From 00bed3325a393c2ceb297b3c23f470a606bb50d7 Mon Sep 17 00:00:00 2001
From: aolingwen <747620155@qq.com>
Date: Mon, 17 Jun 2019 09:54:42 +0800
Subject: [PATCH] werwer

---
 __pycache__/utils.cpython-36.pyc          | Bin 0 -> 3366 bytes
 network.py                                | 102 ++++++++++++----------
 utils.py                                  |  33 +------
 预期输出与实际输出数据表.xlsx | Bin 10564 -> 10842 bytes
 4 files changed, 58 insertions(+), 77 deletions(-)
 create mode 100644 __pycache__/utils.cpython-36.pyc

diff --git a/__pycache__/utils.cpython-36.pyc b/__pycache__/utils.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..fdda1272e6af9ad292645147d8dd3fd74ce03eb1
GIT binary patch
literal 3366
zcmai0|8E>e6`$E}w|BlfI~ON)N+@orqMlStaghMavWPICBHABnXsX50b$v5-&feQw
z&(6gcYY&N@NnDgr11T*~XaY2Cp^5ubs-{W%U-XChwY~G1U-&%|5^rYj&U=m!XV2`+
zn>X{`y!UzU&C|JD=H<_=v;WB>^f^lWbQpgZT6hOKf(Z7I6MwOz!Cdq7s^J(orkRcf
zG~G*8ZO2wL!%J5)PNq8Ij8wBuwwiNtif#fu@8qka&S-Va8B;XNJ5nt;h2B|5ouhD8
z${Vj9bB+PcCh0BY948qv0^f&7mgGo&MPJ5bl#Fd*XM!9d1z0&jj*@ZsK1{zvj;(3z
zHo*(1czhr9N6!_pNH5eFr(}V0n2d4eR{RpDAuol2SK&oNnDf*RsYrjnx<m=7_$$JE
z#t-UTq<-MJoKwH33**NNKY3Q@KYaeDA~nxyL9H%NGoP&y>cJorcr{+C%2^pp!SK_d
zS<t=*Eu4goV4`gyjv}xaF^G9ZU&Spnk6O5iBdN1hKM!n~N%*yYb*+2vUi%ks?p=Mi
zyL+Yc`%7PZe7pV8?e4wzJKH<$?bo~4-&Kq=fy><L<Z{KMfy>X$Odi_P{cyMa$yUtD
zD6ca=!IC&<NNov%1!&<obSt3n*{oD?6}8ZBv~^<vy@%xOL(VEeQ4`wKQ^N2nK4m%J
zSKa-u&d)5C!5)jJ{cwY_)6Xmh(;K`|e|qiwVx3pKa4NWPKQrr9LSC*_gSn$pTx_z<
zM?+o<HfoH7a(QajtCd|ZoP$B1TVS?n*7@?+z72ZV)CWo6H&Df>1=JU5MqMHm3V~$z
zZ6nfya}N|{M?kR14OXNGd5h414p}ELI|1K;bxfE@>nNwb00WLzafDX2mU1K42?lp+
zL~H7eCnG$M&OY8WA|uk)Qfx9Zh`y%7SC5Qup{5=I4CHrx@(kjJI!SNAet?2m{}~(}
zLt*!$_N#BVZ|-zI{Zse)<@WYP07TOFA@u>IWeSc}e21K^_}e?b-uvTL!rz;tQ(qrk
zdWtu=K=*O_KAtR^!d$J;C0FR>I%kj{kTQYvaL9|MWB27b`TnRB(iKYB1n>$ALIZ#y
z4C%nbcFqDLFvt2H2llW49kTM+#s<znFaNTbJqA>E61q(-mk~Hqme7ztY5cwu!Y5(U
zIHk<|$SdHAO9--qj50eyEv>0>9BJSVv~HaN5QACu`e(338L0c|rOwBHZ*Tv(bMx)k
zrm>AWuU~_h+k5{;`_&KsefM%__iFpb%X{zL>0G-D)6Cw@SKxox{3|7C8ZiIxVPD*N
zXYcA;V0Q4=u<?6ATMwy6!%!i8W-=-G5PLsxdph@gDRGeS=m4WK0ZDVIkg{PAp>4FL
z!}alV${`RlxKYIHQCN#Tq=LYn_1x+bap$Ih9zsK#k3;Nm%1Q}Mli?V%LyTE5ANvYy
zD(Vs<u}z$8VrMAuQcr#?T1hQDDB_fqFBM0r3@%7GJOSgu{jxBT5KuArFSsp-ywt=i
zsHs8I8~*^eLqrmM0%8&(@QtRyb)Y~To`3=-sZBFM(G|sly{X0%+~ihd$|!`L6HObk
zUMjL%7^5h)YR9AX7XeNh%u<q`*k}EpeLJvE@9$fG1fo72$$TMmYX*R8v^0$P$ZEE)
zKhMeaWeMW+|Db0Y0mr}xJWA)6nnvTrm?H1nqs})+B2CVt5omZVwS(D@;?tvaOrJ&m
z!=Ok0fUkuQ)^_mvV`mWb*)ymoQROPI(t$+CYeN#%JR$LJPh#ef!~q{j9ISz?t4iX}
z!D{I!v#hDSD1oWMul>Qr{l9K?-h8Y3$?w{i?zDe<xqa&|`+xXlZ@PW4z5P!aTGQug
zIT^2Une4pwLFdXXsG<;R#O3ZXR8bnU^`mWQGm}wOm(WWJRfMEim9vVYOvaxund`4m
zsFUa1fC^n@#R?QN1!tAwgzgd|EH{9vMO5?%s{)S^#)akw!rY{+77E>inr2pfLL0@b
zlVk2iX`r%WOEn><a%XJA^?Auv^;Tshr%>^i-SQAk;EFSnV03arJVNKyKt-o0B)CfD
z8Vl;77$0KpQ}OBpd48#0@knod8f+2HL5D00vb=%EA*c)REkMT4L+g+7I^=$R0#E1#
zd|dKqrw&FM`wpBiNCL7v50b$1Fp=exYmoP$0;qt&jTsr_Dk!EFp&B%GZbSfDqQNV1
zj%X272cj)&vuOQfZ>GmH0GuqsMqO5T1q#nyqF(&?s`>O>jVBSM79oMZvD@Cc(Yg7z
zz3aCCRp7-64Tm6=*t#zfvM)iUkX1C<Q*!(b*-3+mH1+FM%3Kb}src%hs^ma{RSK|g
zN-pd@V0y(h9WxQ&2_<NeB~6~id2I}{({iU>DploExKupJzAOn=;QGW3g%%Jo7OFRM
zx#n}=p)BkH8dM4{C{4vf;hINX=1-|tF96V^WnQWIOjc--T?5s_DbKwiH%n(R=`AK7
z#4IH{ncrDjb{XjU8`NPT8<8Yaogqfzqtv63$-AiMNc=*ObwS}`a3>$axKv6QMLkJ-
n^$kw1dvs1(J)DFter(7G#{_^fE1w>^owmmcS$o<Zv+;ic3K-T=

literal 0
HcmV?d00001

diff --git a/network.py b/network.py
index 59208d7..d4e4294 100644
--- a/network.py
+++ b/network.py
@@ -1,45 +1,57 @@
-from keras.preprocessing import sequence
-from keras.models import Sequential
-from keras.layers import Dense, Embedding
-from keras.layers import LSTM
-from keras.datasets import imdb
-
-max_features = 20000
-# cut texts after this number of words (among top max_features most common words)
-maxlen = 80
-batch_size = 32
-
-print('Loading data...')
-(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=max_features)
-print(len(x_train), 'train sequences')
-print(len(x_test), 'test sequences')
-
-print('Pad sequences (samples x time)')
-x_train = sequence.pad_sequences(x_train, maxlen=maxlen)
-x_test = sequence.pad_sequences(x_test, maxlen=maxlen)
-print('x_train shape:', x_train.shape)
-print('x_test shape:', x_test.shape)
-
-
-print('Build model...')
-model = Sequential()
-model.add(Embedding(max_features, 128))
-model.add(LSTM(128, dropout=0.2, recurrent_dropout=0.2))
-model.add(Dense(1, activation='sigmoid'))
-
-# try using different optimizers and different optimizer configs
-model.compile(loss='binary_crossentropy',
-              optimizer='adam',
-              metrics=['accuracy'])
-
-print(model.summary())
-
-print('Train...')
-model.fit(x_train, y_train,
-          batch_size=batch_size,
-          epochs=15,
-          validation_data=(x_test, y_test))
-score, acc = model.evaluate(x_test, y_test,
-                            batch_size=batch_size)
-print('Test score:', score)
-print('Test accuracy:', acc)
\ No newline at end of file
+import pandas as pd
+from sortedcontainers import SortedSet
+import numpy as np
+from sklearn.model_selection import train_test_split
+from keras.layers import Dense, Embedding, Input, Flatten
+from keras.layers import LSTM, GRU, Dropout
+from keras.models import Model
+import keras
+from keras.utils import plot_model
+import utils
+import time
+
+
+def build_model(want_answer_size, infact_answer_size):
+    inputs_want_answer = Input(shape=(want_answer_size, ), name='want_answer_input')
+    inputs_infact_answer = Input(shape=(infact_answer_size, ), name='infact_answer_input')
+    x_1 = Embedding(want_answer_size, 128, name='want_answer_embedding', embeddings_initializer='he_normal', embeddings_regularizer=keras.regularizers.l2(0.01))(inputs_want_answer)
+    x_2 = Embedding(infact_answer_size, 128, name='infact_answer_embedding', embeddings_initializer='he_normal', embeddings_regularizer=keras.regularizers.l2(0.01))(inputs_infact_answer)
+    x_1 = GRU(128, dropout=0.4, return_sequences=True, recurrent_initializer='he_normal', recurrent_regularizer=keras.regularizers.l2(0.01))(x_1)
+    x_2 = GRU(128, dropout=0.4, return_sequences=True, recurrent_initializer='he_normal', recurrent_regularizer=keras.regularizers.l2(0.01))(x_2)
+    x = keras.layers.concatenate([x_1, x_2])
+    x = Flatten()(x)
+    x = Dropout(0.3)(x)
+    x = Dense(64, activation='relu')(x)
+    predictions = Dense(2, activation='softmax')(x)
+    model = Model(inputs=[inputs_want_answer, inputs_infact_answer], outputs=predictions)
+    return model
+
+
+if __name__ == '__main__':
+    df = pd.read_excel('./预期输出与实际输出数据表.xlsx')
+    want_answer_corpus, infact_answer_corpus = utils.build_corpus(df)
+    onehot = utils.label2onehot(df['是否正确'])
+    x_train_1, x_test_1, y_train, y_test = train_test_split(want_answer_corpus, onehot, random_state=2333)
+    x_train_2, x_test_2, _, _ = train_test_split(infact_answer_corpus, onehot, random_state=2333)
+
+    want_answer_corpus_size = len(want_answer_corpus[0])
+    infact_answer_corpus_size = len(infact_answer_corpus[0])
+
+    model = build_model(want_answer_corpus_size, infact_answer_corpus_size)
+
+    # plot_model(model, to_file='model.png')
+
+    model.compile(loss='categorical_crossentropy',
+                  optimizer=keras.optimizers.Adam(lr=1e-4),
+                  metrics=['accuracy'])
+
+    print('Train...')
+    model.fit([x_train_1, x_train_2], y_train,
+              batch_size=16,
+              epochs=50)
+
+    score, acc = model.evaluate([x_test_1, x_test_2], y_test,
+                                batch_size=8, verbose=0)
+
+    print('Test score:', score)
+    print('Test accuracy:', acc)
\ No newline at end of file
diff --git a/utils.py b/utils.py
index f2c9cc5..63c3f88 100644
--- a/utils.py
+++ b/utils.py
@@ -114,38 +114,7 @@ def label2onehot(label):
 
 
 if __name__ == '__main__':
-    df = pd.read_excel('./预期输出与实际输出数据表.xlsx')
-    want_answer_corpus, infact_answer_corpus = build_corpus(df)
-    onehot = label2onehot(df['是否正确'])
-    x_train_1, x_test_1, y_train, y_test = train_test_split(want_answer_corpus, onehot, random_state=2333)
-    x_train_2, x_test_2, _, _ = train_test_split(infact_answer_corpus, onehot, random_state=2333)
-
-    inputs_want_answer = Input(shape=(len(want_answer_corpus[0]), ), name='want_answer_input')
-    inputs_infact_answer = Input(shape=(len(infact_answer_corpus[0]), ), name='infact_answer_input')
-    x_1 = Embedding(len(want_answer_corpus[0]), 64, name='want_answer_embedding')(inputs_want_answer)
-    x_2 = Embedding(len(infact_answer_corpus[0]), 64, name='infact_answer_embedding')(inputs_infact_answer)
-    x_1 = GRU(64, dropout=0.5, return_sequences=0.2)(x_1)
-    x_2 = GRU(64, dropout=0.5, return_sequences=0.2)(x_2)
-    x = keras.layers.concatenate([x_1, x_2])
-    x = Flatten()(x)
-    x = Dense(64, activation='relu')(x)
-    predictions = Dense(2, activation='softmax')(x)
-    model = Model(inputs=[inputs_want_answer, inputs_infact_answer], outputs=predictions)
-
-    # plot_model(model, to_file='model.png')
-
-    model.compile(loss='categorical_crossentropy',
-                  optimizer='adam',
-                  metrics=['accuracy'])
-
-    print('Train...')
-    model.fit([x_train_1, x_train_2], y_train,
-              batch_size=16,
-              epochs=60)
-    score, acc = model.evaluate([x_test_1, x_test_2], y_test,
-                                batch_size=8)
-    print('Test score:', score)
-    print('Test accuracy:', acc)
+    pass
 
 
 
diff --git a/预期输出与实际输出数据表.xlsx b/预期输出与实际输出数据表.xlsx
index 1238b057b6afa3efe217c836ae376be4385f8d58..3d68820f259e7c20389f0664bc73b2250033da01 100644
GIT binary patch
delta 6176
zcmY*-Wl+^$)b*viyQD)vr0de%c?C&nxHMcqN`4YvQtHyprA0zPI;Fd$ySo%6h1dT)
z&%E!vd*+-Ed#|-;&;GE^oYiZv1gT9y!8l^Ce^f+dqVNE&hdITkq$1#EyCDbRHSFMJ
zT0Hjme)reYew7x&&o<u$_}?frh=1a~Lp>fg-H+lxS1wR~q1iq$HR4Azp}!+FpKSaM
zPM_OVsO%Ml!0AzPvvzYqn@it7%1y`PAd14T;npX;DCBR|oeqf4oGY7%G+PJYu|yqM
zQp|R63V(pFdA|vx-+>7!WDtV_m2sOzDelq>MLlIG#p#o^ATnx}CGr_6uh0MlH?<vC
z1<Yb(=$BND3HwT+$qq|UbRcHkXd?Tp!mDE01V>Jp1A~mRsm86llB1`{7u*^H?!>iS
zOJ!oz>1)aLte5Z%7``|cpOn0Ph9ih8`8mClO|oWYpzM^Cf=pp1W7H<u{V-<=tTGrB
zbU%A*RI{$<82J=^{UQl-E2ShfBR+~sZYtmn5WPjGr-_D+NZ_EYqegQ@!R-n1+V{o;
zf&MTe;?bD^H;*E5;(_&ipLLYW5N2wp>`#Zzm(ISNr|OmMW%DWG1l$qylMdI*>(pSW
zxYNSQS2G*3LO}m^KH_e3JE7(0n$mLW7<6B?yZ^ek!s*ZMVZi>=-}!W+w;l7pZ=w0d
zW9lal`N&_gGLLs7JtfdLEk(rbSR^{D<FbGL-pb$uty^42ZFpIA;@yLV(Brnq##7ir
zu36G~WcCraUG|vv>;2s!W0JGw>s}|q$gjlt*<0Tz%PcXV%Bj~Io#3?nGuqREM~llp
zZXuzL?RS|b51&=~{q=2bDo2oREqjw#%7&4{S+^<6o~{1lE&Qk^F|YPdM~-jN&}Djh
zq*Gr3i%;~=uASAc40A?yNzKPC^m6KBO+t~&dRGtMj<fGzYx|GeJ^X0vOBa{cZxkqY
ze*{&DpFHSJN$Zj^Vqy--mNfa)t+&encZLw;LEp2$zU#ZahgbB0e4lXLbv9#8V#-Pb
zk{TZgksr0hR#*@_%5OTNGu!TUcR!dhL&Y6{+@}!V3;Tm&kE%wh$cJx70fv9g&hEFz
z*YD<C*0V=Ekijy4tX>ip_-4;{aO-YW`jb(t)c(rUf(TP6eaN@87_Po<mwXavvxM|U
zv7==F{nc*6RiGo@#Ngq0=jT7$SSHp$r;zIBuO>m3K;#CyEYM^I-j46m#+=!5y=Hz1
zge3o7y=7<vf>xv=JoFF>3HHCfI7#&H+q%-@G0$fZ=M0m6Sq8<K8uT;?3l(?JDdIN>
zieSpB!Y&QE_uuO7QDSaUpz)sx7xeF8lG<5?H1^2ZymJqj{k_slou42<S~?f;^Apcz
zFs}kZ*M9iS0<xd~+YZ}EMf$OI)}W6GFfa;P;iWKt4k;3U_Tblt8CSEwPtiCCH{ha+
zcLw3viJWB=hn4K^&f1H3ZfD25_*zJACjxpk-A~YFsx?@tN#?&L6l=%*P^-|KoG8k&
zu=nCNa^x#xSzZ)JMEP%z87tJdEcRB)QyO_nlNdJKlURM+0@~UP)oCTC5=T)2;K+uM
zb{LZvv;_Z7<7R8o!z#Lv!Qo;;5cSfn7|LX8;)jMIvIJKSH+E$c%eaMqtw%AGH&!f+
zZ?K-)prB*_H;czK92Oo!{zNHy5ty1)0lP5$p#O=`<sy#!)(NeEhM+cD(>YZE9ojWZ
zM4P=#H$chod~XU<iy-?al&%pdCXl0*CW#4v(Ty^}m2HW)<Lgn>irl#NrbOk0w6R!l
z5*kuaFx%J&)KJ1w^|2~4@vXGFxuz1eW~Q52imhfrP8?_%_{5=t1rmp%R)UFzukye6
z5?~R0;iK02l7r33Spu5xuC<<G3iOOpT%64DS>~s4GU`EgXV}Cb$fN+uCI^?Cl(1Rn
zj4c1*GA<Oa^0n7R>_Si)R$}|}NzX|=1Tyo`T^|uyN)U^Q@=>U7v_fq}-NcqdD8Ol{
z)`?$;nNZ}p%cvcPkgZbR2?H%5t&PE!h_?ocvwly1IZd9}f~$a(yj;NRa7WkgGgnqq
z)S_w;{Vv$4VY3EKQeOe_4^Ibp*yv1>EORGzbc@@2s#+59Gez1gL6PuqcozOg9lsuW
zMqgV*?3c|Rq<qFcapcd!MVgYNZ60e^l%jvRZZn+T!eH~w!z>CsFaZ@M!J4{FpAJ^N
z&6p07mXe=VMoU|e9b$%KhYW=DATF$#|B-&MIdyId+(rul9n&)+JRDMF_1boVn5Md~
zd*l_bRFIp8)o{q?oxx&<hlTVF3v`v|ShTrQr*oGg(gMv<`GR%ONu&Re%E(1^$oGh{
zLo{JSc7ADQtX-=QdC}^+LY>Otf$56$!+oiF@f})Pd|7<Fw4*vvip`TJc3xXIfmvL9
z3Z)fmQ9$e>SeRaTdZ6B)it!uizmdTdG*CuLf>KU9WUy0Ui{G4*SCXH#v%!DtvEkI*
zvu~f8)b9C34k00Z7ZDFTndXa!XN1ZsOOX}oB-RpkEJsPsfi0~N?)P~x#^+nM4pVI9
z2nnaCv&4Oy9!8>^--$7={`IfM>McAfqydg@wqc(@8dU%G$P_7_g7Z;-KmVO!pW0G?
z3tW-jS%;}3hipFpL?2$Q8?78JMiFIOR~Roj`+P$JpagbKVF5>NxbcRYKZHTl^+)uu
z8wNtt$V$1{veHPz%5f_zeL0@^ixqUeWW|}Qc`?u7ivA9Ks@=mQf19Tp*?$jH5*l4y
z3@$Gw&OXMY2p2906oTj^rUaf6menYp7^0uLy)k4VC;jU```X9S-2GlKGaM9C;^64*
zYUfyJ#<wr%EJ{Z}4pMh@xJg^WzaO&2gHwN^S!~GZrMCXbZ?>zGUG413rdbw3FmSH5
z_IkA#*n&fT$(*-O1PSHQI3Dj^b|KCrtamU&%#6e+OuQf5Ct-iX8-Rzu;5ik|G~~jc
zpa!ZN9Wxm(mfO8Q^PLS8eVlo)`$;bwoUDtYhG6OeRWavp<v|DdCr(1*-oZDej1g#H
z_{>ECCBujK)!4+zu1aXkbUA_Re0pQ!!)6z4Z_*C}c*siH$CG5Gj$0ifKIgAWrpM^g
z@4b!h_H^O$4@*=vc|c<L@$@$ji`LSWR7<kp-$fs3-jLoB{0!agd}o~O@|=${LLa}k
zRFF>f?3z9$Tg)+BzpT5KuZi!A=o{U(SA%NOWvk@AbHeTXXm4LcpZzhCy?`c)U?AhF
zT|JaNG_GBVZpv87-lDz4E}?cp*Y7tD4%$crkLS0)d)%VvIGzLuS*;jJeh~N&z+N`P
zHtyu3_~xPre|E;6kvicK{aOBMFLKG4oK<P?ywp?N%vw`Fh0@(wA6I#fy~Dgb?=O81
zsYM_&=O^WDj<4L&nTsflqPP`2)Iv#Ld8RIT4u;~Kf24ls*R+EmwHkdYoArqHl34Yf
zR-MoI=2|46nE3htugfC6dR1Lwi!%52%A&#xn22|Gb3gR{D@C&v?#`i2M#W2E_nhx{
zUo9k94=MA}a-do(f~KYyL?&(J)o{Zo^(HOT@eB4JfUiDHWF#I*oR;~-2xf6hd_T9R
zebDh>ye;R|3-P4nZ7`=Qzrm&T`?xkU*6Xq6d!~9_2mu*Tn;}Q;?@m-Gez0*#LN=>|
zb-1eA;rTZYcGC5}Idnit1k+RBbEYMEyv8J+mfB{3FA!f()OSWW`*&bvbyikk*6xAF
zXAOXITBokgIKGO#nLLlhnARZ2aYQ%qE@sRw!z^-C;CI^M<Y?Nb6uZu5?)$pY#btqX
z3w$Aq+;|hf^e)9<J@dO6-#deJXUd0E!<(#1r1NyTV8v2|mRrR~qe^?)mKgBqZaO!|
ztC5esHnKCUDK5bGZ)Uz%#**_!TP_byoxvIc?QfcywYmCIYElmz4V>BUGE+=(#xjyy
zNbhMLt9l>pjn9g`eQAT{W*Lbt6gi~k&akYB(t?TsyOO>LP7QBp&<N~KAWlf(dV}~~
ze>VxOFS-{!{zk$65>}{3kmqYn_Gw)<oDrp3mLF-`ifu?m2UOBQFzuaUUjy7-+lM@*
z%b!0d3>75RXQ3BHEwQe`FSnBm8R<G~?m_WSp$3|HY0AsQvf$IgN2cN8#pP#^=)q2(
zMI=xT&_vrj+N0y%r6+MYv!1^x?8`JH$wXHR?DkxT@eP1hEy>|}UJCZY#ySpcu`O9i
zTzk$ca#jUgk8w)+bOFXaG$dNu6GcB?t%$4kL5fMAOeuMzH-|=eA~jc-w&~<gGkIVV
zb}(gh`D`8`9#{3|4_m9gQs<)Fb+QMJ6u^oc`$N;jXZ;>p8PgLbm%RgG4sL{^LG=(0
zdL$W3<?*j#w)=xY@RtJL6;TN^j4M~H>z7v#Y5UJ^&>|`PS@CVk=(ZWKoCp?lJml#}
zY#L*Y!djU&6L*)lAITy8GX60dr{Rn8(ISa*f7d)~&m$<;BIM*#@ksmaY1F8T41mPi
z(RgVt%x3fnnt~7+5pEJYv~#!Djxu2=!auqeX$tw$6Zmy>Zn19-nd30m-hOo#P%S{v
z;^kE(f-@y6_THk``9o7JmI~oD>hElFV*ivxgBLY!)8t2MQMwF=6^63(OK#i&oK>>$
zM0M_73tSQ%2f3B!zE`1vGrw`AT!C6M@ZfQ}rOSbxgH~4w9X_!`G2OZ>ZzD_jD1Qr*
zI1}0T@z;EzgM8X6w#hgu{n1L3JL^n7o?>pTV8fI|bs_rZ6?*xevSP(8`A}VBOk8Gl
zedvlO8nRUoGCCUC-&dnCG+eA;?<fZ8N0?h+){IBFS6E<es}i!XdA(v11S%?F>IUpG
zf|;MgbfHs~q-yX4Q~P}@YL%l@jG;i%cpgS~NDMa->7NH9#^UkRA#Qsygt)0a)*}Ak
zH!egfgLy4Q`PV!<WL34Iw@d1~Pc~A`gw#CuD_x~yk`Qm-UDh?))_NVvrA(1Ocj=}!
z>GAhl-~6!ELtn*1vZRpw02vl{qeBRF#94MNQ-aah6eiZ~QRM-)?zq&SEjcw=g3}`K
zxWEZL(=ynh>KO1<s+@;E$6&K*aJ$GB6&zMS_|VoM75EtZ^Xi2}(Z_mOA8CI#4%Vsq
zuAOyV)d}EIad)5sKQOsK@W!_aN~3)XuV?-!^G@J)ApG+DAK)4ACitp*=Tw)iyq#@<
ziwepr8L%q6>Ypr^+ZO!x>O!46G|+(?<@|TgUNW>D-Gw_;#T|m`k&{D7Egg7q>zJU@
z`8eBme=fj*BuvxT${m|Y?l&i!f}y%84S0BIH45IHo5WDg1_Xk&sBlhfH%q<CiHf{;
zFovG1itgDc+X?|oQOtg4z%$FXd+&#AUKancI1L<I|Hpy!jn&Kbb<Is=AF~R@8a=Sc
zldm;yX(1&82A-;h*sG*`BitFDj%%pO&h5AZjzF%VF0gm#N7(jLuDs#qRMR#On}>a>
zK;p)nQdAjfK&$7{+E86*T*y5k+we_zr+t{zyJW_y3;>sRDlPhgJj@!|H9j<x&@Hlk
zuUq!xQuq^IKngV$iB{Y;6iv4u^+UAYp30#4+(&*d9^&DTDy4K2EqXt@ebNdtmBgwf
zrQTZYy<tkJ1mk_`C&Lw$!pV}KJ-$*t;(nl3-$V^e8eGhun0V&MBIRpkWX2@;c=LFp
ztE-X_A_fHBSUqf(W%5ou1Mwz~RZTPHK%ja0zSeu#!%_sIx;*|pChx-`sUxBnC#XcY
zb^}jR4C_yCVyGs#_$?_Nx(<dM7ykweOh=%1G(C+><jamI`Qvum#)Do}J2Dsd)@|U3
z;^7xMggOd6_IPB~5;%RA%!mya?LpwvbMbRF)&NCg7s26+D)K_jPJ%55QtN}~4O+${
z=EC)v&*!3{{I))*xgy8r@7eiPiE}?I$Po>Tg6{D#2zKuNHTgx>FoByj^c%|!zR!E|
zo=ljPpVO9SNoc_59elE|S=>oT%8{VZ3^wxeAK){ymVBs$iAlKY_9)4oPPIfmO<t8R
zV4VmXD;&FUt=eXqQB|g18W3%Bd*5j1XdAR)JaM?J__71>ZZYlf^6jM=3-R?%wb)9K
zd|bBE$kmPS6(!+e!)E5;Z*0q~P+cf>i`%VZ5Lg@bc4Au8bXdowXEOH=ZnD0(T27an
zyJ|545ia=!D|bAITy&?e>@Q(-q-$C=;ExImX<Ph|9uyX`ekz#`<c)pKJS3`@XBA^1
zI>RNiCA(-m{MIm<Lwpl8D)L_YaX{q<@+5M*;fd%!vx6K20%3z9b++VqFT(^@(Lo?r
zF2qYZI>2<*V;MpYsBQt`)C5zJx2({wY{hw##mM#(6r;=65|`Sxdkfvyr54Y4aFy+|
zIGEkgK2>j<6*MR?$uD=mF$AB^JYvMxtoFlBe?SfI-WLsOoc!`l0QOhb+x`CT{f)<o
z?Jb16jbX=d;mLh2t)iRz9>@Ro?Vmtp(6F%PJfOAABBYS7w^@33=i;kDQ&6q!7$IuA
zn#DHdknJPPy<}mtnVNR$BMb*}F{1q#kcai!*vA<L@@XRmszSUdS$-2X2gXL|2PXXk
z(%wAchTM40xwOu?6{_pi{>Kx9Pq)(+tCgBsRho~t8(FRGURSeaL-TEA;Ikse_NsGC
zV4~4BFl0w$X*Nro1^*+dV-mJA##*3Z()th^-Zr*z0d7prsy~ygif~~VGOiGIalN;3
zoXR8}9$Bf{DI>iQ^g6Z1p6FN4+x1S^{dkaw3a>Nx#f;^rF-bqwBYjoD(^3bG(y7Fj
zcXJI@$k)j};ew;L(r%%8&~kDOwHK<#KqC8cGw0%JKyLk>{_<2#$HM76?ygN0^j?J~
z^Nrx^`%}t5+9f%f^;mx!G<~%){7+<U?H$SS#Fm0ub@KGnU#M_U&0T)Ko?vFaQI<K2
z<+LyUP`vW#sYtkALsMIGj>a1cIDNI_^74(8_{uG;_G!in7_qqTv$*FdO*7xk0tU~7
zLRe`zJNqH@hQSAA0fY$~0Q~Z2XC~@Jo-1KPzW*`!?CgAP+rk4Th7&TRE?E^DBJ!#6
z0#D%l2)db}WW*@d@k5cLNCSs^G4fE8(NPQ$Ku+k(ln39DhyQ$)n3{q1DioF6npQ&(
zUXl=Xy3ze*+klb(B*Ti-CG?vMptob&gF`cMNIUcXTHa|PZjr%CG^Eu~ge&zgf2zXh
zYmKNwKF=nSmx)J?{uATdUXp_4pPv;b9vQy#mD1=9wBs)`@H={JbLODk!4}%o{#e3H
zIbBb0bd@P47?VU-^iicSNhDSvcIYNix!H6}eUz`7e23{g6SBuOt=%*N1$ud>v7MMY
zqD!z~-op>rtO5PM@rViRN=Qmc(iTk2(eQB<8&K55y7uWNnTQ1m`FSUqm>339z|@;=
zv4Umuc7xr|nLOZ@RLpRRip%<S-0z#l%T5byl1|dR{weZl_J0beBCDfa=bDS-m@^f=
zk1zeY=_#{SZZ@yBuE`YR1&nx$9R|~7jIvvL(Z}>rKeLO9m?Nk&G|gSmqc74LR|bw>
zzTpj$m3=GLA-=Lkdjf2|dmOX~f^*i{N@6(?>X{4Ww1yd?3^NZB(G*CM=gAq5ZFrzs
z{#?twKAS?AIYZH3EHA(Iy1rgH2fz>8kzY*x>bDah#tA%|@(}h6fbA#l;36KUwDG9r
z&9l8Bj#UESoHeOE$0|Ugp@yowcjNP3Jx7@cqEy5yuCL+7^H_&+WOn-Ma24F^qa_g_
zomC{NdRnYMQsGMqiG}Lz(n?suLOgvRm`G|qgbDULSgdLOvJ8CQF8sRjgNt~_W<Q-n
z6WYMr?J#zu;$JP5K>lXY9a$Dn?Cdn%;xcdg!};nSMT+rc2}F<d=Pcf4HtgK_OQPZV
zgS_@d$M~g-knvLi+~Zd+tYxFA8p0jjlow2c;h}N42t|E(+M<viA4C}cpo(EmCd@nD
zw9a6_S*L_NVf_CJ_zl3c=0E{~R{ov34n_n9Y6%;{kX;s)fCZ7lE{H0?f*52^{$G#%
z4F?3}22sx;hsF6%H2qy)B0h7`Bd9nDQB7D8oSY)4?^zL+oV;}ZPgX&o|HSscKmjrR
zyDJeToSgLk_cN~p0ulUI{1^XIe@%io;FLh6$3w7kiJ&UtA!=|b5FT9Is4Yy0Y%Va0
g8e*P{neP92)F2S`|Iq)J?}H$F&Vwn<@n7=)0B_&2D*ylh

delta 5899
zcmZ8_Wl$Rcvn_!j!J)Vmhu~14G`PDLcMa~80xbl0C=lG;wOA>|-K_;$++B+VfBoLQ
z^WNOMKlaC--JLW0<IL>&qusAnpA0|^9O_Dug=YZRVbv}!#V3RU!fn<gQ2ten&r1|o
z&PXTsXC)`P?RFx)50~fXVb;D{vgfFSyCrG_aTu5qF<QI#3597ZY}rNKLbXiXN;s?z
z-Gw&ZpWu<vQM28H$;4!7Avhw6M$DB&)HzjTgSu3b?v!`oHyfS&(ZXH)=*I3BJ$QjE
zq2u+)upy?8RIT0h6p{O}$(9njd8PYETIOJhAa9bn$j!W3l%nH2^zYU`wc8xoUP>EJ
z(>Y~na8KY95tb$kn$#aDR`WJnum84ua!5b&)<xDxRp<!?oGQUD3f0c-RwirL=WCwQ
zZz01G%u7wLa9iZLvVbq~qBZcdL|T-(KEFHVcb636Vkf%bEPNrH5e~*V>Ks15DGXm%
z*Dj&8!)waJ<8yKM8NvfEc@hq0j!sGf?!fB5ozH~71sf5_(E!X6d-l)Ws7Oe46!2Ie
zEzHri=p9jbS6$N!;KHBwEu(!UV^LI<gI;EeTx080@;e;1b&hWn<6Uc3$8HyKv2xDA
zX7Of|kN_2nhdX&4G@2%REZs9HBF^5xd5AE{^%2+OL*1S_p2hTKwHTDC>VWG}aq0Z;
z^MYB6-u#bdTP0s4f<tf8Tu7J5pnF!HJuIb}pnaM`POZA)4X#<%!nY27eqnDG9h+Fv
z7#(99S?W=YFB77?iwD}<S5Z4>#5HZ$v(s<3*1JSW({4ykhAwJ-hM3!Xn^zxWImAX1
z#a=7nz;qG?t8G|b?^H*Z2>fLeynuS-=(qsCGFFXdw%(4hV0;dW`NTRJ{Bq273x!2q
z(oJ6b#nq<09d+EXOr~v$XJt9#wEgvdu+gpSgzda0>UPNrU_exYN8$G;%xs!LaSk6%
zXha;fqorsHFlG8&%rpTJm8uH`*rwH8^@BK--$7J+CA<c(tXwyLu1PlG&wVJ1RQ#Gv
zN~1lZ$G%_vvi;+noH_2nU?|TOHy75fc0Ui4_ZwYRR@1+8b>H4I`37<w>TEo6ZSoQI
ztWT4f{(hX3Cac;ti)YviYR8X>^3aviMv!ZV?cej1f2CPDtBQ?uqu7|~;u&rSR)P{S
z>D7l0y%05OX1ygDEd3sTPU&TtZkEDLQ1-RbyE?2baSDcjgsCwhW-<XEEnrw{;x{qQ
zj{wNXuSE-y@KP&lk?@BGvfV=#U>Uuv4|Nnp`uo0bJaei~(-Hel)Z1i_#}iENQOWZ}
z)3e9|)h;Lk-OB)ww{H_1&<1UVv5V)ANaRfxPRJRyc!xrjk3u)g9eXdxW+?(=(IXlZ
z;|rQrsbXehu6c+&AQM|X+OX<UOnTz4npC8nxeb6VzPBm^hK+Q@v*8Ss-N=fl4S5;V
z-0^Dgg4>RYMLVwI-Yjfdojs}bfkPBvCSJ!CdW^M#f2&+nvQG}};y$PEK5k4r)pukZ
z{6t{Ay%*Y5CTj;XPP92mVjPj|%?=otfTKSt>X^_nkM}liehA1f2BXvfb>V1MLMZ=I
z1iRT<f#>-t-7<OyJr3+}7#nt{!exu<3?jG2(2{%ea6rOcI5Bc&jo@o23KEOcyH+qG
z=v_krw?P59ki1zf&*4}fl~|&6-WrmfABqnx+_(f^lFaQo1bGj-vsj#1oVylS(bM;N
zbnuTOO8G8#X#eUW4DD^C#SB@$&-ovXZLix9>)bUoyKoOXMmoDKzQCZgRS}FJIS>P$
z6d&0%FKSV4cz<;2BsNA*HDDSW&FhT>FfRzAW(g0{H@c1x<_GfUeh4b)FT`ipbn|2&
z?einSenk>Dh>Pt0nVE>mgjwY|BctBVOU7bk7(fN5Mq!tO3EisGrB8%aD2CrL($U7(
zaTBqwyj9U+rU=FLF_o6sVvpJ^SqdMKn)`Q&o>=pdF*?P$La^y~=X4H!XMKJl?uPA>
zu|X>ChLJHDhRqFJ96($dc!WwIOu|vKPp<ADlgH4>%PqKj&jL3)3Zym{B5SpHzR9PF
z@VYd@fWC(nOh>|*9v7aU1CdO^@B83`Dnx?th?FgFBd{0B9}@R((*Ge5AyqaMp*Yxo
zZZTt*&JuNt%cpsDg{aS;`B&IC0*!BvMfCpb2#X8Hi;-JXB^*&l=Or9aZ!#2ec^78c
zBCet;qylbz6}ffv9#x}BW)qNW6Qj=I#eWI2aL{s|qQf83OQVz|-Ao#%jOz#5RXP*@
z-)p`mtw1gprj#ns891loe{wdfJY(cJ^vpkxnx{k;)Pl#AqQuB8xnT5PT*+1<4n{&j
z8guA&;VS#UFvw?IBFTm#hW@{2zm@d)4bKQ8w>%3F2VNFQGvmRsMRTuKQ5KDd6>YCB
zFRqzBmB6eg95q`dL1%z<cNj3HBoS}Ded;_sUl=p}vjvr@<Z@@uhAMv8&~gmM8xPCa
zwa)G>U&zfeHBlYvy`JjB3n1ZjqYxGc8t6s#g;@p@35fviN<ad0{<1Vk@NAm@5|lp^
zz~jNvc`yOVIhvymLbAv2=F-UL#LkgfA;_0WDO_$R0xSznG69V96%1P?14}{1x6=O3
zbA<xS{m*=fYPS`zf=2G`Q!V8v7slvI0Ew)-cW#C+)Vr;Hd-58dTej>TTyS(tS=m`r
z5k2SnK-~)asR4tzjjdKzjPf{L3v*7L3FBP<B(Gff*!5JCW8rfM5&S~UqMaZ0Twm}y
zVq@K~g5QZoxMSXJiRPU@qT~p@Ll?s=q-fR+|6R=_KIr;KQh|MnR^86cua%tVa@*V9
zg4rASr}_fJ3)Zm5OG9)uib^R%$9w8P52L$Nb@>3qAq-|FLXWPO!d_?GR=A>Tn(}zH
zRRa+xyyLtK+Ufdeq^V56^*S^Kr@!Sbjr`#ma`Mr_UdRmG+=tXy$0w^!%hl%1tcty$
z6t&=l+IWvi;mpd)!M#6;wR8x$h_79^uQgei6Aw|wP}b@{g_n$P2h<$2y|2lYuKnh7
z`8?1_ihxya>GA+?2qfpE_IPghQ$$<O(D$vQ_VtsPIkR%5L**=W+`JYwobT}jnx~s$
zF>OZ6vD~q$eoDtx3JNG3=ui@~S0y)>7&3=e$ByGi+Ckr_Oex7oQp!I>(h62Sj0vdA
z;a83~skVRr*<E6AtEHHeco)n?RiZQK98c*?iGX!!D@|&mWt4L3WvmYw$k8XIw|wtU
zN-0f9vQ7c%Nqbi(F*De~T(0;hOe^{Jz>R69%FB22I?h+EDwP~=wm6ryEyd$~ePML5
zD#lYW=*6W`gKTclkG9j8(k%N+=duMut!n5zS_ged<iCyS7dXG&gY9w-(N3XXLD>FN
zCoozfdDVc=+BzxIvQ|*%cH;;bw8oV8P{5>5m1Z_(C142RZjNu#vsFxJE2G=3_|$av
zO;Zjwd`f&+NL<kCR1;hw(4kfJ>oB##sz|r4Yw{{9zsrzoVbpJ>{>snvW^WVvZNLDp
z#4deI1=i^jKWOUR_(x=zrML>J<}}-53hQXIgj<@ltQa@f{*F%I)`4E{8TnTPq~{no
zR(f=(cM?N3GcEeoqJ60!gwhU;CWbN}+sCi*fl8YGzhqMhLyubLwXnYh2%DJ=fe1;9
zn)mmjXAFf<YRUGh@<{riJKd9x%Sm&Qc|~5GTqVi7a*vLR_O->I`;vPFGu(_m!S>Oq
z^-neTm6WMZ!6Owb)KdrH(*vldGrafMPk_VoN1m%sZqx2Y@&R2tJr=A<kz$FA9Itw!
zHKb1vMbcA2(2*<59p)@iwt|?7)bTfDufmfJfa~)6d}H`lq)V7$3BIhebDlz{<)rO_
z9#PA_wmJmO#ILtd-}GdBLc5pVf&nx5NEODvdjtjcCt@+CS}<fuk2)~0Bu>bw(H6yJ
z%ji<dRw@*+ol6HZCTjNX7+5F}+m5OlRJS=z?mDPRSgs=z8#oi0;H*WOTl*R64u+6m
zO|Tm1C|4QU6iHn=Dt>=O7kg^zl>V)s??gjWGb2UswV_LaPb>9Qd(`ML9<1FRwN(Qo
ze&FHwvp)O_iSSG&L?|>$$xmTa;12neAawZfo}RTtfXt`tZBzUb^O^+Up(Wyh^yVpi
z@AdU3T<&w|dW4#AzcB7Jg&X^KptB6d%3DK_OGv16s;l~UP<gh$9U%J;x-{ylSnySx
z=GipJ=?8j{#V9EQHWouH8qCSUoEE_AolmumNo5-}I4};DiRJv!Y!PyR;W=wu9`w~y
zDu<hf0Ax4WeV#hwenQ6i<HUN-GG5~S$3r(`TG-_W!c;Z-WH%vp1>v(<+|cC&<N;yQ
zf>I*zS5L_BJ`a7(Gz%bPkBF^d%S;k)Kyv#M({H*uc4h1%qUYBv4lGK<sUb$Xi#5T5
zgwgV-RALv+pAs2T6A^gHQ%8>>+V8*~jGLQ$ng0&It<a`ZW*JMX*&MlpwqW<s4v81b
z@ClMEgorNy4tr9y{ow%6Y9>YI=hqohf`_Luh~C9PFolYmEjERuP~yqK3d@lE8xLv=
zHzL0153S+Ryg4>fFp-#Ec5Tsp)IFcs$%Zop%EG?sLRyRuT=KEK@jh(@WE6@deD;WO
zwC(}mh;2&(zbu>h4;C!pf`K3+zl~s_*-#m7W+sR{g_NKPqq>Lh`^7}M2!b0lQt>~2
z7NR5Xzm~^u7nv-^XL%NP68KwMC_`vYRrkNCV)*kuP)@=fVP)1)W=1URCy(?GFU^Cs
z*<L#_ClXovt03yqYHFF;ZQQUdf$1>WzgFUIYr(?2O-}JQx)t1AvAI|nJGPRJZex(3
z8WiiPz8_&vuJH<`IOycPqbJ4qx0?No8q<GgaXvg=9b8=qJ5VX?xNZB@Z;9bLYMI|p
z4Q-WmArf!kFa<Ow8hB4uJN!-f4YiFjIqc5qzV#D)EjKPt<j3PN{NTY1oIn$bV%I-R
z!8<Y+Gh>R6`QiSt@8U^55g6$~cwR1ge5-KCq|X52|6wb14s9;(M)Y*lb@}lGoUM#~
z5#LmiX&Ym_@a$JqbS#Ln1NKtzUZkE_vv0xj<n1(JXWd+a#Q{4TavtR{sK>VbMkoV2
zRzo2jwPSyh@U7V`xA<qVI~^(l{!bH%idH@s4}0q?BWB{!>zkjPN-@vA=Uk7H&iy4#
z!<*t33$PU`(kYJrSp$bX-d!Gt=$9x^2XYpuReq76F5p;w=}ySfGE@;0ZGSxeh^g@f
zGfJBtHua0?BuDIS4zWNZQK%wzr_2GKjp&r8C(tkCBq3l0hR{dl%njv*I5K~vVfh)w
z&!+4v6BmLk5%<a-#mXg0ijAn9;7yt>&Ulkzs8$id(cr6WWB&8qmSU?47u^psr~6u?
zrHAn$(XLDxE{mazya1iLM%%<H>;^J`U&inqjD~xXW{{r#gLzqsjP>Q+$_UNtIGRW3
z=a<#(Ic_V1L{TJt`Ygi|$#h-&L`@s}U4>B*tQG}!pMCw{5bYELN*V7Ubf&NHt~}M#
zXO`X=y>fl?;l5!-tymg=Vl(a=n|dAwO?ThhG&6C3!k9>+U|-cJg-Tnx;{L{UkQf#R
ztO6QkYxUe`1VNygvpw0k)k9^Hf>4+yif=wtlP9|JNTl(@4^z^)N{0n6cN482wPxy}
zgw=2osVL7aAwZ76nQ<T!ul%fpF<n0T4rO2mED25PsV9@*4kT>&Z_9m%0D}jKcZp(V
zV*0B_;uEVaKR@NeX>F>1%NtXF*1#%u(*AYKCza;+XMd}fYys^mc(VMWnvWfx|H?mH
zvHOotOR9Uq?+HrmlRgp=Snx;1A*)a8dP4^-a;84)&kY4%x6ZcE-_o95RSK>M2SjGs
zl^-KMT$y7Zv}|zgUek#TAm@#O-6#*I@NO!vpHP+-EA2lubB=k98eLzm!k%@1JDi@J
zEq$1n&UBR)l|EBy_bxQzJ5xStJL#smUvX>4nOZ`u)Gm2L*V1*le-(W!m1I~5e{D>x
zVa1BbIyRQx>BL7wjxz<SJ`c$yhhMZWd%vJ-D5C(0kdTltkRsGJC1f&X7mR>NNC(tN
zNdKw&IB|R0S$No3zxRT9*t^(za``wrofw!nF6okW(j1Hbu=C<4{T0#W;Mg(>F1{)y
zHR<DubD|zHv{9Uc#2<~-$HrGD_vW9pB$Pft89-JcoQkbxuSQ06UV2Rx)v8C67|K{X
z(t`<Ft^&mNAbTQ06=v-VeCA>Av|Go~ST5+%`<;w3UAP3ll?cOP^eU+|N;5MZ^qKA5
zq@~&%N88VSvf2c8GCRAO3_2O$+qEnEvF&hXr6Y>1AfhQ7>b3^?Ez;U?-XIH6X>g>a
z?0&`f5y_>)tm+;eCfxhdXZt*3JlEvPY^}2ITDGGXA01Z*8;f9F9%p9@sHHzqUjh*0
zE5Aja@BePwuFk<sZfDe2Z*GhBgn7lg+%o<+P*V1eg57m7%CtA0ZW%8^R9bkxxDz|j
zM*R+nxbf8H^G+!K_d2}%rtohh>E-kDc|o+6MLgg+Zk8O|1Uzn>7zw(rB#`q~?08{2
z(sB#ws_(y-=c#USzzAO@Jf=G`geG!hWvY=|2I!eEYp28B#c_tl7u^hZdz@mXuKZLe
zi&~98#P<CJCS{XtbH0-#x5xFG?^y0pfs@&7*>0#?(Si<w)dYh3j)d)p*MnThI5!h=
zo23%g9LlEdamhXN@a_ZrHK1||y}EZaJ!DyiN)-CGi`35@X2jebpD)(!ZtG@xriwzG
zKR)Swd|HB~8(nhf`FejrMVS$LdRw-Yy$csP@w5iMxN5nc^SuYG!fR$GU%6T>tUUC6
z`myJH!ZgE8UTkkRUZqtNLKA7DR~lzUV&Dn9E-`(Bnlh>}v0CctAM8?kZYOdR^bUJO
zHU7krR;d9P4MYV?fkrigdj44+=bQTORbR+z)6HS{o*5r(z-S#C0kOCIDHZtDA=6*U
z8vVLC9v89RB(n!(Xwi{yl=NQ_ExIsb74K%m{`P&7O0Q}&ff4pId$j6FA#pzuT1uNm
zk*C)u6O&B<6)EQ>vt&x?zlas{q%ERq6B4#drZCWql~vh|7>;T7K(h4qte_45X6s>4
zSpF8ap+?I<qA4eVb*HMrxeR}QBcEQuSNkJ;T0)aTlVXe2J9Qo@r&q=qX%y%D8m$Q<
zmKc4Xxg0+_{?!1rx6fSh_(3bVk{c#CtH70E0RI&v@X1eK?Yt=_2&XXU&k*R#+w!lm
zbWP#rl;zMI!BDbj)}Ke!Oa0uPRP50y(Xoy&)%Su14GGZ+uKt@X=V$=KLZE(xYZXGN
zkRcUONr1dj<I?^`(6m$kI>7hTCBV0-QB(*q{;-+Y{*3Jk-SbkL&dx_nx}B0|D@_Lm
z_b{R8undycS3-wB0t_yjh0WS+U0o{PGZ=sEm=L%%UwhiPEfR-iJC4NHBKTHP;2E%U
z+CxGY!aMBYU&^$8t6%*<lRWBp2etRMG*O2;$zxRwo*mOna<)uq0&Uf8b_TEZ^p7L)
zzMRT*YO$x4s-Gpy@$PNk&SW=39Jw=_gPPO|9eCp^gXAX{jA?#|`X>x8NJ&Q1%!3eh
zhCAZ_Q<a)wXco+k6SSGA$X^)XJ&eTg0A@DiTUvM}^BZJJI`}Gc(tmr(A6T>j_wZ2`
z2>>e`3(SKoO%E3Z3m`kt!?PKv;0a)S%Kxo!|3nT6fJFN*lfv7<EL8vR=T<{P!ub#Q
zFa8TJ$A>?IA;?aca0sgaG8_}`#ww0nOapIV6$U85Pgv=w{=XlKghcjVqyIhY4TL8E
PDc~|}_-HIl|3Ur_`l%=u