segment_test

This model is a fine-tuned version of nvidia/mit-b0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1260
• Mean Iou: 0.8447
• Mean Accuracy: 0.9060
• Overall Accuracy: 0.9643
• Per Category Iou: [0.8581731584290526, 0.9316708017224354, 0.6429457606853357, 0.9461543989839276]
• Per Category Accuracy: [0.9425328146084153, 0.9716290009653072, 0.7414471782174548, 0.9684613946705924]

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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
1.0598	0.12	20	1.0566	0.4321	0.5515	0.8279	[0.006337257377741027, 0.7300264416674316, 0.23785776348718268, 0.7540601698014678]	[0.008091933477034156, 0.9660296098187956, 0.4518657444815921, 0.779822911077032]
0.7539	0.25	40	0.6128	0.4529	0.5206	0.8749	[0.00885438037222906, 0.7887343476637405, 0.18630294846181414, 0.827538656414587]	[0.009217881990081138, 0.9747549910409742, 0.23250084632472676, 0.8657848800171862]
0.8122	0.38	60	0.4928	0.4920	0.5507	0.8982	[0.02270113673871408, 0.8263637783329851, 0.2554015068180403, 0.8635887381305902]	[0.023653681019224364, 0.9630036206967801, 0.30407150146655104, 0.9122355221296055]
0.5643	0.5	80	0.4269	0.5063	0.5663	0.9129	[0.007592876375505398, 0.8532772644930355, 0.27466317854807065, 0.8898355351518092]	[0.007684489073480188, 0.9441805272844986, 0.3632055074118278, 0.9501073674832299]
0.4774	0.62	100	0.3648	0.5490	0.6034	0.9218	[0.054267747505144744, 0.8633878614911209, 0.37650159164096025, 0.9020084353341656]	[0.055975413139862956, 0.9428365409577416, 0.45329399467742054, 0.961363980883101]
0.6967	0.75	120	0.3155	0.6158	0.6710	0.9325	[0.19913477919641662, 0.8733583185910088, 0.47153849968591927, 0.9190236996330312]	[0.20509568371799589, 0.9502280021743253, 0.5638919570734214, 0.9648268960892941]
0.4543	0.88	140	0.2964	0.6272	0.7016	0.9310	[0.2268179599804783, 0.8694887954672872, 0.4961026114602021, 0.9165896188501754]	[0.2321184480311235, 0.9344494411072529, 0.6716976656263668, 0.9680582434884342]
0.2924	1.0	160	0.2722	0.6820	0.7592	0.9384	[0.40849985639476094, 0.8843766461027478, 0.5120505100078259, 0.9231923075849184]	[0.4548831992709812, 0.9611082383832005, 0.667739089824088, 0.9532322308907735]
0.3467	1.12	180	0.2517	0.6815	0.7430	0.9414	[0.40527347045732925, 0.8910632241964158, 0.5029203059015287, 0.9266932255262942]	[0.447198710197501, 0.9557005046935547, 0.6038508392823166, 0.9651288642296557]
0.5212	1.25	200	0.2396	0.6752	0.7408	0.9427	[0.3668495492266549, 0.8948194469733646, 0.5099936563424119, 0.9292469056689828]	[0.3789189141825701, 0.965308543339751, 0.6575510698285852, 0.9614078764674099]
0.3599	1.38	220	0.2108	0.6717	0.7162	0.9446	[0.3204229923658836, 0.891874970369172, 0.5386909667961237, 0.9357227289520356]	[0.32437612813907435, 0.9667832593900233, 0.605627503453252, 0.9680877242092517]
0.7198	1.5	240	0.2018	0.7692	0.8618	0.9527	[0.6678205541462868, 0.9118050318864169, 0.5610132418656683, 0.9363469668607539]	[0.8350594956451641, 0.9497174012785288, 0.6927260551877853, 0.9696466684460499]
0.3682	1.62	260	0.1931	0.7812	0.8479	0.9538	[0.6890403931363558, 0.9109837638365132, 0.5887742209415688, 0.93605613049878]	[0.7620327532726986, 0.9497927662356516, 0.7058595323345466, 0.9739980042388998]
0.3358	1.75	280	0.1947	0.7916	0.8837	0.9550	[0.7285153761152414, 0.9167544021441286, 0.5844378827851311, 0.9366799980170186]	[0.8390747945253492, 0.9631461386121828, 0.7712946420849488, 0.9611668227501886]
0.2334	1.88	300	0.2014	0.7843	0.8673	0.9522	[0.7115100993908304, 0.9111127309647635, 0.5831612483285399, 0.9312333415206432]	[0.850726390130207, 0.9751160540868734, 0.6929311499390893, 0.9502407282203037]
0.2996	2.0	320	0.1701	0.8027	0.8896	0.9568	[0.7526564111480359, 0.9206156591940219, 0.5997533073741746, 0.9377288799677301]	[0.8344833780207841, 0.964968540097336, 0.7965929561072522, 0.9622910891033862]
0.4622	2.12	340	0.1897	0.7614	0.8286	0.9509	[0.5700844122547072, 0.9050700733042791, 0.6361100555452245, 0.9344921716585569]	[0.6064108967281776, 0.975059629707727, 0.777640224269875, 0.9551298779200561]
0.338	2.25	360	0.1713	0.8093	0.8947	0.9579	[0.7434863865211815, 0.9179570569529378, 0.6345544665942896, 0.9411654222120547]	[0.9167652419255911, 0.9513808608505936, 0.7380297560360868, 0.9725677707832107]
0.1828	2.38	380	0.1613	0.8102	0.8854	0.9583	[0.7563041981028708, 0.923073161283024, 0.6226932638283688, 0.9387242789473116]	[0.8467351874244257, 0.9575329724998702, 0.7662636431252486, 0.9708901596701879]
0.1848	2.5	400	0.1523	0.8332	0.8882	0.9617	[0.8272877802882616, 0.9256725856927198, 0.6369852470922355, 0.9430524306296677]	[0.8886931987452464, 0.9707900525673887, 0.7260329485953481, 0.9672511041320438]
0.2053	2.62	420	0.1450	0.8382	0.9129	0.9626	[0.8318187789064231, 0.9285519737454524, 0.6480460535396757, 0.9442850494628351]	[0.9293894467518357, 0.9631609731907377, 0.7888883123172988, 0.970120219961517]
0.1596	2.75	440	0.1376	0.8327	0.8953	0.9624	[0.8243205167317957, 0.928415735271021, 0.6344793824796972, 0.9436346449664342]	[0.9292602036345793, 0.9591271598881472, 0.717386845766276, 0.9752671632199643]
0.137	2.88	460	0.1472	0.8260	0.9081	0.9616	[0.7858947726139123, 0.9279875195406946, 0.6464381724674384, 0.9437068326674424]	[0.9422546133221177, 0.9608167124242775, 0.7587838622553998, 0.9706571968763456]
0.1679	3.0	480	0.1471	0.8331	0.9126	0.9624	[0.8265760388185004, 0.9295336697547381, 0.6321998556258659, 0.9441039056088667]	[0.9403510155442231, 0.9637791247810575, 0.7769013889609604, 0.9693229385117723]
0.1736	3.12	500	0.1322	0.8402	0.8960	0.9634	[0.8539363341788129, 0.9291877911278092, 0.632669557547822, 0.9451061440414477]	[0.9248133641764366, 0.9646743651065282, 0.721404726074956, 0.9731695680841902]
0.234	3.25	520	0.1355	0.8448	0.9076	0.9632	[0.8526502893205608, 0.9289085404109714, 0.6534740415947964, 0.944330630190735]	[0.9257553055394915, 0.9662036512136275, 0.7685765188748946, 0.9699754203327694]
0.1183	3.38	540	0.1256	0.8402	0.8954	0.9637	[0.8579969655846054, 0.9294856863546612, 0.6276158375419669, 0.9457592749372996]	[0.9389994392163048, 0.9661690813475308, 0.7035145333106, 0.9727224283186038]
0.2085	3.5	560	0.1279	0.8480	0.9079	0.9642	[0.8653295827282605, 0.9310186709909459, 0.6502631789997088, 0.9455883129741319]	[0.9333061703730964, 0.9630336872086728, 0.7613438401150507, 0.9739094690773278]
0.3477	3.62	580	0.1419	0.8392	0.9182	0.9627	[0.8356838761929862, 0.9295750569194587, 0.6477355195833494, 0.9437915808167578]	[0.9515710705711231, 0.9679427406460035, 0.7873562792352684, 0.9660274217203163]
0.4279	3.75	600	0.1241	0.8453	0.8990	0.9639	[0.8543157504970579, 0.9297143070002641, 0.6517228913538011, 0.9456141839889123]	[0.9299589926922874, 0.955372819449404, 0.7305771563983385, 0.9801914665868114]
0.1622	3.88	620	0.1255	0.8438	0.9206	0.9638	[0.8381374979278546, 0.9308920069470188, 0.6604220534808366, 0.9458239210474509]	[0.9524867251984649, 0.95692011897332, 0.7977592780664754, 0.9753784831657644]
0.1982	4.0	640	0.1229	0.8484	0.9091	0.9647	[0.8568106859830862, 0.9316640237424656, 0.6584440596013207, 0.9466997405108393]	[0.9452293955803235, 0.9595748462766798, 0.754422510013813, 0.9770213126081696]
0.2273	4.12	660	0.1237	0.8444	0.8986	0.9640	[0.8592317269149002, 0.9306231098333563, 0.6421928363458029, 0.9456102319021257]	[0.9358165536337031, 0.9691394407999653, 0.7183357178686949, 0.970921035377795]
0.2371	4.25	680	0.1216	0.8460	0.9096	0.9645	[0.8583380957371707, 0.9321090417297068, 0.6472432759450742, 0.9464390728899327]	[0.9508328514098453, 0.961341618092252, 0.7507901089967407, 0.9753970829896241]
0.2474	4.38	700	0.1283	0.8411	0.9093	0.9635	[0.8497493402399663, 0.9309167522466655, 0.6387399097291454, 0.9448643127623003]	[0.9486861013266039, 0.9710814460747176, 0.7504342819583336, 0.9668204252105732]
0.2696	4.5	720	0.1273	0.8430	0.9049	0.9633	[0.8574838830140721, 0.929602014003094, 0.6403298755318946, 0.9444960387864841]	[0.9355734013283564, 0.9743545898715855, 0.7448102379346218, 0.9649164542411783]
0.1506	4.62	740	0.1255	0.8479	0.9182	0.9646	[0.8525455460809918, 0.9327217587877741, 0.6600164613831294, 0.9464177410883513]	[0.9496433766188248, 0.9639090597950073, 0.7866569802639545, 0.972458775815393]
0.2033	4.75	760	0.1209	0.8501	0.9107	0.9650	[0.8658943552605436, 0.9327154832962604, 0.6546122598692601, 0.9470460950988947]	[0.935571210767047, 0.9689885784341254, 0.76733359526157, 0.9708881136895634]
0.1636	4.88	780	0.1203	0.8483	0.9157	0.9648	[0.8518065574480739, 0.9329685900413733, 0.661433964801116, 0.9468031896359194]	[0.9475010076582023, 0.9648201943117868, 0.7779515729284813, 0.9726341721543897]
0.1047	5.0	800	0.1260	0.8447	0.9060	0.9643	[0.8581731584290526, 0.9316708017224354, 0.6429457606853357, 0.9461543989839276]	[0.9425328146084153, 0.9716290009653072, 0.7414471782174548, 0.9684613946705924]

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.0
• Datasets 2.10.1
• Tokenizers 0.13.2