Hugging Face's logo

glacierscopessegmentation
/
glacier_segmentation_transformer

Transformers
PyTorch
Safetensors
segformer
Generated from Trainer
Model card Files Community
glacier_segmentation_transformer / README.md
aashraychegu's picture
aashraychegu
End of training
29db57d verified
preview code
|
raw
history blame
13 kB
metadata
library_name: transformers
tags:
  - generated_from_trainer
model-index:
  - name: glacier_segmentation_transformer
    results: []

glacier_segmentation_transformer

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0152
  • Mean Iou: 0.9578
  • Mean Accuracy: 0.9770
  • Overall Accuracy: 0.9815
  • Per Category Iou: [0.966974556454048, 0.9306800202753192, 0.9758789229664083]
  • Per Category Accuracy: [0.986415203459849, 0.9545545504757966, 0.9898833477760938]

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: 0.00012
  • train_batch_size: 400
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Per Category Iou Per Category Accuracy
0.0646 1.0 703 0.0566 0.9247 0.9600 0.9667 [0.9494255778418639, 0.867692450590415, 0.9569388095269572] [0.9761190606426763, 0.9255943024845551, 0.9782301302123408]
0.0599 2.0 1406 0.0497 0.9282 0.9616 0.9684 [0.9477319173718886, 0.8753546288784326, 0.9613964795143571] [0.9711497669822643, 0.9303257017214881, 0.983177895106294]
0.0551 3.0 2109 0.0487 0.9297 0.9630 0.9689 [0.9520168030120617, 0.8775963388947718, 0.9593654968038946] [0.9792121369623757, 0.9314351084836408, 0.9782160997896858]
0.0538 4.0 2812 0.0461 0.9301 0.9606 0.9697 [0.9526700363985638, 0.8742364766071521, 0.9633119672224237] [0.9796255350132945, 0.9152524234283989, 0.986801690344372]
0.0537 5.0 3515 0.0459 0.9293 0.9605 0.9693 [0.9480194887278883, 0.8755894234719767, 0.9644158890544811] [0.9695323626038072, 0.9222969866729543, 0.989726930656922]
0.0502 6.0 4218 0.0419 0.9343 0.9634 0.9714 [0.9527888538779837, 0.8847465897421206, 0.9654703601812225] [0.976308799435177, 0.925932784344352, 0.9880231202290629]
0.0485 7.0 4921 0.0426 0.9324 0.9618 0.9707 [0.952703644674004, 0.8794287150835857, 0.9650298528805881] [0.9762879243141335, 0.9201165442085169, 0.9890906398334286]
0.0467 8.0 5624 0.0391 0.9369 0.9657 0.9723 [0.9570939850601026, 0.8892999282991234, 0.9641725721587178] [0.9836122270915137, 0.9306427121583465, 0.9828948498133379]
0.0465 9.0 6327 0.0365 0.9392 0.9662 0.9736 [0.9570166835679954, 0.8930106582106385, 0.9674431118321617] [0.9799191676774507, 0.9305790764906415, 0.9880572447369236]
0.0441 10.0 7030 0.0368 0.9393 0.9671 0.9733 [0.9557842284941621, 0.896113739486814, 0.966114092503835] [0.9781441914622425, 0.9371105515852832, 0.9859413896265214]
0.0449 11.0 7733 0.0363 0.9395 0.9662 0.9737 [0.9587446620984498, 0.8925112162386646, 0.9671193682825042] [0.9841777180627991, 0.9274434801225682, 0.9869006864166631]
0.0429 12.0 8436 0.0362 0.9387 0.9663 0.9733 [0.9568234173066003, 0.8919248505170725, 0.9672752936824205] [0.9807715727354733, 0.9313716624751807, 0.986681375204542]
0.043 13.0 9139 0.0359 0.9399 0.9674 0.9736 [0.9583618298582752, 0.8954311403447485, 0.9658975869823546] [0.9824077197610704, 0.9351835437823924, 0.9845412435692529]
0.0403 14.0 9842 0.0327 0.9431 0.9685 0.9752 [0.9605210947016356, 0.8999327500995064, 0.9687008045143891] [0.9845226714091516, 0.9340792478377873, 0.9869475769904131]
0.0396 15.0 10545 0.0317 0.9430 0.9687 0.9751 [0.9582689044073551, 0.9018446354914778, 0.9689720395673985] [0.9803209527379552, 0.9375974567773319, 0.9880329157127937]
0.0398 16.0 11248 0.0311 0.9441 0.9687 0.9758 [0.9600781473937521, 0.9018120203453935, 0.9705198767201947] [0.9833703518813173, 0.9337154414777602, 0.9891511845852599]
0.0394 17.0 11951 0.0315 0.9428 0.9681 0.9752 [0.9599952844492292, 0.8990866834010299, 0.9692707279290462] [0.9834365741883817, 0.932572065746635, 0.9882736728656242]
0.0373 18.0 12654 0.0298 0.9453 0.9697 0.9762 [0.9604100374735833, 0.9056672192236825, 0.96995081765577] [0.9824364503507406, 0.9378770943608746, 0.9886761889355877]
0.0367 19.0 13357 0.0286 0.9465 0.9703 0.9767 [0.9607981425237361, 0.9074436356815473, 0.9711230723862715] [0.9831492425992236, 0.939067046409727, 0.9888305367096034]
0.0351 20.0 14060 0.0278 0.9471 0.9708 0.9770 [0.9618333883616247, 0.9086763944768684, 0.9708953906361056] [0.9835691087261089, 0.9400169996572258, 0.9886674084372189]
0.0354 21.0 14763 0.0285 0.9464 0.9701 0.9767 [0.9602311140889335, 0.9075971976817253, 0.971288380236899] [0.9805902546456458, 0.9392485003967921, 0.9904217581362069]
0.0346 22.0 15466 0.0272 0.9474 0.9707 0.9772 [0.9633102052165451, 0.9075619288178078, 0.9712912214581088] [0.9877988385981887, 0.9366896178642719, 0.9877408755495605]
0.0338 23.0 16169 0.0274 0.9475 0.9710 0.9771 [0.9621579382676179, 0.9097149647578355, 0.9706881239127263] [0.9833567220330544, 0.9412283931828203, 0.9885273185825774]
0.033 24.0 16872 0.0248 0.9496 0.9719 0.9781 [0.9633388370586602, 0.9126597100292737, 0.9727672608399773] [0.9847982679687262, 0.94123947327555, 0.9897510890585125]
0.0328 25.0 17575 0.0258 0.9484 0.9711 0.9777 [0.962009361478448, 0.9105945715118131, 0.972718363850045] [0.9830181147234991, 0.9396734668722746, 0.990675360103791]
0.0317 26.0 18278 0.0253 0.9492 0.9716 0.9779 [0.9639558319000139, 0.911781219914515, 0.9717356738019733] [0.9860431702641673, 0.939863355704706, 0.9889956162038476]
0.0314 27.0 18981 0.0240 0.9508 0.9735 0.9783 [0.9632096299431748, 0.9178493526029053, 0.9714077953769692] [0.9846306889201024, 0.9490954287160887, 0.9868323630270145]
0.031 28.0 19684 0.0225 0.9519 0.9738 0.9789 [0.9639432214860649, 0.9191291970986417, 0.9727323390586772] [0.9846959927218408, 0.9485333186423948, 0.9882958581706704]
0.0304 29.0 20387 0.0222 0.9520 0.9738 0.9790 [0.9646276170673505, 0.9186786670273801, 0.9728235086526744] [0.9854814689302406, 0.9476905128862128, 0.988319968447955]
0.03 30.0 21090 0.0222 0.9516 0.9735 0.9788 [0.9640884759202785, 0.9181744197938652, 0.9725126150057635] [0.9859090299478983, 0.9467068901317071, 0.9879970937294139]
0.0291 31.0 21793 0.0222 0.9520 0.9737 0.9790 [0.9644441535927152, 0.9183966340639266, 0.973123149831471] [0.9853206084590358, 0.9467745885000796, 0.988864109975414]
0.0288 32.0 22496 0.0216 0.9527 0.9744 0.9792 [0.9635312975815179, 0.9213747265447779, 0.973258211843832] [0.9833080284045524, 0.9510300129067109, 0.9887883360682158]
0.0274 33.0 23199 0.0215 0.9527 0.9740 0.9793 [0.9640157857518139, 0.9206728124821452, 0.9735005954174129] [0.9843878695121234, 0.9483662887400095, 0.9893775350708613]
0.0278 34.0 23902 0.0201 0.9538 0.9749 0.9797 [0.9645863263073253, 0.9230425214866308, 0.9736614810798143] [0.9850184523223797, 0.9508885071458663, 0.9886674084372189]
0.0277 35.0 24605 0.0199 0.9542 0.9753 0.9798 [0.9648387629037706, 0.9242740259952547, 0.9734575868219909] [0.9851679759970793, 0.9527511805364656, 0.9880083198175126]
0.027 36.0 25308 0.0190 0.9546 0.9751 0.9801 [0.9650972384872185, 0.9240644176546186, 0.9747245960463098] [0.9849486973025233, 0.9502703278101333, 0.9899684840482691]
0.0266 37.0 26011 0.0190 0.9546 0.9748 0.9802 [0.9660706980566042, 0.9226321694052272, 0.9749641496350198] [0.9865475517272653, 0.9474771761818693, 0.9902573130078139]
0.0256 38.0 26714 0.0186 0.9554 0.9757 0.9804 [0.9655268292420411, 0.9260322783164191, 0.9746364243430061] [0.9852960464704602, 0.9524953701433246, 0.9893096972992093]
0.0262 39.0 27417 0.0180 0.9555 0.9757 0.9805 [0.966309913000785, 0.9256650052387354, 0.9744492221886044] [0.9861509436967797, 0.9520694054612898, 0.9890100753703193]
0.0248 40.0 28120 0.0177 0.9557 0.9756 0.9806 [0.9660911749634843, 0.9258013861826082, 0.9751373606294378] [0.9859783931577905, 0.9509224960789698, 0.9899530798954259]
0.0249 41.0 28823 0.0178 0.9555 0.9754 0.9806 [0.9665350924463278, 0.9247239384112762, 0.9751124707269705] [0.986670425901285, 0.9496893616146678, 0.9899208453603083]
0.0241 42.0 29526 0.0173 0.9561 0.9759 0.9807 [0.9669016375167447, 0.9264132616994011, 0.9748969691253521] [0.9869660625770613, 0.9512582628169447, 0.9894373142087342]
0.0239 43.0 30229 0.0170 0.9566 0.9765 0.9808 [0.9657028012078982, 0.9293126752941933, 0.9747772529312919] [0.9851113112838957, 0.9551385811653776, 0.9891693536982271]
0.0234 44.0 30932 0.0166 0.9563 0.9758 0.9810 [0.9678707868291188, 0.9255875185979531, 0.975463092811191] [0.9880463982644232, 0.9494462284447401, 0.9899507568039292]
0.0228 45.0 31635 0.0163 0.9569 0.9764 0.9811 [0.9667985174141136, 0.9285307472949479, 0.9753903785250766] [0.9863655913260211, 0.9531677121665817, 0.9897135346001182]
0.0236 46.0 32338 0.0160 0.9572 0.9767 0.9812 [0.9669272848057375, 0.9294111193088093, 0.9753600048069734] [0.98654358224271, 0.9541473420948784, 0.989392186734557]
0.0226 47.0 33041 0.0158 0.9572 0.9765 0.9812 [0.9663055417479156, 0.929422808459136, 0.9759668808926588] [0.9857908703396774, 0.9534334846611139, 0.9903217602033646]
0.0223 48.0 33744 0.0154 0.9579 0.9771 0.9814 [0.9667858095653383, 0.9314465218695431, 0.9755411515813254] [0.9861092962245191, 0.9556930349948158, 0.9895239860837632]
0.0217 49.0 34447 0.0153 0.9578 0.9769 0.9814 [0.9666844542687175, 0.930702020324814, 0.9759177123723752] [0.9859492257963569, 0.95489544799545, 0.9899713627567528]
0.022 50.0 35150 0.0152 0.9578 0.9770 0.9815 [0.966974556454048, 0.9306800202753192, 0.9758789229664083] [0.986415203459849, 0.9545545504757966, 0.9898833477760938]

Framework versions

  • Transformers 4.45.2
  • Pytorch 2.4.1+cu121
  • Datasets 3.0.1
  • Tokenizers 0.20.0