vit-base-patch16-224-in21k-bridgedefectVIT

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: 0.1799
• Accuracy: {'accuracy': 0.9705510388437217}
• F1: {'f1': 0.9705092081728205}
• Precision: {'precision': 0.9710523804561741}
• Recall: {'recall': 0.9704181656558507}

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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
0.37	1.0	8302	0.3462	{'accuracy': 0.8933453778982234}	{'f1': 0.8942100052466936}	{'precision': 0.8984250247518094}	{'recall': 0.8931370564158605}
0.2375	2.0	16605	0.3353	{'accuracy': 0.9053297199638664}	{'f1': 0.9062005892826234}	{'precision': 0.912717242831991}	{'recall': 0.9052684275828231}
0.5678	3.0	24907	0.3114	{'accuracy': 0.9118940078289671}	{'f1': 0.9116597109413729}	{'precision': 0.9165908158739848}	{'recall': 0.9116030141797212}
0.09	4.0	33210	0.2768	{'accuracy': 0.9270099367660344}	{'f1': 0.9272025877193879}	{'precision': 0.9305221603080029}	{'recall': 0.9267551810236085}
0.266	5.0	41512	0.2595	{'accuracy': 0.9312857573020175}	{'f1': 0.9313123811138734}	{'precision': 0.9327488749607135}	{'recall': 0.931043574955592}
0.2037	6.0	49815	0.2123	{'accuracy': 0.9431496537187594}	{'f1': 0.9428749572352995}	{'precision': 0.9435965528419799}	{'recall': 0.9429052318485974}
0.1487	7.0	58117	0.2282	{'accuracy': 0.9430292080698585}	{'f1': 0.9430188942480495}	{'precision': 0.9444609819488103}	{'recall': 0.9428880066548226}
0.1405	8.0	66420	0.2440	{'accuracy': 0.9454381210478772}	{'f1': 0.9455191951029847}	{'precision': 0.9467893516678145}	{'recall': 0.9453224042508239}
0.09	9.0	74722	0.2480	{'accuracy': 0.9436314363143632}	{'f1': 0.9433683232067358}	{'precision': 0.9452971145459653}	{'recall': 0.9433746555197686}
0.2275	10.0	83025	0.2473	{'accuracy': 0.946582354712436}	{'f1': 0.9462472081330006}	{'precision': 0.9479482237973264}	{'recall': 0.9463251646491099}
0.0114	11.0	91327	0.1953	{'accuracy': 0.9551942186088528}	{'f1': 0.954959353992539}	{'precision': 0.9555671952457011}	{'recall': 0.9550120730050532}
0.0778	12.0	99630	0.2246	{'accuracy': 0.948509485094851}	{'f1': 0.9485863094568601}	{'precision': 0.9496017185087666}	{'recall': 0.9484435235390778}
0.1031	13.0	107932	0.2435	{'accuracy': 0.9443541102077687}	{'f1': 0.9443461050911817}	{'precision': 0.9453218450441414}	{'recall': 0.9442028500529185}
0.1419	14.0	116235	0.1751	{'accuracy': 0.9580849141824752}	{'f1': 0.9580811670883926}	{'precision': 0.9586631550970829}	{'recall': 0.9580178560027687}
0.0993	15.0	124537	0.2099	{'accuracy': 0.9542908762420957}	{'f1': 0.9541061721417268}	{'precision': 0.9541191566948424}	{'recall': 0.9541611121516007}
0.0696	16.0	132840	0.2240	{'accuracy': 0.955736224028907}	{'f1': 0.9555782982813351}	{'precision': 0.9563626555520048}	{'recall': 0.9555607789866469}
0.1697	17.0	141142	0.1904	{'accuracy': 0.9579644685335742}	{'f1': 0.9577653922157884}	{'precision': 0.9581933285912818}	{'recall': 0.9578259452834421}
0.0429	18.0	149445	0.2102	{'accuracy': 0.9558566696778079}	{'f1': 0.955829019244906}	{'precision': 0.9570787144559411}	{'recall': 0.955662074541215}
0.0062	19.0	157747	0.1768	{'accuracy': 0.9601927130382415}	{'f1': 0.9601350969183112}	{'precision': 0.9605649770988711}	{'recall': 0.960090994011799}
0.005	20.0	166050	0.1779	{'accuracy': 0.9624209575429088}	{'f1': 0.9622479573311764}	{'precision': 0.9626782993390144}	{'recall': 0.9622658509657924}
0.1395	21.0	174352	0.1801	{'accuracy': 0.961035832580548}	{'f1': 0.9609739947935761}	{'precision': 0.9615134912739316}	{'recall': 0.9609000684385473}
0.0966	22.0	182655	0.1854	{'accuracy': 0.9594098163203855}	{'f1': 0.959384693086552}	{'precision': 0.9602665108685822}	{'recall': 0.9592591268355116}
0.0077	23.0	190957	0.2190	{'accuracy': 0.9573020174646191}	{'f1': 0.9572877808970253}	{'precision': 0.9580176848865115}	{'recall': 0.9571782999468976}
0.1032	24.0	199260	0.2281	{'accuracy': 0.9570009033423668}	{'f1': 0.9568818981129438}	{'precision': 0.9577859752909083}	{'recall': 0.95679636210611}
0.1106	25.0	207562	0.2017	{'accuracy': 0.9615778380006023}	{'f1': 0.9615258017857322}	{'precision': 0.9623198062794668}	{'recall': 0.9614196936259853}
0.0833	26.0	215865	0.2074	{'accuracy': 0.9618789521228546}	{'f1': 0.9618001985746503}	{'precision': 0.9625802607483476}	{'recall': 0.9617264541173526}
0.0257	27.0	224167	0.1716	{'accuracy': 0.9648900933453779}	{'f1': 0.9648046336171575}	{'precision': 0.9653533590655595}	{'recall': 0.9648070647916974}
0.002	28.0	232470	0.2144	{'accuracy': 0.9635049683830171}	{'f1': 0.9634863498105041}	{'precision': 0.9646616314066687}	{'recall': 0.9633283402670114}
0.016	29.0	240772	0.2237	{'accuracy': 0.959349593495935}	{'f1': 0.9594342688149864}	{'precision': 0.9608554784443832}	{'recall': 0.9591930193477335}
0.0575	30.0	249075	0.1847	{'accuracy': 0.9651912074676302}	{'f1': 0.9652324025756626}	{'precision': 0.9661899074568192}	{'recall': 0.9650558808909672}
0.0997	31.0	257377	0.1798	{'accuracy': 0.9686841312857573}	{'f1': 0.9686428828918746}	{'precision': 0.9691104091550086}	{'recall': 0.9685623791125}
0.0017	32.0	265680	0.1985	{'accuracy': 0.9627822944896116}	{'f1': 0.9626870784433683}	{'precision': 0.963172343077798}	{'recall': 0.962659195203449}
0.0538	33.0	273982	0.1605	{'accuracy': 0.9710328214393255}	{'f1': 0.9710267090566379}	{'precision': 0.9715030346291925}	{'recall': 0.9709339306149106}
0.0023	34.0	282285	0.1832	{'accuracy': 0.9674194519722975}	{'f1': 0.9673811237591747}	{'precision': 0.9679330625290327}	{'recall': 0.9672934059576415}
0.0459	35.0	290587	0.1877	{'accuracy': 0.9657332128876844}	{'f1': 0.965749942670487}	{'precision': 0.9664774134203846}	{'recall': 0.9656335047526519}
0.0193	36.0	298890	0.1633	{'accuracy': 0.9677205660945498}	{'f1': 0.9677329659674949}	{'precision': 0.9684419822552822}	{'recall': 0.9675975315398574}
0.0707	37.0	307192	0.1787	{'accuracy': 0.9685636856368564}	{'f1': 0.9684895304986225}	{'precision': 0.9689001010469502}	{'recall': 0.9684451099576021}
0.0985	38.0	315495	0.2076	{'accuracy': 0.9629629629629629}	{'f1': 0.9630524772042474}	{'precision': 0.9642571257654206}	{'recall': 0.9628345133405821}
0.0788	39.0	323797	0.1794	{'accuracy': 0.9702499247214694}	{'f1': 0.9701536210820301}	{'precision': 0.9706833500680011}	{'recall': 0.9700913059580385}
0.0008	40.0	332100	0.1618	{'accuracy': 0.9733212887684433}	{'f1': 0.9732738808256685}	{'precision': 0.9736678524998652}	{'recall': 0.9731998786471756}
0.074	41.0	340402	0.1991	{'accuracy': 0.9668172237277929}	{'f1': 0.9666853676025186}	{'precision': 0.9673504006462602}	{'recall': 0.9666339730453138}
0.028	42.0	348705	0.1556	{'accuracy': 0.9742246311352002}	{'f1': 0.9741506224327396}	{'precision': 0.9743929114728255}	{'recall': 0.9741060958660924}
0.1092	43.0	357007	0.1567	{'accuracy': 0.9740439626618489}	{'f1': 0.9739721593463402}	{'precision': 0.9742787951493688}	{'recall': 0.9739217266482031}
0.0008	44.0	365310	0.1697	{'accuracy': 0.9707919301415237}	{'f1': 0.9707068184898958}	{'precision': 0.9712158191257935}	{'recall': 0.9706396165347172}
0.1728	45.0	373612	0.1791	{'accuracy': 0.9701294790725685}	{'f1': 0.9700180755443455}	{'precision': 0.9704271475318083}	{'recall': 0.9699790872810246}
0.0004	46.0	381915	0.2024	{'accuracy': 0.9672387834989461}	{'f1': 0.9672031338307139}	{'precision': 0.9680962843155184}	{'recall': 0.9670672659468575}
0.0044	47.0	390217	0.1708	{'accuracy': 0.9721168322794339}	{'f1': 0.9720140881144397}	{'precision': 0.9723799188733908}	{'recall': 0.9719693947081535}
0.089	48.0	398520	0.1975	{'accuracy': 0.9686841312857573}	{'f1': 0.9686510789801565}	{'precision': 0.969349692339074}	{'recall': 0.9685439142771983}
0.0774	49.0	406822	0.1778	{'accuracy': 0.9709123757904246}	{'f1': 0.9708794409655027}	{'precision': 0.9714408230271825}	{'recall': 0.9707829629677185}
0.0012	50.0	415100	0.1799	{'accuracy': 0.9705510388437217}	{'f1': 0.9705092081728205}	{'precision': 0.9710523804561741}	{'recall': 0.9704181656558507}

Framework versions

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