sayakpaul/glpn-nyu-finetuned-diode-221122-014502

49次阅读

Training Loss	Epoch	Step	Validation Loss	Mae	Rmse	Abs Rel	Log Mae	Log Rmse	Delta1	Delta2	Delta3
0.7598	1.0	72	0.5809	0.7606	0.9281	0.9834	0.2597	0.3064	0.1320	0.3250	0.6234
0.4481	2.0	144	0.4013	0.3507	0.4879	0.4181	0.1415	0.1950	0.4427	0.7602	0.9021
0.4066	3.0	216	0.3706	0.3081	0.4484	0.3675	0.1269	0.1823	0.5187	0.7977	0.9148
0.3965	4.0	288	0.3641	0.2987	0.4336	0.3607	0.1239	0.1787	0.5294	0.8072	0.9205
0.3942	5.0	360	0.3582	0.2903	0.4251	0.3490	0.1207	0.1753	0.5466	0.8165	0.9232
0.3575	6.0	432	0.3568	0.2898	0.4184	0.3569	0.1211	0.1753	0.5390	0.8171	0.9265
0.3418	7.0	504	0.3490	0.2771	0.4178	0.3248	0.1156	0.1707	0.5783	0.8312	0.9259
0.2916	8.0	576	0.3512	0.2819	0.4172	0.3373	0.1178	0.1725	0.5620	0.8253	0.9262
0.3055	9.0	648	0.3506	0.2808	0.4091	0.3422	0.1180	0.1718	0.5537	0.8248	0.9292
0.2932	10.0	720	0.3518	0.2809	0.4110	0.3441	0.1182	0.1724	0.5548	0.8239	0.9290
0.2518	11.0	792	0.3476	0.2756	0.4115	0.3265	0.1155	0.1700	0.5741	0.8326	0.9264
0.3177	12.0	864	0.3491	0.2784	0.4104	0.3333	0.1169	0.1706	0.5620	0.8290	0.9283
0.3038	13.0	936	0.3503	0.2795	0.4094	0.3410	0.1175	0.1717	0.5596	0.8275	0.9283
0.3299	14.0	1008	0.3460	0.2750	0.4098	0.3257	0.1154	0.1693	0.5721	0.8325	0.9283
0.3325	15.0	1080	0.3476	0.2763	0.4088	0.3308	0.1161	0.1700	0.5682	0.8301	0.9279

glpn-nyu-finetuned-diode-221122-014502

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: 0.3476

Mae: 0.2763

Rmse: 0.4088

Abs Rel: 0.3308

Log Mae: 0.1161

Log Rmse: 0.1700

Delta1: 0.5682

Delta2: 0.8301

Delta3: 0.9279

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

0.7598 1.0 72 0.5809 0.7606 0.9281 0.9834 0.2597 0.3064 0.1320 0.3250 0.6234

0.4481 2.0 144 0.4013 0.3507 0.4879 0.4181 0.1415 0.1950 0.4427 0.7602 0.9021

0.4066 3.0 216 0.3706 0.3081 0.4484 0.3675 0.1269 0.1823 0.5187 0.7977 0.9148

0.3965 4.0 288 0.3641 0.2987 0.4336 0.3607 0.1239 0.1787 0.5294 0.8072 0.9205

0.3942 5.0 360 0.3582 0.2903 0.4251 0.3490 0.1207 0.1753 0.5466 0.8165 0.9232

0.3575 6.0 432 0.3568 0.2898 0.4184 0.3569 0.1211 0.1753 0.5390 0.8171 0.9265

0.3418 7.0 504 0.3490 0.2771 0.4178 0.3248 0.1156 0.1707 0.5783 0.8312 0.9259

0.2916 8.0 576 0.3512 0.2819 0.4172 0.3373 0.1178 0.1725 0.5620 0.8253 0.9262

0.3055 9.0 648 0.3506 0.2808 0.4091 0.3422 0.1180 0.1718 0.5537 0.8248 0.9292

0.2932 10.0 720 0.3518 0.2809 0.4110 0.3441 0.1182 0.1724 0.5548 0.8239 0.9290

0.2518 11.0 792 0.3476 0.2756 0.4115 0.3265 0.1155 0.1700 0.5741 0.8326 0.9264

0.3177 12.0 864 0.3491 0.2784 0.4104 0.3333 0.1169 0.1706 0.5620 0.8290 0.9283

0.3038 13.0 936 0.3503 0.2795 0.4094 0.3410 0.1175 0.1717 0.5596 0.8275 0.9283

0.3299 14.0 1008 0.3460 0.2750 0.4098 0.3257 0.1154 0.1693 0.5721 0.8325 0.9283

0.3325 15.0 1080 0.3476 0.2763 0.4088 0.3308 0.1161 0.1700 0.5682 0.8301 0.9279