sayakpaul/glpn-nyu-finetuned-diode-221214-054706

81次阅读

Training Loss	Epoch	Step	Validation Loss	Mae	Rmse	Abs Rel	Log Mae	Log Rmse	Delta1	Delta2	Delta3
0.746	1.0	72	0.5730	0.7316	0.8964	0.9501	0.2530	0.3005	0.1427	0.3446	0.6453
0.4215	2.0	144	0.3764	0.3140	0.4548	0.3607	0.1286	0.1819	0.5054	0.7927	0.9161
0.3566	3.0	216	0.3568	0.2894	0.4226	0.3436	0.1206	0.1751	0.5475	0.8178	0.9257
0.322	4.0	288	0.3523	0.2818	0.4069	0.3541	0.1188	0.1740	0.5508	0.8219	0.9325
0.3066	5.0	360	0.3464	0.2777	0.4090	0.3283	0.1164	0.1695	0.5609	0.8328	0.9313
0.256	6.0	432	0.3384	0.2687	0.3927	0.3173	0.1136	0.1647	0.5665	0.8381	0.9394
0.2268	7.0	504	0.3407	0.2747	0.4138	0.3091	0.1143	0.1672	0.5716	0.8380	0.9322
0.1721	8.0	576	0.3346	0.2631	0.3850	0.3176	0.1108	0.1636	0.5804	0.8489	0.9452
0.1898	9.0	648	0.3362	0.2656	0.3994	0.3100	0.1112	0.1650	0.5910	0.8455	0.9368
0.1609	10.0	720	0.3355	0.2667	0.3866	0.3242	0.1124	0.1648	0.5693	0.8452	0.9433
0.139	11.0	792	0.3344	0.2660	0.3954	0.3115	0.1112	0.1639	0.5833	0.8453	0.9409
0.1772	12.0	864	0.3360	0.2667	0.3979	0.3137	0.1115	0.1650	0.5854	0.8449	0.9393
0.1631	13.0	936	0.3353	0.2652	0.3938	0.3134	0.1112	0.1645	0.5858	0.8451	0.9403
0.1736	14.0	1008	0.3367	0.2663	0.3954	0.3163	0.1117	0.1653	0.5840	0.8446	0.9391
0.1736	15.0	1080	0.3340	0.2649	0.3917	0.3138	0.1111	0.1640	0.5843	0.8459	0.9413

glpn-nyu-finetuned-diode-221214-054706

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: 0.3340

Mae: 0.2649

Rmse: 0.3917

Abs Rel: 0.3138

Log Mae: 0.1111

Log Rmse: 0.1640

Delta1: 0.5843

Delta2: 0.8459

Delta3: 0.9413

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

0.746 1.0 72 0.5730 0.7316 0.8964 0.9501 0.2530 0.3005 0.1427 0.3446 0.6453

0.4215 2.0 144 0.3764 0.3140 0.4548 0.3607 0.1286 0.1819 0.5054 0.7927 0.9161

0.3566 3.0 216 0.3568 0.2894 0.4226 0.3436 0.1206 0.1751 0.5475 0.8178 0.9257

0.322 4.0 288 0.3523 0.2818 0.4069 0.3541 0.1188 0.1740 0.5508 0.8219 0.9325

0.3066 5.0 360 0.3464 0.2777 0.4090 0.3283 0.1164 0.1695 0.5609 0.8328 0.9313

0.256 6.0 432 0.3384 0.2687 0.3927 0.3173 0.1136 0.1647 0.5665 0.8381 0.9394

0.2268 7.0 504 0.3407 0.2747 0.4138 0.3091 0.1143 0.1672 0.5716 0.8380 0.9322

0.1721 8.0 576 0.3346 0.2631 0.3850 0.3176 0.1108 0.1636 0.5804 0.8489 0.9452

0.1898 9.0 648 0.3362 0.2656 0.3994 0.3100 0.1112 0.1650 0.5910 0.8455 0.9368

0.1609 10.0 720 0.3355 0.2667 0.3866 0.3242 0.1124 0.1648 0.5693 0.8452 0.9433

0.139 11.0 792 0.3344 0.2660 0.3954 0.3115 0.1112 0.1639 0.5833 0.8453 0.9409

0.1772 12.0 864 0.3360 0.2667 0.3979 0.3137 0.1115 0.1650 0.5854 0.8449 0.9393

0.1631 13.0 936 0.3353 0.2652 0.3938 0.3134 0.1112 0.1645 0.5858 0.8451 0.9403

0.1736 14.0 1008 0.3367 0.2663 0.3954 0.3163 0.1117 0.1653 0.5840 0.8446 0.9391

0.1736 15.0 1080 0.3340 0.2649 0.3917 0.3138 0.1111 0.1640 0.5843 0.8459 0.9413