Model Database's logo

sayakpaul
/
glpn-kitti-finetuned-diode-221214-123047

Depth Estimation Transformers PyTorch glpn vision generated_from_trainer License: apache-2.0
Model card Files Community
Edit model card

glpn-kitti-finetuned-diode-221214-123047

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

  • Loss: 0.3497
  • Mae: 0.2847
  • Rmse: 0.3977
  • Abs Rel: 0.3477
  • Log Mae: 0.1203
  • Log Rmse: 0.1726
  • Delta1: 0.5217
  • Delta2: 0.8246
  • Delta3: 0.9436

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: 5e-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: 0.15
  • num_epochs: 25
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Mae Rmse Abs Rel Log Mae Log Rmse Delta1 Delta2 Delta3
0.6103 1.0 72 0.4449 0.3914 0.5513 0.4625 0.1615 0.2186 0.3918 0.6910 0.8549
0.3762 2.0 144 0.4095 0.3583 0.4876 0.4281 0.1505 0.2015 0.4065 0.7121 0.8901
0.341 3.0 216 0.3768 0.3046 0.4061 0.4016 0.1313 0.1840 0.4757 0.7938 0.9309
0.291 4.0 288 0.3853 0.3227 0.4495 0.3724 0.1360 0.1869 0.4646 0.7680 0.9127
0.2861 5.0 360 0.3786 0.3151 0.4257 0.4065 0.1344 0.1876 0.4597 0.7785 0.9329
0.2539 6.0 432 0.3687 0.3158 0.4546 0.3329 0.1316 0.1821 0.4732 0.7869 0.9138
0.2199 7.0 504 0.3705 0.3122 0.4479 0.3378 0.1312 0.1820 0.4784 0.7888 0.9189
0.1728 8.0 576 0.3578 0.2895 0.4008 0.3675 0.1235 0.1766 0.5101 0.8178 0.9420
0.1877 9.0 648 0.3589 0.2846 0.3846 0.3721 0.1235 0.1764 0.5144 0.8170 0.9403
0.1541 10.0 720 0.3521 0.2831 0.3997 0.3283 0.1201 0.1712 0.5241 0.8260 0.9422
0.1414 11.0 792 0.3460 0.2735 0.3772 0.3419 0.1173 0.1691 0.5409 0.8360 0.9469
0.1643 12.0 864 0.3530 0.2878 0.4100 0.3313 0.1214 0.1736 0.5249 0.8214 0.9344
0.1724 13.0 936 0.3606 0.2995 0.4249 0.3459 0.1255 0.1775 0.5057 0.8069 0.9323
0.1514 14.0 1008 0.3477 0.2832 0.3881 0.3596 0.1206 0.1726 0.5174 0.8253 0.9437
0.1535 15.0 1080 0.3535 0.2961 0.4242 0.3412 0.1231 0.1753 0.5186 0.8080 0.9332
0.1233 16.0 1152 0.3508 0.2896 0.4104 0.3391 0.1213 0.1727 0.5225 0.8165 0.9398
0.116 17.0 1224 0.3519 0.2874 0.3989 0.3533 0.1215 0.1731 0.5200 0.8179 0.9407
0.1532 18.0 1296 0.3532 0.2965 0.4200 0.3459 0.1236 0.1747 0.5147 0.8035 0.9353
0.1179 19.0 1368 0.3497 0.2828 0.3896 0.3557 0.1204 0.1728 0.5200 0.8260 0.9457
0.1326 20.0 1440 0.3467 0.2787 0.3848 0.3475 0.1185 0.1704 0.5257 0.8330 0.9479
0.1069 21.0 1512 0.3471 0.2807 0.3922 0.3418 0.1187 0.1707 0.5288 0.8297 0.9452
0.1049 22.0 1584 0.3474 0.2864 0.4048 0.3387 0.1199 0.1717 0.5227 0.8251 0.9428
0.103 23.0 1656 0.3483 0.2840 0.3991 0.3416 0.1196 0.1717 0.5254 0.8269 0.9431
0.1184 24.0 1728 0.3473 0.2839 0.3960 0.3450 0.1198 0.1717 0.5223 0.8251 0.9443
0.1258 25.0 1800 0.3497 0.2847 0.3977 0.3477 0.1203 0.1726 0.5217 0.8246 0.9436

Framework versions

  • Transformers 4.24.0
  • Pytorch 1.12.1+cu116
  • Tokenizers 0.13.2
Downloads last month
0

Evaluation results

Metadata error: specify a dataset to view leaderboard