Edit model card

image_segmentation_text_v2

This model is a fine-tuned version of nvidia/mit-b0 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0615
  • Mean Iou: 0.8781
  • Mean Accuracy: 0.9110
  • Overall Accuracy: 0.9734
  • Per Category Iou: [0.9705763603242805, 0.7855978863350824]
  • Per Category Accuracy: [0.9925963343496583, 0.8293473752101665]

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: 50

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Per Category Iou Per Category Accuracy
0.5172 0.4 20 0.5984 0.6201 0.8312 0.8580 [0.8433386612833417, 0.39680175287609587] [0.866190330235608, 0.7961823484817947]
0.4767 0.8 40 0.4035 0.5258 0.5853 0.8882 [0.8856626480452995, 0.1659390336502284] [0.9810900280003672, 0.18954193123051938]
0.3439 1.2 60 0.3474 0.6995 0.8573 0.9074 [0.8979547121624778, 0.5009503846168084] [0.9228071003413048, 0.7918197268517736]
0.2842 1.6 80 0.2828 0.6923 0.7722 0.9230 [0.9174404161237479, 0.4671543716020125] [0.9692542242141893, 0.5751911816528029]
0.2941 2.0 100 0.2412 0.7274 0.8226 0.9294 [0.9232295355579201, 0.5316079622856148] [0.9621363189327741, 0.6830129526301212]
0.2159 2.4 120 0.2387 0.7471 0.8877 0.9274 [0.9195102054074465, 0.5746316429439954] [0.9395897739623381, 0.8357430695431796]
0.2013 2.8 140 0.2059 0.7660 0.8994 0.9345 [0.9272560031451658, 0.6047195723903955] [0.9453276278068868, 0.8534035600552902]
0.1822 3.2 160 0.1598 0.8016 0.8669 0.9533 [0.9487517424447746, 0.6545065297960031] [0.9797635674372861, 0.7541327875396765]
0.1261 3.6 180 0.1533 0.7870 0.8553 0.9494 [0.9446523952139291, 0.6294161895000621] [0.9782770257479946, 0.7322611643453532]
0.1751 4.0 200 0.1434 0.8060 0.8609 0.9556 [0.9513646199038472, 0.6606627045566984] [0.9846122449614638, 0.7371309163585958]
0.1556 4.4 220 0.1340 0.8102 0.8645 0.9566 [0.9524442279212987, 0.6679515135667692] [0.984852971709032, 0.7440538962567276]
0.1068 4.8 240 0.1253 0.7898 0.8358 0.9531 [0.9490085335911248, 0.6306787839687829] [0.9890518166701155, 0.6826157004965977]
0.1215 5.2 260 0.1073 0.8497 0.9091 0.9652 [0.9614007926531548, 0.7380488065311127] [0.9823778512721868, 0.8358783043120386]
0.091 5.6 280 0.1077 0.8325 0.8819 0.9622 [0.9583632540158915, 0.7067172285496849] [0.9867697098157909, 0.7770472788553937]
0.1009 6.0 300 0.1064 0.8238 0.8680 0.9608 [0.9569876540030186, 0.6906903674483074] [0.9892049828448698, 0.7467735454785426]
0.0984 6.4 320 0.1060 0.8311 0.8761 0.9623 [0.9585729019064059, 0.7037087870923661] [0.9887282833788597, 0.7633723130216779]
0.0813 6.8 340 0.0987 0.8419 0.8883 0.9645 [0.9608922897234371, 0.722840581644497] [0.9878806832041188, 0.7887346836871759]
0.0625 7.2 360 0.0902 0.8475 0.8891 0.9662 [0.9627445497647863, 0.7323412821849365] [0.9898383923982556, 0.7883172763717552]
0.0734 7.6 380 0.0927 0.8469 0.8959 0.9654 [0.961803716924761, 0.7319708674782419] [0.9867425686313254, 0.8049635061174228]
0.0733 8.0 400 0.0876 0.8711 0.9307 0.9701 [0.9666965487625444, 0.7754727726647843] [0.9822345389032576, 0.8790988163056722]
0.0584 8.4 420 0.0918 0.8607 0.9121 0.9683 [0.964859785245781, 0.7564785981949649] [0.9855508631847466, 0.838696128774708]
0.0639 8.8 440 0.0858 0.8735 0.9349 0.9706 [0.967139403611975, 0.7798182886949773] [0.9814883292042429, 0.8884020581691553]
0.0859 9.2 460 0.0863 0.8442 0.8790 0.9661 [0.9627025051107989, 0.7256751249820441] [0.9927421533884905, 0.765291606472024]
0.0676 9.6 480 0.0796 0.8692 0.9225 0.9701 [0.9667776147303454, 0.7716995072886178] [0.9847239214274811, 0.8603714014873224]
0.0856 10.0 500 0.0789 0.8721 0.9231 0.9709 [0.9676681841843127, 0.7764766898351453] [0.9856164687866871, 0.8604845305728103]
0.0732 10.4 520 0.0796 0.8611 0.9052 0.9690 [0.9656431512846645, 0.7564856330988124] [0.9885161672429414, 0.8218307926968024]
0.0642 10.8 540 0.0881 0.8378 0.8718 0.9648 [0.9613570133699026, 0.7141516019397048] [0.9932708878642084, 0.7502987518090901]
0.0806 11.2 560 0.0792 0.8605 0.9039 0.9689 [0.9655732989225007, 0.7554214468864908] [0.9888003717477902, 0.8190597802695073]
0.0642 11.6 580 0.0768 0.8663 0.9076 0.9703 [0.9671300467068968, 0.7654769403248215] [0.9895412223594264, 0.8256966866181296]
0.0503 12.0 600 0.0796 0.8637 0.9056 0.9697 [0.9664822125443101, 0.7610012493038518] [0.9893804497380703, 0.8217891819986919]
0.0711 12.4 620 0.0799 0.8524 0.8876 0.9678 [0.9645090425676253, 0.7402596444414385] [0.9923599294467503, 0.7828006080363261]
0.0622 12.8 640 0.0739 0.8678 0.9150 0.9702 [0.9669781140184328, 0.7687031945706295] [0.9871926455342293, 0.84275642267627]
0.0551 13.2 660 0.0715 0.8741 0.9178 0.9719 [0.9687595252839901, 0.7794271755462193] [0.988435608058477, 0.8472263218873572]
0.0717 13.6 680 0.0757 0.8598 0.8941 0.9694 [0.9662689349888198, 0.7532485600087165] [0.9925298643787859, 0.7955731418547709]
0.0635 14.0 700 0.0718 0.8732 0.9174 0.9716 [0.96850746670701, 0.7778616932414459] [0.9882812317672178, 0.8464279166173623]
0.0523 14.4 720 0.0663 0.8876 0.9358 0.9746 [0.9716791978883015, 0.8036108039561688] [0.9865335642235624, 0.8850107862731508]
0.0577 14.8 740 0.0676 0.8833 0.9366 0.9733 [0.9702350309780267, 0.796362403107551] [0.9846131957672891, 0.8885314414335925]
0.0435 15.2 760 0.0711 0.8671 0.9039 0.9708 [0.9677187286152693, 0.7665006963915808] [0.9913234646603627, 0.8165254286877156]
0.0673 15.6 780 0.0717 0.8637 0.8964 0.9704 [0.9673187596918544, 0.7601149728604968] [0.9930802945146975, 0.7996782972902333]
0.0563 16.0 800 0.0654 0.8845 0.9251 0.9744 [0.971468043944435, 0.7974893910508356] [0.989447006145836, 0.860792709805691]
0.0586 16.4 820 0.0693 0.8732 0.9129 0.9719 [0.9688913807063626, 0.7775406555703152] [0.9900519779613582, 0.8357222641941243]
0.0564 16.8 840 0.0698 0.8698 0.9045 0.9715 [0.9685318541511667, 0.7711066182518113] [0.9921181654564637, 0.8168225550789108]
0.0478 17.2 860 0.0682 0.8736 0.9110 0.9722 [0.9691646685069771, 0.7780200639749164] [0.9909173841360986, 0.831173044589764]
0.0485 17.6 880 0.0646 0.8790 0.9192 0.9732 [0.9702169077246766, 0.7878410180791867] [0.989739854340005, 0.8486430361245876]
0.0566 18.0 900 0.0693 0.8667 0.9029 0.9708 [0.9676668469869958, 0.7656600486037874] [0.991562981291426, 0.8142504938019565]
0.0455 18.4 920 0.0727 0.8621 0.8955 0.9700 [0.9668821926221384, 0.7572945442521496] [0.9928309240778092, 0.7981315496214076]
0.0451 18.8 940 0.0743 0.8616 0.8957 0.9698 [0.9667066236459168, 0.75648894444205] [0.9925660814370375, 0.7987895187852797]
0.0479 19.2 960 0.0648 0.8791 0.9170 0.9733 [0.9704226039041611, 0.787839253664714] [0.9906320559516382, 0.8433539262944503]
0.0429 19.6 980 0.0643 0.8814 0.9217 0.9737 [0.9707538355760997, 0.7919781103864696] [0.9896213493594248, 0.8538053633589195]
0.0412 20.0 1000 0.0685 0.8710 0.9082 0.9716 [0.9685757534109218, 0.7733473306325775] [0.9910794533108528, 0.8252383187717562]
0.0517 20.4 1020 0.0700 0.8681 0.9049 0.9710 [0.9679606643038668, 0.768307799914675] [0.9913238968448287, 0.8184479729738515]
0.0622 20.8 1040 0.0705 0.8661 0.9029 0.9706 [0.967473334999663, 0.7646563357741486] [0.9913285644370616, 0.8145313660142023]
0.0501 21.2 1060 0.0659 0.8735 0.9092 0.9723 [0.9692782576787035, 0.7776627515629433] [0.9915888259224934, 0.826863736666697]
0.0331 21.6 1080 0.0656 0.8740 0.9095 0.9724 [0.969431404124596, 0.7786407327387083] [0.9916812269613265, 0.8273624148768648]
0.0409 22.0 1100 0.0627 0.8812 0.9180 0.9739 [0.9709853246351019, 0.791378914530232] [0.991000449990466, 0.8449500866672821]
0.0507 22.4 1120 0.0670 0.8712 0.9054 0.9719 [0.96890284080009, 0.7735870738116131] [0.992284124291412, 0.8184843823346982]
0.055 22.8 1140 0.0650 0.8740 0.9106 0.9723 [0.969333155377141, 0.7786920795104981] [0.9912426461652186, 0.8299858393592993]
0.0621 23.2 1160 0.0608 0.8851 0.9241 0.9746 [0.97174575429152, 0.7984980757048741] [0.9900816258157266, 0.8580698097480862]
0.0475 23.6 1180 0.0648 0.8757 0.9099 0.9728 [0.969923983536794, 0.781504588462575] [0.9921419356020943, 0.8276972509632227]
0.0719 24.0 1200 0.0597 0.8883 0.9290 0.9752 [0.9723810750370141, 0.8042819715081054] [0.9893649775341871, 0.8686207223877259]
0.0432 24.4 1220 0.0671 0.8745 0.9111 0.9724 [0.9694436234993048, 0.7795285436234692] [0.991234693971044, 0.8309240305682591]
0.0487 24.8 1240 0.0627 0.8794 0.9158 0.9735 [0.9705926889014477, 0.7881517515303139] [0.9911871536797828, 0.840397616227132]
0.0431 25.2 1260 0.0703 0.8607 0.8911 0.9699 [0.9668504181402485, 0.7545441416676626] [0.9940878893787556, 0.7880988202066751]
0.0485 25.6 1280 0.0686 0.8663 0.8992 0.9709 [0.9678797771258784, 0.7647121997104036] [0.992901802330235, 0.8055415047208637]
0.0401 26.0 1300 0.0645 0.8741 0.9094 0.9724 [0.9694702128324741, 0.7787894408819918] [0.9917601438448201, 0.827058136646932]
0.0358 26.4 1320 0.0650 0.8752 0.9080 0.9728 [0.9699118666008384, 0.7804262038251969] [0.9926968604564524, 0.823297569805197]
0.0371 26.8 1340 0.0624 0.8794 0.9138 0.9736 [0.9707671601979381, 0.7880723266892008] [0.9919919675923885, 0.8355421678913648]
0.0373 27.2 1360 0.0593 0.8905 0.9303 0.9757 [0.972963140756407, 0.808064312770677] [0.9896636170002007, 0.8708904559362212]
0.0459 27.6 1380 0.0650 0.8748 0.9074 0.9727 [0.9698368752489902, 0.7797103933881729] [0.9927813957380043, 0.8220466481932505]
0.0448 28.0 1400 0.0585 0.8913 0.9294 0.9760 [0.9732524861024707, 0.8092622781494344] [0.9902627975438784, 0.8685342501557151]
0.0466 28.4 1420 0.0591 0.8868 0.9227 0.9751 [0.9723752980279239, 0.8013039527210407] [0.9912211233788112, 0.8542169191699186]
0.046 28.8 1440 0.0624 0.8782 0.9129 0.9734 [0.9704629995935647, 0.7859491844731833] [0.9919060493205455, 0.8337990697408304]
0.044 29.2 1460 0.0646 0.8738 0.9079 0.9724 [0.9694836939653589, 0.7780494172207805] [0.9922285453690829, 0.8235309798149104]
0.0288 29.6 1480 0.0639 0.8745 0.9087 0.9726 [0.9696471298812925, 0.7793704984065971] [0.9921753866797636, 0.8252409194403881]
0.0613 30.0 1500 0.0599 0.8839 0.9219 0.9744 [0.9715329244431512, 0.7962728238571309] [0.9904880520875633, 0.853244269101586]
0.044 30.4 1520 0.0628 0.8757 0.9097 0.9729 [0.9699453251794836, 0.7814895244296873] [0.9922406465341312, 0.8271010476793583]
0.0293 30.8 1540 0.0617 0.8780 0.9125 0.9733 [0.9704425104125867, 0.7856417917220885] [0.9919949928836506, 0.8329473507638814]
0.0428 31.2 1560 0.0599 0.8818 0.9155 0.9742 [0.9713541972985056, 0.7921493091394499] [0.9921727071360743, 0.8387878023439826]
0.0645 31.6 1580 0.0627 0.8744 0.9059 0.9727 [0.9698458799171074, 0.7789683055241065] [0.9932316455146946, 0.8186261187751371]
0.0567 32.0 1600 0.0633 0.8749 0.9090 0.9727 [0.9697435930361873, 0.7800611650843169] [0.992193538427336, 0.8258657300792034]
0.0285 32.4 1620 0.0620 0.8768 0.9115 0.9731 [0.9701471869845076, 0.783517826129869] [0.991947279718603, 0.8309766941080552]
0.0571 32.8 1640 0.0601 0.8811 0.9153 0.9740 [0.9711643268487495, 0.7909989321972878] [0.9920145276215143, 0.8385108311346847]
0.0401 33.2 1660 0.0642 0.8734 0.9060 0.9725 [0.9695272707444462, 0.7772926276022758] [0.9928285902816927, 0.8192216718918434]
0.0512 33.6 1680 0.0663 0.8706 0.9027 0.9719 [0.9689150151187212, 0.7721857742809729] [0.9930920499321729, 0.8123090946682392]
0.0376 34.0 1700 0.0628 0.8780 0.9122 0.9733 [0.9704377847240172, 0.7854762211546131] [0.9920699336700569, 0.8323290417966459]
0.028 34.4 1720 0.0642 0.8737 0.9078 0.9724 [0.9694875787740898, 0.7780070979332923] [0.9922682199030628, 0.8232540086056125]
0.0529 34.8 1740 0.0597 0.8831 0.9182 0.9744 [0.971576972958672, 0.7947159780070965] [0.9916307478156965, 0.8447452840125196]
0.0405 35.2 1760 0.0633 0.8755 0.9079 0.9729 [0.9700091754702059, 0.7809148433321386] [0.992833430747712, 0.8230108460885294]
0.0586 35.6 1780 0.0653 0.8714 0.9059 0.9719 [0.9689169040586796, 0.7739549474059] [0.9921412441069487, 0.8197053962573778]
0.0572 36.0 1800 0.0628 0.8759 0.9105 0.9729 [0.9699524292504794, 0.7819443592605768] [0.9920137496894754, 0.8289169645515863]
0.0506 36.4 1820 0.0599 0.8820 0.9172 0.9741 [0.9712925656410623, 0.7926478079614093] [0.9916029151360853, 0.8427128614766857]
0.0326 36.8 1840 0.0664 0.8692 0.9008 0.9716 [0.9686386462229684, 0.7696646394310533] [0.9933210212622656, 0.8083313719957401]
0.0332 37.2 1860 0.0611 0.8792 0.9132 0.9736 [0.9707461298544804, 0.7876790150283312] [0.9921293158156869, 0.8343114014613157]
0.0601 37.6 1880 0.0606 0.8808 0.9150 0.9739 [0.9711010689625178, 0.7905405861230448] [0.9920064025535533, 0.8380732686373666]
0.0368 38.0 1900 0.0640 0.8732 0.9050 0.9725 [0.9695594225767187, 0.7769055994221483] [0.9931751157865403, 0.8167887463866961]
0.0422 38.4 1920 0.0617 0.8774 0.9118 0.9732 [0.9702923463415809, 0.7844662873966847] [0.9920254186700578, 0.8315215341864394]
0.0432 38.8 1940 0.0633 0.8759 0.9104 0.9729 [0.9699529639553718, 0.7819260584157408] [0.992026369475883, 0.8288233404808376]
0.0586 39.2 1960 0.0611 0.8793 0.9135 0.9736 [0.9707487378178947, 0.7878378709986136] [0.9920483244467564, 0.834959618117818]
0.0352 39.6 1980 0.0647 0.8741 0.9070 0.9726 [0.9696673702028153, 0.7785742945526019] [0.9927034296603359, 0.8213054576331574]
0.0361 40.0 2000 0.0630 0.8764 0.9101 0.9730 [0.9701097833950302, 0.782591672915463] [0.9923145500778191, 0.8278324857320817]
0.0421 40.4 2020 0.0610 0.8786 0.9117 0.9735 [0.970677588650168, 0.7865422237867158] [0.9924901034079128, 0.8309727931051073]
0.0458 40.8 2040 0.0625 0.8759 0.9084 0.9730 [0.9701118739649276, 0.7817402653153365] [0.9928041286409164, 0.8240530640427653]
0.0414 41.2 2060 0.0619 0.8781 0.9117 0.9734 [0.9705327749561811, 0.7857585839593991] [0.9923163652525764, 0.831171744255448]
0.05 41.6 2080 0.0605 0.8803 0.9136 0.9739 [0.9710563114216053, 0.7895525940174909] [0.9923900094855846, 0.8347476636243178]
0.0306 42.0 2100 0.0601 0.8816 0.9169 0.9740 [0.9712027944676178, 0.792013158538375] [0.9915832939613285, 0.8421550180551419]
0.0523 42.4 2120 0.0613 0.8790 0.9123 0.9736 [0.9707498261404447, 0.7872513405027708] [0.9923932940875263, 0.8322952331044311]
0.0347 42.8 2140 0.0591 0.8831 0.9180 0.9744 [0.9715892664031848, 0.7947023514847699] [0.9916940196215205, 0.8443525830491019]
0.0433 43.2 2160 0.0607 0.8802 0.9135 0.9739 [0.9710164270074672, 0.7893017221767921] [0.9923620903690803, 0.8346481880491474]
0.0408 43.6 2180 0.0595 0.8826 0.9176 0.9743 [0.9714433752474874, 0.7937025271738631] [0.9916428489807448, 0.8435957884772175]
0.0318 44.0 2200 0.0626 0.8772 0.9107 0.9732 [0.970326566740187, 0.7840985118681589] [0.9923819708545168, 0.8290287933027581]
0.048 44.4 2220 0.0598 0.8812 0.9156 0.9740 [0.9711679522425044, 0.7911830955700955] [0.9919216943982151, 0.8392585233663575]
0.0332 44.8 2240 0.0612 0.8785 0.9111 0.9736 [0.9707002963862882, 0.7863211248436672] [0.9927155308253842, 0.8294058902543844]
0.0609 45.2 2260 0.0597 0.8811 0.9152 0.9740 [0.9711846631400414, 0.7910596755855982] [0.992068291369086, 0.8382553154416]
0.0351 45.6 2280 0.0610 0.8785 0.9113 0.9735 [0.9706826467363787, 0.7863583642701348] [0.9926178571360657, 0.830022898887304]
0.0444 46.0 2300 0.0592 0.8825 0.9168 0.9743 [0.9714948390728185, 0.7935759738110126] [0.9919480576506418, 0.8416394354988668]
0.0508 46.4 2320 0.0607 0.8794 0.9122 0.9737 [0.9708919841673713, 0.7879353992275995] [0.9926144860972308, 0.8317074819936205]
0.038 46.8 2340 0.0601 0.8816 0.9159 0.9741 [0.9712634066834908, 0.7918441714173262] [0.9919517744370495, 0.8397806075942125]
0.035 47.2 2360 0.0606 0.8800 0.9134 0.9738 [0.9709804766649791, 0.7890313829990002] [0.992362781864226, 0.83435821349669]
0.033 47.6 2380 0.0607 0.8800 0.9134 0.9738 [0.9709604358552277, 0.7889657724192867] [0.9923135128351007, 0.8345812208318759]
0.0411 48.0 2400 0.0614 0.8779 0.9107 0.9734 [0.9705508928119433, 0.7853294217376947] [0.9926405900389779, 0.8288025351317824]
0.0378 48.4 2420 0.0596 0.8816 0.9162 0.9741 [0.9712621476572101, 0.7919987894404409] [0.9918544464953037, 0.8405243988229374]
0.0376 48.8 2440 0.0596 0.8816 0.9160 0.9741 [0.9712701752402172, 0.7919569993962999] [0.991914779446759, 0.8401206450178341]
0.0405 49.2 2460 0.0599 0.8808 0.9148 0.9739 [0.9711090571855628, 0.7904560916024725] [0.9920912835826778, 0.8374790158549763]
0.0357 49.6 2480 0.0601 0.8805 0.9142 0.9739 [0.9710507537711823, 0.7898657116838633] [0.9921822151943266, 0.8363132661407248]
0.0508 50.0 2500 0.0615 0.8781 0.9110 0.9734 [0.9705763603242805, 0.7855978863350824] [0.9925963343496583, 0.8293473752101665]

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.1.0+cu121
  • Datasets 2.15.0
  • Tokenizers 0.15.0
Downloads last month
22
Safetensors
Model size
3.72M params
Tensor type
F32
·
Inference API
Unable to determine this model’s pipeline type. Check the docs .

Model tree for Caraaaaa/image_segmentation_text

Base model

nvidia/mit-b0
Finetuned
(314)
this model

Dataset used to train Caraaaaa/image_segmentation_text