lombardata commited on
Commit
fd90411
1 Parent(s): be54219

Evaluation on the test set completed on 2024_11_14.

Browse files
README.md ADDED
@@ -0,0 +1,156 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: facebook/dinov2-large
4
+ tags:
5
+ - generated_from_trainer
6
+ model-index:
7
+ - name: drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs
8
+ results: []
9
+ ---
10
+
11
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
12
+ should probably proofread and complete it, then remove this comment. -->
13
+
14
+ # drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs
15
+
16
+ This model is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large) on the None dataset.
17
+ It achieves the following results on the evaluation set:
18
+ - Loss: 0.4672
19
+ - Rmse: 0.1550
20
+ - Mae: 0.1155
21
+ - Kl Divergence: 0.3295
22
+ - Explained Variance: 0.4649
23
+ - Learning Rate: 0.0000
24
+
25
+ ## Model description
26
+
27
+ More information needed
28
+
29
+ ## Intended uses & limitations
30
+
31
+ More information needed
32
+
33
+ ## Training and evaluation data
34
+
35
+ More information needed
36
+
37
+ ## Training procedure
38
+
39
+ ### Training hyperparameters
40
+
41
+ The following hyperparameters were used during training:
42
+ - learning_rate: 0.001
43
+ - train_batch_size: 16
44
+ - eval_batch_size: 16
45
+ - seed: 42
46
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
47
+ - lr_scheduler_type: linear
48
+ - num_epochs: 150
49
+ - mixed_precision_training: Native AMP
50
+
51
+ ### Training results
52
+
53
+ | Training Loss | Epoch | Step | Validation Loss | Rmse | Mae | Kl Divergence | Explained Variance | Rate |
54
+ |:-------------:|:-----:|:-----:|:---------------:|:------:|:------:|:-------------:|:------------------:|:------:|
55
+ | No log | 1.0 | 438 | 0.4934 | 0.1825 | 0.1294 | 0.9903 | 0.3297 | 0.001 |
56
+ | 0.5313 | 2.0 | 876 | 0.4789 | 0.1716 | 0.1262 | 0.6847 | 0.3731 | 0.001 |
57
+ | 0.4831 | 3.0 | 1314 | 0.4788 | 0.1709 | 0.1271 | 0.5498 | 0.3824 | 0.001 |
58
+ | 0.4773 | 4.0 | 1752 | 0.4766 | 0.1695 | 0.1278 | 0.3131 | 0.3979 | 0.001 |
59
+ | 0.476 | 5.0 | 2190 | 0.4765 | 0.1687 | 0.1277 | 0.4013 | 0.3970 | 0.001 |
60
+ | 0.4746 | 6.0 | 2628 | 0.4765 | 0.1689 | 0.1243 | 0.6370 | 0.3924 | 0.001 |
61
+ | 0.4738 | 7.0 | 3066 | 0.4763 | 0.1694 | 0.1292 | 0.4314 | 0.3911 | 0.001 |
62
+ | 0.4727 | 8.0 | 3504 | 0.4755 | 0.1681 | 0.1267 | 0.3379 | 0.4076 | 0.001 |
63
+ | 0.4727 | 9.0 | 3942 | 0.4734 | 0.1662 | 0.1250 | 0.4916 | 0.4072 | 0.001 |
64
+ | 0.4715 | 10.0 | 4380 | 0.4755 | 0.1677 | 0.1277 | 0.3348 | 0.4062 | 0.001 |
65
+ | 0.4714 | 11.0 | 4818 | 0.4731 | 0.1659 | 0.1255 | 0.3524 | 0.4154 | 0.001 |
66
+ | 0.4713 | 12.0 | 5256 | 0.4768 | 0.1690 | 0.1306 | 0.2383 | 0.4103 | 0.001 |
67
+ | 0.4722 | 13.0 | 5694 | 0.4737 | 0.1666 | 0.1223 | 0.6968 | 0.4028 | 0.001 |
68
+ | 0.472 | 14.0 | 6132 | 0.4737 | 0.1658 | 0.1254 | 0.3983 | 0.4099 | 0.001 |
69
+ | 0.4697 | 15.0 | 6570 | 0.4739 | 0.1664 | 0.1248 | 0.5620 | 0.4036 | 0.001 |
70
+ | 0.4721 | 16.0 | 7008 | 0.4720 | 0.1648 | 0.1231 | 0.6049 | 0.4159 | 0.001 |
71
+ | 0.4721 | 17.0 | 7446 | 0.4741 | 0.1664 | 0.1265 | 0.3072 | 0.4171 | 0.001 |
72
+ | 0.4709 | 18.0 | 7884 | 0.4738 | 0.1650 | 0.1253 | 0.3350 | 0.4239 | 0.001 |
73
+ | 0.4711 | 19.0 | 8322 | 0.4763 | 0.1672 | 0.1282 | 0.2746 | 0.4162 | 0.001 |
74
+ | 0.4696 | 20.0 | 8760 | 0.4756 | 0.1670 | 0.1245 | 0.5659 | 0.4060 | 0.001 |
75
+ | 0.4715 | 21.0 | 9198 | 0.4734 | 0.1662 | 0.1230 | 0.6154 | 0.4061 | 0.001 |
76
+ | 0.4714 | 22.0 | 9636 | 0.4744 | 0.1677 | 0.1223 | 0.7974 | 0.4027 | 0.001 |
77
+ | 0.4697 | 23.0 | 10074 | 0.4721 | 0.1639 | 0.1252 | 0.2307 | 0.4337 | 0.0001 |
78
+ | 0.4653 | 24.0 | 10512 | 0.4706 | 0.1631 | 0.1217 | 0.4219 | 0.4314 | 0.0001 |
79
+ | 0.4653 | 25.0 | 10950 | 0.4688 | 0.1612 | 0.1195 | 0.5242 | 0.4371 | 0.0001 |
80
+ | 0.4665 | 26.0 | 11388 | 0.4693 | 0.1620 | 0.1190 | 0.6159 | 0.4338 | 0.0001 |
81
+ | 0.4638 | 27.0 | 11826 | 0.4685 | 0.1607 | 0.1206 | 0.4046 | 0.4416 | 0.0001 |
82
+ | 0.4647 | 28.0 | 12264 | 0.4694 | 0.1616 | 0.1220 | 0.2860 | 0.4443 | 0.0001 |
83
+ | 0.4644 | 29.0 | 12702 | 0.4689 | 0.1614 | 0.1197 | 0.4270 | 0.4401 | 0.0001 |
84
+ | 0.4638 | 30.0 | 13140 | 0.4699 | 0.1619 | 0.1225 | 0.2625 | 0.4436 | 0.0001 |
85
+ | 0.4636 | 31.0 | 13578 | 0.4684 | 0.1607 | 0.1197 | 0.3876 | 0.4431 | 0.0001 |
86
+ | 0.463 | 32.0 | 14016 | 0.4678 | 0.1600 | 0.1195 | 0.4060 | 0.4467 | 0.0001 |
87
+ | 0.463 | 33.0 | 14454 | 0.4676 | 0.1596 | 0.1193 | 0.3688 | 0.4494 | 0.0001 |
88
+ | 0.4628 | 34.0 | 14892 | 0.4677 | 0.1600 | 0.1194 | 0.3900 | 0.4491 | 0.0001 |
89
+ | 0.4616 | 35.0 | 15330 | 0.4670 | 0.1593 | 0.1189 | 0.4282 | 0.4500 | 0.0001 |
90
+ | 0.4634 | 36.0 | 15768 | 0.4668 | 0.1591 | 0.1180 | 0.4446 | 0.4506 | 0.0001 |
91
+ | 0.462 | 37.0 | 16206 | 0.4669 | 0.1590 | 0.1185 | 0.3942 | 0.4528 | 0.0001 |
92
+ | 0.4631 | 38.0 | 16644 | 0.4665 | 0.1588 | 0.1177 | 0.4783 | 0.4512 | 0.0001 |
93
+ | 0.4603 | 39.0 | 17082 | 0.4674 | 0.1597 | 0.1190 | 0.3868 | 0.4500 | 0.0001 |
94
+ | 0.4614 | 40.0 | 17520 | 0.4677 | 0.1599 | 0.1195 | 0.3627 | 0.4498 | 0.0001 |
95
+ | 0.4614 | 41.0 | 17958 | 0.4682 | 0.1602 | 0.1211 | 0.2655 | 0.4540 | 0.0001 |
96
+ | 0.4612 | 42.0 | 18396 | 0.4665 | 0.1589 | 0.1172 | 0.5072 | 0.4514 | 0.0001 |
97
+ | 0.462 | 43.0 | 18834 | 0.4664 | 0.1585 | 0.1177 | 0.4306 | 0.4555 | 0.0001 |
98
+ | 0.4603 | 44.0 | 19272 | 0.4671 | 0.1594 | 0.1192 | 0.4115 | 0.4529 | 0.0001 |
99
+ | 0.4599 | 45.0 | 19710 | 0.4666 | 0.1590 | 0.1171 | 0.4417 | 0.4528 | 0.0001 |
100
+ | 0.4612 | 46.0 | 20148 | 0.4663 | 0.1585 | 0.1179 | 0.3686 | 0.4574 | 0.0001 |
101
+ | 0.4596 | 47.0 | 20586 | 0.4658 | 0.1582 | 0.1172 | 0.5090 | 0.4567 | 0.0001 |
102
+ | 0.4603 | 48.0 | 21024 | 0.4663 | 0.1589 | 0.1175 | 0.5279 | 0.4548 | 0.0001 |
103
+ | 0.4603 | 49.0 | 21462 | 0.4666 | 0.1591 | 0.1183 | 0.4497 | 0.4532 | 0.0001 |
104
+ | 0.4599 | 50.0 | 21900 | 0.4676 | 0.1595 | 0.1205 | 0.2712 | 0.4580 | 0.0001 |
105
+ | 0.4594 | 51.0 | 22338 | 0.4664 | 0.1586 | 0.1172 | 0.4008 | 0.4552 | 0.0001 |
106
+ | 0.4593 | 52.0 | 22776 | 0.4659 | 0.1583 | 0.1163 | 0.4922 | 0.4557 | 0.0001 |
107
+ | 0.4614 | 53.0 | 23214 | 0.4657 | 0.1579 | 0.1178 | 0.4274 | 0.4601 | 0.0001 |
108
+ | 0.4592 | 54.0 | 23652 | 0.4663 | 0.1585 | 0.1158 | 0.4574 | 0.4570 | 0.0001 |
109
+ | 0.4601 | 55.0 | 24090 | 0.4664 | 0.1586 | 0.1189 | 0.3486 | 0.4580 | 0.0001 |
110
+ | 0.4589 | 56.0 | 24528 | 0.4662 | 0.1584 | 0.1184 | 0.3016 | 0.4612 | 0.0001 |
111
+ | 0.4589 | 57.0 | 24966 | 0.4663 | 0.1587 | 0.1181 | 0.4163 | 0.4553 | 0.0001 |
112
+ | 0.4588 | 58.0 | 25404 | 0.4674 | 0.1593 | 0.1189 | 0.3399 | 0.4557 | 0.0001 |
113
+ | 0.4595 | 59.0 | 25842 | 0.4650 | 0.1572 | 0.1170 | 0.3686 | 0.4646 | 0.0001 |
114
+ | 0.4594 | 60.0 | 26280 | 0.4660 | 0.1584 | 0.1172 | 0.4535 | 0.4567 | 0.0001 |
115
+ | 0.4599 | 61.0 | 26718 | 0.4662 | 0.1585 | 0.1179 | 0.3751 | 0.4584 | 0.0001 |
116
+ | 0.4584 | 62.0 | 27156 | 0.4661 | 0.1583 | 0.1173 | 0.3534 | 0.4588 | 0.0001 |
117
+ | 0.4575 | 63.0 | 27594 | 0.4660 | 0.1583 | 0.1163 | 0.4048 | 0.4577 | 0.0001 |
118
+ | 0.4598 | 64.0 | 28032 | 0.4671 | 0.1588 | 0.1188 | 0.2471 | 0.4629 | 0.0001 |
119
+ | 0.4598 | 65.0 | 28470 | 0.4654 | 0.1577 | 0.1166 | 0.4526 | 0.4604 | 0.0001 |
120
+ | 0.4582 | 66.0 | 28908 | 0.4657 | 0.1582 | 0.1161 | 0.5259 | 0.4592 | 1e-05 |
121
+ | 0.4592 | 67.0 | 29346 | 0.4654 | 0.1574 | 0.1173 | 0.4252 | 0.4623 | 1e-05 |
122
+ | 0.4573 | 68.0 | 29784 | 0.4649 | 0.1572 | 0.1154 | 0.4989 | 0.4614 | 1e-05 |
123
+ | 0.4576 | 69.0 | 30222 | 0.4651 | 0.1570 | 0.1161 | 0.4023 | 0.4644 | 1e-05 |
124
+ | 0.4556 | 70.0 | 30660 | 0.4660 | 0.1576 | 0.1166 | 0.4118 | 0.4622 | 1e-05 |
125
+ | 0.4591 | 71.0 | 31098 | 0.4661 | 0.1578 | 0.1177 | 0.3075 | 0.4644 | 1e-05 |
126
+ | 0.4563 | 72.0 | 31536 | 0.4658 | 0.1580 | 0.1171 | 0.3836 | 0.4621 | 1e-05 |
127
+ | 0.4563 | 73.0 | 31974 | 0.4649 | 0.1569 | 0.1154 | 0.4544 | 0.4640 | 1e-05 |
128
+ | 0.4577 | 74.0 | 32412 | 0.4647 | 0.1567 | 0.1163 | 0.4538 | 0.4660 | 1e-05 |
129
+ | 0.4576 | 75.0 | 32850 | 0.4656 | 0.1573 | 0.1166 | 0.3348 | 0.4658 | 1e-05 |
130
+ | 0.457 | 76.0 | 33288 | 0.4647 | 0.1571 | 0.1158 | 0.4976 | 0.4645 | 1e-05 |
131
+ | 0.4574 | 77.0 | 33726 | 0.4651 | 0.1570 | 0.1163 | 0.3934 | 0.4653 | 1e-05 |
132
+ | 0.457 | 78.0 | 34164 | 0.4650 | 0.1571 | 0.1161 | 0.3936 | 0.4654 | 1e-05 |
133
+ | 0.4566 | 79.0 | 34602 | 0.4653 | 0.1573 | 0.1159 | 0.3759 | 0.4653 | 1e-05 |
134
+ | 0.458 | 80.0 | 35040 | 0.4647 | 0.1567 | 0.1162 | 0.4189 | 0.4660 | 1e-05 |
135
+ | 0.458 | 81.0 | 35478 | 0.4649 | 0.1571 | 0.1158 | 0.4751 | 0.4647 | 0.0000 |
136
+ | 0.456 | 82.0 | 35916 | 0.4654 | 0.1572 | 0.1161 | 0.4335 | 0.4651 | 0.0000 |
137
+ | 0.4564 | 83.0 | 36354 | 0.4647 | 0.1566 | 0.1161 | 0.3906 | 0.4667 | 0.0000 |
138
+ | 0.4575 | 84.0 | 36792 | 0.4643 | 0.1564 | 0.1157 | 0.3855 | 0.4677 | 0.0000 |
139
+ | 0.4557 | 85.0 | 37230 | 0.4653 | 0.1571 | 0.1173 | 0.3372 | 0.4669 | 0.0000 |
140
+ | 0.4587 | 86.0 | 37668 | 0.4655 | 0.1572 | 0.1184 | 0.2969 | 0.4686 | 0.0000 |
141
+ | 0.4564 | 87.0 | 38106 | 0.4652 | 0.1571 | 0.1173 | 0.3572 | 0.4670 | 0.0000 |
142
+ | 0.4565 | 88.0 | 38544 | 0.4656 | 0.1578 | 0.1151 | 0.5179 | 0.4627 | 0.0000 |
143
+ | 0.4565 | 89.0 | 38982 | 0.4654 | 0.1574 | 0.1177 | 0.2948 | 0.4670 | 0.0000 |
144
+ | 0.4569 | 90.0 | 39420 | 0.4650 | 0.1569 | 0.1167 | 0.3427 | 0.4674 | 0.0000 |
145
+ | 0.4561 | 91.0 | 39858 | 0.4655 | 0.1572 | 0.1173 | 0.2790 | 0.4691 | 0.0000 |
146
+ | 0.4575 | 92.0 | 40296 | 0.4646 | 0.1566 | 0.1153 | 0.4153 | 0.4672 | 0.0000 |
147
+ | 0.4569 | 93.0 | 40734 | 0.4649 | 0.1571 | 0.1153 | 0.4664 | 0.4645 | 0.0000 |
148
+ | 0.456 | 94.0 | 41172 | 0.4653 | 0.1568 | 0.1159 | 0.3859 | 0.4662 | 0.0000 |
149
+
150
+
151
+ ### Framework versions
152
+
153
+ - Transformers 4.41.0
154
+ - Pytorch 2.5.0+cu124
155
+ - Datasets 3.0.2
156
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 94.0,
3
+ "eval_explained_variance": 0.4649323523044586,
4
+ "eval_kl_divergence": 0.3295058012008667,
5
+ "eval_loss": 0.46722450852394104,
6
+ "eval_mae": 0.11545699089765549,
7
+ "eval_rmse": 0.15503521263599396,
8
+ "eval_runtime": 63.8487,
9
+ "eval_samples_per_second": 36.915,
10
+ "eval_steps_per_second": 2.318,
11
+ "learning_rate": 1.0000000000000002e-07,
12
+ "total_flos": 9.743166860733663e+19,
13
+ "train_loss": 0.46343856458701926,
14
+ "train_runtime": 26810.1674,
15
+ "train_samples_per_second": 39.181,
16
+ "train_steps_per_second": 2.451
17
+ }
logs/events.out.tfevents.1731579971.datavisu2 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ce197a3964b5b48570a9227ef122b5c81bf61528c8bf4985c6143dbedf03efdf
3
- size 74004
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1500f50c57053313422ea4d93f432942296a509f10b72b4d993797c46d389007
3
+ size 75744
logs/events.out.tfevents.1731606943.datavisu2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b89cf9e214139c5c296f78293cb37febf4da9130e0e3cdb1b41af35b90fc9986
3
+ size 40
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1b5c7e833d8c3a978ee56e19569c0153147472cf8ff76d18918d83ec1844b827
3
  size 1222956704
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9111fac77c578e4f47628d8742c5a82e04d0d229294950f2d300cca771a7095
3
  size 1222956704
test_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 94.0,
3
+ "eval_explained_variance": 0.4649323523044586,
4
+ "eval_kl_divergence": 0.3295058012008667,
5
+ "eval_loss": 0.46722450852394104,
6
+ "eval_mae": 0.11545699089765549,
7
+ "eval_rmse": 0.15503521263599396,
8
+ "eval_runtime": 63.8487,
9
+ "eval_samples_per_second": 36.915,
10
+ "eval_steps_per_second": 2.318,
11
+ "learning_rate": 1.0000000000000002e-07
12
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 94.0,
3
+ "learning_rate": 1.0000000000000002e-07,
4
+ "total_flos": 9.743166860733663e+19,
5
+ "train_loss": 0.46343856458701926,
6
+ "train_runtime": 26810.1674,
7
+ "train_samples_per_second": 39.181,
8
+ "train_steps_per_second": 2.451
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1848 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.46430692076683044,
3
+ "best_model_checkpoint": "/home/datawork-iot-nos/Seatizen/models/multilabel/drone/drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs/checkpoint-36792",
4
+ "epoch": 94.0,
5
+ "eval_steps": 500,
6
+ "global_step": 41172,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 1.0,
13
+ "eval_explained_variance": 0.3296825587749481,
14
+ "eval_kl_divergence": 0.9903448224067688,
15
+ "eval_loss": 0.4933677613735199,
16
+ "eval_mae": 0.129361093044281,
17
+ "eval_rmse": 0.18249033391475677,
18
+ "eval_runtime": 70.7501,
19
+ "eval_samples_per_second": 33.272,
20
+ "eval_steps_per_second": 2.092,
21
+ "learning_rate": 0.001,
22
+ "step": 438
23
+ },
24
+ {
25
+ "epoch": 1.1415525114155252,
26
+ "grad_norm": 0.5407673120498657,
27
+ "learning_rate": 0.001,
28
+ "loss": 0.5313,
29
+ "step": 500
30
+ },
31
+ {
32
+ "epoch": 2.0,
33
+ "eval_explained_variance": 0.37307724356651306,
34
+ "eval_kl_divergence": 0.6847184896469116,
35
+ "eval_loss": 0.47886165976524353,
36
+ "eval_mae": 0.12624560296535492,
37
+ "eval_rmse": 0.1715712547302246,
38
+ "eval_runtime": 71.8777,
39
+ "eval_samples_per_second": 32.75,
40
+ "eval_steps_per_second": 2.059,
41
+ "learning_rate": 0.001,
42
+ "step": 876
43
+ },
44
+ {
45
+ "epoch": 2.2831050228310503,
46
+ "grad_norm": 0.43918001651763916,
47
+ "learning_rate": 0.001,
48
+ "loss": 0.4831,
49
+ "step": 1000
50
+ },
51
+ {
52
+ "epoch": 3.0,
53
+ "eval_explained_variance": 0.38236939907073975,
54
+ "eval_kl_divergence": 0.5498285293579102,
55
+ "eval_loss": 0.4788369834423065,
56
+ "eval_mae": 0.12710730731487274,
57
+ "eval_rmse": 0.1709037572145462,
58
+ "eval_runtime": 66.8689,
59
+ "eval_samples_per_second": 35.203,
60
+ "eval_steps_per_second": 2.213,
61
+ "learning_rate": 0.001,
62
+ "step": 1314
63
+ },
64
+ {
65
+ "epoch": 3.4246575342465753,
66
+ "grad_norm": 0.28072813153266907,
67
+ "learning_rate": 0.001,
68
+ "loss": 0.4773,
69
+ "step": 1500
70
+ },
71
+ {
72
+ "epoch": 4.0,
73
+ "eval_explained_variance": 0.39789319038391113,
74
+ "eval_kl_divergence": 0.31306192278862,
75
+ "eval_loss": 0.47656363248825073,
76
+ "eval_mae": 0.12784750759601593,
77
+ "eval_rmse": 0.16946662962436676,
78
+ "eval_runtime": 68.7002,
79
+ "eval_samples_per_second": 34.265,
80
+ "eval_steps_per_second": 2.154,
81
+ "learning_rate": 0.001,
82
+ "step": 1752
83
+ },
84
+ {
85
+ "epoch": 4.566210045662101,
86
+ "grad_norm": 0.23626942932605743,
87
+ "learning_rate": 0.001,
88
+ "loss": 0.476,
89
+ "step": 2000
90
+ },
91
+ {
92
+ "epoch": 5.0,
93
+ "eval_explained_variance": 0.3970318138599396,
94
+ "eval_kl_divergence": 0.40128740668296814,
95
+ "eval_loss": 0.4765072166919708,
96
+ "eval_mae": 0.12769053876399994,
97
+ "eval_rmse": 0.1686621755361557,
98
+ "eval_runtime": 69.4052,
99
+ "eval_samples_per_second": 33.917,
100
+ "eval_steps_per_second": 2.132,
101
+ "learning_rate": 0.001,
102
+ "step": 2190
103
+ },
104
+ {
105
+ "epoch": 5.707762557077626,
106
+ "grad_norm": 0.2325536459684372,
107
+ "learning_rate": 0.001,
108
+ "loss": 0.4746,
109
+ "step": 2500
110
+ },
111
+ {
112
+ "epoch": 6.0,
113
+ "eval_explained_variance": 0.392393559217453,
114
+ "eval_kl_divergence": 0.6369896531105042,
115
+ "eval_loss": 0.47649845480918884,
116
+ "eval_mae": 0.12429385632276535,
117
+ "eval_rmse": 0.16892628371715546,
118
+ "eval_runtime": 69.6508,
119
+ "eval_samples_per_second": 33.797,
120
+ "eval_steps_per_second": 2.125,
121
+ "learning_rate": 0.001,
122
+ "step": 2628
123
+ },
124
+ {
125
+ "epoch": 6.8493150684931505,
126
+ "grad_norm": 0.1943524181842804,
127
+ "learning_rate": 0.001,
128
+ "loss": 0.4738,
129
+ "step": 3000
130
+ },
131
+ {
132
+ "epoch": 7.0,
133
+ "eval_explained_variance": 0.3910720646381378,
134
+ "eval_kl_divergence": 0.43142321705818176,
135
+ "eval_loss": 0.47629299759864807,
136
+ "eval_mae": 0.1291760504245758,
137
+ "eval_rmse": 0.16937647759914398,
138
+ "eval_runtime": 68.3166,
139
+ "eval_samples_per_second": 34.457,
140
+ "eval_steps_per_second": 2.166,
141
+ "learning_rate": 0.001,
142
+ "step": 3066
143
+ },
144
+ {
145
+ "epoch": 7.9908675799086755,
146
+ "grad_norm": 0.161671444773674,
147
+ "learning_rate": 0.001,
148
+ "loss": 0.4727,
149
+ "step": 3500
150
+ },
151
+ {
152
+ "epoch": 8.0,
153
+ "eval_explained_variance": 0.407635897397995,
154
+ "eval_kl_divergence": 0.33790889382362366,
155
+ "eval_loss": 0.4755041003227234,
156
+ "eval_mae": 0.1266549676656723,
157
+ "eval_rmse": 0.1681331843137741,
158
+ "eval_runtime": 71.2607,
159
+ "eval_samples_per_second": 33.034,
160
+ "eval_steps_per_second": 2.077,
161
+ "learning_rate": 0.001,
162
+ "step": 3504
163
+ },
164
+ {
165
+ "epoch": 9.0,
166
+ "eval_explained_variance": 0.40718480944633484,
167
+ "eval_kl_divergence": 0.4916018843650818,
168
+ "eval_loss": 0.47342246770858765,
169
+ "eval_mae": 0.1250395029783249,
170
+ "eval_rmse": 0.16618509590625763,
171
+ "eval_runtime": 68.5459,
172
+ "eval_samples_per_second": 34.342,
173
+ "eval_steps_per_second": 2.159,
174
+ "learning_rate": 0.001,
175
+ "step": 3942
176
+ },
177
+ {
178
+ "epoch": 9.132420091324201,
179
+ "grad_norm": 0.19357167184352875,
180
+ "learning_rate": 0.001,
181
+ "loss": 0.4715,
182
+ "step": 4000
183
+ },
184
+ {
185
+ "epoch": 10.0,
186
+ "eval_explained_variance": 0.40620386600494385,
187
+ "eval_kl_divergence": 0.33481186628341675,
188
+ "eval_loss": 0.47546806931495667,
189
+ "eval_mae": 0.1277414709329605,
190
+ "eval_rmse": 0.167744979262352,
191
+ "eval_runtime": 67.1085,
192
+ "eval_samples_per_second": 35.078,
193
+ "eval_steps_per_second": 2.205,
194
+ "learning_rate": 0.001,
195
+ "step": 4380
196
+ },
197
+ {
198
+ "epoch": 10.273972602739725,
199
+ "grad_norm": 0.17573006451129913,
200
+ "learning_rate": 0.001,
201
+ "loss": 0.4714,
202
+ "step": 4500
203
+ },
204
+ {
205
+ "epoch": 11.0,
206
+ "eval_explained_variance": 0.41538166999816895,
207
+ "eval_kl_divergence": 0.35237056016921997,
208
+ "eval_loss": 0.4731104075908661,
209
+ "eval_mae": 0.12549267709255219,
210
+ "eval_rmse": 0.1659194976091385,
211
+ "eval_runtime": 67.2129,
212
+ "eval_samples_per_second": 35.023,
213
+ "eval_steps_per_second": 2.202,
214
+ "learning_rate": 0.001,
215
+ "step": 4818
216
+ },
217
+ {
218
+ "epoch": 11.415525114155251,
219
+ "grad_norm": 0.13790665566921234,
220
+ "learning_rate": 0.001,
221
+ "loss": 0.4713,
222
+ "step": 5000
223
+ },
224
+ {
225
+ "epoch": 12.0,
226
+ "eval_explained_variance": 0.41025033593177795,
227
+ "eval_kl_divergence": 0.23825553059577942,
228
+ "eval_loss": 0.47679492831230164,
229
+ "eval_mae": 0.13060903549194336,
230
+ "eval_rmse": 0.16900604963302612,
231
+ "eval_runtime": 68.7453,
232
+ "eval_samples_per_second": 34.242,
233
+ "eval_steps_per_second": 2.153,
234
+ "learning_rate": 0.001,
235
+ "step": 5256
236
+ },
237
+ {
238
+ "epoch": 12.557077625570777,
239
+ "grad_norm": 0.1497349739074707,
240
+ "learning_rate": 0.001,
241
+ "loss": 0.4722,
242
+ "step": 5500
243
+ },
244
+ {
245
+ "epoch": 13.0,
246
+ "eval_explained_variance": 0.40282678604125977,
247
+ "eval_kl_divergence": 0.696779727935791,
248
+ "eval_loss": 0.4736888110637665,
249
+ "eval_mae": 0.12227921932935715,
250
+ "eval_rmse": 0.16662709414958954,
251
+ "eval_runtime": 66.9685,
252
+ "eval_samples_per_second": 35.151,
253
+ "eval_steps_per_second": 2.21,
254
+ "learning_rate": 0.001,
255
+ "step": 5694
256
+ },
257
+ {
258
+ "epoch": 13.698630136986301,
259
+ "grad_norm": 0.11125459522008896,
260
+ "learning_rate": 0.001,
261
+ "loss": 0.472,
262
+ "step": 6000
263
+ },
264
+ {
265
+ "epoch": 14.0,
266
+ "eval_explained_variance": 0.4099460542201996,
267
+ "eval_kl_divergence": 0.3982764780521393,
268
+ "eval_loss": 0.4736703634262085,
269
+ "eval_mae": 0.12543360888957977,
270
+ "eval_rmse": 0.16584907472133636,
271
+ "eval_runtime": 70.7133,
272
+ "eval_samples_per_second": 33.289,
273
+ "eval_steps_per_second": 2.093,
274
+ "learning_rate": 0.001,
275
+ "step": 6132
276
+ },
277
+ {
278
+ "epoch": 14.840182648401827,
279
+ "grad_norm": 0.13986219465732574,
280
+ "learning_rate": 0.001,
281
+ "loss": 0.4697,
282
+ "step": 6500
283
+ },
284
+ {
285
+ "epoch": 15.0,
286
+ "eval_explained_variance": 0.40361204743385315,
287
+ "eval_kl_divergence": 0.5619756579399109,
288
+ "eval_loss": 0.4738818407058716,
289
+ "eval_mae": 0.12480553239583969,
290
+ "eval_rmse": 0.16644911468029022,
291
+ "eval_runtime": 67.989,
292
+ "eval_samples_per_second": 34.623,
293
+ "eval_steps_per_second": 2.177,
294
+ "learning_rate": 0.001,
295
+ "step": 6570
296
+ },
297
+ {
298
+ "epoch": 15.981735159817351,
299
+ "grad_norm": 0.10489701479673386,
300
+ "learning_rate": 0.001,
301
+ "loss": 0.4721,
302
+ "step": 7000
303
+ },
304
+ {
305
+ "epoch": 16.0,
306
+ "eval_explained_variance": 0.41589656472206116,
307
+ "eval_kl_divergence": 0.6049214601516724,
308
+ "eval_loss": 0.47202879190444946,
309
+ "eval_mae": 0.12306753545999527,
310
+ "eval_rmse": 0.1648208647966385,
311
+ "eval_runtime": 68.6318,
312
+ "eval_samples_per_second": 34.299,
313
+ "eval_steps_per_second": 2.156,
314
+ "learning_rate": 0.001,
315
+ "step": 7008
316
+ },
317
+ {
318
+ "epoch": 17.0,
319
+ "eval_explained_variance": 0.41705062985420227,
320
+ "eval_kl_divergence": 0.3072437644004822,
321
+ "eval_loss": 0.47406336665153503,
322
+ "eval_mae": 0.12645679712295532,
323
+ "eval_rmse": 0.16636626422405243,
324
+ "eval_runtime": 67.693,
325
+ "eval_samples_per_second": 34.775,
326
+ "eval_steps_per_second": 2.186,
327
+ "learning_rate": 0.001,
328
+ "step": 7446
329
+ },
330
+ {
331
+ "epoch": 17.123287671232877,
332
+ "grad_norm": 0.11510255187749863,
333
+ "learning_rate": 0.001,
334
+ "loss": 0.4709,
335
+ "step": 7500
336
+ },
337
+ {
338
+ "epoch": 18.0,
339
+ "eval_explained_variance": 0.4238598346710205,
340
+ "eval_kl_divergence": 0.3350007236003876,
341
+ "eval_loss": 0.4738321006298065,
342
+ "eval_mae": 0.12525837123394012,
343
+ "eval_rmse": 0.1649676412343979,
344
+ "eval_runtime": 68.1862,
345
+ "eval_samples_per_second": 34.523,
346
+ "eval_steps_per_second": 2.171,
347
+ "learning_rate": 0.001,
348
+ "step": 7884
349
+ },
350
+ {
351
+ "epoch": 18.264840182648403,
352
+ "grad_norm": 0.1282125860452652,
353
+ "learning_rate": 0.001,
354
+ "loss": 0.4711,
355
+ "step": 8000
356
+ },
357
+ {
358
+ "epoch": 19.0,
359
+ "eval_explained_variance": 0.4161747395992279,
360
+ "eval_kl_divergence": 0.2745565176010132,
361
+ "eval_loss": 0.476326048374176,
362
+ "eval_mae": 0.12823572754859924,
363
+ "eval_rmse": 0.16715575754642487,
364
+ "eval_runtime": 68.1658,
365
+ "eval_samples_per_second": 34.533,
366
+ "eval_steps_per_second": 2.171,
367
+ "learning_rate": 0.001,
368
+ "step": 8322
369
+ },
370
+ {
371
+ "epoch": 19.40639269406393,
372
+ "grad_norm": 0.15439559519290924,
373
+ "learning_rate": 0.001,
374
+ "loss": 0.4696,
375
+ "step": 8500
376
+ },
377
+ {
378
+ "epoch": 20.0,
379
+ "eval_explained_variance": 0.40602585673332214,
380
+ "eval_kl_divergence": 0.5658758878707886,
381
+ "eval_loss": 0.4755523204803467,
382
+ "eval_mae": 0.12445954233407974,
383
+ "eval_rmse": 0.1669723242521286,
384
+ "eval_runtime": 66.5326,
385
+ "eval_samples_per_second": 35.381,
386
+ "eval_steps_per_second": 2.224,
387
+ "learning_rate": 0.001,
388
+ "step": 8760
389
+ },
390
+ {
391
+ "epoch": 20.54794520547945,
392
+ "grad_norm": 0.1288071572780609,
393
+ "learning_rate": 0.001,
394
+ "loss": 0.4715,
395
+ "step": 9000
396
+ },
397
+ {
398
+ "epoch": 21.0,
399
+ "eval_explained_variance": 0.40606603026390076,
400
+ "eval_kl_divergence": 0.6153907179832458,
401
+ "eval_loss": 0.47340598702430725,
402
+ "eval_mae": 0.12303853780031204,
403
+ "eval_rmse": 0.16621015965938568,
404
+ "eval_runtime": 66.3328,
405
+ "eval_samples_per_second": 35.488,
406
+ "eval_steps_per_second": 2.231,
407
+ "learning_rate": 0.001,
408
+ "step": 9198
409
+ },
410
+ {
411
+ "epoch": 21.689497716894977,
412
+ "grad_norm": 0.13872142136096954,
413
+ "learning_rate": 0.001,
414
+ "loss": 0.4714,
415
+ "step": 9500
416
+ },
417
+ {
418
+ "epoch": 22.0,
419
+ "eval_explained_variance": 0.402675986289978,
420
+ "eval_kl_divergence": 0.7973779439926147,
421
+ "eval_loss": 0.47443991899490356,
422
+ "eval_mae": 0.12229780107736588,
423
+ "eval_rmse": 0.16771160066127777,
424
+ "eval_runtime": 65.4256,
425
+ "eval_samples_per_second": 35.98,
426
+ "eval_steps_per_second": 2.262,
427
+ "learning_rate": 0.001,
428
+ "step": 9636
429
+ },
430
+ {
431
+ "epoch": 22.831050228310502,
432
+ "grad_norm": 0.12462610751390457,
433
+ "learning_rate": 0.0001,
434
+ "loss": 0.4697,
435
+ "step": 10000
436
+ },
437
+ {
438
+ "epoch": 23.0,
439
+ "eval_explained_variance": 0.43371883034706116,
440
+ "eval_kl_divergence": 0.2306971251964569,
441
+ "eval_loss": 0.47205689549446106,
442
+ "eval_mae": 0.12522436678409576,
443
+ "eval_rmse": 0.16393429040908813,
444
+ "eval_runtime": 67.2945,
445
+ "eval_samples_per_second": 34.981,
446
+ "eval_steps_per_second": 2.199,
447
+ "learning_rate": 0.0001,
448
+ "step": 10074
449
+ },
450
+ {
451
+ "epoch": 23.972602739726028,
452
+ "grad_norm": 0.15716814994812012,
453
+ "learning_rate": 0.0001,
454
+ "loss": 0.4653,
455
+ "step": 10500
456
+ },
457
+ {
458
+ "epoch": 24.0,
459
+ "eval_explained_variance": 0.43139490485191345,
460
+ "eval_kl_divergence": 0.42188531160354614,
461
+ "eval_loss": 0.4706146717071533,
462
+ "eval_mae": 0.12165608257055283,
463
+ "eval_rmse": 0.1630544662475586,
464
+ "eval_runtime": 69.0079,
465
+ "eval_samples_per_second": 34.112,
466
+ "eval_steps_per_second": 2.145,
467
+ "learning_rate": 0.0001,
468
+ "step": 10512
469
+ },
470
+ {
471
+ "epoch": 25.0,
472
+ "eval_explained_variance": 0.4370768964290619,
473
+ "eval_kl_divergence": 0.5242214798927307,
474
+ "eval_loss": 0.46876564621925354,
475
+ "eval_mae": 0.11949952691793442,
476
+ "eval_rmse": 0.16123180091381073,
477
+ "eval_runtime": 67.8342,
478
+ "eval_samples_per_second": 34.702,
479
+ "eval_steps_per_second": 2.182,
480
+ "learning_rate": 0.0001,
481
+ "step": 10950
482
+ },
483
+ {
484
+ "epoch": 25.114155251141554,
485
+ "grad_norm": 0.12071619927883148,
486
+ "learning_rate": 0.0001,
487
+ "loss": 0.4665,
488
+ "step": 11000
489
+ },
490
+ {
491
+ "epoch": 26.0,
492
+ "eval_explained_variance": 0.43381467461586,
493
+ "eval_kl_divergence": 0.6159161925315857,
494
+ "eval_loss": 0.46925392746925354,
495
+ "eval_mae": 0.11895165592432022,
496
+ "eval_rmse": 0.16197769343852997,
497
+ "eval_runtime": 66.3554,
498
+ "eval_samples_per_second": 35.476,
499
+ "eval_steps_per_second": 2.23,
500
+ "learning_rate": 0.0001,
501
+ "step": 11388
502
+ },
503
+ {
504
+ "epoch": 26.255707762557076,
505
+ "grad_norm": 0.13487227261066437,
506
+ "learning_rate": 0.0001,
507
+ "loss": 0.4638,
508
+ "step": 11500
509
+ },
510
+ {
511
+ "epoch": 27.0,
512
+ "eval_explained_variance": 0.4415856897830963,
513
+ "eval_kl_divergence": 0.4046337604522705,
514
+ "eval_loss": 0.46849024295806885,
515
+ "eval_mae": 0.12057051062583923,
516
+ "eval_rmse": 0.16070334613323212,
517
+ "eval_runtime": 67.413,
518
+ "eval_samples_per_second": 34.919,
519
+ "eval_steps_per_second": 2.195,
520
+ "learning_rate": 0.0001,
521
+ "step": 11826
522
+ },
523
+ {
524
+ "epoch": 27.397260273972602,
525
+ "grad_norm": 0.122219979763031,
526
+ "learning_rate": 0.0001,
527
+ "loss": 0.4647,
528
+ "step": 12000
529
+ },
530
+ {
531
+ "epoch": 28.0,
532
+ "eval_explained_variance": 0.4443088471889496,
533
+ "eval_kl_divergence": 0.28601282835006714,
534
+ "eval_loss": 0.46939656138420105,
535
+ "eval_mae": 0.12197214365005493,
536
+ "eval_rmse": 0.16160771250724792,
537
+ "eval_runtime": 68.24,
538
+ "eval_samples_per_second": 34.496,
539
+ "eval_steps_per_second": 2.169,
540
+ "learning_rate": 0.0001,
541
+ "step": 12264
542
+ },
543
+ {
544
+ "epoch": 28.538812785388128,
545
+ "grad_norm": 0.1166076809167862,
546
+ "learning_rate": 0.0001,
547
+ "loss": 0.4644,
548
+ "step": 12500
549
+ },
550
+ {
551
+ "epoch": 29.0,
552
+ "eval_explained_variance": 0.44008609652519226,
553
+ "eval_kl_divergence": 0.42698290944099426,
554
+ "eval_loss": 0.46892231702804565,
555
+ "eval_mae": 0.11971781402826309,
556
+ "eval_rmse": 0.16136477887630463,
557
+ "eval_runtime": 67.0859,
558
+ "eval_samples_per_second": 35.089,
559
+ "eval_steps_per_second": 2.206,
560
+ "learning_rate": 0.0001,
561
+ "step": 12702
562
+ },
563
+ {
564
+ "epoch": 29.680365296803654,
565
+ "grad_norm": 0.1203789934515953,
566
+ "learning_rate": 0.0001,
567
+ "loss": 0.4638,
568
+ "step": 13000
569
+ },
570
+ {
571
+ "epoch": 30.0,
572
+ "eval_explained_variance": 0.44359269738197327,
573
+ "eval_kl_divergence": 0.2625378370285034,
574
+ "eval_loss": 0.46987923979759216,
575
+ "eval_mae": 0.12246056646108627,
576
+ "eval_rmse": 0.16186781227588654,
577
+ "eval_runtime": 67.1358,
578
+ "eval_samples_per_second": 35.063,
579
+ "eval_steps_per_second": 2.204,
580
+ "learning_rate": 0.0001,
581
+ "step": 13140
582
+ },
583
+ {
584
+ "epoch": 30.82191780821918,
585
+ "grad_norm": 0.11070472747087479,
586
+ "learning_rate": 0.0001,
587
+ "loss": 0.4636,
588
+ "step": 13500
589
+ },
590
+ {
591
+ "epoch": 31.0,
592
+ "eval_explained_variance": 0.44310399889945984,
593
+ "eval_kl_divergence": 0.38758543133735657,
594
+ "eval_loss": 0.46842578053474426,
595
+ "eval_mae": 0.11974402517080307,
596
+ "eval_rmse": 0.16068722307682037,
597
+ "eval_runtime": 68.317,
598
+ "eval_samples_per_second": 34.457,
599
+ "eval_steps_per_second": 2.166,
600
+ "learning_rate": 0.0001,
601
+ "step": 13578
602
+ },
603
+ {
604
+ "epoch": 31.963470319634702,
605
+ "grad_norm": 0.18743829429149628,
606
+ "learning_rate": 0.0001,
607
+ "loss": 0.463,
608
+ "step": 14000
609
+ },
610
+ {
611
+ "epoch": 32.0,
612
+ "eval_explained_variance": 0.4466596841812134,
613
+ "eval_kl_divergence": 0.4060189425945282,
614
+ "eval_loss": 0.46784707903862,
615
+ "eval_mae": 0.11951278895139694,
616
+ "eval_rmse": 0.16003534197807312,
617
+ "eval_runtime": 66.0349,
618
+ "eval_samples_per_second": 35.648,
619
+ "eval_steps_per_second": 2.241,
620
+ "learning_rate": 0.0001,
621
+ "step": 14016
622
+ },
623
+ {
624
+ "epoch": 33.0,
625
+ "eval_explained_variance": 0.44936656951904297,
626
+ "eval_kl_divergence": 0.3688030540943146,
627
+ "eval_loss": 0.46755874156951904,
628
+ "eval_mae": 0.11934500932693481,
629
+ "eval_rmse": 0.15963692963123322,
630
+ "eval_runtime": 72.5096,
631
+ "eval_samples_per_second": 32.465,
632
+ "eval_steps_per_second": 2.041,
633
+ "learning_rate": 0.0001,
634
+ "step": 14454
635
+ },
636
+ {
637
+ "epoch": 33.10502283105023,
638
+ "grad_norm": 0.15971647202968597,
639
+ "learning_rate": 0.0001,
640
+ "loss": 0.4628,
641
+ "step": 14500
642
+ },
643
+ {
644
+ "epoch": 34.0,
645
+ "eval_explained_variance": 0.4491260051727295,
646
+ "eval_kl_divergence": 0.3900001645088196,
647
+ "eval_loss": 0.46766504645347595,
648
+ "eval_mae": 0.11943826079368591,
649
+ "eval_rmse": 0.15995797514915466,
650
+ "eval_runtime": 67.335,
651
+ "eval_samples_per_second": 34.96,
652
+ "eval_steps_per_second": 2.198,
653
+ "learning_rate": 0.0001,
654
+ "step": 14892
655
+ },
656
+ {
657
+ "epoch": 34.24657534246575,
658
+ "grad_norm": 0.15495429933071136,
659
+ "learning_rate": 0.0001,
660
+ "loss": 0.4616,
661
+ "step": 15000
662
+ },
663
+ {
664
+ "epoch": 35.0,
665
+ "eval_explained_variance": 0.44995394349098206,
666
+ "eval_kl_divergence": 0.42819926142692566,
667
+ "eval_loss": 0.4670174717903137,
668
+ "eval_mae": 0.11885016411542892,
669
+ "eval_rmse": 0.1593056619167328,
670
+ "eval_runtime": 69.6729,
671
+ "eval_samples_per_second": 33.786,
672
+ "eval_steps_per_second": 2.124,
673
+ "learning_rate": 0.0001,
674
+ "step": 15330
675
+ },
676
+ {
677
+ "epoch": 35.38812785388128,
678
+ "grad_norm": 0.1067986786365509,
679
+ "learning_rate": 0.0001,
680
+ "loss": 0.4634,
681
+ "step": 15500
682
+ },
683
+ {
684
+ "epoch": 36.0,
685
+ "eval_explained_variance": 0.45064660906791687,
686
+ "eval_kl_divergence": 0.44459816813468933,
687
+ "eval_loss": 0.46679624915122986,
688
+ "eval_mae": 0.11801303178071976,
689
+ "eval_rmse": 0.15909221768379211,
690
+ "eval_runtime": 70.3703,
691
+ "eval_samples_per_second": 33.452,
692
+ "eval_steps_per_second": 2.103,
693
+ "learning_rate": 0.0001,
694
+ "step": 15768
695
+ },
696
+ {
697
+ "epoch": 36.529680365296805,
698
+ "grad_norm": 0.17229855060577393,
699
+ "learning_rate": 0.0001,
700
+ "loss": 0.462,
701
+ "step": 16000
702
+ },
703
+ {
704
+ "epoch": 37.0,
705
+ "eval_explained_variance": 0.4528321325778961,
706
+ "eval_kl_divergence": 0.39424213767051697,
707
+ "eval_loss": 0.46689239144325256,
708
+ "eval_mae": 0.11852964758872986,
709
+ "eval_rmse": 0.15902292728424072,
710
+ "eval_runtime": 68.7832,
711
+ "eval_samples_per_second": 34.223,
712
+ "eval_steps_per_second": 2.152,
713
+ "learning_rate": 0.0001,
714
+ "step": 16206
715
+ },
716
+ {
717
+ "epoch": 37.67123287671233,
718
+ "grad_norm": 0.16489605605602264,
719
+ "learning_rate": 0.0001,
720
+ "loss": 0.4631,
721
+ "step": 16500
722
+ },
723
+ {
724
+ "epoch": 38.0,
725
+ "eval_explained_variance": 0.4512017071247101,
726
+ "eval_kl_divergence": 0.47831401228904724,
727
+ "eval_loss": 0.4664570987224579,
728
+ "eval_mae": 0.11769810318946838,
729
+ "eval_rmse": 0.15879899263381958,
730
+ "eval_runtime": 69.0328,
731
+ "eval_samples_per_second": 34.1,
732
+ "eval_steps_per_second": 2.144,
733
+ "learning_rate": 0.0001,
734
+ "step": 16644
735
+ },
736
+ {
737
+ "epoch": 38.81278538812786,
738
+ "grad_norm": 0.24003809690475464,
739
+ "learning_rate": 0.0001,
740
+ "loss": 0.4603,
741
+ "step": 17000
742
+ },
743
+ {
744
+ "epoch": 39.0,
745
+ "eval_explained_variance": 0.4500379264354706,
746
+ "eval_kl_divergence": 0.386750727891922,
747
+ "eval_loss": 0.4674011468887329,
748
+ "eval_mae": 0.11898898333311081,
749
+ "eval_rmse": 0.15971434116363525,
750
+ "eval_runtime": 69.047,
751
+ "eval_samples_per_second": 34.093,
752
+ "eval_steps_per_second": 2.143,
753
+ "learning_rate": 0.0001,
754
+ "step": 17082
755
+ },
756
+ {
757
+ "epoch": 39.954337899543376,
758
+ "grad_norm": 0.1894679218530655,
759
+ "learning_rate": 0.0001,
760
+ "loss": 0.4614,
761
+ "step": 17500
762
+ },
763
+ {
764
+ "epoch": 40.0,
765
+ "eval_explained_variance": 0.4498337507247925,
766
+ "eval_kl_divergence": 0.3627259135246277,
767
+ "eval_loss": 0.4677062928676605,
768
+ "eval_mae": 0.11947864294052124,
769
+ "eval_rmse": 0.15986855328083038,
770
+ "eval_runtime": 68.3668,
771
+ "eval_samples_per_second": 34.432,
772
+ "eval_steps_per_second": 2.165,
773
+ "learning_rate": 0.0001,
774
+ "step": 17520
775
+ },
776
+ {
777
+ "epoch": 41.0,
778
+ "eval_explained_variance": 0.4539731740951538,
779
+ "eval_kl_divergence": 0.2654862701892853,
780
+ "eval_loss": 0.46822381019592285,
781
+ "eval_mae": 0.12105310708284378,
782
+ "eval_rmse": 0.16022303700447083,
783
+ "eval_runtime": 68.0126,
784
+ "eval_samples_per_second": 34.611,
785
+ "eval_steps_per_second": 2.176,
786
+ "learning_rate": 0.0001,
787
+ "step": 17958
788
+ },
789
+ {
790
+ "epoch": 41.0958904109589,
791
+ "grad_norm": 0.13239748775959015,
792
+ "learning_rate": 0.0001,
793
+ "loss": 0.4612,
794
+ "step": 18000
795
+ },
796
+ {
797
+ "epoch": 42.0,
798
+ "eval_explained_variance": 0.45140552520751953,
799
+ "eval_kl_divergence": 0.5071607828140259,
800
+ "eval_loss": 0.4664672613143921,
801
+ "eval_mae": 0.11723508685827255,
802
+ "eval_rmse": 0.15891815721988678,
803
+ "eval_runtime": 66.3853,
804
+ "eval_samples_per_second": 35.46,
805
+ "eval_steps_per_second": 2.229,
806
+ "learning_rate": 0.0001,
807
+ "step": 18396
808
+ },
809
+ {
810
+ "epoch": 42.23744292237443,
811
+ "grad_norm": 0.17189915478229523,
812
+ "learning_rate": 0.0001,
813
+ "loss": 0.462,
814
+ "step": 18500
815
+ },
816
+ {
817
+ "epoch": 43.0,
818
+ "eval_explained_variance": 0.4554661214351654,
819
+ "eval_kl_divergence": 0.430560827255249,
820
+ "eval_loss": 0.46638762950897217,
821
+ "eval_mae": 0.11766376346349716,
822
+ "eval_rmse": 0.15850575268268585,
823
+ "eval_runtime": 65.8594,
824
+ "eval_samples_per_second": 35.743,
825
+ "eval_steps_per_second": 2.247,
826
+ "learning_rate": 0.0001,
827
+ "step": 18834
828
+ },
829
+ {
830
+ "epoch": 43.37899543378995,
831
+ "grad_norm": 0.1566254049539566,
832
+ "learning_rate": 0.0001,
833
+ "loss": 0.4603,
834
+ "step": 19000
835
+ },
836
+ {
837
+ "epoch": 44.0,
838
+ "eval_explained_variance": 0.45294663310050964,
839
+ "eval_kl_divergence": 0.4114503562450409,
840
+ "eval_loss": 0.46708741784095764,
841
+ "eval_mae": 0.11923228949308395,
842
+ "eval_rmse": 0.15942519903182983,
843
+ "eval_runtime": 70.2132,
844
+ "eval_samples_per_second": 33.526,
845
+ "eval_steps_per_second": 2.108,
846
+ "learning_rate": 0.0001,
847
+ "step": 19272
848
+ },
849
+ {
850
+ "epoch": 44.52054794520548,
851
+ "grad_norm": 0.1487286388874054,
852
+ "learning_rate": 0.0001,
853
+ "loss": 0.4599,
854
+ "step": 19500
855
+ },
856
+ {
857
+ "epoch": 45.0,
858
+ "eval_explained_variance": 0.45276129245758057,
859
+ "eval_kl_divergence": 0.441719114780426,
860
+ "eval_loss": 0.46663177013397217,
861
+ "eval_mae": 0.11713916063308716,
862
+ "eval_rmse": 0.15901675820350647,
863
+ "eval_runtime": 67.4478,
864
+ "eval_samples_per_second": 34.901,
865
+ "eval_steps_per_second": 2.194,
866
+ "learning_rate": 0.0001,
867
+ "step": 19710
868
+ },
869
+ {
870
+ "epoch": 45.662100456621005,
871
+ "grad_norm": 0.17072182893753052,
872
+ "learning_rate": 0.0001,
873
+ "loss": 0.4612,
874
+ "step": 20000
875
+ },
876
+ {
877
+ "epoch": 46.0,
878
+ "eval_explained_variance": 0.4574218690395355,
879
+ "eval_kl_divergence": 0.3686365485191345,
880
+ "eval_loss": 0.4663327634334564,
881
+ "eval_mae": 0.11786039918661118,
882
+ "eval_rmse": 0.15854045748710632,
883
+ "eval_runtime": 66.0464,
884
+ "eval_samples_per_second": 35.642,
885
+ "eval_steps_per_second": 2.241,
886
+ "learning_rate": 0.0001,
887
+ "step": 20148
888
+ },
889
+ {
890
+ "epoch": 46.80365296803653,
891
+ "grad_norm": 0.17141404747962952,
892
+ "learning_rate": 0.0001,
893
+ "loss": 0.4596,
894
+ "step": 20500
895
+ },
896
+ {
897
+ "epoch": 47.0,
898
+ "eval_explained_variance": 0.4567171633243561,
899
+ "eval_kl_divergence": 0.5089961886405945,
900
+ "eval_loss": 0.46577510237693787,
901
+ "eval_mae": 0.11723583936691284,
902
+ "eval_rmse": 0.15815743803977966,
903
+ "eval_runtime": 68.7513,
904
+ "eval_samples_per_second": 34.239,
905
+ "eval_steps_per_second": 2.153,
906
+ "learning_rate": 0.0001,
907
+ "step": 20586
908
+ },
909
+ {
910
+ "epoch": 47.945205479452056,
911
+ "grad_norm": 0.13870340585708618,
912
+ "learning_rate": 0.0001,
913
+ "loss": 0.4603,
914
+ "step": 21000
915
+ },
916
+ {
917
+ "epoch": 48.0,
918
+ "eval_explained_variance": 0.4547780752182007,
919
+ "eval_kl_divergence": 0.5279051065444946,
920
+ "eval_loss": 0.46634000539779663,
921
+ "eval_mae": 0.11751776933670044,
922
+ "eval_rmse": 0.15890929102897644,
923
+ "eval_runtime": 66.5777,
924
+ "eval_samples_per_second": 35.357,
925
+ "eval_steps_per_second": 2.223,
926
+ "learning_rate": 0.0001,
927
+ "step": 21024
928
+ },
929
+ {
930
+ "epoch": 49.0,
931
+ "eval_explained_variance": 0.4532240927219391,
932
+ "eval_kl_divergence": 0.44965481758117676,
933
+ "eval_loss": 0.46656596660614014,
934
+ "eval_mae": 0.11831226199865341,
935
+ "eval_rmse": 0.15911053121089935,
936
+ "eval_runtime": 67.2369,
937
+ "eval_samples_per_second": 35.011,
938
+ "eval_steps_per_second": 2.201,
939
+ "learning_rate": 0.0001,
940
+ "step": 21462
941
+ },
942
+ {
943
+ "epoch": 49.08675799086758,
944
+ "grad_norm": 0.16372357308864594,
945
+ "learning_rate": 0.0001,
946
+ "loss": 0.4599,
947
+ "step": 21500
948
+ },
949
+ {
950
+ "epoch": 50.0,
951
+ "eval_explained_variance": 0.45799562335014343,
952
+ "eval_kl_divergence": 0.27116483449935913,
953
+ "eval_loss": 0.46755433082580566,
954
+ "eval_mae": 0.1204712763428688,
955
+ "eval_rmse": 0.15946535766124725,
956
+ "eval_runtime": 67.5446,
957
+ "eval_samples_per_second": 34.851,
958
+ "eval_steps_per_second": 2.191,
959
+ "learning_rate": 0.0001,
960
+ "step": 21900
961
+ },
962
+ {
963
+ "epoch": 50.22831050228311,
964
+ "grad_norm": 0.15486378967761993,
965
+ "learning_rate": 0.0001,
966
+ "loss": 0.4594,
967
+ "step": 22000
968
+ },
969
+ {
970
+ "epoch": 51.0,
971
+ "eval_explained_variance": 0.4551831781864166,
972
+ "eval_kl_divergence": 0.4008127450942993,
973
+ "eval_loss": 0.46639156341552734,
974
+ "eval_mae": 0.11721646040678024,
975
+ "eval_rmse": 0.15861637890338898,
976
+ "eval_runtime": 64.0368,
977
+ "eval_samples_per_second": 36.76,
978
+ "eval_steps_per_second": 2.311,
979
+ "learning_rate": 0.0001,
980
+ "step": 22338
981
+ },
982
+ {
983
+ "epoch": 51.36986301369863,
984
+ "grad_norm": 0.2131696492433548,
985
+ "learning_rate": 0.0001,
986
+ "loss": 0.4593,
987
+ "step": 22500
988
+ },
989
+ {
990
+ "epoch": 52.0,
991
+ "eval_explained_variance": 0.4556889832019806,
992
+ "eval_kl_divergence": 0.4921821057796478,
993
+ "eval_loss": 0.46591076254844666,
994
+ "eval_mae": 0.11625734716653824,
995
+ "eval_rmse": 0.1582706868648529,
996
+ "eval_runtime": 68.9844,
997
+ "eval_samples_per_second": 34.124,
998
+ "eval_steps_per_second": 2.145,
999
+ "learning_rate": 0.0001,
1000
+ "step": 22776
1001
+ },
1002
+ {
1003
+ "epoch": 52.51141552511415,
1004
+ "grad_norm": 0.3151676654815674,
1005
+ "learning_rate": 0.0001,
1006
+ "loss": 0.4614,
1007
+ "step": 23000
1008
+ },
1009
+ {
1010
+ "epoch": 53.0,
1011
+ "eval_explained_variance": 0.4601232707500458,
1012
+ "eval_kl_divergence": 0.4273536205291748,
1013
+ "eval_loss": 0.4656851887702942,
1014
+ "eval_mae": 0.1177864745259285,
1015
+ "eval_rmse": 0.15793031454086304,
1016
+ "eval_runtime": 68.0941,
1017
+ "eval_samples_per_second": 34.57,
1018
+ "eval_steps_per_second": 2.173,
1019
+ "learning_rate": 0.0001,
1020
+ "step": 23214
1021
+ },
1022
+ {
1023
+ "epoch": 53.65296803652968,
1024
+ "grad_norm": 0.1899169534444809,
1025
+ "learning_rate": 0.0001,
1026
+ "loss": 0.4592,
1027
+ "step": 23500
1028
+ },
1029
+ {
1030
+ "epoch": 54.0,
1031
+ "eval_explained_variance": 0.456957072019577,
1032
+ "eval_kl_divergence": 0.4574394226074219,
1033
+ "eval_loss": 0.46629655361175537,
1034
+ "eval_mae": 0.11577478051185608,
1035
+ "eval_rmse": 0.15850554406642914,
1036
+ "eval_runtime": 65.2277,
1037
+ "eval_samples_per_second": 36.089,
1038
+ "eval_steps_per_second": 2.269,
1039
+ "learning_rate": 0.0001,
1040
+ "step": 23652
1041
+ },
1042
+ {
1043
+ "epoch": 54.794520547945204,
1044
+ "grad_norm": 0.27214938402175903,
1045
+ "learning_rate": 0.0001,
1046
+ "loss": 0.4601,
1047
+ "step": 24000
1048
+ },
1049
+ {
1050
+ "epoch": 55.0,
1051
+ "eval_explained_variance": 0.45804765820503235,
1052
+ "eval_kl_divergence": 0.34863969683647156,
1053
+ "eval_loss": 0.46644341945648193,
1054
+ "eval_mae": 0.11888903379440308,
1055
+ "eval_rmse": 0.1585645228624344,
1056
+ "eval_runtime": 62.989,
1057
+ "eval_samples_per_second": 37.372,
1058
+ "eval_steps_per_second": 2.35,
1059
+ "learning_rate": 0.0001,
1060
+ "step": 24090
1061
+ },
1062
+ {
1063
+ "epoch": 55.93607305936073,
1064
+ "grad_norm": 0.32561782002449036,
1065
+ "learning_rate": 0.0001,
1066
+ "loss": 0.4589,
1067
+ "step": 24500
1068
+ },
1069
+ {
1070
+ "epoch": 56.0,
1071
+ "eval_explained_variance": 0.46123751997947693,
1072
+ "eval_kl_divergence": 0.3015596568584442,
1073
+ "eval_loss": 0.4661739766597748,
1074
+ "eval_mae": 0.1184321939945221,
1075
+ "eval_rmse": 0.15838663280010223,
1076
+ "eval_runtime": 62.9298,
1077
+ "eval_samples_per_second": 37.407,
1078
+ "eval_steps_per_second": 2.352,
1079
+ "learning_rate": 0.0001,
1080
+ "step": 24528
1081
+ },
1082
+ {
1083
+ "epoch": 57.0,
1084
+ "eval_explained_variance": 0.45534956455230713,
1085
+ "eval_kl_divergence": 0.4162982106208801,
1086
+ "eval_loss": 0.46634721755981445,
1087
+ "eval_mae": 0.11808174103498459,
1088
+ "eval_rmse": 0.15866883099079132,
1089
+ "eval_runtime": 63.3411,
1090
+ "eval_samples_per_second": 37.164,
1091
+ "eval_steps_per_second": 2.337,
1092
+ "learning_rate": 0.0001,
1093
+ "step": 24966
1094
+ },
1095
+ {
1096
+ "epoch": 57.077625570776256,
1097
+ "grad_norm": 0.21059946715831757,
1098
+ "learning_rate": 0.0001,
1099
+ "loss": 0.4588,
1100
+ "step": 25000
1101
+ },
1102
+ {
1103
+ "epoch": 58.0,
1104
+ "eval_explained_variance": 0.4557063579559326,
1105
+ "eval_kl_divergence": 0.3399294316768646,
1106
+ "eval_loss": 0.4673805236816406,
1107
+ "eval_mae": 0.11887022852897644,
1108
+ "eval_rmse": 0.15934236347675323,
1109
+ "eval_runtime": 65.9311,
1110
+ "eval_samples_per_second": 35.704,
1111
+ "eval_steps_per_second": 2.245,
1112
+ "learning_rate": 0.0001,
1113
+ "step": 25404
1114
+ },
1115
+ {
1116
+ "epoch": 58.21917808219178,
1117
+ "grad_norm": 0.1977730244398117,
1118
+ "learning_rate": 0.0001,
1119
+ "loss": 0.4595,
1120
+ "step": 25500
1121
+ },
1122
+ {
1123
+ "epoch": 59.0,
1124
+ "eval_explained_variance": 0.4646029472351074,
1125
+ "eval_kl_divergence": 0.3685876429080963,
1126
+ "eval_loss": 0.4650005102157593,
1127
+ "eval_mae": 0.11696922034025192,
1128
+ "eval_rmse": 0.15720517933368683,
1129
+ "eval_runtime": 63.6041,
1130
+ "eval_samples_per_second": 37.01,
1131
+ "eval_steps_per_second": 2.327,
1132
+ "learning_rate": 0.0001,
1133
+ "step": 25842
1134
+ },
1135
+ {
1136
+ "epoch": 59.36073059360731,
1137
+ "grad_norm": 0.22561948001384735,
1138
+ "learning_rate": 0.0001,
1139
+ "loss": 0.4594,
1140
+ "step": 26000
1141
+ },
1142
+ {
1143
+ "epoch": 60.0,
1144
+ "eval_explained_variance": 0.4566896855831146,
1145
+ "eval_kl_divergence": 0.45347684621810913,
1146
+ "eval_loss": 0.46599113941192627,
1147
+ "eval_mae": 0.11718535423278809,
1148
+ "eval_rmse": 0.15840259194374084,
1149
+ "eval_runtime": 62.901,
1150
+ "eval_samples_per_second": 37.424,
1151
+ "eval_steps_per_second": 2.353,
1152
+ "learning_rate": 0.0001,
1153
+ "step": 26280
1154
+ },
1155
+ {
1156
+ "epoch": 60.50228310502283,
1157
+ "grad_norm": 0.19251562654972076,
1158
+ "learning_rate": 0.0001,
1159
+ "loss": 0.4599,
1160
+ "step": 26500
1161
+ },
1162
+ {
1163
+ "epoch": 61.0,
1164
+ "eval_explained_variance": 0.4583926200866699,
1165
+ "eval_kl_divergence": 0.375108540058136,
1166
+ "eval_loss": 0.4662201702594757,
1167
+ "eval_mae": 0.11791958659887314,
1168
+ "eval_rmse": 0.1585090458393097,
1169
+ "eval_runtime": 61.9039,
1170
+ "eval_samples_per_second": 38.027,
1171
+ "eval_steps_per_second": 2.391,
1172
+ "learning_rate": 0.0001,
1173
+ "step": 26718
1174
+ },
1175
+ {
1176
+ "epoch": 61.64383561643836,
1177
+ "grad_norm": 0.2111298143863678,
1178
+ "learning_rate": 0.0001,
1179
+ "loss": 0.4584,
1180
+ "step": 27000
1181
+ },
1182
+ {
1183
+ "epoch": 62.0,
1184
+ "eval_explained_variance": 0.4587836265563965,
1185
+ "eval_kl_divergence": 0.35335177183151245,
1186
+ "eval_loss": 0.46614503860473633,
1187
+ "eval_mae": 0.11732319742441177,
1188
+ "eval_rmse": 0.15833334624767303,
1189
+ "eval_runtime": 63.6491,
1190
+ "eval_samples_per_second": 36.984,
1191
+ "eval_steps_per_second": 2.325,
1192
+ "learning_rate": 0.0001,
1193
+ "step": 27156
1194
+ },
1195
+ {
1196
+ "epoch": 62.78538812785388,
1197
+ "grad_norm": 0.21951091289520264,
1198
+ "learning_rate": 0.0001,
1199
+ "loss": 0.4575,
1200
+ "step": 27500
1201
+ },
1202
+ {
1203
+ "epoch": 63.0,
1204
+ "eval_explained_variance": 0.4576869010925293,
1205
+ "eval_kl_divergence": 0.40475067496299744,
1206
+ "eval_loss": 0.4660026431083679,
1207
+ "eval_mae": 0.11627298593521118,
1208
+ "eval_rmse": 0.1582755148410797,
1209
+ "eval_runtime": 62.5657,
1210
+ "eval_samples_per_second": 37.624,
1211
+ "eval_steps_per_second": 2.366,
1212
+ "learning_rate": 0.0001,
1213
+ "step": 27594
1214
+ },
1215
+ {
1216
+ "epoch": 63.926940639269404,
1217
+ "grad_norm": 0.21228961646556854,
1218
+ "learning_rate": 0.0001,
1219
+ "loss": 0.4598,
1220
+ "step": 28000
1221
+ },
1222
+ {
1223
+ "epoch": 64.0,
1224
+ "eval_explained_variance": 0.46285080909729004,
1225
+ "eval_kl_divergence": 0.2471023052930832,
1226
+ "eval_loss": 0.46711620688438416,
1227
+ "eval_mae": 0.11878780275583267,
1228
+ "eval_rmse": 0.15883709490299225,
1229
+ "eval_runtime": 62.9069,
1230
+ "eval_samples_per_second": 37.42,
1231
+ "eval_steps_per_second": 2.353,
1232
+ "learning_rate": 0.0001,
1233
+ "step": 28032
1234
+ },
1235
+ {
1236
+ "epoch": 65.0,
1237
+ "eval_explained_variance": 0.46037113666534424,
1238
+ "eval_kl_divergence": 0.4526395797729492,
1239
+ "eval_loss": 0.46536803245544434,
1240
+ "eval_mae": 0.11657585948705673,
1241
+ "eval_rmse": 0.15768831968307495,
1242
+ "eval_runtime": 65.6932,
1243
+ "eval_samples_per_second": 35.833,
1244
+ "eval_steps_per_second": 2.253,
1245
+ "learning_rate": 0.0001,
1246
+ "step": 28470
1247
+ },
1248
+ {
1249
+ "epoch": 65.06849315068493,
1250
+ "grad_norm": 0.23406238853931427,
1251
+ "learning_rate": 1e-05,
1252
+ "loss": 0.4582,
1253
+ "step": 28500
1254
+ },
1255
+ {
1256
+ "epoch": 66.0,
1257
+ "eval_explained_variance": 0.45918479561805725,
1258
+ "eval_kl_divergence": 0.5259271264076233,
1259
+ "eval_loss": 0.46573594212532043,
1260
+ "eval_mae": 0.11610028892755508,
1261
+ "eval_rmse": 0.15816400945186615,
1262
+ "eval_runtime": 63.8294,
1263
+ "eval_samples_per_second": 36.88,
1264
+ "eval_steps_per_second": 2.319,
1265
+ "learning_rate": 1e-05,
1266
+ "step": 28908
1267
+ },
1268
+ {
1269
+ "epoch": 66.21004566210046,
1270
+ "grad_norm": 0.20290181040763855,
1271
+ "learning_rate": 1e-05,
1272
+ "loss": 0.4592,
1273
+ "step": 29000
1274
+ },
1275
+ {
1276
+ "epoch": 67.0,
1277
+ "eval_explained_variance": 0.4623439311981201,
1278
+ "eval_kl_divergence": 0.4251798987388611,
1279
+ "eval_loss": 0.4653942584991455,
1280
+ "eval_mae": 0.11725304275751114,
1281
+ "eval_rmse": 0.1573910117149353,
1282
+ "eval_runtime": 65.929,
1283
+ "eval_samples_per_second": 35.705,
1284
+ "eval_steps_per_second": 2.245,
1285
+ "learning_rate": 1e-05,
1286
+ "step": 29346
1287
+ },
1288
+ {
1289
+ "epoch": 67.35159817351598,
1290
+ "grad_norm": 0.210421621799469,
1291
+ "learning_rate": 1e-05,
1292
+ "loss": 0.4573,
1293
+ "step": 29500
1294
+ },
1295
+ {
1296
+ "epoch": 68.0,
1297
+ "eval_explained_variance": 0.46140334010124207,
1298
+ "eval_kl_divergence": 0.49885210394859314,
1299
+ "eval_loss": 0.46487176418304443,
1300
+ "eval_mae": 0.11542203277349472,
1301
+ "eval_rmse": 0.15722759068012238,
1302
+ "eval_runtime": 63.8681,
1303
+ "eval_samples_per_second": 36.857,
1304
+ "eval_steps_per_second": 2.317,
1305
+ "learning_rate": 1e-05,
1306
+ "step": 29784
1307
+ },
1308
+ {
1309
+ "epoch": 68.4931506849315,
1310
+ "grad_norm": 0.3686019480228424,
1311
+ "learning_rate": 1e-05,
1312
+ "loss": 0.4576,
1313
+ "step": 30000
1314
+ },
1315
+ {
1316
+ "epoch": 69.0,
1317
+ "eval_explained_variance": 0.46442678570747375,
1318
+ "eval_kl_divergence": 0.40228068828582764,
1319
+ "eval_loss": 0.465110719203949,
1320
+ "eval_mae": 0.11607488244771957,
1321
+ "eval_rmse": 0.15701350569725037,
1322
+ "eval_runtime": 68.8094,
1323
+ "eval_samples_per_second": 34.21,
1324
+ "eval_steps_per_second": 2.151,
1325
+ "learning_rate": 1e-05,
1326
+ "step": 30222
1327
+ },
1328
+ {
1329
+ "epoch": 69.63470319634703,
1330
+ "grad_norm": 0.3072531223297119,
1331
+ "learning_rate": 1e-05,
1332
+ "loss": 0.4556,
1333
+ "step": 30500
1334
+ },
1335
+ {
1336
+ "epoch": 70.0,
1337
+ "eval_explained_variance": 0.4622003734111786,
1338
+ "eval_kl_divergence": 0.41175633668899536,
1339
+ "eval_loss": 0.466043084859848,
1340
+ "eval_mae": 0.1165715679526329,
1341
+ "eval_rmse": 0.15760619938373566,
1342
+ "eval_runtime": 67.6756,
1343
+ "eval_samples_per_second": 34.784,
1344
+ "eval_steps_per_second": 2.187,
1345
+ "learning_rate": 1e-05,
1346
+ "step": 30660
1347
+ },
1348
+ {
1349
+ "epoch": 70.77625570776256,
1350
+ "grad_norm": 0.2429497092962265,
1351
+ "learning_rate": 1e-05,
1352
+ "loss": 0.4591,
1353
+ "step": 31000
1354
+ },
1355
+ {
1356
+ "epoch": 71.0,
1357
+ "eval_explained_variance": 0.4643818140029907,
1358
+ "eval_kl_divergence": 0.3074641227722168,
1359
+ "eval_loss": 0.46608996391296387,
1360
+ "eval_mae": 0.11772569268941879,
1361
+ "eval_rmse": 0.15784674882888794,
1362
+ "eval_runtime": 72.04,
1363
+ "eval_samples_per_second": 32.676,
1364
+ "eval_steps_per_second": 2.054,
1365
+ "learning_rate": 1e-05,
1366
+ "step": 31098
1367
+ },
1368
+ {
1369
+ "epoch": 71.91780821917808,
1370
+ "grad_norm": 0.1753670871257782,
1371
+ "learning_rate": 1e-05,
1372
+ "loss": 0.4563,
1373
+ "step": 31500
1374
+ },
1375
+ {
1376
+ "epoch": 72.0,
1377
+ "eval_explained_variance": 0.46214058995246887,
1378
+ "eval_kl_divergence": 0.38358867168426514,
1379
+ "eval_loss": 0.46582332253456116,
1380
+ "eval_mae": 0.11707664281129837,
1381
+ "eval_rmse": 0.1579793244600296,
1382
+ "eval_runtime": 66.3861,
1383
+ "eval_samples_per_second": 35.459,
1384
+ "eval_steps_per_second": 2.229,
1385
+ "learning_rate": 1e-05,
1386
+ "step": 31536
1387
+ },
1388
+ {
1389
+ "epoch": 73.0,
1390
+ "eval_explained_variance": 0.464040070772171,
1391
+ "eval_kl_divergence": 0.45435720682144165,
1392
+ "eval_loss": 0.4648771584033966,
1393
+ "eval_mae": 0.11544410139322281,
1394
+ "eval_rmse": 0.15694700181484222,
1395
+ "eval_runtime": 64.6485,
1396
+ "eval_samples_per_second": 36.412,
1397
+ "eval_steps_per_second": 2.289,
1398
+ "learning_rate": 1e-05,
1399
+ "step": 31974
1400
+ },
1401
+ {
1402
+ "epoch": 73.05936073059361,
1403
+ "grad_norm": 0.2539682686328888,
1404
+ "learning_rate": 1e-05,
1405
+ "loss": 0.4577,
1406
+ "step": 32000
1407
+ },
1408
+ {
1409
+ "epoch": 74.0,
1410
+ "eval_explained_variance": 0.46597161889076233,
1411
+ "eval_kl_divergence": 0.4537515938282013,
1412
+ "eval_loss": 0.46466848254203796,
1413
+ "eval_mae": 0.11628029495477676,
1414
+ "eval_rmse": 0.1567227691411972,
1415
+ "eval_runtime": 64.3001,
1416
+ "eval_samples_per_second": 36.61,
1417
+ "eval_steps_per_second": 2.302,
1418
+ "learning_rate": 1e-05,
1419
+ "step": 32412
1420
+ },
1421
+ {
1422
+ "epoch": 74.20091324200914,
1423
+ "grad_norm": 0.19458268582820892,
1424
+ "learning_rate": 1e-05,
1425
+ "loss": 0.4576,
1426
+ "step": 32500
1427
+ },
1428
+ {
1429
+ "epoch": 75.0,
1430
+ "eval_explained_variance": 0.46580255031585693,
1431
+ "eval_kl_divergence": 0.33480265736579895,
1432
+ "eval_loss": 0.46563470363616943,
1433
+ "eval_mae": 0.11664699763059616,
1434
+ "eval_rmse": 0.1572960466146469,
1435
+ "eval_runtime": 65.0705,
1436
+ "eval_samples_per_second": 36.176,
1437
+ "eval_steps_per_second": 2.274,
1438
+ "learning_rate": 1e-05,
1439
+ "step": 32850
1440
+ },
1441
+ {
1442
+ "epoch": 75.34246575342466,
1443
+ "grad_norm": 0.23031042516231537,
1444
+ "learning_rate": 1e-05,
1445
+ "loss": 0.457,
1446
+ "step": 33000
1447
+ },
1448
+ {
1449
+ "epoch": 76.0,
1450
+ "eval_explained_variance": 0.4644790589809418,
1451
+ "eval_kl_divergence": 0.49758121371269226,
1452
+ "eval_loss": 0.4647076725959778,
1453
+ "eval_mae": 0.11578535288572311,
1454
+ "eval_rmse": 0.15706199407577515,
1455
+ "eval_runtime": 64.4752,
1456
+ "eval_samples_per_second": 36.51,
1457
+ "eval_steps_per_second": 2.295,
1458
+ "learning_rate": 1e-05,
1459
+ "step": 33288
1460
+ },
1461
+ {
1462
+ "epoch": 76.48401826484019,
1463
+ "grad_norm": 0.21923455595970154,
1464
+ "learning_rate": 1e-05,
1465
+ "loss": 0.4574,
1466
+ "step": 33500
1467
+ },
1468
+ {
1469
+ "epoch": 77.0,
1470
+ "eval_explained_variance": 0.46530094742774963,
1471
+ "eval_kl_divergence": 0.3934381902217865,
1472
+ "eval_loss": 0.4650570750236511,
1473
+ "eval_mae": 0.11632921546697617,
1474
+ "eval_rmse": 0.15704087913036346,
1475
+ "eval_runtime": 65.0363,
1476
+ "eval_samples_per_second": 36.195,
1477
+ "eval_steps_per_second": 2.276,
1478
+ "learning_rate": 1e-05,
1479
+ "step": 33726
1480
+ },
1481
+ {
1482
+ "epoch": 77.62557077625571,
1483
+ "grad_norm": 0.27394065260887146,
1484
+ "learning_rate": 1e-05,
1485
+ "loss": 0.457,
1486
+ "step": 34000
1487
+ },
1488
+ {
1489
+ "epoch": 78.0,
1490
+ "eval_explained_variance": 0.4654146730899811,
1491
+ "eval_kl_divergence": 0.39355090260505676,
1492
+ "eval_loss": 0.46495845913887024,
1493
+ "eval_mae": 0.11605050414800644,
1494
+ "eval_rmse": 0.15708379447460175,
1495
+ "eval_runtime": 64.2371,
1496
+ "eval_samples_per_second": 36.645,
1497
+ "eval_steps_per_second": 2.304,
1498
+ "learning_rate": 1e-05,
1499
+ "step": 34164
1500
+ },
1501
+ {
1502
+ "epoch": 78.76712328767124,
1503
+ "grad_norm": 0.20854905247688293,
1504
+ "learning_rate": 1e-05,
1505
+ "loss": 0.4566,
1506
+ "step": 34500
1507
+ },
1508
+ {
1509
+ "epoch": 79.0,
1510
+ "eval_explained_variance": 0.4653468132019043,
1511
+ "eval_kl_divergence": 0.37590739130973816,
1512
+ "eval_loss": 0.46530288457870483,
1513
+ "eval_mae": 0.11588699370622635,
1514
+ "eval_rmse": 0.15725918114185333,
1515
+ "eval_runtime": 62.4443,
1516
+ "eval_samples_per_second": 37.698,
1517
+ "eval_steps_per_second": 2.37,
1518
+ "learning_rate": 1e-05,
1519
+ "step": 34602
1520
+ },
1521
+ {
1522
+ "epoch": 79.90867579908675,
1523
+ "grad_norm": 0.21241584420204163,
1524
+ "learning_rate": 1e-05,
1525
+ "loss": 0.458,
1526
+ "step": 35000
1527
+ },
1528
+ {
1529
+ "epoch": 80.0,
1530
+ "eval_explained_variance": 0.46603894233703613,
1531
+ "eval_kl_divergence": 0.41887620091438293,
1532
+ "eval_loss": 0.4647064805030823,
1533
+ "eval_mae": 0.11623198539018631,
1534
+ "eval_rmse": 0.15669378638267517,
1535
+ "eval_runtime": 62.0038,
1536
+ "eval_samples_per_second": 37.965,
1537
+ "eval_steps_per_second": 2.387,
1538
+ "learning_rate": 1e-05,
1539
+ "step": 35040
1540
+ },
1541
+ {
1542
+ "epoch": 81.0,
1543
+ "eval_explained_variance": 0.4646754562854767,
1544
+ "eval_kl_divergence": 0.47510233521461487,
1545
+ "eval_loss": 0.46485888957977295,
1546
+ "eval_mae": 0.11576662212610245,
1547
+ "eval_rmse": 0.15710778534412384,
1548
+ "eval_runtime": 61.3344,
1549
+ "eval_samples_per_second": 38.38,
1550
+ "eval_steps_per_second": 2.413,
1551
+ "learning_rate": 1.0000000000000002e-06,
1552
+ "step": 35478
1553
+ },
1554
+ {
1555
+ "epoch": 81.05022831050228,
1556
+ "grad_norm": 0.23253624141216278,
1557
+ "learning_rate": 1.0000000000000002e-06,
1558
+ "loss": 0.456,
1559
+ "step": 35500
1560
+ },
1561
+ {
1562
+ "epoch": 82.0,
1563
+ "eval_explained_variance": 0.4651080071926117,
1564
+ "eval_kl_divergence": 0.43354716897010803,
1565
+ "eval_loss": 0.4654049277305603,
1566
+ "eval_mae": 0.11612025648355484,
1567
+ "eval_rmse": 0.1572476029396057,
1568
+ "eval_runtime": 62.2235,
1569
+ "eval_samples_per_second": 37.831,
1570
+ "eval_steps_per_second": 2.379,
1571
+ "learning_rate": 1.0000000000000002e-06,
1572
+ "step": 35916
1573
+ },
1574
+ {
1575
+ "epoch": 82.1917808219178,
1576
+ "grad_norm": 0.23126010596752167,
1577
+ "learning_rate": 1.0000000000000002e-06,
1578
+ "loss": 0.4564,
1579
+ "step": 36000
1580
+ },
1581
+ {
1582
+ "epoch": 83.0,
1583
+ "eval_explained_variance": 0.46666058897972107,
1584
+ "eval_kl_divergence": 0.3906138837337494,
1585
+ "eval_loss": 0.46466442942619324,
1586
+ "eval_mae": 0.11605874449014664,
1587
+ "eval_rmse": 0.15662842988967896,
1588
+ "eval_runtime": 61.3486,
1589
+ "eval_samples_per_second": 38.371,
1590
+ "eval_steps_per_second": 2.412,
1591
+ "learning_rate": 1.0000000000000002e-06,
1592
+ "step": 36354
1593
+ },
1594
+ {
1595
+ "epoch": 83.33333333333333,
1596
+ "grad_norm": 0.2322608381509781,
1597
+ "learning_rate": 1.0000000000000002e-06,
1598
+ "loss": 0.4575,
1599
+ "step": 36500
1600
+ },
1601
+ {
1602
+ "epoch": 84.0,
1603
+ "eval_explained_variance": 0.46770432591438293,
1604
+ "eval_kl_divergence": 0.3855327367782593,
1605
+ "eval_loss": 0.46430692076683044,
1606
+ "eval_mae": 0.11573296785354614,
1607
+ "eval_rmse": 0.15642325580120087,
1608
+ "eval_runtime": 66.0549,
1609
+ "eval_samples_per_second": 35.637,
1610
+ "eval_steps_per_second": 2.241,
1611
+ "learning_rate": 1.0000000000000002e-06,
1612
+ "step": 36792
1613
+ },
1614
+ {
1615
+ "epoch": 84.47488584474885,
1616
+ "grad_norm": 0.23277735710144043,
1617
+ "learning_rate": 1.0000000000000002e-06,
1618
+ "loss": 0.4557,
1619
+ "step": 37000
1620
+ },
1621
+ {
1622
+ "epoch": 85.0,
1623
+ "eval_explained_variance": 0.46685031056404114,
1624
+ "eval_kl_divergence": 0.3372012972831726,
1625
+ "eval_loss": 0.46528080105781555,
1626
+ "eval_mae": 0.11732091754674911,
1627
+ "eval_rmse": 0.1570904403924942,
1628
+ "eval_runtime": 64.5003,
1629
+ "eval_samples_per_second": 36.496,
1630
+ "eval_steps_per_second": 2.295,
1631
+ "learning_rate": 1.0000000000000002e-06,
1632
+ "step": 37230
1633
+ },
1634
+ {
1635
+ "epoch": 85.61643835616438,
1636
+ "grad_norm": 0.40314728021621704,
1637
+ "learning_rate": 1.0000000000000002e-06,
1638
+ "loss": 0.4587,
1639
+ "step": 37500
1640
+ },
1641
+ {
1642
+ "epoch": 86.0,
1643
+ "eval_explained_variance": 0.46858659386634827,
1644
+ "eval_kl_divergence": 0.29690617322921753,
1645
+ "eval_loss": 0.46546733379364014,
1646
+ "eval_mae": 0.11843166500329971,
1647
+ "eval_rmse": 0.15721069276332855,
1648
+ "eval_runtime": 63.2115,
1649
+ "eval_samples_per_second": 37.24,
1650
+ "eval_steps_per_second": 2.341,
1651
+ "learning_rate": 1.0000000000000002e-06,
1652
+ "step": 37668
1653
+ },
1654
+ {
1655
+ "epoch": 86.7579908675799,
1656
+ "grad_norm": 0.2404189556837082,
1657
+ "learning_rate": 1.0000000000000002e-06,
1658
+ "loss": 0.4564,
1659
+ "step": 38000
1660
+ },
1661
+ {
1662
+ "epoch": 87.0,
1663
+ "eval_explained_variance": 0.4670089781284332,
1664
+ "eval_kl_divergence": 0.3571859300136566,
1665
+ "eval_loss": 0.4651782214641571,
1666
+ "eval_mae": 0.11727560311555862,
1667
+ "eval_rmse": 0.15713386237621307,
1668
+ "eval_runtime": 63.9279,
1669
+ "eval_samples_per_second": 36.823,
1670
+ "eval_steps_per_second": 2.315,
1671
+ "learning_rate": 1.0000000000000002e-06,
1672
+ "step": 38106
1673
+ },
1674
+ {
1675
+ "epoch": 87.89954337899543,
1676
+ "grad_norm": 0.27116990089416504,
1677
+ "learning_rate": 1.0000000000000002e-06,
1678
+ "loss": 0.4565,
1679
+ "step": 38500
1680
+ },
1681
+ {
1682
+ "epoch": 88.0,
1683
+ "eval_explained_variance": 0.4626586139202118,
1684
+ "eval_kl_divergence": 0.5179023146629333,
1685
+ "eval_loss": 0.4655611217021942,
1686
+ "eval_mae": 0.11513540893793106,
1687
+ "eval_rmse": 0.15777353942394257,
1688
+ "eval_runtime": 64.6969,
1689
+ "eval_samples_per_second": 36.385,
1690
+ "eval_steps_per_second": 2.288,
1691
+ "learning_rate": 1.0000000000000002e-06,
1692
+ "step": 38544
1693
+ },
1694
+ {
1695
+ "epoch": 89.0,
1696
+ "eval_explained_variance": 0.4670461118221283,
1697
+ "eval_kl_divergence": 0.29475125670433044,
1698
+ "eval_loss": 0.4654468297958374,
1699
+ "eval_mae": 0.11768680065870285,
1700
+ "eval_rmse": 0.15736806392669678,
1701
+ "eval_runtime": 64.3025,
1702
+ "eval_samples_per_second": 36.608,
1703
+ "eval_steps_per_second": 2.302,
1704
+ "learning_rate": 1.0000000000000002e-06,
1705
+ "step": 38982
1706
+ },
1707
+ {
1708
+ "epoch": 89.04109589041096,
1709
+ "grad_norm": 0.35173681378364563,
1710
+ "learning_rate": 1.0000000000000002e-06,
1711
+ "loss": 0.4569,
1712
+ "step": 39000
1713
+ },
1714
+ {
1715
+ "epoch": 90.0,
1716
+ "eval_explained_variance": 0.46744585037231445,
1717
+ "eval_kl_divergence": 0.34272825717926025,
1718
+ "eval_loss": 0.4649873971939087,
1719
+ "eval_mae": 0.11670031398534775,
1720
+ "eval_rmse": 0.15687011182308197,
1721
+ "eval_runtime": 63.1964,
1722
+ "eval_samples_per_second": 37.249,
1723
+ "eval_steps_per_second": 2.342,
1724
+ "learning_rate": 1.0000000000000002e-06,
1725
+ "step": 39420
1726
+ },
1727
+ {
1728
+ "epoch": 90.18264840182648,
1729
+ "grad_norm": 0.27881479263305664,
1730
+ "learning_rate": 1.0000000000000002e-07,
1731
+ "loss": 0.4561,
1732
+ "step": 39500
1733
+ },
1734
+ {
1735
+ "epoch": 91.0,
1736
+ "eval_explained_variance": 0.4690658748149872,
1737
+ "eval_kl_divergence": 0.27898427844047546,
1738
+ "eval_loss": 0.4655340015888214,
1739
+ "eval_mae": 0.11734825372695923,
1740
+ "eval_rmse": 0.15720801055431366,
1741
+ "eval_runtime": 62.9781,
1742
+ "eval_samples_per_second": 37.378,
1743
+ "eval_steps_per_second": 2.35,
1744
+ "learning_rate": 1.0000000000000002e-07,
1745
+ "step": 39858
1746
+ },
1747
+ {
1748
+ "epoch": 91.32420091324201,
1749
+ "grad_norm": 0.18466131389141083,
1750
+ "learning_rate": 1.0000000000000002e-07,
1751
+ "loss": 0.4575,
1752
+ "step": 40000
1753
+ },
1754
+ {
1755
+ "epoch": 92.0,
1756
+ "eval_explained_variance": 0.4671539068222046,
1757
+ "eval_kl_divergence": 0.41531136631965637,
1758
+ "eval_loss": 0.4645555317401886,
1759
+ "eval_mae": 0.11530810594558716,
1760
+ "eval_rmse": 0.15662522614002228,
1761
+ "eval_runtime": 65.5509,
1762
+ "eval_samples_per_second": 35.911,
1763
+ "eval_steps_per_second": 2.258,
1764
+ "learning_rate": 1.0000000000000002e-07,
1765
+ "step": 40296
1766
+ },
1767
+ {
1768
+ "epoch": 92.46575342465754,
1769
+ "grad_norm": 0.18533748388290405,
1770
+ "learning_rate": 1.0000000000000002e-07,
1771
+ "loss": 0.4569,
1772
+ "step": 40500
1773
+ },
1774
+ {
1775
+ "epoch": 93.0,
1776
+ "eval_explained_variance": 0.464511513710022,
1777
+ "eval_kl_divergence": 0.46641936898231506,
1778
+ "eval_loss": 0.4648568034172058,
1779
+ "eval_mae": 0.11531826853752136,
1780
+ "eval_rmse": 0.1570696085691452,
1781
+ "eval_runtime": 64.6542,
1782
+ "eval_samples_per_second": 36.409,
1783
+ "eval_steps_per_second": 2.289,
1784
+ "learning_rate": 1.0000000000000002e-07,
1785
+ "step": 40734
1786
+ },
1787
+ {
1788
+ "epoch": 93.60730593607306,
1789
+ "grad_norm": 0.27212274074554443,
1790
+ "learning_rate": 1.0000000000000002e-07,
1791
+ "loss": 0.456,
1792
+ "step": 41000
1793
+ },
1794
+ {
1795
+ "epoch": 94.0,
1796
+ "eval_explained_variance": 0.4661947786808014,
1797
+ "eval_kl_divergence": 0.3858942687511444,
1798
+ "eval_loss": 0.4652610421180725,
1799
+ "eval_mae": 0.11592459678649902,
1800
+ "eval_rmse": 0.15684206783771515,
1801
+ "eval_runtime": 63.3717,
1802
+ "eval_samples_per_second": 37.146,
1803
+ "eval_steps_per_second": 2.335,
1804
+ "learning_rate": 1.0000000000000002e-07,
1805
+ "step": 41172
1806
+ },
1807
+ {
1808
+ "epoch": 94.0,
1809
+ "learning_rate": 1.0000000000000002e-07,
1810
+ "step": 41172,
1811
+ "total_flos": 9.743166860733663e+19,
1812
+ "train_loss": 0.46343856458701926,
1813
+ "train_runtime": 26810.1674,
1814
+ "train_samples_per_second": 39.181,
1815
+ "train_steps_per_second": 2.451
1816
+ }
1817
+ ],
1818
+ "logging_steps": 500,
1819
+ "max_steps": 65700,
1820
+ "num_input_tokens_seen": 0,
1821
+ "num_train_epochs": 150,
1822
+ "save_steps": 500,
1823
+ "stateful_callbacks": {
1824
+ "EarlyStoppingCallback": {
1825
+ "args": {
1826
+ "early_stopping_patience": 10,
1827
+ "early_stopping_threshold": 0.0
1828
+ },
1829
+ "attributes": {
1830
+ "early_stopping_patience_counter": 0
1831
+ }
1832
+ },
1833
+ "TrainerControl": {
1834
+ "args": {
1835
+ "should_epoch_stop": false,
1836
+ "should_evaluate": false,
1837
+ "should_log": false,
1838
+ "should_save": true,
1839
+ "should_training_stop": true
1840
+ },
1841
+ "attributes": {}
1842
+ }
1843
+ },
1844
+ "total_flos": 9.743166860733663e+19,
1845
+ "train_batch_size": 16,
1846
+ "trial_name": null,
1847
+ "trial_params": null
1848
+ }