Edit model card

14-classifier-finetuned-padchest

This model is a fine-tuned version of microsoft/resnet-50 on the imagefolder dataset. It achieves the following results on the evaluation set:

  • Loss: 0.9826
  • F1: 0.6620

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: 32
  • eval_batch_size: 32
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 100

Training results

Training Loss Epoch Step Validation Loss F1
2.0848 1.0 18 2.0904 0.0451
2.0822 2.0 36 2.0842 0.0848
2.0681 3.0 54 2.0742 0.1092
2.0568 4.0 72 2.0624 0.1443
2.0363 5.0 90 2.0420 0.1747
2.0235 6.0 108 2.0288 0.1533
1.9984 7.0 126 2.0008 0.1816
1.9647 8.0 144 1.9752 0.1849
1.9257 9.0 162 1.9481 0.2075
1.8791 10.0 180 1.9060 0.2219
1.8561 11.0 198 1.8678 0.2734
1.803 12.0 216 1.8322 0.2039
1.7461 13.0 234 1.8045 0.2939
1.7169 14.0 252 1.8216 0.3012
1.6773 15.0 270 1.7588 0.3166
1.6724 16.0 288 1.7379 0.2726
1.6286 17.0 306 1.7109 0.3248
1.5533 18.0 324 1.6492 0.3294
1.5075 19.0 342 1.5951 0.3394
1.4789 20.0 360 1.5657 0.3643
1.4077 21.0 378 1.5287 0.3665
1.4146 22.0 396 1.4897 0.4099
1.3583 23.0 414 1.4704 0.3765
1.3486 24.0 432 1.4469 0.3813
1.2947 25.0 450 1.4228 0.4049
1.3272 26.0 468 1.4035 0.4203
1.3048 27.0 486 1.3907 0.4316
1.2898 28.0 504 1.3992 0.4520
1.2204 29.0 522 1.3751 0.4952
1.2298 30.0 540 1.3658 0.4771
1.2036 31.0 558 1.3464 0.4723
1.2314 32.0 576 1.3276 0.5061
1.2201 33.0 594 1.3068 0.5027
1.1737 34.0 612 1.2978 0.5161
1.2102 35.0 630 1.2962 0.4961
1.156 36.0 648 1.2793 0.5172
1.1707 37.0 666 1.2715 0.5125
1.149 38.0 684 1.2728 0.4986
1.1685 39.0 702 1.2525 0.5101
1.1212 40.0 720 1.2446 0.5100
1.095 41.0 738 1.2365 0.5119
1.1166 42.0 756 1.2241 0.5294
1.0775 43.0 774 1.2175 0.5234
1.0768 44.0 792 1.2041 0.5165
1.0395 45.0 810 1.1995 0.5284
1.0857 46.0 828 1.2031 0.5316
1.0447 47.0 846 1.1954 0.5096
1.0504 48.0 864 1.1708 0.5349
1.0229 49.0 882 1.1656 0.5468
1.0715 50.0 900 1.1625 0.5505
1.0401 51.0 918 1.1619 0.5458
1.0477 52.0 936 1.1373 0.5608
1.009 53.0 954 1.1425 0.5740
1.0078 54.0 972 1.1397 0.5622
0.9709 55.0 990 1.1503 0.5813
0.9989 56.0 1008 1.1271 0.5761
0.9704 57.0 1026 1.1332 0.5691
0.9537 58.0 1044 1.1113 0.5910
0.9722 59.0 1062 1.1047 0.5832
0.9889 60.0 1080 1.1005 0.5815
0.9682 61.0 1098 1.0862 0.6137
0.9609 62.0 1116 1.0737 0.6148
0.9688 63.0 1134 1.0580 0.6238
0.9488 64.0 1152 1.0645 0.6253
0.926 65.0 1170 1.0576 0.6188
0.9689 66.0 1188 1.0438 0.6210
0.9445 67.0 1206 1.0409 0.6319
0.938 68.0 1224 1.0302 0.6397
0.9134 69.0 1242 1.0346 0.6337
0.9125 70.0 1260 1.0221 0.6575
0.8879 71.0 1278 1.0146 0.6633
0.9212 72.0 1296 1.0206 0.6384
0.9259 73.0 1314 1.0255 0.6213
0.9224 74.0 1332 1.0190 0.6417
0.9249 75.0 1350 1.0063 0.6371
0.8888 76.0 1368 0.9951 0.6458
0.8799 77.0 1386 1.0045 0.6436
0.9186 78.0 1404 0.9871 0.6449
0.9087 79.0 1422 1.0031 0.6611
0.914 80.0 1440 0.9893 0.6501
0.9012 81.0 1458 0.9876 0.6441
0.8748 82.0 1476 0.9873 0.6533
0.8736 83.0 1494 0.9951 0.6524
0.892 84.0 1512 1.0012 0.6563
0.8746 85.0 1530 0.9944 0.6684
0.8769 86.0 1548 0.9841 0.6558
0.8816 87.0 1566 0.9930 0.6551
0.8889 88.0 1584 0.9880 0.6497
0.8705 89.0 1602 0.9874 0.6564
0.8607 90.0 1620 0.9850 0.6471
0.86 91.0 1638 0.9851 0.6572
0.878 92.0 1656 0.9835 0.6553
0.8592 93.0 1674 0.9784 0.6577
0.8699 94.0 1692 0.9783 0.6568
0.8413 95.0 1710 0.9909 0.6519
0.8944 96.0 1728 0.9759 0.6581
0.8404 97.0 1746 0.9834 0.6640
0.8954 98.0 1764 0.9785 0.6582
0.8539 99.0 1782 0.9746 0.6528
0.8732 100.0 1800 0.9826 0.6620

Framework versions

  • Transformers 4.28.0.dev0
  • Pytorch 2.0.0+cu117
  • Datasets 2.19.0
  • Tokenizers 0.12.1
Downloads last month
4
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Evaluation results