From 76c9b7f05eb3e4c06e49dece77983d1879e0f61c Mon Sep 17 00:00:00 2001 From: colin Date: Mon, 23 Sep 2019 10:02:01 +0800 Subject: [PATCH] Refine line display visdom --- FilterEvaluator/Evaluator.py | 47 ++++++++++-------------------- FilterEvaluator/image/0-1024.png | Bin 413 -> 409 bytes FilterEvaluator/visdom.server.log | Bin 3918716 -> 3946415 bytes tools/WebVisual.py | 35 +++++++--------------- tools/utils.py | 2 +- 5 files changed, 27 insertions(+), 57 deletions(-) diff --git a/FilterEvaluator/Evaluator.py b/FilterEvaluator/Evaluator.py index a93c1c8..5f82c15 100644 --- a/FilterEvaluator/Evaluator.py +++ b/FilterEvaluator/Evaluator.py @@ -28,10 +28,8 @@ from tools import utils, Train, Loader, WebVisual import EvaluatorUnsuper - batchsize = 128 - # model = utils.SetDevice(Model.Net5Grad35()) # model = utils.SetDevice(Model.Net31535()) model = utils.SetDevice(Model.Net3Grad335()) @@ -44,8 +42,6 @@ layer = 0 - - # traindata, testdata = Loader.MNIST(batchsize) # traindata, testdata = Loader.RandomMnist(batchsize, style="Vertical") # traindata, testdata = Loader.RandomMnist(batchsize, style="Horizontal") @@ -59,24 +55,27 @@ traindata, testdata = Loader.Cifar10Mono(batchsize, num_workers=0, shuffle=True) # weight = EvaluatorUnsuper.UnsuperLearnSearchWeight(model, layer, traindata, NumSearch=100000, SearchChannelRatio=8, Interation=10) # np.save("WeightSearch.npy", weight) - -# weight = EvaluatorUnsuper.UnsuperLearnTrainWeight(model, layer, traindata, NumTrain=5000) -# np.save("WeightTrain.npy", weight) - - - - # weight = np.load(CurrentPath+"WeightSearch.npy") # bestweight,index = EvaluatorUnsuper.UnsuperLearnFindBestWeight(model,layer,weight,traindata,128,100000) # np.save(CurrentPath+"bestweightSearch.npy", bestweight) +# bestweight = np.load(CurrentPath+"bestweightSearch.npy") # utils.NumpyToImage(bestweight, CurrentPath+"image") +# EvaluatorUnsuper.SetModelConvWeight(model,layer,bestweight) +# utils.SaveModel(model,CurrentPath+"/checkpointSearch.pkl") -# weight = np.load(CurrentPath+"WeightTrain.npy") -# bestweight, index = EvaluatorUnsuper.UnsuperLearnFindBestWeight( -# model, layer, weight, traindata, databatchs=16, interation=1000000) -# np.save(CurrentPath+"bestweightTrain.npy", bestweight) -# utils.NumpyToImage(bestweight, CurrentPath+"image") + + +# weight = EvaluatorUnsuper.UnsuperLearnTrainWeight(model, layer, traindata, NumTrain=5000) +# np.save("WeightTrain.npy", weight) +weight = np.load(CurrentPath+"WeightTrain.npy") +bestweight, index = EvaluatorUnsuper.UnsuperLearnFindBestWeight(model, layer, weight, traindata, databatchs=128, interation=1000000) +np.save(CurrentPath+"bestweightTrain.npy", bestweight) +bestweight = np.load(CurrentPath+"bestweightTrain.npy") +utils.NumpyToImage(bestweight, CurrentPath+"image") +# EvaluatorUnsuper.SetModelConvWeight(model,layer,bestweight) +# utils.SaveModel(model,CurrentPath+"/checkpointTrain.pkl") + @@ -95,22 +94,6 @@ traindata, testdata = Loader.Cifar10Mono(batchsize, num_workers=0, shuffle=True) - -# weight = np.load(CurrentPath+"bestweightSearch.npy") -# EvaluatorUnsuper.SetModelConvWeight(model,layer,weight) -# utils.SaveModel(model,CurrentPath+"/checkpointSearch.pkl") - - -weight = np.load(CurrentPath+"bestweightTrain.npy") -EvaluatorUnsuper.SetModelConvWeight(model,layer,weight) -utils.SaveModel(model,CurrentPath+"/checkpointTrain.pkl") - - - - - - - # utils.NumpyToImage(weight, CurrentPath+"image") # utils.SaveModel(model,CurrentPath+"/checkpoint.pkl") print("save model sucess") diff --git a/FilterEvaluator/image/0-1024.png b/FilterEvaluator/image/0-1024.png index 73aede367d6de54bd913d2a771baef82281cc5b7..be955325af856df998dab45c0cb3f5b66d155e52 100644 GIT binary patch delta 218 zcmV<0044vO1DOMmB!6H@L_t(|0mZ?wN)-SAK+$^#X$TA?9BXiKNPGxQE+GUtG}Vxb zD7dt=xD+IWTjcUk0;9$12f|H1prOgZKxuYssY!^;uXx;Zy#07)bFekB@ZkB?^6~KK z-(Yd&-pQ*Ecc0!_xJ zjmwQJ_y1z|-RkD>)6UP&myg~|PVJq&=j-sv3r3R>0d|v60X7zP4Ip1F Uff(u=Z2$lO07*qoM6N<$f_1KNf&c&j delta 224 zcmV<603ZLE1DykqB!6T{L_t(|0mZ?wN)-SAK+$^#X$T1-8Xa09Legw^?91M@&2QgN z$7W@;T)lsFus-#9i<7O_E6<1T-~8JDGyme|t+zk6FFy_+Kb6t)lK+#@l9BO}lOO>X a7O{T}*e-$Q@cS760000fQF@b1G0?JD5 z;;f7xENst4n%TH=(Z;VtV@QdCgoXSIvUXv?#3&Rg=biiRyZ5~e-Q9QZ`S_jR`JFTK zbY(NMw)rr#gc)qWUAP+?u?d@T54NC#d$ASw;eO0w8@A(1cmQ9<4(!B(*oEELgS~hN z`>-Dea1amU5a#d*4&y6$6ko+-_!=I^6L=C&;Rw1oig_Hv0*>PZPNIkY55df%@28uJ zf6~mOV4Mc$z@gf2@HK6phZcJ0XYd?9|L(V1;g(rh4`?wHEe!l_M>J|6G*t6R~n%>0>l!fOV4ic~OHeM zszm#G!K0~R_#kw;M&)~vXZX1CJ`40_1+uF6%F|%3ULB$!Zl1(ZdXdnxA8FdIG(so6 z)oz1g`#I4Vb#DblV%Q$_xlmLIbO9pMU(ZExzJht-z>sX&DSG&+Ol=F$r68M*>e8bmK|zsj>)%ow5te zvz*+jVmwDN_m(kF!VxRVMn}|^Y@-IDv!=}~ueO%ROGEc)b&=&Z@*U(!in~!@(KuPS zN`1}t4iN{ItxtpD(8{ocvbBQqn9kFT1Cvy`m#BFC7;`JLF3~@G&z^OK-^_XkuP;-- z2F$(s&?t5JStM#xJY#|Sex*9slyFpox#N{^YxiAnA-8OFjq^g yVEJ<(JnFf;94h^jSoOS=uR}ikADi)#Lp6AO#Ji1#NbCuYlw*tjHQ8yB3jp8d+)Q)otZm@7D&p> z-e>LgUTd$l_MANb%E+51&X0V(8mUIBHPu-4p6aUV>gv7K+G@O7S6x%RuUcPis5VyD zR@YV6SDUKM)s||a+FIREO;%IYjn%eld$pr#RXeNcYNooWdVh6ubxU5@^rNM$@B3|{F!aB&-ZMM=lHq5MLqt)Yf;EI z{54w351#=X8FiDu$%z-b5=V1G;rRTB&AsA`t7G;5DQ_Bx- zYdIFBoScm&K0^tWFuhLprJ}zKl30#?v^Re5z7Or+-R*SxyZ0{^?ILfri-%k4NZIWi zZ7nSBZVeFOf*bzm*=YR6qJ4b!XtK~=0XY+@n%=ewgii#cQ}0JX%H;S$)@fUd?SspZ zIhIfPD=$S6e_|7;@bNYA4RNi3;>-P_-CO9id$W6!{-M1^`0vfdbDN4NFDI~=K$5Ys{s|zRll9P z9+1LbkDt9d4#kH)f94vXYpDJ2>2*PE%5<<{w0uR zZd3eij_d>Kh=1`+G|tIu;QPo9k*5W-eESrT220HFr#6Z++r)1dL6irl7598fh>YB= zp8=i#B;b>4g!>CAH1B=M&5kLcXL%sSk1PNN<|4lSIzTEY^Ya&B34F=rh++hQK<#fv z2P8&BvFB9~=-G48^=lw3`20q|h5P2PQ7TF?C#~|rHehW2&mh@9E=I^=k3YT%*a5va z6@j`!4cs>ZCmvlVw7m)B0jmLvAmgYP@H;ogC;9vsY=b&te*RT34EippP06nX4n}>E zNlw z?2WifU5@a@3(<3Y`yE0Mw-I${a5NB7+%pEc0pXfvL7eDazI2G^uLZ}#;K1MTAR59% z&Cx%B&!9tqLX=yhYBUGvuv!JN^ohOO;%|+CeSPVW`1LysyyN-TqWc9%%-ic>34XA7 z)VGB>xM2B>3p(wg5Ce+V*~OR2LE~yajNr-?nU{_F%Io0#H2>}Tc*{u0K~@zp=q2dJ>#x!# zMLnWwz;g~rH-2gtIM>T#panp2zsyCTxC0IKpgVxowCT_t2>+RI0!TB;7HDr&1|}55 zflMO^{)WK=Srd{psbeDGkqtOzg?|_;Znz3KLU%{?2u~Sh%27@D1r5(dF&oAR_($K2 zx1Zo|{jcypsJf;(++yQc zb+J&m(9R{_g|z=V9UT%utj+>HJUhcrot71wOi2TXzt=bb)e0UkdlMqeRoMsVMA?^! z%Vb(o*9RIDRwdu~_-9QD;}BRnz`@~BLXl}ugbCihAgO;(C|m6TnBqu_C1lqsnWziZ z=~0JBmB4kT&Vmy0?DTSCzmc~*aB`|vi9t3q;s{*b8pwQ_7FqAsoT3trgQ?-ge1~n4M zY8W}i;i6X>)Yz0l4A4Y_DheP< zQPl?T=J(jd3LGKlgkYyN=2Va5qzCrkiH z$3A6;O~|rV33!B7$%g&_l{EEvY(aYcq`#2OX1(4lAkX@{xFMu!>}wU^uwkTR^}xZ2@v&wVuX1IH6yKBWKNn2)slkxEb(bTbJuh8 zvP_D^rpr}hhy3Te;_aVa!nbst9D2xF>!C;7(91H9W?32%M(wO5WdZhd-7+Yh<*ZXH z#EWif?-J~y*_Nhs9SU6fd682u4egW_LExs0q`qA;k1j__DdRi;9Zjw>7?6r{EatxL z!2o)?7z})*4mcH0(pe53d1UdB=)}pW+RL;+tPp1i z+gF$&NKUC^5t(dOnlmUju%H3Z8QTb>?hWJDDpA3g&!5a=|vt%zs%3_;?- ziH2s%+@@x{BpT&mSH8Iio0p_2Ah1GNE5}lqR@G=v+JX_*H#j-uYs5G*Wg z_%F~@ih4ssgas+Q#I8jt>J0@^x@8Kf->>;F8v+7UUcy(0W)7z>Ms9Q6c`@3v4!)PH z@Zn$0h1rl87$FV|10W%tt6pn$F+!cB%$?e~8XFpgTb3lre8mJ@v1%=t7|$4C;Cpa2 z)+E&0kO)ZO`oz_8GE#3yO@Pl(Uw~*i8XJ;{G|0R(55aS!9Q9fgNrQq=KS>jfqOqY- z6qySu)1CtWa$Tc zjIPy{>UGUVMnbPh{alACAk!O)NEW6EW4U3&Y-ostB~Iwnt_57^4Gob(R!Q8zP^mW* zqZuTc94<#?_q1M1yMsAA-^qeNf3MflInz9+NgjBR`(+?DHWUwqQ&P`Q43}m@@lb$O zQ>f~++4Wj0D|b>t^ODuk-q?^Raxu^^p)yn`>a|2Mp5;#Ar7nYa0~usyE#)X2sMI_^ z54A}f8&ZzKNz&4>9Vj%Vq>T*);RZpFIiz-Q$R~P3LnK8R=4Gv|!mT&zVgF#)i`|C2 zXU4Mv>J29$siVZK*O_@MkjJ4fTGFhTD;ZISYp4~7Yu|%k2v5#LcMANTm)TJ3)cDm} zTLJ>d4m0K_+T4we0Rf(``^Bwd5Qn@hYragcwyg6tpG>>83=_^FVmVtEqv)$_%l41< z)Qws|C76VnpK0fBY)G#uyfRDE)J~LBn6+{WX8VLeOx4nw4P{W~K$f#1=WC}h8ww)v zyxex2Oy`@zY%I|j&P;VZvP1#BF)vh-ftivlU}c!qeAjCkv_&XsV4->YE@qN>vhnl^<;k#+m!=iuPHi@O+K>g|1qW6ITxM)q%W)c3* zwugcavE)qRoo#^xW}8~FS)UqcfuOs((k?|{r_daVNWF1?Q$K+GGMf~ElZBR#74~ZH zmo~gy;!Ee{v?)*v3;*8xjh*wC>fs$e z&44W!yd^&LcR5-#$rh*SIPv#jG6?ma$fv{2B^`Lx358%7rj&30PZtKwHQXYuMdW~U zI0&`iaYi`S&_5y={$~h+4Iu%h&aGEqGe=##d&Vgxi7$#}tSH>(TG(uRo$3azILSuf zz10Y@f!-BG$mEa+0d4SaTo>Oo{5$2zwejR-Wk%`W&?yUZ4aKn9yhdW6sHdrk<)s3C z=NI0%#AHo?5zBq|{FYj-FQnYLnjU|3E}9I7-4B;?T1UJcOkTFjLNQ`omU)Tq!XzEr za&TF*9!N7-uj4zdHwOKH6mYN;nHJ9>`pG7SX9&;%8-<1meh#;nkmQni2F9?E7_Z3b zhbIw80hh`%wnb!k4t3m1nhgFf3|;@Stp-_J6$N|NQkDlc3}C2eOuWVt>ihX|>f0&w zuLZmq;MLJsFSjYSjscwUiU7t$xheu0>wz=oMj3o=or979?G2LvVecVV3(Se}>Uk@b zfX&`Dlf$bVvN5DolqteTCm)`xE^TsG$~yFNx^%X~#``X?j_)Ki&*Y#HqSUMhtb@