sayakpaul/glpn-nyu-finetuned-diode-221122-044810

71次阅读

Training Loss	Epoch	Step	Validation Loss	Mae	Rmse	Abs Rel	Log Mae	Log Rmse	Delta1	Delta2	Delta3
1.3864	1.0	72	1.2016	3.4656	3.5204	5.1101	0.6881	0.7346	0.0	0.0011	0.0764
1.0082	2.0	144	0.4607	0.4107	0.5420	0.5254	0.1697	0.2234	0.3596	0.6460	0.8465
0.4656	3.0	216	0.4071	0.3431	0.4758	0.4359	0.1425	0.1992	0.4567	0.7481	0.8958
0.4093	4.0	288	0.3953	0.3261	0.4622	0.4197	0.1363	0.1947	0.4841	0.7624	0.9103
0.392	5.0	360	0.3916	0.3211	0.4463	0.4116	0.1338	0.1896	0.4810	0.7756	0.9176
0.3466	6.0	432	0.3807	0.3075	0.4451	0.3658	0.1293	0.1839	0.5026	0.7921	0.9180
0.3297	7.0	504	0.3811	0.3047	0.4448	0.3534	0.1290	0.1835	0.5066	0.7920	0.9137
0.2768	8.0	576	0.3779	0.3057	0.4283	0.3894	0.1280	0.1832	0.5046	0.7996	0.9256
0.2849	9.0	648	0.3753	0.2978	0.4341	0.3496	0.1259	0.1806	0.5149	0.8041	0.9184
0.2571	10.0	720	0.3825	0.3068	0.4305	0.3896	0.1289	0.1849	0.4998	0.7974	0.9206
0.2246	11.0	792	0.3718	0.2951	0.4235	0.3678	0.1240	0.1803	0.5249	0.8105	0.9248
0.2703	12.0	864	0.3716	0.2945	0.4317	0.3593	0.1235	0.1808	0.5324	0.8122	0.9215
0.2596	13.0	936	0.3692	0.2921	0.4185	0.3690	0.1229	0.1798	0.5294	0.8167	0.9264
0.2604	14.0	1008	0.3684	0.2893	0.4171	0.3601	0.1223	0.1785	0.5325	0.8179	0.9252
0.2679	15.0	1080	0.3690	0.2909	0.4208	0.3635	0.1224	0.1793	0.5323	0.8179	0.9258

glpn-nyu-finetuned-diode-221122-044810

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: 0.3690

Mae: 0.2909

Rmse: 0.4208

Abs Rel: 0.3635

Log Mae: 0.1224

Log Rmse: 0.1793

Delta1: 0.5323

Delta2: 0.8179

Delta3: 0.9258

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: 2e-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: 0.2

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

1.3864 1.0 72 1.2016 3.4656 3.5204 5.1101 0.6881 0.7346 0.0 0.0011 0.0764

1.0082 2.0 144 0.4607 0.4107 0.5420 0.5254 0.1697 0.2234 0.3596 0.6460 0.8465

0.4656 3.0 216 0.4071 0.3431 0.4758 0.4359 0.1425 0.1992 0.4567 0.7481 0.8958

0.4093 4.0 288 0.3953 0.3261 0.4622 0.4197 0.1363 0.1947 0.4841 0.7624 0.9103

0.392 5.0 360 0.3916 0.3211 0.4463 0.4116 0.1338 0.1896 0.4810 0.7756 0.9176

0.3466 6.0 432 0.3807 0.3075 0.4451 0.3658 0.1293 0.1839 0.5026 0.7921 0.9180

0.3297 7.0 504 0.3811 0.3047 0.4448 0.3534 0.1290 0.1835 0.5066 0.7920 0.9137

0.2768 8.0 576 0.3779 0.3057 0.4283 0.3894 0.1280 0.1832 0.5046 0.7996 0.9256

0.2849 9.0 648 0.3753 0.2978 0.4341 0.3496 0.1259 0.1806 0.5149 0.8041 0.9184

0.2571 10.0 720 0.3825 0.3068 0.4305 0.3896 0.1289 0.1849 0.4998 0.7974 0.9206

0.2246 11.0 792 0.3718 0.2951 0.4235 0.3678 0.1240 0.1803 0.5249 0.8105 0.9248

0.2703 12.0 864 0.3716 0.2945 0.4317 0.3593 0.1235 0.1808 0.5324 0.8122 0.9215

0.2596 13.0 936 0.3692 0.2921 0.4185 0.3690 0.1229 0.1798 0.5294 0.8167 0.9264

0.2604 14.0 1008 0.3684 0.2893 0.4171 0.3601 0.1223 0.1785 0.5325 0.8179 0.9252

0.2679 15.0 1080 0.3690 0.2909 0.4208 0.3635 0.1224 0.1793 0.5323 0.8179 0.9258