Hugging Face's logo

Ventalik
/
segformer_rust

Transformers
PyTorch
segformer
Generated from Trainer
Inference Endpoints
Model card Files Community
1
segformer_rust / README.md
Ventalik's picture
Ventalik
End of training
9bcc543
preview code
|
raw
history blame
No virus
7.19 kB
metadata
license: other
base_model: nvidia/mit-b3
tags:
  - generated_from_trainer
model-index:
  - name: segformer_rust
    results: []

segformer_rust

This model is a fine-tuned version of nvidia/mit-b3 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1438
  • Mean Iou: 0.7943
  • Mean Accuracy: 0.8621
  • Overall Accuracy: 0.9463
  • Per Category Iou: [0.9404316292856895, 0.6481512759943856]
  • Per Category Accuracy: [0.9766097421311045, 0.7476136089750051]

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 30

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Per Category Iou Per Category Accuracy
0.2683 1.0 514 0.1919 0.7319 0.8203 0.9254 [0.9180806580950376, 0.5456807971830461] [0.9632417779264837, 0.6772761097220785]
0.2221 2.0 1028 0.1848 0.7247 0.7906 0.9285 [0.9222502676042136, 0.5272319457425635] [0.9780221080448669, 0.6032531950921799]
0.2051 3.0 1542 0.1798 0.7523 0.8476 0.9301 [0.9226057238240142, 0.5819673887381934] [0.9598172268335441, 0.7353887410800054]
0.1996 4.0 2056 0.1662 0.7530 0.8151 0.9366 [0.9305887422755489, 0.5753549591140691] [0.9802446445708194, 0.6499256145141094]
0.1878 5.0 2570 0.1613 0.7695 0.8443 0.9386 [0.9321977696827667, 0.6067612442121683] [0.972541727740175, 0.7160658070940397]
0.1848 6.0 3084 0.1579 0.7659 0.8299 0.9395 [0.9335431699668292, 0.5982432303010027] [0.9789606327074537, 0.6808199676860336]
0.1739 7.0 3598 0.1543 0.7799 0.8516 0.9419 [0.9357233119518715, 0.6240100443072397] [0.9744037993799354, 0.7287987165292924]
0.1742 8.0 4112 0.1607 0.7737 0.8667 0.9369 [0.9296752366973121, 0.6176824708083143] [0.9620707480253384, 0.7713870499643393]
0.1631 9.0 4626 0.1553 0.7803 0.8636 0.9402 [0.9335684843016983, 0.626952054351963] [0.9678138198093251, 0.7593407854928076]
0.1631 10.0 5140 0.1564 0.7679 0.8406 0.9386 [0.9322118932460649, 0.6035637849191774] [0.9738013828280326, 0.7073043940670594]
0.1577 11.0 5654 0.1499 0.7836 0.8520 0.9434 [0.9373956576514434, 0.6297874494960901] [0.9763014923228973, 0.7277053858903674]
0.1522 12.0 6168 0.1515 0.7781 0.8454 0.9422 [0.9361248378060324, 0.6200926437502748] [0.9769566337905029, 0.7138379976069665]
0.1486 13.0 6682 0.1531 0.7766 0.8485 0.9410 [0.9347900233191081, 0.618411826576312] [0.9743083731504567, 0.7226475641364578]
0.146 14.0 7196 0.1568 0.7835 0.8667 0.9412 [0.9345511743761084, 0.6325140238903502] [0.9679584694275379, 0.7654769806378111]
0.1453 15.0 7710 0.1485 0.7837 0.8473 0.9442 [0.938357945982732, 0.6290898844632448] [0.9790292010707323, 0.7156414504087399]
0.1439 16.0 8224 0.1492 0.7896 0.8623 0.9444 [0.9382510027549076, 0.6409378390750833] [0.973914066583286, 0.7506284726295933]
0.1385 17.0 8738 0.1494 0.7790 0.8369 0.9439 [0.938230866841315, 0.6196920559815114] [0.9823649345853678, 0.6913889357135595]
0.1375 18.0 9252 0.1515 0.7847 0.8549 0.9435 [0.9373782356302373, 0.6321131725666339] [0.975314658131637, 0.7344851844673589]
0.1353 19.0 9766 0.1450 0.7929 0.8610 0.9459 [0.9399812074698386, 0.6457742806194801] [0.9764190916833184, 0.7456795800307741]
0.1317 20.0 10280 0.1453 0.7906 0.8584 0.9453 [0.9394006951099269, 0.6417152113828789] [0.9765905703089925, 0.7402706100619615]
0.1292 21.0 10794 0.1565 0.7788 0.8416 0.9431 [0.9372341895291594, 0.6204583221235739] [0.9795794314033889, 0.7035825637871398]
0.1284 22.0 11308 0.1487 0.7879 0.8532 0.9450 [0.9391595949869292, 0.6366277680932542] [0.9780565365202, 0.7282789363894756]
0.1279 23.0 11822 0.1461 0.7927 0.8629 0.9456 [0.9395641382795358, 0.6458609695526066] [0.9752807298744423, 0.7506032283294566]
0.1262 24.0 12336 0.1436 0.7934 0.8633 0.9458 [0.9398055111519206, 0.6469847970011983] [0.9754426177792707, 0.7512221554580607]
0.1223 25.0 12850 0.1465 0.7945 0.8622 0.9464 [0.9404797675363489, 0.6484203969264617] [0.9766238155760367, 0.7478641586538628]
0.1234 26.0 13364 0.1435 0.7925 0.8570 0.9463 [0.9405453729701521, 0.6444525239879496] [0.9784625404961454, 0.735513574144182]
0.1223 27.0 13878 0.1464 0.7937 0.8618 0.9461 [0.9402290147962161, 0.6472276210560935] [0.976462283595653, 0.7471743581526247]
0.1196 28.0 14392 0.1450 0.7929 0.8589 0.9462 [0.9403789254222912, 0.645380249186489] [0.9776359685007636, 0.7400721898628863]
0.1199 29.0 14906 0.1451 0.7919 0.8563 0.9462 [0.9403753807784087, 0.6433302809940281] [0.9784796056303284, 0.7341607320998506]
0.1208 30.0 15420 0.1438 0.7943 0.8621 0.9463 [0.9404316292856895, 0.6481512759943856] [0.9766097421311045, 0.7476136089750051]

Framework versions

  • Transformers 4.33.1
  • Pytorch 2.0.1
  • Datasets 2.14.5
  • Tokenizers 0.13.3