segformer-b5-random-init-morphpadver1-hgo-coord-v9_mix_resample_40epochs

This model is a fine-tuned version of random_init on the NICOPOI-9/morphpad_coord_hgo_512_4class_v2 dataset. It achieves the following results on the evaluation set:

  • Loss: 1.3932
  • Mean Iou: 0.1216
  • Mean Accuracy: 0.2761
  • Overall Accuracy: 0.3021
  • Accuracy 0-0: 0.0
  • Accuracy 0-90: 0.4718
  • Accuracy 90-0: 0.6292
  • Accuracy 90-90: 0.0032
  • Iou 0-0: 0.0
  • Iou 0-90: 0.2342
  • Iou 90-0: 0.2489
  • Iou 90-90: 0.0032

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: 6e-05
  • train_batch_size: 1
  • eval_batch_size: 1
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 40

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Accuracy 0-0 Accuracy 0-90 Accuracy 90-0 Accuracy 90-90 Iou 0-0 Iou 0-90 Iou 90-0 Iou 90-90
1.3906 1.3638 4000 1.3826 0.1056 0.2465 0.2708 0.0010 0.5901 0.3924 0.0027 0.0010 0.2307 0.1879 0.0027
1.3995 2.7276 8000 1.3857 0.1089 0.2601 0.2673 0.0051 0.6062 0.0089 0.4203 0.0050 0.2424 0.0087 0.1794
1.389 4.0914 12000 1.3877 0.1035 0.2537 0.2540 0.0001 0.0091 0.5170 0.4888 0.0001 0.0088 0.2170 0.1881
1.3914 5.4552 16000 1.3853 0.1124 0.2572 0.2587 0.0475 0.4953 0.0007 0.4854 0.0413 0.2177 0.0007 0.1900
1.3922 6.8190 20000 1.3829 0.1068 0.2485 0.2724 0.0 0.4926 0.5013 0.0000 0.0 0.2142 0.2129 0.0000
1.3338 8.1827 24000 1.3827 0.0865 0.2484 0.2748 0.0 0.8928 0.1009 0.0001 0.0 0.2670 0.0789 0.0001
1.4147 9.5465 28000 1.3838 0.1115 0.2508 0.2742 0.0 0.5011 0.4838 0.0184 0.0 0.2185 0.2102 0.0173
1.3939 10.9103 32000 1.3884 0.1249 0.2523 0.2712 0.0048 0.4879 0.3971 0.1195 0.0047 0.2164 0.1927 0.0857
1.4044 12.2741 36000 1.3832 0.0718 0.2501 0.2772 0.0 0.9894 0.0089 0.0021 0.0 0.2765 0.0087 0.0021
1.3871 13.6379 40000 1.3827 0.1042 0.2491 0.2722 0.0000 0.3397 0.6565 0.0003 0.0000 0.1791 0.2374 0.0003
1.3797 15.0017 44000 1.3841 0.0851 0.2514 0.2732 0.0 0.0927 0.9130 0.0001 0.0 0.0751 0.2653 0.0001
1.3914 16.3655 48000 1.3831 0.0824 0.2511 0.2728 0.0 0.0748 0.9296 0.0001 0.0 0.0630 0.2664 0.0001
1.3861 17.7293 52000 1.3851 0.1132 0.2495 0.2721 0.0344 0.5293 0.4342 0.0 0.0307 0.2211 0.2010 0.0
1.3791 19.0931 56000 1.3836 0.1088 0.2519 0.2763 0.0006 0.5300 0.4760 0.0009 0.0006 0.2251 0.2088 0.0009
1.3958 20.4569 60000 1.3844 0.0722 0.2505 0.2718 0.0000 0.0198 0.9823 0.0000 0.0000 0.0189 0.2700 0.0000
1.3817 21.8207 64000 1.3848 0.1107 0.2548 0.2792 0.0000 0.4950 0.5207 0.0034 0.0000 0.2196 0.2196 0.0034
1.3664 23.1845 68000 1.3866 0.1280 0.2586 0.2787 0.0064 0.4989 0.4251 0.1041 0.0062 0.2197 0.2100 0.0761
1.395 24.5482 72000 1.3839 0.1177 0.2689 0.2949 0.0 0.5445 0.5307 0.0006 0.0 0.2394 0.2309 0.0006
1.3716 25.9120 76000 1.3856 0.1175 0.2690 0.2953 0.0 0.5923 0.4833 0.0002 0.0 0.2479 0.2218 0.0002
1.399 27.2758 80000 1.3909 0.1150 0.2687 0.2957 0.0004 0.7046 0.3698 0.0001 0.0004 0.2636 0.1960 0.0001
1.3874 28.6396 84000 1.3893 0.1199 0.2728 0.2993 0.0000 0.5819 0.5084 0.0007 0.0000 0.2499 0.2287 0.0007
1.3876 30.0034 88000 1.3992 0.1195 0.2738 0.2997 0.0002 0.4573 0.6373 0.0003 0.0002 0.2284 0.2492 0.0003
1.4595 31.3672 92000 1.3937 0.1152 0.2689 0.2958 0.0 0.7006 0.3749 0.0000 0.0 0.2641 0.1966 0.0000
1.3356 32.7310 96000 1.3841 0.1217 0.2765 0.3034 0.0 0.5904 0.5155 0.0000 0.0 0.2550 0.2317 0.0000
1.3761 34.0948 100000 1.3939 0.1225 0.2756 0.3018 0.0060 0.5329 0.5634 0.0000 0.0058 0.2452 0.2390 0.0000
1.3373 35.4586 104000 1.3953 0.1151 0.2693 0.2963 0.0 0.7132 0.3634 0.0005 0.0 0.2661 0.1939 0.0005
1.3926 36.8224 108000 1.3947 0.1222 0.2711 0.2969 0.0 0.6567 0.4024 0.0253 0.0 0.2603 0.2052 0.0232
1.3749 38.1862 112000 1.3909 0.1140 0.2690 0.2962 0.0 0.7359 0.3399 0.0004 0.0 0.2686 0.1871 0.0004
1.3604 39.5499 116000 1.3932 0.1216 0.2761 0.3021 0.0 0.4718 0.6292 0.0032 0.0 0.2342 0.2489 0.0032

Framework versions

  • Transformers 4.48.3
  • Pytorch 2.1.0
  • Datasets 3.2.0
  • Tokenizers 0.21.0
Downloads last month
6
Safetensors
Model size
84.6M params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support