sayakpaul/glpn-nyu-finetuned-diode-221122-082237

155次阅读

Training Loss	Epoch	Step	Validation Loss	Mae	Rmse	Abs Rel	Log Mae	Log Rmse	Delta1	Delta2	Delta3
0.7577	1.0	72	0.5814	0.7579	0.9267	0.9816	0.2592	0.3062	0.1341	0.3276	0.6248
0.4447	2.0	144	0.3947	0.3412	0.4785	0.4051	0.1381	0.1912	0.4585	0.7697	0.9072
0.4034	3.0	216	0.3657	0.2988	0.4357	0.3629	0.1242	0.1802	0.5321	0.8054	0.9202
0.3726	4.0	288	0.3576	0.2896	0.4176	0.3647	0.1215	0.1769	0.5376	0.8178	0.9270
0.3656	5.0	360	0.3547	0.2818	0.4098	0.3504	0.1186	0.1732	0.5551	0.8225	0.9283
0.3211	6.0	432	0.3495	0.2743	0.4160	0.3250	0.1155	0.1708	0.5773	0.8317	0.9258
0.3027	7.0	504	0.3471	0.2724	0.4123	0.3226	0.1146	0.1695	0.5801	0.8345	0.9283
0.2438	8.0	576	0.3494	0.2735	0.4097	0.3304	0.1157	0.1708	0.5741	0.8328	0.9267
0.2512	9.0	648	0.3448	0.2700	0.4121	0.3160	0.1134	0.1683	0.5880	0.8395	0.9266
0.2416	10.0	720	0.3439	0.2688	0.4017	0.3255	0.1135	0.1682	0.5781	0.8397	0.9324
0.1971	11.0	792	0.3456	0.2730	0.4059	0.3348	0.1148	0.1703	0.5730	0.8370	0.9305
0.2382	12.0	864	0.3446	0.2708	0.4069	0.3283	0.1139	0.1696	0.5818	0.8394	0.9295
0.237	13.0	936	0.3417	0.2674	0.4038	0.3218	0.1123	0.1680	0.5901	0.8424	0.9307
0.2378	14.0	1008	0.3421	0.2686	0.4030	0.3241	0.1130	0.1683	0.5852	0.8408	0.9311
0.2409	15.0	1080	0.3421	0.2700	0.4042	0.3279	0.1132	0.1688	0.5839	0.8408	0.9309

glpn-nyu-finetuned-diode-221122-082237

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: 0.3421

Mae: 0.2700

Rmse: 0.4042

Abs Rel: 0.3279

Log Mae: 0.1132

Log Rmse: 0.1688

Delta1: 0.5839

Delta2: 0.8408

Delta3: 0.9309

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

0.7577 1.0 72 0.5814 0.7579 0.9267 0.9816 0.2592 0.3062 0.1341 0.3276 0.6248

0.4447 2.0 144 0.3947 0.3412 0.4785 0.4051 0.1381 0.1912 0.4585 0.7697 0.9072

0.4034 3.0 216 0.3657 0.2988 0.4357 0.3629 0.1242 0.1802 0.5321 0.8054 0.9202

0.3726 4.0 288 0.3576 0.2896 0.4176 0.3647 0.1215 0.1769 0.5376 0.8178 0.9270

0.3656 5.0 360 0.3547 0.2818 0.4098 0.3504 0.1186 0.1732 0.5551 0.8225 0.9283

0.3211 6.0 432 0.3495 0.2743 0.4160 0.3250 0.1155 0.1708 0.5773 0.8317 0.9258

0.3027 7.0 504 0.3471 0.2724 0.4123 0.3226 0.1146 0.1695 0.5801 0.8345 0.9283

0.2438 8.0 576 0.3494 0.2735 0.4097 0.3304 0.1157 0.1708 0.5741 0.8328 0.9267

0.2512 9.0 648 0.3448 0.2700 0.4121 0.3160 0.1134 0.1683 0.5880 0.8395 0.9266

0.2416 10.0 720 0.3439 0.2688 0.4017 0.3255 0.1135 0.1682 0.5781 0.8397 0.9324

0.1971 11.0 792 0.3456 0.2730 0.4059 0.3348 0.1148 0.1703 0.5730 0.8370 0.9305

0.2382 12.0 864 0.3446 0.2708 0.4069 0.3283 0.1139 0.1696 0.5818 0.8394 0.9295

0.237 13.0 936 0.3417 0.2674 0.4038 0.3218 0.1123 0.1680 0.5901 0.8424 0.9307

0.2378 14.0 1008 0.3421 0.2686 0.4030 0.3241 0.1130 0.1683 0.5852 0.8408 0.9311

0.2409 15.0 1080 0.3421 0.2700 0.4042 0.3279 0.1132 0.1688 0.5839 0.8408 0.9309