9.242618741976893,
1095
+ "grad_norm": 0.2656523287296295,
1096
+ "learning_rate": 0.0005996137339055794,
1097
+ "loss": 2.5636,
1098
+ "step": 144000
1099
+ },
1100
+ {
1101
+ "epoch": 9.306803594351733,
1102
+ "grad_norm": 0.23091475665569305,
1103
+ "learning_rate": 0.0005960371959942776,
1104
+ "loss": 2.568,
1105
+ "step": 145000
1106
+ },
1107
+ {
1108
+ "epoch": 9.370988446726573,
1109
+ "grad_norm": 0.26248905062675476,
1110
+ "learning_rate": 0.0005924678111587983,
1111
+ "loss": 2.5739,
1112
+ "step": 146000
1113
+ },
1114
+ {
1115
+ "epoch": 9.435173299101413,
1116
+ "grad_norm": 0.21125547587871552,
1117
+ "learning_rate": 0.0005888912732474965,
1118
+ "loss": 2.5724,
1119
+ "step": 147000
1120
+ },
1121
+ {
1122
+ "epoch": 9.499358151476251,
1123
+ "grad_norm": 0.22728748619556427,
1124
+ "learning_rate": 0.0005853147353361946,
1125
+ "loss": 2.5749,
1126
+ "step": 148000
1127
+ },
1128
+ {
1129
+ "epoch": 9.563543003851091,
1130
+ "grad_norm": 0.22006183862686157,
1131
+ "learning_rate": 0.0005817381974248928,
1132
+ "loss": 2.5791,
1133
+ "step": 149000
1134
+ },
1135
+ {
1136
+ "epoch": 9.627727856225931,
1137
+ "grad_norm": 0.3410221338272095,
1138
+ "learning_rate": 0.0005781652360515022,
1139
+ "loss": 2.5892,
1140
+ "step": 150000
1141
+ },
1142
+ {
1143
+ "epoch": 9.69191270860077,
1144
+ "grad_norm": 0.22049233317375183,
1145
+ "learning_rate": 0.0005745886981402003,
1146
+ "loss": 2.588,
1147
+ "step": 151000
1148
+ },
1149
+ {
1150
+ "epoch": 9.75609756097561,
1151
+ "grad_norm": 0.22515858709812164,
1152
+ "learning_rate": 0.0005710121602288985,
1153
+ "loss": 2.5873,
1154
+ "step": 152000
1155
+ },
1156
+ {
1157
+ "epoch": 9.82028241335045,
1158
+ "grad_norm": 0.20887359976768494,
1159
+ "learning_rate": 0.0005674391988555079,
1160
+ "loss": 2.5967,
1161
+ "step": 153000
1162
+ },
1163
+ {
1164
+ "epoch": 9.88446726572529,
1165
+ "grad_norm": 0.2514614760875702,
1166
+ "learning_rate": 0.0005638662374821173,
1167
+ "loss": 2.5889,
1168
+ "step": 154000
1169
+ },
1170
+ {
1171
+ "epoch": 9.948652118100128,
1172
+ "grad_norm": 0.28176066279411316,
1173
+ "learning_rate": 0.0005602896995708154,
1174
+ "loss": 2.5968,
1175
+ "step": 155000
1176
+ },
1177
+ {
1178
+ "epoch": 10.0,
1179
+ "eval_accuracy": 0.4265262381548364,
1180
+ "eval_loss": 2.9411489963531494,
1181
+ "eval_runtime": 112.5273,
1182
+ "eval_samples_per_second": 467.797,
1183
+ "eval_steps_per_second": 7.314,
1184
+ "step": 155800
1185
+ },
1186
+ {
1187
+ "epoch": 10.012836970474968,
1188
+ "grad_norm": 0.2588973641395569,
1189
+ "learning_rate": 0.0005567131616595136,
1190
+ "loss": 2.577,
1191
+ "step": 156000
1192
+ },
1193
+ {
1194
+ "epoch": 10.077021822849808,
1195
+ "grad_norm": 0.23765428364276886,
1196
+ "learning_rate": 0.0005531366237482117,
1197
+ "loss": 2.5107,
1198
+ "step": 157000
1199
+ },
1200
+ {
1201
+ "epoch": 10.141206675224646,
1202
+ "grad_norm": 0.4186215102672577,
1203
+ "learning_rate": 0.0005495636623748211,
1204
+ "loss": 2.5149,
1205
+ "step": 158000
1206
+ },
1207
+ {
1208
+ "epoch": 10.205391527599486,
1209
+ "grad_norm": 0.2386707216501236,
1210
+ "learning_rate": 0.0005459871244635193,
1211
+ "loss": 2.5323,
1212
+ "step": 159000
1213
+ },
1214
+ {
1215
+ "epoch": 10.269576379974326,
1216
+ "grad_norm": 0.24114102125167847,
1217
+ "learning_rate": 0.0005424141630901288,
1218
+ "loss": 2.5318,
1219
+ "step": 160000
1220
+ },
1221
+ {
1222
+ "epoch": 10.333761232349165,
1223
+ "grad_norm": 0.2539213299751282,
1224
+ "learning_rate": 0.0005388376251788269,
1225
+ "loss": 2.5398,
1226
+ "step": 161000
1227
+ },
1228
+ {
1229
+ "epoch": 10.397946084724005,
1230
+ "grad_norm": 0.22287596762180328,
1231
+ "learning_rate": 0.0005352646638054363,
1232
+ "loss": 2.5398,
1233
+ "step": 162000
1234
+ },
1235
+ {
1236
+ "epoch": 10.462130937098845,
1237
+ "grad_norm": 0.2541950047016144,
1238
+ "learning_rate": 0.0005316917024320458,
1239
+ "loss": 2.5513,
1240
+ "step": 163000
1241
+ },
1242
+ {
1243
+ "epoch": 10.526315789473685,
1244
+ "grad_norm": 0.24435748159885406,
1245
+ "learning_rate": 0.0005281151645207439,
1246
+ "loss": 2.5499,
1247
+ "step": 164000
1248
+ },
1249
+ {
1250
+ "epoch": 10.590500641848523,
1251
+ "grad_norm": 0.2417035698890686,
1252
+ "learning_rate": 0.000524538626609442,
1253
+ "loss": 2.5547,
1254
+ "step": 165000
1255
+ },
1256
+ {
1257
+ "epoch": 10.654685494223363,
1258
+ "grad_norm": 0.22597502171993256,
1259
+ "learning_rate": 0.0005209620886981402,
1260
+ "loss": 2.5551,
1261
+ "step": 166000
1262
+ },
1263
+ {
1264
+ "epoch": 10.718870346598203,
1265
+ "grad_norm": 0.24076290428638458,
1266
+ "learning_rate": 0.0005173855507868383,
1267
+ "loss": 2.5559,
1268
+ "step": 167000
1269
+ },
1270
+ {
1271
+ "epoch": 10.783055198973042,
1272
+ "grad_norm": 0.23913373053073883,
1273
+ "learning_rate": 0.0005138125894134477,
1274
+ "loss": 2.561,
1275
+ "step": 168000
1276
+ },
1277
+ {
1278
+ "epoch": 10.847240051347882,
1279
+ "grad_norm": 0.22141295671463013,
1280
+ "learning_rate": 0.0005102360515021459,
1281
+ "loss": 2.5657,
1282
+ "step": 169000
1283
+ },
1284
+ {
1285
+ "epoch": 10.911424903722722,
1286
+ "grad_norm": 0.23373030126094818,
1287
+ "learning_rate": 0.0005066630901287553,
1288
+ "loss": 2.5634,
1289
+ "step": 170000
1290
+ },
1291
+ {
1292
+ "epoch": 10.975609756097562,
1293
+ "grad_norm": 0.30720722675323486,
1294
+ "learning_rate": 0.0005030865522174534,
1295
+ "loss": 2.5705,
1296
+ "step": 171000
1297
+ },
1298
+ {
1299
+ "epoch": 11.0,
1300
+ "eval_accuracy": 0.4267387806186304,
1301
+ "eval_loss": 2.9455363750457764,
1302
+ "eval_runtime": 112.5954,
1303
+ "eval_samples_per_second": 467.515,
1304
+ "eval_steps_per_second": 7.309,
1305
+ "step": 171380
1306
+ },
1307
+ {
1308
+ "epoch": 11.0397946084724,
1309
+ "grad_norm": 0.28773027658462524,
1310
+ "learning_rate": 0.000499513590844063,
1311
+ "loss": 2.5192,
1312
+ "step": 172000
1313
+ },
1314
+ {
1315
+ "epoch": 11.10397946084724,
1316
+ "grad_norm": 0.2327316552400589,
1317
+ "learning_rate": 0.000495937052932761,
1318
+ "loss": 2.4889,
1319
+ "step": 173000
1320
+ },
1321
+ {
1322
+ "epoch": 11.16816431322208,
1323
+ "grad_norm": 0.24159583449363708,
1324
+ "learning_rate": 0.0004923605150214592,
1325
+ "loss": 2.4953,
1326
+ "step": 174000
1327
+ },
1328
+ {
1329
+ "epoch": 11.232349165596919,
1330
+ "grad_norm": 0.22464554011821747,
1331
+ "learning_rate": 0.0004887875536480687,
1332
+ "loss": 2.4973,
1333
+ "step": 175000
1334
+ },
1335
+ {
1336
+ "epoch": 11.296534017971759,
1337
+ "grad_norm": 0.24839630722999573,
1338
+ "learning_rate": 0.0004852110157367668,
1339
+ "loss": 2.5079,
1340
+ "step": 176000
1341
+ },
1342
+ {
1343
+ "epoch": 11.360718870346599,
1344
+ "grad_norm": 0.28848448395729065,
1345
+ "learning_rate": 0.00048163805436337627,
1346
+ "loss": 2.5107,
1347
+ "step": 177000
1348
+ },
1349
+ {
1350
+ "epoch": 11.424903722721437,
1351
+ "grad_norm": 0.24984413385391235,
1352
+ "learning_rate": 0.0004780615164520744,
1353
+ "loss": 2.5192,
1354
+ "step": 178000
1355
+ },
1356
+ {
1357
+ "epoch": 11.489088575096277,
1358
+ "grad_norm": 0.260431170463562,
1359
+ "learning_rate": 0.0004744885550786838,
1360
+ "loss": 2.5257,
1361
+ "step": 179000
1362
+ },
1363
+ {
1364
+ "epoch": 11.553273427471117,
1365
+ "grad_norm": 0.25838425755500793,
1366
+ "learning_rate": 0.00047091201716738197,
1367
+ "loss": 2.5248,
1368
+ "step": 180000
1369
+ },
1370
+ {
1371
+ "epoch": 11.617458279845957,
1372
+ "grad_norm": 0.25366073846817017,
1373
+ "learning_rate": 0.0004673354792560801,
1374
+ "loss": 2.5294,
1375
+ "step": 181000
1376
+ },
1377
+ {
1378
+ "epoch": 11.681643132220795,
1379
+ "grad_norm": 0.2690897583961487,
1380
+ "learning_rate": 0.00046376609442060085,
1381
+ "loss": 2.5306,
1382
+ "step": 182000
1383
+ },
1384
+ {
1385
+ "epoch": 11.745827984595635,
1386
+ "grad_norm": 0.22692571580410004,
1387
+ "learning_rate": 0.000460189556509299,
1388
+ "loss": 2.5306,
1389
+ "step": 183000
1390
+ },
1391
+ {
1392
+ "epoch": 11.810012836970476,
1393
+ "grad_norm": 0.26649022102355957,
1394
+ "learning_rate": 0.00045661301859799714,
1395
+ "loss": 2.5373,
1396
+ "step": 184000
1397
+ },
1398
+ {
1399
+ "epoch": 11.874197689345314,
1400
+ "grad_norm": 0.23332759737968445,
1401
+ "learning_rate": 0.0004530364806866953,
1402
+ "loss": 2.5345,
1403
+ "step": 185000
1404
+ },
1405
+ {
1406
+ "epoch": 11.938382541720154,
1407
+ "grad_norm": 0.2456694394350052,
1408
+ "learning_rate": 0.0004494635193133047,
1409
+ "loss": 2.5402,
1410
+ "step": 186000
1411
+ },
1412
+ {
1413
+ "epoch": 12.0,
1414
+ "eval_accuracy": 0.4266437213183146,
1415
+ "eval_loss": 2.9544591903686523,
1416
+ "eval_runtime": 112.4087,
1417
+ "eval_samples_per_second": 468.291,
1418
+ "eval_steps_per_second": 7.321,
1419
+ "step": 186960
1420
+ },
1421
+ {
1422
+ "epoch": 12.002567394094994,
1423
+ "grad_norm": 0.27514010667800903,
1424
+ "learning_rate": 0.00044589055793991416,
1425
+ "loss": 2.5287,
1426
+ "step": 187000
1427
+ },
1428
+ {
1429
+ "epoch": 12.066752246469832,
1430
+ "grad_norm": 0.23540076613426208,
1431
+ "learning_rate": 0.0004423140200286123,
1432
+ "loss": 2.4584,
1433
+ "step": 188000
1434
+ },
1435
+ {
1436
+ "epoch": 12.130937098844672,
1437
+ "grad_norm": 0.2580265402793884,
1438
+ "learning_rate": 0.00043873748211731045,
1439
+ "loss": 2.4659,
1440
+ "step": 189000
1441
+ },
1442
+ {
1443
+ "epoch": 12.195121951219512,
1444
+ "grad_norm": 0.2443842738866806,
1445
+ "learning_rate": 0.0004351609442060086,
1446
+ "loss": 2.4725,
1447
+ "step": 190000
1448
+ },
1449
+ {
1450
+ "epoch": 12.259306803594352,
1451
+ "grad_norm": 0.24378903210163116,
1452
+ "learning_rate": 0.000431587982832618,
1453
+ "loss": 2.4779,
1454
+ "step": 191000
1455
+ },
1456
+ {
1457
+ "epoch": 12.32349165596919,
1458
+ "grad_norm": 0.24195650219917297,
1459
+ "learning_rate": 0.00042801144492131616,
1460
+ "loss": 2.4849,
1461
+ "step": 192000
1462
+ },
1463
+ {
1464
+ "epoch": 12.38767650834403,
1465
+ "grad_norm": 0.24276842176914215,
1466
+ "learning_rate": 0.0004244349070100143,
1467
+ "loss": 2.4877,
1468
+ "step": 193000
1469
+ },
1470
+ {
1471
+ "epoch": 12.45186136071887,
1472
+ "grad_norm": 0.26197871565818787,
1473
+ "learning_rate": 0.00042086194563662377,
1474
+ "loss": 2.491,
1475
+ "step": 194000
1476
+ },
1477
+ {
1478
+ "epoch": 12.51604621309371,
1479
+ "grad_norm": 0.2582203149795532,
1480
+ "learning_rate": 0.0004172854077253219,
1481
+ "loss": 2.4985,
1482
+ "step": 195000
1483
+ },
1484
+ {
1485
+ "epoch": 12.58023106546855,
1486
+ "grad_norm": 0.24834005534648895,
1487
+ "learning_rate": 0.0004137124463519313,
1488
+ "loss": 2.498,
1489
+ "step": 196000
1490
+ },
1491
+ {
1492
+ "epoch": 12.64441591784339,
1493
+ "grad_norm": 0.3061864376068115,
1494
+ "learning_rate": 0.0004101359084406295,
1495
+ "loss": 2.4973,
1496
+ "step": 197000
1497
+ },
1498
+ {
1499
+ "epoch": 12.70860077021823,
1500
+ "grad_norm": 0.30300888419151306,
1501
+ "learning_rate": 0.0004065593705293276,
1502
+ "loss": 2.4992,
1503
+ "step": 198000
1504
+ },
1505
+ {
1506
+ "epoch": 12.772785622593068,
1507
+ "grad_norm": 0.24682293832302094,
1508
+ "learning_rate": 0.00040298640915593703,
1509
+ "loss": 2.5085,
1510
+ "step": 199000
1511
+ },
1512
+ {
1513
+ "epoch": 12.836970474967908,
1514
+ "grad_norm": 0.24502244591712952,
1515
+ "learning_rate": 0.0003994134477825465,
1516
+ "loss": 2.5122,
1517
+ "step": 200000
1518
+ },
1519
+ {
1520
+ "epoch": 12.901155327342748,
1521
+ "grad_norm": 0.23922552168369293,
1522
+ "learning_rate": 0.00039583690987124464,
1523
+ "loss": 2.5128,
1524
+ "step": 201000
1525
+ },
1526
+ {
1527
+ "epoch": 12.965340179717586,
1528
+ "grad_norm": 0.2427544891834259,
1529
+ "learning_rate": 0.0003922603719599428,
1530
+ "loss": 2.5136,
1531
+ "step": 202000
1532
+ },
1533
+ {
1534
+ "epoch": 13.0,
1535
+ "eval_accuracy": 0.4266956463436438,
1536
+ "eval_loss": 2.9615485668182373,
1537
+ "eval_runtime": 112.5448,
1538
+ "eval_samples_per_second": 467.725,
1539
+ "eval_steps_per_second": 7.313,
1540
+ "step": 202540
1541
+ },
1542
+ {
1543
+ "epoch": 13.029525032092426,
1544
+ "grad_norm": 0.24927350878715515,
1545
+ "learning_rate": 0.00038868383404864093,
1546
+ "loss": 2.4798,
1547
+ "step": 203000
1548
+ },
1549
+ {
1550
+ "epoch": 13.093709884467266,
1551
+ "grad_norm": 0.2634033262729645,
1552
+ "learning_rate": 0.00038511087267525035,
1553
+ "loss": 2.4386,
1554
+ "step": 204000
1555
+ },
1556
+ {
1557
+ "epoch": 13.157894736842104,
1558
+ "grad_norm": 0.2785243093967438,
1559
+ "learning_rate": 0.0003815343347639485,
1560
+ "loss": 2.4426,
1561
+ "step": 205000
1562
+ },
1563
+ {
1564
+ "epoch": 13.222079589216944,
1565
+ "grad_norm": 0.2686097323894501,
1566
+ "learning_rate": 0.00037795779685264664,
1567
+ "loss": 2.4475,
1568
+ "step": 206000
1569
+ },
1570
+ {
1571
+ "epoch": 13.286264441591785,
1572
+ "grad_norm": 0.3374476432800293,
1573
+ "learning_rate": 0.0003743848354792561,
1574
+ "loss": 2.4584,
1575
+ "step": 207000
1576
+ },
1577
+ {
1578
+ "epoch": 13.350449293966625,
1579
+ "grad_norm": 0.2582445740699768,
1580
+ "learning_rate": 0.0003708082975679542,
1581
+ "loss": 2.46,
1582
+ "step": 208000
1583
+ },
1584
+ {
1585
+ "epoch": 13.414634146341463,
1586
+ "grad_norm": 0.25310373306274414,
1587
+ "learning_rate": 0.00036723533619456366,
1588
+ "loss": 2.4634,
1589
+ "step": 209000
1590
+ },
1591
+ {
1592
+ "epoch": 13.478818998716303,
1593
+ "grad_norm": 0.2605305016040802,
1594
+ "learning_rate": 0.0003636587982832618,
1595
+ "loss": 2.4691,
1596
+ "step": 210000
1597
+ },
1598
+ {
1599
+ "epoch": 13.543003851091143,
1600
+ "grad_norm": 0.2543717622756958,
1601
+ "learning_rate": 0.00036008226037195995,
1602
+ "loss": 2.4781,
1603
+ "step": 211000
1604
+ },
1605
+ {
1606
+ "epoch": 13.607188703465981,
1607
+ "grad_norm": 0.23427140712738037,
1608
+ "learning_rate": 0.00035650929899856936,
1609
+ "loss": 2.4727,
1610
+ "step": 212000
1611
+ },
1612
+ {
1613
+ "epoch": 13.671373555840821,
1614
+ "grad_norm": 0.27259138226509094,
1615
+ "learning_rate": 0.0003529327610872675,
1616
+ "loss": 2.4774,
1617
+ "step": 213000
1618
+ },
1619
+ {
1620
+ "epoch": 13.735558408215661,
1621
+ "grad_norm": 0.30411496758461,
1622
+ "learning_rate": 0.000349359799713877,
1623
+ "loss": 2.4781,
1624
+ "step": 214000
1625
+ },
1626
+ {
1627
+ "epoch": 13.7997432605905,
1628
+ "grad_norm": 0.24120020866394043,
1629
+ "learning_rate": 0.0003457832618025751,
1630
+ "loss": 2.4833,
1631
+ "step": 215000
1632
+ },
1633
+ {
1634
+ "epoch": 13.86392811296534,
1635
+ "grad_norm": 0.36322614550590515,
1636
+ "learning_rate": 0.00034221030042918453,
1637
+ "loss": 2.4847,
1638
+ "step": 216000
1639
+ },
1640
+ {
1641
+ "epoch": 13.92811296534018,
1642
+ "grad_norm": 0.2745082974433899,
1643
+ "learning_rate": 0.0003386337625178827,
1644
+ "loss": 2.4843,
1645
+ "step": 217000
1646
+ },
1647
+ {
1648
+ "epoch": 13.99229781771502,
1649
+ "grad_norm": 0.31372272968292236,
1650
+ "learning_rate": 0.0003350572246065808,
1651
+ "loss": 2.4814,
1652
+ "step": 218000
1653
+ },
1654
+ {
1655
+ "epoch": 14.0,
1656
+ "eval_accuracy": 0.4261248435544431,
1657
+ "eval_loss": 2.9729630947113037,
1658
+ "eval_runtime": 112.6175,
1659
+ "eval_samples_per_second": 467.423,
1660
+ "eval_steps_per_second": 7.308,
1661
+ "step": 218120
1662
+ },
1663
+ {
1664
+ "epoch": 14.056482670089858,
1665
+ "grad_norm": 0.2546994984149933,
1666
+ "learning_rate": 0.00033148068669527897,
1667
+ "loss": 2.4229,
1668
+ "step": 219000
1669
+ },
1670
+ {
1671
+ "epoch": 14.120667522464698,
1672
+ "grad_norm": 0.29706746339797974,
1673
+ "learning_rate": 0.0003279077253218884,
1674
+ "loss": 2.4175,
1675
+ "step": 220000
1676
+ },
1677
+ {
1678
+ "epoch": 14.184852374839538,
1679
+ "grad_norm": 0.40730276703834534,
1680
+ "learning_rate": 0.00032433476394849785,
1681
+ "loss": 2.425,
1682
+ "step": 221000
1683
+ },
1684
+ {
1685
+ "epoch": 14.249037227214377,
1686
+ "grad_norm": 0.2597196400165558,
1687
+ "learning_rate": 0.000320758226037196,
1688
+ "loss": 2.4344,
1689
+ "step": 222000
1690
+ },
1691
+ {
1692
+ "epoch": 14.313222079589217,
1693
+ "grad_norm": 0.2576155662536621,
1694
+ "learning_rate": 0.00031718168812589414,
1695
+ "loss": 2.433,
1696
+ "step": 223000
1697
+ },
1698
+ {
1699
+ "epoch": 14.377406931964057,
1700
+ "grad_norm": 0.2582731246948242,
1701
+ "learning_rate": 0.0003136051502145923,
1702
+ "loss": 2.4325,
1703
+ "step": 224000
1704
+ },
1705
+ {
1706
+ "epoch": 14.441591784338897,
1707
+ "grad_norm": 0.25023168325424194,
1708
+ "learning_rate": 0.0003100321888412017,
1709
+ "loss": 2.4451,
1710
+ "step": 225000
1711
+ },
1712
+ {
1713
+ "epoch": 14.505776636713735,
1714
+ "grad_norm": 0.3350090980529785,
1715
+ "learning_rate": 0.00030645565092989984,
1716
+ "loss": 2.4445,
1717
+ "step": 226000
1718
+ },
1719
+ {
1720
+ "epoch": 14.569961489088575,
1721
+ "grad_norm": 0.30197829008102417,
1722
+ "learning_rate": 0.0003028826895565093,
1723
+ "loss": 2.4478,
1724
+ "step": 227000
1725
+ },
1726
+ {
1727
+ "epoch": 14.634146341463415,
1728
+ "grad_norm": 0.2431783527135849,
1729
+ "learning_rate": 0.00029930615164520746,
1730
+ "loss": 2.4494,
1731
+ "step": 228000
1732
+ },
1733
+ {
1734
+ "epoch": 14.698331193838253,
1735
+ "grad_norm": 0.2774644196033478,
1736
+ "learning_rate": 0.0002957296137339056,
1737
+ "loss": 2.4535,
1738
+ "step": 229000
1739
+ },
1740
+ {
1741
+ "epoch": 14.762516046213094,
1742
+ "grad_norm": 0.27773821353912354,
1743
+ "learning_rate": 0.000292156652360515,
1744
+ "loss": 2.4548,
1745
+ "step": 230000
1746
+ },
1747
+ {
1748
+ "epoch": 14.826700898587934,
1749
+ "grad_norm": 0.35124674439430237,
1750
+ "learning_rate": 0.0002885836909871244,
1751
+ "loss": 2.4565,
1752
+ "step": 231000
1753
+ },
1754
+ {
1755
+ "epoch": 14.890885750962774,
1756
+ "grad_norm": 0.28358736634254456,
1757
+ "learning_rate": 0.00028500715307582257,
1758
+ "loss": 2.4572,
1759
+ "step": 232000
1760
+ },
1761
+ {
1762
+ "epoch": 14.955070603337612,
1763
+ "grad_norm": 0.27410945296287537,
1764
+ "learning_rate": 0.0002814306151645207,
1765
+ "loss": 2.4607,
1766
+ "step": 233000
1767
+ },
1768
+ {
1769
+ "epoch": 15.0,
1770
+ "eval_accuracy": 0.4256197478991597,
1771
+ "eval_loss": 2.9826996326446533,
1772
+ "eval_runtime": 112.5916,
1773
+ "eval_samples_per_second": 467.531,
1774
+ "eval_steps_per_second": 7.31,
1775
+ "step": 233700
1776
+ },
1777
+ {
1778
+ "epoch": 15.019255455712452,
1779
+ "grad_norm": 0.27103665471076965,
1780
+ "learning_rate": 0.00027785407725321886,
1781
+ "loss": 2.4425,
1782
+ "step": 234000
1783
+ },
1784
+ {
1785
+ "epoch": 15.083440308087292,
1786
+ "grad_norm": 0.26552310585975647,
1787
+ "learning_rate": 0.00027428111587982833,
1788
+ "loss": 2.3947,
1789
+ "step": 235000
1790
+ },
1791
+ {
1792
+ "epoch": 15.14762516046213,
1793
+ "grad_norm": 0.2552954852581024,
1794
+ "learning_rate": 0.0002707045779685265,
1795
+ "loss": 2.4,
1796
+ "step": 236000
1797
+ },
1798
+ {
1799
+ "epoch": 15.21181001283697,
1800
+ "grad_norm": 0.27387532591819763,
1801
+ "learning_rate": 0.0002671280400572246,
1802
+ "loss": 2.4109,
1803
+ "step": 237000
1804
+ },
1805
+ {
1806
+ "epoch": 15.27599486521181,
1807
+ "grad_norm": 0.3243711292743683,
1808
+ "learning_rate": 0.00026355150214592277,
1809
+ "loss": 2.4071,
1810
+ "step": 238000
1811
+ },
1812
+ {
1813
+ "epoch": 15.340179717586649,
1814
+ "grad_norm": 0.27082788944244385,
1815
+ "learning_rate": 0.0002599785407725322,
1816
+ "loss": 2.4118,
1817
+ "step": 239000
1818
+ },
1819
+ {
1820
+ "epoch": 15.404364569961489,
1821
+ "grad_norm": 0.27366554737091064,
1822
+ "learning_rate": 0.0002564055793991416,
1823
+ "loss": 2.4123,
1824
+ "step": 240000
1825
+ },
1826
+ {
1827
+ "epoch": 15.468549422336329,
1828
+ "grad_norm": 0.2563120722770691,
1829
+ "learning_rate": 0.00025282904148783974,
1830
+ "loss": 2.4177,
1831
+ "step": 241000
1832
+ },
1833
+ {
1834
+ "epoch": 15.532734274711169,
1835
+ "grad_norm": 0.2995336949825287,
1836
+ "learning_rate": 0.00024925250357653794,
1837
+ "loss": 2.4236,
1838
+ "step": 242000
1839
+ },
1840
+ {
1841
+ "epoch": 15.596919127086007,
1842
+ "grad_norm": 0.2699737548828125,
1843
+ "learning_rate": 0.0002456759656652361,
1844
+ "loss": 2.4274,
1845
+ "step": 243000
1846
+ },
1847
+ {
1848
+ "epoch": 15.661103979460847,
1849
+ "grad_norm": 0.2669052481651306,
1850
+ "learning_rate": 0.0002420994277539342,
1851
+ "loss": 2.4262,
1852
+ "step": 244000
1853
+ },
1854
+ {
1855
+ "epoch": 15.725288831835687,
1856
+ "grad_norm": 0.2465328872203827,
1857
+ "learning_rate": 0.00023852646638054364,
1858
+ "loss": 2.4262,
1859
+ "step": 245000
1860
+ },
1861
+ {
1862
+ "epoch": 15.789473684210526,
1863
+ "grad_norm": 0.27883973717689514,
1864
+ "learning_rate": 0.0002349499284692418,
1865
+ "loss": 2.4294,
1866
+ "step": 246000
1867
+ },
1868
+ {
1869
+ "epoch": 15.853658536585366,
1870
+ "grad_norm": 0.29509106278419495,
1871
+ "learning_rate": 0.0002313769670958512,
1872
+ "loss": 2.4359,
1873
+ "step": 247000
1874
+ },
1875
+ {
1876
+ "epoch": 15.917843388960206,
1877
+ "grad_norm": 0.2847062945365906,
1878
+ "learning_rate": 0.00022780042918454934,
1879
+ "loss": 2.4355,
1880
+ "step": 248000
1881
+ },
1882
+ {
1883
+ "epoch": 15.982028241335044,
1884
+ "grad_norm": 0.2528593838214874,
1885
+ "learning_rate": 0.0002242238912732475,
1886
+ "loss": 2.4394,
1887
+ "step": 249000
1888
+ },
1889
+ {
1890
+ "epoch": 16.0,
1891
+ "eval_accuracy": 0.42555724417426544,
1892
+ "eval_loss": 2.993483543395996,
1893
+ "eval_runtime": 114.5219,
1894
+ "eval_samples_per_second": 459.65,
1895
+ "eval_steps_per_second": 7.186,
1896
+ "step": 249280
1897
+ },
1898
+ {
1899
+ "epoch": 16.046213093709884,
1900
+ "grad_norm": 0.300592303276062,
1901
+ "learning_rate": 0.00022065092989985693,
1902
+ "loss": 2.3904,
1903
+ "step": 250000
1904
+ },
1905
+ {
1906
+ "epoch": 16.110397946084724,
1907
+ "grad_norm": 0.26522037386894226,
1908
+ "learning_rate": 0.00021707439198855508,
1909
+ "loss": 2.3843,
1910
+ "step": 251000
1911
+ },
1912
+ {
1913
+ "epoch": 16.174582798459564,
1914
+ "grad_norm": 0.27688267827033997,
1915
+ "learning_rate": 0.00021350143061516454,
1916
+ "loss": 2.3842,
1917
+ "step": 252000
1918
+ },
1919
+ {
1920
+ "epoch": 16.238767650834404,
1921
+ "grad_norm": 0.28256699442863464,
1922
+ "learning_rate": 0.00020992489270386266,
1923
+ "loss": 2.3902,
1924
+ "step": 253000
1925
+ },
1926
+ {
1927
+ "epoch": 16.30295250320924,
1928
+ "grad_norm": 0.25413182377815247,
1929
+ "learning_rate": 0.00020635193133047212,
1930
+ "loss": 2.3907,
1931
+ "step": 254000
1932
+ },
1933
+ {
1934
+ "epoch": 16.36713735558408,
1935
+ "grad_norm": 0.2933899760246277,
1936
+ "learning_rate": 0.00020277539341917024,
1937
+ "loss": 2.3951,
1938
+ "step": 255000
1939
+ },
1940
+ {
1941
+ "epoch": 16.43132220795892,
1942
+ "grad_norm": 0.2763221263885498,
1943
+ "learning_rate": 0.0001992024320457797,
1944
+ "loss": 2.3993,
1945
+ "step": 256000
1946
+ },
1947
+ {
1948
+ "epoch": 16.49550706033376,
1949
+ "grad_norm": 0.27912837266921997,
1950
+ "learning_rate": 0.00019562589413447783,
1951
+ "loss": 2.3998,
1952
+ "step": 257000
1953
+ },
1954
+ {
1955
+ "epoch": 16.5596919127086,
1956
+ "grad_norm": 0.25394317507743835,
1957
+ "learning_rate": 0.00019204935622317595,
1958
+ "loss": 2.3953,
1959
+ "step": 258000
1960
+ },
1961
+ {
1962
+ "epoch": 16.62387676508344,
1963
+ "grad_norm": 0.264137327671051,
1964
+ "learning_rate": 0.0001884728183118741,
1965
+ "loss": 2.4006,
1966
+ "step": 259000
1967
+ },
1968
+ {
1969
+ "epoch": 16.68806161745828,
1970
+ "grad_norm": 0.3014145791530609,
1971
+ "learning_rate": 0.00018489985693848356,
1972
+ "loss": 2.4017,
1973
+ "step": 260000
1974
+ },
1975
+ {
1976
+ "epoch": 16.752246469833118,
1977
+ "grad_norm": 0.26256245374679565,
1978
+ "learning_rate": 0.00018132331902718168,
1979
+ "loss": 2.4029,
1980
+ "step": 261000
1981
+ },
1982
+ {
1983
+ "epoch": 16.816431322207958,
1984
+ "grad_norm": 0.2925412654876709,
1985
+ "learning_rate": 0.00017775035765379114,
1986
+ "loss": 2.4056,
1987
+ "step": 262000
1988
+ },
1989
+ {
1990
+ "epoch": 16.880616174582798,
1991
+ "grad_norm": 0.26922452449798584,
1992
+ "learning_rate": 0.00017417381974248926,
1993
+ "loss": 2.4098,
1994
+ "step": 263000
1995
+ },
1996
+ {
1997
+ "epoch": 16.944801026957638,
1998
+ "grad_norm": 0.27031683921813965,
1999
+ "learning_rate": 0.0001705972818311874,
2000
+ "loss": 2.4062,
2001
+ "step": 264000
2002
+ },
2003
+ {
2004
+ "epoch": 17.0,
2005
+ "eval_accuracy": 0.42530469634662377,
2006
+ "eval_loss": 3.0018820762634277,
2007
+ "eval_runtime": 113.4358,
2008
+ "eval_samples_per_second": 464.051,
2009
+ "eval_steps_per_second": 7.255,
2010
+ "step": 264860
2011
+ },
2012
+ {
2013
+ "epoch": 17.008985879332478,
2014
+ "grad_norm": 0.2683101296424866,
2015
+ "learning_rate": 0.00016702432045779687,
2016
+ "loss": 2.4048,
2017
+ "step": 265000
2018
+ },
2019
+ {
2020
+ "epoch": 17.073170731707318,
2021
+ "grad_norm": 0.3228018581867218,
2022
+ "learning_rate": 0.0001634513590844063,
2023
+ "loss": 2.3623,
2024
+ "step": 266000
2025
+ },
2026
+ {
2027
+ "epoch": 17.137355584082158,
2028
+ "grad_norm": 0.2701365351676941,
2029
+ "learning_rate": 0.00015987482117310446,
2030
+ "loss": 2.3628,
2031
+ "step": 267000
2032
+ },
2033
+ {
2034
+ "epoch": 17.201540436456995,
2035
+ "grad_norm": 0.28526586294174194,
2036
+ "learning_rate": 0.00015629828326180258,
2037
+ "loss": 2.3659,
2038
+ "step": 268000
2039
+ },
2040
+ {
2041
+ "epoch": 17.265725288831835,
2042
+ "grad_norm": 0.2874760925769806,
2043
+ "learning_rate": 0.0001527217453505007,
2044
+ "loss": 2.3639,
2045
+ "step": 269000
2046
+ },
2047
+ {
2048
+ "epoch": 17.329910141206675,
2049
+ "grad_norm": 0.2576092779636383,
2050
+ "learning_rate": 0.00014914878397711016,
2051
+ "loss": 2.372,
2052
+ "step": 270000
2053
+ },
2054
+ {
2055
+ "epoch": 17.394094993581515,
2056
+ "grad_norm": 0.3425317108631134,
2057
+ "learning_rate": 0.0001455722460658083,
2058
+ "loss": 2.3713,
2059
+ "step": 271000
2060
+ },
2061
+ {
2062
+ "epoch": 17.458279845956355,
2063
+ "grad_norm": 0.2755325436592102,
2064
+ "learning_rate": 0.00014199928469241775,
2065
+ "loss": 2.3775,
2066
+ "step": 272000
2067
+ },
2068
+ {
2069
+ "epoch": 17.522464698331195,
2070
+ "grad_norm": 0.2874724268913269,
2071
+ "learning_rate": 0.0001384227467811159,
2072
+ "loss": 2.3802,
2073
+ "step": 273000
2074
+ },
2075
+ {
2076
+ "epoch": 17.586649550706035,
2077
+ "grad_norm": 0.34283754229545593,
2078
+ "learning_rate": 0.00013484978540772533,
2079
+ "loss": 2.3774,
2080
+ "step": 274000
2081
+ },
2082
+ {
2083
+ "epoch": 17.65083440308087,
2084
+ "grad_norm": 0.2768910825252533,
2085
+ "learning_rate": 0.00013127682403433477,
2086
+ "loss": 2.376,
2087
+ "step": 275000
2088
+ },
2089
+ {
2090
+ "epoch": 17.71501925545571,
2091
+ "grad_norm": 0.263095498085022,
2092
+ "learning_rate": 0.00012770028612303291,
2093
+ "loss": 2.3822,
2094
+ "step": 276000
2095
+ },
2096
+ {
2097
+ "epoch": 17.77920410783055,
2098
+ "grad_norm": 0.2954932749271393,
2099
+ "learning_rate": 0.00012412374821173103,
2100
+ "loss": 2.3805,
2101
+ "step": 277000
2102
+ },
2103
+ {
2104
+ "epoch": 17.84338896020539,
2105
+ "grad_norm": 0.28619909286499023,
2106
+ "learning_rate": 0.00012054721030042918,
2107
+ "loss": 2.3861,
2108
+ "step": 278000
2109
+ },
2110
+ {
2111
+ "epoch": 17.90757381258023,
2112
+ "grad_norm": 0.3277396857738495,
2113
+ "learning_rate": 0.00011697067238912733,
2114
+ "loss": 2.3902,
2115
+ "step": 279000
2116
+ },
2117
+ {
2118
+ "epoch": 17.971758664955072,
2119
+ "grad_norm": 0.3048025965690613,
2120
+ "learning_rate": 0.00011339771101573678,
2121
+ "loss": 2.3845,
2122
+ "step": 280000
2123
+ },
2124
+ {
2125
+ "epoch": 18.0,
2126
+ "eval_accuracy": 0.42455025627272186,
2127
+ "eval_loss": 3.015594482421875,
2128
+ "eval_runtime": 113.2739,
2129
+ "eval_samples_per_second": 464.714,
2130
+ "eval_steps_per_second": 7.266,
2131
+ "step": 280440
2132
+ },
2133
+ {
2134
+ "epoch": 18.035943517329912,
2135
+ "grad_norm": 0.2825532853603363,
2136
+ "learning_rate": 0.00010982117310443491,
2137
+ "loss": 2.3589,
2138
+ "step": 281000
2139
+ },
2140
+ {
2141
+ "epoch": 18.10012836970475,
2142
+ "grad_norm": 0.319431871175766,
2143
+ "learning_rate": 0.00010624463519313304,
2144
+ "loss": 2.3481,
2145
+ "step": 282000
2146
+ },
2147
+ {
2148
+ "epoch": 18.16431322207959,
2149
+ "grad_norm": 0.27330195903778076,
2150
+ "learning_rate": 0.0001026716738197425,
2151
+ "loss": 2.3508,
2152
+ "step": 283000
2153
+ },
2154
+ {
2155
+ "epoch": 18.22849807445443,
2156
+ "grad_norm": 0.30694159865379333,
2157
+ "learning_rate": 9.909513590844063e-05,
2158
+ "loss": 2.3498,
2159
+ "step": 284000
2160
+ },
2161
+ {
2162
+ "epoch": 18.29268292682927,
2163
+ "grad_norm": 0.282497376203537,
2164
+ "learning_rate": 9.551859799713878e-05,
2165
+ "loss": 2.3433,
2166
+ "step": 285000
2167
+ },
2168
+ {
2169
+ "epoch": 18.35686777920411,
2170
+ "grad_norm": 0.29808005690574646,
2171
+ "learning_rate": 9.194206008583691e-05,
2172
+ "loss": 2.3527,
2173
+ "step": 286000
2174
+ },
2175
+ {
2176
+ "epoch": 18.42105263157895,
2177
+ "grad_norm": 0.27210718393325806,
2178
+ "learning_rate": 8.837267525035765e-05,
2179
+ "loss": 2.356,
2180
+ "step": 287000
2181
+ },
2182
+ {
2183
+ "epoch": 18.485237483953785,
2184
+ "grad_norm": 0.3066132962703705,
2185
+ "learning_rate": 8.47961373390558e-05,
2186
+ "loss": 2.3604,
2187
+ "step": 288000
2188
+ },
2189
+ {
2190
+ "epoch": 18.549422336328625,
2191
+ "grad_norm": 0.3081810176372528,
2192
+ "learning_rate": 8.121959942775393e-05,
2193
+ "loss": 2.353,
2194
+ "step": 289000
2195
+ },
2196
+ {
2197
+ "epoch": 18.613607188703465,
2198
+ "grad_norm": 0.31945279240608215,
2199
+ "learning_rate": 7.764306151645208e-05,
2200
+ "loss": 2.3577,
2201
+ "step": 290000
2202
+ },
2203
+ {
2204
+ "epoch": 18.677792041078305,
2205
+ "grad_norm": 0.2833667993545532,
2206
+ "learning_rate": 7.407010014306153e-05,
2207
+ "loss": 2.3567,
2208
+ "step": 291000
2209
+ },
2210
+ {
2211
+ "epoch": 18.741976893453145,
2212
+ "grad_norm": 0.30316558480262756,
2213
+ "learning_rate": 7.049356223175966e-05,
2214
+ "loss": 2.3606,
2215
+ "step": 292000
2216
+ },
2217
+ {
2218
+ "epoch": 18.806161745827985,
2219
+ "grad_norm": 0.2820417284965515,
2220
+ "learning_rate": 6.69170243204578e-05,
2221
+ "loss": 2.3615,
2222
+ "step": 293000
2223
+ },
2224
+ {
2225
+ "epoch": 18.870346598202826,
2226
+ "grad_norm": 0.2879752516746521,
2227
+ "learning_rate": 6.334406294706723e-05,
2228
+ "loss": 2.3626,
2229
+ "step": 294000
2230
+ },
2231
+ {
2232
+ "epoch": 18.934531450577662,
2233
+ "grad_norm": 0.3219081461429596,
2234
+ "learning_rate": 5.976752503576538e-05,
2235
+ "loss": 2.363,
2236
+ "step": 295000
2237
+ },
2238
+ {
2239
+ "epoch": 18.998716302952502,
2240
+ "grad_norm": 0.28323307633399963,
2241
+ "learning_rate": 5.619456366237482e-05,
2242
+ "loss": 2.3632,
2243
+ "step": 296000
2244
+ },
2245
+ {
2246
+ "epoch": 19.0,
2247
+ "eval_accuracy": 0.4244418618511234,
2248
+ "eval_loss": 3.0269229412078857,
2249
+ "eval_runtime": 113.5061,
2250
+ "eval_samples_per_second": 463.764,
2251
+ "eval_steps_per_second": 7.251,
2252
+ "step": 296020
2253
+ },
2254
+ {
2255
+ "epoch": 19.062901155327342,
2256
+ "grad_norm": 0.2774474322795868,
2257
+ "learning_rate": 5.261802575107296e-05,
2258
+ "loss": 2.3301,
2259
+ "step": 297000
2260
+ },
2261
+ {
2262
+ "epoch": 19.127086007702182,
2263
+ "grad_norm": 0.30794039368629456,
2264
+ "learning_rate": 4.90414878397711e-05,
2265
+ "loss": 2.3322,
2266
+ "step": 298000
2267
+ },
2268
+ {
2269
+ "epoch": 19.191270860077022,
2270
+ "grad_norm": 0.3270117938518524,
2271
+ "learning_rate": 4.546494992846924e-05,
2272
+ "loss": 2.3376,
2273
+ "step": 299000
2274
+ },
2275
+ {
2276
+ "epoch": 19.255455712451862,
2277
+ "grad_norm": 0.2921634614467621,
2278
+ "learning_rate": 4.189198855507869e-05,
2279
+ "loss": 2.3358,
2280
+ "step": 300000
2281
+ },
2282
+ {
2283
+ "epoch": 19.319640564826702,
2284
+ "grad_norm": 0.29304492473602295,
2285
+ "learning_rate": 3.8315450643776827e-05,
2286
+ "loss": 2.3367,
2287
+ "step": 301000
2288
+ },
2289
+ {
2290
+ "epoch": 19.38382541720154,
2291
+ "grad_norm": 0.2906641662120819,
2292
+ "learning_rate": 3.4742489270386264e-05,
2293
+ "loss": 2.3334,
2294
+ "step": 302000
2295
+ },
2296
+ {
2297
+ "epoch": 19.44801026957638,
2298
+ "grad_norm": 0.3057336211204529,
2299
+ "learning_rate": 3.1165951359084404e-05,
2300
+ "loss": 2.3384,
2301
+ "step": 303000
2302
+ },
2303
+ {
2304
+ "epoch": 19.51219512195122,
2305
+ "grad_norm": 0.2955686151981354,
2306
+ "learning_rate": 2.7589413447782547e-05,
2307
+ "loss": 2.3385,
2308
+ "step": 304000
2309
+ },
2310
+ {
2311
+ "epoch": 19.57637997432606,
2312
+ "grad_norm": 0.2858632206916809,
2313
+ "learning_rate": 2.401645207439199e-05,
2314
+ "loss": 2.3363,
2315
+ "step": 305000
2316
+ },
2317
+ {
2318
+ "epoch": 19.6405648267009,
2319
+ "grad_norm": 0.2869555652141571,
2320
+ "learning_rate": 2.0439914163090128e-05,
2321
+ "loss": 2.3394,
2322
+ "step": 306000
2323
+ },
2324
+ {
2325
+ "epoch": 19.70474967907574,
2326
+ "grad_norm": 0.28836971521377563,
2327
+ "learning_rate": 1.6866952789699572e-05,
2328
+ "loss": 2.3366,
2329
+ "step": 307000
2330
+ },
2331
+ {
2332
+ "epoch": 19.76893453145058,
2333
+ "grad_norm": 0.2759711742401123,
2334
+ "learning_rate": 1.329041487839771e-05,
2335
+ "loss": 2.3377,
2336
+ "step": 308000
2337
+ },
2338
+ {
2339
+ "epoch": 19.833119383825416,
2340
+ "grad_norm": 0.2844057083129883,
2341
+ "learning_rate": 9.71387696709585e-06,
2342
+ "loss": 2.3386,
2343
+ "step": 309000
2344
+ },
2345
+ {
2346
+ "epoch": 19.897304236200256,
2347
+ "grad_norm": 0.302503377199173,
2348
+ "learning_rate": 6.140915593705294e-06,
2349
+ "loss": 2.3388,
2350
+ "step": 310000
2351
+ },
2352
+ {
2353
+ "epoch": 19.961489088575096,
2354
+ "grad_norm": 0.2948252558708191,
2355
+ "learning_rate": 2.5643776824034334e-06,
2356
+ "loss": 2.3437,
2357
+ "step": 311000
2358
+ },
2359
+ {
2360
+ "epoch": 20.0,
2361
+ "eval_accuracy": 0.4239814649263961,
2362
+ "eval_loss": 3.035736322402954,
2363
+ "eval_runtime": 113.6992,
2364
+ "eval_samples_per_second": 462.976,
2365
+ "eval_steps_per_second": 7.238,
2366
+ "step": 311600
2367
+ },
2368
+ {
2369
+ "epoch": 20.0,
2370
+ "step": 311600,
2371
+ "total_flos": 1.30265052217344e+18,
2372
+ "train_loss": 2.6546336687942524,
2373
+ "train_runtime": 44210.8645,
2374
+ "train_samples_per_second": 225.529,
2375
+ "train_steps_per_second": 7.048
2376
+ }
2377
+ ],
2378
+ "logging_steps": 1000,
2379
+ "max_steps": 311600,
2380
+ "num_input_tokens_seen": 0,
2381
+ "num_train_epochs": 20,
2382
+ "save_steps": 5000,
2383
+ "stateful_callbacks": {
2384
+ "TrainerControl": {
2385
+ "args": {
2386
+ "should_epoch_stop": false,
2387
+ "should_evaluate": false,
2388
+ "should_log": false,
2389
+ "should_save": true,
2390
+ "should_training_stop": true
2391
+ },
2392
+ "attributes": {}
2393
+ }
2394
+ },
2395
+ "total_flos": 1.30265052217344e+18,
2396
+ "train_batch_size": 32,
2397
+ "trial_name": null,
2398
+ "trial_params": null
2399
+ }