From ac6754fb4b8ad222b56b9413b2bee69ff50eb7a0 Mon Sep 17 00:00:00 2001 From: Kevin Maik Jablonka Date: Sat, 14 Oct 2023 18:50:14 +0200 Subject: [PATCH] add plot for results on scaling experiments --- .../scaling/kappa_learning_curves_scaling.pdf | Bin 0 -> 18667 bytes experiments/scaling/load_results.ipynb | 185 ++++++++++++++++++ 2 files changed, 185 insertions(+) create mode 100644 experiments/scaling/kappa_learning_curves_scaling.pdf create mode 100644 experiments/scaling/load_results.ipynb diff --git a/experiments/scaling/kappa_learning_curves_scaling.pdf b/experiments/scaling/kappa_learning_curves_scaling.pdf new file mode 100644 index 0000000000000000000000000000000000000000..10743fb457d1d120cda55a1f6e2a8f0970469fd3 GIT binary patch literal 18667 zcmeHvc|4U*^uHo>ZCOg9JP1kLeY5XN5|J&$#l^L|uB>e;WtSvUwk(l~Rzh~EM3NGv zBqqma`T$H%qPkz7e@6XTm8Z*y4^UR!c<~--T=gizIYp!RYgjU8OWJ@1Hr>-E- z5DM~l@j|GnLCDQv0aOU7OJPua{5>F~ImLq-1Ytk}3rIr)LG^Qk8?$Z5TEvXF1 z9t^8z07#*RG9aX}FCal@D%YLLjUmSw8^kun0_$6 zDb_fe#Rf9@W*~jS906#Gq3tzKGCcZ?>2QNnGQr z={Wo1M5ME@)=o*;o0JDuC3+^-bfa%tmfo?v#wPj8<-dIjQ?I(2G5Yes*|Nyk65HN- z=kNt;YzgY2)=||pFfvl?8XQ;atv|ACXRV^O$JeS+ZksR1W$&f( z4nFm(Sc2$P6|LT%dgEq?;}`Fc@5@Zo26E4zId80b)7q}cj>Ng2+PwUYt5mi4UWo*L z@t4ov=~Vm_dRePyFM49R{Pm{|LlYT$LMq;lyj#)#(48D9erlU=ck_exiz-~+DKZqQ z!DxJb;M<~z{dGV4x4)r(+{nF+RC`B8($d_8K_3~p8WGZY{AbM4Ls!0DO#0DWG&(lI zx4meGwWwyWrUOH=s(*K)-GvkBdvHI0mWQ5Nu))2+QzvPVpC^3_A++f9fgl#u`B z1)E+h736ShQx?ysvRmJ8`HDg833&B|r{Hk!8GphMG0D;So*aiz(XMK}G?$gi_rHA( z4f_!AA>eeo|IhQszPsF8JQ{ULW8o2pZ5B6s58mb6&A5>p5U)3^=H>bHVAtY_Jiaul zk@6c(%d65g3P_X2{ih=K$!^>c`DFkZa&zZ#Ck?gZ>(cIbgd9nd|3Vb#FdG^79~o?l z*?Sus!7n4vX(!CTRrO$L+LK75#dcRtuVcta5gvEcX^^~6$MY^HJg8!g*0bFk zPYKo2b?;o=^td>#ZzDZ({qz4wp9EAPLK zX$?hIUwzl}xa{Z#NFCCMu-|kWJhpDm4&saHoby3)b!pDfzTx^XkUVLz&UhDR_|^Ir7jzYdWU7$5r_C z#q)Pus6B?uZdG1dEuC^bdh>%vM?_+mKaC!|rgN}yBoDuww`nN$VN@s_cg<#PItmqyd3GT*j|>?>AXVL=A(dI(ez}jT);S(%ka^&$B2`(YT>9-N>b-NHH7qdj;E^8{uJkV1e{@KM??|OuMG%AQx_-6a zH5ym8X9;s#f@pXjv=i^FWxqpW9J(qw2}VG!>;C zCuE1hTo>x|pFN$)r-j|>Ua(Yzf93A{OB_mZ18+rQcil59^i)LdkW5P2YdUO>gct>1 zmu4C#)|M@`8Oum-l)jUBBl-El-AijeugMSK8pL&r@WR2BGcZzsLn!x51Ky*zN?YpQva2naDyEfQ z`ceIv05>)xIH|S9wabv>;3b7KqgIdHpD(sje>cJRDPXhS_=T}2S4oyb#WvW75%JFm z{z5xazO=seluozGxLL{8$xXlcqFuZHQjdLC^ti(Hqd%x4#K0S8PF_$M*K#&MhUa9K ztrE|Dw#L$g-K}$o{b*@{Phn!uJGxfk6@G?GaE_iFQA6p)g0vJa{RahnvBL4odE&2` z>{P??Jy*~nCaGg}#J&gJ49~*IU0M8fljQ|Bd3#^B`x^@%Dx;*uTy-Q|YY$(0okzAz z<@z9)wiNrb?gJ{@?d(E%(XwxsE||1xE0ms3m$WgIy0~qz>z#V`51x^|#ooy#_ncR( zJ|{DzE0tZA(rou-1%8Lm4~IPur7e^eUbN}F(VllWjKh9~tk8a%1iQJm`z1EFI&~pK zQ_Yg zD!0w-NPI5NLPO$`AURUD=XSlcMz_d=XO{dWUU;cI+uhf44zIi9b~a-7^N^_Lmp?u} zK)G90I_jfawRd23_OL;$LLT+?-DO3tyRhrD?->2yzWX^h=WJ8R&uey`{-3Kf&-Www z8rpH(s>6k4L2G+0T^GHJE%4cKz&z^1d(Ix$ET`JGRsC{^mXupJUw;sl=tEvLtSuy^ z2RzR<4w#UM8Hn-w8PhB^R@ZV?WguhM`iAq7|L}OOi#|~{WHl@ha&y=}`$&@iPPOs( zWB%Fx%4W8GP8&6pG}U+7?O40DX;nwrIn_t2I7;73ts2{dbL@%JT-S@FZdoSRJ9av} zQOoR0(6Uj9EYUYd#BMFWGU(dV9(6zK+p_)D1kWp;N!vCtjs?gCkh*7JYq4LKvC5Uzx~ij+RTB$uYmuJ9AZ-d&Ou;oK z1c6}*293c^V>l*ufh+AH$LTr@49sTX7cv%LmcL*ZeA{KgIu!S!+gS?Bcf0NV{O+5> zlW}{VeVqM*SH)#=U#w~0-OuhHEj_^1>ys$d)VJx89IAOl$-H%c8|{_F(W<3~-bT?3 zV`85PTC6wU#>f!rR|)y_;UCd=s|4n9gMs1VEFwvmzY~eQh_XhRmvIOdMyDa%)0QC= zq8*^Z-8#zY1mdRF0Su04h>GXCwQHTQiHTcU!GaCT+A-LLFWJQ7oEIpy4lL~D@>s4# z)nahktY90{kx*kn8H{^pQAQ-pgEBPA9Q?5hw#aEk7wprCl@dKb%Sq-byu~L%uZZ6e zzl0Yn!vpb@O4Og@C~r#^b0+Ir@~mFg!;NJp`LcEM-AvqB#G?bvD&?D2b5aj!)E{dNFMHS>WQ(j2#L88_d7|#3zCPS# z3pC_N^>C%R|9FmQtGS#VSMwn!(t9PLJGnf!=Z<~~zX8NJV8-{O_f^G`HeW9#nNLHO zkB!70528lDNL*==ynW(^{@$gx6)$ikolz?m->R#vuyk$G$^q585&tA+?-Y4zu!Dt2 zSfF$aSVR0(&GnOw{T4~_>U}xZAbQ~0dyZrKUns`4k6$i-WMp^yQ}2`aPrRDk44l}= zs+H@$x2$pd?t6Sg2P$r-`DK%3>s1Q6nyg=my}j?L_&^9Df5iNFI(qXV0-!J>B_ImXtSEslNF+jcoVs1D8HyW6E6FVBrR{+eX~%$lPl`{kjq>l>IKFOWxw9Wi(Dr!xsqVPqkT}bMvX>_7 z4AqkxByjl&-8way$79&EHj8jkdL zX7Oew{5sm)eL=BX>)gJ^^A3l#RNVgIue97o>4GT*c}-`vU&-DCUoL^AHy@A4v=1Gi zy!{cnSyXwksi#Q)(;{=8jn|YJsR1!r8Qm+r(4H%llD}c(_8-nYT^@8U;YesHWyeCw z>!IAN{u*PdY*%NN+Qa(~)o(TVl~E$swTb&=U(XBW>)!El;+ZTmY{!D8Wb|1U%)#r< zIWvj>J9GLoXHstm@3H+}cNc+~G(K9Qs7YQ-AgHR84r+5aMw_T+-A!#958*jAc;WGj zXV1*m9@EpubZ1vWI;gBCot7ad=xQc*Iz0^`wo&N@`+5W(ncfieZ{0EY!hdW-Fv;rO z1)IU^%RAGo7hY#W7h!180!B$u#wLQNyfm_%K53oL|0m69U##O&*E6EgXTE!|gm=muZqJKRcRRFCyDTc}dJ!-}dvJYxU_9ggUPiA_TLRMqa-K zQA%p*1FHibguYK+I{s1Yiu)n5w!DvGx?Ca!Q(qh@kf7y8^!OI-bBfN}U$yIS{Zq$% zr&s7-v9wFvnQASX@ob=*wsw8@9mmwdaf){D8Y$lH(>Wm@U8}?SGHVR_^;;_0H`x@d z2~N?YzHrdAc%LcVwg1A;EZeRQeIv)+u^2>E3 zvcp8j)}}>k>RS%svXw7cx||HvquBD)}|`lx`NXV}Wqc#Ch3%Q+^#Z_uen+%?@>Y7xZehe`t&2 zshP0WIuYl7HLW(+gd?gZPzG&U_O$MJqb0&=&xmP<;DQtK0X!Ryj@(*ym)o8*;Xyb* z4~^czU0@aFmTz1_5v(+?eEjrG_s5jFn!1mM9as;}JMz@~4xDO8)!tC6H0z-8({$T7 z_fc}JkILG5CC3rNpreL|FivjUEHj5v?K8d1n(y>hXg)U^M<)f=ETB}Y7qQ@(1z;i* z|859O6Y2%)Aa0HcYwg95wy9>uhE=^2M2FjG$({k*HEm7#x8m$_>4Yn*SGsE~cM?5) zSGd76L)(a`{l@sPDE@}@0*co~$#)-bOJ-LTjc6X+?V>much}fkpnZIy$NklDEC^+R zpRgbR_gBYknsU>Z9nGsZX1_?>R_h<*&FcK|60d)r&B%3GUFR~a{WN>MqL`5K<>K|d z$)yTT;eunw-up#awWY3jdUZ|G&C-Ejg`Zz>>r1w5Ch{J&(7&nY5%W^SG44!a*HV}A zTseZrtHJlNh3T<O`CIi6B@Z5HE^916O65n$XmL} zg=ptK zcmGl^B-;ATB&d<^~S`e!;-Hw8v=?W{8#DXdPBv1ghSo) z*85%2?tN77J6LALl!Ph%GFgwmnxt3e$myWhLbGtFSj+tVV|Dr3O4CNsGAO* zc}QhDTWp_(hq5&Ga@ljYlK8lrw}kw&*QPdxe`xs|G34FA>__swm>02oEJ`YzqFuO& zPWeZMq#tMyxa^eL&$~?M>MD~Zp=c2f76h>ZH}f&$rpeku3(?#hV_}QTjM6z;BQC5H z_&#xp-OdE5>PVp(HAp$Syc-lhhkK^cTc#Fj%(_HiWkp0v1*TT^v~WN7zu@gw687qgt)0u)qO z(9aSl%)?gT6IF%abx_2{0huSrfUH|?7O(qj6FAQEg@2Vf6@Ek+|2Dsh_s#oGgLqUJ z!7BS}vtREP`?G~%+*JkoG3Dl5p}d`pazuIHN3X0P%er>e7Tx9$`8+MT5Z8_+*Ol~t z2(6S=bI0a{e6U)mTZl0W3)=eT{xR>9>bFAK^2MG8`6bX)v7Vg=Qc*)bWNFvT*z}tO zw21u4hh+52pzOMU=j)I)moo-`aJ`AvOlQG23#@>~&ST!x6D#?2VQgudyJ(^9xo58M z%w>=F9cVIJt6{5l>;$bV{X44b!kL#Q*>YD`WNQdr`gv`oNJoL8*1#DQw2%AafqjQc z(xSuU5hV+3M1IO8R+EoZ-QyLuKOKS(Dp6I-KOTNC&&teBMm{3P@cZ)ox5iu|8Cy85 zI`nn0)p&-$pukxqrz-j<3!YfO3evprPbD9SYU#sfmtVY9V^eCwgId0vHT1@vcQjGb ztB*N&ka`6}F(N`5r!@k0_i`@X0(nVNBW`SXA|b9!4L>Sogm`q4D7~YgRWMuTdz=poxo}vcD^> zLQqYnul7~u*J!t{Il438UIhMDOs6-k8@jz!7BQCP7N~j1L4py^TXr}cpPBh^W!lw? zufA{owr@f%X6=j9#|}kwBRY^9A84^M-7K6ic^)s8eu4@AE-h{lV8|0im4DD}$F@`~ zAFEw@AVz#EyRX8QGF?@I5N=U1NugB6CdoW$!dA%FdHn%DTZK~TNWnz6fY^&7TSq;z zN3))O{E7(Av_JRb2Vp4f)uJb_#dpn}6+8=PO`ZqRF-_3she7k$RhR51vC>zIOuP1s zPh8Byzg)2GT%Gh5K?)V8Ah<8~(C(2IZ1!kr_f9j6^K zV}!5q$z5Dl!GfxpzI!^-M4X2WQcu$M)76HxK8j~^L*3hnV;&_Z<39fEb57YtwGtv??K>g`k&0`smI(V^`qaImO!KacX2PRvsf&H4jT(?4 zIb4_&y=z@rhGF)wy_axum7}|9)D5Cxi`at~p&1XNcEnzP8@hmdRMUk8=PZB#ejdgR zdx;sX-+UX7!^3#n0!2id1I&F6wU>K{Ue6I!5#Pp_v+m?~iHdU)XL!%x*d!nE?`APR zD@Q&LeZXP}mZ)8gBD$dGmG{^;vFMqVt3=IXPEFgZ`05~D4zOFIw$Q^z8=b_vxLIzQbjE^|q!+BiFCZ8Eo<2 zaO@@9EBeLVNzndt${ex>Y>nefg01{U1HW)JyR3JWVoy=@Rd?)rueXFf{l)N9+PYfz0BkrlTKVAVfvJ^JH_UH{RWa!NJa zMIQJ=Cql8j?&E}nExe&micFjKaR}9Ki193bdM@{M=k2GAHMdsy`2Q3Q(%MKo`qG$= zJ!gKbp(mjoRjn_fJML_?H|R?}%HO`M38jkf^L_{AJwMGT-8&Y|;d*-)s@hr6R#r@b zT)x!h^@`V>Ds2yPO5#-F2#2hRAC_=jtv#)8kg%I3nG4QL1H<^!M7B>e0nvbFDA72F98h|@b@ui(}aB{p~!p9&1+4qH~th+ zyc%HCH~wtRX0Gkg!OsDHj6L3Vjv z8m68_zh21}*($2ff+!Xa5{;dQ*-}pgJ3Jr?!E{AYd(DEfHd~2y{qPsxQ>du9+m~B# z8UMikwN;CY5Vfg=A77X(`{p#_dHhnSa4S}fzbSdQ-yL#X9$%n*o?@B7&~~@3B7V-% zmlnHv4|ShslQ)@ouLzKD#aMV zxg23;I@igCG4oI&Qil>moh<2DXzLi8DjMsUn)^^gEh&CMuSEupwpPA z50z0USXaQ`uaYAI2@_`y4n=|kn6MWGCl>Ynz_~&A$mH*%n$uMXq#n)PoeB;-!Y3T< zAuQPA4GN&Rg2R^pU3-QFcvAf!WB`>;^LGPxd?-Pl5Hg%f_lJ=Fet-=|2q-Z;=~Pg5 z2j?Wgf7&)G0vQA-Lg0FVjBqk#+He0?A!h*E$K!H@wUE&)QkV7wqOK73xvj^FCwZm=pOw%@WW6ABsm15Y&9;OFPIFK1zy4 zfin_FcnJ7P2}1-|x|9H;Ny(rPNZaWmkO(+KOLNus^YEbp%8`}~s;@O%Fr|btS%T15 z6k*0ORwjcmKlTugi247K5dW|JV6X%TL&8Bg5V9eH5G6PWiy&i&V7hTY7cmePi)SVr zSSArrCR?q3pO`W}Xif#5WK4d*yuB1vR4L;xWx0v_}QApkV|cSY{4lu6Po77SKlk%K!q3JlQvs zC(s`W1!2)d2n+fKIg>6t&}^6vJlrq1j|b|@%yDF(E1(WTg75?qgaN$62@m7~^G^bA z0bGv<-U_rwF#9F|Izg-fiw87-dpN)=5pDz159EadTkBY`b%zD70GvBi2G1Ra z23<@|UV=%{)b(Y|joh(tnOEpVk58x)1i+EGQ6vKq zc!n%`eq2WL7k}JXBoO2H`d*kIj^8vT!1{vey&5%1>wIqOLmxRJE#6pGpgHN3FVMmK zEg3QskMj4Aiax4bzG5!a0;X{`SF@p(p{|Xsp0%PL)r(>s3?nTiGZ!CPU@(9cGY}Wz z55xs-{fW51TgIH7U;7$N#D$nbT>mx%=w}*uv6Sbeum5-C1zh?s{*`9{)F$x?8JL9`;03(Y{~PcE zUcnh`O@nZIrhS8L?EfW5)foSEbRASz=kEmz6#i6 zKqGt~ZjS|TGcyO~4YdEa1OUh6eJp4TC%BL4ga<|zje?m6=!pP3O5os#L~spkE*=a{ zgcC4Q7&0CK5^#tF;L$-RKv9Um+7SU-AOHu$%;EdM3=_cNL?U<|?vqRc2naZPxDL*l z%wUNKIERUVnSv7mcFN#cJYamX56~8j%Pa#sj{%MkPIw}W+DHJl1A0L{aP)w|$#HRD z3_NhQSQw$f4iV%eAS^g%Cg6==VZcIwb9j8_WWfRgGys7j;4$&w8KzLCiGy=Ac8W6C zU87O3h+wi|*5Di@=6t}zO(q0>YR-OjHCX@zPIdvjAN&{|rV3CFKMt$G^mI+f zA#jnqz=G(@MmJm z8t4VA6EhN1A7GuBPE1{y(g|1}W+s>}<~8U8){W`J)C&{|%psr^)&UlY48~yU1sayr z?=-Cw(>W9L0T9H@VE{NIG3Oe-hIM8}g7t!_cOd@-VwgsB%I5tx8PgU6UdtMQZ{1CK zYXi{1q;vk=bf(SJM$fdtzrJFK)YS%t73oTW(6CuTYGeM#W?*^wyBYQ`%}~l1*q;BU zWrhp$CZ|A2~3`8QJx1mr(iiv@0fs)&PtI!@Mty?Z7E0^e37 zKwvdCSqoMiQ#ix`tu!O5p=3ez2=<}Sm3(P_!9fsMQGoWqLtvEV`eP+;NW8*(8i;~^#Z-*oy7ThLf682>NdAMi~k!-t3UBeJ7-BHiNHbXBQmw1xGL z{1A=Lq{*wcOX_x<#1A^(4{*wcO>5>1F0|JhO z{3(XXACW+s(A?mS2>{Ju5Cb5Bu75BbQlEjXK*Zzs!@A&?5#a5YHDI+K;N#Emp}9cY zu*xK5)H=wM!3YRaK_Y#B-BR|adnkaDEOdXjU{~-%3p1YzaC3)TD6Zb%`;b#TfR@be zY5snC@DEh1(Nh86io=4C0a)~4iFlkM3L}p~$t%ntE|gAnM*vh!Mt}w0)CWLSJXkHc zL%-@^HvxYjzo|MjfeaQdv+8j0qGNU)9BiFkM}Wgt@aK1bAU+B(~TQ@7)Srn z4uyh)v9s#H8PB=xFd%3;ryU%+`=dWJh#${xhX!l_9r&{y9tQ%ev+D?8(Kow}4Av5} z>Ii7CR+?RhgClP6=eJx4L^vk;XB`?1hwlEYLy_T-;jB6^A+zf+7!Vl$vmMy|0SnaG zbwsd|on1$s`z#g(2PfgrFPc$UEF8U^RR^!(X4jG6Na&yKfJkTahQeV%fP8j4JiHE= z^(<+QTu@*$WllS=xiN=k5U-vyF4$L?BLlD)pF<0n%Q<5ZuwX4ai$1V7F=t$$_H$$p zHl*f^K}3V~?X3QYxVf|t@$kCzPq}~?(cEXrb7er9tK%pV4y=b~^+&?b(Nh!&-l>}1 zjtmFQ|D=VCf@Aly>M-#4wrADhz^>fvI^tX%0%mA7Zy?kGHiZ6s7Hn~Xc+RXkEZ97n zQ-=ndPqW%#NN~^`{`}H$G!6%c?fIEX=^GnghCO$0ySgOHWgH#0!|A0w#rF#rGn literal 0 HcmV?d00001 diff --git a/experiments/scaling/load_results.ipynb b/experiments/scaling/load_results.ipynb new file mode 100644 index 0000000..e312523 --- /dev/null +++ b/experiments/scaling/load_results.ipynb @@ -0,0 +1,185 @@ +{ + "cells": [ + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [], + "source": [ + "%load_ext autoreload\n", + "%autoreload 2" + ] + }, + { + "cell_type": "code", + "execution_count": 39, + "metadata": {}, + "outputs": [], + "source": [ + "from glob import glob\n", + "import os\n", + "import pandas as pd\n", + "from fastcore.xtras import load_pickle, save_pickle\n", + "\n", + "import matplotlib.pyplot as plt\n", + "import matplotx\n", + "\n", + "from gptchem.plotsettings import *\n", + "from gptchem.plotutils import *\n", + "\n", + "from scipy.stats import sem" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [], + "source": [ + "results = glob('results/*')" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "metadata": {}, + "outputs": [], + "source": [ + "all_results = []\n", + "\n", + "for result in results:\n", + " result = load_pickle(result)\n", + " compiled = {\n", + " 'accuracy': result['accuracy'],\n", + " 'kappa': result['kappa'],\n", + " 'f1_macro': result['f1_macro'],\n", + " 'f1_micro': result['f1_micro'],\n", + " 'fine_tune_time': result['fine_tune_time'],\n", + " 'inference_time': result['inference_time'],\n", + " 'model_name': result['model_name'],\n", + " 'train_size': result['train_size'],\n", + " }\n", + " all_results.append(compiled)" + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "metadata": {}, + "outputs": [], + "source": [ + "df = pd.DataFrame(all_results)" + ] + }, + { + "cell_type": "code", + "execution_count": 22, + "metadata": {}, + "outputs": [], + "source": [ + "import numpy as np " + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "metadata": {}, + "outputs": [], + "source": [ + "df['kappa'].replace({'None': np.nan}, inplace=True)\n", + "df['f1_macro'].replace({'None': np.nan}, inplace=True)\n", + "df['kappa'] = df['kappa'].astype(float)\n", + "df['f1_macro'] = df['f1_macro'].astype(float)" + ] + }, + { + "cell_type": "code", + "execution_count": 34, + "metadata": {}, + "outputs": [], + "source": [ + "grouped = df.groupby(['model_name', 'train_size']).agg(['mean', 'std', sem])" + ] + }, + { + "cell_type": "code", + "execution_count": 49, + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiwAAAEeCAYAAACpLU7SAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAACs2ElEQVR4nOyddXhTZ/vHP/E0TdqmLtRb3N11ONvQjTnMjfneubv9Zu9cmMKAwcbYxnArru3QFihUqLdJ2saT8/sjNDQ1rAXe7XyuK1eTc84j5+Rpzvfc9/3cj0QQBAERERERERERkUsY6cXugIiIiIiIiIjI6RAFi4iIiIiIiMgljyhYRERERERERC55RMEiIiIiIiIicslzToLlja9/4rc1m7yfdx88zNMfzmbn/sxm65iIiIiIiIiISA3ysy1gdzgoN5iIDg8BYPW23azbnsGMK0eTEhdd7/jDOfn8sX4rxeUVBAcG0LVtMsN6d0MqkQDgcrtZtWUXew4ewVhVTUx4CBOG9CMuKhy3ILBhZwbb9x46uS+U8YP7EBsZ7q3/0/lLSIiOwGKzs/dwNlcOG0DHlATW7khn575MTNVmYsJDmTCkL60iws71OomIiIiIiIhcRM5asBSUlCMAESF6Fq5YT+bxPO6efgURIfp6x7rcbr5bsoIBXTty7fgRlBtMrN62hzYJcbSKCAXgzw1byTqez9RRg/H3U7N97yG+WvQnT9x6Let3ZpCReZRJIwaiD9ByMDuXT+cv4YEbphKmD/S2s/fwMYb17sqQnp0J8Pdn+eadHDqWy9SRgwnU+bPvyHE+nf87j8yYRpBOe+5XS0REREREROSicPaCpbQcgKUbtnE49wT3Xz+5QbECYLc7sNrsdGuXQpg+kDB9IG0SY0/tdzjYtGcft0+dQGJMJADjB/ehb+d2yGRS1m5P5+ZJY0mO9Vhu+nftQE5BMWu372HaqCHeejokJ9CjfWsAHA4n63dkMOvaSUSFBQMwqHsnDmXnsu3vg4zq3/NsT1lERERERETkInPWgqWwpAwAhUKOXCZjx75MYsJDGzzWT61iQNcOfLbgd9omxBIbFUbbxDivlaPMYMLlchMeHOQtI5FICNUHUlBSjtPlIjLUVwxFhgaTkXnUZ5tMdioUp9Rgwuly8cm8xXDS7QQeIaNWKc/2dEVEREREREQuAc5asJwoLad1fCtuunI0v63ZxNaMAwzr1ZUArabB468cPoDhfbqx/8hxsnLy+XP9VmZMHENSq6gm26mlNc6J26dNQKNW+2xTKs76dEVEREREREQuAc56llBhaTlJsVFIJRKG9uqCgMDa7XsaPd5itaHz19Cnczuun3AZHVMTvRaSkKAAZFIpxeUGnzJ5RSWEBAYgl8koLK2o136Nq6chQoM85arMFoIDdd6XsaoajVp1tqcrIiIiIiIicglwVoKlwlSJ1WYnMtQjGIJ0Wnq0a82WjAOYqsz1ji8qq+C1r+aStmsvxeUGDufkczgn3+vmUSoU9O/WgV9WpZGdX0hxuYEVm3fyybwlmG02hvXuyq+r0ziSe4JyYyWb9uzj76xshvbq2mgfFQo5g3t25pdVGzmUnUtJhZH1OzP4/OffKSoznM3pioiIiIiIiFwinJWPpKDEE3AbFRri3Tasd1d27D/E2u17uGJYf5/jI0L0XDd+BOt3ZrB803aUCgU92remd8e23mPGDezDKsUu5v+1FmNVNdHhIdw2ZRyBWn9G9O2OQi5n0coNmKrNRIeFcMe0CT4zhBpiVL8eKBVyFq/ZhLGqiqiwEG6ZNLZJy4yIiIiIiIjIpYtEXK1ZRERERERE5FJHTM0vIiIiIiIicskjChYRERERERGRSx5RsIiIiIiIiIhc8oiCRUREREREROSSRxQsIiIiIiIiIpc8omARERERERERueQRBYuIiIiIiIjIJY8oWEREREREREQueUTBIiIiIiIiInLJIwoWERERERERkUseUbCIiIiIiIiIXPKIgkVERERERETkkkcULCIiIiLNyNDtj/BN/vIWq3/G3rd4PPOrFqv/XDhUnUubtJtbpO7nD3/H9IxXGt3/wMFP+L9jP7dI22fKjL1vce+B/553Pd/kL2fo9kfOuXxLj72mWFueTsL6G1q0DVGwiIiIiJwlQ7c/gnrleIJWT/J57TYdbva2EtbfwDFLYbPX2xCX7XgM6fLRZFXn+2x/K3s+AasmIlk+ioBVE/kkd4nP/t+KN3N5WN9m6cPQ7Y+wtjz9jI9/r+1dPJQw9bzadAtuHj30OcoV4+i48bZ6+wtsZVyV/jLha6YRvmYa0zNeochWcV5tipw9omAREREROQdeTpmBYfgvPq9uASnN2obZZcXkNDdrnY2RYynmYHUud7Qaz7cnfJ/SH028ioz+nwJwdNC33BV7uc/+JSVbmkWw2N0Oyh2V513P2WB2WZmW/jK7K4/wRutb6u0XBIHJe14kWKEjZ/CPHB30HQ63i3sOfHhB+ykiChYRERGRFmV/1XFG7niMkDVTaLfxFr7O/wtBEICGXQAJ629gbXk6f1dmE73uGiqclXTedCfXZbzmPabaZWXynhcIXTOVXlvuJaPyqHdfhaOSm/e+Q9Ta6cStv45HDn2OzW0H4JilEMnyUfxUsIbOm+7g0UOfe8t9V7CCmTGjuCf2cr4rWIlLcJ3R+ZXZTRyszmVAUAfglGvgq7ylpKbNIGrtdP6T+QUOtxOry07Q6kk+Fhyb207w6imkVx4hfO1V7K06xoTdzzBo20PeY5xuF7fte5eItVfRfuOtPhaYGXvf4vnD3wEeS8nTWbNpk3Yz+tWTGb/raXIsxU32f4cxkwhlEEu7v4Jerqu3X0DgheQbeKf17ahlSrRyP26OGU2aYZ/PcTa3g2syXiVszTQ6b7qDZaU7Tnvt9lZmM3Dbg4SsmcLIHY9R6jD67G9q7AiCwEc5v5G04UZi113Li0d+8Cnb1NgCj6vtxr/fZNaBj4haO53WaTP5/sRK77EuwcVrR+eSvOEmwtdMY/KeFzhhLfPuz7UWM3bnkwSvnkK/rfdzxHzitOd7voiCRURERKSFqHSaGbnzcaZHDqVoyHyWdHuRV4/+xF+l209btpMuEcPwXwDI6P8pP3Z+wrtvhymTt1rfRuGQeXQPSOHO/R9499209y1kEilZA2fzd7/POWzO59mTN/QazG4bab3/j7fa3A54bn7fn1jFrTFj6ahLJFYdxpozdMv8WbqNUaE9kEtl3m1F9gqCFFoy+n3G1j4fsKpsN+/n/IJapmRqxCDmFq7xHvtX6Q7aaWPpokvGMPwX4tTh/N7tJTb0/j/vMXsqj3Jv3BWcGDKX66KGc93fr3tv3LXJqMymwF5OWu//I3Pg10Sq9Nx38KMm+z84uDMft78PhVTe4H6pRMqo0J74y/282zYa9tFaE+Nz3FbjQR5PvJrCoT/xaupMpqa/RJ61pNF27W4Hl+9+lonh/SkcMo/P2j/Ahoq93v2nGzurynfz4tEfWNTlOY4O+o5+Qe3YX3W8yXOtyxbjAa6PGk7ekB/5puMjzDr4EduNhwB4//gv/Fq8ibW93iJ/yFx6BKQyOf0FBEFAEASuyXiN1v6tyB8yh4VdnmWb6dBZtX0uiIJFRERE5Bx4+vA3PvErSRturHfMHyVbiVGFckurscilMlI0MdwdO4Ev8/86r7ZHhnQnWRONXCpjZvRodpqyACi0lfN7yVbebXMnWrkfgQp/nk2+ni/zl/rc4GdGjyZA7u/9vNGwjxRNNPF+EQDc0Wo835w4s+DNhtxBYcpApkQMwk+mIs4vnMcTp/NjwWoAro8awZyC1d7+zClYw/VRI5pso39Qe7rokpFJZNwSM4YTtjIK7eX1jusakMxXHR4mTBlEmDKIayKHkVGV7d3fedMd3u9rwq5nzuj86rLDmMmXeUt5JXWmz/bLQrp5+zghrC+D9Z1YWJQGwIRdz3jb7bzpDgC2GA5gdTt4OH4qCqmcJE0UV4b389Z3urEzp2ANM6NH0zUgGYVUzsiQHrT1jz2rc+kb2I4+Qe2QSWT0D+rAdVHDmXPye/o870+eS76BWHU4CqmcJxKnc6A6h71Vx8i1lrDJsJ/XUm/GT6YiWh3C1RFDzul6ng0NS0oRERERkSZ5OWUGjyRMa/KYHGsJuyqzCFo9ybvNJbhp5x93Xm1LkXjfR6r02AXHyfaKERBotf5a734BAZvbgdFZ7d0mkZwqD/DNieUsL9uJeuV4TxkBZBIpxrbVBCr8aQy728Hq8j180f7BOv3zfRYOUwaSZy0FYLC+E9UuK7srD5OqiWFZ2Q4+bjer6fOV1D7fYMDjgqlLga2MBw99ysaKfZhcZlyCm1BFoHd/Rv/PmmzndGys2MdVGS/zTcdHGaTv5NvHBs/ZY2H5vftL9eo6YSsjRKHz+S5q13G6sXPCVkZ7re84kkrOzgYhrTMOwhSB7Ks+7m3/qoyXkUtOWc7sbifZlkLClUFoZCo0MvU5t30uiIJFREREpIWIU4fRL7C9j3ujNiqpAuvJ+JIaBOq7Os68vXBkEilFQ+ehkirr7Tc4q+ptq3ZaWFS0kaKh8whWBHi3X7H7WRYUrefWVmMbbW9dRQZddcn1RE3d+JcSu5EYdQjgubFdEzWMHwtW01WXzFB9Z0KUATQHj2V+hcFRzZY+HxCtCmFtRToz977TLHV/f2IlT2Z9zbzOTzFQ37He/obOuasuudH6olUhlDkqEQTBK1pq13G6sROtCqHUbqrTB7f3/ZmMrdrHA5Q4jMSoQr3tf9RuFiNCutVrO8dSjNllw+yyekVL3bpagkvCJZR+6AjP/nc2//m/z6m2WBs9zi0IrNi8kze++okXPvmO2b/+hbGyutHjRURERC4m48P6cMxSyNyCNTjdLgpt5VyV/rI38LGrLpm9Vce805b3VmZTYKvv6jhTIlXBjAvtzRNZX2N2Wal0mnn56I+8enRuo2V+Kd5Ij4AUH7ECMDl84GndQo3NDip1mFhUlIbVZSfHUsxr2T9xXeRw7/7ro0awuHgzPxWurecOktSt7CxwCi78ZWqC5P4U2w38WrzpPGo7xdNZs3nr2AI29P6/BsUKwNqKDP6uzMYluPi9ZAvrKjKYHD6w0Tr7BLVFJVXw9rEFONxOss0F/Fayxbv/dGPnmqihzD6xjD2mIzjdLlaV7eZQda63/JmMra3Gg2wzHsQluNhk2MePBau5NmoYALe3GsfLR+dwwlqGw+1kcfEmJu/xxLDEqsPoF9SOJ7K+xuqyU2ArY37RunO7uGfBRbWwCILA2u3pbMk4wPRxw/nm12VNHr9+RwYHj+Zwx1UT8FerWbczgy8X/clDN06tZ+IUERERaUmePvwNLx+d47NtTc+3mBw+gEczv8DqtnNn7ASW9XiN+w5+zL0H/0uMKpTpkUMZou8MQDttHK+mzGT4jv+QoI5gdGhPolUhPnVeGzmMHlvu4daYsbzR+tbT9uvbjo/y0KHPSNpwExqZimHBXXivzV2NHv/NiRVMiah/Y708rC+37X+XrOp8fi3eyEsnzzVpw0280foW7mw1gSXFW1jV8816ZSOUekxOM502347JaeaGqMu4P/6Ua6OzLgk/mZItxgMs7PKsT9mrI4cyJf1FRof0ZE6tQOMz4aWUm5ix922i1l1Dol8kH7S9mx2mTK7LeM0naLk2Vpcd7eorAM9DsYCAfMUYAOZ1foopEYN4JXsuGqmKrpt9r+OPnR5nfFgfwOPqeuf4z/xRso1wZRDzOz9NnF94o31VSZUs6fYid+x/n9ey59E7sA2jQrp7RYtOrmly7FwW3J2nk65l4p7ncQku7oydQHttvLf+MxlbfQPb8WPBaq7c/Tz+MjXvt7mL3oFtAbg/fhJWt4MB2x7A4rbTK6A1b7W5zXuvndvpSW7b/y7R666hnTaWW2PGsqx055l9UeeIRGgo1PoCUWW2sGjlBiaNGIjD6eL1r+by3F034u+nbvD417+ay+QRg2id0ArwDK7Xv5zDNWOHk9gq6kJ2XURERORfzd+V2VyT8Sp7B3zhs31teToz9r7NscHfN1n+kUOfY3RW80WHB5s8TqRleP7wdxyzFvFNx0cvdlfOmItqYdFq/LjxilEAlBubThbkcDopN1YSHhLk3SaVSAgP1lNcbmhQsOw+eJg9B09lnuzaNoVubZs3sZOIiIjIv5ElJVu4PPzck8WtKd/Du23ubMYeifzT+Z8JujVbbQAo5L5dVshlVFsbjnvpJgoUERERkRbhyaRrzqlcjqWYXlvvZWRI93ozbUREmuJ/RrBo1CrAY2mpjcPpwl/dsAtJREREROTCMjS4S5PuoDi/cIqGzr+APRJpiOdT6ucNutS5JGYJnQkKuRx9gI7icoN3m1sQKC6vICw46KL1S0RERKQ2Y19dyQ8bjp7+wHPkjs838+y8PS1W/7mQWWCi23+WnP7Ac+DVRRnM+Cit0f2P/bCTD5ceaJG2z5UODy3m1205513PHZ9v5tVFGedcXnfjHI6X1J/KfiF4dVEGd3y+uVnrvKQtLOt2ZJBXVMJ14z3T3vp1ac+yjduJCNaj8VOxfkcGKqWSxJjIi9xTERGRfxNjX13J9iOlqBUyn+1/PN50xtZzocNDi/nziRHEh2mbve66XP76KtYdKGLXGxNIiTw1zfm9P/bz5uK9VFqd6NRyXry6G7eOSPXu/3NXHmO7xTRU5Vkz9tWVPDmpE4PaRZzR8W9c3+O823S7BZ6Zt4dPlh8iJVLHttfGn1G5/HIz/Z9eSnmVjcrvrj19AZHz4qILlo9+WkxuQbE3nc2Ln36PBI84QSLxsagM7tkZu8PBJ/N/w2Z3EBsZxi2Tx4pTmkVERC44z0zpwv3j2rVoG2abk0pL/YyuLUFuaTWZBSZuHpbCnLRsnp3axbvvgfHtmdQ7jo4P/0bGO1cQqvN1wy/dk88zUzqfdx/sThcV1fbTH9iMmG1ObvtsMyaznRev7sp3646cUTlBELj3q63MHJbMO0v2t3AvReASECz3TL/yjI+VSiSM6t+TUf17tmCPRERERJqPg/lG/vPDTvYcKycsQM3949pxw+AkJBIJP2w4yo8bjrL0ycu8x3d4aDGf3taXYK2K0a+swGh20O+pPxnbLYav7hoAeG6y176/no2HSkgI8+ejW/rQMU4PQEW1nSfn7GJFxgkUMimT+sTx3NQuqBQyjpdU0fHh35h9d3/eXrKfER2jeOUaTybTuRuzuX5QElP6xjPl7bU8NbkTMunpowbKKm1knjDRNzUMgA0Hirjziy3858qOvPv7PqqsTqYPSOS5qV1wuQVS7lvE2udHey04NoeL1Pt+4Y8nRjD21ZWYLA6m/d86OsfrWf70SACcbo84+HN3PsFaJe/e1Mtrgbnj883Eh/rz5OTOuN0CLy/K4JdtOZSarPRODeO9m3oRG9r48gK7sssJD1Dzzd0DmLf52Bl/r7PXHkEQBGYMTWlQsBwvreayl5aTecJEl3g9783sTXJE/dWga7N4ew4v/JyB0Wznih6x9RZ4/GNXHq8syiCvzExKpI7Xr+tB7xRPZlqL3clTc3ezaFsOYQFqnp/Wxads7esEeMdCjWVo7KsrGdcthlV/F7Aru5zkSB3v3NCT7kme3C1NjSuAtIPFPPbDTvIrzAxuF0F8WOPX/Fz5n4lhEREREflfo9Li4Io3VzOlbzxH/zuZ+Q8N4Z0l+1iRUXDash1ig8j71LNW0eZXxnnFCsCu7DJent6NIx9OoktCMPd/c2r15zs+34xMKmHPW5ez5dVxHC2q4uU6cRBmm4vlT4/0ihVBEJi7MZubhqbQvlUQMSEa1u8vPqNzXJ5xghGdopDLTt1Oio1WgjQKNr8yjjXPjWbd/kI+Xn4ItVLGlb3i+HnLqVWFV2QU0Do6gE5xevI+nUZsiD8LHhriFSsAfx+v4I6Rrcn6YCJX90/glk83Nbha895cA4UGC8ufHsmuNy8nIlDNoz80ncxsYNtw3p3RC4X8zG+Hx0uqeHPxXv57S59GM/NmHC/nm7sHkPXBJHqnhHLdBxtwuxtPe3bohJE7Pt/Cezf1IvP9iVw7KJFth0u9+3ceLePuL7fy3ozeHPtoCk9P6cy0/1tHsdECwFu/7WNfnoFtr45j9bOjKDI2njW+MTYeKuH/burF0f9O5oZBSUz9v3VUWT0WvqbGVVmljenvrePBCe048uEknpjYkY0Hz2z8nA2iYBERERE5B15amE6rOxd4X50eXlzvmGXp+UTr/bhpSDJymZTkCB23jkjl2zN0OzTG8I5RJEXokMukXD8oiT3HPCnXiwwW/tqTz+vXdUerVhCoUfL4xI58e9IaUMMNg5MI8FN4P2/OLCEpQkfcSUvEzcNS+DHtzAKHl+7OZ2xX3/iVUJ2KK3vF4aeUExvqz0MTOjB/0zEApvdPYP7m497+LNh8jOn9E5tso09qKJ3i9MikUm4cnExBhaXBG3LneD0f39qXsAA1YQFqpvVNYF9uhXd/36f+9H5fU99Ze0bnVxe3W+DuL7fy9JTOtApp3IowuU88rUL8USlkPDm5E8dLqtiXZwDwGTcPnhSbv2zLYUzXaAa3j0AmldIrOdRrPQH4bt0RbhicRO+UUKRSCcM7RtEnNZRFWz3BvQs2H+ORyzsQHuiHzk/BjCGNr2PUGOO6xXjH1c3DUwnRqVi7r+i042pZej6pUQFM7ZuATCqlXasghnds/mSuF90lJCIiIvK/yJnEsOSWmdlzrIJWdy7wbnO5BdpEn99if9Jaj/URgWrsTvfJ9qoRBGj7wK/e/YIANqcLo/lULEzduL8f07JZvbeA0Ft+8paRSSUYzXYCNfUXUazB7nSxbn8RH97c27d/Ut/6Q3Uq8ivMAAxoE47Z5iT9eAXJETpW7S3g/27q1eT51u5vRJAf4HEl1aXQYOHxH3eyJauUSosDl1sgRKfy7t/yyrgm2zkTPl+ZiZ9SxnUDmxZZtVdClkmlBGtV5JebvZakuhRUWHz6Cr7nnVdWzboDRXyz9lQyVIfTTUqkx81UYLAQoj1Vvu53cCY09L2dqDCTW6ZuclwVVPi2Db5jtLn4RwuWRSs3MPmyQRe7GyIiIv9SYkM09E4J9XFv1EYll2Ktc+M9n9VSYkP8kUklHPlwsje2oDZGc/2A1mqbkyU7cjny4WSCa910rnp3Hb9sy2HG0MaTb6YdLKZzXFA9UeOq4/oorbQRrfcIDalUwtS+8czbdIzO8XoGto2od6M+V56dtxuj2cHqZ0cRpfdjw4Fi7vpyy+kLngVL9+Sz7XApsXf9DOB187S6cwGvXtOdG09aNmpfA5fbTXmVjWi9ptF6o/R+HMw3+myr7UJqFeLP/WPb8czULnWLesoH+VFWZfNpszYqhQyr49S2hkZZQ99blN7vtOMqSu/bdkN1NQf/aJeQsUpcyVlEROTiMbpLDDml1SzYfAyny02RwcKN/01jw4EiADrF6TmQZ/TmytifZ6DQcPaxBzVEBPkxqks0z83f451h9Obivbz1275GyyzZkUvXhGAfsQJwRc9YftyQ3WR7S3fnNziduazSxuLtuVjtLnJLq3lnyT6u6pfg3T+9fwJ/7srj5y3Hmd4/wafsea3W7BLQqOQE+ispMVn5fVfu6QudATM+SvPmeln8n+EUfH4VeZ9OI+/TaWw+abXJ+3SaV6wA/LYjl/xyMzaHi1cX/U1cqD8dYgMbbWNirzj+2nOCdfsLcbnd7DxaxvYjp2JYbhySzA8bjrLnWDlut8C+XANjXllJaaVnvEztm8DbS/ZRbLRQaXHw3Tpfl17XeD1r9hVgsTsRBIE1ewvr9WHp7nyyi6twutx8tTqLskobwzpEnnZcjeoSTeYJEws2H8PldnMw38jqffXrP1/+0RYWERERkZbipYXpvPXbXp9tfzw+git6xvL0T7uxOVzcMjyVXx8dxqPf7+Dh73YQHaxhSp94Brb1rOLbNiaQ56Z1YcLrq4gL1TKiUxRRJy0RNVzVL55Bz/7FTUOTeenqbqft12e39+OJObvo9MhvaJQyBrWL4I3rGs9V8uOGo1zZK67e9rFdY5j19VYOF5r4fWceby72nGvnh3/jxau7ccvwFJbuzmdJA7lnwgPVVFod9H3qD0wWJ9MHJHD36Dbe/R3j9KiVMrYfLuXHWb5W8Ml94rnugw1c1imKr+8eULfqJnl6Smfu/GILqbMWER+m5a0berAru5xbPtnoE7RcG6vdReTtnsy7bkFAECBoxlwAvr1nAFf2iiOzwER44NllVO+bGsotn2xif56BjnFB/HjfoCZnXbWNCeST2/py/+ztVFodTOoV5xPD0iMphHdu7MVdX2whv9xMu1aBzBrTzjvF/D9XduCJOTZ6PfEnEYFqXry6q0/91w9OYtuRMvo++SdJETrGNSA0B7YN5+HvtrPjSBlJEVoWPDQErdoT69TUuArVqfnpgcH854edPPL9ToZ3jGRQ2wiKTgYENxcXdbXmlmb2r38xc+KYi90NERERkX8c+3INzPx4Y70kazXTmvf9X9MpK56cuwuT2cF/b+nTkt0UOUPGvrqS6wYlcf2gpIvdlUb5R7uERERERERahj8bcQedKRsOFDF9QNOBqyIitRFdQiIiIiIiZ82jV3Q4p3K5pdUMeX4ZwztGMqBNWDP3SuSfjOgSEhEREREREbnkEV1CIiIiIs3I0qVLycrKarH6N2zYwI4dO1qs/nPBaDSycOHCFql79+7drF27ttH9W7duZe/evY3uvxgsWLCAY8eOnXc9GzZsYPfu3edcfvbs2VRWVp53P86F3bt3s2HDhmatU3QJiYiIiJwlS5cupaSkBJnMNyfFmDHNb9FdsGABY8aMQadreh2a5uCvv/6ioKCAyZMnExh4agru33//TXp6Og6HA4VCQc+ePWnbtq13f05ODrGxsc3Sh6VLl9K1a1eios4sU2qfPs0TtOtyudi+fTtFRUUMHz68yettNBrZtm0bJSUlyOVyEhMT6datG3K5eEttScSrKyIiInIOdOvWjU6dOrVoG06nE7v9wqxeXFVVhdFopE2bNhw+fJgePU5Nhe7UqRMJCQn8/PPPTJ06FbXad4pvbm4u3bt3P+8+uFwubDbb6Q9sZgRBIC0tDYlEwvjx45sUHg6Hg+XLl9O+fXtGjBiBw+Fgw4YN7Ny5s9nEk0jDiIJFREREpAUxGAxs3bqVsrIy1Go1HTt2JDU1FYlEQlZWFocPH2bs2LHe4xcsWMDAgQNRqVQsXboUu93O4sWLiY2NZciQIYBHyKxatYqioiJ0Oh0DBgwgODgYAJvNxvbt28nLy0MqlZKQkECPHj2QyWRUVlby888/M2TIEDIyMoiJiaFXL09a/MOHD5OamkpiYiIrVqygW7duSM9gtWar1YrRaCQ83JNbpqCggLS0NLp06UJGRgZOp5Pk5GR69OiB2+1m3rx5TJgwwWvBcblc/PTTT4wdO9Z7vitXriQkJIRx4zxJ2dxuNxs3biQnJweVSkW/fv28FpgNGzag1Wrp1q0bgiCwa9cujh07htVqJTw8nH79+qHVaps8h0OHDmGz2Rg5cmS9ZQvqUlxcjMvlon379kgkElQqFV27dmXZsmX07t3bW76yspI//vgDo9FIcHAw/fv3JyCg6SUZjh07xq5du7Db7cTHx9fLepyTk8Pu3bupqqoiMDCQ3r17e6+70+lk+/btZGdn4+fn5yM4616nmv79/PPPzJw5E/BYtmJjY8nPz6esrIyAgAD69u1LaKgnF0xT4wqgsLCQrVu3Ul1dTVRU1Gmv+bkgxrCIiIiItBAOh4Nly5aRmJjI9OnTueyyy8jIyCA/P/+0ZYODg7nuuusAuPLKK71iBaC0tJRevXoxffp0goOD2bRpk3ffhg0bkEgkTJkyhYkTJ1JZWcmuXbt86nY6nYwbN84rVgRB4MiRI7Ru3Rq9Xo+/vz+FhWeWqTQvL4/o6GgfcWOxWFAqlUycOJEJEyZQUFDAvn37kMvlxMfHk52d7VM+KCjIe75arZbLLrvMK1YAysvLadeuHVdffTXJycmsX7++wSUMysvLsVgsjB8/nilTpuDn58fWrVub7L/L5SI9PZ3o6Gj++OMP5s6dy7p167BYGk565nLVX8NIEATsdrvPvvLycoYMGcLVV19NeHg4q1evbnLZBYPBwIYNG+jXrx9XXXUVKSkpFBefWvG4pKSEtLQ0+vXrx7XXXku3bt1YuXKlt5/p6elUVFQwadIkJkyY0Gj/m6KoqIh+/foxffp0UlJSWLFiBQ6HZw2qpsaV1Wpl1apVdOrUienTp9O1a1eKiorOuv3TIQoWERGRSw5BEHC4nZhdVoyOakrtRgpsZeRaizlqLmBdeTqLizeRZy25aH3cvXs3P/74o/e1YMGCesfk5uai0Who3bo1UqmUgIAA2rZtS2Zm5nm1HR0dTUBAAFKplNTUVMrKygAwm83k5ubSu3dvFAoFSqWSLl26kJWV5XOzTE1NRak8tf5PcXExAQEB3qfiNm3anHHgcG5ubr34FbVaTUJCAnK5HK1WS6dOnTh61JMqPjk5maNHj3r7c/ToUZKTm15ZODw8nODgYO/5ms3mBm/IISEhDBw4ELVajVqtJikpifLycu/+X3/91ft9rVixAvAIBZvNhkQiYcyYMVx55ZXYbDY2btzYaF9cLhd79+7F6XRisVjYv38/Go3Gx5WUmJiIVqtFJpPRtWtXKisrqajwrBxde9xs3rwZ8FhXYmNjiYqKQiqVEhYW5rWeAGRlZZGamkp4eDgSiYSYmBjCw8O94u/o0aN07twZPz8/FAoFrVu3bvKaNkRsbKx3XLVt2xa1Ws2JEydOO67y8vIIDAwkKSkJqVSKXq8nOjr6rNs/HaJLSERE5ILhElw43C4cgtPzOvneWWu73e3A6nTidAs4XQJOlxuX243j5Pvlhm18WroIAQEpEj5v/wC3tBp7+sabmTOJYamurqasrIwff/zRu83tdhMUFHRebdd2W/j5+eE+udBddbVn/bT58+d79wuCgNvt9omFqev2yMrKIj8/n++++85bRiKRYLfbfYRNXVwuFwUFBQwY4Jv2vm79arXa27fIyEgcDgdlZWUEBgZy4sQJ+vXrd8bnq9FovG3XxWw2s23bNoqKinA4HAiC4BNvM3HixHplTCYTSqXS6+KRy+V069aNP/74A5fLVS+wWq1Wc9lll7F9+3b27t1LWFgYWq2WmBjfJHq1+yyVSr3XoLblrG7fVarGV2uuqqqisLDQR+y63W6vm8lisfic6+lcWw3R0PdmNptPO65O1/fmQhQsIiIi54UgCCfFhxOH4PIKEadQX5g4nG6cbvdJMeL2ESW1P9fGKbiodluodlsocJTwSelC7wp5bgRu3/8+o0N70kp96SUh8/f3Jzw83Me9URuZTFbvxns+qbH8/f2RSCRMnz693o0WaDCA1+FwcPz4ca655hqfm87KlSs5duxYk0/qhYWFBAcH1xM17jorBVutVvz9/QHPjSwpKYmjR48SHBxMZGRkvSDec2XHjh3Y7XYmTJiARqOhsLCQtLS0Jsv4+fnV2yaRSJDJZI3edCMiIpgwYQLgERpLliyp9x3XvgZutxur1eoVWw2h0WgwGAw+22qPBa1WS8eOHRsNbvbz88NqPbVwZt3vQCqVNijyGmsP8Pb5dONKo9HUC5ZuiRRvomARERFpEGcDlhCH4Dy5/ZQwceHyiA63+6T4EHwsIk63G4vDgdFVTbXLQrXb6hUg1W4LZndD28wn31ux4fDtmAT0Lj8inDqK5JVUyCxsLjzKtIRLT7DExsayfft2jh49SkJCAlarla1bt9K2bVuioqIIDg6moqKCyspKdDodFRUV5xR7UINGo6FVq1bs3LmT7t27IwgC+/btQyKR0KVLlwbL5OTkEBoaWu8JOT4+nszMzCYFS0PuIPAEaB47doxWrVphtVrJyMggKenUGjXJycmsXr0ao9FIamrqOZ5tfQRBQC6Xo1QqsVqtHD9+/LRlwsLCUCgU7N+/n3bt2uFwOEhPTycxMdEbl7N27VpCQ0Pp2LGjT1mj0cjq1avp2bNnvWnQx48fJzQ0FLVaTXp6OlqtFr1e32g/EhISWLJkCSdOnCAyMpKysjKKi4u99aamprJ69Wri4uIICQmhoqKCLVu2MHz4cK/7KyMjg5CQEORyudcFV0NISAiZmZk4nU5kMhknTpyo14fc3FyioqLw9/cnMzMTq9VKdHQ0CoWiyXHVqlUrtmzZ4h3nJpOJEydO+EyNbw5EwSIi8i/CJbh83C+1rSK1tzsFFwICbjc4XG4cThdVThsGRxVGuwWj04zRWU2l04zJaabKKzR8hUe124rZbcGBs8H+yAQZKpcKhUuJ3KFEalcisSsRrMFgU6KyKJFYVGBTebZLXfTqe4SbjL2QIsWNm28DdyCtbHr2RUuwe/duMjIyfLaNGTOG+Ph4tm/fjsvlom3btowePZotW7awZcsWNBoNiYmJREZGAhAUFESPHj3466+/vG6Fuk/hSUlJLFmyhNTUVG+QbFMMGjSIbdu28fPPPyOTyYiKimpyum1WVhYJCQn1tsfGxrJx40aMRiM5OTmkp6cD8PPPP9OzZ0/atGlDbm5ug7ln/Pz8cDgc/PrrrzgcDpKTk2nfvr13f3BwMDKZjJKSEoYNG+ZTNjExkTVr1hATE+MTaHwmdOvWjbS0NObNm4dWq6Vv376UlZWxbt26RuuSyWSMHDmSrVu3kp6ejkwmIz4+3meWjdForGcFysnJYfPmzfTq1ctHjNUQHh7OunXrqKioIDg4mOHDhzc56yooKIiBAweyefNmHA4HCQkJPjEsYWFh9O3bl7S0NMxmM0FBQXTo0MHbry5durBt2zZ++eUX/Pz86Nmzp0/9qamplJSU8OuvvxIQENCg0IyMjGTz5s2Ulpai0+m47LLLUCg8qzU3Na7UajUjRoxg69atbNmyhejoaCIjI89LfDeEmJpfROQSI89aQpY5n1RNzBm5OWpcMs4G4kO8bhm3E5vbQZXLgslpxugwU+GoxuioxuiwYHJWY3SYMbnMVDrNVLksnlctC4iLhs3JCkGBSlCjcKuQO1XIHEokNiXYFAhWJS6LAodZic2sBJsSiV0FJ4WJxCVHJhHQKsFfCTol+CsF/BWgVYJWeWqfv0IAl5totdvHVC8IboaMvpLkmNBm+w5ETk95eTnr1q1j0qRJPttrpjVPmzatyfLbtm3D4XDUi38RuTgsXbqUlJSUZrV4NTeihUVE5BLiq7yl3L7/fdy4kSLho3azuC5quFeAWN12yu0myh1VlDlNlDsqMTirTlo6asRINaaTlg9TLfFR7bbgpuHnExVKVIIapVuNwqVC5lQhtQegtIegsKrQWhU4zQocVXJsVQqoLTrcHp+2WwJKpYBWVSM2PIJDp8SzTXdqm79SwF8m4C93opZKENwCcpsVZZUJVXUV6mojyqpKVNUmlNWVqMxGVGYTJfpItg73DViUSKRoDCdAFCwXlMbcQWdKYWEhvXv3bsYeifzTEQWLiMhFxul2UeUys6FiL7ftfw/hpKhwI3DXgQ94M3s+Vrcdk8tMtcvaYB0SJKhRoUbtsXa4VMidaqQOPf72CDQ2Je4aa0e1AnuVAlulHMGmBIcSiSDFDVgBhxT8VVI0agm6k1YPnRIC5AI6HQSEQoBKgk4pJUApIUApQaeQ4CeTIBUkgATBLUFmtyGvMqGoMqKoMiKvMCKvNqCoNiKvNqI0G1CYTSjNRhTVRmQuT6yKANikaiwyf6r8Q6nWhFGij+NEjw4YY6O8M1hqEAQBSX4+dOjcsl+UiA+NxcScjqqqKpYsWUJ0dDQRERHN3CuRfzKiS0hE5AJic9sps1ey23SYPVVH+Lsym4PmXA6bTzQqRuJsyWitIWBTIlhVuCxynBalR3RUybGYZAh2JRJ8ZzTIpRK0fnJ0NS+1DJ1KSqBKSqBSQqACAhQS9AqBQLmAXg4BStDIpAiA4BYQBHCf/Cs4HMiqjcirDCiqDcirTgoQs9HzueZvtRFptRGnC6wyDRaZBqtMg02moVodhNkvGLMqCItCh1Xhj02qwSZVYRMU2NwyHC6p1w4kU4E2QY4mSorgBnOeC6ddILC1HIlEgiAIGA84mXxlb+IHdm3R705EROTiIlpYRERaAEEQsLhsHLUUsrMyk4zKbPZWHSOzOo8cazHukzlEWqnCiBQi6V7dn7ICOfsTluOjO9wSKv7oDtJgAvwUnpfG81cXrTi1TS0nQCkhUCmgV4BeKeAvcSF12MHhQOq0U6MCJIJHgOAWkJpNyCorkJdVIDkpQORVBuTVBqg24aw24zDbsdsc2J2nBIjhpAixKHVYlOFYFclYVP7Y1GqsoUrsQuM/LQqlFKVSikIpQ6HwvNcqpQQrZSe3S5GooFppp1JiQ4aECLWWKH8tzmCBv349hrXEjdxPgtMiINhB3y6lJb/Os2PmULhyBkyc0TL1PzUDQiPhwddbpv5zIfsQ3HcFLDnU/HV//DxkH4S3fmp4/xsPQGQc3PRQ87ddw+gEePhtGDX1/Op5agbEJMDdz59b+U4S+CvbU8eF5uPnIf8YvPLNhW/7JKJgERE5T9yCG4OjioyqbHabDpNeeZQD1TlkmvMxOKsA8JepSfGLpotfawa4B2IvCKI4W0PWcSv7LQ6kEogK8kNV2J/4uC10KFSwL9LB8Zy+fH/LKPol6ZE47EjtdnDYwW47+dcBDguC0wVuAYmlGkrKwVSBpMqA22jAWVmFvaoah9mKzWLHbnNiswvYXBJsMj+sMj+sNQJEHohVHoNV5odDqgQdnlctJBJQKSQoVTKUKjkqpQylUkaQUoJcIUWukKCoJT583iukTSaUsjgd5FdXUmI1o5BKSQgMITowCIVCgUQmA6mEPpdJ2boqG7vN4xoad8tgAkKaf92SJpk5FDK2gKpO/pCv1jR/W6MT4Ou1F+YmdetlsG21R3jE1wq+nP0WfPYSVFeCvw4efAOuvuvU/rW/wZDLm6cPM4d6bui9hp7Z8Y+9d/5tut3w7mPww/uQ0Bp+2Xv+dYo0O6JgERE5C5xuF3m2YnYYs9hTeYS9Vcc4UJ3DUUshTsEziyZGFUJrTSuuCh9KjBCDsyCYvMMyDuRUsf6EEYdLwE8J7VspmdY3nJ5xAfRupcVpczD7uW3oM69AkEjpudeNqfoYCbEbMW834TBW4qiqwl5twW62Ybc6sNld2B14xIdbflJ81AgQHU5piO8JqD0vGW5UMjdKOahUMpQqBUqNEn8/lefzSRGiUkpRqmQeUXLyr1wuQSKR4BI8OVYcbjcuBJBKkUqlJ4WFFKlMhkQm9bykMiRyGRKJ1PNXJkMilSKVy0Amo9pq4Uj2MU6UFaFWq+nQoQNxcXENrprbq10bekzoh7G4En1E4IUXKzXMehlmPNKybVjMUGVq2TZqKMjxWDKm3QGLv4X7Xj61b+ajMGoajEmEpUdBXyfAee0SuPel8++Dww7G8tMf15xYzPDkDVBp9AixX766sO2LnDGiYBERaQSzy8LequPsMh0mo/Ioe6uOk2nOpchuAEAlUZCsiaKTNpGrI4bRRhWPnzGMQ0cdpO8ykJZTQU6ZGSggzF9Bxwg1I3uF0jNCSUc/B/6GAhyl+6laa8JotFBmcBCkSUE4aYEQJFJ0/on8vrrmhuVr7lAIDlQyByqFC5VcQKmQ4K+So/dToNT4ofD3Q6nVoFQrvOKjRozI5VKPqaRGTMikSGVyj5iQSpEqZEhkcqQyj6CQKuQgleKWgFMQcAhuJIBCJsdf7rF8nEsqbqPRSNbBAxQUFKBWq+nUqROxsbENZtOUSCT4+/t7804EhV743CvnxJH98Pr9cGAXBId7bv4TZ3qu/6/fwOJvYPbaU8ePToCXv4HAEJgxyHMjndLZY8F442SKf0s1PDAZdq6HmER44StoczLo2FgBbz8MaUtBroDRV8F9r4BS5THpj0mEN+fCF6/CgNHw8Fuecr995+nXmKvh7nFwzwvQwPdQD0MZHDsI3U5OT96+Fp6eAXc8A1+9DuYquPwGmPUKuF0wLBLmbj9lwbHbYFgUfL0GZg7xCLR7JkC7bvDtBs8xLic8fxus+Q2CQuDpj09ZYGq7Wdxu+O+zsGIBlBVD1/7wzCcQFdd4//ftgJAIePMn+OPHxo87cQxuGOARdW27wbOfQtxpXJErFsIHT0GlAS6bDIJv9lnW/AYfPesRiwlt4D/vQpe+nn1WC7z9CCyf7xk397/mW7aue6nmu/37pO935lAYegVsWuY5x/hUePK/0OFkfpamxgnAjvXwxv1QlAe9hnnG2UVGFCwi/3oEQaDYXsEu02F2VWaRUXmMA9XHyTKfwOr2pDIPVQTQxj+WK0L70UUTR0dFK6IcIRw8WsXWveVsz61kdlEBZssxohwGeiiN3K4w0l5hJAo7VLmoLhUw7VFidGtYpdBjUIZil/kBntTgSpcFZHVu+hIJPQIKCUuOQRkQgCIwEKVWjVKtQK5UeKwY8pNiQ+axYni2yb2WDOnJ96e2y5Eq5B7Lh6RpoeF2u3E4HNjtdux2uzfd9hncxprEYDCQmZlJUVERGo2Gzp07Exsb22hirZrspQ6HA7PZjN1uR6fTedO9X7JUV8LtI+GeF+GTpZ6b3p1jIDQKBp1m/aPWnWCTwRO3sDDD1yW0bwe8/gNExcMr98BLd8IPJ1dsfvomzw34jyxwueCpmzw38YfeOFXeaobv0kB7UvQJAvz+PXy2HKLjITIWtq2Bfped/hw3/An9RkFta1hZEeiCPP2uKIH7J3luujMegZFT4c+5cNeznmPT/oKkdtCmi+d8awRbbZfQwT3w7kJ45lOPCHr8OliZ5xF9tcnMgNIC+PZkOv73HofX7oMPfm28/z0He16n48BuTxxNcDh89rJHMP68BxpLBnf04MlrvwR6DPZ8Z09cD61OJpnbux2evRn++zt06g1bVsG9E+CXfRAaAZ+/All/w6K94KfxXLOzZed6j7iLiodFX8Hd42HpEdBomx4nhjJPTNIzn3osa9kH4blbPaLqIiIKFpF/FU63k8zqfHaYMtlTdYS9lR6XTp6tFAA5UhJV4bRTRXJFcHu6yKPoIo1Eb1ZSeryAA1l5FOTuYEvJX0hNZYTbKxjkrGKiTIJMosQm1VChCsOgDKVCGcZ2ZRKCxHN7l6jcBPjZCFA6CdVISQxUogvV4h8dTkCkHgzl/PDTEQTJqR9AieCi5y2TCGzbxiMw5DKP20Xa/AuL1eByubwCpWZp+eaivLyczMxMSkpK8Pf3p2vXrsTExCCVSj0J8BwOnE6n9+VwOJDJZA0KmeZaf+ac+fBp+LyW2yQgGP7yTYfO+j8gPAYm3+L5HJcCV98Ni748vWBpin4jIfbkCscTZ3osEwClhbDud9hi8tyUAO581iOaagfp1lh4ati9EWJTPGIFTrqFvjkzwbJ2CYyc4rtNH3ZqW1Qc3PI4fPmqR7BMuB5evBPufMbThz/neLY1Rdf+HkEDnmv54dOecw2L8j2ubVd4sZZLZ+w18MLtpz5P7gyFOZ733QbCR7+f/vxqGHO1R8iBx6rxw3seQdGmC/QPOnXcuGs9ImHZfBgyAXqfzOTbuQ90qbXI46KvYNLNpywq/Ud6znPZfLhulue6PP2xR7wATLkVXrzjzPsLMOyKU+PkqjtgzgceYdS5T9PjZP0fkNgWxk737Evp4BlzBTln134zIwoWkX8slQ4zuwyH2GU8REblUfZV53DQeoJKt2f6cIBERTtZGFfYwuhlakUno4wkoxuFwYC7NBNL6Uac5SUoqyqQudxolCEkqULRK8OoUIVRFtCeUmUIedJTi6cppW4C1AI6nZLYYC3tQwPRBft7Xnp/FBo/5GoVMo3a81chR66QIZN7RM24rC/4c6cdQSJDIrgY10NJaI9zy3dxNtS2opxugbRzobS0lKysLEpLS72LuIWEhOByuSgrK/MKlNrIZDJUKlWDYqWiooLi4mJat25NWNhFWkPoTGJYCnI8rqDaNzSXy2NROB9qX5PQSE/sR017ggCXtTq1XxA8bpdK46ltdS0Ti7+Bzcuhh/pUGZnMU0bXxHowDrsnSPf5LxrvH0BwmMe1AB5rg6XaY7GIT/W4LJ7+uOnzldQ5X/CcU11KCuDNBz0CrNrkuda1420WZdQvc6bUPieZzOOaKso7ZRmq15cTEFQn1qf2eRTkeK7dz5+f2uZ0QHzrU+Vr972JtP6NIqlTRn/yeyiIanqcFDfQ93Npv5kRBYvI/yYul+ef22EHp4PjVSfYWXmY3dXZ/G3JZb+9kHxrCREmKzFGG11MMm42SWljgjijgzCjGY3RiNRUhtRuxSFRUKEKo0QZSoFfDCdU0ZSp2lEdrscZra31jy+gUcvQBakJDNLQKsiPgGB/dHotulAdKq0fcj+PGJErZMgUMuRyGXLlKVHSFF0fuo2kg1lUHMhE3641AW1bJk22IAg+VpS6K7ueLy6XC4fDQXFxMdnZ2ZhMJvz8/EhISCAwMBCJRILRaGy0vFKprLcCcA2bN29m7ty53gRy999/f4Pr2VwSRMV5nqprYjHqolSBvU7+nfNJjRUV57mZri06FYtQm0pD/W3mali5yFMmMPjU9llXwPIFnif7xti+Dtp0rS9q6ore8hKPpQk8N75x13jiRdp2hZ5DPTf/5uDdxzzn+OMWCI/2xNM8M7N56q59Ti6Xx21Sc04NERYNR/f7bnPXqiMqDmY8CrMaCVYOi4aK0obbB8/3a6s1dhoaN+46ZSpOfg+nGyfh0WAo9d3WAg8yZ4soWEQuHIV5kJMFcakQ2arhY5xOjxCp+3J4/jrsFjKqszlYvJ+80iMYy3JxVBQRYKwixmhjsMnOjUYnkSYrOvOpf2YBqPILpUyfTJk2gSxNBIXaYMqidVhQI5EovMe6AJdKjr9ORVS4lvBwHfqwAAJCA9AG65CrlcjkUq9lRK6QI1NIPX/lTU/bPRMC2qa2iFCpERE2m61ZXD1ut9vHdVPz1+FwYDKZKCwsxGw24+fnR2JiIgEBAae9NlKpFLVajVQqxWazUVFRQXl5ufdvYWGhz4KDgiDwwQcf0KNHj4tnaWmKwePhnUc98QejpnluAq/dB9Pv9sRotO0KWXs9AZMxCZ73JQXn3l5oJAwaB+894bEAuV2eqbpSKdz2ZMNlVv0C7Xv4ihWAEZM9rqumBMu6JTC0genMhlKPCBo0DsqL4avXYHytJRXGXw8PTvbERkysIyjO5//H6QQ/f0/8TFkxrP713Ouqy6pFnusUHOaJYYlOgNROjR8/ehpM7wVbV0PPIR5LW8aWUy6aybd4YntGTPKMg8P74NV7PbE6+lCPqPviVU+Ar58GltbJQ9Ouu8c6c9dznun1W1bW78PaJZ6A2ag4z3dZUepx82m0TY+TQePgtVmecTv6Kjh2CLasgIS2zXU1zwlRsIhcGBZ95fElu92ef4qH3vT8gNcSIzjsUGX0PAWUl1JVlkt+8RGqyvJwG4rxMxjQG6vpWmmjR62HCZdUglkXiCMwHEtgMoaEGNJVYRikQRgEDUa7ApMV3K5ThaqQYECKWS5DpVURHKYjJSGEzqlhhIfqUCnlSGVSH+tIbYtJS8aQNCc1AsJut9dzuZwJteNK6saX1LXKCIKA0WikqKgIi8WCv78/SUlJ6HS6ekJFEASqq6sxGo1UVFRgMBiorKzEYDB4xUl1dbX3eIlEQlBQUINxK263mxMnTlx4wVI3hgU8eVgumwz/96jHcnLVnfDZMo9IefVez9Pt2OmeGxh4XEP3vwq3DPcIlv6jPU+3tRl3LVzdAybf6hs42xgvfwtvPQRjk0Ct8cRQNJWrZPE39WNQwCNEXrgNjmd5bvyfnbQEjE3yTP+96k6PYPliVf2yIRGe2T6TO3n+Xn4DXH//qf1tOoPKz3MDf3ehb9nRV8ODUzzX4s05pz/f2sx6yTNDaXiUZ1bL4x94gl0fu+7UDKu62KzQ52Qch9vtsVR0PXlrfGveqWvTpb8n2LcmbuW9RU3PokpqBy/N9gREV5k8v3e1Y1g69oKnPvJYgIryILkD3PTwKTfQ7U97vsdJHU4mCqzz3U+a6bl+kzt5YqOGXlG/Dz2HeIKy927zCKWPfj8Vs9LUONGHwgeLPbPbXpvliV/pOdQTN3QREVPzi7Q8hXkwOt7zY1CDRALDJuI2V0F5CY6KQmSGMuR1nvxL/BUUBKioCNRhD9Dj1rZCqklCoYjHLgnG6FRSaZNiqnJgrnacsopKJLhUcowSGQUOgUIXGJCiDvCjVVQgHWOD6BEXRGqkjgB/NWo/uUeUKOXI5R5ryf+KKKlNjcCocfeciatHEIR6ga41f8+0vMFgoKioCKvVilarJSwsDJfL5RUktYVJzau2lUehUBAcHExwcDB6vb7e+8DAQGQyGRUVFTz33HPU/tmSSqV8++23l6aF5Z9M5t/w2DX1k6zVTGtedqzp8m8/4nlAqRv/ItI8zBzashmXLwKihUWk5cnJArcbkyKIcmU4wfZiAhwGSneu4GioP0cDZeREq8gPTMKk06FWxRMoSyTEHU2wVY+yWkV1pYNKkw17mQvKPNUq1TL8g1QodH44dRJK7AKHq53sNdopcwrIXVJah2lpH6VjcnwQfVPDiAn1R6dRoVR5gl2lsosfSHa+uN1un3iUxp5B6lpIar/OFrvdTnl5OWVlZdhsNqRSKdXV1Rw/fpycnBxMJpNPP/z9/QkMDESv19O6dWuCgoIICgoiLCyM8PBwtFrtGbnS9Ho911xzjTeGRSqVct9994li5WKwbsn5ZbfdtsaTd0RE5AwRBYtIi1MQoqMguD9/tboOQSJFIrjpX/Qn706IIIw4oqqCCKzWkGCQYc1x4HZ7bnQOCVQFStEFKwmJ0xMXpEGpVWOUSDlgtLOjxMz+wkqOHzEjAAFqOR1iApjSKYheKaH0SQ0lJFCDVqM477iSSw2n0+kVKbUFR02cSkMunDOlxl1TYwmpaxkxGo2Eh4fTrl07dDodhYWF5OTkeN023bp1Q6/Xe0VJUFBQvQBaqVSKSqVqMEHc6ejXrx9t27bF4XBc3FlC/3Yai4k5HQU5ntiOfiOhx6Dm7ZPIPxrRJSTSsrhcrFj7E9u/NvrkF6lBrpSh0/ujC9aenG3jmQKs1fuj8FdjdrrYm2did66BvQWVHCiqpKza40qIDfajU6ye7knB9G8TQfvYQHR+CpRnMBvnf43arh6r1eq1ptR14ZwJLpcLk8lUT4jUuG4actfo9Xr0ej2tWrVCr9cjk8mQy+WEhYURGhp6VsJDoVCgVCpPKyJlMpm3ndrvaz6fq+ARERH530S0sIi0HIIAOYeJ/mEJgmRovd1dp3WgW6c2SCQS3G43ZruLskob2/NNZGTms6+gkoOFJqwONwqZhLbRgYzu1oqeySH0Sw0jNtQfrVqO7BLID9DcuFwubDYbZrMZs9mM1Wr1BtCe7hnDZrPVEyK1PzfkrqmxhLRu3bqedUStVlNeXk5xcTEOh4OgoCAiIiLw8/Nrohf1kUgkaDQalEqlV3TU/Vv7/f+qVez7lxfTeXAbugxumRkVSz5bjX+ghuHT+7ZI/edC2QkD8/9vKXe9fU2z171+4XbKCgxMundkg/uXf7+RwBAtfca1fL6iM+W/D/zAiGv70a538nnVs+Sz1QSG6hg8pdc5lX/l+k+5591rCQq78MtYrF+4HWNpJZffMbzZ6hQFi0jLkZeNbfGPKA/tQ0gagoRaNyAJxLWK5WCZhV05FezJNbIv38TR4krcAgRqFHSM1XPz8FR6J4fSOyWEYJ0ajfJ/90bWFDVThI1GI9nZ2eTl5XmDTWsjCAJVVVU+AqSuMLFYLN7jpVIpgYGBBAUFERwcTFJSkleI6PV6AgMDG813UpPULTs7G6fTiV6vJyIiosGZOnK5HOnJxQ9r3tcIEJlMhkajISAgoNHU+/9rfP/yYvKPFCNX+Fp4rn+ygZka58l/H/iB65+64oLcdH58bQnH9udz11vTCY4M8m7f/Pse0n7did3qQKlWMHx6X3pc1sG7P3PXMVK7JzRLH75/eTGDJ/ckvn0TOU5qMeqGAefdZv7hIlbN2UxJfgUKlZxOA1ozZFqvRsdrxoZDbF2agaHYhFQqISI+lGFX9yEmJeK8+yLSOKJgEWkZCvOwpu/EMP8HFibNwqi1EFDlhxQJbiA9SM9Xs3dRYvLkSmkVoqFznJ6rByTQJyWUjnFB6NRK1Mp/lsnf7XbjcrlwOp24XC6vJcVkMlFdXc3WrVv55ZdfvBaQdu3a4e/v7yNMasejKJVKrwCJi4ujc+fOPtaRcxEJLpeL0tJSSkpKcLlchIaGEhsbi1ar9REiNeKkKQEpk8nQarUoFIpGj/lfZejUXvQd37VF23DYHNgszbs8QmMYSyspO2Gg+/D2ZGzIZOi03t59/SZ0pV2fJD56cA73vHstGp2vdS1r93GGTD03K0BtXE4X1uoGMti2IFUGM3Pf+INRNw6g04DWmMqrmPP672j1/vQa1bHe8ZUV1Sz5bA3THx1HfPsYXA4XW5ams/iTVdz9zrUXtO//NkTBItL8lBVjPZaN4ZN3mBN/N/lBVn7UyFBX69DjpgIplUYXV/SMYlDbCPq1DiMuTItWJUch/99/AhcEwStG6r5qu2IsFgtVVVVYrR7RlpWVxaJFi3zqOnDgAJGRkYSGhtK2bVuvZaRGkPj5+Z2VxakpK4jL5SIvL4/jx4/jdDqJi4sjJSUFjUZzTtdBrVbj7+//j7SInQ0l+eWs+H4jhcdK0QT40Xd8V7oM9rhC09cfJGP9IW54+krv8f994Acuv30Yfjo13720GJvZzhdPLCC1ezwT7/as7eOwOfn5vWXkHCwgKEzH+NuGEhHnyRZrqbax8sdNHEnPRSaX0q5PMkOn9UaukGEoMfHRg3OYeM9lbPptF0mdYhlxrSc3yN9pmXQe3Ib2fVOY9/afDJ7S84zErrnSSllBBbGtPSnzj+/PZ8nnaxg4sQebluzGbnXSaWAqQ6f1RnALvHfvd9z84mSvBcfpcPH+vd9x/ZOX8/0rv2Gz2Jn3zlIi40O58dmJgCeH0h9friVr13H8tGrGzBjotcDUdpsIboF1C7dzYOsRzCYLMamRjJ05iMBQXUNdB0CpVjBp1kiSO3vWCQoM1dG6ewJ5mQUNChb/QD/0EQE4HS7cLjdOpwu3002r1Eif44wllXz7wi+UFRiIiA9l7MzBBEc2scwBcHDbUdYu2IbVbKNNz8R6yWszdx5j3cLtmMqqCIkKYuT1/b1WHYfdyao5mzmw9QiaAD+GXdXHp2xd91LNWHjqhzsBj2WrdfcEjv6dS0F2CfqIQEbfNJDopHCg6XEFkHPwBMu/30hleTXx7aIJCm9+i6AoWESaF2MFtuPHKP/2C37WXkmJzkX6ZeU4FnXDIYFKTv0A3j6iNUM6RDZR2aVNjbWkoVdTZaqrq6mqqvImX8vKyiItLY3Dhw83WObyyy8nKSmp0TprREhty0ddK0jNMQ2JB7vdztGjR8nOzsbtdnuFytnGqNTuj06n+0daVc4Wm8XOnNd/Z8iUXkx/dDyGkkp+evMPtEEaUrrENVk2PDaERz6/mVeu/5TbXpvm4xIqyC7hyruGExiq469vNrB09npmPDcJgCWfrsY/0I+737kGt9vNks/WsH7hdp+YF4fdyY3PTESl8bgDBUHg77RMrn18AoGhOgKC/Tm+/wSJHRvJSF2LI+k5JHWM9UkRUG20oNaouO21qzCbLPz83jK2B/xN3/FdadsriX2bDzNoUk9P+YwcQqODiIgP5ZHPb/YKttouoaLjpUy5fxRjbx7MpiW7+fWTVdz3wQ31xnNRbhlVBrNX6KyZt5Xl329k2oONT75QqhVesVJDblYhiR0aPnepVMr1T13BnNd/Z+H7yxEEgeQucUx9YLTPcYXHS5l070g0AX6k/bqThe8v49ZXpjWa36n0RAW/fbaaqx4eS1zbKAqOlrD4k1UEhXnE1okjxfz+xVqufmQs0UnhZO/LY97bS7n9javQBmrYuHgXxbnl3Pb6VSiUcvZtbvj3pClyDhUwZoZH4O1Ze5D5by/l7v+7FqVa0eS4Mldamf9/fzF25mDa9Umi7ISBP75cR0hU0wLtbBEFi0jzYa7CcewwRWvWsaS0NRZ/OV+P28Libm8xatEWaj8syKQSUqIufCDYudCYKDmb9XccDgeVlZXe7K12u53du3ezadMmiouLiYmJ4fLLL+f333/3scJIJBJiYmLquWNqC5NztWDYbDavUAGIj48nOTn5vFZB/jdZVdb+vJ20xbu8n/38Vdzz7nU+xxzek4NO70/XoZ7FDoMjA+lxWQf2rD1wWsHSFIkdW6GP8NwMugxpy/cvLwY87o2sPcd59ItbUKo9gnHgxB7MfeN3hl196om7xsJTQ15mIfqIQK8lotvw9mSsP3RGgiVr1zHa9vYV1JoAP++2wFAd/S/vxsbfdtF3fFc6DUhl6ewNDJzYA4lEwr5Nh+k4oHWTbbRqHUlEvCcDbNch7Vi3YDtVBjM6vb/PcZHxoUy4baj3c4d+Kfz51Trv5y+emI+xrAqA2NaRXP3IuHpt7Vq1nyqDmZ4NWFcALFVWvn/5NwZN6kH7PsnYbU5Wzd3M4o9XMeX+Ud7j2vdJISDEk1V28JSebF/2N8W5ZUTEh/L27V/79HHszMEc2HqE1G7xJJwUajEpET5Wmz3rDtJlSBuvRSWpUyytUiM4sPUIvUZ1Yt+mLMbMGIQ20GMR7Ta0HUu/Xt/kda1LavcE77jqPqI925f/Tfa+PGKSI5ocV4f3HCckKogO/VIACGsVTGLHVpjKKs+q/dMhChaR5sFmxXk0k4L9Ofy51YJD6cfrl6fxVPtriQ+IRC6T4HQJCHjEyvszexMTfG6uhpbgTN04Z1unxWKhsrISu92zmq7JZGLz5s1s27YNi8VC+/btmTRpEvHx8UgkEvz9/Zk/fz5utxupVMrNN99MSkpKc54qVquVI0eOcOzYMSQSCYmJiSQlJaFSNbAI2hkilUrRarWNBvD+EzmTGBZTWSWFx0p9blCCWyAkOui82q6tB/0D/XA5PeLZWFYJAnxw3/enDhDA6XRhM9trlfcVlOnrD5G9N4/XZ57MOisISKRSRpttqDWNjwuX08Wx/ScYd+uQRvsHoNGpqSz3iPW4ttHYbQ4Kj5USHBnI0b9zGTOj6XwstevTBmm8bdelsqKalT9uIjezELvFk6lZozslwG977aom29mxYi9b/kzn+icuxz+gYQvj/i1H0Ab60XlQGwDkSjmXXduP/7vzG4yllV7RV9uSIpVK8dOqMJVXey1J9ftuxk/n+7BQ+7xNZZUc23+C3WsOeLe5XW6Co4I85Q2+5c8lU3dj35sxqOlxVVVh9rnODdXVHIiCReT8cThwHjlEfo6BZUsOYpX68fvIdOLCwngwdSq3frIZjUrOsqcuo7zKTlKE7qKJlXNx45wtLpeLqqoqqqqqvFaYEydOkJaWRkZGBnK5nJ49e9K/f3+Cgz0Lzvn7+6PT6YiNjaVfv34UFRURERFBSEgzrWKLJ2bm8OHD5OTkIJVKSU5OJikp6bxFhkqlwt/f/x8zA6g5CQjR0SolwuuiqItcLsPp8B1755MZKzBEh0Qq4YGPbqo3gwnAaq4f0Gq3Oji0I5sHProRP+2pm878d5ZyYOtRug1r12h7xw+cICI+pJ6oqUn+WIO50or2pDVEIpXQoV8K+zZlEREfSny76Ho3u3NlzbytWKvtzHh+Ejq9P8cPnOD3z9ectpzb5Wblj5vIyyrixmeuJCBY2/ixbne9u3FDFsXaFli3242lyoYu2L/ecTXo9BpK8yvq1HHqOgaE6Og7rotPMLRP+SANlspTC77WtQDL6o61BsaZ0MD3ptP7n3ZcafUazLXartv35uKiCxa3ILBqyy527c/CarcTFxXO5BGDCNTV/2J37DvEwhUbUCh8u/3cXTf+I3Nx/E/gcuE6lklhXgUrFuymyq1C2+MYa6JL2NrlQ/YcrWDh1hxevrobHWL1F7Bb5+/GOVusVitVVVXeacVut5uDBw+SlpZGdnY2er2eMWPG0LNnT9RqNRKJBK1Wi06n80mAFhIS0qxCxWw2k5WVRW5uLnK5nNTUVBITE887xuTfaFU5W1K6xrFq7mb2bcqiXZ9kzJVWln+XRo/LOhDfPoaI+FBK8soxlJgICgugOLecKoP5nNuriY1ZM28rQ6f1wu0W2L7sbyQSCQOu7N5gmUM7solKDPURKwBteiWyZ+3BJgVL1u7jpHaLr7fdUmnl4PajpHSJo9pkYeNvu+nY/9QK5B0HtGbhe8soPWGgy5A2voXP49Hc7XKjUMtRa1RUmyxk7sg+bRmbxc6iD1eg8lNyw9NXoFDV/7/45b8riE4Kp8+4LqR0iWPt/G0c2HqENr0ScdpdrJm/lejkcK8LCODQ9myiEsK8MSyBoTrCY4Pr1V1Du97JfP3sQo7tyyeuXRSF2aXkHy5CfzJ4teuQtvz83jLa9EwkMj6Ukvxyln2bxpT7R6HR+dG+Xwobf9tNREIoCqWc/Vt8Y1giE8LYvWY/DrsTuUJG9r68en3I2n2c+PYxBIZo2b32AJZKK4kdW6FUK5ocVyld41j+3UbPOO+bTFmBgey9eYSepyWxLhddsKzfkcHBoznccdUE/NVq1u3M4MtFf/LQjVMbVK1d26Zw9ZihF76jIvURBNzHD1OcU8byxQcwWKWMCv2bvl0KuT16DD0CUhn+3nISw7XcPbrN6es76+ab341zttQNogVPbMjOnTvZuHEj5eXlxMfHc91119GuXTtv7ElAQECLWyWqqqo4fPgweXl5KBQK2rZtS0JCAnL5+f/bi1aV+jEs4MnD0rZXEqvnbsHpcNFjRAeueWw8y7/byLLv0tDp/WnfN4W4dp4VmUNj9Ay7qg8/vrqEwDAdSZ1i0el9rY8d+qfw9TML6Tq03Rkli7v8zuGs/GETHz00B4VSTnz7GEZe37/R4zM2HKJtr/pB3andEvjzq/WUFxo4tOMYab/uBOCjB+cwfHpfuo9oT9bu41z3RP31hPwD/bBZ7Hz+xHzsFgcdB7am95hO3v0RcSHIlXJOHCliaq24D4D2fZJZ+MFykjrFMvGey057vrUZMrUXSz5bw/v3fkdQeACjbhhAQXYJv3680jvDqi6F2SUczchF5afk/Vnf++y7973rUGtUlJ4weF1E+ohArn5kLGvnb2Pp7A3IFFIS2sUw9YHRPvesVqmRLP5kFcW55UTEhTD1gVFN/r+Exui5/PZhLJ29HpvFTrveyT4xLNHJ4YyeMZDfv1hLZXkVoTHB9BnX2TvFfODEHqz8cROfPzYfbZAfw672HStdhrQh/3ARXzw+H31EIKnd6wvNuLZR/PXNBgqOFhMUHsBVD4/1xqw0Na40Oj+mPTia5T9sYtl3G0ns2Ir4dtFUG89dfDfERU/N//pXc5k8YhCtEzzBXW5B4PUv53DN2OEktoryOXbHvkMcyS04Y8EipuZvWdy52ZQczmXpH0cpPmFiWvViZt2bynZZKVmDv2X59hJmfrKJ+Q8OZmy30wfvNdrOBXDjnC0Oh4Oqqiqqq6u94shgMLBp0ya2b9+O3W6nU6dODBgwgNhYzwwEuVyOTqdr8aDUyspKsrKyyM/PR6VSkZycTHx8fLMIFalUir+//3nFu4j8MyjOLePXj1Zy++tX+2yvmdZ873vXN1l+5ZzN2Mw2xt86tAV7KXKmtHSG5ubgolpYHE4n5cZKwkOCvNukEgnhwXqKyw31BAuA1Wbnh99XcjSvgCCdljEDe9E6/txvhiLnhlCYT+mRfJYvz6GwsJqr8r9m70PX8Iewip/aPYHSrebpeXsY0j7CK1aqq6sxmUxe60JdLoYb52yoCaKtqqrCZjsVC5CTk0NaWhr79u1DqVTSu3dv+vXrR1BQEOBJ7hYQEHDO04TPFJPJRGZmJgUFBajVajp27EhcXFyzrbejVCq9s5VERLJ2HSe1W8I5lz++P79Jy4+ISF0uqmAxWz0/+oo6T34KuYxqq7Xe8UE6HSAwok93po8JZM+hI3y7eDmPzJiGPqB+YiBTlZnZv/7l/dy1bQrd2jbvjIt/I0JZMWWHslm1Ooe83EqmHv2U4OumcKf/Ni4L6MxVkUN5c/FeCg0WfnlkKACZmZls2rTJa43o06cPCQkJF8WNc7Y0FETrcrnYt28faWlp5ObmEhISwoQJE+jevbvX+qBWq9HpdOc1TfhMMBgMZGZmUlRUhJ+fH507d6ZVq1bNJlRqYm1Eq4pIbRqLiTkdxtJKZj+7iMSOrYhtU/+hVESkMS6qYNGoPT+Ajlqpxj2fXfg38COfEhdNSly093PPDq1ZvyOdwzkn6NWxfoxEgFYjuoSam0ojhoOHWbM2j+yjJibmfkNC1yT+00WgzGLhs44PU2Ky8u4fB7hhcBLtWgVRXV3tI1YAtm7disPh8KZtVyqVl9yCdzabjaqqKszmU35Yq9XK9u3b2bRpEwaDgaSkJG688UbatGnjtTzUrJvT0onTKioqyMzMpLi4GH9/f7p27UpMTEyzWkBEq8rZ8+ijjzJy5EhGjRp1+oPPgbfffpvg4GBuvrn+1NiLRW5uLs8//zxfffUV8e1jmnQHBYbqeODjm8647u+//568vDyeeOKJBvd/+umnhIWFMWXKlLPu9z+BG2+8kYcffpguXc5+8cf09HTeeecdvvvuO59syxeKs/1fuaiCRSGXow/QUVxuIEjnia52CwLF5RWEBXerd7zL5UZAQF7nyVEh/2etN3PJYq7GsHc/a9fmkZllYJzpT9oqi9h74yN8aPqS5xKvJ0kTxe2fbUImlfD8tK4A9VYHrmHXLt+ARYlEgkKh8AqYht43tq25rAmCIFBdXU1lZaXPmj1lZWVs2rSJHTt24HK56NKlCwMGDCA6+pSArpnx0xyxIk1RVlZGZmYmpaWlaLVaunXrRnR0dLOKipqcMC1tHfpf5dFHH+XgwYP1Zki98cYbzd7WjTfeyJtvvklkZMtnhX788cdJT0/nyy+/JCbmVKbZBQsWMHfuXMxmMxqNhptvvpkJEyZ492/ZsoW+fZtn9ehHH32U66+//oxvwHfeeWeztAtQUlLCPffcg8lk4q+//vLZt379et577z3MZjPz5s2rtzBpWloac+fOpaioiFatWnHrrbfSsWPDyedEzo2LPkuoX5f2LNu4nYhgPRo/Fet3ZKBSKkmMiWT3gcPs2H+ImVeOQS6XsTRtG4bKKsYP7kOg1p/dBw9TabaQEndmq3qKnAd2G6Z9+0jbkM/+A+UMU+2na+4yhJdnc4v5D5LVkTyWPJ30Y+XM23SMF6/uSojOY0ELCGg4o22/fv2QyWQ4HA4cDgd2u937vuazxWLBZDJ5PzcWzyKVSuuJmDMVPRKJpMEgWkEQOHbsGGlpaRw4cACNRsPAgQPp27cvOp3O226NUGlJK4QgCJSWlpKZmUl5eTkBAQH06NGDqKioZrdKKRQKtFpts4nAfyo33XQTU6dObdE2rFarj4WvJSkuLiYvL49x48axYsUKZsyY4d03bdo0Bg0axIwZM5g9e3a9m/XWrVu58cYbz7sPNRmhLwaCIPDee+8xduxY5s2b57N9/vz5/Pnnnzz22GM899xz9coePHiQDz/8kOeee442bdqwY8cOnn32WT777DPCwsIu5Gn8o7nogmVwz87YHQ4+mf8bNruD2Mgwbpk8FolEgqm6mpJyAy63GzkyRvXvyZJ1m/ngR89qtlFhIdwyeRxaTcsGM/7rcTqp2rePTRvySM8opV8rM/3/+BBm/odPQsvZYTzBmq5voZDIeeT7HcSHarl71KlIc7lcjkKhwOHwrDorkUjo168f8fHxXpFypvErLpernqip/bn2tqqqKp/9jbVRO919zbRjo9FIXl4e5eXlqFQqJk+eTHJyMiqVynsjDwwMRKfTNbtgsFgsVFdXey0cxcXFZGVlUVFRQWBgID179iQyMrLZ2xWtKi3D8ePH+fTTTzl8+DBBQUFMnTqVUaNGIZFIWL58OStWrOCtt97yHl9j4g8ICOCRRx6hurqau+++mz59+vDYY48BHiHz4osvsnfvXiIiInjwwQe9601VVlbyxRdfsH37duRyOYMHD+amm25CqVRSWFjIjBkzePzxx/npp5/o0aMHt912GwArV65k5MiRDBkyhGeeeYYbbrjhjESryWQiNzeXDh06AKfcDNdeey3z58/HYrEwYsQIZsyYgdvt5pprruGDDz7wWnDsdjvXXnstb775Jo8++ihms5nnnnuO5ORk3nnnHcDzf//ee++xZcsWAgICuOeee7wWmLfffpuIiAhuuOEG3G433333HRs2bMBoNNKuXTtmzZpFeHj4ac9j6dKlCIJQT7AYjUYyMzN57733fILta/Pnn38yZswY2rdvD3hi9Pr06cPy5cu57rrrvNdk8uTJ/PLLLyiVSmbOnElAQABff/01OTk59OrViwcffLDJvEYul4sff/yRZcuWoVQqmTZtWr39CxYsYNmyZZjNZjp06MA999zjzetUUlLC+++/z8GDB2nVqhVjxviGTNR1L9Udn2PGjOHJJ59k3rx5FBUV0blzZ2bNmoVe78mx1dRYFwSBJUuWsGjRIpxOJ2PHjj3td1KXiy5YpBIJo/r3ZFT/nvX2DenZhSE9T5kFlQo5Uy4bxJTLmk7jLNKMuN1UH9jPlg3H2bmrmO5tNQz9/THoOYSiEWN4ovj/uC58KEODu/Db9hy2ZJXy0/2DfVZd3rhxo/eHQBAEn1lCarUaQRB8hEZT05VrRMXZ3lRrcrbY7Xbsdjsmk4nKykpve06nE5vNRklJCSaTCalUSnh4uHdKMnj+GWtT24XVmCWnsc+N3QhycnJIT0/3fvbz88NisaDX6+nduzfh4eEtEucjWlVaBrPZzJNPPskNN9zAyy+/TFFREU899RTBwcH06tWrybKJiYksXLiQMWPG8PHHH/u4hDIzM/nPf/5DREQE//3vf/nwww959913Ac8NXK/X8/XXX+N2u3n77bf5/vvvueWWW7zlbTYb77zzjvf/UBAEVq1axauvvkpERARhYWGkp6fTvfvpA2u3bdtG9+7dfcaOwWBAq9XyySefYDQaefHFF/n111+ZOnUqAwcOZO3atVx3nWfdpZ07dxIXF0dSUhILFy5sMCbjyJEjPPPMM8yaNYv58+fz5ptv8sMPP9T7X8jOzqa8vNwrdL7++ms++eSTBq0itSksLGTOnDm8++679X5/goKCeOaZZ7zHNUROTk49F1Z8fLzPb4bBYCAmJoavvvqKLVu28MYbbzBp0iReeOEFXC4X9913H8uXL/dxtdVl+fLlbNiwgbfeeovQ0FC2bdtGRcWp7Li//vormzdv5s0330Sv17NgwQJeeukl79h47bXXSE1N5ZlnnqGqqooffvihyevS2Lm++uqrKJVKPv74Y959911efPHF04713bt3M2fOHF555RXi4+PJyMjgt99+O6u2L7pgEbmEEQQsmYfYsSGbrduK6NRBz2XbX0eq8oM7nuE+4xJkUhnvtb8Hu9PFE3N3M7BtOOO6n3LRHT16lOPHj9O3b99GffASiQSlUul9snA6nT4CpjmQSCSe9NgWC2azGYlE4nVVFRcXs3XrVnbv3o0gCHTr1o0BAwYQERFx8jIIXpFU48JqzLJjtVq9awc1Jb6kUmmDrqm6P4gWi4Vu3boRExPTIkJFIpGg0WhafMr1P5Fvv/2WuXPnej/rdDq++eYbn2O2bdtGSEiI90k2OjqaCRMm8Ndff51WsDRF9+7dvfFTo0aN4j//+Q8A5eXlbNu2jUWLFnm/02uvvZYnn3zSJ0i35qm3hn379hEdHe0d8zVuoTMRLFu3bmXgwIE+2wIDA73bwsPDueqqq5g3bx5Tp05l+PDhfPjhh1x77bVIJBLWrFnD8OHDm2yjffv2XgvS6NGj+fbbbykvL6+XETo5OZmHHnrI+3no0KF88MEH3s933nknJSUlAHTo0IEXX3wRt9vNu+++y4033khYWFijoqQpKisr61lGVCoVRqPR55rUfOc17vCuXbt63cu9e/fmyJEjAOzdu9dHZN17770MGzaMtWvXMnHiRO93P3DgQD777DPvcUuXLuX222/3uqGuvvpqFixYwLFjx/D39+fAgQO8+uqrqFQqVCoVQ4YMYefOnWd1riNGjPC6BG+55RamT59OVVUVO3bsaHKsr127llGjRpGcnAx4xnDtB8IzQRQsIo1izT7Kzg2HSdtUQNu2ekaaVqA4nAHPfspyRTHzTXv5vN39hCoDeff3/eSVVTP/wcHeH0KLxcKWLVuIioqibdszT0Ykl8uRy+X4+fnhdrt9BMLZ5mQRBAGz2ey1ptTefvjwYdLS0sjMzESn0zF06FD69OnjkyNGrVYTGBh4zunn6/a/KXdWVVVVg3XUpPFvbuRyOVqttsWDhP+pnEkMS0lJCYcPH/aZweJ2u8/6h7outceDXq/3ju2SkhIEQeD660/N0qmxYNasFF63PMCKFSvYtWsXl19+KmutVCr1uiYbw+FwkJ6ezgMPPNBo/8Bzsy4tLQWgU6dOWK1WDh8+TExMDDt37uTee+894/OtWX+roYeZsrIyPv/8c/bt24fZbMbtdvvE0H366af1yixZsgSVSsXIkSOb7ENT6HQ67wKnNdhsNp9Yn9rnIJPJ0Gq1PtsCAwMpKioCoGPHjixcuLDB86sbE1i7jpKSEl599VUfa5fT6aSwsJCgoCBUKpWPdfpcfldql6nJNVVaWnrasV5WVkZcXFyjdZ0J4i+VSIPYcnPYs/YA69fnk5IcyMiYMvzenQ2Tb8HathN3Fb1LX20qt7YaR4nJypu/7eX6QUne9YIEQWDDhg0IgsCgQYPO+YYrlUq9TwOAz83eWWc6fG1qBEDNj1bt7Xv27GHjxo0UFRURFRXFtGnT6Ny5s8+Nu2YxwuZYb6d2/5vCYrGwcuVKn201cSXNTY1V5VKaRv5PJCwsjHbt2nldFHWpHdvVXO1JpVLmzp3boMhuSBRbrVY2btzITz/95H3aB3juuedYv359k7EGf//9N0lJSfXGaN0HC6PR6LWGSKVShg4dypo1a0hKSqJz586NBuafLV9//TVVVVW89957hISEkJGR0ei1r2Hr1q0cOHDAKz5rYt2mTJnC7bffzujRo0/bbmxsLLm5uT7bjh8/TqtWzZvUNCQkBJPJ5LOt9rUOCwvjnnvuoVu3+rNsi4uLsdlsWK1Wr2ip+z2dyXisXcZgMAAQGhp62rF+ur6fCWJyBZF62IuK+HvNPtaszSc+PoBR/YII+PZFaNMFJt/Cy6ZV5LiMfNnpUSQSCc8v8MRcPH9VV28dNanh61oszheFQoFGoyEoKIjg4GDv4nu1rTolJSUUFhb6JHqrrKxkxYoVvPHGG/zyyy8EBwdz2223MWvWLLp3745cLvcmSIuOjiY4OLjF86jUxc/Pjy5dunjPRSKR0Llz52Z118jlcoKCgtBoNKJYuQD07t2boqIi1qxZg8vlory8nFdeecUbp5ScnMyxY8e8bohjx45RXl5+zu3VxAvMnj3bO8Nozpw5/PTTT42W2bhxI6mpqT5iBWDAgAGsWLGiyfYam85sMplIS0vDbrdTXFzMvHnzGDZsmHf/8OHD2bx5M+vWravnDjqfcelyuVCr1Wi1Wu9SGafj1Vdf5ZdffmHhwoUsXLiQjz/+GICFCxeekVgBGD9+PEuXLuXgwYO4XC62bdvG1q1bmz0Xz9ChQ/n11185ceIEdrudtLQ0H7fT2LFjmTt3LmVlZTidTjZv3syLL76IIAheQTF79mzsdjtlZWWsX7/ep/6UlBS2bNmCIAg4nU4yMjLq9WH16tWYTCYsFgtffvklvXr1QqvVnnasDx06lOXLl3PkyBFcLhe7d+8mL6/+AoxNIVpYRHxwlJezf9UeVq7KISban9GXtSLou+fAaoF7X+KQu4K3qtK4P3YiHXQJ7M8z8MP6Izw/rSthAR7VXlVVxbZt24iLiyMlpeUyC0ulUtRqNQqFAqPRSEVFBVar1cfyUlhYSFpaGnv27EEmk9GjRw/69+9PaGioTz06ne6SSJAWFxdHWFiY1xTfnGJFtKo0L3VjWMCTh2XAgAF89dVXOBwOxo8fzyuvvMInn3zCxx9/TEhICEOHDqVz586A5/uumbUTERFBjx49vC6PGoYNG8asWbMYM2aMT+BsYzzyyCN8/vnnzJw5E5VKRefOnZvMVbJixYp6MSgAffv25f333yc/P59NmzZ5z3XmzJncfPPNjB8/nq1bt/L666/XKxsUFITZbObOO+/EbDYzYsQIJk6c6N2flJSESqXi4MGD3oDWGoYMGcLLL79Mjx49ePzxx097vrW58cYbvTOUIiIiuOuuu8jKyuKNN97wzrA6Fx566CEOHjzo/XzNNdcAMGHCBO6++27atm3LrFmz+OCDDyguLiYmJoYXXnih2ac0jxo1iqKiIh5++GHUajXTp0/3ztABmDhxIg6Hg4ceegi73U7r1q259dZbvf/zjz/+OO+99x7XXnstcXFxjBkzxieG5ZZbbuHtt9/mrrvuIiIigpSUFK+bqoaEhASefPJJCgsL6dSpkzdmSKPRNDnWu3XrxjXXXMMLL7yA2+1m/Pjx9VxEp+OiL37YkoiLH54dTlMlB5Zt4c8/sgkL8+Py8QmE7v4DyddvwENvIfQaytCSz8mmkoODZ+MnVTHutVXklFaz+80JKOUyBEFg6dKlGI1GJk2a1KJTZK1WK0ajkcrKSp8py263m3379rFmzRqysrIIDAykf//+9OrVy0cA1MRw1PUl/9MQY1VEWoLs7Gxef/11n6BP8M2e2hRffPEF1dXV9eJfRC5dxowZwzfffHNBkhg2hPgLJgKAy2Ila/V2/lqaTXCwivFjEwg15yP5/l0YORV6D+P76p2sdxxjSdcX0cjULN2VR9rBYn68byDKk9mG9+/fT1FREcOHD28RsSIIAlVVVRgMBqx11puy2Wxs27aNtWvXUlxcTHx8PDNmzPBON3Q6nbhcLuRyOQEBAWg0mmbv36WGaFURaSm2bt1Knz59zrl8eno6d9xxRzP2SOSfjihYRHDbHRxdvY0/fz+CTqdkwrhEwgMlSN56CqLi4IYHqHBbeNj4J1eG9GZCeF8cTjePzdlF/9ZhXN7DEwVuNBrZuXMnycnJxMfHN2sfHQ4HRqMRk8lUb6qwwWBg/fr1bNy4EYvFQpcuXbjuuutITEz0uVEHBAR4I+Vr8rE0Fbj7v0zNLIQLHYcj8u9h+vTp51SuuLiY++67j27duomp60XOClGw/Mtxu1wcW7ed3xdn4qeWc/n4RCIjNUi+fBWK8+HV70Gp5tHyn7FIXHzc4UEAPluRyfGSKubcN8ib42T9+vWoVCr69evXbP2rrq7GaDT6TMmsIScnhzVr1rBr1y6USiX9+vVjyJAh9XIzaLVa9Hq9j8VHLpej0Whwu93e6cV2u/2SXTH6bPDz8xODakUuGl26dGnSHRQeHt5kELDIpUvd9ZUuNKJg+RcjuAVy03azZOEB5HIpl09IIDraH8nWVbDqF7jtKWiVxEb7Mb627OTt1NuJVodQVmnjtV//5pqBiXSM8wR8/f3335SWljJ69Ojzfqp3uVxUVlZiMBjqTbFzu91kZGSwZs0ajh49SkhICJMmTaJPnz4+8Sk1ieGCgoKazKFSE7hbk3HX6XR6rS9NZdy9FBGtKiIiIv9kRMHyL0UQBPK3/81v8zJAgMsnJBATo0NSUgCfvwx9L4PhE3EKLm4vX0QnTRz3J0wC4KWF6bjcAi+enMZcXl7Onj17aNu2rc/qxWeLzWbzun3qWjqsVitbtmxh7dq1lJWVkZyczK233kqnTp18ZvZIpVICAwMJCgo66yDT2qn2/f39fVL5N2eujJZArVbj7+8vWlVERET+sYiC5V9KUXomv/2wE4fDzZVXJBIXG4BUcMGHT4NG57GuSCS8bVrPQVcpWzp9gEwi42C+kW/WHuHpKZ0ID/TzuoL8/f3PKdV4U0G04MmOuG7dOjZv3ozdbqd79+7cfPPN9abDyWQy9Ho9AQEBzbYejkwmw8/PDz8/PwRB8HEdnW3Co5aiZkq2aFURERH5pyMKln8hJQeyWfzNZixmJ5dfnkhiQiBSqQTmfQ5H9sFzn4O/juPOCl6qXMOtUaPpFdgGgMd+3EmU3o9ZY9oBsHv3bgwGA+PGjTsri4bT6cRoNGI0Guu5XgRBIDs7mzVr1pCeno6fnx+DBg1i8ODB3lTQNSgUCoKDg1tk1eTaSCQSn4y1tV1HFytwV61Wo9FoLnruGBEREZELgShY/mWUZZ9g8ZcbMJnsTBifSEpykEes7N0Ov86Gq+70ZLQF7qr4Ba3cj7faeaYersg4weq9hXx3z0BUChklJSX8/fffdOzY8YyWbwePa6eioqLBFOEul4s9e/awZs0ajh8/Tnh4ONOmTaN37971Utur1Wr0ej1arfY8r8i5UbPeUU3gbo3l5UIE7kqlUm+GXxEREZF/C6Jg+RdhyC/ht09WUVZmZdy4eFqnnhQrJgN89Ay07wFXzgBgoflvltqzmNPpcQLk/jhdbh77YSd9UkOZ2DsWp9PJ+vXrCQwMPKMVXcGTAbegoKDedrPZzKZNm1i3bh0Gg4HWrVtzxx130L59+3rWA39/f/R6/SW1unDt9YJqB+46HI5mt76oVCr8/f1Fq4qIiMi/DlGw/EswFRtY8tEKiorMjBkdT/u2wchkUhAE+PQFcDrhnpdAKqPKbeM+4xKGBXZieqRn/Y+vVmdxuKiS9XcPQCKRsGPHDqqqqrj88svP6OZZVVXF/v37KSkpISwsDL1eT3FxMWvXrmXr1q243W569uzJ0KFDiYmJqVdep9Oh1+vPaBHBi0ntwF3wWI1qr8x8rtYX0aoiIiLyb0cULP8Cqisq+f2jv8jLMzFqZBydOoZ4xArA0p9g1wb4z3sQ7Fn34injMsoEC190fBiJREJFtZ2XF/3N1f0T6JoQTGFhIQcOHKB79+711j1piMrKSn755Rfmzp2LIAhIJBKioqIoKCjA39+fESNGMHDgwAaXTa+Z8fO/GlQqk8mQyWTeadO1XUdnGrgrWlVERERERMHyj8ZUVkVBdjE7lmzn+FEDI4bH0qVz2Cmxkn0Q5nwAY6+B7p7Fz/Y4TvCReQtPJV5Lsr9nivLrv2Rgd7h46epuOBwO1q9fT2hoKJ06dTptHyorKzlw4IBXrIAnqPbEiRNMnDiRwYMH1xMjMpnMK1Saa8bPpYBEIkGpVHqtJLVdRw1Nm5ZKpfj7+1/yViURERGRC4EoWP6h7F6zn6Vfb/CKhDatg+jRPRy5/KRYsZrhgyehVRJcOwsAt+Dm1vKfSVRH8FTytQBkFZj4fGUWj0/qSGSQH2lpaVitVsaMGXPaJ/7KykoKCwspKSlp0BUSFxfnI1bkcrl3avK/wZpQE7gL+ATuOhwO74KF/4brICIiInImiILlH4TT4cJqtlF2wsDSr9dTWyNkZhkwm50EBJyMgZj9FpSXwGs/gMKz7ePqzex0FrC6y5sopR4h8cScXYQHqrlvbDvy8vLIysqiT58+9dw3dakRKwB5eXn19kskEkJDQwFQKpUEBwf/41dNboragbsiIiIiIvURBcv/OHarA6vZhs1sx+lwYa60sv2vDOoaNAQBKiqsHsGS9hesWwJ3PgfRnkUKC10mnjKt4JrwIQwL6QrA2n2FLEs/wey7+yMVXKSlpREZGUm7du2a7JPJZKKoqAhBEFi6dClLly4lJSWFI0eOeGNYrrnmGqKjo9Hr9fj7+7fEpRERERER+QchCpb/MQS3gM1i94oUt9ujTMoLjezblMWRjFwk0vpWCokE9Ho1FObBV6/BgDEwZIJ3/30VvyGVyvig/b0AuNxuHv1hJz2TQ5jSJ55169bhdDoZPHhwk1aQGrHidDqZO3cu27Zt44orruCyyy7DYDBQWlpKXFwcKSkpPosRioiIiIiINIUoWP4HcDldWM12bGY7duupqbGCWyDvcCF7N2ZRcLQE/wA/elzWgdY9Esg/XMy6BdtOWjRg3NgEAjQSeONJ0AXBLY97VAywzHqIBbZ9fNr2PkKVgQB8u/YIB/ONrH1+NMePHyc7O5uBAwc2aQ0xGo0UFxdjsVj48ssvOXr0KDfddBM9e/YEPKu0dunSRXR7iIiIiIicNaJguURx2BxekeKw+yYfczpcHN5znH2bDmMsrSQ0Rs/Qab1J6BCD0k9BQLCWhPYxdOoSQcW+Q+j1ao8r6Mf34dgheOEr0HgyxFoFB3dV/EofbSq3x44HwGi28+LPGVzVL572URoWLfqLVq1akZKS0mh/a8RKRUUFn3zyCQaDgXvuucdbRqfTER4eLgaRioiIiIicE6JguUQQBAG75WQ8isWOy1k/R4e50srBbUc4uO0oVoud+LbRDJjYnYg4T14Vrd4fjU7tddkE6DUExJ8Mjk3fAku+h+vug5SO3jpfNK4k121icafXveXeXLwXs93JC1d1ZePGjUgkEgYOHNioK8hgMFBSUkJeXh6ffvopMpmMhx56iMjISADCwsLqrQEkIiIiIiJyNoiC5SLidrk9VhSL5yW4G86CWl5oZN/mLI6k5yKVSWjdPYH2/VIICPZYSTQ6NTq9P1JZI9YLQxl8/Bx06Qfjr/duPuQs5u3qjcyKvYJOuiQAjhZV8snyTB65oj2W8hPk5uYybNiwRlPh14iVAwcO8NVXXxEeHs6dd95JQEAACoWCyMhIMVZFREREROS8EQXLBcZpd56KR7HVTxZWgyc+pYh9m7I4caT4ZHxKe1r3SETl55mGrFQpCAjxR6FqIgus2+0RKwhw1/Nw0iUjCAK3lS8iUhHIy61v9h7+1NzdhOpU3D4knr/+XEJiYiIJCQkNVl0jVjZv3sxPP/1E+/btmTFjhjcza0RExD8q8ZuIiIiIyMVDFCwXiGqTBbPJgtPhavI4p8PFkfQc9m3KwlDiiU8ZMq03iR1ivBYUmVyKTu+Pn/Y0lovifPjqdcjYAk/8F4JCvLu+Ne9gg+M4v3Z9Do3MU0/awSJ+35XHF3f0Zce2zcjlcvr169dg1RUVFZSUlPDHH3+wbNkyBg4cyNSpU5HJZISGhqLX68/i6oiIiIiIiDSNKFguEDV5UhrDUmXlwNajHNx2xBuf0v+K7kTEh/jEjmiDNGgDNQ1OXfZh0Vfw/O0gnIyFKSv07qpwW3jE+BcTgntyZfgA4OQ05u930j0xmE5BVrZlFjBy5MgGZ/RUVFRQWFjInDlz2L59O1deeSUjRozwuoAupZWURURERET+GYiC5QLhpmGxUlFkZO+mwxxJzzkVn9I3hYAQrc9xao0SXbAWueIMXCyFefBCLbEC8OWrnhiWkAgeMizBKnHxSYcHvbt/3JDN3lwDfz46gJ0719G6dWtatWpVr+ry8nJyc3P58ssvyc7OZsaMGfTo0QONRkNkZKToAhIRERERaRFEwXIBOHTiOCv2bCc8UE/bkHi0MhVl2Qb2bcoi/3AxmgA/uo9oT5uep+JTapArZAQEa1FplI3U3gBbVnpiV2rjdkNhLmlaM99advN66i208vOszlxpcfD8gj1M6ROL8fjf+Pn50bt373rVlpWVcfjwYT799FOMRqN32nJwcDAhISH1jhcREREREWkuRMHSwnz4y3wMi8qQChIqKGFv2yyCStVYSq0ERGrpO7krbTolIJP7WiYkUgnaIA3+AX5nt77O3u3w1oP1t0ulOCOiuaPiJzpq4ng4Yap31ztL9mGyOLilhx+HDxxm3Lhx9VZQLisrIz09nU8//RS5XM6DDz5ITEwMkZGRaDSas7omIiIiIiIiZ4soWFqQQyeOe8UKgBQJyoMCVXFW2l6TREicHolEwhFbAVqnGq3cD61Mjb9Og07vX0/EnJbNK+H+idC6M4ydDm8+BG6XZ2bQrU/yluIgB22lbOz4LjKJp+7jJVX896+DPDw6gaOH9tGhQwciIiJ8qi0rK2PDhg3Mnj2biIgI7rjjDiIiIoiMjPSuNiwiIiIiItKSiHebFuRAzjGvWKlBgoRv2q4nT7GYtmWxdFEm0lWVRHtlHFUyK346P/QaFw63myC3FoX0DL+iFQvhsWuh93D4v59B4w99RsCuNIiM5ViggpeL3uPmqFH0DWrvLfbMT7sJ0SlooywEhY7u3bv7VFtWVsaSJUuYP3++d9pyZGQkISEh/9qVlUVERERELjyiYGlB2sUl8Ldkv49ocUvc3Bk7njxNOXtsR1lcvYUfqtaglMjpqE2ghyOVnoGt6ahNQCVV4C9TEyTXEij3x0/WyBo8P38BL90Jo6+GV74Bxcl4l/AY6OBZx+fukq/xl/vxdts7vcU2Z5bwy/Zc/js5ApMhlwkTJvhYTEpKSvj+++9Zvnw5gwYN4uqrryYqKkpcXVlERERE5IIjCpYWpE10PEGTQzAsKkUqSHFL3NiGyRgQ1oaewET//ii0MvKU5eyszGKnKYt5hev4In8pKomCTrpEegak0iMglQ7aeHQyDUEKLUFyLVq5HwgCfPUGvP8EXHMvPP6+NzFcbRaY01lqz+L7jv8hUOERG263wH9+2MGI1loEQy5dunQhNDTUW6agoICPPvqIHTt2MHHiRMaNG0dUVFS92BYREREREZELgShYWphZk67iUJ/j7Nx3iHh9OEkhMTjdLtxKAalehl3qJMgVSGttK66JGoZLcHPYfIIdpkx2mDL5oWA1n+b9gUqqoIs2iR6BHgHTRZNA6qfvEzjnM9x3PYf0rue8qy/Xpspt437j7wwJ7Mh1USO823/alM2+nHI+GK1Ap9HTpUsX775jx47xzjvvkJ2dzcyZMxk+fDihoaGiC0hERERE5KIhCpYLQJvoeMKkQdgsdmRyGfrgANT+vu4dp9uF2W3F7LIRqgikszaR66KG4xLcZFbnsdOUxQ5TJt+dWMnnx35j9oJD9Nyex/ybp2K+sj99q3MJUwYRKPdHLvUE1ObZyphV/hNlbgtfdHzIKziqbU6em5/O/f38cNkrGTJ2lHcV5QMHDvDGG29gMpmYNWsW/fv3R6fTXdgLJiLyP8zvr8+m9cCutB7YrUXqX/vlL2gCtfSeNrJF6j8XDAWlLP9gLle9NqvZ69756xoMBaWMuGtag/s3z1mKf0ggnUf3b/a2LwW+mPk809+6H13o2WcPz0zbTWbaHiY8PrMFenZ65j7yLkNunUh028RmqU8ULBeImmnK2iBNg5YKuVRGgNSfAPmp+BCX4MLispOgjmCgviNml41qcwWRT91N9K4TPH/baN5qb8W87200UhVddcl0D0hlsL4jxy3FPJb1JW4EJMD6ir9J9fckgnvvj/3oZWbilHa6d+/pXUl527ZtvP322ygUCh5//HG6du2KUnkW+V9ERP4l/P76bIqP5CFT+P6Ejn9sRrO3NfeRd5nw+IxzumGdLX+89S0nDmRz1auzCIw8lVspfelGdv+2DofVjkKtpPe0kbQf3su7//ieQ8R1ad0sffj99dl0nzj0jG9y/a4de17tnTiYzZ9vfYdCdeq3ThAEHFa7j1DYv3o7+1Ztw2KsIjg2ggHXj0MfE35ebYucHaJguUAEhenO2qUik8g8U53lJ1PdVxoR/nMj7NuD6d2fmNl3MFOcZnadtL7sNGXxVf5S/pu72KceAbhj//uMDu2JUKXhk7/282x/CA8Lp0OHDgCsWLGCDz/8kMjISB555BFSUlK8VhcREZH69Jw8nM5jB7RoG06bHYfV1qJt1FBVZsBYUEq7oT3J3LiHXlNOuZC7jB1AUq/2/PTo+0x/837UOt/A+5w9h+g5afh598HldGKrtpx3PWdLZGqcjxXi6PZ9bJ2/An99gPdzxl+bGP3AtQSGB5O1OYPfX/+Ga95+ALlKfKi7UIiC5QJx3vEfpUVw1xgk+cfgi5UEdu1P4MldHbUJTIscjNllw+SsZnb+cl7OnuNT3IUnNua7+VVMSnGilgoMHjwYqVTKvHnz+Pbbb+nQoQOPPvoo4eHiU4OISHNRkV/M5jl/UXq8AL8AfzqP6U/rQd2QSCQNmuxrzOhqfw1LXvsau8XGwmc+Ia5rG4bfMQUAh83Big9/oiDzOLpQPYNvvoKQ2EgAbNUWtvy0jNy/DyOVSUnq1YFeU0YgU8ipLK3gp0ffZ/idU9jz+wZadUyhz9WjAMjcmE7rgd1I6t2Bv979kR6Thp3RQ4u1yoyhoJSI1FjAY7FY9+WvdLtiCOl/pOGw2Unt35lek0fgFtz8eP/bTHz2dq8Fx+Vw8uODbzP+PzP4/fXZ2K02lr03h5C4KK540rOSvOB2s372b+TsOYRK68eAG8Z7LTBrv/wFXWgQPSYOQ3C72fHLGrJ37MdiqiYiJZaBN45HGxLUaP81gVpadUrxfhYEgfQ/N9JpdD+kJ5ca2b96O13+v737jovqTBs+/pth6L0KCChFQUQUFRF7LNFETTHZJJreNmWzb/Lsm2STfbObYt7sk919d5N9drObtsYUY5qxR6PYCyiCoiKgIiDSex2YYeb94+AwQ7UgTMz1/Xz8BO7T7nPPCeea+77OfW6eimegMjt45PQ4svemce7IKUZMGUvOvnSy9x1lSHgwpw8cw9HNman33UxTbQNHN+2lvryaEVPGMnnpgl7vBfpWHSlf/UjuoZM4uDkz6c45XZanrtlB7uGTGNoMBI+JIHHpAuyclBfYVhWWsu/TjdQUVeAzLICgmHCL7TsPLx1Zu5P6ihpmPXa76dq48dmlHP52O021DQyLi1L276ikMJScLiDlqx+pLanA2cudiUtmM2xcJAAGg4G0dbvI3pOOxk5zTYJ5+Qr9U3AhDx6cBpWl8MkeGGc5VqtSqXCyccDHzp0wp0CeCF6IutNHa4Oa+jIHTuTkMt5Xx6RJ8Tg7O/OPf/yDFStWMGPGDF5//XUJVoToR63NLWz+y2eEJcRw37vPc+OzSzm6aS+Fx8/0ua1X8BAefO9lAO5Y/pQpWAGoyCsi4e4bue+d5/EZFsC+lRtNy3Z99D0qtZq7//vX3Ln8aerKqkj9fqfFvvUtOhb/7hFTsGI0GjlzMIPIGePxChqCi5c7xafOXdI5ns84TdDocNPNHaC5rhF7JwfuWP4Ut/3+MYoyz3FiWzIaW1tCJ0ZzNuV4x/bHz+AR4It3iD8PvvcyLl7uzH9umSlYAagsKGH0nEks+9v/JmJyLDvfX4PRaOxSl6rCUppqG1j88iPc9cdf4+juwoEvfui1/h4BvoxbON30e9Gpc9SXVxM5vSMHqaa4whSsXOQZ6EdNcYXp99riCkLGjWTp//svIqbE8uPfV1NXVsWi3z7Eba8+QdbutD4/9/QNe6gqLOWON5/m1lceo6m2wWL5oW+2UVtSyZLXn+SePz2Lja2GvZ9sAJTeqa3vfsnwuCjue+d5pj24iJKcgl6P153m2gZu+T+Pctdbz9BUXcfhb7cDSg/c1ndWEbd4Bvf//UVmPnIre/6zjqrzpQDk7E3n3OFMFr/8MHe8+TQOLk40d6r/1ZKAxdqdPgEPTFUeYf50P4wc0+cmQQ6+fBD9LDbtH68Nav4d/Sz/+DqXB0YbCAwMJDQ0lDfeeIONGzdyzz338MILL8gU+0JchtQ1O1j59B9N/1a/+E6Xdc5n5ODs6UrUjPGobWxwH+LNqNnxZO1Ju6pjDx0dhpufF2obG0ZOH0dFfjEATbX1FBzLIXHpfGwd7LFzcmD8LTPJ3pNmcYMfOT0OO0cH0++lpwtw8/PC1ccDgKhZE8jZd/SS6pJ/NJuQ9m/ZFzm4OhE6MRqNnS0u3h6MXTiNM8lKkBIxJZYzycdN9TmbfJyIxNhej+EXEYx3iD9qtZrIGXE01dR3ezP0Dglg5iO34ujmjKObMxEJMVQVlpqWf/f790yf15Z3vuj2WMc27yd6drxFTktLQ1OXfCWNnQZtfZPpd3d/b/xHhKBWqxmRGIu2oYnQidHYOtjj5utJQOQwKgtKADhzMMPi2inJyVfaIuU44xZNx8ndBTtHe6JmdEzkaTQYyNqTxqRfzMXBxQmNnS0Tl8wm9/BJWpu1lJ0tpE2nZ8yCKag1Nrj5eTEszvJzuRSR7deGg6szE2+fbfqsziQfZ+joMELGjkSlVuMzPJDwhBhy9h9Vzin5OKPnJeDm52UKTB3c+nfOLhkSsmZHD8KvFkJACPx7K/gM6Xubdo8G3cR8n4mcaSoiwimQA2mNRNjtxtVeTVxcHC+++CJnz57lueeeY8GCBdfwJIS4Pl1KDktDZS0V+cWsfPqPpjKj0YhHgE8vW/XNfFjByd0Fg77NdDyMsOo3f+04HmDQ6Wlt1na7PUDOvqMUnjzLfx5fbtpGpVYxtUlrGm7oTpteT9Gpc8x46JYe6wdKANNYVQdAwMhh6Ft0VOYX4zbEm8ITZ5j6wMLLOF9X07E7a6qp5+CXWyg9fZ5WbQtGgwEHl44vYncsf7rX41TkF1OSk8+sx2+3KLd3caJNZ3k8faseB9eOfZvX0dHNRSlTd5Q5uDmb6hyRGNttkNZUXW9RX5XZkJy2oYm2Vj3r3vzI4lg2Ghvqy2toqqnHwcXy3XPmx79U5sd0cHOitUmLvlVHQ2Ut+WlZFteywWAgeMwIpe41lnWHfkiF6EQCFmu1bwv85g4YNQH+sQFc3fveppMgB1+CHHxpatHzxdad3DbcQMzoGF566SVqa2tZvnw5cXHX5tFLIQS4eLvjFx5sMbxhTq3RdLkRXu3xVGoV9737QpceAYDWJm2XMl1LK3lpp7jvnectbjhb311F7uGTRM2c0OPxirPz8Q7x7xLUGA2WwzXa+iacPJVAQ6VWEz45hjPJx/EO8ScganiXG92VSvlmG61NWm79/WM4ebhSnJXH7o/XXvL2GT/sZ8SUsTi5u1iUewT4UFNcgf/IYaay6qKyXtvmSjh5uqJt6Oi1MRgMpp8dXJywsdOw5PUncfPz6rJta7MWbUMzRqPRFCgYOn0Oao2N5fXWdVQNg8Fgyl3S1jVh52jf3lPmTtikGG745ZLu6+5hWXfoeh1cLRkSskabv4RfL1beBfT+1isKVsz9c1MG84Y24ujqzttvv41Op+Nvf/ubBCtCXGPBsSNpqKjhTPJxDG1tNNXWs/29rynKUvJDvEP8qbpQRn1FNdCeg1FTf8XHc3J3JTh2BIe+3Y6+pZXW5hbS1u8mfeOeHrfJO3IKn2GBXYKG0AmjTN39PSk4ms2wsV2HHbQNTZxLzUSv09FQWcPRjXstehQiEmPJT8/ibMqJrj0NV/Gt3NhmQGNvh52TA811jeSlZV3ytnVlVZxLzWTMgsQuy6Jnx3Ns8z5qissx6NvI2ZdObXElYROju9nTlQtPGMPRjXtpqm2gtbmFnL3ppmUqtZqoGeM5/F0SLY3N6Ft1ZO0+wr5Plfwlv7AgbDQ2ZGw5gEHfRl15NQXp2Rb79xkWQP7RHEBJ4C053TXHJWdvOq3NLWjrG0n9fgcRibGoVCoiJo/hwsmzXMjMxWgwUFdezeY/f2rK44mYPIaT21KoK6tCr9NxLjUTbX1jv7aP9LBYm9XvwVvPwOL74bWP4Cqnwr9Q2UjZuROM8IBvV6/C39+fN9980zT3ihDiyqSu2UH6BstAYOFvH2L4hFGkfLUNvU5P9A3x3PS/7+PAqh848PlmnD1dCUuIISByOACegb7E3zGHTX9aiYu3B0ExETh5WE7UGD55DN+/9gGRM8aTcFffk8XNeux2kldvZfWL76KxtyUgKpTEpT0P++bsP0rohK433pBxkez5ZAO1JZXkpWeRvn43AKtffJdJv5jHqBsmkn80h4UvPNBlW0c3Z3TaFr575V/otC1EJMYSMy/BtNw72B8bW1vKzhYy95m7LM930mi2/+NrgmLCmf3knX2er7mJS2az66O1fPHcX3D19WTKspuoyCtix/vfWSQtd+f41oMEjx2Jh3/X4bqw+NFo6xvZ9j9f0VzXiFeQHwtfeqjfH2mOWzyD5NVb+faVf3Y7OeCkX8zj8HdJfPf7fwHgFxHElPZ5aGxsNcx/dhl7V27g2Ka9+IYOZWhMuEXQMu3BRez5zzpyD53A1ccTz6G+tDZbPjbv7OnG+v/7EU01DQyLiyT+zrkAuHh7MPeZu0lZvZXa0krc/LyImjnBNLw5cnoc9RU1rH/rP9ja2zJu4XQcO/VUXS2VsbtU6wFkMBpJSk4jLfM02tZWQgL8WDJnOu6u3SfrHDyWyYGjJ6lvbMLfx4vb50xjiHf3EyqtWLuFh28b/PyMlF1JnM04TnjsGBJmzel+JaMR/r0c3nsV7v8veP4v3b4X6HKPu3d3Kl5DvNh/OJmgoCBefvllHBx6HpO+lhrOZ1OXdxK34aNxCb78ZLCfC2mnvkkbDb6q86XseP9b7nzzVxblFx9rXvqX/+p1++TVW2ltbmHGw7f0up4YGBcfa358xWuDXZUeDXoPy57UDLJyC3jirkU4Oziw+0gGH63ZzG8euLNLwk5GTi67U4/x8G0L8PFwJ+3Uaf799QZefmwZdt2M11qD93//GupC5eeM9L0cTdrLE8tfs1zJYIC3n4NV/wPPvgWPvnRV3aLmx3UH2vIuMMLNhxf+8AdszB49HEhZq99l39ZKjKhRcZRp872JuufZQamLNZN26pu0kXXIP5ZNSDfDQZeqOCuPyUvn92ONxPVu0O/yyRmZLJkzHQ9XpetodkIch46fIu9CCaFBARbrHjyWyaz4caYelfiYSA6fyOL46VwmRPfPtND9KWVXkilYuUhdCH9+5U2MFwMsoxHH5nrsdLY03/IqrRdU8MbbV3VclU6PxwXLMu86R/av+pDomIFvp+aKUvZureRiypQRNfu2VuLs/SWOl/Hk0/VO2qlvPbVR0NRs6WkZYHGLZlzRdg2VNax940OGRodZJLEK0ZdBDVh0ej1VtfX4mc1CqFap8PPypKyqpkvAUlZV3WX4Z4i3J+VVNd3uv66hiRVrt5h+HxcVQVxURLfrXgtnM453W+5xQQ+YPxlgB9jhXA3OdM3i7y/ZO0rI3lFyzfbfO8vhLSNqtqzKBrK7X/1nS9qpb13bqC4vUwIWKxEYFdrrcJCLtwf3vfvCANZIXApXH0+rHg6CQQ5YmtrfkWGrsayGrcaGRm3XG3dTcwuaTkMathoNjc3d3+TdXJwGNYclPHYMGel7u5RHzvYnOmwo/OMvUFgATz0Ho/qeEO5SZZ7I6TYwiZztP2g9LFtWncL8RqPCwPxlo6TnwIy0U996aiO34f37tMZVWTsLoh5S/l0LSQ+Bkz8k/ve12f+VqM6GH26BZdcgsD70GtRkwY2ru1++7zlwCYFxv+n/Y/8UfDYcZn8CQ2dd/rYXdsGOh+D+vP6s0aW7zP9XBjVgcXJQ3k+g6zQBkE7fhnM3iaFOjvbo29o6ravH2XFwkkj7kjBrDkeT9loMCxmCYMaCW+GXN0J5EfzrBxgzqV+POyNuNqdyXut63Puf7NfjXI7pleZ5BwamzfcmeN7SQauPtZJ26lt3bTTgvStrZ0FpMth0+ttz685uV78qnw2HW3eB2/D+33dn6+dC4Q4l8PAY0VGe/mdIXQ66erB1hcS3IeapjuV562HY4v6pw9pZEP/apd+Ap71z9cc0GuDgbyHjXfAYCfec6H69Uyvg8Ksw6yMIudFy2dk1cGQ51Ocr+0j8MwRO734/4ooMasBiq9Hg6eZKWVWNKYfFYDRSVlWNr1fXOUJ8PT0oq6wmdKi/qay0spqE2FEDVufL9cTy15SnhI4fJ3zMGBLCh8P9U0Gvg0/3Qdi1qXuX4/b0dNIAibrnWYKmZlOXl4nb8Gjpvu+BtFPfrKaNJr0Jcc9f22PomqC17toe46L6AqjOgtFPQPZKSHizY1ncCxD+C/g8FO7LBcdOj/7mbYBJy6++Dm2toK26+v1cDl0TJN0PrbVKIHbq4+7XO/k+ZH4At+3pGjyWpsCeJ+GmdeA3CQq2wOZFcM9JcAm65qfwczHoE8cljo1m6/7D1NY3otPr2ZmSjr2dHaFD/Uk/dYYPv9uEvn3a6SnjRrPr8DHKqmpoazOQejKb8upaxowIG+Sz6F1CVCTLEsaSYNemBCu2dvDZ/msWrJiOO2sOy3793KAHKxe5BEcSOP12uQn3Qdqpbz+pNqrKhPXz4GNvWDUKTv1HmcYAIOsTpUfB3GfDla76yuOwMhBaquGrWNh2b8c6+kb4YQn8xwe+iYeKjI5l2mrY8Qh8EgCfhsD+56Gtfa6Nujx4TwWnV8PqWDhglkuS/SlEPQwxv1J+Nlj2ZvdIW6kM2QS0v6bgwi7lHDI/hi9GKPU48CK06UCvhY88oOZ0x/ZtLfCxF1QcgxV+UHUCNi2C7816Jwx62Pk4rBgCX0Yrx7go6SFl2AiUnpKUV2BVJHzsCRsXKoFYb8pTwXEILPwB7LufIoOqk3DkLVj8Y/c9XSffh1GPgX8iqG1g+EKlxylrhWWbZPwdPgtVroPctVC0F9ZMU+q67V6lfXpjaINDf4CVQ+HzCDj5QdflR/4In4crbfnDEmgs6lhefx423qS093eJUHvWcvuL195Fna/P91Rw5hv4eryyjx+WQFPHu5p6vdaNRjj+T/g8DFYGw+E3ej/Xbgx6wDJjYiyRw4P519freevDVeQXl/LokptQqVTUNTZSXlVDW/v0xLEjw5g+YQyfrv+R5e9/RurJHJ74xSKrfaQZgDUfw/xh8OhseGI+2DnAyr3K+4GEENe31nrYMA8i7oGHS2HhBuXGV7Cl7229x8BjNcrPd2fAPLOX9ZWlwpQ/w0Ml4DsedpsN9+54EFQ2cO9puPs41J1RbnLm9E2wZJ+yD1BuJjmfQfRj4B0DLsFw4RKHt/I3Q9CNoDb7O9xUCvYeSr3vSIELScpwi8YBwu+E0192rFuwBTxHgc9Y5XxdQ2DhRrjdLP+v4iiMeQYeLIIR98L2eztuhOYqM6CxGG7fB8tylFyfff+r9/oHzoCZ74FNL5N0Hn5N6WHa/oASNK2dBeVmL7CsPgWenfKovEYrPVbmbeI+QqlXwpuwfZly7jevV8qK93UEOD3JWqEEDLftUYatHLyhySxfMeNdOLdWGUJ84AL4ToAtS5S2Mhph21JwHwkPXoAF30HZod6P153qTCVwe6AA7N1h56NKeV/XemESpL4BC9YovXT+icq+LsOg3+nVKhU3TpnIjVMmdlk2c+JYZk4ca1E2ZdxopowbPVDVuzolhfD6L5V5Vi4qLYTWlp63EUL8NBx6BY6YDZs4eCl/iM3lbwLnoRDd/kfdPQJinoZTH8Gwm6782MHzwD1c+TnqYVg7U/m5qQTyNsLjdWDbPsvoxD8o33onmyXpRj1sOddTyX5wiwDX9seMo59Qvl0Hz+27LnkbIKzTLLKOvhDeXuYaAnEvQdpbyhDayPuUAGvi75U65KxSynrjP0UJaEBpy0OvKOfqbPkkKT7jYLbZkM6IpbD7lx2/r46FhvYel4BpSmDUF6MRzm9TznHOStA4QtofYdNCJSi0dVF6mTSd8pk0jqCtsGyTi5956K2g0kDQHOW6ARi2UAnMQAleNi3q2HbGezByGZxeBbHPdnz24XfAfrMnsjI/gKl/Bddg5fcJL8PRPym9VnbuUHJACTY0jsq/iLvh/Na+28DcyPs7hgQT/6T05LTUKIFJb9f66VXKdeczTlkePA88oi7r0IMesFzXCk5bBiugdNmdPwP+Mq4pxE/apeSwNBQo38Q/8ugoM7YpPQpXxaxz3MkfDK3Kz/UFgBFWmv99MSrDLq21HUWdJ6bM+gQKf4T322+6RqMytNFSq3yL7klbK1zYAbM+tCxXdeq8d/SFhvanAAJngK4RKtKVHofzW5Uejt6oOp0vdAxzmWssVm7gxftBV6f8vTXPt7kno+s2fWkuV9pu3PMd+5r0htKbUXYYht6g9HR0Hs7RN4OD2bHNz0GtUQIV8zJHP6jPU34OmNbRu2ZxfkVd84fM99FQAFvvsuztamuFunPK/jVOYOvU/baXyqLOvu31utD3td5Y1LUX6jKPLwHLtRQyQple3zxoUdtA8MDNBSOEGEQuIUrXt/nwhjkbe2jrdKO7mreluIYow0EPlyr77qylpmuZrhFy18BDpR3f9gE23wJnv1GGiXpStFv5xtw5qOmc/9Jcrnz7BuUmNWIp5HyhbDt0lnLD7w8Hf6uc4x3J4BwIRbtgx8NXt08HL1DbdnNzVSkBAIBnlDIsZK7qJHj0c46VcyA0V1iWmbe1SwjM+KfSc9NZfYEyFKhr6ghajJ0+p87XY3fXovk2zeXt9Rra97XuHGjZ49Td8fsw6Dks1zX/IHj1AyVIAeW/r74vvStC/FwMW6gkup7+UkkcbSpRvgFfTGz0HgeVJ5R1QPm5qfjKj+fkD8NuhuSX258wqofUN5Vcgp7kfq/kOpgHKwBhS5Sel97kbej+cWZthfKYr16r3CjT/ggjzZKGR94H59Ypyb9dhoOu4rUkRj3YOiv5M81lSmLr1VJrYMQySH1dac82nZLY6jy0Y3hj9JNw6kPlaSFDm5LXk79BGQLpTyOWKj07tWeVtj27BrTlHcujf6l83o1FSj3PrVMSY41GJS/JP1G5NvRapTfqzNeW+/cZr3ymRqOyfdGurnXI+VwZAtM1KEnbITcr7d3XtT5iqZKDU3FUWV6YBDWXN2+P9LBca0sehSnzlWGg4AgJVoS4XnTOYQFlHpawJcofcr0WYp6ExVuVxM89zyg3uRH3QGB7zonXKJj8FqyfDa7DIXi+8k3U3Ihl8M0Epacj8RJe2zF7JRz4jfI0hq0TBN7Q+1wl2Z90zUEBGL4Ydj2uPNFzbq0yDwso+018W7lJ522AW5K6bus0RBmS+WqM8lj2yPuV3IuLfGKVHIrSZCX501zE3bDlDgiZD/NW9X2+5iYtVyZC+yQA3EJh2t+Vp4C23WuZtGxOr4UP2/N9jAbACP9qvzXe+JWSJzL9f5Temy9HKUNR/lNg4aaOXqwhCTDjX7DrCWjIVxJbb97QkUvSX6IeUeZ5+X4aaJxhwu86hshAaeM2LayZqgxJ+cUridUXhwDnfal8pisDlaGaUY9Z5rAkvg1JD8JXY5WnoXzilOOZ84qBDTdCXa5yHc/+j1Ju59r7tR40Fya8Aj/cpgR1o5/sOkTUh0F/W/O1ZC1vaxZCiOtO5XHlqZPOk6xd6uyp+59XckNu+LD39YT1eE8F950bmEkMuyFDQkIIIS5fT8NBl6poJ0Te33/1Edc9GRISQghx+Sb87sq2qy+Ab+MhaB4EyNT14tJJwCKEEKL/DJ3V+3CQa4jyFJP46Xl6cDNIZEhICCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD0JWIQQQghh9SRgEUIIIYTVk4BFCCGEEFZPAhYhhBBCWD3NYB783IUSNu1Opry6Fk83F+YkxDFmZFiP67/41w9wsLezKLv1hilMiB55rasqhBBCiEE0aAFLTX0DK77/gaU3zyFyeBDnS8r5ZN1W3F1dCAnw63G7N3710MBVUgghhBBWYdCGhA6fyCYqNIRRYSGo1WqGBQ4hIXYUKRmnBqtKQgghhLBSg9bDUlZVg7+3p0XZEG9Pzp4v6nW7/ekn2Jd2AoPRQFxUBPMSJ2JjI6k4QgghxPXsmgUs9Y1N/PmTr7tddvvsaTRptWg0NhblthoNjc3aHvcZOTwYB3s7fvPAnVTV1fPJuq042NszK35st+vXNTSxYu0W0+/joiKIi4q4grMRQgghxGC6ZgGLq7NTr/kmmbn56PVtFmU6vR5nR4cet3l0yU2mn4d4ezI5NprMs/k9BizB/r4smTv98iouhBBCCKszaGMpfl4elFbVWJSVVlbj6+ne4zatOl2XMttOvTTmJFgRQgghrg+DFrDEx0SSlZvPqdwCDAYD+UWlpGScYnJsNKAMKf376w2cKbgAQFFZBX9e8TXnCotpMxgor64lOSOTcVHhg3UKQgghhBggg5Z06+HqwsO3LWDjnmRW/7ADDzdXlsyZZnqkuVWnp6yqhob2nJZAPx9unp7Ad9v3UtvQiKuzE9PiYmQOFiGEEOJnQGU0Go2DXQkhhBBCiN7I88BCCCGEsHoSsAghhBDC6knAIoQQQgirJwGLEEIIIayeBCxCCCGEsHoSsAghhBDC6knAIoQQQgirJwGLEEIIIayeBCxCCCGEsHoSsAghhBDC6knAIoQQQgirJwGLEEIIIayeBCxCCCGEsHoSsAghhBDC6knAIoQQQgirJwGLEEIIIayeBCxCCCGEsHoSsAghhBDC6knAIoQQQgirJwGLEEIIIayeBCxCCCGEsHoSsPSj9Kwzg10FqydtdGmknfombSTEz4sELP3oqPwB7ZO00aWRduqbtJEQPy8SsAghhBDC6knAIoQQQgirpzIajcbBrsT1Ij3rDHFREd0uW7N9L7UNjQNcI+tT19CEm4vTYFfD6kk79U2nb+OXdy4c7GoIIQaIBCxCCCGEsHoyJCSEEEIIqycBixBCCCGsngQsQgghhLB6ErAIIYQQwuppBrsC1wOD0UhSchppmafRtrYSEuDHkjnTcXd1HuyqDahj2Wf5btsetK06Xn3qAZwdHUzLjp8+R1JyGtV1Dfh6urNwRgKhQQGm5bX1jaxJ2ktBcRkOdnbEjYpgbuIE1CrVYJzKNVHX0MT6XQc4e74IgIiQQG6ZNQVXZ+VpIGkjKC6vYuuBwxSWlKPT6wkJGMItN0zB19MdgHMXSti0O5ny6lo83VyYkxDHmJFhpu2bW1pZm7SPM+eLsFGriQ4LYdHMRDQam8E6JSFEP5Eeln6wJzWDrNwCnrhrEb97bBnB/n58tGYzP5cHsIxGIzsPHWXz3kPcc/PsLssListYs30vt8+ZxqtP3c+cyeNZsXYLNfUNpu0/WrOZYH8/fvf4Mp64axHZ586zJzVjoE/lmjEajXy64UccHez53ePLeOnRpbS1GVi7Yz8gbXRRdl4B8aMjeenRpbz06FLcnJ1Ym7QPgJr6BlZ8/wNzJo/n1afu5/Y501iTtI+C4jLT9l9uTsLOVsMLD9/Fs/ctobq+gQ27Dw7W6Qgh+pEELP0gOSOT+VPj8XB1wdZWw+yEOFpaW8m7UDLYVRsQjc1azpeU8czSW/H39uqyPDkjk0ljohgWOAS1Ws2osBBGhQ0j9WQOoHxr1ra0MjshDluNBg9XFxZMiyc5I3OgT+WaMQI3Jk5k8czJ2Go02NvZEh8Txbn2a0TaSDErfhyjI4ZjY6OmSduCtrUVf1/lmjp8Ipuo0BBGhYWgVqsZFjiEhNhRpGScAqCqtp4zBUUsnDEZBzs7nB0dWDwzkSOZOeh0+sE8LSFEP5Ahoauk0+upqq3Hz9vDVKZWqfDz8qSsqsaiS/965eLkyAO33AgoN43OyqpqiAgealE2xMeTkorq9uXV+Hp5WAxtDPH2oqq2Hr2+7brozlerVIwcHmRRlldUYhrqkDbqcPhEFut2HqBVp2fGhFgWzkgAlDby9/a0WHeIt6dpiK28ugZ3F2cc7O1My7093GgzGKhpaDS1tRDip0l6WK5Sk7YFAFuNZexnq7GhUasdjCpZnabmli43VFuNhqZmbcdym87Lld8vtu/15nxJOYeOZ7FgajwgbWQuPiaK5c88zMuPLaOwtJwdKekANGm13bZRo3kbdVquUqks2lEI8dMlActVcnKwB5SeFnM6fRvODg7dbfKz4+Roj17fZlGm0+txak/KdXK0R9/Webny+8X2vZ7kXShh5bqt3DV/lqkHTtrIkkqlMiXV7j96EgAnB4du28jZvI06LTcajRbtKIT46ZKA5SrZajR4urlSVlVjKjMYjaYufAF+Xh6UVVVblJVWVJu66H09PSivqsFglqRcWlmFl7vrdTXUAXAkM4cvNiVx76K5jAoLMZVLGykqa+osC8yGwPy8PCg1+/8MoLTSso1qGxrRtrZa7E+tUuPh8vN6Yk+I65EELP0gcWw0W/cfpra+EZ1ez86UdOzt7Agd6j/YVbMKk2OjSTmeRUFxGQaDgVO5BZzKzSc+JhKA0KAAHOzt2JmSjk6vp7a+kS37DzM5NnqQa96/tuw/zO7UDJ66+5Yu14a0EVRU1/LO59+RnnWGVp2ehqZmdh46ytjIcADiYyLJys3nVG4BBoOB/KJSUjJOmdrAy92ViJBANu1JoaVVR1Ozlg27DzJx9EhsbSVdT4ifOnn5YT8wGI1sP3iEtFOnaWnVEezvy5K50/FwdRnsqg2Yf65ex/niMowo3fAqlQoVSjB36+ypHM/JZXtKOjV19fh4urNwxmTCzBKSa+obWLN9H+dLyrC3s2X8qBHX3RwjL/71A2w1GmxsLL8nLL1pNqPCQqSNgNMFF9i8N4WK6loc7e0ZFxnOvMQJpoDjXGExG/ckU1Fdi4ebK3M7z8OibeH7Hfs5e/4CapWa6PBhLJZ5WIS4LkjAIoQQQgirJ0NCQgghhLB6ErAIIYQQwupJwCKEEEIIqycBixBCCCGsngQsQgghhLB6ErAIIYQQwupJwCKEEEIIqyfTP4qfrK+37kKFil/Mn3lJ658puMDnG7fz8mPLsLezvSZ1yi8q5ceDqVworcDGRk2wvx/Tx48hPDgQgNSTOSSlpPHbR+65JscXQojrlUwcJwZUVW09//3xl/zpN7+86n21GQwA2KgvvaNQr2+7ZrOeVtbU8bfPvmPKuGgmxUShbW3lxJk89qRm8OiSmwgPDsRoNNLWZpCZV4UQ4jJJD4sYUPWNTf22r8sJVC662kAh9WQ2G3cnMy9xAlPjYiyWZeedx8nBjpumTULVPl1+0BBfYiKGE+jnAyhvIZZgRQghLp8ELGLA/Hggle3JaYDyXp2woACevGsxqSez2XbwCAtnTGbHoXTcXZx5+LYFFJdXsmX/YfIulGJnq2FsZDg3TZtkehfPV1t2AXD3glkA/PGjVcyKH0duYTE5eedxd3Xh9tlTCW1/H8/Z80W8/81GU+/OjwdSyS0sZlxUBLtTj9HSqiNhzChunDLBFHBU1daxbucBzhUWY29nB8Cvlt6Gi5Njl/NTq9U0t7TSpG3B2dHBVB40xNf088VzffmxZab6dDYheqTpnDLP5rMt+QjlVTV4u7sxa9I44qIirvQjEEKInywJWMSAmRU/jqF+Pqxc/yOvPvUAarMekvrGZorLq7hj7gx8PN0xGo1s3nuIUWEhLJqZSEV1LRt2HcTZ0YEbJo3r8RgZOblMnzCG+VMmsj35CJ9vSuLlR5f22KtRXl2DTq/n4dsWUFRWweofdhLs70t0+DAAvtiYhLurM0/ctZiWVh0bdh0k9UQ2N02f1GVfY0aEkpScxl8++Zr40ZGMCgshJHBIjz1BwwP9efWpB0y/5xeV8vnG7UwZp7x9OCe/kG9+3M3tc6YR7O9LUVkl327bg62NDTEjQvtsbyGEuJ5IwCIGjJ2tBgd7pZfCvAcCwNHBjvlTJ1qUPXL7AlNPh6+nO/lFJeQWFvcasIwZEUp0mBJs3DQ9gbQPvqC8uoYAX+9u13d3cWH6+DEA+Hl5cPhkNmfPFxEdPoy2NgOFZRXMnxrP0PYhnbGRYaSfOtNtwOLs6MCz9y1h5+GjZJzOZVfqMVycHJmXOIHEsdFd1rexUZvaobFZy/dJ+1g8K5Fgfz8AdqSkM3fyeGLb30bs6eZKWVUNe9NOSMAihPjZkYBFWAWNTdcekKxz59lzJIOK6lpadDr0+jZCAvx63Y+dbccl7ebsBECTtuWS1r+4zcX1lad8fMnMzSfY3xedvo3T+Rfw8/bscX8uTo4snpnI4vZeob1px/k+aR/urs6mQKozg9HIl5t3EBoUwOTYUabykooqzpeUs/VAqqmsra3NNDQlhBA/JxKwCKtUXl3DyvVbmT0pjgVT43F1dmJf2nGKyisveR8Xe2cuj+U29y6cw4ffbubgsUyMRiPhwYEsnpl4SXvy8XTn9jnTKCqr5HReYY8BS1JyGrUNjdy/eF6XOi+ckUBUaIhlDa/ovIQQ4qdNAhZhlYrLq1Cr1MydPN6U66K+gqeCrlZGzjliR4YyNS4GO1vbXudvKa+uoby61iIwMRqNtOp0OHeTpAtwOr+QvUcy+NXS27rs29/Hi5q6BrzcXU1lxeWVuLu6XOVZCSHET4/MdCsG1MWgo1nbQqtO3+N6Q7w9aWtrY1fqMcqqajh8IouscwW0GQxoW1oHqrqknTqNh5sSIOjb2tDr23pc91h2LivX/ciPB1IpLC2nrKqGTXtSqKytY1xkeJf1a+sbWbV5B/OnxuPi5Ehjs5bGZi3N7ec3JyGO/UdPkJJxisqaOo7n5PLhd5s5ln322pysEEJYMelhEQMqeIgvIQF+vPreSovHdzsb4u3JkrnTSUpJZ/fhDEYOH8pDt85n5fof+W77Hu5dOHdA6hs7IpQ12/exZvs+U1mAjxd333QDgZ0SeeckxOHj4ca+9BPsTs3AVmND0BBfnvjFInw83bvs+3RBIY3NWtbtPMC6nQdM5Rcf9x4xLIh7F85l28EjrN91AHcXF+ZOHm+R5yKEED8XMtOtED04X1LOpj3JPHTrfNPTTfWNTWzak0KrTscDt9w4yDUUQoifDxkSEqIH50vKKKuqoaC4jPrGJmobGimrqqG2oVHySIQQYoBJD4sQPWjV6di8J4XM3AIamppRqcDV2YnosGEsmBaPne21eYGiEEKIriRgEUIIIYTVkyEhIYQQQlg9CViEEEIIYfUkYBFCCCGE1ZOARQghhBBWTwIWIYQQQlg9CViEEEIIYfUkYBFCCCGE1ZOARQghhBBWTwIWIYQQQli9/w8VxgRqbnLg3AAAAABJRU5ErkJggg==", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "metric = 'kappa'\n", + "\n", + "fig, ax = plt.subplots()\n", + "\n", + "for model_name in grouped.index.levels[0]:\n", + " subset = grouped.loc[model_name]\n", + " ax.plot(\n", + " subset.index, \n", + " subset[metric]['mean'], \n", + " label=model_name,\n", + " marker='o', \n", + " )\n", + "\n", + " ax.fill_between(\n", + " subset.index,\n", + " subset[metric]['mean'] - subset[metric]['sem'],\n", + " subset[metric]['mean'] + subset[metric]['sem'],\n", + " alpha=0.2\n", + " )\n", + " \n", + "\n", + "range_frame(\n", + " ax, \n", + " np.array([df['train_size'].min(), df['train_size'].max()]),\n", + " np.array([df[metric].min(), df[metric].max()])\n", + ")\n", + "\n", + "matplotx.line_labels(ax)\n", + "ax.set_xlabel('training Size')\n", + "ylabel_top(\"$\\kappa$ score\", ax, y_pad=0.01)\n", + "\n", + "fig.savefig('kappa_learning_curves.pdf', bbox_inches='tight')" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "chemlift", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.9.18" + }, + "orig_nbformat": 4 + }, + "nbformat": 4, + "nbformat_minor": 2 +}