sayakpaul/glpn-nyu-finetuned-diode-221121-063504

97次阅读

Training Loss	Epoch	Step	Validation Loss	Mae	Rmse	Abs Rel	Log Mae	Log Rmse	Delta1	Delta2	Delta3
1.3991	1.0	72	1.2199	3.6023	3.6519	5.2780	0.7010	0.7461	0.0	0.0007	0.0616
1.1099	2.0	144	0.7471	1.2562	1.5028	1.6644	0.3550	0.4165	0.0965	0.2342	0.4292
0.5036	3.0	216	0.4876	0.5019	0.6198	0.7615	0.2000	0.2599	0.2878	0.5643	0.7803
0.4157	4.0	288	0.3789	0.3008	0.4211	0.3793	0.1291	0.1840	0.4961	0.7961	0.9261
0.4043	5.0	360	0.3795	0.3025	0.4117	0.4028	0.1303	0.1850	0.4889	0.7892	0.9278
0.3638	6.0	432	0.3790	0.3022	0.4019	0.4175	0.1313	0.1862	0.4851	0.7889	0.9262
0.3532	7.0	504	0.3605	0.2756	0.3864	0.3447	0.1201	0.1732	0.5397	0.8202	0.9330
0.3087	8.0	576	0.3599	0.2781	0.3896	0.3365	0.1206	0.1722	0.5312	0.8183	0.9332
0.3232	9.0	648	0.3613	0.2772	0.3879	0.3444	0.1204	0.1733	0.5341	0.8237	0.9334
0.3072	10.0	720	0.3570	0.2752	0.3794	0.3582	0.1195	0.1731	0.5341	0.8270	0.9374
0.2673	11.0	792	0.3633	0.2747	0.3838	0.3390	0.1207	0.1728	0.5330	0.8221	0.9333
0.3222	12.0	864	0.3548	0.2713	0.3783	0.3441	0.1180	0.1711	0.5448	0.8315	0.9367
0.3072	13.0	936	0.3532	0.2668	0.3700	0.3441	0.1168	0.1701	0.5502	0.8353	0.9387
0.3214	14.0	1008	0.3553	0.2674	0.3747	0.3322	0.1177	0.1701	0.5472	0.8324	0.9355
0.3406	15.0	1080	0.3533	0.2668	0.3716	0.3427	0.1167	0.1703	0.5522	0.8362	0.9382

glpn-nyu-finetuned-diode-221121-063504

This model is a fine-tuned version of vinvino02/glpn-nyu on the diode-subset dataset.
It achieves the following results on the evaluation set:

Loss: 0.3533

Mae: 0.2668

Rmse: 0.3716

Abs Rel: 0.3427

Log Mae: 0.1167

Log Rmse: 0.1703

Delta1: 0.5522

Delta2: 0.8362

Delta3: 0.9382

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-05

train_batch_size: 24

eval_batch_size: 48

seed: 2022

optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08

lr_scheduler_type: linear

lr_scheduler_warmup_ratio: 0.1

num_epochs: 15

mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Mae Rmse Abs Rel Log Mae Log Rmse Delta1 Delta2 Delta3

1.3991 1.0 72 1.2199 3.6023 3.6519 5.2780 0.7010 0.7461 0.0 0.0007 0.0616

1.1099 2.0 144 0.7471 1.2562 1.5028 1.6644 0.3550 0.4165 0.0965 0.2342 0.4292

0.5036 3.0 216 0.4876 0.5019 0.6198 0.7615 0.2000 0.2599 0.2878 0.5643 0.7803

0.4157 4.0 288 0.3789 0.3008 0.4211 0.3793 0.1291 0.1840 0.4961 0.7961 0.9261

0.4043 5.0 360 0.3795 0.3025 0.4117 0.4028 0.1303 0.1850 0.4889 0.7892 0.9278

0.3638 6.0 432 0.3790 0.3022 0.4019 0.4175 0.1313 0.1862 0.4851 0.7889 0.9262

0.3532 7.0 504 0.3605 0.2756 0.3864 0.3447 0.1201 0.1732 0.5397 0.8202 0.9330

0.3087 8.0 576 0.3599 0.2781 0.3896 0.3365 0.1206 0.1722 0.5312 0.8183 0.9332

0.3232 9.0 648 0.3613 0.2772 0.3879 0.3444 0.1204 0.1733 0.5341 0.8237 0.9334

0.3072 10.0 720 0.3570 0.2752 0.3794 0.3582 0.1195 0.1731 0.5341 0.8270 0.9374

0.2673 11.0 792 0.3633 0.2747 0.3838 0.3390 0.1207 0.1728 0.5330 0.8221 0.9333

0.3222 12.0 864 0.3548 0.2713 0.3783 0.3441 0.1180 0.1711 0.5448 0.8315 0.9367

0.3072 13.0 936 0.3532 0.2668 0.3700 0.3441 0.1168 0.1701 0.5502 0.8353 0.9387

0.3214 14.0 1008 0.3553 0.2674 0.3747 0.3322 0.1177 0.1701 0.5472 0.8324 0.9355

0.3406 15.0 1080 0.3533 0.2668 0.3716 0.3427 0.1167 0.1703 0.5522 0.8362 0.9382