as-cle-bert's picture
Update README.md
02e9036 verified
metadata
license: other
base_model: nvidia/segformer-b0-finetuned-ade-512-512
tags:
  - generated_from_trainer
model-index:
  - name: segformer-breastcancer
    results: []
datasets:
  - as-cle-bert/breastcancer-semantic-segmentation
pipeline_tag: image-segmentation

segformer-breastcancer

This model is a fine-tuned version of nvidia/segformer-b0-finetuned-ade-512-512 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1986
  • Mean Iou: 0.4951
  • Mean Accuracy: 0.5647
  • Overall Accuracy: 0.5716
  • Per Category Iou: [0.41886373003284666, 0.5713219432574086]
  • Per Category Accuracy: [0.542773911636187, 0.5866474640793707]

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: 2
  • eval_batch_size: 2
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 50

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Per Category Iou Per Category Accuracy
0.9179 1.25 20 0.8275 0.1056 0.2990 0.2215 [0.15928433223106872, 0.05189369644942194] [0.5449101796407185, 0.053152424747755486]
0.7951 2.5 40 0.7554 0.3808 0.6154 0.6539 [0.2962250026735109, 0.46535774064135604] [0.4931218643793494, 0.7375983290380178]
0.6317 3.75 60 0.5784 0.2076 0.3576 0.3005 [0.24602488191071786, 0.16910477266308951] [0.5386308464152776, 0.17651220374955784]
0.5525 5.0 80 0.4935 0.3310 0.4279 0.3908 [0.3572223576675606, 0.30487703968490387] [0.5453956950962939, 0.31031549514039786]
0.4365 6.25 100 0.4277 0.4259 0.5007 0.5093 [0.3753112405986087, 0.4765198093920762] [0.473248098397799, 0.528071150639244]
0.3658 7.5 120 0.3757 0.3739 0.4207 0.4501 [0.2934911929427469, 0.45430117531467024] [0.32736688784592977, 0.5140397864133273]
0.357 8.75 140 0.3155 0.4305 0.5273 0.5652 [0.31276016750127367, 0.5482260296446353] [0.40734746722770676, 0.6473124799973049]
0.2889 10.0 160 0.3121 0.4761 0.5439 0.5495 [0.39972203089638886, 0.5525428502787649] [0.5259588930247613, 0.56174305590648]
0.2536 11.25 180 0.2611 0.4607 0.5411 0.5586 [0.37248963582652733, 0.5489196143472734] [0.4856772940605276, 0.5965098455370829]
0.3375 12.5 200 0.2522 0.3905 0.4676 0.4535 [0.3615823724169426, 0.4193968866718472] [0.512558666450882, 0.4227348526959422]
0.1835 13.75 220 0.2393 0.4343 0.4809 0.5004 [0.3816968232451229, 0.4869246466631396] [0.41924259588930246, 0.5425994239223811]
0.1878 15.0 240 0.2364 0.3883 0.4769 0.4591 [0.3594858252766199, 0.4170536161683648] [0.5331607056157954, 0.42058719490626106]
0.1804 16.25 260 0.2388 0.3503 0.4221 0.3934 [0.3722368961671656, 0.3283766624340039] [0.5131736526946108, 0.3310593427324945]
0.2296 17.5 280 0.2108 0.3845 0.4523 0.4383 [0.36382381172455475, 0.4051134890024848] [0.4968765172357987, 0.40781915879192143]
0.1752 18.75 300 0.2065 0.4408 0.5307 0.5278 [0.37362255868123995, 0.5080655748465653] [0.539941738145331, 0.5215102666464534]
0.1404 20.0 320 0.2025 0.4192 0.5049 0.4948 [0.37603680369849973, 0.4624047452321127] [0.5370771969574365, 0.4727289571647548]
0.1044 21.25 340 0.1993 0.4134 0.5006 0.4938 [0.36164057945015027, 0.46514651056315] [0.5219938501375627, 0.4791635083463877]
0.1047 22.5 360 0.1995 0.4409 0.5612 0.5654 [0.35316826827766823, 0.5286988461568266] [0.5477909046771322, 0.5746205804571564]
0.0969 23.75 380 0.1934 0.4208 0.5256 0.5171 [0.3610564616784075, 0.480532337904731] [0.5524356692021363, 0.49872824970101237]
0.1198 25.0 400 0.2100 0.4047 0.4892 0.4726 [0.377810637529348, 0.43159533203482664] [0.5416895937854022, 0.4366988394225748]
0.116 26.25 420 0.2038 0.4208 0.5123 0.5040 [0.3659432240473206, 0.47558361909786334] [0.5386632141123159, 0.48590968046220967]
0.0803 27.5 440 0.2035 0.4643 0.5486 0.5520 [0.3885018236229309, 0.5400125204269953] [0.537854021686357, 0.5594101099937676]
0.1031 28.75 460 0.2068 0.4193 0.5268 0.5199 [0.3565531095848628, 0.48207738324971056] [0.5486324648001295, 0.5049522461973824]
0.0652 30.0 480 0.1906 0.4799 0.5572 0.5719 [0.39256244632789455, 0.5671483599490623] [0.5104709499919081, 0.6039045260835144]
0.0865 31.25 500 0.1946 0.4660 0.5319 0.5360 [0.4022848534304187, 0.5297039831736081] [0.5185952419485353, 0.5451176579581248]
0.0781 32.5 520 0.2018 0.4170 0.4977 0.4881 [0.37508619500758517, 0.4588260589120619] [0.5281922641204079, 0.46729664628497314]
0.0922 33.75 540 0.1932 0.4649 0.5558 0.5608 [0.39512968947922955, 0.5346638407173079] [0.5401683120245995, 0.571521215490087]
0.0802 35.0 560 0.2029 0.4519 0.5364 0.5344 [0.3877223005943433, 0.5161263869184783] [0.5426606246965529, 0.5300756312429464]
0.0737 36.25 580 0.1983 0.4605 0.5598 0.5666 [0.3930664524057094, 0.5280028151990147] [0.5383719048389707, 0.5812993750736941]
0.0766 37.5 600 0.2097 0.4902 0.5645 0.5701 [0.41298901286924217, 0.5674679408239331] [0.5468846091600582, 0.5821500160021561]
0.0663 38.75 620 0.1926 0.5041 0.5653 0.5781 [0.42229021548076295, 0.5859655697770101] [0.5249069428710147, 0.6057405629390065]
0.0572 40.0 640 0.1944 0.4884 0.5550 0.5643 [0.41379925802215733, 0.5630840363400389] [0.525295355235475, 0.5846429834756683]
0.1065 41.25 660 0.1949 0.4713 0.5603 0.5687 [0.4052270716602772, 0.537297205601135] [0.5337271403139666, 0.5868664409520441]
0.0881 42.5 680 0.1945 0.4557 0.5355 0.5362 [0.38861418270649184, 0.5228113541121006] [0.5329341317365269, 0.5379672208465983]
0.0616 43.75 700 0.2055 0.4851 0.5479 0.5493 [0.4288067420034476, 0.5413945423770796] [0.543486000971031, 0.5522512506948305]
0.135 45.0 720 0.2017 0.4950 0.5702 0.5770 [0.4186215922560253, 0.5714192766576933] [0.5487133840427254, 0.5917428874627318]
0.0683 46.25 740 0.1986 0.4880 0.5579 0.5633 [0.41617258731503165, 0.5599071727881785] [0.5407347467227707, 0.5750585342025031]
0.0962 47.5 760 0.2010 0.4907 0.5660 0.5730 [0.41037067786677084, 0.571094427269902] [0.543955332578087, 0.5881213468762106]
0.0534 48.75 780 0.2061 0.4941 0.5671 0.5740 [0.4158937943809818, 0.5723742349360128] [0.5450234665803528, 0.5891404315528829]
0.069 50.0 800 0.1986 0.4951 0.5647 0.5716 [0.41886373003284666, 0.5713219432574086] [0.542773911636187, 0.5866474640793707]

Framework versions

  • Transformers 4.38.2
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2