vit-base-patch16-224-in21k-crack-detectorVITmain50epochs

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0670
• Accuracy: {'accuracy': 0.9845817447858264}
• F1: {'f1': 0.983869840492846}
• Precision: {'precision': 0.9840346899184906}
• Recall: {'recall': 0.9837409101507315}

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.5622	1.0	1114	0.4109	{'accuracy': 0.8736263736263736}	{'f1': 0.8653360894930927}	{'precision': 0.8669392826942903}	{'recall': 0.8641606609285752}
0.3696	2.0	2229	0.2493	{'accuracy': 0.9249271137026239}	{'f1': 0.9207602119769538}	{'precision': 0.9199949527238288}	{'recall': 0.9219224853720074}
0.321	3.0	3344	0.1849	{'accuracy': 0.9410742318905584}	{'f1': 0.9378351696354423}	{'precision': 0.9376663800198108}	{'recall': 0.9381803360170522}
0.3403	4.0	4459	0.1828	{'accuracy': 0.9382148463781117}	{'f1': 0.9358183038090213}	{'precision': 0.9351989305945833}	{'recall': 0.9374254756058774}
0.2399	5.0	5573	0.1283	{'accuracy': 0.9539694998878673}	{'f1': 0.9516195088904471}	{'precision': 0.9525864748624857}	{'recall': 0.9506985041313745}
0.2241	6.0	6688	0.1133	{'accuracy': 0.9605292666517157}	{'f1': 0.9584812990028835}	{'precision': 0.9604748305069212}	{'recall': 0.9567171466267748}
0.2375	7.0	7803	0.1138	{'accuracy': 0.9638932496075353}	{'f1': 0.9615520166695122}	{'precision': 0.9628392910079963}	{'recall': 0.9605232709871921}
0.1946	8.0	8918	0.0935	{'accuracy': 0.9682103610675039}	{'f1': 0.9665726403969513}	{'precision': 0.9677788948564787}	{'recall': 0.9654971437363139}
0.1513	9.0	10032	0.0908	{'accuracy': 0.967593630858937}	{'f1': 0.9657913168767667}	{'precision': 0.9657840712769861}	{'recall': 0.9659100270116727}
0.174	10.0	11147	0.0851	{'accuracy': 0.968490692980489}	{'f1': 0.9669266120574112}	{'precision': 0.9663934578407292}	{'recall': 0.967857117571674}
0.1376	11.0	12262	0.0848	{'accuracy': 0.9692756223368468}	{'f1': 0.9676560659554989}	{'precision': 0.9695619317426214}	{'recall': 0.9663476308224139}
0.1681	12.0	13377	0.0789	{'accuracy': 0.9743776631531733}	{'f1': 0.9731609997444961}	{'precision': 0.973142077383281}	{'recall': 0.9733127731609976}
0.0998	13.0	14491	0.0750	{'accuracy': 0.9749943933617403}	{'f1': 0.9737724540493713}	{'precision': 0.9739505389554447}	{'recall': 0.9737043141281871}
0.0968	14.0	15606	0.0837	{'accuracy': 0.9737609329446064}	{'f1': 0.9722837146803432}	{'precision': 0.9729901044455692}	{'recall': 0.9716272189580044}
0.0841	15.0	16721	0.0689	{'accuracy': 0.9758914554832923}	{'f1': 0.974522946671967}	{'precision': 0.9749603470712005}	{'recall': 0.974152667413215}
0.095	16.0	17836	0.0705	{'accuracy': 0.9762278537788742}	{'f1': 0.9749967013469043}	{'precision': 0.9757537211246345}	{'recall': 0.9743092871293193}
0.1055	17.0	18950	0.0709	{'accuracy': 0.9753307916573223}	{'f1': 0.9739943140941046}	{'precision': 0.974174353159638}	{'recall': 0.9738453333404872}
0.0644	18.0	20065	0.0738	{'accuracy': 0.9766763848396501}	{'f1': 0.9755727076296268}	{'precision': 0.9750313848950637}	{'recall': 0.9764061086777133}
0.0839	19.0	21180	0.0691	{'accuracy': 0.9770127831352321}	{'f1': 0.9758154254345139}	{'precision': 0.9761822643915987}	{'recall': 0.9754580155091578}
0.0752	20.0	22295	0.0737	{'accuracy': 0.9765642520744562}	{'f1': 0.9753527993174993}	{'precision': 0.9756389600577889}	{'recall': 0.9751662665128649}
0.0777	21.0	23409	0.0594	{'accuracy': 0.9806010316214397}	{'f1': 0.9797190165517261}	{'precision': 0.9797694892510824}	{'recall': 0.9796726937896714}
0.0753	22.0	24524	0.0673	{'accuracy': 0.9798721686476789}	{'f1': 0.9791539833000008}	{'precision': 0.9788167338624194}	{'recall': 0.9796378458227846}
0.0564	23.0	25639	0.0670	{'accuracy': 0.9795357703520969}	{'f1': 0.9785354369309488}	{'precision': 0.9785832048371994}	{'recall': 0.9785660480900121}
0.0721	24.0	26754	0.0685	{'accuracy': 0.9800964341780668}	{'f1': 0.9791442297494849}	{'precision': 0.9800176266790614}	{'recall': 0.9783165451081763}
0.0637	25.0	27868	0.0683	{'accuracy': 0.9804888988562458}	{'f1': 0.9793615860016096}	{'precision': 0.9795645667425841}	{'recall': 0.9792228423789527}
0.0845	26.0	28983	0.0687	{'accuracy': 0.9789190401435299}	{'f1': 0.9779974196750578}	{'precision': 0.9777924373303344}	{'recall': 0.9782512107344294}
0.0443	27.0	30098	0.0660	{'accuracy': 0.9821148239515587}	{'f1': 0.9812042927695707}	{'precision': 0.9814603709525461}	{'recall': 0.9810002705393822}
0.0544	28.0	31213	0.0778	{'accuracy': 0.9791993720565149}	{'f1': 0.9782906130895705}	{'precision': 0.9787040955747366}	{'recall': 0.9779528847328706}
0.0537	29.0	32327	0.0674	{'accuracy': 0.9818344920385736}	{'f1': 0.9811984625176332}	{'precision': 0.9815390282130447}	{'recall': 0.9809323144878648}
0.0493	30.0	33442	0.0701	{'accuracy': 0.9814980937429917}	{'f1': 0.9805602239528695}	{'precision': 0.9811340066266928}	{'recall': 0.9800627824611866}
0.0522	31.0	34557	0.0710	{'accuracy': 0.9814980937429917}	{'f1': 0.9808148178357844}	{'precision': 0.9806080192879051}	{'recall': 0.9810839837288577}
0.0159	32.0	35672	0.0724	{'accuracy': 0.9820026911863646}	{'f1': 0.981223779316131}	{'precision': 0.9815852582883647}	{'recall': 0.9809265201908598}
0.0469	33.0	36786	0.0681	{'accuracy': 0.9827315541601256}	{'f1': 0.982079515043356}	{'precision': 0.9821689199854957}	{'recall': 0.9820640757052289}
0.0469	34.0	37901	0.0678	{'accuracy': 0.9817223592733797}	{'f1': 0.9809328125490557}	{'precision': 0.9810701495429422}	{'recall': 0.9808658173842284}
0.0385	35.0	39016	0.0634	{'accuracy': 0.9842453464902444}	{'f1': 0.9836492800003742}	{'precision': 0.9834341260971065}	{'recall': 0.9838961344763829}
0.0386	36.0	40131	0.0687	{'accuracy': 0.9839089481946625}	{'f1': 0.983239373896663}	{'precision': 0.983372610840443}	{'recall': 0.9831603310143421}
0.031	37.0	41245	0.0643	{'accuracy': 0.9837968154294685}	{'f1': 0.9830873488066041}	{'precision': 0.983437856778449}	{'recall': 0.982767386441964}
0.0188	38.0	42360	0.0672	{'accuracy': 0.9840771473424534}	{'f1': 0.9834705260789471}	{'precision': 0.9841321671664716}	{'recall': 0.9828412386527916}
0.0406	39.0	43475	0.0575	{'accuracy': 0.9851424086117964}	{'f1': 0.9846872838275066}	{'precision': 0.9845262374841439}	{'recall': 0.9848621233267102}
0.0725	40.0	44590	0.0654	{'accuracy': 0.9843574792554385}	{'f1': 0.9836921118378804}	{'precision': 0.9838071847209022}	{'recall': 0.9836248084545489}
0.0456	41.0	45704	0.0652	{'accuracy': 0.9837407490468715}	{'f1': 0.9831247046792615}	{'precision': 0.9830251647671387}	{'recall': 0.9832768782292008}
0.0272	42.0	46819	0.0670	{'accuracy': 0.9835725498990805}	{'f1': 0.9828974576921549}	{'precision': 0.9829831457743452}	{'recall': 0.9828424029347932}
0.021	43.0	47934	0.0633	{'accuracy': 0.9847499439336174}	{'f1': 0.9841868836767669}	{'precision': 0.9842200567065219}	{'recall': 0.9841697876663609}
0.0231	44.0	49049	0.0724	{'accuracy': 0.9825072886297376}	{'f1': 0.9819634359798524}	{'precision': 0.9814536283421093}	{'recall': 0.9825728366747504}
0.0419	45.0	50163	0.0652	{'accuracy': 0.9844135456380354}	{'f1': 0.9837880332204787}	{'precision': 0.9837369807084215}	{'recall': 0.98386986331642}
0.0257	46.0	51278	0.0673	{'accuracy': 0.9842453464902444}	{'f1': 0.98363705023352}	{'precision': 0.9835068164678353}	{'recall': 0.983839536345849}
0.0238	47.0	52393	0.0670	{'accuracy': 0.9850302758466024}	{'f1': 0.9844204182477418}	{'precision': 0.9846537574771261}	{'recall': 0.9842341338822049}
0.0185	48.0	53508	0.0675	{'accuracy': 0.9844696120206324}	{'f1': 0.9837444887203743}	{'precision': 0.9840638645916341}	{'recall': 0.983456407378361}
0.0196	49.0	54622	0.0664	{'accuracy': 0.9844135456380354}	{'f1': 0.9837492030151856}	{'precision': 0.9839059161676366}	{'recall': 0.9836254408912738}
0.0352	49.97	55700	0.0670	{'accuracy': 0.9845817447858264}	{'f1': 0.983869840492846}	{'precision': 0.9840346899184906}	{'recall': 0.9837409101507315}

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0
• Datasets 2.17.1
• Tokenizers 0.15.2