From 073fc335093c5efda0814a88616e9215434659af Mon Sep 17 00:00:00 2001 From: Giorgio Gilestro Date: Mon, 7 Sep 2026 10:58:46 +0100 Subject: [PATCH] Accessibility pass: introduce every pop-gen term at first use, with real-world anchors The CS reader is now walked into the biology as it arrives: drift (surname extinction, island alleles), Wright-Fisher and heterozygosity defined in place (collision-probability reading), Muller's ratchet (Y-chromosome decay), immigration (the one-migrant-per-generation rule of conservation management), blending inheritance (Jenkin's 1867 swamping argument, which the Proposition makes exact), Fisher-Muller (clonal competition vs assembly), genotype/locus, epistasis and NK landscapes glossed, outbreeding depression (the Tatra ibex restocking collapse), panmixia, BDM incompatibilities (mule sterility), hybrid load. Three new literature anchors (Mills & Allendorf 1996; Jenkin 1867; Templeton 1986), all verified; references renumbered to first-appearance order (now 72) and re-verified 1..72. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_01BkRLcc18rwT2Lysu6PbG7v --- paper/pnas/body.tex | 37 ++++---- paper/pnas/main.md | 219 ++++++++++++++++++++++++++------------------ paper/pnas/main.pdf | Bin 303180 -> 307583 bytes 3 files changed, 149 insertions(+), 107 deletions(-) diff --git a/paper/pnas/body.tex b/paper/pnas/body.tex index 7f045ac..09c96ca 100644 --- a/paper/pnas/body.tex +++ b/paper/pnas/body.tex @@ -14,17 +14,17 @@ Machine learning has become a population-scale phenomenon. Public repositories h The generations are coupled through data as well as through weights. Successive models increasingly learn from model output rather than from fresh human experience: frontier alignment pipelines are now predominantly synthetic (over 98\% in documented cases; 11, 12), self-generated instruction data seeds whole lineages of descendants (13), a large and growing share of the public web is machine-generated or machine-translated text (14, 15), and the stock of human text is projected to be exhausted by frontier training within this decade (16). Meanwhile persistent multi-agent systems and emerging agent economies put many interacting models into sustained contact (17--20). A population whose members inherit from one another, recombine, and retransmit under these conditions is an evolving population in the technical sense, and that observation motivates this work. Here I transfer the quantitative framework of the branch of biology built for exactly this situation, the population genetics of the evolution of sex, and use it to treat multigenerational model populations as systems whose inheritance, diversity, and compatibility can be measured, predicted, and managed. -Training each generation of a model on the previous generation's output degrades it (\emph{model collapse}): rare capabilities vanish first, and the lineage drifts toward its own most common behaviour (21). That degradation is, mathematically, \emph{genetic drift}, the loss of rare variants that any finite population suffers when each generation is a finite sample of the last. The identification has been made repeatedly and independently: for sequential inference chains before deep learning (22), for language-model text ecosystems (23), as a closed-form first-extinction law placing collapse onset at the Wright--Fisher first-extinction time (24), and in quantitative-genetic form for self-consuming diffusion models (25). A diagnosis reached so often, from such different starting points, marks population genetics as the natural mathematics of the setting, though only as its entry point: population genetics is not, at heart, a theory of decay; it is a theory of the mechanisms that maintain and build populations despite decay (immigration, recombination, selection, population structure) and of where those mechanisms reach their limits. This paper develops that fuller structure for model populations: the arc from drift through its remedies to its limit, reproductive isolation, carried as one framework from closed forms to trained networks to language models. +Training each generation of a model on the previous generation's output degrades it (\emph{model collapse}): rare capabilities vanish first, and the lineage drifts toward its own most common behaviour (21). That degradation is, mathematically, \emph{genetic drift}, the loss of rare variants that any finite population suffers when each generation is a finite sample of the last --- the same sampling accident by which rare surnames vanish from small villages and rare alleles (gene variants) drift out of island populations with no selection against them. The identification has been made repeatedly and independently: for sequential inference chains before deep learning (22), for language-model text ecosystems (23), as a closed-form first-extinction law placing collapse onset at the Wright--Fisher first-extinction time (24), and in quantitative-genetic form for self-consuming diffusion models (25). A diagnosis reached so often, from such different starting points, marks population genetics as the natural mathematics of the setting, though only as its entry point: population genetics is not, at heart, a theory of decay; it is a theory of the mechanisms that maintain and build populations despite decay (immigration, recombination, selection, population structure) and of where those mechanisms reach their limits. This paper develops that fuller structure for model populations: the arc from drift through its remedies to its limit, reproductive isolation --- the point at which diverged lineages can no longer produce working offspring, biology's boundary between species --- carried as one framework from closed forms to trained networks to language models. An operator of a model population faces recurring decisions for which there is no principled guidance: how much verified real data does retraining need before a lineage decays; will combining two particular models compose their abilities or damage them; can incompatibility be detected before paying for a failed merge; and when should specialists be kept separate rather than consolidated? In practice these are settled by convention and by trial-and-error search. They are also, recognisably, machine learning's oldest problem at a new scale: \emph{continual learning}, the struggle to acquire new abilities without losing old ones (26, 27), transposed from a single network to a population whose members inherit from one another. Population genetics, I will argue, prices these decisions. Table 1 summarises the correspondences on which the argument runs; the sections that follow develop them from closed-form theory to experiments in trained networks and language models. \section*{The minimal model, and where its exactness ends} -Knowledge is modelled as a distribution \texttt{p\_t} over \texttt{K} discrete items (capabilities, facts, modes of behaviour), with a fixed true distribution \texttt{p*} whose rare tail carries the knowledge most at risk. One generation is: *draw \texttt{n} samples from the parent's distribution, optionally mix in \texttt{m} verified real samples (``grounding'', \texttt{g = m/(n+m)}), and refit the child\emph{. In this minimal inheritance model the resampling step }is* the Wright--Fisher process: the same equations, which I exploit as an engineering gate: the simulator reproduces the classical closed forms (heterozygosity decay \texttt{E[H\_t] = H\_0(1 − 1/n)\textasciicircum{}t}; the exact immigration--drift equilibrium; the closed-form multi-teacher union) to within 0.5\%, and these are standing tests in the codebase, not one-off checks. +Knowledge is modelled as a distribution \texttt{p\_t} over \texttt{K} discrete items (capabilities, facts, modes of behaviour), with a fixed true distribution \texttt{p*} whose rare tail carries the knowledge most at risk. One generation is: *draw \texttt{n} samples from the parent's distribution, optionally mix in \texttt{m} verified real samples (``grounding'', \texttt{g = m/(n+m)}), and refit the child\emph{. In this minimal inheritance model the resampling step }is* the Wright--Fisher process, population genetics' canonical model of neutral evolution, in which each new generation is a random sample of size \texttt{n} from the previous one and every statistical property of drift follows from that one step. Diversity throughout this paper is \emph{heterozygosity}, \texttt{H = 1 − Σ p\_i²}: the probability that two random draws differ (one minus a collision probability), high when many items share the mass, zero at total collapse. The identity is exploited as an engineering gate: the simulator reproduces the classical closed forms (heterozygosity decay \texttt{E[H\_t] = H\_0(1 − 1/n)\textasciicircum{}t}; the exact immigration--drift equilibrium; the closed-form multi-teacher union) to within 0.5\%, and these are standing tests in the codebase, not one-off checks. -The boundary of the exactness matters, and I measured it rather than assumed it. Real training adds approximation, optimisation noise, and inductive bias, and when trained networks are fit against the exact drift null they deviate in \emph{opposite, architecture-specific} directions: a smoothing recurrent network resists collapse (keeping spurious variants alive), while a sharpening image generator accelerates it. A one-parameter \emph{learning kernel} (a smoothing knob and a sharpening knob on the refit) reproduces both. Throughout, a real learner is therefore treated as Wright--Fisher \emph{plus a signed, measurable estimator bias}, and the drift signs (rare-first loss; the grounding response) survived that bias in every architecture I tested, including a convolutional VAE retrained on its own generated digits, where the dry lineage collapses to a single blurred digit class while 10\% grounding holds all thirty modes (Fig. 1). Retraining on a single parent is \emph{asexual reproduction}, and sustained loss under it carries the defining consequence of \emph{Muller's ratchet} (28): once every copy of a rare capability is gone from all parents and sources, no recombination can rebuild it, so remedies must act before fixation-by-loss (a consequence-level correspondence; the minimal model lacks the ratchet's recurrent-mutation driver). +The boundary of the exactness matters, and I measured it rather than assumed it. Real training adds approximation, optimisation noise, and inductive bias, and when trained networks are fit against the exact drift null they deviate in \emph{opposite, architecture-specific} directions: a smoothing recurrent network resists collapse (keeping spurious variants alive), while a sharpening image generator accelerates it. A one-parameter \emph{learning kernel} (a smoothing knob and a sharpening knob on the refit) reproduces both. Throughout, a real learner is therefore treated as Wright--Fisher \emph{plus a signed, measurable estimator bias}, and the drift signs (rare-first loss; the grounding response) survived that bias in every architecture I tested, including a convolutional VAE retrained on its own generated digits, where the dry lineage collapses to a single blurred digit class while 10\% grounding holds all thirty modes (Fig. 1). Retraining on a single parent is \emph{asexual reproduction}, and sustained loss under it carries the defining consequence of \emph{Muller's ratchet} (28), the mechanism by which lineages that never recombine decay irreversibly --- the reason non-recombining genomes such as the Y chromosome have shed most of their ancestral genes. Once every copy of a rare capability is gone from all parents and sources, no recombination can rebuild it: each such loss is a click of the ratchet, and remedies must act while copies still survive somewhere (a consequence-level correspondence; the minimal model lacks the ratchet's recurrent-mutation driver). -\textbf{Table 1.} The dictionary. Each correspondence is stated with the level of support it currently has (exact = closed form in the minimal model; empirical = measured in trained systems; hypothesis = stated with a falsifier, untested or unconfirmed). The full claim-by-claim ledger with assumptions and known limits is SI Appendix, Table S1. +\textbf{Table 1.} The dictionary. Each biological term is introduced in the section that develops it; each correspondence is stated with the level of support it currently has (exact = closed form in the minimal model; empirical = measured in trained systems; hypothesis = stated with a falsifier, untested or unconfirmed). The full claim-by-claim ledger with assumptions and known limits is SI Appendix, Table S1. \medskip\noindent\begin{center}\footnotesize \begin{tabular}{p{0.307\textwidth} p{0.307\textwidth} p{0.307\textwidth}} @@ -45,7 +45,7 @@ Selection on a fitness function & Verifier-anchored selection (``reality that ca \subsection*{Grounding is immigration: cheap, with a floor} -In the minimal model, grounding from a fixed real source is \emph{immigration} into a drifting population (29--31), and the equilibrium diversity has a closed form the simulator matches exactly. That equilibrium is \emph{smooth} in the grounding fraction (there is no phase transition in aggregate diversity), so the practical number is an operational threshold, and I define it as such: under the tested population size and Zipf source distribution, \texttt{g \(\approx\) 0.05} retained most (\(\geq\)95\%) of equilibrium diversity indefinitely, with the required fraction depending on sample size, source distribution, and the chosen retention target (dependencies in SI). Verified real data remains, on any of these definitions, cheap insurance at fractions far below one. But the same analysis yields a floor the field's average-loss framing misses: under unstratified sampling from the source, a capability of rarity \texttt{p} appears in a real-data batch of size \texttt{m} with probability \texttt{1 − e\textasciicircum{}{−m\(\cdot\)p}}, so \texttt{m\(\cdot\)p \(\approx\) 1} marks roughly a 63\% chance of one example per batch: a soft observation floor, with higher confidence priced accordingly, and with distinct consequences for continuous retention, stationary occupancy, and reintroduction after loss (immigration can restore an absent item; SI separates these). Protecting the rarest knowledge under unstratified grounding is therefore priced per item at cost \texttt{\(\propto\) 1/p}; targeted or stratified sampling changes that cost, and recombination can recover rare capabilities \emph{that are still retained across complementary parents} (next section). In trained networks the \emph{sign} of the grounding response transfers everywhere I looked, with two deviations, both traced to the estimator bias above: sharp thresholds soften, and support-counting metrics decouple from truth (forward-KL is the operative collapse metric for a smoothing learner). On real images (Fig. 1B), dry self-training collapses a convolutional VAE to one mode while \textasciitilde{}10\% grounding holds all thirty (the trained model needs roughly twice the exact-operator fraction, the measured price of the estimator bias). +In the minimal model, grounding from a fixed real source is \emph{immigration} into a drifting population (29--31). Immigration is what conservation managers prescribe when a fragmented reserve loses diversity, and its striking property there is how little is needed --- the field's rule of thumb is that one migrant per generation holds an isolated population's diversity (32). The same economy appears here: the equilibrium diversity has a closed form the simulator matches exactly. That equilibrium is \emph{smooth} in the grounding fraction (there is no phase transition in aggregate diversity), so the practical number is an operational threshold, and I define it as such: under the tested population size and Zipf source distribution, \texttt{g \(\approx\) 0.05} retained most (\(\geq\)95\%) of equilibrium diversity indefinitely, with the required fraction depending on sample size, source distribution, and the chosen retention target (dependencies in SI). Verified real data remains, on any of these definitions, cheap insurance at fractions far below one. But the same analysis yields a floor the field's average-loss framing misses: under unstratified sampling from the source, a capability of rarity \texttt{p} appears in a real-data batch of size \texttt{m} with probability \texttt{1 − e\textasciicircum{}{−m\(\cdot\)p}}, so \texttt{m\(\cdot\)p \(\approx\) 1} marks roughly a 63\% chance of one example per batch: a soft observation floor, with higher confidence priced accordingly, and with distinct consequences for continuous retention, stationary occupancy, and reintroduction after loss (immigration can restore an absent item; SI separates these). Protecting the rarest knowledge under unstratified grounding is therefore priced per item at cost \texttt{\(\propto\) 1/p}; targeted or stratified sampling changes that cost, and recombination can recover rare capabilities \emph{that are still retained across complementary parents} (next section). In trained networks the \emph{sign} of the grounding response transfers everywhere I looked, with two deviations, both traced to the estimator bias above: sharp thresholds soften, and support-counting metrics decouple from truth (forward-KL is the operative collapse metric for a smoothing learner). On real images (Fig. 1B), dry self-training collapses a convolutional VAE to one mode while \textasciitilde{}10\% grounding holds all thirty (the trained model needs roughly twice the exact-operator fraction, the measured price of the estimator bias). \begin{figure*}[p]\centering % fig1 \includegraphics[width=\textwidth]{figs/fig1.pdf} @@ -54,11 +54,11 @@ In the minimal model, grounding from a fixed real source is \emph{immigration} i \subsection*{Recombination: a conservation law, its operators, and offspring that exceed every parent} -The largest returns from the transfer concern merging. \textbf{Proposition (blending inheritance, rare-item regime).} Let K parents independently retain a rare item (mass \texttt{p} when retained), and let the child draw \texttt{n} samples either from one parent chosen at random or from the \emph{mean of the parents' output distributions}. Expected item mass is identical under the two schemes; and in the rare-item regime \texttt{n\(\cdot\)p/K \(\ll\) 1}, where per-item survival is first-order in sampled mass, expected \emph{survival} is also identical: the 1/K dilution of averaging cancels the K-parent union gain to first order, so in this regime adding parents through the output-mean does not increase expected tail retention. Two boundaries: outside that regime, survival is a convex function of mixed mass, so the variance reduction from averaging can \emph{reduce} extinction relative to a randomly chosen single parent; the cancellation is a first-order result about rare items, not a universal impossibility; and the contrasting union operator (keep each item's strongest source, then renormalise, which itself redistributes mass and presupposes a verifier or oracle to identify the strongest source) increases expected retention with K in all regimes in the minimal model. The practically important operators, \emph{weight averaging} (a nonlinear network's weight-mean does not compute its parents' output-mean) and \emph{routing among intact specialists} (32) (different storage and inference budgets from a single child), are its empirical cousins, and the measured bridge is a \emph{headroom rule}, stated qualitatively: in language models, union-preserving operators beat the weight-average where that average falls short of attainable performance, and add nothing where it does not (easy-versus-hard contrasts at two scales; a quantitative form of the relationship is untested). On easy tasks a capable base's average is already at ceiling and refinements add nothing; on hard tasks the average dilutes a fragile specialist below even the best single parent and routing wins by a wide margin (Fig. 6A--B). +The largest returns from the transfer concern merging. \emph{Blending inheritance} --- offspring as the average of their parents --- is the failure mode at the root of population genetics' founding controversy: the swamping argument pressed in Jenkin's 1867 review of \emph{The Origin of Species}, that under blending a rare advantageous variant is diluted toward the common type faster than selection can multiply it (33), an objection dissolved only by Mendel's particulate inheritance, in which discrete variants pass through generations undiluted. Refitting a child model to the mean of its parents' output distributions is blending inheritance, and the proposition below is Jenkin's dilution made exact. \textbf{Proposition (blending inheritance, rare-item regime).} Let K parents independently retain a rare item (mass \texttt{p} when retained), and let the child draw \texttt{n} samples either from one parent chosen at random or from the \emph{mean of the parents' output distributions}. Expected item mass is identical under the two schemes; and in the rare-item regime \texttt{n\(\cdot\)p/K \(\ll\) 1}, where per-item survival is first-order in sampled mass, expected \emph{survival} is also identical: the 1/K dilution of averaging cancels the K-parent union gain to first order, so in this regime adding parents through the output-mean does not increase expected tail retention. Two boundaries: outside that regime, survival is a convex function of mixed mass, so the variance reduction from averaging can \emph{reduce} extinction relative to a randomly chosen single parent; the cancellation is a first-order result about rare items, not a universal impossibility; and the contrasting union operator (keep each item's strongest source, then renormalise, which itself redistributes mass and presupposes a verifier or oracle to identify the strongest source) increases expected retention with K in all regimes in the minimal model. The practically important operators, \emph{weight averaging} (a nonlinear network's weight-mean does not compute its parents' output-mean) and \emph{routing among intact specialists} (34) (different storage and inference budgets from a single child), are its empirical cousins, and the measured bridge is a \emph{headroom rule}, stated qualitatively: in language models, union-preserving operators beat the weight-average where that average falls short of attainable performance, and add nothing where it does not (easy-versus-hard contrasts at two scales; a quantitative form of the relationship is untested). On easy tasks a capable base's average is already at ceiling and refinements add nothing; on hard tasks the average dilutes a fragile specialist below even the best single parent and routing wins by a wide margin (Fig. 6A--B). -The generative payoff is the \emph{Fisher--Muller effect} (33, 34): recombination assembles, in one offspring, complementary variants that arose in different lineages, producing a genotype fitter than any parent. In the multi-locus model, sexual merging of decorrelated specialists climbs to the global optimum, a genotype no parent held, while the best single parent and the blended average both plateau below (Fig. 2). In real language models the signature replicates under seed replication: merges of three LoRA (35) specialists beat every parent overall (decisively at 7B: 0.87 vs 0.77), and on the sharper worst-family metric the merged models are the only ones competent everywhere, in every seed (Fig. 6A). +The generative payoff is the \emph{Fisher--Muller effect} (35, 36), the classical account of why sex speeds adaptation: in an asexual population, beneficial variants arising in different individuals can only compete until all but one lineage is lost, whereas recombination assembles them in one offspring, producing a \emph{genotype} (an individual's combination of variants, one at each \emph{locus}, or position) fitter than any parent. In the multi-locus model, sexual merging of decorrelated specialists climbs to the global optimum, a genotype no parent held, while the best single parent and the blended average both plateau below (Fig. 2). In real language models the signature replicates under seed replication: merges of three LoRA (37) specialists beat every parent overall (decisively at 7B: 0.87 vs 0.77), and on the sharper worst-family metric the merged models are the only ones competent everywhere, in every seed (Fig. 6A). -Sex has risks and, for AI, an unfair advantage, both quantified on rugged (epistatic) NK landscapes (36) (Fig. 3). When skills are entangled, blind recombination produces offspring \emph{below} their parents (outbreeding depression), worsening with ruggedness, and the optimal recombination rate shrinks as entanglement grows. But an engineered population can do what biology cannot: recombine unbounded parents, choose complementary mates, and \emph{screen many candidate offspring against a verifier before keeping one}. This directed sex converts the outbreeding catastrophe into a reliable gain in the model (tracking or exceeding the best parent at every ruggedness) and replicates as a sign in language models: bred-and-screened merges beat the a-priori blend in every seed on headroom tasks, including one seed where the blend failed catastrophically and selection was immune (Fig. 6A). Finally, population \emph{structure} is itself a knob: sweeping the mate-pool breadth from monogamous (local) to promiscuous (panmictic) against ruggedness, wide mixing maximises the population mean while monotonically destroying diversity, and the best \emph{champion} shifts from wide breadth on smooth landscapes to intermediate breadth on rugged ones (Fig. 3C), the mating-system phenomenon known to structured-population search, mapped onto merging populations. +Sex has risks and, for AI, an unfair advantage, both quantified on Kauffman's NK fitness landscapes (38), the standard model of \emph{epistasis}, biology's term for interaction between genes: the fitness contribution of a variant depends on which variants occupy the other loci, much as a component's value in an ML system depends on the components around it. Each of the landscape's \texttt{N} sites interacts with \texttt{K} others (the model's eponymous parameters), and raising that interaction count tunes the landscape from smooth and additive to rugged and many-peaked (Fig. 3). When skills are entangled, blind recombination produces offspring \emph{below} their parents, worsening with ruggedness, and the optimal recombination rate shrinks as entanglement grows. Biology knows this failure as \emph{outbreeding depression}, the reason conservation practice warns against crossing locally adapted populations: in the textbook case, an ibex herd in the Tatra Mountains restocked with animals from Turkey and Sinai produced fertile hybrids that bore their young in the coldest month of winter, and the herd died out (39). But an engineered population can do what biology cannot: recombine unbounded parents, choose complementary mates, and \emph{screen many candidate offspring against a verifier before keeping one}. This directed sex converts the outbreeding catastrophe into a reliable gain in the model (tracking or exceeding the best parent at every ruggedness) and replicates as a sign in language models: bred-and-screened merges beat the a-priori blend in every seed on headroom tasks, including one seed where the blend failed catastrophically and selection was immune (Fig. 6A). Finally, population \emph{structure} is itself a knob: sweeping the mate-pool breadth from monogamous (repeated local pairings) to promiscuous (\emph{panmixia}: any model may merge with any other) against ruggedness, wide mixing maximises the population mean while monotonically destroying diversity, and the best \emph{champion} shifts from wide breadth on smooth landscapes to intermediate breadth on rugged ones (Fig. 3C), the mating-system phenomenon known to structured-population search, mapped onto merging populations. \begin{figure*}[p]\centering % fig2 \includegraphics[width=\textwidth]{figs/fig2.pdf} @@ -72,7 +72,7 @@ Sex has risks and, for AI, an unfair advantage, both quantified on rugged (epist \subsection*{The society: grounding, recombination, and diversity make complementary contributions} -Composing the operators (Fig. 4) requires one definitional distinction first. In the inheritance model, grounding is \emph{grounded inheritance}: external samples added to the reproduction process (the data channel). In the society model, grounding is \emph{grounded evaluation}: selection weights true fitness against conformity to the population's own consensus, \texttt{g}\(\cdot\)true-fitness + (1−g)\(\cdot\)conformity, the analogue of scoring models by the crowd's approval (the fitness channel). These are related design ideas, since both couple the lineage to a non-drifting external signal, but they are different operators, and I name them separately. In the tested society (a finite agent population on a rugged NK landscape), a four-arm ablation separates the failure modes: the full system (grounded evaluation + directed recombination + diversity-preserving selection (37)) climbs to near the global optimum while keeping its specialists; removing grounded evaluation converges the population confidently on an unfit consensus (self-consumption); removing recombination strands it on local optima; removing diversity converges it prematurely to a worse answer. Each removal fails differently; the three implementations make complementary contributions \emph{under the tested conditions}; general joint necessity is not established (alternative mutation, restart, archive, or selection schemes could alter the picture). At language-model scale this composed loop remains unbuilt; it is the paper's largest stated gap. +Composing the operators (Fig. 4) requires one definitional distinction first. In the inheritance model, grounding is \emph{grounded inheritance}: external samples added to the reproduction process (the data channel). In the society model, grounding is \emph{grounded evaluation}: selection weights true fitness against conformity to the population's own consensus, \texttt{g}\(\cdot\)true-fitness + (1−g)\(\cdot\)conformity, the analogue of scoring models by the crowd's approval (the fitness channel). These are related design ideas, since both couple the lineage to a non-drifting external signal, but they are different operators, and I name them separately. In the tested society (a finite agent population on a rugged NK landscape), a four-arm ablation separates the failure modes: the full system (grounded evaluation + directed recombination + diversity-preserving selection (40)) climbs to near the global optimum while keeping its specialists; removing grounded evaluation converges the population confidently on an unfit consensus (self-consumption); removing recombination strands it on local optima; removing diversity converges it prematurely to a worse answer. Each removal fails differently; the three implementations make complementary contributions \emph{under the tested conditions}; general joint necessity is not established (alternative mutation, restart, archive, or selection schemes could alter the picture). At language-model scale this composed loop remains unbuilt; it is the paper's largest stated gap. \begin{figure*}[p]\centering % fig4 \includegraphics[width=\textwidth]{figs/fig4.pdf} @@ -81,11 +81,11 @@ Composing the operators (Fig. 4) requires one definitional distinction first. In \subsection*{The limit of sex: model speciation} -Recombination presupposes compatible parents. In biology, lineages pushed far enough apart become separate species (\emph{reproductive isolation}) through Bateson--Dobzhansky--Muller incompatibilities (38, 39): changes harmless on their own background but deleterious in combination. A merged model is exactly the exposed hybrid. I built the analytic model (Fig. 5A): hybrid fitness tracks the parents while compatible, then peels off and crashes below the ancestor; the isolation cliff arrives earlier the denser the incompatibilities; and the incompatibility \emph{count} snowballs quadratically with divergence (39). Note that a super-linear count does not by itself entail a sharp performance cliff without the count-to-effect-size link, which the analytic model supplies under its assumptions and any neural test must establish separately. +Recombination presupposes compatible parents. In biology, lineages pushed far enough apart become separate species (\emph{reproductive isolation}) through Bateson--Dobzhansky--Muller incompatibilities (41, 42): changes harmless on their own genetic background but deleterious in combination --- the mechanism behind the mule's sterility and the inviability of many between-species crosses, in which two genomes that each work perfectly cannot run in the same cell. A merged model is exactly the exposed hybrid. I built the analytic model (Fig. 5A): hybrid fitness tracks the parents while compatible, then peels off and crashes below the ancestor; the isolation cliff arrives earlier the denser the incompatibilities; and the incompatibility \emph{count} snowballs quadratically with divergence (42). Note that a super-linear count does not by itself entail a sharp performance cliff without the count-to-effect-size link, which the analytic model supplies under its assumptions and any neural test must establish separately. -In trained networks, the claim must survive a known alternative: merge barriers between independently trained networks are famously \emph{coordinate artefacts}, removable by re-aligning hidden units (40); richer symmetry groups remove more (41), with known failures beyond the shared-data regime (42). I therefore aligned under the composition of permutation matching and exact per-unit rescaling (the unit symmetry group of plain ReLU MLPs, as the search space) and decomposed the barrier (Fig. 5 C and D): two networks trained from different initialisations on the \emph{same} task have a barrier that this alignment removes essentially entirely (residual \(\approx\) 0.001, the aligned merge performing at parent level): coordinate, not functional; two networks trained on \emph{conflicting} label maps have a barrier the same alignment leaves largely unchanged (0.502 \(\rightarrow\) 0.497), with the merged model functionally dead. The tested alignment removes the same-task barrier but leaves the conflict-associated barrier intact, supporting a functional-conflict interpretation without proving optimal alignment: exact recovery of a permuted-and-rescaled copy validates a special case, so the removable share is a lower bound and the residual an upper bound. Sweeping conflict traces the cliff as hybrid fitness, 0.97 \(\rightarrow\) 0.03. The conflict floor itself is information-theoretic (no single model can satisfy contradictory conventions; SI Appendix, Proposition S2), with the framework's role being the \emph{structure around it}: which divergences generate conflict, and what moves the cliff. +In trained networks, the claim must survive a known alternative: merge barriers between independently trained networks are famously \emph{coordinate artefacts}, removable by re-aligning hidden units (43); richer symmetry groups remove more (44), with known failures beyond the shared-data regime (45). I therefore aligned under the composition of permutation matching and exact per-unit rescaling (the unit symmetry group of plain ReLU MLPs, as the search space) and decomposed the barrier (Fig. 5 C and D): two networks trained from different initialisations on the \emph{same} task have a barrier that this alignment removes essentially entirely (residual \(\approx\) 0.001, the aligned merge performing at parent level): coordinate, not functional; two networks trained on \emph{conflicting} label maps have a barrier the same alignment leaves largely unchanged (0.502 \(\rightarrow\) 0.497), with the merged model functionally dead. The tested alignment removes the same-task barrier but leaves the conflict-associated barrier intact, supporting a functional-conflict interpretation without proving optimal alignment: exact recovery of a permuted-and-rescaled copy validates a special case, so the removable share is a lower bound and the residual an upper bound. Sweeping conflict traces the cliff as hybrid fitness, 0.97 \(\rightarrow\) 0.03. The conflict floor itself is information-theoretic (no single model can satisfy contradictory conventions; SI Appendix, Proposition S2), with the framework's role being the \emph{structure around it}: which divergences generate conflict, and what moves the cliff. -The pre-registered \emph{emergent test} constrains the claim most: true BDM incompatibilities are emergent (each lineage's changes harmless alone), so I let children diverge with \emph{no conflicting signal anywhere}, using complementary class specialists and divergent input conventions, to 6.4\(\times\) the base training. No isolation emerged (residual 0.000 throughout); instead the merge \emph{rescued} the two catastrophically-forgetting specialists (parents \(\approx\) 0.50, merge \(\approx\) 0.955, a sustained Fisher--Muller rescue). The same double result appears at the language-model tier (Fig. 5 E and F): conflicting conventions produce \emph{function-specific} hybrid breakdown (the merge scores below both parents on the conflicted function, while a budget-controlled design shows the disjoint skills merge unharmed), and over-training disjoint specialists 1\(\rightarrow\)12 epochs (cf. the merging literature's expert-duration effect; 43) produces no isolation at all --- the merge improves. Across every tier tested, isolation had to be provoked by functional conflict; specialisation alone did not speciate --- a bound on the analogy that sharpens the design rule: what breaks merging is conflicting conventions on shared circuitry, not divergence per se. +The pre-registered \emph{emergent test} constrains the claim most: true BDM incompatibilities are emergent (each lineage's changes harmless alone), so I let children diverge with \emph{no conflicting signal anywhere}, using complementary class specialists and divergent input conventions, to 6.4\(\times\) the base training. No isolation emerged (residual 0.000 throughout); instead the merge \emph{rescued} the two catastrophically-forgetting specialists (parents \(\approx\) 0.50, merge \(\approx\) 0.955, a sustained Fisher--Muller rescue). The same double result appears at the language-model tier (Fig. 5 E and F): conflicting conventions produce \emph{function-specific} hybrid breakdown (the merge scores below both parents on the conflicted function, while a budget-controlled design shows the disjoint skills merge unharmed), and over-training disjoint specialists 1\(\rightarrow\)12 epochs (cf. the merging literature's expert-duration effect; 46) produces no isolation at all --- the merge improves. Across every tier tested, isolation had to be provoked by functional conflict; specialisation alone did not speciate --- a bound on the analogy that sharpens the design rule: what breaks merging is conflicting conventions on shared circuitry, not divergence per se. \begin{figure*}[p]\centering % fig5 \includegraphics[width=\textwidth]{figs/fig5.pdf} @@ -94,7 +94,7 @@ The pre-registered \emph{emergent test} constrains the claim most: true BDM inco \subsection*{A controlled predictive test: functional conflict, measured pre-merge, predicts merge damage} -The framework's prediction-level claim was put to a designed test (Fig. 6C). Thirty-nine parent pairs (13 conditions \(\times\) 3 seeds; rows are not independent --- parents share task-data seeds across conditions, so inference is condition-clustered, and because shared seeds also couple rows \emph{across} conditions I report per-seed and leave-one-seed-out sensitivity alongside) span three axes decorrelated by construction: \emph{conflict} (contradictory conventions on shared prompts, private budgets fixed), \emph{compatible overlap} (the same shared prompts under the same convention --- overlap and volume without conflict), and \emph{duration} (weight divergence with zero conflict). Before merging, six predictors are computed: \emph{confidence-weighted functional conflict} (bilateral confident disagreement on probes drawn blind to where conflict lives --- a proposed proxy for merge-relevant interactions, motivated by the observation that raw disagreement counts harmless complementation, one parent merely ignorant, as conflict), raw disagreement, gradient alignment at the shared base (44), LoRA-delta cosine and distance, and a cross-task performance baseline. The pre-registered outcome is the merge penalty against oracle parent potential (the hybrid-load analogue), also reported against best- and mean-parent references because the predictor ordering is sensitive to that choice. +The framework's prediction-level claim was put to a designed test (Fig. 6C). Thirty-nine parent pairs (13 conditions \(\times\) 3 seeds; rows are not independent --- parents share task-data seeds across conditions, so inference is condition-clustered, and because shared seeds also couple rows \emph{across} conditions I report per-seed and leave-one-seed-out sensitivity alongside) span three axes decorrelated by construction: \emph{conflict} (contradictory conventions on shared prompts, private budgets fixed), \emph{compatible overlap} (the same shared prompts under the same convention --- overlap and volume without conflict), and \emph{duration} (weight divergence with zero conflict). Before merging, six predictors are computed: \emph{confidence-weighted functional conflict} (bilateral confident disagreement on probes drawn blind to where conflict lives --- a proposed proxy for merge-relevant interactions, motivated by the observation that raw disagreement counts harmless complementation, one parent merely ignorant, as conflict), raw disagreement, gradient alignment at the shared base (47), LoRA-delta cosine and distance, and a cross-task performance baseline. The pre-registered outcome is the merge penalty against oracle parent potential (the analogue of \emph{hybrid load}, the fitness a hybrid loses relative to what its parents' genes could jointly supply), also reported against best- and mean-parent references because the predictor ordering is sensitive to that choice. Across this controlled grid, pre-merge functional disagreement predicted merge penalties (clustered bootstrap CIs excluding zero; held-out leave-one-condition-out \(\rho\) \(\approx\) 0.35--0.40), whereas LoRA-delta cosine and L2 showed no statistically detectable association; gradient alignment carried intermediate signal. Head-to-head predictor differences are not individually significant at this sample size; only these baselines were tested; and with three seeds, uncertainty about seed generalisation remains substantial --- though the seed sensitivity favours the functional measures (per-seed \(\rho\) stable at +0.37 to +0.53 in each seed alone, geometry \(\approx\) 0 in every seed, gradient alignment seed-unstable at −0.11 to −0.55). Two further results bound the claim: the initial two-axis grid's best predictor was delta-cosine (\(\rho\) = +0.60) --- an overlap artefact that the compatible-overlap control was added to expose, and did (collapse to +0.03); and the pre-registered internal prediction that confidence weighting would beat raw disagreement \emph{failed} (they are statistically indistinguishable as rank predictors), so the present evidence favours functional disagreement generally, not the DMI-specific refinement. The framework motivated the measurement and the controls; their success does not validate the specifically population-genetic mechanism. Whether the prediction improves a budget-matched operator choice, and whether it generalises to unfamiliar conflict structures and real task pairs, are the experiment's open front. @@ -123,15 +123,15 @@ Predictive test & 13 conditions \(\times\) 3 seeds (0.5B) & Merge penalty vs ora \textbf{Design rules.} As engineering guidance, the results reduce to rules that an operator of a model population can apply, answering the four decisions posed in the Introduction. \emph{Ground every generation} in verified reality --- a few percent retained most diversity in the tested settings --- but price the rarest capabilities individually (observation probability \texttt{1 − e\textasciicircum{}{−m\(\cdot\)p}} per batch under unstratified sampling), consider targeted sampling for the deep tail, and use recombination to recover rare capabilities still retained across complementary parents. \emph{Merge, don't blend, when there is headroom}: keep specialists intact and route, or breed-and-screen candidate merges, whenever the naive average is far from ceiling; plain averaging is adequate only where a strong base has already composed the skills. \emph{Match the operator to entanglement}: merge freely when skills are additive; sparingly, with offspring selection, when they entangle; and expect the champion-optimal mating breadth to narrow as landscapes roughen. \emph{Preserve diversity as a first-class objective}, because selection can only preserve variety that exists, and in the tested society its removal produced a distinct failure mode. \emph{Before merging, measure functional conflict} --- cheap, pre-merge, and in the controlled setting predictive where the tested weight-distance baselines were not; and \emph{do not treat divergence or specialisation alone as evidence of incompatibility} --- in every regime tested here, what broke merging was conflicting conventions on shared circuitry, which is the thing to detect. -\textbf{Continual learning at the population scale.} Within a single network, the discipline's remedies for forgetting are this framework's operators writ small. Rehearsal and replay of stored data (26, 27) is grounded inheritance within one lineage, and the replay fractions the field settled on empirically, on the order of 1\% for instruction tuning (45) and 5\% to 25\% by distribution-shift strength in continual pretraining (46), sit where the minimal model's operational threshold lies. \emph{Pseudo-rehearsal}, the replay of a network's own generated samples, proposed as a cure in 1995 (47) and revived as generative replay (48), is precisely the ungrounded null studied here: immigration from a drifting source, benign for one hop, compounding over generations, with verifier-filtering (29, 49) converting it back into grounding. Parameter isolation (50), including frozen-base adapters, which forget far less (51), is engineered decorrelation; complementary-learning-systems consolidation (52--54) is the periodic adapter-into-base merge; the recent turn to merging as a continual-learning mechanism (55--58) applies recombination within one lineage over time, where this paper applies it across lineages; and the observation that rare examples and long-tail knowledge are forgotten first (59--61) is tail extinction seen one model at a time. The mechanisms differ (forgetting is largely deterministic interference, collapse is sampling drift) but the victims and the remedies coincide, and to my knowledge no prior work carries population-genetic formalism into continual learning. Read into that field, the results offer: (i) an equilibrium theory for the replay ratio, with the sharper prediction that the required fraction is set by the rarest capability one refuses to lose (the \texttt{1 − e\textasciicircum{}{−m\(\cdot\)p}} law) rather than by average loss, testable against published replay sweeps; (ii) a \emph{failure theory for generative replay}: self-generated rehearsal is safe for short horizons and compounds into collapse across generations unless verifier-filtered back into grounding (29, 47--49); (iii) \emph{pre-merge interference prediction with a mechanism}: where the current state of the art fits regressions over candidate metrics (44), the functional-conflict measure arrives at a convergent signal from principle and comes with an operator prescription --- when conflict is high, do not average; route or breed-and-screen; (iv) a candidate \emph{decision rule for the consolidate-versus-stay-modular question} that currently splits the field's practice (keep adapters separate vs merge them; 54--58): union-preserving operators where headroom exists, fusion where the base composes, consolidation as the slow-store step; and (v) \emph{tail monitoring as the leading indicator}: continual-learning evaluation that averages over capabilities hides exactly the losses that drift theory says come first and, past a threshold, become irreversible. On that last point I note the standing objection that apparent forgetting can be skewed task-inference over latent capability rather than erasure (62); the irreversibility results here concern oracle-measured behavioural distributions, and distinguishing latent from extinct capability at language-model scale is an open experiment whose outcome would be decisive for both readings. +\textbf{Continual learning at the population scale.} Within a single network, the discipline's remedies for forgetting are this framework's operators writ small. Rehearsal and replay of stored data (26, 27) is grounded inheritance within one lineage, and the replay fractions the field settled on empirically, on the order of 1\% for instruction tuning (48) and 5\% to 25\% by distribution-shift strength in continual pretraining (49), sit where the minimal model's operational threshold lies. \emph{Pseudo-rehearsal}, the replay of a network's own generated samples, proposed as a cure in 1995 (50) and revived as generative replay (51), is precisely the ungrounded null studied here: immigration from a drifting source, benign for one hop, compounding over generations, with verifier-filtering (29, 52) converting it back into grounding. Parameter isolation (53), including frozen-base adapters, which forget far less (54), is engineered decorrelation; complementary-learning-systems consolidation (55--57) is the periodic adapter-into-base merge; the recent turn to merging as a continual-learning mechanism (58--61) applies recombination within one lineage over time, where this paper applies it across lineages; and the observation that rare examples and long-tail knowledge are forgotten first (62--64) is tail extinction seen one model at a time. The mechanisms differ (forgetting is largely deterministic interference, collapse is sampling drift) but the victims and the remedies coincide, and to my knowledge no prior work carries population-genetic formalism into continual learning. Read into that field, the results offer: (i) an equilibrium theory for the replay ratio, with the sharper prediction that the required fraction is set by the rarest capability one refuses to lose (the \texttt{1 − e\textasciicircum{}{−m\(\cdot\)p}} law) rather than by average loss, testable against published replay sweeps; (ii) a \emph{failure theory for generative replay}: self-generated rehearsal is safe for short horizons and compounds into collapse across generations unless verifier-filtered back into grounding (29, 50--52); (iii) \emph{pre-merge interference prediction with a mechanism}: where the current state of the art fits regressions over candidate metrics (47), the functional-conflict measure arrives at a convergent signal from principle and comes with an operator prescription --- when conflict is high, do not average; route or breed-and-screen; (iv) a candidate \emph{decision rule for the consolidate-versus-stay-modular question} that currently splits the field's practice (keep adapters separate vs merge them; 57--61): union-preserving operators where headroom exists, fusion where the base composes, consolidation as the slow-store step; and (v) \emph{tail monitoring as the leading indicator}: continual-learning evaluation that averages over capabilities hides exactly the losses that drift theory says come first and, past a threshold, become irreversible. On that last point I note the standing objection that apparent forgetting can be skewed task-inference over latent capability rather than erasure (65); the irreversibility results here concern oracle-measured behavioural distributions, and distinguishing latent from extinct capability at language-model scale is an open experiment whose outcome would be decisive for both readings. -\textbf{What is borrowed and what is new.} The collapse-as-drift diagnosis is established prior work (21--25); so are the empirical facts that merges can beat parents, that decorrelated parents merge better, and that naive averaging loses to interference-aware or routed merges (4, 63, 64), that model populations can climb (5, 8--10), and that merge success admits ML-native predictors (44, 65), correlational where this framework supplies mechanism; the reading of sex as an algorithm for mixability in the theory of computation (66) anticipated the transfer before model merging existed. New here is the framework-level synthesis --- inheritance, diversity, and compatibility as managed quantities --- together with: the conservation law for blending inheritance and its operator boundaries; the per-item grounding floor; the society ablation with its complementary failure modes; model speciation as a named, tested question, with the coordinate-versus-functional decomposition under permutation-and-rescaling alignment and the emergent null that bounds it; and the controlled predictive test with its controls. I claim the framework generated these measurements and experiments; I do not claim that their outcomes validate a uniquely population-genetic mechanism, and one refinement it proposed was not supported. +\textbf{What is borrowed and what is new.} The collapse-as-drift diagnosis is established prior work (21--25); so are the empirical facts that merges can beat parents, that decorrelated parents merge better, and that naive averaging loses to interference-aware or routed merges (4, 66, 67), that model populations can climb (5, 8--10), and that merge success admits ML-native predictors (47, 68), correlational where this framework supplies mechanism; the reading of sex as an algorithm for mixability in the theory of computation (69) anticipated the transfer before model merging existed. New here is the framework-level synthesis --- inheritance, diversity, and compatibility as managed quantities --- together with: the conservation law for blending inheritance and its operator boundaries; the per-item grounding floor; the society ablation with its complementary failure modes; model speciation as a named, tested question, with the coordinate-versus-functional decomposition under permutation-and-rescaling alignment and the emergent null that bounds it; and the controlled predictive test with its controls. I claim the framework generated these measurements and experiments; I do not claim that their outcomes validate a uniquely population-genetic mechanism, and one refinement it proposed was not supported. \textbf{Limits and open problems.} The demonstrations are deliberately small: exact where small is a virtue, sign-level and seed-replicated at the language-model tier, on constructed task families with a trivially separable router and one model lineage (Qwen, 0.5B--7B). The composed society has not been built at language-model scale. The predictive test's next bars, in order of value: generalisation to \emph{unfamiliar} conflict structures and real task pairs; a demonstrably better \emph{budget-matched} merging decision; then scale replication. Beyond engineering, the framework's hardest open problem is the fitness function itself: selection optimises what is measured, and for knowledge systems the persuasive and the true compete --- grounding against a reality that can refuse is the only anchor I trust, and institutionalising that anchor (verification, replication, and challenge among models) is the society-level problem this paper poses but does not solve. What biology receives in return is a new model system: populations of learners where every genotype, environment, and mating decision is observable and manipulable --- where the evolution of sex can be studied with interventions (unbounded parents, offspring preview, directed mating) that no living system permits. -\textbf{Creative diversity.} Collapse is not confined to facts and skills. Homogenisation of \emph{style} is already measurable: models trained on model output lose lexical and syntactic diversity across generations (67), writing produced with model assistance is individually better but collectively less diverse than writing produced without it (68, 69), and the house styles of the large assistants are recognisable enough that their tics serve as signatures. In this framework these are the same phenomenon at a different locus. A voice is a distribution over rare stylistic variants, exactly the tail that drift erases first and that blending inheritance averages into a common register. The remedies transfer unchanged, though they are untested here: grounding on stylistically diverse human sources, decorrelated lineages maintained as distinct voices rather than merged into one, union-preserving recombination over blending, and selection that rewards being different as well as being good. Whether these preserve measured stylistic diversity at scale is an open experiment that the framework specifies. +\textbf{Creative diversity.} Collapse is not confined to facts and skills. Homogenisation of \emph{style} is already measurable: models trained on model output lose lexical and syntactic diversity across generations (70), writing produced with model assistance is individually better but collectively less diverse than writing produced without it (71, 72), and the house styles of the large assistants are recognisable enough that their tics serve as signatures. In this framework these are the same phenomenon at a different locus. A voice is a distribution over rare stylistic variants, exactly the tail that drift erases first and that blending inheritance averages into a common register. The remedies transfer unchanged, though they are untested here: grounding on stylistically diverse human sources, decorrelated lineages maintained as distinct voices rather than merged into one, union-preserving recombination over blending, and selection that rewards being different as well as being good. Whether these preserve measured stylistic diversity at scale is an open experiment that the framework specifies. -\textbf{Outlook: the evolution of language models.} The Introduction's premise, that the model ecosystem is an evolving population, is also a forecast about where these results matter next. Language-model development is consolidating around exactly the operators studied here: synthetic-data flywheels (inheritance), merging and routing of specialist fine-tunes (recombination and population structure), verifier-gated data pipelines (grounded selection), and periodic consolidation of adapters into new bases. If coming model generations remain what the tested regimes found, freely recombinable in the absence of conflicting conventions, then the ecosystem evolves as one interbreeding population, and the levers that matter are grounding budgets priced per rare capability and diversity preserved deliberately. If instead long-horizon specialisation at scale begins to produce emergent incompatibility, as the expert-training-duration observations hint (43) and the small-scale null here does not rule out, then lineages will begin to speciate, and the ecosystem's future is a set of diverging species connected by routing rather than by merging. Which of the two it will be is measurable now, with the pre-merge conflict instruments this paper tested. +\textbf{Outlook: the evolution of language models.} The Introduction's premise, that the model ecosystem is an evolving population, is also a forecast about where these results matter next. Language-model development is consolidating around exactly the operators studied here: synthetic-data flywheels (inheritance), merging and routing of specialist fine-tunes (recombination and population structure), verifier-gated data pipelines (grounded selection), and periodic consolidation of adapters into new bases. If coming model generations remain what the tested regimes found, freely recombinable in the absence of conflicting conventions, then the ecosystem evolves as one interbreeding population, and the levers that matter are grounding budgets priced per rare capability and diversity preserved deliberately. If instead long-horizon specialisation at scale begins to produce emergent incompatibility, as the expert-training-duration observations hint (46) and the small-scale null here does not rule out, then lineages will begin to speciate, and the ecosystem's future is a set of diverging species connected by routing rather than by merging. Which of the two it will be is measurable now, with the pre-merge conflict instruments this paper tested. \section*{Materials and Methods} @@ -177,11 +177,14 @@ Predictive test & 13 conditions \(\times\) 3 seeds (0.5B) & Merge penalty vs ora \item B. Yi, Q. Liu, Y. Cheng, H. Xu, Escaping model collapse via synthetic data verification. arXiv [Preprint] (2025). https://doi.org/10.48550/arXiv.2510.16657. \item M. Gerstgrasser, et al., Is model collapse inevitable? Breaking the curse of recursion by accumulating real and synthetic data. \emph{Conf. Lang. Model.} (2024). https://doi.org/10.48550/arXiv.2404.01413. \item S. Wright, Evolution in Mendelian populations. \emph{Genetics} \textbf{16}, 97--159 (1931). +\item L. S. Mills, F. W. Allendorf, The one-migrant-per-generation rule in conservation and management. \emph{Conserv. Biol.} \textbf{10}, 1509--1518 (1996). +\item F. Jenkin, The origin of species [review]. \emph{North Br. Rev.} \textbf{46}, 277--318 (1867). \item J. Pari, S. Jelassi, P. Agrawal, Collective model intelligence requires compatible specialization. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2411.02207. \item R. A. Fisher, \emph{The Genetical Theory of Natural Selection} (Clarendon Press, 1930). \item H. J. Muller, Some genetic aspects of sex. \emph{Am. Nat.} \textbf{66}, 118--138 (1932). \item E. J. Hu, et al., LoRA: Low-rank adaptation of large language models. \emph{Int. Conf. Learn. Represent.} (2022). https://doi.org/10.48550/arXiv.2106.09685. \item S. A. Kauffman, S. Levin, Towards a general theory of adaptive walks on rugged landscapes. \emph{J. Theor. Biol.} \textbf{128}, 11--45 (1987). +\item A. R. Templeton, ``Coadaptation and outbreeding depression'' in \emph{Conservation Biology: The Science of Scarcity and Diversity}, M. E. Soulé, Ed. (Sinauer, 1986), pp. 105--116. \item J. Lehman, K. O. Stanley, Abandoning objectives: Evolution through the search for novelty alone. \emph{Evol. Comput.} \textbf{19}, 189--223 (2011). \item H. A. Orr, The population genetics of speciation: The evolution of hybrid incompatibilities. \emph{Genetics} \textbf{139}, 1805--1813 (1995). \item H. A. Orr, M. Turelli, The evolution of postzygotic isolation: Accumulating Dobzhansky--Muller incompatibilities. \emph{Evolution} \textbf{55}, 1085--1094 (2001). diff --git a/paper/pnas/main.md b/paper/pnas/main.md index f31e162..6c5cade 100644 --- a/paper/pnas/main.md +++ b/paper/pnas/main.md @@ -69,7 +69,9 @@ as systems whose inheritance, diversity, and compatibility can be measured, pred Training each generation of a model on the previous generation's output degrades it (*model collapse*): rare capabilities vanish first, and the lineage drifts toward its own most common behaviour (21). That degradation is, mathematically, *genetic drift*, the loss of rare -variants that any finite population suffers when each generation is a finite sample of the last. The +variants that any finite population suffers when each generation is a finite sample of the last — the +same sampling accident by which rare surnames vanish from small villages and rare alleles (gene +variants) drift out of island populations with no selection against them. The identification has been made repeatedly and independently: for sequential inference chains before deep learning (22), for language-model text ecosystems (23), as a closed-form first-extinction law placing collapse onset at the Wright–Fisher first-extinction time (24), and in quantitative-genetic form for @@ -78,8 +80,9 @@ points, marks population genetics as the natural mathematics of the setting, tho point: population genetics is not, at heart, a theory of decay; it is a theory of the mechanisms that maintain and build populations despite decay (immigration, recombination, selection, population structure) and of where those mechanisms reach their limits. This paper develops that fuller structure -for model populations: the arc from drift through its remedies to its limit, reproductive isolation, -carried as one framework from closed forms to trained networks to language models. +for model populations: the arc from drift through its remedies to its limit, reproductive isolation — +the point at which diverged lineages can no longer produce working offspring, biology's boundary +between species — carried as one framework from closed forms to trained networks to language models. An operator of a model population faces recurring decisions for which there is no principled guidance: how much verified real data does retraining need before a lineage decays; will combining two particular models compose @@ -98,10 +101,15 @@ Knowledge is modelled as a distribution `p_t` over `K` discrete items (capabilit behaviour), with a fixed true distribution `p*` whose rare tail carries the knowledge most at risk. One generation is: *draw `n` samples from the parent's distribution, optionally mix in `m` verified real samples ("grounding", `g = m/(n+m)`), and refit the child*. In this minimal inheritance model the -resampling step *is* the Wright–Fisher process: the same equations, which I exploit as an -engineering gate: the simulator reproduces the classical closed forms (heterozygosity decay -`E[H_t] = H_0(1 − 1/n)^t`; the exact immigration–drift equilibrium; the closed-form multi-teacher -union) to within 0.5%, and these are standing tests in the codebase, not one-off checks. +resampling step *is* the Wright–Fisher process, population genetics' canonical model of neutral +evolution, in which each new generation is a random sample of size `n` from the previous one and +every statistical property of drift follows from that one step. Diversity throughout this paper is +*heterozygosity*, `H = 1 − Σ p_i²`: the probability that two random draws differ (one minus a +collision probability), high when many items share the mass, zero at total collapse. The identity is +exploited as an engineering gate: the simulator reproduces the classical closed forms (heterozygosity +decay `E[H_t] = H_0(1 − 1/n)^t`; the exact immigration–drift equilibrium; the closed-form +multi-teacher union) to within 0.5%, and these are standing tests in the codebase, not one-off +checks. The boundary of the exactness matters, and I measured it rather than assumed it. Real training adds approximation, optimisation noise, and inductive bias, and when trained networks are fit against the @@ -113,11 +121,15 @@ estimator bias*, and the drift signs (rare-first loss; the grounding response) survived that bias in every architecture I tested, including a convolutional VAE retrained on its own generated digits, where the dry lineage collapses to a single blurred digit class while 10% grounding holds all thirty modes (Fig. 1). Retraining on a single parent is *asexual reproduction*, and sustained loss under it carries the defining consequence -of *Muller's ratchet* (28): once every copy of a rare capability is gone from all parents and sources, -no recombination can rebuild it, so remedies must act before fixation-by-loss (a consequence-level -correspondence; the minimal model lacks the ratchet's recurrent-mutation driver). +of *Muller's ratchet* (28), the mechanism by which lineages that never recombine decay irreversibly — +the reason non-recombining genomes such as the Y chromosome have shed most of their ancestral genes. +Once every copy of a rare capability is gone from all parents and sources, no recombination can +rebuild it: each such loss is a click of the ratchet, and remedies must act while copies still +survive somewhere (a consequence-level correspondence; the minimal model lacks the ratchet's +recurrent-mutation driver). -**Table 1.** The dictionary. Each correspondence is stated with the level of support it currently has +**Table 1.** The dictionary. Each biological term is introduced in the section that develops it; each +correspondence is stated with the level of support it currently has (exact = closed form in the minimal model; empirical = measured in trained systems; hypothesis = stated with a falsifier, untested or unconfirmed). The full claim-by-claim ledger with assumptions and known limits is SI Appendix, Table S1. @@ -139,7 +151,10 @@ known limits is SI Appendix, Table S1. ### Grounding is immigration: cheap, with a floor In the minimal model, grounding from a fixed real source is *immigration* into a drifting population -(29–31), and the equilibrium diversity has a closed form the simulator matches exactly. That equilibrium is +(29–31). Immigration is what conservation managers prescribe when a fragmented reserve loses +diversity, and its striking property there is how little is needed — the field's rule of thumb is +that one migrant per generation holds an isolated population's diversity (32). The same economy +appears here: the equilibrium diversity has a closed form the simulator matches exactly. That equilibrium is *smooth* in the grounding fraction (there is no phase transition in aggregate diversity), so the practical number is an operational threshold, and I define it as such: under the tested population size and Zipf source distribution, `g ≈ 0.05` retained most (≥95%) of equilibrium diversity @@ -164,7 +179,14 @@ fraction, the measured price of the estimator bias). ### Recombination: a conservation law, its operators, and offspring that exceed every parent -The largest returns from the transfer concern merging. **Proposition (blending inheritance, rare-item +The largest returns from the transfer concern merging. *Blending inheritance* — offspring as the +average of their parents — is the failure mode at the root of population genetics' founding +controversy: the swamping argument pressed in Jenkin's 1867 review of *The Origin of Species*, that +under blending a rare advantageous variant is diluted toward the common type faster than selection +can multiply it (33), an objection dissolved only by Mendel's particulate inheritance, in which +discrete variants pass through generations undiluted. Refitting a child model to the mean of its +parents' output distributions is blending inheritance, and the proposition below is Jenkin's dilution +made exact. **Proposition (blending inheritance, rare-item regime).** Let K parents independently retain a rare item (mass `p` when retained), and let the child draw `n` samples either from one parent chosen at random or from the *mean of the parents' output distributions*. Expected item mass is identical under the two schemes; and in the rare-item regime @@ -178,7 +200,7 @@ contrasting union operator (keep each item's strongest source, then renormalise, redistributes mass and presupposes a verifier or oracle to identify the strongest source) increases expected retention with K in all regimes in the minimal model. The practically important operators, *weight averaging* (a nonlinear network's weight-mean does not compute its parents' -output-mean) and *routing among intact specialists* (32) (different storage and inference budgets from a +output-mean) and *routing among intact specialists* (34) (different storage and inference budgets from a single child), are its empirical cousins, and the measured bridge is a *headroom rule*, stated qualitatively: in language models, union-preserving operators beat the weight-average where that average falls short of attainable performance, and add nothing where it does not (easy-versus-hard contrasts at two scales; a @@ -186,25 +208,36 @@ quantitative form of the relationship is untested). On easy tasks a capable base add nothing; on hard tasks the average dilutes a fragile specialist below even the best single parent and routing wins by a wide margin (Fig. 6A–B). -The generative payoff is the *Fisher–Muller effect* (33, 34): recombination assembles, in one offspring, -complementary variants that arose in different lineages, producing a genotype fitter than any parent. +The generative payoff is the *Fisher–Muller effect* (35, 36), the classical account of why sex speeds +adaptation: in an asexual population, beneficial variants arising in different individuals can only +compete until all but one lineage is lost, whereas recombination assembles them in one offspring, +producing a *genotype* (an individual's combination of variants, one at each *locus*, or position) +fitter than any parent. In the multi-locus model, sexual merging of decorrelated specialists climbs to the global optimum, a genotype no parent held, while the best single parent and the blended average both plateau below (Fig. 2). In real language models the signature replicates under seed replication: merges of three -LoRA (35) specialists beat every parent overall (decisively at 7B: 0.87 vs 0.77), and on the sharper +LoRA (37) specialists beat every parent overall (decisively at 7B: 0.87 vs 0.77), and on the sharper worst-family metric the merged models are the only ones competent everywhere, in every seed (Fig. 6A). -Sex has risks and, for AI, an unfair advantage, both quantified on rugged (epistatic) NK landscapes (36) -(Fig. 3). When skills are entangled, blind recombination produces offspring *below* their parents -(outbreeding depression), worsening with ruggedness, and the optimal recombination rate shrinks as -entanglement grows. But an engineered population can do what biology cannot: recombine unbounded +Sex has risks and, for AI, an unfair advantage, both quantified on Kauffman's NK fitness landscapes +(38), the standard model of *epistasis*, biology's term for interaction between genes: the fitness +contribution of a variant depends on which variants occupy the other loci, much as a component's +value in an ML system depends on the components around it. Each of the landscape's `N` sites +interacts with `K` others (the model's eponymous parameters), and raising that interaction count +tunes the landscape from smooth and additive to rugged and many-peaked (Fig. 3). When skills are entangled, blind recombination produces offspring *below* their parents, +worsening with ruggedness, and the optimal recombination rate shrinks as entanglement grows. Biology +knows this failure as *outbreeding depression*, the reason conservation practice warns against +crossing locally adapted populations: in the textbook case, an ibex herd in the Tatra Mountains +restocked with animals from Turkey and Sinai produced fertile hybrids that bore their young in the +coldest month of winter, and the herd died out (39). But an engineered population can do what biology cannot: recombine unbounded parents, choose complementary mates, and *screen many candidate offspring against a verifier before keeping one*. This directed sex converts the outbreeding catastrophe into a reliable gain in the model (tracking or exceeding the best parent at every ruggedness) and replicates as a sign in language models: bred-and-screened merges beat the a-priori blend in every seed on headroom tasks, including one seed where the blend failed catastrophically and selection was immune (Fig. 6A). Finally, -population *structure* is itself a knob: sweeping the mate-pool breadth from monogamous (local) to -promiscuous (panmictic) against ruggedness, wide mixing maximises the population mean while +population *structure* is itself a knob: sweeping the mate-pool breadth from monogamous (repeated +local pairings) to promiscuous (*panmixia*: any model may merge with any other) against ruggedness, +wide mixing maximises the population mean while monotonically destroying diversity, and the best *champion* shifts from wide breadth on smooth landscapes to intermediate breadth on rugged ones (Fig. 3C), the mating-system phenomenon known to structured-population search, mapped onto merging populations. @@ -223,7 +256,7 @@ the analogue of scoring models by the crowd's approval (the fitness channel). Th ideas, since both couple the lineage to a non-drifting external signal, but they are different operators, and I name them separately. In the tested society (a finite agent population on a rugged NK landscape), a four-arm ablation separates the failure modes: the full system (grounded evaluation + -directed recombination + diversity-preserving selection (37)) climbs to near the global optimum while +directed recombination + diversity-preserving selection (40)) climbs to near the global optimum while keeping its specialists; removing grounded evaluation converges the population confidently on an unfit consensus (self-consumption); removing recombination strands it on local optima; removing diversity converges it prematurely to a worse answer. Each removal fails differently; the three implementations @@ -236,18 +269,20 @@ language-model scale this composed loop remains unbuilt; it is the paper's large ### The limit of sex: model speciation Recombination presupposes compatible parents. In biology, lineages pushed far enough apart become -separate species (*reproductive isolation*) through Bateson–Dobzhansky–Muller incompatibilities (38, 39): -changes harmless on their own background but deleterious in combination. A merged model is exactly the +separate species (*reproductive isolation*) through Bateson–Dobzhansky–Muller incompatibilities (41, 42): +changes harmless on their own genetic background but deleterious in combination — the mechanism behind +the mule's sterility and the inviability of many between-species crosses, in which two genomes that +each work perfectly cannot run in the same cell. A merged model is exactly the exposed hybrid. I built the analytic model (Fig. 5A): hybrid fitness tracks the parents while compatible, then peels off and crashes below the ancestor; the isolation cliff arrives earlier the denser the incompatibilities; and the incompatibility *count* snowballs quadratically with divergence -(39). Note that a super-linear count does not by itself entail a sharp performance cliff without +(42). Note that a super-linear count does not by itself entail a sharp performance cliff without the count-to-effect-size link, which the analytic model supplies under its assumptions and any neural test must establish separately. In trained networks, the claim must survive a known alternative: merge barriers between independently -trained networks are famously *coordinate artefacts*, removable by re-aligning hidden units (40); -richer symmetry groups remove more (41), with known failures beyond the shared-data regime (42). I therefore aligned under the composition of +trained networks are famously *coordinate artefacts*, removable by re-aligning hidden units (43); +richer symmetry groups remove more (44), with known failures beyond the shared-data regime (45). I therefore aligned under the composition of permutation matching and exact per-unit rescaling (the unit symmetry group of plain ReLU MLPs, as the search space) and decomposed the barrier (Fig. 5 C and D): two networks trained from different initialisations on the *same* task have a barrier that this alignment removes essentially entirely @@ -270,7 +305,7 @@ catastrophically-forgetting specialists (parents ≈ 0.50, merge ≈ 0.955, a su rescue). The same double result appears at the language-model tier (Fig. 5 E and F): conflicting conventions produce *function-specific* hybrid breakdown (the merge scores below both parents on the conflicted function, while a budget-controlled design shows the disjoint skills merge unharmed), and over-training -disjoint specialists 1→12 epochs (cf. the merging literature's expert-duration effect; 43) produces +disjoint specialists 1→12 epochs (cf. the merging literature's expert-duration effect; 46) produces no isolation at all — the merge improves. Across every tier tested, isolation had to be provoked by functional conflict; specialisation alone did not speciate — a bound on the analogy that sharpens the design rule: what breaks merging is conflicting conventions @@ -290,9 +325,10 @@ divergence with zero conflict). Before merging, six predictors are computed: *co functional conflict* (bilateral confident disagreement on probes drawn blind to where conflict lives — a proposed proxy for merge-relevant interactions, motivated by the observation that raw disagreement counts harmless complementation, one parent merely ignorant, as conflict), raw disagreement, gradient -alignment at the shared base (44), LoRA-delta cosine and distance, and a cross-task performance -baseline. The pre-registered outcome is the merge penalty against oracle parent potential (the -hybrid-load analogue), also reported against best- and mean-parent references because the predictor +alignment at the shared base (47), LoRA-delta cosine and distance, and a cross-task performance +baseline. The pre-registered outcome is the merge penalty against oracle parent potential (the analogue of +*hybrid load*, the fitness a hybrid loses relative to what its parents' genes could jointly supply), +also reported against best- and mean-parent references because the predictor ordering is sensitive to that choice. Across this controlled grid, pre-merge functional @@ -350,16 +386,16 @@ conventions on shared circuitry, which is the thing to detect. **Continual learning at the population scale.** Within a single network, the discipline's remedies for forgetting are this framework's operators writ small. Rehearsal and replay of stored data (26, 27) is grounded inheritance within one lineage, and the replay fractions the field settled on empirically, -on the order of 1% for instruction tuning (45) and 5% to 25% by distribution-shift strength in -continual pretraining (46), sit where the minimal model's operational threshold lies. -*Pseudo-rehearsal*, the replay of a network's own generated samples, proposed as a cure in 1995 (47) -and revived as generative replay (48), is precisely the ungrounded null studied here: immigration from -a drifting source, benign for one hop, compounding over generations, with verifier-filtering (29, 49) -converting it back into grounding. Parameter isolation (50), including frozen-base adapters, which forget far -less (51), is engineered decorrelation; complementary-learning-systems consolidation (52–54) is the periodic -adapter-into-base merge; the recent turn to merging as a continual-learning mechanism (55–58) applies +on the order of 1% for instruction tuning (48) and 5% to 25% by distribution-shift strength in +continual pretraining (49), sit where the minimal model's operational threshold lies. +*Pseudo-rehearsal*, the replay of a network's own generated samples, proposed as a cure in 1995 (50) +and revived as generative replay (51), is precisely the ungrounded null studied here: immigration from +a drifting source, benign for one hop, compounding over generations, with verifier-filtering (29, 52) +converting it back into grounding. Parameter isolation (53), including frozen-base adapters, which forget far +less (54), is engineered decorrelation; complementary-learning-systems consolidation (55–57) is the periodic +adapter-into-base merge; the recent turn to merging as a continual-learning mechanism (58–61) applies recombination within one lineage over time, where this paper applies it across lineages; and the -observation that rare examples and long-tail knowledge are forgotten first (59–61) is tail extinction +observation that rare examples and long-tail knowledge are forgotten first (62–64) is tail extinction seen one model at a time. The mechanisms differ (forgetting is largely deterministic interference, collapse is sampling drift) but the victims and the remedies coincide, and to my knowledge no prior work carries population-genetic formalism into continual learning. Read into that field, the results @@ -367,26 +403,26 @@ offer: (i) an equilibrium theory for the replay ratio, with the sharper predicti fraction is set by the rarest capability one refuses to lose (the `1 − e^{−m·p}` law) rather than by average loss, testable against published replay sweeps; (ii) a *failure theory for generative replay*: self-generated rehearsal is safe for short horizons and compounds into collapse across generations -unless verifier-filtered back into grounding (29, 47–49); (iii) *pre-merge interference +unless verifier-filtered back into grounding (29, 50–52); (iii) *pre-merge interference prediction with a mechanism*: where the current state of the art fits regressions over candidate -metrics (44), the functional-conflict measure arrives at a convergent signal from principle and comes +metrics (47), the functional-conflict measure arrives at a convergent signal from principle and comes with an operator prescription — when conflict is high, do not average; route or breed-and-screen; (iv) a candidate *decision rule for the consolidate-versus-stay-modular question* that currently -splits the field's practice (keep adapters separate vs merge them; 54–58): union-preserving operators +splits the field's practice (keep adapters separate vs merge them; 57–61): union-preserving operators where headroom exists, fusion where the base composes, consolidation as the slow-store step; and (v) *tail monitoring as the leading indicator*: continual-learning evaluation that averages over capabilities hides exactly the losses that drift theory says come first and, past a threshold, become irreversible. On that last point I note the standing objection that apparent forgetting can be -skewed task-inference over latent capability rather than erasure (62); the irreversibility results here +skewed task-inference over latent capability rather than erasure (65); the irreversibility results here concern oracle-measured behavioural distributions, and distinguishing latent from extinct capability at language-model scale is an open experiment whose outcome would be decisive for both readings. **What is borrowed and what is new.** The collapse-as-drift diagnosis is established prior work (21–25); so are the empirical facts that merges can beat parents, that decorrelated parents merge better, and that -naive averaging loses to interference-aware or routed merges (4, 63, 64), that model populations can -climb (5, 8–10), and that merge success admits ML-native predictors (44, 65), correlational where this framework +naive averaging loses to interference-aware or routed merges (4, 66, 67), that model populations can +climb (5, 8–10), and that merge success admits ML-native predictors (47, 68), correlational where this framework supplies mechanism; the reading of sex as an algorithm for mixability in the theory of computation -(66) anticipated the transfer before model merging existed. New here is the framework-level +(69) anticipated the transfer before model merging existed. New here is the framework-level synthesis — inheritance, diversity, and compatibility as managed quantities — together with: the conservation law for blending inheritance and its operator boundaries; the per-item grounding floor; the society ablation with its complementary failure modes; model speciation as a named, tested question, with the @@ -411,8 +447,8 @@ offspring preview, directed mating) that no living system permits. **Creative diversity.** Collapse is not confined to facts and skills. Homogenisation of *style* is already measurable: models trained on model output lose lexical and syntactic diversity across -generations (67), writing produced with model assistance is individually better but collectively less -diverse than writing produced without it (68, 69), and the house styles of the large assistants are +generations (70), writing produced with model assistance is individually better but collectively less +diverse than writing produced without it (71, 72), and the house styles of the large assistants are recognisable enough that their tics serve as signatures. In this framework these are the same phenomenon at a different locus. A voice is a distribution over rare stylistic variants, exactly the tail that drift erases first and that blending inheritance averages into a common register. The @@ -432,7 +468,7 @@ the tested regimes found, freely recombinable in the absence of conflicting conv ecosystem evolves as one interbreeding population, and the levers that matter are grounding budgets priced per rare capability and diversity preserved deliberately. If instead long-horizon specialisation at scale begins to produce emergent incompatibility, as the expert-training-duration -observations hint (43) and the small-scale null here does not rule out, then lineages will begin to +observations hint (46) and the small-scale null here does not rule out, then lineages will begin to speciate, and the ecosystem's future is a set of diverging species connected by routing rather than by merging. Which of the two it will be is measurable now, with the pre-merge conflict instruments this paper tested. @@ -504,41 +540,44 @@ publication; every figure in this paper regenerates from committed artifacts wit 29. B. Yi, Q. Liu, Y. Cheng, H. Xu, Escaping model collapse via synthetic data verification. arXiv [Preprint] (2025). https://doi.org/10.48550/arXiv.2510.16657. 30. M. Gerstgrasser, et al., Is model collapse inevitable? Breaking the curse of recursion by accumulating real and synthetic data. *Conf. Lang. Model.* (2024). https://doi.org/10.48550/arXiv.2404.01413. 31. S. Wright, Evolution in Mendelian populations. *Genetics* **16**, 97–159 (1931). -32. J. Pari, S. Jelassi, P. Agrawal, Collective model intelligence requires compatible specialization. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2411.02207. -33. R. A. Fisher, *The Genetical Theory of Natural Selection* (Clarendon Press, 1930). -34. H. J. Muller, Some genetic aspects of sex. *Am. Nat.* **66**, 118–138 (1932). -35. E. J. Hu, et al., LoRA: Low-rank adaptation of large language models. *Int. Conf. Learn. Represent.* (2022). https://doi.org/10.48550/arXiv.2106.09685. -36. S. A. Kauffman, S. Levin, Towards a general theory of adaptive walks on rugged landscapes. *J. Theor. Biol.* **128**, 11–45 (1987). -37. J. Lehman, K. O. Stanley, Abandoning objectives: Evolution through the search for novelty alone. *Evol. Comput.* **19**, 189–223 (2011). -38. H. A. Orr, The population genetics of speciation: The evolution of hybrid incompatibilities. *Genetics* **139**, 1805–1813 (1995). -39. H. A. Orr, M. Turelli, The evolution of postzygotic isolation: Accumulating Dobzhansky–Muller incompatibilities. *Evolution* **55**, 1085–1094 (2001). -40. S. K. Ainsworth, J. Hayase, S. Srinivasa, Git Re-Basin: Merging models modulo permutation symmetries. *Int. Conf. Learn. Represent.* (2023). https://doi.org/10.48550/arXiv.2209.04836. -41. T. Li, Z. Shen, Scaling linear mode connectivity and merging to billion-parameter pretrained transformers. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2606.23607. -42. E. Sharma, D. M. Roy, G. K. Dziugaite, The non-local model merging problem: Permutation symmetries and variance collapse. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2410.12766. -43. N. Kozodoi, Z. Afolabi, J. Butler, Are we merging the right models? Impact of expert training duration on model merging for LLMs. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2607.11997. -44. L. Zhou, B. Zhao, R. Yu, E. Rodolà, Demystifying mergeability: Interpretable properties to predict model merging success. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2601.22285. -45. T. Scialom, T. Chakrabarty, S. Muresan, Fine-tuned language models are continual learners. *Proc. Conf. Empir. Methods Nat. Lang. Process.* (2022). https://doi.org/10.48550/arXiv.2205.12393. -46. A. Ibrahim, et al., Simple and scalable strategies to continually pre-train large language models. *Trans. Mach. Learn. Res.* (2024). https://doi.org/10.48550/arXiv.2403.08763. -47. A. Robins, Catastrophic forgetting, rehearsal and pseudorehearsal. *Connect. Sci.* **7**, 123–146 (1995). -48. H. Shin, J. K. Lee, J. Kim, J. Kim, Continual learning with deep generative replay. *Adv. Neural Inf. Process. Syst.* **30** (2017). https://doi.org/10.48550/arXiv.1705.08690. -49. Y. Feng, et al., Beyond model collapse: Scaling up with synthesized data requires verification. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2406.07515. -50. A. A. Rusu, et al., Progressive neural networks. arXiv [Preprint] (2016). https://doi.org/10.48550/arXiv.1606.04671. -51. D. Biderman, et al., LoRA learns less and forgets less. *Trans. Mach. Learn. Res.* (2024). https://doi.org/10.48550/arXiv.2405.09673. -52. J. L. McClelland, B. L. McNaughton, R. C. O'Reilly, Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory. *Psychol. Rev.* **102**, 419–457 (1995). -53. D. Kumaran, D. Hassabis, J. L. McClelland, What learning systems do intelligent agents need? Complementary learning systems theory updated. *Trends Cogn. Sci.* **20**, 512–534 (2016). -54. J. Schwarz, et al., Progress & Compress: A scalable framework for continual learning. *Proc. Int. Conf. Mach. Learn.* (2018). -55. G. Ilharco, et al., Editing models with task arithmetic. *Int. Conf. Learn. Represent.* (2023). https://doi.org/10.48550/arXiv.2212.04089. -56. D. Marczak, B. Twardowski, T. Trzciński, S. Cygert, MagMax: Leveraging model merging for seamless continual learning. *Proc. Eur. Conf. Comput. Vis.* (2024). https://doi.org/10.48550/arXiv.2407.06322. -57. A. Alexandrov, et al., Mitigating catastrophic forgetting in language transfer via model merging. *Findings Assoc. Comput. Linguist.: EMNLP* (2024). https://doi.org/10.48550/arXiv.2407.08699. -58. S. Dziadzio, et al., How to merge your multimodal models over time? *Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit.* (2025). https://doi.org/10.48550/arXiv.2412.06712. -59. M. Toneva, et al., An empirical study of example forgetting during deep neural network learning. *Int. Conf. Learn. Represent.* (2019). https://doi.org/10.48550/arXiv.1812.05159. -60. N. Kandpal, H. Deng, A. Roberts, E. Wallace, C. Raffel, Large language models struggle to learn long-tail knowledge. *Proc. Int. Conf. Mach. Learn.* (2023). https://doi.org/10.48550/arXiv.2211.08411. -61. X. Liu, et al., Long-tailed class incremental learning. *Proc. Eur. Conf. Comput. Vis.* (2022). https://doi.org/10.48550/arXiv.2210.00266. -62. S. Kotha, J. M. Springer, A. Raghunathan, Understanding catastrophic forgetting in language models via implicit inference. *Int. Conf. Learn. Represent.* (2024). https://doi.org/10.48550/arXiv.2309.10105. -63. L. Yu, B. Yu, H. Yu, F. Huang, Y. Li, Language models are super Mario: Absorbing abilities from homologous models as a free lunch. *Proc. Int. Conf. Mach. Learn.* (2024). https://doi.org/10.48550/arXiv.2311.03099. -64. M. Wortsman, et al., Model soups: Averaging weights of multiple fine-tuned models improves accuracy without increasing inference time. *Proc. Int. Conf. Mach. Learn.* (2022). https://doi.org/10.48550/arXiv.2203.05482. -65. Y. Cao, et al., An empirical study and theoretical explanation on task-level model-merging collapse. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2603.09463. -66. A. Livnat, C. Papadimitriou, Sex as an algorithm: The theory of evolution under the lens of computation. *Commun. ACM* **59**, 84–93 (2016). -67. Y. Guo, G. Shang, M. Vazirgiannis, C. Clavel, The curious decline of linguistic diversity: Training language models on synthetic text. *Findings Assoc. Comput. Linguist.: NAACL* (2024). https://doi.org/10.48550/arXiv.2311.09807. -68. V. Padmakumar, H. He, Does writing with language models reduce content diversity? *Int. Conf. Learn. Represent.* (2024). https://doi.org/10.48550/arXiv.2309.05196. -69. A. R. Doshi, O. P. Hauser, Generative AI enhances individual creativity but reduces the collective diversity of novel content. *Sci. Adv.* **10**, eadn5290 (2024). +32. L. S. Mills, F. W. Allendorf, The one-migrant-per-generation rule in conservation and management. *Conserv. Biol.* **10**, 1509–1518 (1996). +33. F. Jenkin, The origin of species [review]. *North Br. Rev.* **46**, 277–318 (1867). +34. J. Pari, S. Jelassi, P. Agrawal, Collective model intelligence requires compatible specialization. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2411.02207. +35. R. A. Fisher, *The Genetical Theory of Natural Selection* (Clarendon Press, 1930). +36. H. J. Muller, Some genetic aspects of sex. *Am. Nat.* **66**, 118–138 (1932). +37. E. J. Hu, et al., LoRA: Low-rank adaptation of large language models. *Int. Conf. Learn. Represent.* (2022). https://doi.org/10.48550/arXiv.2106.09685. +38. S. A. Kauffman, S. Levin, Towards a general theory of adaptive walks on rugged landscapes. *J. Theor. Biol.* **128**, 11–45 (1987). +39. A. R. Templeton, "Coadaptation and outbreeding depression" in *Conservation Biology: The Science of Scarcity and Diversity*, M. E. Soulé, Ed. (Sinauer, 1986), pp. 105–116. +40. J. Lehman, K. O. Stanley, Abandoning objectives: Evolution through the search for novelty alone. *Evol. Comput.* **19**, 189–223 (2011). +41. H. A. Orr, The population genetics of speciation: The evolution of hybrid incompatibilities. *Genetics* **139**, 1805–1813 (1995). +42. H. A. Orr, M. Turelli, The evolution of postzygotic isolation: Accumulating Dobzhansky–Muller incompatibilities. *Evolution* **55**, 1085–1094 (2001). +43. S. K. Ainsworth, J. Hayase, S. Srinivasa, Git Re-Basin: Merging models modulo permutation symmetries. *Int. Conf. Learn. Represent.* (2023). https://doi.org/10.48550/arXiv.2209.04836. +44. T. Li, Z. Shen, Scaling linear mode connectivity and merging to billion-parameter pretrained transformers. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2606.23607. +45. E. Sharma, D. M. Roy, G. K. Dziugaite, The non-local model merging problem: Permutation symmetries and variance collapse. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2410.12766. +46. N. Kozodoi, Z. Afolabi, J. Butler, Are we merging the right models? Impact of expert training duration on model merging for LLMs. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2607.11997. +47. L. Zhou, B. Zhao, R. Yu, E. Rodolà, Demystifying mergeability: Interpretable properties to predict model merging success. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2601.22285. +48. T. Scialom, T. Chakrabarty, S. Muresan, Fine-tuned language models are continual learners. *Proc. Conf. Empir. Methods Nat. Lang. Process.* (2022). https://doi.org/10.48550/arXiv.2205.12393. +49. A. Ibrahim, et al., Simple and scalable strategies to continually pre-train large language models. *Trans. Mach. Learn. Res.* (2024). https://doi.org/10.48550/arXiv.2403.08763. +50. A. Robins, Catastrophic forgetting, rehearsal and pseudorehearsal. *Connect. Sci.* **7**, 123–146 (1995). +51. H. Shin, J. K. Lee, J. Kim, J. Kim, Continual learning with deep generative replay. *Adv. Neural Inf. Process. Syst.* **30** (2017). https://doi.org/10.48550/arXiv.1705.08690. +52. Y. Feng, et al., Beyond model collapse: Scaling up with synthesized data requires verification. arXiv [Preprint] (2024). https://doi.org/10.48550/arXiv.2406.07515. +53. A. A. Rusu, et al., Progressive neural networks. arXiv [Preprint] (2016). https://doi.org/10.48550/arXiv.1606.04671. +54. D. Biderman, et al., LoRA learns less and forgets less. *Trans. Mach. Learn. Res.* (2024). https://doi.org/10.48550/arXiv.2405.09673. +55. J. L. McClelland, B. L. McNaughton, R. C. O'Reilly, Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory. *Psychol. Rev.* **102**, 419–457 (1995). +56. D. Kumaran, D. Hassabis, J. L. McClelland, What learning systems do intelligent agents need? Complementary learning systems theory updated. *Trends Cogn. Sci.* **20**, 512–534 (2016). +57. J. Schwarz, et al., Progress & Compress: A scalable framework for continual learning. *Proc. Int. Conf. Mach. Learn.* (2018). +58. G. Ilharco, et al., Editing models with task arithmetic. *Int. Conf. Learn. Represent.* (2023). https://doi.org/10.48550/arXiv.2212.04089. +59. D. Marczak, B. Twardowski, T. Trzciński, S. Cygert, MagMax: Leveraging model merging for seamless continual learning. *Proc. Eur. Conf. Comput. Vis.* (2024). https://doi.org/10.48550/arXiv.2407.06322. +60. A. Alexandrov, et al., Mitigating catastrophic forgetting in language transfer via model merging. *Findings Assoc. Comput. Linguist.: EMNLP* (2024). https://doi.org/10.48550/arXiv.2407.08699. +61. S. Dziadzio, et al., How to merge your multimodal models over time? *Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit.* (2025). https://doi.org/10.48550/arXiv.2412.06712. +62. M. Toneva, et al., An empirical study of example forgetting during deep neural network learning. *Int. Conf. Learn. Represent.* (2019). https://doi.org/10.48550/arXiv.1812.05159. +63. N. Kandpal, H. Deng, A. Roberts, E. Wallace, C. Raffel, Large language models struggle to learn long-tail knowledge. *Proc. Int. Conf. Mach. Learn.* (2023). https://doi.org/10.48550/arXiv.2211.08411. +64. X. Liu, et al., Long-tailed class incremental learning. *Proc. Eur. Conf. Comput. Vis.* (2022). https://doi.org/10.48550/arXiv.2210.00266. +65. S. Kotha, J. M. Springer, A. Raghunathan, Understanding catastrophic forgetting in language models via implicit inference. *Int. Conf. Learn. Represent.* (2024). https://doi.org/10.48550/arXiv.2309.10105. +66. L. Yu, B. Yu, H. Yu, F. Huang, Y. Li, Language models are super Mario: Absorbing abilities from homologous models as a free lunch. *Proc. Int. Conf. Mach. Learn.* (2024). https://doi.org/10.48550/arXiv.2311.03099. +67. M. Wortsman, et al., Model soups: Averaging weights of multiple fine-tuned models improves accuracy without increasing inference time. *Proc. Int. Conf. Mach. Learn.* (2022). https://doi.org/10.48550/arXiv.2203.05482. +68. Y. Cao, et al., An empirical study and theoretical explanation on task-level model-merging collapse. arXiv [Preprint] (2026). https://doi.org/10.48550/arXiv.2603.09463. +69. A. Livnat, C. Papadimitriou, Sex as an algorithm: The theory of evolution under the lens of computation. *Commun. ACM* **59**, 84–93 (2016). +70. Y. Guo, G. Shang, M. Vazirgiannis, C. Clavel, The curious decline of linguistic diversity: Training language models on synthetic text. *Findings Assoc. Comput. Linguist.: NAACL* (2024). https://doi.org/10.48550/arXiv.2311.09807. +71. V. Padmakumar, H. He, Does writing with language models reduce content diversity? *Int. Conf. Learn. Represent.* (2024). https://doi.org/10.48550/arXiv.2309.05196. +72. A. R. Doshi, O. P. Hauser, Generative AI enhances individual creativity but reduces the collective diversity of novel content. *Sci. Adv.* **10**, eadn5290 (2024). diff --git a/paper/pnas/main.pdf b/paper/pnas/main.pdf index bc14f7efbc6b8deea30d0c375668f927f20f8141..0d9ce17bb5a7f1e8eba83c9373386fec9f52f368 100644 GIT binary patch delta 99078 zcmY(}V|Q2$qeam+w(T^wZQDuXG`8&%+qP}9v2EM78{OyQegDN?bL`1pl<$3%x-B?> zorRO1AI8bq(Zs+8#(lk5XDsHh4Y})`<{XlG%I9YgH&z||t7)EH2x4|(qSJlfH&8#S zZ;e)5Wo@LGCLeCaxQr>}^YJp^oAxR29=|0a|MXjW>h7TR`*P-M_517VE|riF_{#jO z+dCF%VDtHKzn&4OZeP+kABi35VK)K-w!R7eY+ruMyL__@i@rat5w@Lt?bd&PBQOM7 zF?fXn->)P5o*(9g>6X&ig`JL#LnVF#->k!5%i^z&9^Y5FjMA$TPuCXwKKcA7kiLod{OBCRyp-HoGvO;R!i)r*607(f%a&NkPB%; zw8*2oZDA}xYq)dkEN-{r>m7grk@+1Cs#jezoVIPCgpWs@n^t}0Ru@Pwr9A`PANXI7 z<;}nOeb8HpF(s*g={gwXh}!}r#BvP<+*ty)%?KqnqP@?&<6OvfL(>A;G)@JHtM7*} z-wN+v`}|d2tPWvq`>qi)9^-yR2w1z7*VL@9sgg3$5yR&1jo2UX4`vb~mA3iq>dJW77uqGSrpc7rzU{fgY zouJaI2e-1hx8aHIMqRH`0(JP z=5C)EFYQjIqm4CTSXO{okCGY-T@yMv+0Bu^1)33b2A6~XjJ=^r;%bKz_a<)G^M0-1 z*A-nZ>or#64fKg$eM%|$(awmxzro5s>GQF>$CN8k$AB+EktUEp={dQEK81f&Fn1yK zb;0{&E?J*!%$6#-!hOBa_wo&l9ADwtoe&7K=~oz1?Djt$1qlI2Ym#wZD|d#=Ld=AV zMd2=^r=qlUsl`co=V4dnPcZZ@+;vy|+*Yi_nTL84SaPVdh17d6xs_C=C_bDtgOc8% zdhvHosvU2K;j2w19lW?sD1?R7q3{g(FugM@bJn%7Vv6Q>CoasHMn8}T$g8U&Vb)9N z<^gQGF@^wWQXkO|x=Nwd*j1PD;b*z^NQe+Mt*TAmlH7Zu^5h(sO-OZ|fe0jcIjOFxvzM@2 zg_%&`nmt{WVOmc1_+dUO*}wg9>$r=e(&1Y*^PPGZE;XRw=&sLAK~~by2yd~N(NO?# zZw>U}r=~E@IYhKAr~OcS0olkYf6`81dscLEwlE|G8b8$#I~!`kCu|f*hpEayhups+ z^8v-DS8i1Wq-}%jE)jDEv86^BPg(9>3!Y;20E(z(SZvu10qd(Eh6s5B0fr+xMJLsR z57-ZA8d*T}0lHDYl^&GBcIwiTc&{@ms6rgK${#u9K}FOQVuh`h?S?J zixJ2T2Sz*p&-AG0=Tj@*lHM%w-%Pgx!e zp1jia=*T)W9^*?EGs5RF>EAgW&(!tw`OiD>B*~I<8u@={6wVbbYMd2#Dtk;I2L!W( z-V4A!JgErF4V36rTr1A@ixQ-Ql77A+;4ua$ahebQSj z@UzXR%#3KmwMH>y;0@dRtXKYwPVGhay zVKZsz#y`I{RNU5p7{E@)MQkeCYd9>hKmc6KXB9X^ zNWlt4Y{Xp2b-(9|OQ#U1p2^6nD+vJ^RY`M64dhB=!4jyAGjyj>0~&`Ir3DvZe79;D z-ERGQ23Qzz_DqI-ovhJm5o{B$V+aUyS(8@-f9Lgc4s30NP|h3^Q%ptos=&8Q&dJ~S z6j?f|+y>!7N!dCJq&0`di+?Amo&u7Nj*+^(D!iDfe-(Y2LLibI|4s+(5qm9cYjWzX z{x~72K~1O*Rp#V){0w1p7|RLzf;qaGSP-JCR1(TQkwtttymERC@sW%m;R*$%h7urV z%#Zu*bChyQg@w+7XE4zMyDJiM9#T;J;}(MC$*z^&oFn!mT3I|ySeHxFI1l(!tR7FE zocMlqYdNGf8Y0?QQAL8aHQpxfhU+_+X76KS#i$a_$K6VQPI#m2_Da~BrRVIzszaw#kQR747+RIm9{Xyss-IzpL&_5k52h(e5)75KD4g312iChLLDC??nlnyp{r4Lp4Xk^1(G z?L08~XgS#bMomp^Ug!mG2R~#oXzm8(OrQIRK-*rGd))Wd4Y&uFnr+iTzEwDiMVLXd z*2j>MPq~ltXr;J%vb`Ap8DpCEVKV68Nh`xMMK#lW?gYyl<4s79+5o22R>𝔛j%C&h>7dMFqs@kA>GG_rYdA_;w(IHwd{L?ZH7DuLL~hG<3H zirol1PA*A>ki`5ZulPcI`S@FF>2kj5yq06$8C%9jKw-?YQP(8xus`gzhMq!ijISTs zQeGpE{LQOI$g5HK;>q=Hk z_zmJtW)9?{Jfbe?7I9fx}w{5Di>(Clo>Hogh??2dez5hpA&j*y>-( zwj(kc82N8RfRjj4n!sp0f@vPVBsYFi75)z!D2qkh{6z{;tbeX(DM;f{WRGgPOO^J) zD8wC_qYcw&htSno*`Oa=!XCbE);CvKwyMkHtypfskV2Wc3 zvWff=SR|(vtv_=NRONF>r4kXQLf&f*HI$)IBbT!Fvc&DSUnYrG0J+7E>B5QD1I(wJLFMFX%jmiyZq<{+^;WMI*re3?OfVL+>;-ojuxWji^juT zUMBx>ILhD1*GDSxjCu#o^Mh#@S+Es6vgb&_1jlk8^(c!|+ zBb={D+3sOb6qJg!rdxM^jQG*a%Di@ZC*37k#!nMbooJngQCsnT>^9w2qouG2f|&ev z6nXVPqu4b81o0ZLiib3t3L*@O?~!#-AaM@AQeOcf+~D!7 zPxPy?5`k=oTI#}MmY>4>FKEuWq3=XJy>Ki&Jsd9_v#ZKc|KqsuY z3!zYuYZ9$PzwV@M=J@@I@@na(XXq_+VJq_r=in;s7GBzjL3xOMM?*RkNxVa6S2qW# ziuo9xlUw)tnh+<|!6zKK4F#Fs(-kfBW%0qfAOw2gfhzG%;d% z3a18?42y`t@yLS9e0TW=roBD{AOKeRs8gR}^hXXJ^M_6^_Vqn#SR;Q}3%iQEnkWA?+X!|S;+)F_zt((%rvPjY_sN5>rEmEXaaO5H(2w5QqqV8j(e_uINc7j0 zwXV5%3*RaWn)ZMj0QqdL78yHzg1-y%n>yX*=D=N!Xp2WJw%|9Bi|wMEZ(uHh-^qz( zH;4C(I^(@^cVfX7q{@pwVaAx`xNI6;?e6ieR!liuYP+iXv0_Ci`RwRE4_0bYOuVJo z7N|?;yJl)?d>jc9EI(Q>E9;_mM?Zyag`lZ+(!y@>a(68r&|3IU%s8DPplFQZJajYB zE>s4cYsyXT)lrVWGE*m6?Q!{Do{HfJ$8hSkI*Vy=i2~cad7=!uwnH_Yq<3)gl!u>C z2-?y~cg%Iw6nu->3N8dECJaJ%n+gu&9Y;?{oH4oPD!qFSR0-gGaG(KstBt_SBOCjj zykf@~Q^-;V_{Jk!IET)}Q5NcbIXn{Bf?>H z7pc!?x|S+ij>h|--_WJlu|p8t=)^>7E72c4)Q}<%U^Aw-`UA^<(N2|ZS=z@BRqg)s zhy%4Gsf`k}$z1b_JfWGr=9a^w6cfL*JjbA$Hc#InK4}RY>|yA;aZbsl@__vx30xYx zAdlQF7dpP=rHrm1gL)48iNxJ{b-JB31dRY2i|eY-#V*DV7+o3{9cMqxWzp)XQy(9T zsyB-#;7DRO2vNZuM>4I=Q1i@2A7;N}m=3o6eVk|H_WBjO?V6|RBFy!{gNBQz zrbR`~2@xK$G*q@i{8ztv5+v#M%{4@m&z~6+FW0Akmx0Xeo5IOYZhkTN{FFv9JK8uE zSJQ5-Vlq{7B{zjG7|1o4e_9kCMs(X|Dm(ZS6)@Z;jh8^9$c(U(g_x!4;aYNg?wQub zzP>W^=5>JwBX4MHhb{d*;!!EYH++`ynDurp6}#n(Q-SA`)~@m?%0c@&;PdWoM&J|C zw-G2v>?7a=SeY3A2bHRII%8)A3#-7$~b1zdi`vA)BQ8C5n5W^@VIWjDN-ck z3sftvyY+4i8CBG%`CfUw7e8NTLO&UwfcN-ak|D=K9p%^8r=6$mp7WVM0$+~@2eAM& z;WFM3TH?bg7(jB4XU?L zO;2F;i>wcUhcWFH{hzfz8-FB&t-&v8hUTFSmhatEm9B&C_g}pm7EZ4` zHX}P0JiF(ISBRU176T@fxo2LyPtFP5Q8Lb$m^~A|bh(~c!PmR4~1kdYMu-~*oPIA#~tIp$6LNR>PPHmm*fR##_k_M?6V-e$?r*2m7Y zDu|}F(bP$MdZcJXGk?eC#vRLX;L&nLG8)b%=OeghL$8;rTw1cSp1@gp>n!x;+T^pF zXLDdR`|_OM74+%f1N033qIM#Tv1Ess@`h*N_O-bEPvV(c(yn zUcIXs22_wVD!>H0?eUw&b#a_$)DsqM`)qXv-m!{(qQS#guRWim0|U)9xzN+3AoWOI zDSMI&E`QD_S()sw=_XcJ9P_NhJOyGqX^E~fssf@q77QFLXm^DIsCj~1f5f^ zlN1OcpCQpqzSweo1hTyXy;rRZ&8)CjJy~*Bah@I+yKMG8n`szl$^^*GMd&tucpz^P zGLEgL)hp)fIt<j~OG zt?QMxpAC9sxqc8qV3#UHPO=U(w+sDl7a=8ZhV|dJT7=VU2M|j-9TRDZD8qW6B_iiO zoF`l`SD#C&`Ib_f8IeP#!k{Cy3`@BtxxJ-PHdEUvthTouGHsg?BobY zB8pa3skl-y|LUv%rpTw32&3whD(k)E(v_2uCosOmD4a`)4j{@y{58xvA(S+6A_Apd zrTt0Ry#RON1;D`fyO5*L5bG(q(#DZT=za!yn*SkeQ-cdo!b09hm`H~`3@sEBqkieK z3;4q@T7yld@*nh69lLS zPt_pyE5OoNU7|73l8lPXA?}0xvV1Wavts@eAy9Nq3lRTf4RZ%#!*Fo!4CZ?wDY>_Y zNib5cpf#L5>r_!ZC$3B5W9Ohy03~7Z4X-k9^M{c}IPZWFdp;1YK2azfI~w=ai_>5Jz>~ERZ>nt| zDJw~=5_jB*5)^sdc2O)x@c4#-7zz~a!{$_;-Xn$CBHr8eB&RX& z9Jk{Q<&h*BaJn%ul@!V(*$G+Zf{A!;!Jq$0&r=Lzoi~Y{*vfNSM;q*J*m`rinkPYr z(PyoipEVnGa;WI+e$+YavxWBhx}XWRxf6NY0ccSXjOm31dLIjzvqq6H{2*pp7v=>) z8>Mg)nl0#une`tqf@WckoxQ`9uu@M!5ynob=>gDtV%DI^&M1CgRfz@qTxQzBljcR- z+Xk{fre@V`P(RZV-CeFd-*vKFah}CBoUW|(tL5_q$Psrr$G(+Cm_m5#X8-x48t%K3 z1gK+Ji}_YPGZAu)K=D{0-P$w9I-5oJl7X=6oi!bc>X7OjE)PmAu*Wxu);0V<&1N5H z154(0*)`r@a2AIA!MOpgyjxWY25$2xnziN^FZF`(@}!L!60E2c$Bs=R*4AC(nrGaC z_HE1%nYelkjRip`TCGhQ3qq}`yOmt41f)P_hQ9h-m|JMPw=@T7gtkn$r|>Cs+>b9U zyzfdNxkJOuBRhdKC+gfwU=eEc|+?mk#X7ui75U{3~fK z67eo+4@E!cg-)GukyQ|XhnW(-%>#NLGkU8lu6u?#^|V>2K{k4bCk;zwv|jL|7WnC` za3Dl$Ejst)RFdd&l@(}1a;1=HtIsCrHjT2gnX3=za7Sd34! zXpNjFGb-cH!Ou}SedXU=?h`DG8$z?BCznHgwiaV&5b_EI(S1ciRe+|_RHQfW>Erz< zYmPNia4pLAFqIEiLQ#fm5_A}MA7C6hG21BlDc*iTg*jd~>-cy4dccA7yq{_}rbF17 z5Tzm?(}dT<>=JWk_c~KAyWz&Qhx8Y)pP7boxsR{bZxee!GV4iI!Ga6WvCE^Ny0J^MbP?5i`GHmbXjT32Y zXE-k(=UIie4?_{yRdEjP1!$wV{KT;Lkf)brIWLO&E8yC`?%bhb@nB(>$rz!VT(@8G zK~=0lp8)qqrb^J|GdTAgEg-NDk9t2hHPEW*(9Ak>VqhnF zV_sN2k#m7_Xhvd?azjNYF79c6XksHTLm511t>xKVwvR%;1R>x-(JG!)FGba! zuiAx!L=_Tg?uF;E=8aJ7sUxBeCA_lbta<{E6%*|l>y{o1D6s6P{o8jEo=`;qoc-YM zQcDnt%R<;6{0usVKycmmsz&Kw0mQf7DAS{k5G*-~5`>tys4aU8AVJ)h`#Gf{BOEv8)1Se0jv**8?E#`hA}Pi$-r%9h z3v|6<{q?<0%m9kKya@@B;wkFH!AdBzJq(AI6>HG7)@B!zje(JjHEX!uAQ*wtlqEIx6fN>fc}Ut9kO#t z56^qOr109wvg+F4cz$&0d!BZo8Y#&u-}Na+pu&IhGDV4jleu zt-j0iAyIqCRvwnOGph$YO?4b;QOnzPEgL+8v!shGmD?EIdKsU|5 zO-UgP8h{VtywsyDRW<{!%qLEG$2vCR=I@4(`5Gzp=S2r|toOPp6?RieIkK%&Q6Hyt zCuIYVsXB7l^JG4MrD*ybKTWDA{y|N%@mw0Tu!TTyc8de%jbqJ^Y9U&j`km0t8)!)g z@0egUgeADL{s*16O&=rCM7g4qm(=DmPPy=^P2e}5rh42Cu1(IFbnEq<(~|$g?prR` zi-)gb`7x81IzuKs4)oa;7g3$`6jXQP&t-}Q!J?jrU&wf+Pi3~fw3mxP=|cmzH=4u4 zmP2)mSe7lS@f4+1lmctNc)aJC{nDtVy-(=mO5=9|x$(q#%-FUkd0nYfs>Q@872a>~ zKmqe-&NL$%M_C&X{Lu=Yot0nS{2_U86xHd|I>6g)({w95m!XH|d1PofJ0fBMs_t0q z^&e96F_}upe7t-2YNNsJ=XC$_9lI+-X^9`1_lJYx-~Y^amu%itnC)!><)05dUu#lm z{CUU99us7q)xLMiDGy5?Jw2_la>%Qf5dbM4IHvBOyBw+yC+Qx!%kRsCDt%Kp4(^M5MGRtxve5Or{2sgYq zS=T@9XZQp1))?d}Yd&(VK7?A+WeGA@mi&DgMxGUX9wQoVe5!-ppc~`6&e8M>H-Spl zTy%Ey)Xnu#BW+Tt^7OIKS$q~V^&yWCdu^@ANw3nyfU7`l9h$D8YcHZh`@auE^6L9y9HI z5wi!-1-Gp-s`8AH;bFIEMgh!R=l~KTc2tLdcn#FM%|?+)&dJox#RrOrcuHJ(XvlN7 z>#k@k_Z`$srYHJy4hzJj!PZ+1m*7q~)|@{q>3azOWy3`m`az8q5_k28+;D}PbGeWr zs&t!q)P>D2p6UWkjF-~m5M&ZZlc1kjE#-O{>N!gh_sWr)ky=^lw{b0MM8Mvu?h+S8 zTxROOdIicgJ|B|g1>+Tu@#%%0YTO&@n>&juoj(5xMv zgp_ehrZ4;eUTe-B@QyKeSNenH;Ezu+o4C$X@wMiV!yuB#XZt!BMV1$xH|Q0oJCF8G z`1(bCmi~t8Bew<$RV5X|0niy8vHOwG*Iskwi$2}}tZgJnIi?%MMjl$5?+#IeTXk6${6#?32$&C#8k{6eZ8?ov@wMU<3nqBYZrSfV|{`+vx=uqfp+( zd@2p;Kz%7%8jsB9sLs1E7d5X&TNxMOlLQk&E^YYt3VlpSE5~{ zAf4&mBJbEcJZWzs;IiN*$M?AbmZub!&JK{3JUXMRcy_J?P3N|h`jZrM{UQ58`iytv zHKpP)Q_%)h{~Ha*msF$zJ+AJAp;L%Gh7=E~T#{xg23K|d#o8%fF2xZ#cW2+jAAR64 zlS^=a?e@oKzM>N$eJj*z(Ky%xmpt-A=kLlu^G>nJ{hd#u0T3BO=Fy=$LRiK9T!5}jzs z0A_ZM|Nl#=#%r%RY-4&+cW%RN3ZdQhy9EK&1D(dmnRLBn<+~chwLaY8&@_%0i?}pe zb~pco%hGZ#SZ;^7+A6!^77y4 zFWCui9pXXQwp4U)EtTRe7mr80D9B8Wbcz!24)kit%H66S?<*#N>c}*znUDTr3bT%C zprmgMB;cP8F$IB|Fqd}##g-bv9<>518tXywhxpDWU2W-vC(P*>uK9bLG!#4W7~uY2 z=+864+qeXA3&RXw+x)_6_4_4IulSpr843CaFVAgqQsEt^Z9goZQmwLe->aavmU(7N zNg_r<3)d$adMPK6Z6I4N6R}fyWxK45fC4{cwEW~{k@k$j(w2#crnX%d{#B-bN?N3JF z3`4Q6#VX$rV7x4}fXCUt52H@D4?@g|o~o=p8>1SX0)7EVh+yUKR$0ztP!bkw@}1I4 zi$mtt5(wO5~jh;-3Ve12_Uw+VM4_45qYexpTI1?zr>Tjo#tLt|Q$CG+9WY~#S&Y*8Eu zA>aLo7|w$~=-TLau~uA}q}&am^DXvN>yg5}@tZ+wqJD6!{(c*<)8$1k_e;_@$kwEl z-Dmbm{G*wYG{!j3W7Zi&ay+Xb`xeY%YR`q{nr5&-;B1Twbva~M6?VWXE>h77o9fnA zoPUHgF$x9~iTlGa_$P9Y{~o5vZjiQWr`S+(*NV5vwKpy)lx^??M$}|d%6Rq}yI93!2W977qVu9j zuRoChl-%>jPf>(8?XZY4Zn2b0lti6)^|T8=%O5Du0=R4n7`gAMc(6rL*(8X?X7`1f ziiTF&HZ~dJzjDnL<`*!k6(EmxFV`ayJV%++PwF2AV;@1!j^@(X{L@v@UnosPzeps& z;R6~N?X8-LB@#ioap~5g21}gl40x3qAYR`9Z!sb#c7k zjyF@BOp1D?1nI~8aT3ggIOI&jn&gsIah;I1hOfWBfGZV70DB)#;V=}*5|*7xb#pV~0GL`rL%E}-bxL7Ub-CwQYD z70$xxf0rNLhe{i&bJ1(o8cD}mLLOjq$LitkexIJVpFt&=;T1$wi(Of*-G)B{G6_`8 zeU}Ydf@JKygzGn;e7Q|oM$@gr*3vGP@lsQgei~8Q=A2dcJ#hCe7DYokvJIfPn}=pZ zWhZ+D+5^AjsD#P!^Yw?py(?Xp%mnO*qu(WTuZ_nH0*5QH7vDo$BR3)RU`>ZKY2~p< zYKL*A_Ddxsnud+}6W!P>TgEQ|SWH%m9})7^;iE5Itf z`#a^DCP85ZE*fUO451T80ys`)9Q|n`H5P5dS=Df-QoRri+fMW{$%s!fkyH`17pNN5 z*<@5m6gfqwJj%3g8Pfs>CB#y_7}-JEE2LE`r_?+B!_M{n;chF0^rB-qf{I;trLV;&km4x{ENuyjny%=LI}xpKv^lB*MS4MFMIp9;~D zpK>E{UUd9S6J`;6Ug(vs+-CA&Gy{4V6?hxUYYQXd%|?I3NKMe?=CmR0m-PJmKq<5% zPW1M<>qPvad;^b&ct-^cP0Al&xGR6Sj$ko#Hw2yDq3I6MQBFAD zXEhOuU#fUy8}5*Q$sbYAP2h9pRHAnC+_fvSL39`p@~+K|(Ej6TXy0g(E^agik49je z3Di|{;j!kven~Kurm921%t-QExZm;=c2$=&6|6@p)`Gzs$W8$I+SJ9|^|B2Ol@nDn zETrR?30ZLL86joSbt$5aPkWip=Q8$*gkXUeR36sdnx^}X>=+~?&^pqsXb9}3d zyO1Dgb}@eErD?>5V3BPjV2DB2bBHBA;2J_Xu+7A)VxaLLN@DT+78Hh{U3-Od#I%jC z-TqOgP6jpIc;OFtK>2&`uF%Apv8Qa^mM1;^=($kaa?%@OKWORgfm#D${qLV91)Rk~R6b6=FA)L?77Hf;N`pfbB1 z)DZvCyd7QmG`B*GKZ1?aphm3gh^|q#9bcX7LFM>*OGzG2rc*=rCiG$H`KVTPf{myP zU#{=;o5vj3BUCLN52NQ|bE8SxDS8}_u@zg(Hu{^j3+`4GW$bE$!Evlasfy~do!!0o z)Zn+6p5cU*yX0cJ$r(CLq#vA6>1_DG-=V}3P}snuaaBmuO`3bm=akVb$z0m8)J9sp zkgo@6N-7a_U!*{w`Q&EDRs)yEoy1rLy3*Higa-*&@kUDg+k(EImcw53g;3IjzC4LVqok%?Gc<8+#@34a+uVgQBdwNo+8Sq%@*o6=->MQV7=^$R0r)5e|6lDtx0! zfAIo;SRf9Fgfcov)k+qCqmv349c8s%W!-{i^{I?HX6~oHe+a_Rrhm7TFYUxlnM{3~ zplW}&WW!C)s9=4TbRCOb-L$XU>Rc<%>8E9;)^0aKAM36Zhsb43%~bn!;fvfY+G^o4 zQ*`RPyJTbsouE!==B|S+B*&y*r{m&mA87zT(w~_zo^m+Rin7nb$iv=VZxu&-(+N=< z8t5_a|6)j3Xl7f`UXvSq&Xx;-49C@1vKx|znGsqU$ zFU`D9YFyp#r*6tmyf+iXOdQT}rbCx< z@?e5C`+UR^rj8amLxQ{q$_uxD?`u`>QdymuTs$hM>U!;xW{k6|AIm@)5cR;%n#{!q z;|Nd6^HmkFwKiaSv&cO!8T+2jYL0;RER@!>sM&O{gc(PpER^XQ&6SyoM`5NM1Du31 zC$0ncCeLzmDtO(@z3Od`87M3COX^Np90&1a@k$Sla?a>K?zrUE!`y3=j%aP=ZwQ&F zL>F}@2CYkfK@4dqv>Iom4Nji-0#MjsWfvA|AeS;47uq-*R5Byu^KElR@8*Gcg(Z;6 z&_}$!e}2|ZX@6zW)5HT8jdN*R9*1Zar0srD3QWY0-dN!A4DaUs+!ZS$$k2Gpu|%cz zEK%_26GdeDT5=d27jNRU>E;cS1Gm`1zI*QbCjfT)J@d(4XNW*j%zjBL`$)?{!%OB@ zai@nruF74eX7?yBO)$mTF%S*xZ*H;$UWU+`c8f&kpw7e|M`6o0_&#JIo$FJ0_4@xH zE<9QG0jpF7ZSZ8&hSJKyn+w0A&MKXhscDkqO*b9Xizx!h&;{Ellx5J(K||GU~rM@%17h0)+yP z2BQ(n-$&n*A&%*ku7gu3$sWrrg4SkkII;%>_#33~w|odTv$lIPoa3N6(?_N0jy#8R zYilcgu$126IfGp+R0h^TTw4zJC9@ePvV8T68<~XUD4~8UR1{5>l=Er6^kjSPWT>{l zb&qyFV7u<6>j?pEQ<6isuT5;M?;YO??C4RBc^fbF zo~zdq_m0*B*aRM7X{Ei0#q6akyPMSiK=+i&6)QqO-#dI%?E%x$LYT(1UAFrTKQqBIYynJq7N=|Xri9er< z7+HlPCoLi_sJ{$sPe`2Z90<2u%4Ry+aH58LBedr(G=ep!6b4hJG#V4a30X9dkS<`I=m zJ!~}_MS07pZdT>WSI{{Z<1h~xj(Ha`=#&cd%IJ4_Os9U-roi8f;Jr=p_D-}k4 zG^@T7um|J2u&d5V_E41XHHH~K9Mx<=74;ISD|$@i%QytF!BMyBYQh<%| ze~qJ^fmGrqgKKyK_5?FN*xiC7S&ZX4v_s)mUzIEBpUdq*>&4@x|4vMQ?;3vM=f zj0Edt=;L(<@iTv1;1hXD0`5(cO5*$OV&`bBXZ#!R{kR$#`SYDG(8S;H!d(yK{kH@lMNdy-(VeOsB1(%+t*&JukW zTmR(Bk3eb{{zZgXhLkja^TrOwv@i5cnNT@-D2V9thpMYrTLzT{Wd>s zmT+ZqeQc?RK@dNM0)IUvsB_ZR?%N26V{f=rQ-8k@_>d@08K-7D_aR`<6uwqChyke{ z))fZfv=K{i2%7D73}s34g;7z8@nS#h=7!M2pR);yzHMp*pPuk$@D5-lnwQ{Zh}oa? zPYdL82m*{0XeI**Oia(Q=A;V-2kqPNbkQLMLzwKnt>R8mE%@%$yzZw2&rXS4J$yZg zHgoW_+6OcWQLn3?P@!*A`qtxd!}XsOG8YYOB=hg$+}4O(u!pN%h#X5C4fkjh$*F!= z9*n|-L_eh#=?YZ!bKuHb^W>d2FRkC{Z?SJwT>$%u(%&J&@^FWBO#N)Wum7we84Z~u zaIKr^5oFg&*~4d?5gfKF$jucC`y<1Li^92gXZpA(32H7zgk?VQUy(PCYw2>~U3w+C zg8cF^9| ztN^);%iavwCbvEeGt@uoR3w26-}uu*K8C4CuqS7BJVG!1cDkHZk>Oy9=EArYx1MZ4 zcIL_demnVE^nj#cpn2Bvj3HIoS3q6vHMKZU=3_!NJH3VR;{xQ1qZi~yp@(Fuxwj>2 z2e8!;g%xzyG*tAnM;5=cLC+`q&9)N_zb>!01l*T9 ziN8hzGr60vjq|A3Kr8p?9nnx)Yi0FdsnLk&V{T|dYG8Uj53ClI`rzVl`WrK#4)weo z#Onq7S;?3x(Q~=;e2Tw{G*po~Y7)$I0(YilZsT=`vQ8>*KZ3sb{ZfrUNyuXsk^zg3 zFQ;odza7bD+eT+d5g%-4&y2@~{p+EhHdY#g9R6P_j@A;(v#@nIEd0Vd^6TsKa28!X z{7(my0dAuTP?F<$9t}$mO*?IXzzawB5tijF6J&)Dgw@~Vc;9>W4EdO2R_tNVQ}5tw zFnA3Nn&M|YZf&O)9==oz_n-b$3!`Z?2$Ws7!PqRu z$OX?wbUggbV3l!V;|akQf&;i9dV(T`-j4W^Tb`x-?-&W!X=dPuVUN z;I&ruoVh%Le0Y^7gb0V|WXyHTsDen+Zs|`X|L|06PhrWJlZ{E1>uysIXU(mdLZ(q% zw;!y}%4gg^e1ZMLo>lV~7Qn2yJDtt05_a~G4{H*dswrPLWXLZ&zc$awN}1BjCjh51 zV7@d(JNC&Lu0lE~B;@5-5l4*iEv^bZrW^ZZAxxc9tz>Etxsp!fZGLjuE5xXxkuaM2 z%UmOwPD z_;aA+p0~ltIm!~Y~ZJ7dUjlRZOlSx-`3a+z2l8o(+!8q*L{`Ui4xF8 zRuIXm@)uBj>u=_Mcly1iCZZ#n!0x4c(h&ulnzz8>F8CU2t#u?OyKn};yY`;Xa3&1= z6a2KIz~`Tg@Q|fl7Ztj)j=mloLBl%c5{1*tHYjXsTxzws;VQ%LuD|66iT{025)H@V-*ZcoYPEZ3fCJniIwkyNgVLCNFErydSCLDA${v)pjH@_ER|= zlgXSvB{6fC-rR^^3xeOz`-@z8|Ay>NE744X%mi4WaLa<{%x}bmBE$iw{tej5yngK@ zeV3#p>Iw>Co<_;|O?P|+dGm{+1CQXF{Z|GNh$)#FfYT|jQ^0UNjM*bg0u#}|?M!hK z&AJl5GbWna@S}YlF86x)Wl3RHW#ZcC3iavSUSDSAtD$q&z*q~0hu2l=g`&hr7LlJx z;Q*g;23!Hp)dF>>t+;c+ss{+H3(`w0lPR=WsFK44e(26(C}W_B(#x_$sZN20Rp?$> zPMzcl008+q-846ArlyRAG16$q^E%tVDq@rQ{N|;Fa8qmgSwyG#4%xai0iO zkw%rJLoy=0^G}fCNZiz6EG0pR$C^A%qd9XQYN|G2sDQnfEhs`$2pdJa>9oh=gF&4K z$T=bCkAG6|YM%EnxVPAsW@Ge6n})b`r3v2SpiJaH%0200uq>(qGTUa7c7Oj_+EjQ3 z#RnW16!X9dleod{gHrw2i<_Fp8lr`Twre-5L9e^vGHk9TbEC%{W7R2ypuf#gh8C)i zD}Q4*Y+X~Z{E}Ckfcb@ujP{_W$j6xjC_-aC5lMoQjmDRFXc|l9Eae(;GDm^VQnc|U zIG`)EsAx2iEyBBqiyPb|y8DP)EhV9|7n0-;VJ{6SP(~@m7b!~H-G{Xh$TUd!lf|rd zTCS$xBW7sQRBhf5I?dpu#2@!{{yBF3jd{`w!!9Hs4jI9_TIOL3nwO^lN(2%JfaK|y z1bj_15Gkt{_$4qtcKn8R6H6)VZ1S?Q**+cKc_Um9#WO#9q4UCDQs_G0&Zwa*b-q&AJu zosc?J@7s`CW;`3PAQazR7P|Xu1zm|FCdk{75lsnwKjC^C5;4;ea$sFnA? zpW@j+QDzLPkj*ylqTy5oV5t!WQ?7q@IrrkmQ;{8$SWU2|><05u!(S*}$kwr(`Rp>D z2{tBG@J^Gj$SkRt443=h$wkOBSw43QWe!XswYI}@-CrMg|DL_DBR@&Ryw6fwO`RQv z;bVHc;qi(0^o25&VP42s?Z`;&-o<0MSJ)33=3vL)=T5Q?Rr`Yi=vf(elHxSxfGvlH z_tA?lzOVzCTSlbYkT!Sl$CP)`B8V(r{Uv5pji6F79jXR+#&Zwm&iY^N;Qx(d6*R{hF5O;SNLs$#^nOwtg>JoPvlX{g7-;zvyaOD zgmUG_>cRZt-X3ZM5HRuk{cD%Lg;Qn0#KLcIC&}x!dD=O}(yGhLs|U=@w7+oN(-r7b z5_$fQi?@!eqU-*@or8pQO9=u3(%lUrf}}`yNq6S~0YOqYz|bk(d882q2>}7=P6=s5 z+Gpgt?%#cXpMQL<*?aAnnfF;6=d1IXra4g4yMXn@V@Zo;sb=o^Qu%^WO;YM+d<-PQ9&{EWPHv*0G*ADN~g+a}b z{b%cW*`Ka=rRNk{l0;tP_Emh;*+#wA60m!)n)NIBT+V{oPZcf1DqA$=KN4$jG9vH9RX2 zXY)f@+~su-PJXO+5vj*91&gsQOX#YnMw9Z`;}QqwP_g68hx0Q+9IhE#h@-C1NxAl=&il_9wLC({+qr&6fWA<5)K%d++Pu@Or*1 zMSgO);aV;I5vS(Bygt$5LR`@QBkJo;; zM=+ewZ0$QT2Rbc~v=Ukwq=_+SWzJ>)!r^`cZ_o?=&8svlM_*JE?wR@v);?Hd(ueLp zNjZu8umB^Ly(3mefcNrOW3qEqYFpo)vc!iACZJNurab|>>Z<#?BAG-b^H=_7T4nLR zMZWdi@zt*FwVm`)A+x*TvAfcZdNCXo!cPnw>x{{l1gBSxUdtABDeGZA{ao;Y>AB|? zH=N4y!}lKBAJS6`Y`;52E7d|Y^Q4-#pt2z0*+~M@wdA)hCOKvPH9^8s8#*%Sq z?(``_;_OkhH`Y>eH4m;P)~7PWj^g)c6TByaD{X=4nsW<#0UqQ#{=AZn9mXi{%5-p= ziv&#Nt&OC;UlfmIV@v<|HB9Tg{nV&2PGlH9kd@D0J;#J3|Jie8Npxo=%5N+0VE^lM zx0zLHVJ=qF!O7ru9p^kk9$`emrL%y8C0dR9d{+NHpfx#W_>0i&dPOQb=C_d{ClA3% zM^XI6c#Co}g;ssKXXb?R(pS;GaDod1jAl{H@g9UhD!82{dT7T03|@2H#7yDJY(Isk zoIVm5n{Zkb^)2PHUozjemvZ$ID$nG}Fc+WW7TbaG)o$al`y^ zXjIJEdax-5*)BE=%((mz8frMuqsQ{NKdsVKGe4h9s>(c!dM{ z5k6_Wl~(VGvx7b)%@PP_I$Ka07`chSeffH4w9+3Ha-ThUQWxB@yc@r3uhEs!BQ8CH z-Ela5H;68SqN#C7Q~Y5&Ssr3FVIojC4vip>TkDKG_g?hXIHz5s z@unDF7UeIerw@DvI2DMG)56z=lBJKt{#cyxfTnb zBR4T;6&jf9hn=!D9x5gq^s*|pdZ4!#5M0JgG{yPcnR5xZ}PeVe}tYLf~mJ-i;} zBgD_Q2LpBz23PaGA>#c$*&SIQ!|A*o?ElQHG6`fFIPPw=AH2 z7g^B7CZU9Qu)}Hh3K5Q%9^h^IvifaW?ep*HM4zpa7vv(MW^p{%e)KxHyA>4SsXb=< zFjAzgv!9@;kvNw9wfUWuOi#ti9s#pa2pO}7!A{(SSpSZf2dbH<27W$NT?mMLXPU-|9-KdzPyRw|85DF{Dq|o_V$t;$)XKUraGZn=F#ge}2Y6#UcEo zfMhpT_~dtg;h{^a>o#Gu-UloH!k}ct~dqPw+lF ze0~4>3^@ZLbU%$%%VYKCmsCH;bAuXLK5iU2HZeeR1{)R zodU)H=NAzEKj}^eEtQXMcT5S^hv=96MzMA`>9m8kO!e8LkB#{~5T5#@+V~_AujJ)5 z8CleHYNeJ!;*4&JH`Sr{TXPp@VdGka5O=F*FnQ z<1aP(zaXow&#f-0o?oAN{((<+lU;rygMaD>nf)^tqj0%cVk8-$;)jKVr$=*-oH}LH zp9tELDB8JPyBBo89*|xkH0+Wuo%gW*#OC7Ed9BRwZa;prpg@V2HdlipMA~PNc7*_v&`cf+wFTas0JY>siB-1(nx3#bG+Gz!KT?uBaLYpWxr*0~t zN?Z>h&h0B=JQXl2ld1^cqZ^5RC&HiGhH)MCJYHjVf+KGyz?F4|e~#X_^kpPzdO=kl zZgn!6z=@k%HFvORm2z!!>1;@)*0W_$^ou{Zpjj~nfU8!*1hlwkj~FH{dG zY)n;cwSBHQ|NgP%LKRWd#_G3%>#iH0d{=ee`>E!0lDpf~fwO%hhmr;LD|VdJNtfNjyT2h8|WocOsF_aYpuZREHK$raO+bPofzN1N!$ z^Cl)KJ!8=cC5KVIb;3W;-Ox3L|4NXPQvMM5+DmHQp7L~{Y9c!D&@HNIif7-{b?r~( zND#GQzGc8nkwy$fN6ekmGQA9uR{9?@Un8D9ubBOL>_rt{o_bn4wH=QvB;7S0Y<@+5 zF+Z~*8u03#_#oyW-3(hAZ3E`LAIj{eaJ0U!a$-jdgxcOP>V;`Id(rh`AtyVvky_{L z3*ae%QkdpbY1X=@SW}S(ZpR<4+xn_4>4(DX17&C>)72XHXz(s84SDwQ^vGN=vm7Qu zBTC0Nb1@?yTB*FTq%v=;F<~v;YKe3>>m{+r3+j);`lOW~t8}fQ8ZMltszyzWzQPbe z5o6uUrMg&+Vi7C?|49e;Z?~7PL9-rj2)GwEH29(}`WIcJz}?Y}LTcR-dC`tD_Q%hE zliw6RX=hcK@mUhAN5_`lC!MZedfIkixV|26^@OSE`RFyLca>;>YRg-PUh#PFxmOKo z)}`3ADKN^MJW#yEWyzb$9r9X_g!)_M=#jgI-{&&Hbx)_-yKsp&K2_7aUD_dLokaM3 z(WeQ{+Cf@}d8*+*aodM~01}xtvNZEH-33MS-_-Do=>&u1c*98r2^A^BJOigJyM5y| zMkYu-f3yp+e~aU3!9!)JrVWyRJ_>p?84s-UHqzS&@W{njFL1*zo?##6)r`e=G4V-K zA9{9L_&s+H+JpC-Vp^wg-@m7YLCpPOgLH7{{EGkmgGk(22#-F3SZh|hfu9dNSw zu6_$AbfdtN*+%l2sMW(@^DXlQ`D|=P>T}CYqi2K;AJ5=Ir}*glUvo>M<*Xue1b=QR zOW%#^3KDsz67bc{joYzJ#b9@u@(>}QaSClLXS>OIzoeyi~F5Msj$ALZOL??kdrTMIJ6u{AkFdJm0=<*3F>M4o#7VHxpDD zr9NX%OO~JQtTQ)JDOW%93+Pqz345;BjNszu9zh{McmKKu=5~?+)lMe zEU&g6mjr0pn^^v5E-W{mam=2SI1Eb}E#@XYM`E!k`cR zn>v13^R};4mYXH^j@VZ3;do&2hoZPW>?}ubxFOcl`Bou*>a|Fqcc*CMhY3cuwN-JZ z?50R20-k|uq2~QF=j=<1Q1~)$99v~;Hiwy5tGeh&fj@?H57!F&@hG}me@R-r#_NJe zUsJv8kedutI^N5BBaL2Icsx2JcV0U&F`vdb%3SSIzOwuo-k={d!r0$0w2!=wG1*eO zj~YT)!BZPP?}%^NJWHj+v{aTZKUk@8C!?_`ju=kNj3Y$D<%iY(E&VNz!61&Gv)e)X zoV9P8Zn{wlme4#OzRlZ(oW2ez{8}@X*J=XoZ!(ewcheUan_iU1rc#;u_Ba@%uQvLa z&g$z?FSU?rG2JcvPF^_|#+f9DnJ0b`!LhGS2JJ_bMCh`s=f9HtU?gncwDVjbv`hW#*Eh zoEY$H5>9Ex`t_gzxtx0NiZoc-Td&W3T(~}efreD4^HL>DHsOuBdispNWEY7y9OL=7 z$yPx1r}pT|koRy4&)XE@k@Uy=k@5E04+WXkD)mb(@dBTO_>S?(3CvQ{H#x@MTXZT2!kZpm?sVKdc?l5B|B|!qZW+ zLzvPm+ZOR0({F^FenatuujGB;zFcmb@N@(_^&_m}KI%U#V9v)fnRdO#qB zF<&&%lDo`IYAc>@O||7eFt$@jG2Yeb_~56!`d%{ORO$4dbeWvB;l@!ZnVw~-zcI0Pbr$HE=<+DRNa#lTtOw|OhPW)eTCd>Pg zBiso%BiP+KvZO|xRFB6$804MI-`KbB30Nd`=fL7kc3;M#IF2{l&>Y^A=U)8c8c*9~ zy}MJ~CaVC)NMp-0+>ZU>IktkW`9v_V^fBBctxGDk%>79S{XH(~PXvBa6Gxw8{9ix2 zpwOvF6h`}Mac_HvYlZFj(Y8QkqkNBu`}QtF7LQJ51y+W}54MT}Q8MuZ_3odxv*FKn zJt}|EddOMC;suhu+Z8A2B%c=OB|v*KbTj=v%R!v1)48`iToX+dgEDKoAN|DQ(8i)0 z9`f+c*IGg6NL@cI0}`zhsV9Cg;{sxZrFG2&j34-J@K1R=`}tz5+*=9lYRV1NgNN38 z?2Nu&ZT*zzDDRGi?AGU@i64op#+YfDx(T1xzZQ$(>TMAHl)Dz48Jrfl0^{(!D`5CE z9N>R=8l=C(4kv0Iy)c(`LP$(`P?0At!v(+*92>MmI`JXFmCJ8+gpp${qp&IeQwmv@}Z-&b=W^%}2EH6$fEQ$y-T?y}hD8&rKM`ep#FfP2d zGGZ%We?seIMUNZENquGX^|?vF?v{gZc!d5?oqvVR@4EM)9+q`Y#NF}UicaUVIwfCD z^PORACj_{?vn=}(wMAH!{j9hdz8i^>&+{jw>5VG2zp0Ff#xgedbIQo+(R0HEADn)# z-d%7tM#yrb0#RiFIgR*@a(&6azBW>l;A4h1F<3VJXl3a8f{(J`EMo4c7vLUHv%>CN z_%bdJk0GT$*y&LRJ9F4-sq3(a(C2E_XEifFBq}j$b!a?u}PYh`gmN47x4SXU4~Ra zR1D0`MRBfyBhN~;$^xFD*p@I83QgH7F})(eA@w64P)YG$K7H?Era4@SC=yaG)zhQq z7GrqkUS|1?Y3u7jvC8OPTAgfc4pk-b;?c#)0DlXiykA1!*Uu8ZZyIbtBTY*?16Hb& zgqbj&J4B3Xf>jcY&U@auk)v?ArF=LdMI4NOQ51RlKu>1YQ&Mx2W?fq;g=M};uc7EL zV?clIkG9Ib&!%kd-sf*_3(aZ7Z3HbXF%rKp-+q7IOxrTWN$|i8RrafT{zdYbfTbeS zB<;^qn*glD*_c$ML=X0a^^l-%Tm(_T1FJy}wQLmf*Z}_+eDfepM;0=8k|NTW)5~uR8-TKyJ?D*J183F$JUwt7X=-9L-!`++A#K?oZCkj33@MQFbj2 zX~lKDUd3~}%8MeCOFMUjd%}m=W%f+&1L9YA?-XKh_~RPVSRs__NjN_pI&wr=lHr^t zc7AwBFTG!QK_ST|un4fVMG{Be{M`NZT-*6A&$_(j>9rk;sj@D^mNUZ_xlU5qk?n&a z=C6B$)CR3zcJdmndaJJql(VC_ih6_3)f!IyVTU_WbJ(_y@U`hsy(o|Vw7mY~ zlm6`+hM%g!N^ReZFybrstGwN>Hu%M6>|E?pGv^G~zP-|Q(Ym(#0JL542fs*qyKKC6 z_F{Z|N`|t9x-K4PYd*G&7u!7@PM%LQMVggbZ7_X*WO`IO!NExb_hZ={ukqA`JF~&^ zcP(r^63Zx@U%a2UKf>*l+jm}$`dpH*_~E%PH;3urPMZD6rmsbcmKY4U%Llg z-hk#sW_=`CHKC^p# z<1F~ZSMQowjcYqr(+ieJZgRi!h&y$fmpANBFeZgg4?~cs+MB7HRJ_xh!z%n^f7bnwwfR$B`Z;C6yr4xXr{z*wgZ-EiANxH;w6ig<_tP@YPL-HhQRh@M0u} zRhdTSD4`)&?h(*Na#Q})+j;)&OdG;sGm)W?;XbJD^5lqW2EO{-Eb+&U-qi#VRjpn4 zz?|MSx-3c7+A4})a9t+uczf$uMMLV&&(c>IU(zSozRY|QjO?9K8}0hUJ3djztp2Ao zdZCR`HqjW9t9JMi5%gKMQa!Yy^f|E>rrkT z-k5(`(P!3;HhF}3bOU`ALUBl^W&wjuP>2z8ss!MY7D_BA9RLs^|LnqK0T`b!GMEiu z1rWrVM1L!35u)l203V{tk+*J6y%0d~Y7y2YY6Js-0LV>_LR81Bi3xxsXpr?(iv>V@ z*CRr9vIDOG#2(jTfcXFjFK;5S zGa0Z)5c?8=QXn9U(Doxnbb5V2DESj2os(dZPzU}*Aar|lNG=ct&BZH#7*r)iSwbiT zLTxW#bci5-U69)wMuhn3AB8Z36N64;PdQau=j96;gb;;plU*8+fG9a-=<01!q< z+Y!}Wy(R^~bxMG?jyIGZ1bD&tA^>{#lz=6`_g@=~F9fb(+}vP1GYSK!#*RV)Ry_bD zpYZWKFNkw3*KI9Seshd zI9g$Ha|?m-IWPsJA2ooD4F<6RgnyBWfpPQv4{|8nJh#lT17Hj{K=_v)YXI!z0sdn1 z2uS7wi2uO`9O8#eh#Oh}%n*Rgs$du!04@kZ<}Ep3h!8;hUt~aP5vb;t84yDhAcmlc z2SXv_A8J4XF{q5gasmv84A96P#TpzFhpfNofb9|h@n3d81u1~!FFe8k=pYT{x0~$dl_(vWH0NI`cM98>jfDZunDM9`I1qj4e0Z1T% zkRYH6HMvCyjDyVoVF)x+gRUQ{0IOi-1&j{@NC)&+hl>7!1LGFBeFb%ffoKrb2p4T4 zQ1UrIi>y@#Yyp55`9K3801&}eL`dnEzT2H#e8;2GYf;@CUT^a1o#d_19l{=$XAEzE~>)<)3;zyvRV6lv@MPy%2VDheSo z+7rO|Csu(u5Jd<)HGFW&3)lsCz>k+Kw z<_O3UiePU6@IwSZc*`Q&Td2h?i{6ot1+nN0T0{Z2aSYsxh6cF}WFUSlRC~)}F=YM2 z;!+&c-)$@dTi!v{e}fsA5)WDbVeu{ik|aP63SzMbyhwm*AQt<<_(Z4-V$m9Wm;_n3 zES4ujok1*y0w4znK?||y2Zn%9;cZX@Ra2nWP*igS_fw&2h)Ow-Hyt3lrIIlN^81I% zddU1=D#J6OYlzBa6dpcqu<8QFdrM|+7F7I~OsJR_39T&m0T>S-sGb7B-l;m@D|3;1#ad5E1+IJG}-xFzzz|tK?HJW05{;f0ssfRh=DO9&VW13*5!prbU91Q~qDgBTbz*%}-a zUl=bJV#IYFDis84(qTkMd;|~#K&7BGG*I$02x?Am=sqEod;xGd2zrmVVePRm>^6XZ zK;ebf2xxXhNU7&Q1ORIL!AOvJAPl-^Zaz>t1$GQ@|G%&X1%BLbbtP};$8e;9F{ z0Tk6h5jZqLS32x9^q51TelK!>VbCKCvLI%l|0sn+)lFG2YG_`!JUsfl_K|`_IWS>4 zo_}N2e^B7%`DbbSFABUoLbq`ICwB4j`~%N_7C&BIE+iDYj-SDJdEyBk!MXW``2SaZ zYtU(mT@%2E&MU;+HK}jKliKX8o(y`x&Xu}r;=ATq&vp8TsE#8cPcIrVHCjv_l8{KD$u#Z90;yX5tIO<_<|acl?mce+qZMpBm1=0-pGH+g^23ZM6LF8byU%MO;Bx4dHo~>9vClmhpnvN|3EWP+CAHKk%J&RR<(TCp= zA8s;UrvYD~R3O~Cyfg)4T+x(Z==~CW%%*(a9h(u%KyxPZmxK4ue%xIpdi4xTX#2_h zcPm59eiv*8ZY>pyh?o4cJt|MPv&m)+(c@Yfff~nYKLWgU4v#1a{O9~=hsmC3^G?E# zmmZe9(o4ERWPB_w!)+Apc6_#?L{)a7)^D!KI!g=x^}zl&v?q+yo}j2)4h*<3=yVQI zpC}b({&9yZ>VC(cUGy12Dcy5*NB3Y)_Ln~H0me7-0$rTaF9yO$z8~IDNLT9ZL?&|- zalN_HYBWZnEM>&!RAsFe`n16R^eam7y|j$gzD2xto$(mnCL{L zQs7KKJl-?nb8&B=E8A|I-M%&FM9nL4qX?SEq07>I)0g+JJsbE0G14w;X}z6o%L9t8 zcxE`u+}-BQPfum1YrlRe4Y8CPW3sO5qYl2L%=(>_QE6gxmyMZ5R*kSmuwQ=kQ=3D9 z!caAboN?QnQ-PAfsmu0r^#YWWB>C=do#Qc#8PFN%C8$NAbSHeWU5=W2T?&5m=?~Y( zUNXmaGX-oV@Ax*49O8mc0*BF;*z(vYSVYqrQ~dx_KPE@NrF*60`noQz)uP&%qas@D z>k>ADXeyO&4hkDynOA5cFGwVkx{^1EUMkY_z`0+pEEI4j?GYVNzK)|?&8bS)-I~fE>sW!noVYQ0$X{zvDFU#a(=HaldFQ4Ic8Ew1+VP?8t3;Y&S*X?EQ4tQBLiTSM> zEIh|82~kR%NN=;LV8-07Jnv7Ob=5QF#N*TS=KMrkAXKzBQ`>rPJ@I(-K@EcH#5&sI zX^zI-Vw$ZJ<$km&we%uP3_}4G-3EJ+^fzV&E|qUAQ-0j#d2M|MH;%Gr%C#}u#SbGH z-iYdpTXgv`oRShY-~9V!m~y>twxwgNU5LWXm)5~V#3+)Yiybz(Vcv(=9Hqjsn)@-+ww^*H0m7XN zL!u*HQ-(uL_Kzx5`=`j;;fD#}2DM_99ZxJFB|I*P2Tq)1Z3z08e`D?Zw2ntiLw z(QTtB$B*}|KSV1CVXa<9EobU|vXDE#&!<>$J@CWxEA?C)z-#_}FZd zT1hU3y`Yq&61j31Uyl6K)+ZQ$(AI=F*1wH{-Zp6g9yC2Ih~|&>bQBe_+BV4 zf-V$EGkkI-{>MSq|MKu`QW=hMP2BaE_B&Ca`^CJN#6knnR)ybMM^0GOmi)RPPg{*Z z!=S3f*gnT~)flCdD~-B*EBvvsRd2qCbbeunzwN8NgM1hF`0Q2e3tZLmP!FAu<0z5m zx;x^?pw!kgHQL|rwGK%2S2Ya$Zu;!BEjyZuv>mE0-m-Gz7E2vJjIoD*t@EK>Q{A+J zm6K>sZWa&*Y_Ie)4oLaf_?Y~9?;!VR>PrXwDxV6JD1Tc@t<7i{n9cMVI;MkRCh^a} zampW6`lkaJNqsiNgj9=!%&>P|Z{oa)1_~Ks@Xu){My{l$HXW0&feq$kapy3;2^6od zDP}3C@*2Mm`f1Y?+j*Gbq6Jry@8TOD9UpCSe%dtBsjIaVE}VKifoFkv?B6<|Z}*YB zS&e6ZuU(=@=B;mSzmS+0iqp}M!Dqo^!fx;5YMWJ7vQVN{4s`T}YHyVZksjZ*`(Cw< zX3Yq`?9g+bbr=0=fo4iP-w;3k7DvJq?H%}zOIp*n1@#je+=m?#HYJvScJ0#Ph9DP*8J-&XXn+Z94vn&2&-J2H_-juVMJkO9g4?mxMe;I(Hs+~u+?;#)PU}UR7H941dBAx7HjuTE= zI+tT){yeSDE*1L2)zv!P!Un~B@{{wYdwPCk>@P#d&epwh*Nx%HUl`ZppStXp2MD0R zu8afGml2{Jl$I|hb9UcHoTJ8FgdK&DKGH_5w+>(;>HJmJga^wiWJT9O-*kAr7m?rm zaWoQ_P*X^2H2*V~B4TJgPr2UCo_d#Xn7^EIqLffDcQGO=aSR9?~MH%SOQZEME+4x0$pofxc3@tvFA+nDe#nG&&#FIj&UhU?FxpJ^ zN&VfYp2m-ya)oV?M-!yaaCHRE%|Fyok}z=!Wi82*nnwE%n#_LqK<@joruQhecJ>K( zm4#E?J2~T?M`3Y46{ro#bB+&<>@CdMyy!Cx=@#*9m%CefOPSZi-@J9A#<;c64{Cm(e{%krw0!}ahhGzgww_06 zHl%P-h*pMr27S4Vs22Gs`bdC7{2i=Vpx@Q9{Dc8>qbLL?E|W5LmaXEEbR_BcC?nI9 z2$HZzuJ3KH*sl>;U7sIMPc}_uvGw&mzQosUX_-!3!0-A{D2ung^I#Bd!7xG`_#P~u z@ff1=L{%sRlv1nH9|RlWZ;Q@hpWsF<7`XnhT5g1Kb)xeGX(-$=G8OYNHqb~I*L?d@ z)x*=qUDkplw`l8kGtA5B4OaR}3P$~r7U{LI}l@OgEWbr$vg;ct0##RH<$2$LBnQqR&4v!>%gdXoeMM!O|RpIfb_EXz+OW&IC$;g$U1hLd=F{iV0Wxt&Pa)P zoauDaq%|jB7Tsm7PlZausBJSyVMIk2Gjtd#P0hBXC9l8RhW({ies}-u&5Cj9bk0ze zJ{-^FJ#304D1MMBrqjF9bg_+!y)nVe#KFQ*VO^d4vuP5B3o25BSYi6@OvdQf)R)9< zGEB*ELxyith9>VDOU>i&jACMZ2q(=+YWmfa-6U8x#VkRp*X*5xZ!7h@i+7B81{Bht z#)-MFy(0s=f2mR}<)Ay;G}TFt#@+t0%Umb{{EMs>GCE`QYA0fo1ePvU}=0D>|WLMT8-~wFa2BG9VopU)_OM|^XydSg1;il$# zTLdlFr9ZRJowo}w1YcC>b9&4(pa#D|5CHasc1RN z#=zkv;*sd=9Crn52e_%1SIIQjXs38I(4Ff;u{Cqv70RRq!a_VBRABF@b?fQ9#|reO z@ppSQ)$qQ>Q110eUpEFz|U9SYoy59riB9$egerI)E%{JU9OO$k(2I z7Y#1f=6ZB$2AYG~=48%ApJ11z0I~3w+wSnVL_~Paj3H()>gxOK+@JlP>#v^&^k95i zo1igL%x{aC!HeKDLft8z`S~*!r{)&n1zH@e5OJpCjk0G4N9pnbcgwQtzLc zr&tIn_sCq)Di~k4F`yp&dQ6cN{Q&<;?~>4__K|;jj2oH<1I^(?7Tu4vP3G#gck_%~ z+Ld6fD^cGY!PxaacU@<*(=9*mdIKZ4L@)^Y`!Zt8M)^|9VD=K-5Z6-05!#qc0WU$i2 z(mVMIhh|01W*^C|rl^v!;Qi!N^flLc@8w?4WgzYNl6zcaOh-n{;sLibnX40Q1&(pw z(?CDa5Gm{JThxLWV{o{*M0GW|c^6sfH>BxO?_VKO9=Gy#Fv|YRJ5Emmi*`ku2Twz6 zeGJ4$K4i}e!~58jFxC4M-s5o*48?Eb}S_3W&y<3a$*Ui3_41>16Xn|f^Oabh6FTzwqK=)9G36Jd%F?Vu9NHQbr`&mjTYxAKDCBt*}+Ns zzl(5P>;Kq(Xp%M5@0D~;SxwT>8yM>HrT8Z?tBbAVmWtuTX`b#<7`3j=EdSB2;2X{K z>zA?HZv-;rr;KOXQ1brBwIG*{cZ*LNS8`sx?yAI48PUw=UsYi#FS;rEvZ|rY<*tPb8A?^*de-GbuKB=dZ z@PtgxY?*bnh=o?ADOJv!KVW3hmVMMX z_-u$N>W9$X3T1jSS_9Ydp?|7upZ45YYvsWp>B;_0n1& zp9K!ekk)0RYF7-<9BG`^gJSUQQ>tiR@z%|wRyXZ?wMs_L(Wb%!^^BZFCw(3C!!I882`L$I*(aL{yb^4*+3-45Em_VzoG@D)@z)>0Qioxwec z&yTxi@fStBo)juBy?ln2$}y@_)7RWnNlzO)yNns+nQ8HdVkF%6hl@TXX8S#tRT4fo z72J(H0hif|SpDnY5*aU9XQ!whX_V-NgvZyfY#KU(0ee6L<@RDwAmz_+&0o>?zV^rA zY6aBx!JS1zcYomeCG8z(!F_#dUYZZ)h)~I#Pvt3LB>6M(DkY_D?o+Hu(O~lXdNGpv z?;r5+J>>Br7qipEd1>d=mLAU3(8H(@r0;h2TQd)5&@rnF2f;6@`g00W@M)S1 zg3n0h$FoQb39Y3vPO8+UzPCK?3{!G4g$&_Wh)*T!LUD(r_?zFZ~>S0dbWMf-XdHRi~Xm;S61lXgwh zmIO=E%yqKXb1_oGKM$@c+m9c&=8oF%XZe|MuUQduWP_?# zXg57+r%C9I1b(+_>9-#*x;00s1F`O~XIgr=aC0cYJ(i>cXLP&Z6`?SiQQCN)59A`#Xi;^z3fxqVkP@BKV4vmC%*?PllI8xM2IUP?#YtJ-V_@xKs0Z+#HZa7X*q)gFu1Q%$7H#TJGO<+!fcrOP%rt8_U?Yf#s7&I|rv`f{`Ptce{?o6cK~@%R9j)T>#?g3SjpCG>Ovgt%f8m9T0v9Gfd^NN4?#eqlD+Z^f63cEsr-HV2 z?T}xJbf{kURaMda*_fdi+Q^n{y2CM0z~#GJvrK%|!WByHPE#DOp?M-a z)2FL-YKdz0dp)!LXEc*At@#MpZj^|~T>JP)j*fIuhevC^Jtf=fTo4Cm_;o5QZs!i$ zkni{wA6zSNY&$H=>gt5tA0HHZ0J9{#%+G;b_=rdhD>Tz60ye4u!5#;d8kjXpP?BHJ63 zP210WNp|-^UcoZHN)MJL>N%dE)#;A}&j$}jvfxa7UwDnB$m`sv`>O7c_60qO)mYj3+P$7(XA#`4&>G zdrx5Z0kH6I<XQ&cAO!d+v^FUJCZ>Q*M^N*>VlGr{Y6nSi|B-WH&?E(K2fS5uQFsM~!1i~* z40w19wYW_#NH~G4TS0f>7c|a4$pwR_Fp@fB4{rdJKZB}orMIAS$b!;sK7i#HkRv4C zMggGcC1l>pZlL28G`GLHEm#F9l>f;|(E$VgKv$4*P74wTq1YrTK%33}%D7+*fby>n zH;95lj5MG?DF8rSG^qMK2lxgK-&#~CZ6GBERCX)gKE{MBD8GXayrD+90nf2eu)%Z| zlw%kdq^h!@Tmz7N3#F=nthgvR|5iW>E>r=Q;Q#CT*M%8s8$!I0E&_{?--&!ne*Sd_1?}@vTUxD$Dzi`1c=q zkdNyhUHHH9Am6{aF#nMU`MCdS@gI2*(!Ss2z!Z^e94Iq%kn9Vv-(|SV&BrB-A@S^{w|SuIet`P9WWc=*LVpo6 zrH_UZ@V?(SnSX9v&&QAhs*){HTJkRxy1ma&dw(w;%yhe-d;a;N31`Mo6wdrZt+=J% zuBDiIHJf%N%CfG}{^$JWRmNccV9ZlL|46tORiElFrg?6>_K>SuF-GJKC=s zs)wm@6H0!v4l7B@?6}~PZIpe0S zL;L+htO_Mhs&WXM8CKwI(Ry06oImDJR-CHXAt;tMT(lqbIeZANp7kcBDU19?k<0LJ z&Ct^#^@VnNh!zq zBs@5XlUR8RSl^0ciLiQWn=25MzWShui(QAsU=d5GSx3G!y&a2Xsf07uM@8*rACp4) zr)~AB@W}2R+Bk_HJkg(r)17CzztlLX1*5NxCk}%md7k&%LbK4(%8QEY34_)SX;U4d zjr2z#E_JlIpBIys!YtExs^)#V_M|w29eqE^nwSC~%(l4m68@Mxm*Ll#v(`Ib5!)eO zD7Ozs4Frf$b6U0o#4Nj8>aiYFwzHCprq8w;8@$CKFPqgkhFf23vn}1}wfmtqUiz5t zvba&^U~E_<+L14mK5k+&_*e#hW<`B1Pa%9h>%ZJdmJ^JSL>NUB)AiZ#ZMXn9k1m$& zaRUnPXEfJN&&AYPvPqU~t1a5TI@d~~JjZciLXAl?yYc|NUdm!9Sj+hM?zEHQ@eJI3 z5Ka^K>8?;Y4!rL)WK0!zh+5!cZRTTGy?v>Hu51@$g(iW*Gs@=)eP7lW?485!%}iwm zcDQb=Zne5@_{l^T&$qmBu*`F==(}sQ^)R zAtz8?UeaAt(nv4omB#zwcf^QbDXH5)RxZ&zHqhprZv#*B`8v6AJeQt~GGdgQyb6oumpP9ajr+xyr0ObrO@HSa zcs}>NSia#=e^dd=a-$CJ6vtRK?NuJTq;KbQuWy(93HfvHFvtZTJ-gQlA7RbaWd4|q zGdqro)u_nEU!Wt0i|%CJ$b<;yX{og;iH#sH=i%gozj$iPw@;_ik85A~w$ofi#rcsE zz~>k7(@B|_+hAm#a_ox}pDejg_A3E?D8aijO>^}=wYgM(a&+dm9!<@$+@~S$R_o|5 ztOJl3(kWI)nikE%25Cx)xHDb3z-uG#{T}xMi+!N*bL`#~}iNHgJ68&Q3N5do26%}T4gu-3TS4NvwJsR2|tx={Pc6n5;7;Yj&9oh}h8%_)Eop0byEWrvQVCj`04Tc#VJb>HH*op(qU7L)>Wr?g2QP zlpB3-XKw4Vz*e~JgKws%FvcvvJix!M9!PyZ;i*CLG7pfFd5Ib<#w#cIZdl;G$EvqM z+f3d>p$x@YqqUf)D~SxL{3Dk(d0IU1cOk#DO6#Bb=MzoZfzuFL-=X!nsAg+Yqx;KL zDJBeM6KFwZ_t`-k|GGZu;6CZ%b_W2xkwSp|f~}GxpfwGp#=z(@If+=KI8q<9da5pD z53)g?z=qaA2hGj?f!CfyCI?~~Wk4____giH&vh|jx^T|!;^csl-@#|ik>66#A7Ivf zj;m2!ul-%14X#f4Bv}kx3m%_*7ZgHgCI^Y8 zOpBovKWI#=vLO4_T-7eVbms{3c9d?EY5m}UVyKrV&0d8*lh)nYC{yNDKCp}br6R}g zuRp#QI$mc9vw5q&eK>dfqzq>A8|rsNMVy^A>mKD~lyVa410LQFZPf38GNiD4T;T8b z28ORmemNBjcwm(b_UO#JYR5v43Y%hPhAPomE5I@pExDz*%N(Nk@8`Y}^x{=}st7&B z-?sWt$YHO#=K4|fB~RqVx{lYkuMcsNcBjI)30HLEDuxN7ONZC=QhS(fec zy+mnKH1MPS>@K=#A!l(cslTY9n^~9^R0bpVQDLebTIYTzw^vcESEAFg)S8s4D@ogY z=}uhdSGMt~rC}Gi$@S~~n;Ll3fhi9|9 zdPd0lO~!95DXdx0RLfOCIaF;e$$%2AaYM64Hq)_2eP`#cp^P*jG$fFWwuQrG_q%G- z&8F)Z6JJ>m|J=F^;Ku)}e^+ehA?(6%41mT!@K$EVWZ8&g5?v}k_F{0tk9$>bk7>VVpBw)jMt6rgVqzSjF z8l!)jr=AT(u%P=Tn@(_V$y(kle~YN$S)KL_3nG=jkmwB;AXKp+q-o<_XX;9juCg_I zH+oP)R(Jj_XW&Ue2wHYvOs*2B@9EBFc~fHNMHK#oAhskx<)iUeP;@Gd#jFlZWP*&0 zT@hCN&yg}lH@bKen?h$w2}N)eu;{r;fX!<4kwoZT^b8nc_u3cHXa9jmX7 z>EmlE)Q+SpAoun`+t@OQtn|+Y7WJh?X;NHiF&_uqCOLAgN8OyCnn31ZobmNC084zZ9RU}qs2w&LqiI@~joST&OEHsm*_b~$ABOc+YXgw^L0~l+z`T|8^#pVrd<-CVD@=uW? z?|#G%fI}%al9pnnKM@6f-IV)XyLNiLfDwnfif(^oy^F#5&{npG&s9!4vbpOQ9baz4 z-B-N|?~evt=}hH3pmZpBb#;I=^V^0^ z8A2`xKJ8B;mvwwkXBFoT=3C3H(%coBE@h<^KpvyJwQh#L2aH|iHd|mzo7kE+WUu3{ml9{h7(c53fOhOr;|cqs;GrwqTEltM zBGE2-UP1=ETaONf&VaV7OhGq;uaK>X3AXp&uuz2$S6hGiW8X|G`*LR@np0s5IkC#@ z14P_h_-&2G1k?)>-Avgv#Njc}Gpi%*IB?7id%_-n6(``$T+pO*mRRw*-AFv;*4PvD z)aaf8V+{Hi{Lj4*FGlVsU^fn`a;z#DA4+ESUekeW1x$<_2B->fR^|iE9y7X z`dKe=#55&-F3A4e^#v|-r?5bE1^VN{^Akd}F#{-!4@zb~k%2(SiUvdyI4j#fW`zbA z5R45J%l{YJL9ns@D}u2CD?PyW!P!{ySvq4@uxxnwg?P0LZto}iDV(y?P@4uGtv9-~NzS6|lec5b+> zp<^T3ijKjY&{>QC!%wY%Hlpx*3wMDbCq{1a_{)L z0Fzi76VX&10wuz7c8kVfg%j|@twcdIdp7LfmPjF{ZAOloN<@*PdlIt!Pd0FgTuv$q z{O?$rS1eZ7>;f&bWYW?O*OF+;`F_tBh55f0EUucnrPikasGp1?T;|z{;6hR&!?T9u z&;qrL_6^HP2V zhvO~KglGE{id$Ryh5C<*uyMmhp_c;;`Ar_tV>;ztd^TH0mN7H%|M4JR6dtUgkzEfcMj+ZOBepn5K}yxQ$p>dm;IB?WCrbmQ}4h%=U?2Fuz;DL0#n;N ze0d(}F6ovVfq7y2IXY>~6V2?d+?jxw z4Z1)UN%_0dzRV40j_M!)Gd>msl7oyzI9+g`dqMUMuv6ZF2>&um7shzj8=%g+lJ z>C-KQgq*CK7~b+IM$(T~G5TRB&ZC(PkY05B$C7lEvrNF#i$7CxeT9)2%DH5Bl^4be zjuez<`x!S7!s%EKk39lm_tvp`@LFKcSIn{inr z*DKVQ@>ljiodO6dfFq-Y9`L&F;NYT9US;btyzsZfvtqbwbFkzF7g~FnVfQ_^vPpH6 zt8~fgQ4%gk7sucF(5L`eh~J?^7nXy|@P95PccMO|noQ_VA4|+uJ^VRG!P z9pdG@n2yW0Y92hi;>Sjvl}g5=>Ixdfx>rnO%#a=O{8+807%$c~8CJs(tl zpY+w9#hKj!OhcgVA&sBIaRtO*RSdbB4CqX!@&yK^bNB%bVHVsJl>H5U>t`wETaK?) z;ZuVC5En*GzkxSDLEl=tAP5BDyDb8^@(Q1E)oDge?|ps6eKlkw>8PP_WyBL?vP-sHOeClu-LZ&Fci%0V!H{PUqA$3F7A zz5;xtC`Z-oQUA;ezm}c!*L;Oj*2Zus-@@gg+DC=F)vaSdY_+A!Ll}QHm+gWnFoCY6pQ%hJ%h+t=FHWe<81 zAI?~@$5N5@QLf{O*8EnqwMhY8;MpY+oSOL{)K#=6Q*}-6^)t2cD$!**>XM(4YeVTRq5_dQ>+LHwWGE%O`4d8fNApiui9W18 z7Sa)E1g99FTaLlJ?@5VAxoNCpuV8uy-8j@0R#W@?FZ_5)gMFjA37T+{**BxKt8uxC zQa9`JDhbO*dwVOW2SEnKYg#y0+A&9TXe2-=lT>+-#Rf zhI}tz$j4I*0_>SwPkl+r^L%eQxu3}>j4Cmu^y%0#J&;^hzq~=Tq!i!2UfIfM#Bn+F zpP-~+Mv6M0C--z~8N^f(3DjV9Y z0+)Q|v(awqsPzRfu495Q#S%j15VnjnqBQsUS1ab_m+*Xw5mmcV#10O>xk_c3W9(VL z)QXc9Ao@@KRyBQ^{ta9}q2RS!;}Kq)8s6-$S5{e<&(hv%#u?tSNCob)Wy_gXnR2se zC7I}i{!!h@JzZ(^Hp@mVLDHeVpmh2hb)IxL@AU`!PU{BVEVh!*f`#(kr%`T>YBRz; zslTVTlTnXp%`W~)Yz}RlqWuX0Q8)JhEf+>NJo3Q?Y7y3+N50$q3Ow<2zct#cm&T#W zq~)W(BaZcQbC`4e7kFQ(80c%gv_=!)s6CF6lpC1brQptN@jr@(HKL{OgX8 zy4B2<1sIM7#i-Dkc`s7G?jN;jDO8NDrlmuFv`I~@qzF@G*p3aS_w^tD_^W{fA@dl|6_56p-` z%l#sS_kh3U;Fa6YQFjfkH{w`>(Lmq<&96O=CGfBshFy85^tXU^VMh+%cmlx-LKp3d zM0ifF-)+?=Pc#J=^R3#4v2KV0kYyJb@2q3LbWH4j`@(2f?kTE;t_5zpJge3jw$!Qe zyQycfTR_ZGsNJBKd3aWMK~@NhD~yu*=$I5Tk+rw*n>;G92xs#Tjj*ttlFi4zfvSpd zr{|L2^tN{By_E2rpV?K4>dWlY@O0t%W@f^byJDEr8T^L{ZM{cf&iQ4w0Y;F{MO!cd z?ge{-nTkcaXjtCS$6AjGW|SqmjCB{q+AHzSrNhfB2g^JYJXF5ItIx99BK}b0vwauj zvdjpm(-J?6nvLvH3)$=LoVT=**bXO4J9KU^S$Ee|pQ^=YQ;_I|CGFPxFU4N`Y&@|_ zoicoyjh7-Q`N|JQe4S>00piba8ZoRVB#uJLLyH|9rnnyUM!eg>O+_s`d1f?_WEHPz zQd58a(A7<;sGaKuBMLVnWSTXE;#{|4A4{U_fRgHgeF`GPkZd#Jgrt-9%!}jFIS9mo zWeV&sVr@W=4rh6XntZUX5G3d~YVsO}UG7PmYuS*>9x z-M}IQToBJyldGN5aU&vk=$4~PavLaFt231`%59AIvZx-L;-Ov+%hRjX=mk7Y$4#VL zS08Ze>Jnh-CkE<6ipV@iOyI7T2l8mhnr9a{na969vrdc%@={fb2j!{K1NGcy{0KfFLCP5J@#oz`Er$AeyC@tic; zb0W*8jU;k}|9juW2hNvJVaHZfQ7s0pl#$hMyE?sJDj{kKGXSon#5?UZ;$Ni4lxVb7 zgc+Wxleg34`Lm9t2F8Pz4ST;V8xrMYg>EzIppJZT=BsbmEE8qQMhN1##$MqIb8=vP zgdj!wTe;^>nsA3hrF2^W2NY#a9aLrzPLGN9`V-h^M2ueyZ&5B*sRx@16_+4nS0K1V zH;bnqnUpCisSPmA`*_Vafps@vtZ{vZ^1L7)6VhIbg30-tPjl^iRuof)io5g88!%_n zT4huF6Iz4TGjDQbDkF%DpWpi{de4R3lQYlRF$S6)X>CpTO5sI&Ou+14EbS4oMouUO zGO~45HgBrv{ROVNyD&0z@saL5uQ@q_cXdUQ8Ls%38-Bn>bbs)k>D9Vv%;ISc{Hws0 zbdi!Kf>h?Snm0*vyxdtltjec*_7*Qnv~wiBV3H!1gpfFy@e&u8vP<3{8gY{>6qneV zz-WJC+QYyU#`o`xQRq1Wgjv9)9SN*d2Y%y>V1?yR0q{A7we-;!uC?T{=57&Xf66E8 z#Wp?L&Rf8V;w>Qy-=QXp-;w;DjU$HJgZvfw<8)1WHvD+qY^+qU{hTh%0i{rpF0BVbs5?Uv}1V;)*G7`5e*%D1sF; zo!gWMPS{=T(oxxo3-6Q9rWLnP=#?E~BJCcNLN)=XmnlEUuACY`*f3(gONlgH>9==j zH@!YzNbPURu-Z6HYX9m*8mVgRH$L7zsBPGng^scZ4rYZxv&5dh&B-%5_$xL7Q2v}w z!V2|;r79Bod@A>Xx%T_3!@3gk>?46cb~gIX&E0BH41%F2u9!<$vgZZrxU;?+hfchK z?-igE(kp}G1{ngpnWEK|#SmPTm)t;Mx0V~mhB~7r_=l-6!2^e3-v(;1A+#JZk5iSV ze=3LRI(|O&Tq*(ON&e*NH|Qm=ODEjjT>?Sb8oMr}zYbo`GZ(*HZGTSaYI`=%CL!&1 z)X_I%I|(Jsu>SsH%E@(ce4&ZC5tx1%{fTx9KAyH>SYVbbbV38|x0{1Ow z?;MndNmk)z>i*6?1=}_=BZS)GOAS|YqCRavs^Ji?Y{*74*4ySJJ`r1^`-?|hoJkmv zUl1Tjc`hBjp?Yp^23b(7ECis`qqD!o#of=Y4aR_5C7*h!2TQjFhU$jXYss##oe>cj zH=;YE8Jl|WoLpqjq;m2WZm81to~^lk-#$+;Z&2x*ELk@V&F5+jt8FC+X-qLBDOwIH zM+Cy9^&VaE;(yo~kht^1Ma*wx7+eEzdPnY7Q^3Da|LMiBDjQyasWPUQ$iGc9ZyjOk z7$zUs1jdR@vXaN$sY>mR=4{a}?xgsJwXnko7(!Jg=JmCsnhzd*mt1Ynbz)W57C($U&7VtUnJR6ML8p z>dV%vJqn?5fLIReYN7Ts3U6=N0F_&!qCnwGi6XzYkQEfOE?5g4YQ9mF)7nLEgsSe| z^PRs`y=;|Y$mVJ>HdP=`BujvDQ>tof`rj5shj48%*wHta65FS5q`CL9DSaUii|NHp zM|!bbi}6eA0k^f)0|aql7EoFMv+xjAR_Uf_le-<=M8@c8L@j63}! zM$--3fk(hw(;yM`tfQf>k2!R6i9a=Rs6c8MjJ&7%pf~uZXkFW!1PVGQsIP)Z-TiKB!SE0n zSqa{ZN2$BF5&ZMQhuA9sF*|Yx`zT;~Vvqk0Z1i4hG@E|24dRkFrgo`fFTu{TSVUi4 z>k!H&O;DPpQY*g{S|MLERRI-@Fa8`PBOQ%LnNGSz$1%oX!>{QOmkgV~?>C<{4$O&G zVyHjzJtc>k<=DQmU=)4Z>+kTgdM*9%Euc@4u$x1r6WRm=>l+0*)w>yx{fTjK;pBNp zmpr%q?t^+yvrgpGCPa`z>o;?hK_(>SEa=fX2}e7^;r$F)FloN18*n##=`1d#4Do7` z5-Fi0V_y#o3&b-t4QMm>28rP(NvQT%I7k+Ag4f?m2-p_^ix4Z@`>jXUtX{N^ zL7nuN4tkvi{{ju%nYTn%7z1;Lq~bOISmeD6hF95MuJ3H%kS%8-J$grZE+q#9ufHX$ zZbCRfd^@LD7(O%KbvK5#Z*FfU8%PvKS+c#HJ(=d0ePHkg%tWYEEVQAK$$M97@su9D z;&ny^+wpj3rU1LklsCbb4^8yN}^fe%qI@0RU@;E z!5-K7$GmC_{N~*rqpM6@&dJyLMp}KU$uMJXzTnp$z~(ChiWCzzZ2nW^O!QamWvrSp z#71YD&n9q>G-gMzJP}XEs0T?6!>jeyTfS+ z+Hf=j3@@KE2fR5YxwY3T>>5AHKR-8?P-VQ10J1PX+yJUm1GAU=h1*;*A)&w-3t%LJ zFG@?}Stw<3*Q--(9-|SQb^EA_V@!gx*ueuq)h$>*_GCL%S)2mt8(tnt&kW-SVI3Ie z+11doveDfHBU(>_!xryJ5i6Pk7VT(%^yCM6(A`FY3s0>~^&khO0tK@Zn8K&S5I#>9 z;8gyP0pQjVX3+jGM7CLPy!msQK+(WMKMVLc8qb)D^~VcJo$$cDUWAK_1^%4Q&=Lu1 zGuRK5S@FC50mOnw+4Uuxz&u?$ByvWuk&Esuuu>`pRiaoja;ib$hh zMH;vee>EYlNO*W2_iPka5n|Ys{VdL&1K2xRCx23&S~{P|!Yfuc$-7_a8DJAwvfZED z*dk!(+)=j0Q{%rmdMo2-a6LIBZEntx=iyB*$G8v zeq|1uc6!KQ9;9h}SIZ5X-HG3KeS-VtY79B>DI%4nZ8$>+hx(zFOdVyh3fj1v2*7w{ zkCpnEFtkKCp)S3}k-xlHw@hcOaN^qQBNm^;Y#v{f5pIXc{Ymy2ST==ySSlE>&=sEd#v zNhB)v9XU5lsFtmTw^MJLD%Ot0JD_3td2iv-RKl+6qSLD^KG6K!XVY0bjhZE$b9PqMm)Ua4YuN3^08(iH_ESuv$r8MIfm-|M+wJ`U5koo_ zu^zZsOSOeBeS`7kO@QWA6}-$F<7*;lSgEb=A3?IVo&x2LT^}6;n&3A`Q_JXw zxX!^Pa#&?@mtB)(6^Fa3GNWULKPQ!FZl`}gTC^=?qDF~ExU>N zJid9<9d@Q~?EAg_^?qx9+aM$q7@ecPY64tU0r(@OFVkg*wkGhyBeIdTe#7t7GY`T$*>DnzLxgxbHzJXPR1mN-p*!|&4HR8YS^>6 z!yYWX^O)bB_8GOWV(v=U46SfWg*dAg&=Iv{+I#dWt}tL@1?zfV_nX-;*ZFh{2`g zt?~*skS%r7D@GctLM8J>vkl1lSWMK|i|X5jJ)p_-*T)kA!qhj^m-@^_&K1xCqlB8x zCE+BmJ|740dkgG>;|}6VvHI5RF#SeZbcv z$V8FFIz%TQvzpOaqx5yMA zO8Kt*T$u-Vro3x$O#I#B#Oz~Sxypy7Ji;1j^qH-;GQ_@OzmaVQZ4^Xu*p%U;Al3|` zf5bS~(%<3))FRHu`C9kRNQGYm^F=Z*;`P)!63%>$@z*9tbyUyo)-C5e>?wwHZL{T2 za|5*mo!QR};^%XddMrt`ZChjRwf2r8*Hb=B?Yr_lp4u|6OZKN&OK#dwZzD`YEYpAR z!S2f1)$l8$S?I|Fv;nqBL_*x>vajhE&H2qa8Y9;9I{`$E{EyN0O;4b^T|_1CH6mB- z0sp+ks(Gh-wbLA)U`M~7ZkjAi*d=2GE>19BIinVRFSKRwiqm7Ul?B%LW#@;+9eXLv=u4T*I)YNSgYW&N^mD`_dw^a0*j1J=jQ_O)RF2T(qB)ha|GE6k1nJK5HJd|CTuLTpS zN-&b-=`s*DM5{&vuTZT-~Gz<%N4Yk&Sld?8e{)rMkf^I$E1B-R8H|#@RF|0 ztgq}De{d{cB?HVK|5|x!d4%7V)+pz>Nc()A_9@Q0g0 zH*iqf>BN$4l6FM;d8RE%kfPI2iHDHwcEoNaPEF2|mSFk@*?P6`2v0kDUl?QN2XA?b zti?mk%^*_m*g+a%oao*2W2OCfzR-emGgHs~^d6nt2?W5ah5`!}&k`k!%_>ip7E(Vp z0t{+tU^F0~%ywD(==m0>3Yue5V4pOGBZB(qz5fFoz6U=IDk*V3)wKd)-nm12pk`Q` z@W7T^%N!3oERYGstLeQ=4m^G9@@Y%JV)AfMN%+@Z3imO ztPZnvBYV3hO!IJk(+UjbsAnNoBSyv4K;xL5SOs*a#JT~efV{UH~u!O+Z?IW zfI6qEARv+4(fC-PXO;qjzYl5_Tg>mEKOuAF&v>n92{rv@k%Hg$zjcgD@=*h^4p&HH zWz%>02l@mi74#MROMeDYK0{Te1XIh6Z9teC(D&a*zYCbRx1n%_p?DP%2TMIwf1ok^ zl#>I*(K4Uj6fQ!ihrf;48!!e*GpY>R!T{aPop_3(UsF4D7JFEKh4(huMtq&M_B85G zhK`NQj;v_8N40=nud$4}vZg=AVEfEyMy^#kpY$!F#bg=3;X9&BKxQyMCM2?BT%c&w zM(gZMj?POs8WeRH`V>qP*{^PSZb#`C?d1V1C2}mZfWPC%H{efwXNQUwXkK|Mjyux+ zq(#ktGFT>z6(;uCyX=cKD+`Y|P&btVBl0xq=@$M>#PNz(Kbd_yJ%fTRC4Z>?(CsWC z_E$9(G*!e_35(~5k}t8-T%#2kF^WPL3b3_a<|M#h3>k=Gehb?g39A<8IyWONPtyg= zrPiC$FCNCGNs}F!v$*e-!Qw#rN1zpPr81KwoxrkYT#?!fBTAruGMxCx zhT_M39u%EGwcELWD%56?4PT!uwm{8?w~y3B4QCx>nUuPq_S5lZ^MIB~-Gw~>WqcqW zbmLA)2mT7Pa7dS2xSxfuGP|rvK6GS#CDj-|d&(mEjW47<;#O8|#f;`VG8Yz|z`ZiB z{SQn{%RwiahNOXMm7K<+xppETgC11AP#6TS-bj$CEX!jCi5DD4RROL+#_Vz(_%Q}n zou^Z9i>;YX1yiTaB^0|`LUIm(3b#OJ#-{_5TZeAyU_&Y_k>fjRmKRvdv|I(SFe$cS ziAK9Di8(uHRXHKm&H7_t>OyrS=MYyT9M>c1k#pfC5o<1ifQ3>USef>Y`YCOT#AYnF z)PGTK;I?mp_~4>Ct?eDE9irOx)~>ep$5^(avT!Nl+f~6F{>1}%$(#=$L2_o=A@SbU z)Y%l5e4e5u`BzdoI0KJZ*$B;HI&mrXd93aS#()5Vg$bgDp}EI=Eh#xUy3A5dhq=fl zZ&|KB&xifI`FL~oiFb7f!v}mcJdA@?)W?ig5N!sdHAV3P;y|$Ddp^fjEU7qG`8DuV zniZiBtN5(cjnOrlCA|;8qqa{mBHe#nvh;Iv%!J<;dOy1DyxS)r;S;DplRhRuX(dp%1p|BMKV7ELfboPNKS+?o899 zR6-*J9X3=C4okL>{7dkK!V$gPMnxjkH@`|J|HicJ-QAZ=->Qq7iy4z3Wn#$g&mR=A5P2X#+ z!V0d+@LfvtO9M~S8jCV6x3H~80uI^!6czPl;D)g?ta2~1UJ4PJ?Vd>h9xdoVg4RMeZhyma)g5_fT_2>bYHw zV4EV(r#Sk6Va=EtBv6Yn`N3kSq{_2tx3M3W6Jk_2KYco=MvGb*=E<&|+}Mi}98HMF z&9&iFi%(+xdMMAF$L+7fn)*(k$ z7u+RNu5JvTV@4E|%8w)R@E!TqZeW4m5#5h@0N^5YREFTp|n-XSz9GG$`Z6yi?Odha84Z6l`-!D>vO@&~? zb~2*_44e5hGC@f8va~LR}~t?fK8R@7o%EV9w?5vJF>k zePI)nP}i;SLN#b+=jn@v*a{G@T-)=(9x2gBCFdg*T?f#R%fEc;u?w^ntCTHOF%Quo z?7};o7mS`PI1wsRk(VDn9sg-qwkFLcei&fqn8gF`V-CEz5MHc|n!Vnu?b}CCvfB6I;~- z!LP=?N6}L-w{X;ae8b!p=Tc(nqXB|k7{}5O%_)!7PtN{o64H@zYw+Qk{)>4qKojol zk4s~ZR0(X^>_`g6))xK5P6&fwHXkQSVO2KD_Cc}kXfB$90dEY8-hI!~*sL-WZ4?f9 zd5{~yF1z)Ke^^Cdp?m}q60)?CoT~FP0PR*z@zoD2b4-hOf4;Og;VIT?GSgjC)NNMz z*Td@3#nr<2ZckbzVeB?GL;dI&fX}j_vGwRU5m#(M>i2E}gv=c#BZIvQHEfr{sndxmkGAR-dGIlh(yX+A_KF11;~yl^t8@LR9n3Is?CSfqKssLpSOLJU-F6kxde*@mw!>S zS=KQ>7Na+-^Xz$cZGL^+X`P`HOIl(|YRegT4#xsjY}YzwT$*nlFkomm@}?1lH;k?p z<^@Y)eVc>gdf2C)(3Yf-sNs^U(YrZbX z=hK55c`{@ z(4EqAckN-wGol>4bHPx^v*H?qtR7&Mzec%s!?Qb7YL~dj!5D&YP#NGRUF7~C_tii4 zQ9X`JtJp2sEytIagViitjwwEv!IA=v>>l#J00$(f|2y54nUndy5zd5OTQqqa5~#oQ9fWFfG3P&xKROi;B37$XEW%H$!OG__Tq8J#k+SDsxC_0ge-M9WeS!~IDOm$Ltfv)7Zff$>e ze|5J$ivGU5tr-Tif4n&EOr%wSa$o^(H^o<5Yp%lw7egP0oB-4HKAf))Qy&m2{?P|} z8|Ipym?9qy^PCS?C~LZgTdP$Zi#T5OoM4vIG{|0Zc4rs3dn;z_o`- zd$k||8RY3>FhJ3;UBY|4+gf$zPshu>4UXORG)RAyHOv!;{O2vfzdz31bg!X!c-ej1 z^;h=n)ogc><^e@gfVG8#Z9wS zlV3hgkCW19B#6?)$a&o`9|MdZ=um=NwSHS0bBpDc&wU~G0u=Ld&3Qu4G^_tSsX#N{ z4Dwchc!u?2XEMv@55`YhlFa+XBAo9Z8@`%_%=FtZ2Ayw=1ULBsen^TQ`GnFiTY$xo z(^kl|$Q+jI$L$KZ+m-FT9E6eWV#OuV1 zmF&3$9}tNo0@iX4hRkX1aR?bS zO37QImKfJzFH6aSC?7A0E8>x#h0#*ovp6~q(>$9H+ZAl$_s)md>O?43P(TAyHeUNi zZG_VPvOva`;xgL&=O9>0J(!%20`-j;B`rGH9BQRGZuvbG>Qy}%u*+_ zaCXV9ibAu`4$lguN*M{-MzZpqlq@}Zu~wYKrvOUw-1r)d?-c%T=h0G@lt+Pk%kYK1 zPjz01lpeNJxUz@ihtL#$;+S%k_`}iREtmqptx< z-nx>7+>_mbF~Qs61j0C0s@Ehrf=Kpn{)0Iaaw}%QxN!)~JIok_vqM+rFG2W-{AB7S zzPZDtAp&ZyC1i`p7(DvY!E3``;A_Yj0xutO>f6c#OQZNw{9U$l3`JZcv*D|@4jutr z{kH_nvWvEp!vP-d8(nQyj&v}6wBcWudL&`}bS;e366aNEOP8{vaZ#q%VRXd{*|*l9 ztxT{W1m=GNPsyk2;rx=p$isY$-|HbzvJe()iuQ#_``7%ay<+vb4+GQ^Aqgf(NBYlD z9B4YbDfz5sWXvYDN7K4UFg;u)CBOhXvqF^^Q*-;pIZ8~l_a>N-ECDDf`IzrbviV6A zcDC0_?J0#eX$X9dVwrr34)>FBWSLUqdE+q19*8l!O@3&Uds?Qu{bVb32r&4QoifeW zlng`kLu%{Ss2Yq+S)xs3CYijpB8NC(!!rGquAye@i$u`hWmZ7r8c60TGK~;F;0pA>QZ#;v0QFF0E17a-U zH^X$hpSd!swSQ=+{m58)CdY0>_SRxV1t-{ojJzg=n^}fYWr)dJZJH_Rvo8AP*`<{| z2|sD5XT^hSlNvHYB=%)s^H(e2&EIIx^TxPz?*4Y;=aK9&RDau=6=9O+n&bh4zcst6NOju_ntZWqtTlS3^i;VQ&yL^RZb zNGP(?_QB_ut1X$Y=RhNjQOmQb6=mmKqf3XZ%^?kMh!O*2D%`y~-YL98A;VknIp@Vg zjzPZbJWeI>imAib3Gf4G*|vIm%0p#%qZ<5%ZK8Gh!$iw(9J{?NTFAw!fG;?S>oqPR z2TPn{SRY%9K%0SLu=fpz!78BbZ0%H=q^0I;e?vrL$ZeA5X6{lRr6ZW#H`>+B%SFwP z3lnFuakw)>U0?+w5w(4MZkR^&XC3eX=u6xRL#+lk|dpRbqX#cL2Xj%-PU~@bUbxT8h8|h zf#zWL8GFo-*!a}z;DkZcVOA*&y4aZXm(oU|PO|nOsG5kkAlBl2Oi+&s-%aH-w6R@)TqMf?P2XC zMPO-;kPQ#Rj|tsYM)OtT>Gms7QJIuKoHH~=A=P&S{u20nz~#(|cbG)QWYmjiLFaFG{(zcc41rASU*9 zfouQq=VGFh-0a)&AAyr;LrpHV8o4(|4-+@or;>2J=kHdN)W$~hio8S*I_9rj5q_}b zq;!kUvj$1J0CE*9>{~YGyzQMf*~dgN);v-?_@;o^Q;703Sfv8D59R`~ zyK=C>Az;I1boyPwEE^iBGFb3}szKJZx}R}#$Po6EzL{Q~j7dhX^DWoHj@e}RMH_v{ z#`^zIbxw_823i(vY};yV+qP}n<`>&;Y^!k^HCAKWww+Gz%*)*0aGuWDXRig$Bn=qI zQ97fr%`z6?r8Y7o&CHX`wojw1N_kgcN6)z_7U>!P`2KAuD)!YHe4S7YP0~MT>$M+T;Xoa^%hgm_g+0@*3c@2+RIMdgKNcG z(Ul(uQre~6+|Xx!q3<#0J0u?e#|`53a!zKtsYIA6-&eNP>xorpj_3fqrCm-{72@*x zyzmdLV)D( z{n?}xNxc-#v9Kd?GJM3ahqkig(RY(#Av_-^y9%8{_9~yzP-N$6Ln<|ou8X*S-wW4` z)WEbMoYuXR2mYhoreXHm3Jpz|?b_;gK6ZASW5L7f1n-vV`{Xwr4Og!<<3q_a*_!Ty z$gCpY^_YHoUEvhUm*Y#4pA5jhkgz%q?@%U}f?=XjEG2I(_gy8J@tU@KM7QtR!-S7` zQ}b}2C}&shLJXApy!9msM>dFpAEvI1Ja1Qlgk`iO#D}`BArVf``auyMtQ)&d!<7Gu zgwA{7HuNnA)e=9osk=m<0}%pZ7i1@MU7`&J_Ek^Bx6#GLTm56g5^q4r^0`#;M7k7! z9d6P{LxEu|`QZf8R>-u`%PJv%RidmDfU{ClS9oSa1uV2M_a zRNh|GmY-?7S{sCSZNQcsg{8=v(0p6OaeG=F?mtU-^5_zntuQ^lVJVL@_YJz?X$tHZ zI5qUqjrUI!H`f;rLBE<8d(v!5Z(u^Zw7s>2Aky>>Mnf9i1@zVfD)}L%TyXG*r|oxC zLBrvgEh`tev6xVQbt%PYQ3=fIe{*$s@pOSeSi2+V)Cfn%n!r^-JyLiD{i;+L_)u1F zr80!=`&Lcq8D71>OMCnDy$#t|cBy}q{$#7esD{+`E?%A()aF#37B|j`5}8)v-u+18 z+|tyENhdoL53r6Ow*J0cu$(D?-QM6zN91jWPkTO&>Ca=`7C^OeGb1ytjv~?PkRM~2 zOvp}K7Pi-3p?x&t!JON9MY8DHTdEc@TG%cN*4IlYpi$UUbZVNmmnwvB15apCizmtV z1o6C~M1=a#!JuItC_6u+n`*nmcN`6+l_B@zARwzQ9LN*z>(XRXwQUFmdFJ+pdlz3= zNtC`*#AUIbEZ(_coO8jQ)Rr)y;`xh8ai$QZsonN-2c%r`UjCXjE6lKht`yMY9%J}J z!zUWzkg5Lt{>(M0vh;GX?caYsz1BI)GZ?=?As9D22YCNAiw?dO}i1rds12~fnrlSF$-afTj=<~v-w(XpR7#Qg$+ z6vUQQJhtfz>3;oBBQ4I(-^9Nc!jg4W=-JjHIQO;5TUM$!Mc`b(GEvH@_&~>cg+HyZ z*jfL~9N+1t@;4ASUl{#N_t%N6`n}eV^}f?O_n(JG%9JduG8=J?a@6-DlZDm4GeZy# z6M$91S*<^e5%Z720AcNNydg~~ASi7qixfJwW$v{mnvo0pjDK^y-r$FIrJhLVU3Af7 z2&M;XXK^{nm|Z)?k+6s9J4IL{HbS8P%fE_oMF4$Gg&G-2CrbbnCp2Nkl*%J)6Q#>H zbZNH_Ok33|SoOK`ggb0@sD}RUH2V0vAP_C`SOlZ8+?nPhnG%&>gt4tsv4L+*FbVrO z7D4|BcAkh+>0N%7T*X%K%L6vSyc|O-2ze5U(+QoBNmZk>qi@9*M%c)mHfhd=CqqEJ z_vjPc{sj$gi8|gjVf&JBaI5^ zuTcZ;Gi3)73Cl5>5#$<&X4{ZI-!h&dSb{_!BJxytA*l;h^OxJs7CyTgPs0oG@^4^r zv_zf4MgM=;j(9_j?Xcz!X0C27884w=@c+z?NhtrUB&?kO{}gekJLbUJfzoqK{Q!DJ z62oisf=rA?-1hA+{iyJp`C+Q))?d^{%*K_N4Ta#SxF8a(vpR|xGkQaM|JUUq1G&$p z4Dbx}iDXyl-Mjnz$2GyTj45EY>_ZZhQJ%hEy#EC;-4ZhxOC`_-nR7mqe(R63_s~5z8^YOV`Wk?I?;}t|(r(#a9{E$Bp z>E25z3$e}ZjmNR`hdk-5{`={V`(Bj~gM`be|APb0-ezC1`P-j?+~JXH?F52zMng-_ zxRAHo@L?;mk+)6tD}L97o`Gqg4a<<_K1faIqs@6e5B{t*3_(y z!toaOHDSS;@*(yQaq_A#d-0C(4wBBJeY!E?^M{$~QAhJb#UywmN>n{TtBoGxZJQ7= z35Pijvg1gW9MR}XZ8pGdWnDA2;{Im^w zAI2QSuW-|rf1RN2xDhtlls7jSI@W(PPzpeZ`1a$$nngXdaSUkNr+DHw1 zvemjXds41j)qELov$MT8#hM4F-LQQ>MS93ETD91I$1}U5{CP^DD;j}B61|@UU5Wp+ zDc%GtdmfXjk4AIq44JmmO~a5{ihboik>l@15a6 zKE|Bc>qQz7z3p~**mq<15|wOuNrdwbZHNxhogQ)M;avs?t72)you_#Zv0I%zYEtaQ zGgU1{uBr9cT*XE_VzJ!FJZ{pa3ghCXF+<2x_Fi$YxO8Wo0w$qoF=)bCsS94xT_HZ!rI!Ktua!;?1-XsBk1Z@wGWc z9h7q;X~?e2VTJ4@3fq&0U|r$sF_7UJI;hz?dZSZDchOE>4i^9A^cI)cl>ty~n(gU)hy;FOF*T zS|;!U4jXR77oiMo))iv=9;4L89sIQgy!o@R5(y*i4ZmdnsyCBP3rLxCJ|#IT*g#+# z@^=9A!cJeE1zv8j3Z`%2{6TFghyiJ%MoqXi%R0$-gjwQQ@y!nBJ0*Ww8)Q$gW8==- zNhwJ$*ztP)c8OV&BB={FlgLX(h7b+v?(|6#`Xd2tofr%XodCOPy|E=@*5`S*jG%>cP(kM zdWfK11tFseskHRm;+pih3rsw1ff^hIO%x84E;AQ?DnDYR#~ehr_-pPe&{G?VxO>a3 z>LMui1jQ)BYnQ6%x=P;Z_;3s5aFp24GbfA(wnB?KJIZx9lq|Car|$C7n3_d!wwskn7zhl|LhrVzi`#AMFiacqp?si!rfejl zWuum`N^wBkUP2)6g@x@VHe3GE8}G{RAcWxH-4_;y9+2QbkNvieFk5P4+u>Vse|U?wnY2trN5ZLt*m>xP)%g|{i?$_R1QZR+&+UE~6omub%Xy3~ z0}xUC(a%B7vuLLLkhDxWL%;c^{M91{e+WizW}azZrDb$F&qLc9b%uhlhdX1gFznJO z!cAlGTI%JQ2iguMsK+Xn1Xfy6AIeQvl*V=)!*6f zQo+$vVorUiyH1339^@$VrA$|E3asZTWMZekYxvv6W+VyQ%q5{sE(LnM|qox%5W182B3{|dAPI|>zH5vg=W3E5IbNEJ{rG2-i4J6 z?)Ivp`C!I--3YxHAFQ6a?A?bs7R2e9wKJNX6Y-%n29whc@!X-a^%^Esm5iTV`lR;j zU7=+bHR75_VQ;WGm&k#p{Xl2KE1F*0UwW(dA_|g=&Dbi2uNahSW5pWYRTsT7&E)$D ze;W1TXjlX71(mNJ*sy1*S%-A1Sis0spBIoHO4z&>>6U6|({g<GWC>-eT2lpVYQH(by&w<#_`WUMU-aXute_frs!+(i5H^&MDLDIs z(4&Qc_wt@1XF&HzwN{wJgsav^-%#S?v0V_bj)o;<%G@ewE-j)%48<7+x!WE#=h{C# za3i^CqJ^O?gY^B80ymD=Lsnq};Z!1wvftnBX;L)k=hLT%Kzm7QfvB zpYAVFFJ+uSnKT7$(5o^%ZP05(mpdA@RW~c;Li024TQxOc~rrT!bDAV?0s;Nn46;*!-4}c~obrjH~3fO*HG!>1kA-XL%$MsElmAANgQ6!CU0(t~-9xy~R;d z4_EPv4VE7m5gpdMIJ&WNN>L8|)zMw+ezdEUs6Q1GhQFI* zUh2QRJD=t(CoYiax?z@BvZ$w2V*L8lHcC2%0x(=~DE`O2`o}={| z+CKdeLEJw8(rR6tSF3C;ncs4}hbFqq{#gtyzcT>HWLFc>PCY>Xl5%*IIAzYd4J&R^!oIK%Jv>v$1I+ovMPeAM{23OYbxb-M%+8i3vN|^xSI%c+X z(`o^RqLX(kI~jPK#iNY)Bc7{lLt8hO7{zU*^hd3WRv$_78Lj-$m++i)70?HRwRNi& zY=}eUu@o$s?;z!AQU;b}j&5ZIl*+jd$M~f@Ui=QN1$L!fB_t|jdY?pVLiZWy=w)Ph zwZkE^jsYr4RfhUBzLiwM+U|tVZpC8l{DA;1anSn*;&*U@`T?m6+d>!(^3up7>so}3 z)FagnFm41;M2lUJz3 z6I=!qYJit}c0PP4T@Nc32N1X_w{V>iBj($NVy06Q@Efe4>B8!jX>;;GL9Bque(*S) zu?PCWCZ4!O*X`oN`RLXBw1%X}l(PalZ4L=8juEm@BYan)z&PsrvUp1CdnLRl{5haK z?=UfS55^L9TKM%(%(oOzM8ZpHGQn!$H0|4bN9F5vrF6u4uY;z3xu;Ctd49|dFN}VB z=O6kYef-#K88?{9g}Q9x9ZKkA31njj(P9rxSBVu2w-ezXdlO@upfIJ z`gO$V=3wysvL1Nh#0}oN|4&$qmT?*Nk8e&V362M_bN&}d{8D$q;h#Bj^O*Jx${|kd z3%{V)HYSgHXR z}~UnmOlISC&=#?JZk;R&Swe1;*@m39d_H@-S*|PO4^*|=R99k z(yj?iUIl~y#QS?#FCTsj$OnGW#J^3@d8K?kN?-2Uccg4feLE#I`Y<$j$>09#@zc8x zet*qvCWv%LF(@WVx7gdaCx!U?_tKK1v+;r~Z zba5%jkk~&^a%Dap-eW#Q)#s-=wNpMAB@X`Gldf`~bRKQ5f{*Ib8n0%l9YM-U-->If z0E9KVp?df4PCmzU5<=j6(xw{T1sQN>3nULw$#SduwkMG;MP>c`360Hq4L9tP2H&fp zT0~n<^T$G=c8^l=Pdw?(BXUKRw{ucv)jF8MrgDbV)f?WJRGn*Psp)G=HK9wC+$F-KR%hb-`@gd!yNLmGYrSDWiv=sS$a(ZtIth@5noXiJ7 zi^QRtX1`rm2t=nB-0YL)3(;-Nk9BTBq7%T(5T{P@edr>?t`_7FD`Z;~dP=;fibl_K zdMtY}4Y6okyb-v`(@}d%?RFtzkuc$8uMG(COa?ddShl$Qp>2u1;`P|`~_xB~tlX#I(afM&9TmLA0Ra=#{P z?6NtGZ!vA>q>t8(GWkp<&R!8GG393YYAib}=d9BiIf1rf#P(gmgbShrQ#E8z^sF9I zhjXAJ=d<_5yhNqrLYNvW7<%9f*)drfM?QrG+l|srTYcVM4TTTr3-$Wo>Z~QCy_a#7 z;)oi%JOgMg2KM=^IU?2~#qnZB9tMiBI&HE__CpS63wP6Z!?eg8Md4y7>-_1v14Ga* zY4g~I@WTbd&U^O4N;74}YCX|uK(InMj$67p7dzR>@v6M|Rm5NY#W6(n8n7FeEkblK zAgoNF$3yNhLg*-IJ0gUr7*UKuPv`@mv3cp~ZweCcK?tGfw{Jm#gw^>ffShcz z@r{!vj>o_NO7_1i-e~LJwxnVCK@?w|f9kOGkn#q1qTpX{c$U<6%GshA7NFNbu=m7F zCT$f>UYv2?{gtQhVlVaC%A#;I7I0<*qm8idpbc3ji2H(ltBeZQ6-$WXB4gQu4O%g; zCc#WZ+G9~ZpH+~C@|L)mwU-LO%(e#-w{HXLh>8G_=(dKu2c0R1#OjsbX z9G1a)Z-;7j%9?p~#4X3t@(@(+W9-iEfH~B4kWXl(iGlc6sgV4`I0Tdb7y~hrp_GC8 zi^y#WLdZ4@0;l6zN*ZqE;A%`M?5|8CWb@$Qg>||s9xxO6Z_MipuA&|&$t)MvoffZ% zEy#WyG*q5eBsA0(7>@BaM4w!3CCLCpWpmT(afWL*sJ~8~=7u-avbx~eCw=?X`V;8W znDOmTyZW_Pr|g_wtvomxB*@)8qnGV(A_-msPuTYR3r&QJR#FVU8*$&ylGwz`zcv}a z&9;lIo(Zx2L=0~+6Ct4SzgV|@a^~sn!THrPGBMajUrbkQW7k8lIo& zR93SZDZPeUr&@*6Oq9wRH;fhf3LmN?H%1<1W!cO9-j%0Tgs)`>{T9FggL@$JfEhUd zRnI?_`pr(S2#Kk=AgPd!1mX#RT}|m16UB04@RIQ#YW&^vwe4^7mCOm>c7@O>iJsrK z7=){c_uZrZG?4KpGCof!*gti9(L6Yq`C#k}k2uv2<9HqiiuyqqXY zHvBw$i+pg21ny39+P^-2jr$6$;;6n6KhyhsUksU-*qjtQ!Q+Aj^3!htC`*x9U!Kj8 zUHi(%Lgi5zKA~ZIm^tITG3hOUz4VXZd563x64Q-YvA^RR-!V7dxW&(%7QD3k3k>fK z4Yf0llPXQwz;RkROVWSvve(iTTcwh!yN*du0#mLuNjDR*M@L2TLR&-$`d|N$n40<2 z<6^5DR}~rr0{5f5Lxi1x0E(Pe&9t|I@+2 zGL&5J#N<<8W!+S}_CJl9S2&6YXQq6ULWwdFW%)Wtd~t2+0e*zsZ}QMo*KODdnFELU z)tM2o{41#8WIM0lOrwfLPCf9+To;pke1xU22*HpwG28xm*RDE@~;QsaN zX$-=`PSFVceXLX*k!w=+>W^6Ia74%w?iryX{dZt5ZLHgKb2JWJDWb1|QBX59AUCZn zjI({GQ8P%!p!?m6$|wV`+{#Sk{;{h>7>wuy0|PV=etrR~CvI*H=;|0x%{q`iW{3jnkTNO)V@Jfh*kC%`@_WT-)Y|9jYI1UDG@8j^v=)bnayU=Q} zk+*&}1aR1yXTfhM7s3ZpvWD%|1Sopl)n z#M`Y}#Podtk8Qq*)sr;&hAOSP)J(LI02jW{k8Z7xVO$O?9W4BcGej8NPV66V@OpG8 z;=`4VNPe|#A*&HY1M24kl#8B*kERBsVk_jD^)>HgkC|EyAo-xm*PICQwz}=-H;bS- z`UKCwRo~iJS!R)nBEjGIByOs-o%x@+IWPRSF;ffxp}t#maMXLV1T9^f)?Bos#!sl+ zU1;%3UFXW#?hpdZV}XcG2?qG0jL8d-Du3(0HL^|;D4FOBppCK~SF!8Z6&U zYEmgYC4_ocq|pMO`vOeMKjYSntP5ivy?>NfZAKJU4vg@{LrDMN4bT{TF|N65H*>2( zjKZ`8>QU=>bJw}FxhJiQbXAs2QyPKXmm%RdSUY%4fz|J%ro6z0@YbVPF_CF))vT@K7AlyQ zZL7ezk~`J3l?PQ>8gqC7_47Idt7+HKj@*i}eX|CZcPmHm^cxf@tw!eC5B5+i|9o}k zZPp}wrz$m!sOv~RvI=QYV8uOMb}k{zPBV&xVX9o-jLQ4t0?|=RRm?` z`cIa}&cX8kv%E{aiTLw&lz(gD2UsmijAOyBc%WS11?U6AXupl&|80qlgl5X=X1uOw zT*6FD4zi>%^*oP9IbY5!n5zElLNBI4P_U`ic;ovOGMzRjS098z7FiFVFpxuN?DO z&bUHHSy&58cr5dJ(`X@5~kuh93f#p*`Nswaw!0Pl@JghM8okNv350?fP4RWK-K#{g;`7k|oLs6lWgq zRLWO6S?xCe4D=%&=m09$8ImNTZgtU6|6Efg>C9g z7sv3*cZ+snFmA=_dfMpzvs+g^mm#>9HExN`)rHEDfEpmuvGZ!b=3y+2!L1I%png60 zQs1J=?e`pIuoK{qMzjn2PQ2upC1ZLjPV}p+9mH%8!Z66ni8(x*$KZ-o4E4Gwj>V`H zS>Jxb%zU_8^mF27Lrf;=fP3(+U2g|xzm=uXzBxhhy|LPVYuhw+K)GP%P3%kGYyXR^ zKmrorRRWf^jd_=a^?@s>&y!Ep0tKd_M7fy)53mZH3{3pug}XBv)q)-}wUFlfDuQT} z;`PF>%ZCr{zVb4R)E-KmnxDL^QA+J4#-M^?-BB*(At(X0bYX8&%y{x9lku1mLHlB- z#|k(Y6_dfbLvkY0neqVUvpU%Pau1iQy3GkUgs-UoUpWO{2F zos{j1AjiM1?E>!?OfBV4Yq>0uP%iA@nmauwJ2O%x=Bi)JM@LyTqH7=-s75^H$SAN^iN%U^XmL2|6_xL#tKKCN~j zEme{N2ZDZL~$Qz*0HF_)Pp~hT$Y%9@igL8pXwkA_MSP zR$Oj#{T}(Yye$xxcR0A6xv`-Zj$s~DmuE`CEDnU%v*8*pm&TJtNfO*tXW$dSZm3juZi z6VQt&2u(s*Lm@CRspCi!PYSIuyR4}=)N5p@{;h);NT(mQ%Pc5lXtim*($%KFadq=9 zx)PTA$ULs%_)P3P3q6yp*%DqLx{kbEV2+MZDSxct)(=HHq}Q;}yLU82AsPAir~Q3QIT(>%1p zI4d>XALuHW%CaK}qkQ4V%_GkIvjegc;^+x&CsyWq@LHn_KpQx{X-U)Z;k^HuOlVyR zW~rv6;UI4nXpTeE28>{?W3Kw?)Kfe?IAl4SrSO?`Cmo`3EL_W3dw2&2>uh)rOz-*9 zKCuj&2egdU0%o`F7{cMzB7pD5E5dG-sE(3?p<{z)b7cWSGwq)9NG!DW)9jx+FkQ@o zNE_e8D{w|(+l^=6NMOOiSgZ=HBK61YVn!D1+ncQiq|qTmLJFY_I~UXA@qxDws&s>W z!Yway_w3!$pE%(2kv$Dj3yJROLXt`jxvIBr$#<|Jp^2SYcKa|os{j|;wImAyUqbL* z0UDuu=tz7HojW{WbF>-lp7;TfFB#K+JpWMX5iRYPL|=gWi8yM_vT9je=(5UE%4NvL z=e-(R4J&87vQ4v7A(sM*xZ^Q?Yn0=UV~QPzWSL9HlrNJ~@zM@kq`$sW!oIU%^6K>= z95K<~%d8jfa(UTr4zN_S!%17fJc98}ikRzCT+0VJSdwJ7{P^x|Y)dyG1Emjrrsu+% zEjO2p^}Z=>Kz4+tbHcgsxZyDoDXeUk^M)%XjftE}w3{v!EJiwcD%}GbX5biMP*;FyFZ;eAo*S4 zf;`dv^Vm#7+lSDCuwsmBK~#X*+F!+_E@7#|BA}wJY4OYeYlsKv!}rYIDK9 zYrt23qqagVuto?=K~L-ScL<74aCN^glZH>-$DdV%MJ%Ac5;hfsD2yYHiXq@9uiO!=f{?`W59uH9Jtc06tm2xeZjg{qHLJm>8%1q2AYOE4VnS`VhLypcY9BL0AJ5Jz zhQ#|v$dK^0Z9yhmE_9B0k3qycvc4Yi1pqTKb|xYrf)S{s_v&>b4sR6}+2BtnLvdp; z_lh3UR|Xk^TD+=L6N$CJd8t6AtszilwC*Q1Y#y(KPz+Nt;Mw^Vo{eW+DLzG4U+UDf zmv*+GlCs`f-MlN7HYey#>{&r7JmN7|EOQQKLpT;Y1(Oww7;Kw=|61 zrrb!l6~~AjVL>Inga=M4OZv=P0J&?fEuZfxx5ugZycQj- z>9Tzw{z>S=Ih+ECt(uV#`uv?@lc&7HY}nrQY>874V@%Z;A>M5 zq)g9z=l_yKnL;oMfQ5vp1I+Nk0Db6mWa! z6_j9{FO8E4Ya4u)FADOZl?PFZLnF%0FO0N{*#CAIwh_~L36HNZC-=V+HTbbQay=>q4R`4P6F zIZa#iqu##^eb6Kb?>)ue#86mdA`~NJpc3oudb-sxRe43@`S0BOy}&qAFrZ z%xY@5?>8?rr-vgXHL4I*H{PxBi>V%~H2?;qGD_;baTnfp zpB{{P%n^D1VuKw@<8#$?T#Hv8aFx4hIvVNw#aMeuJBH#vK~s+ShNcvAGaa?M-ybn! zT&Bnjl6zM$Cc%1JCBvt0D7<97_d?GnCP1qOYCWC8KcLQg*;TtCca>%}6yh!-|EL`H zjGiIC<&gSXJNTPia{)|QOL~|p8(oPtdEiENt#mu2h>o=L`&@HX6C#{YR@yZN7=lOz zhU}=S$g9b2Q~Qs5MD4>AlmmNwEkT;tNP>%fY#artl7z45LKIF)P=r^;FcTw*O%`7rp{K#4v@>a7 z99}B~85b`^1_?NH&#&WC#8oJEe^hAvVBHp2g;SBrb!v_m`p#5QM+LJ%PH^Ru*FV;~ z>2yK{3rc3-Mvj{U{e%kO>BLf*q|nLL_wwGlnYj| zJ~U|L(?}kpK{>Tl8CmJKO$xA4kbnGEVaI^TeA*7d7j3SWaPa~f3ucrSd6pop*t2j) zY792wxji~EC+PVU7HMdG2J@dMY}#IcM8p_deAvqP4U-u6UA~MtaW}@>2YiY2-I9QT zG{fz|LV&T!M(CDM;M$DKs@e{4Ma6=kE^3f=&zsI>NTS>Xg);`t3c8UEj5LNJSt2nG zF?r+(F**)OsgNEc9_1!bTnZQU>h3<=Q-U1!xJ=HpbN#U@R_ThK#?ZvJ=5G$xa}}c) z(s8db9A@&aO)OvI0Rq$c)_cj~Z~PVMXY-yyB1f(2sVAM(1$*^vwbrtv~oRwTeH}?_FyS+bXY4 z>>D$L0(e=9hVo8&1EeV@9&%&R1h0%(ga8AV*tpQE)ZteaOg(*(As1JIbw;B@O`4-u zpgJ?euqfifusRc`54tw;`$3S9=I26ix!3TLOm^?9x_0FgXZ=5Ha4fB-nlAaSTHA*L<@!P<@#ErjhTb{V3XKW%y3JzPgYu>eo3*2E56p z^xJ)`@Er!%ZZ6m}f+D#JKojjvt78&cq45S(ElcO{F06(^ti zdh_)u6GoH{EW_tnTGAL&1_S%!_WZy$Wes1LRaJ?;E2v9tXnjmmOY4GcYHp{RR9Vx4 z@jCum#v*?hQX5+-vhySZu(vOOf5cgd3szavRL>YLq%Ed3=Fyzf9-{N~D{Y;Rv!^+L zsw)^XojWzvH9I;9*Hk$Bfs*O$IS_@(4{{15|kE|ND*PBCJ90`O$<)%iH zNogO_F}Z-|IvEQl_;{Tx^c&D`EI`qxDXgt(okxcUI<-P0_9uv=Cpz3eJ}L&Y%~n?3 zEj$|^9FCo-8wE8B@JhzZtTJY`fHk%KL;YQ=##lhMed^xs^!G98G%C+*P(5ps)6{*5 z2ql3R;;I?Ji{7AriPxM|9}75EY{XY8MP(mzU!fDt8(AKDEto8VEQ+aDo@k7Lbg|$i z)8gP(M5tB2AcVVfeAf3}d`U?3V^=@Nz%;XVcAt39JV%IW06Btvfmjbqd(T(ZN>>cE zRH_eL9se{EW~BCHNDCbG_Ioz;JGBXXHL_9o@%o*da_F+%6^^Oax)UuUCR{CiKh^|f z56cA>9{L_ik8u=Oc0F7c);B{eyc*Hi3cY%<5%KEM8->I?s|F*>GeFp-%cdjQqiIWow728WJS- z#R9=Z3N;;VI6O460U-g!nNbqL1Rrk0gj#e@LqS?uTwp{t=(6D=M5>Ew6$&LjeDIh{ zRjZY=zhvA+CmLkyXS#1xq~ru#q4fC*$rZ{_mL?k`K-{dAuK`6#JcyExY_dhlgAy<^ zX~IBKXc=)>#yy$@-T;fvC^Mp5%~ghPqC)006HUuTfriRgp2VJr$JC?)VReAaty6B` zRKi}bHu7d|z>lb0#2$>$ddnTpI?-*Ul0>>gUW@S7MA2x*l}!OtImUA}&5@(sfhB8K zDlfeR?BOTalIf0la!q31a!34oRlIi3bc{!;d{hb-V%tR4h ze^0k7y%Xt+7#Wl*&sMH%+1?8cm%BS_GX8M_K>~C6S}7X0JBX{b#~0m#1VcMyT*mI) z*m?u~olizDohz&{Vh+|F8w%A|)O9)wt>q{WsNC)P!|%%A6n;(Kx_m=0_zSydz(^u@ zm)DDcYrqq`HS$xAqVeww;?jUKDmmSHv@87gY1c8HCeMZiEz$s{64&^A47-Q{+_w?I zvumK2aLHFwbKlzE<=taka#})DaQS_ItaJB(T`(z|^Kt)8+CLDCE|c?-(1EwoFyJ{9 z`0MY^jjb;?0tgbtQNBC_$Hg(blWv|YTd!~DZ=JiR5A4C_oW!ZS>G}b|96lLjgszRY z>R$Jc%fM3eEz_<0jzI(GBqS;0nQk@O;z8~ZHI6JiS0m((G-`*?ub@K)U?cqQE zHj6#NcF>`V%bwp|Y>wtx+Tfi)YS=pu>3iJYJ^A=-wmBoRoJ%>~F>(9l=^ns5I6LO% z^$H6lKoJFCcW3w1-xy8dUaM1ddK0a%Wq)KWf2=$y^lCFzZV_$3r_5N6TDh?DdkJO| zt1nJN)?TFJ_pjtZ*{Ome2|WTjNVT5uoJ2@vDYFsFoxhWE)T=bSRGU!wA~@wJyfz{< z%lAEttM8XupV|fRb8`_W2dd|HCvCMFG_OrH&JXJGEQtN?pZ7B~$@B!l50D-G?be70 z`97XjQ|2m;9T_EhHxSK_iglTgTluAk6L&dRq*w0#jCdRo?N{d8@r(e)0x%gu%+l@s zApw|h8kD0q;$O}l*IYzslq~+Y9mD%g$d5jRMZ{hc%r@$Vv2~|@pL7nmWg3ozWet1G zzbOBZL@M}e8)A^yeyWVFH5iW)2DsjRygD13dM1NL`WWG-Xq4go=EchjPRDF(f$i z|JE~rpDt4`c#V=);W-elk4QNfGg=>*ebL9bn_0$c<5V9l{;L9n@x6ib@xe$jdjAhw zZy8ic6Ksp3gS)%CySux)ySu}{#@%h>u7eHk4uiY9!!WozT)y+}i+CsE{m9kX)m48w zsJr_ZH-0#HrgUi{G@blQ$nYY@M5El(7mD7_C@4}hD_2Cdpo;IX6DT5kN zV@e3qeVi0vg_>XT1y5y$U%?ro4E8EVw>oVR{gNM+(=QmGD_fG=MlW9bWi*Z&M*p=Q za5*JZhCFHsGp`o@D&LhXPl>?=Shv$jhX+z?WW!_%2QHMmg0!w>Y&7XW8Z!&z{GDTOiOZ>jX)a(x;zL51r6rP^lNy zk*8IisSNlrg@d_;?U%ZD4aHzcw-hZzx=CIz!-R%XD(t^XhOZy#- z^twKxo^akem3b*#>(AaVS<4c4xwzENl(wXKqe%Q4zZfMeDSaV5SgURtAzp8n+xx$B z{$SbBy6%+~VL*NB@n=;PIA)AAXf+jN^-Ah#lTy|xe68E;TzGmw+FUSwTD5yny6oSZ zNVpF$#x&=9IcdY)QdG&TJaO7am4JEa7UR&$&|&?%&OnOamFKYQ>dUFmsxUaA%b5sZ z^0a&CT!ThBOY+q7DJn9K<4EQdqbB2ZN3i0R5F>0Q>cM+XBsidHvdZsQ@;R`~r|-PX zI-hp^hXXvIDR9Ps8?&!VH1AEy40PEoC`w4>$FkO?;SQP%^~H%wKHshR#;kj)zA-}h z_F!B(G(lLLJ5?H$gONT`CM^rhz9QYwXh({*j<^^VB*6OH-WZlr3*%7r`Nh&->wF`< zSIqMsNj)bMD~-1){Ae4%BV;(=yp2^5u~@(j^%a*o zqwQf;{6csn4q@s(#DNHn68>G)cda{1FOvoUo8pnAk0`(QSr;TfZ$@aOXSyPW`mgquY{q1t&4O9%AgAJQ@DbqiIBDVzmKM)I`=!Q_B+l~u- zd(%|O2V`y`b5zrpt#)R-`fj6&$Gz4fHw#uEO;BI}<%(z>b{Qpy0SwWZC<$-phtA4c@tex-F3^WP+#))V)d9Mb9n)c{rpq2)&Gyk&H{(pvzB z(y+M?SQ!|DInfhTgSqeRdo4lTx64AhFNEf+wTIW6Awt1@>9xy(AU4UhW{g2@WE z`g^vyG=!MVs@T%T>0Bee?aVIg##r0t`s|+cpYPYtH+urV+>|H}cE~w0TMpH1D*N=~ zYaLDM7L9d_ifb7E?a7DU`d&oX5oiLuk_dHSjA<}Dy1R8QT;)LNB`OhQlw4KLvIcUU z&2U&fk6s+6L~HcNK8IvH=GB1yVykLC=jNE3Dt=S;U^`@`i(kjS<}~OZ(9H=Wqf#59$6xknQR@D zNK~lv;gqT}_r0++YO*c8r?bDVd@7Q!iW2vYDvy#pr>c(Blm?xU*A=CL(~&Hb$YiHJ z7#}eXS=4gJ@8FVim&t(T(o0p=o#TL)B6y(<9w9u}s)uJOCZh|X_r2|<0~uR^w?t`V zYM2^8-4>ZlMuO|NL zjN54jPfa&1Z&G!SMua>lBKMm%!6U4gFq<_S(7xp8n+zKnY4sbcR zpiN2_$;;@ZTh?}3b~GpzF_TBMm7hx;TM>4dPcjCx<22ENZU$4=Zr@pJ-f63cHMC4; zg6r4H=H=dA${cqYkK9?al9z^;6^9NGizXW7um!b3o1$%8bhKGMm4mdfSY~hBnkdV> zFyvHa;h~JIt(nBF%*5A=Ed0*_;@^FjPPMSwts~4Q3GJ~|eG9i)+QhD0rYq)){caqu zR=JFE%w^O80a%orN`3M!8(!q*xRbv}7W!`@)`RH-&?rYDd@`H=k+3s;U~KFwtvvO&>&SDZ|_6n za8S#0Zpqi2VtyQ>)I#=(5kb5Xk+Fq*8K04W2<(`jZv72dYO^(^o}^&;{> z)WJH_#sH|Fc>>+mOgI}xvfR+{ zzR1F>a5&2plYNqFnBF?W3AS6OM${o#r$ss2IGthfk2j}UO2Uwu>iOy4ttq=~F2C{_ z#xg)}CS9o#<70-@;v*_>QY@x<`23qU8qN*Iu_>$_f@^(VzilM_aX_H93F~rfs6wue zQ!>s#6;8L!^cKGBZ*&IRbrD??=l{U(aSUC&%eqyr733 zXkPF$|4Odvr?748Wh`60j3c;amb@vzd6gVPm0t5evaWS~PX+|sofr!Y=5d6m1Dt?p zVM~}!q4LH+wel~7M%Hs3Jbjru{Gv?I@~fZ6y(C`<-I@K8r^A^RXS+kn@`QAH*x!!! z{|h<)dEXFoV}GsyKOL>xqoWGPz=@f1QlLd|I^7>tua2h}3n<-o8Q}=q4)}(3>=}*@ zC5AfB9$viM9mVj@rV^|yeKq;K6~0=UV8-*dGUe>%QC&H}Pd{p<>yLVV5k5v>0+2n$ zlK!6^KkUh&;fJh2LM-vl;l4t+^Y$y~)AFhb|F+aI;H6D&C}q|Lcw=cB!-;8q`2H zrgFZtIfL%v?S8&UQ+pnkaSSQ;Cp5v)X+B~hDF&^W7%h8vusSvTN5w4GzoZK_sGc5B zQ0%z6PhC-D8@HD=j$Q~f8l|rn8u=FDz!x5UDr#du03s$VYDRuYX7CSKhMZGh(P_p4 z(IZpG7*8Nk^PQS=eqhKNP1`Iz4ir}w#lfy@qdDHD4NyL$xLOx067IDu(h~=@Y+mOz zTSV#g{-&!K|GoV=v!^wOaqOJVA5Vu7a&;8{egyC_v=cC>=82`0T(uDs!)2Z7B{yHu z^(O_;QkOK}YF(=GOK7vKNfhyzY3qu!TYw7z#OP$o7uZWfNT4}$Zdp2>-5t~EZ7f|3 zdzY66+xdIT_^-wsX@5p)rAT?NbRj7N#W+%=YbJ8vs%vZPf5Cs6+c`D0NKOHelQ7LZhD@9 z1N#~7N}I6>=gL1=2m|pQQ8bC}vlxkb<07f(428pe@;m-%EFOkGX{~)n0uGW_rbo?` z4xTPuVJ#eg7%E^iTh7dYFlSc_%rg;;cbb=aCx1T!UjXYgMp;0VneW!!^$?3`7~wDtv980@u?3Fco?zFogVhwaeQ^>HGHMS zsIzHxf=Z*|d^@He|H>BILI`gh>3CBiQf1Nq&`|{R{sg>zKD+ZTI%sSQlx`i>TwWRG z0X3*OZ6+3Gt_)~ny*Z3}^QTXL+qRCF@)M#DU-Q}nc%QC@fv3mMSv}jeALm?^JKte) zV!}SRuA0f{I zKr!^a-P6(UOC?a7sRCHx*VjKUAc9zpbuTz85VWWwZd-*R?T{9>Ncbd7P$d2R;0QpR z3_ezbFa8VMk3a(o1u-u-fm*=kLiFxFjf$Vh;S9hKuKarS2P~8pTvGBM|2qZ=4{vLs#R3qW}SLnrf0)2&;v5bPvrl|=r3Aayy^^H1=xFeVvc+(66B49l{}cjTSljDPj;oxmp7PTciyEcm zKf4q)OC~+qJM*mg7B9J@A9ztoKcW5dtpFHa*^d0HLKCamSJhWrGMg7M1aaq#3-=aK zXO>CgK8hPMZFVEmdF5rzJuXyuH7VDJrUqeD3Dsu-!%N`l1o0PNf0!1I)H3qOGbf{9 zNfBh#hExHwuqPytA^TWx~oU^0+JrtdT7BSm35i z^3_iurUJ_8eL66v4^qfgY$Q+cIFNVfX2EP95W$V`Yd>_NS?ZOR&GO~B8|~o+GHnur zXR5*a%PoT&nZux`;ZsBdMm77bqGQo>N~=Mbes}=1xGBP9?JDpnE9lg94^rXL==w6l zgGC(%Mdsxx&8w94;_MO7s)}2}mol}HnmKY9ZrPN%MrVY#3M4HDza}p7S&n9)>sCKw zmAlD3#ImHv6*Y11rQ?PPnwnN!IW`@`ANT#O)jNd#!fJOkdp&v%%O5mV7d_#j_n@GD zv$6rvXTVW_#S8z<1Qxdc6~0oGoPvdshp&D{3$)nat^x|8NK_UVn_3hcn7T<~)5WnG z_WPZTGFndyvxRRr)_mCb`1eg4H`~gvZaST+2p;xl+twFKyI-S+^5WxSN>Dx29)g|? zawiz0V;t7H)2&c%-d`%80|LC)9Q9Vl4*@qJ@hfVlM7@G5tQZ^w&d+ztNyN=aYEGY&m&YER%mQ!9sfUg%y%!|`68E46%cnfgr0)H6DYMV-HD zPUq7wUgimCvM2_}hwTNb`AN0DV$RTQ-ozf%UT@my!w+PV$8&6k&9&rdO;p`xAprG| z*08$4n96en9HY!OW9IbIB$(e$g|+%D2_`!mGA|n%C7OB}aSS#u8wpa@k0-XaBpuQp z4a#$Eof1qsV8Lag0@=aJGet@m>dY`<|1g)>tjpRuhxJt#C%_=5IG|~9EYpi-;8Uar zwfe{CD~uU&ZrQyUsK{t2mZW7}86X8td}T%43RkiFL5n3M?Y?+AJA8$(<346E3AdKH zU*9A;TJRd5Um$-0=RN*g<>){XDF}*I*%|N+P~f*hOBIM+ODLY=d*OBCdlsg8@DFzWbhTf|Gy(OZLC}3T}gQg9Bau z`^e?&fkT4==Z3ypSNFlYK*4yJf%WPTe64eb;NwVOJl`yS@J-VS=KMtlY==$QZ>!IfuY2kH#D@V!^D*TMPG5#{e0{9 z^=AO=X?j}VWK*qYhfOO`N81=IsLyh0J+43KQbesBAiw_if)2(Ri7kQgAl!8@5Mz)9 zdjyty!*J7?glhmSt_77-Psg37KupZ#NG=pGUBlv5=v3ZI5F0gwE zD#L1(&b`D?4?QpB*O>fC;1eXSNQeqYLKBOU><;o9tMWU%BpkB$*avl-8rL|;frsQuWv(Ob7Mji_3((m5^wLs zp>uQ1Cgq}Gw*Y#Bqf?+eWT4xDN}-_{;pwpj@DLLToyq0n8alMX$UW)RfCNx@J`Bu@ z8KBA*;lW_&HS^?TS?X$vlXX1|T=LDo1i&Fz6CIvCX zlkeJ72jiv_%fx=RltiE?9p;DC!CmhkVQU9#^t1uP+`J4Rmt zvu{*CM&K{;zqegW8%~eznZ&dZ9k4m=w4Z-ROwu2-b|wng^Q~%{yuDqQp=?O$#&z+I zyQq&qu?Z^2W5x=yT0UgVj9A9veG|R3Ssc8NpdF}IcO|2yHhdM-hSZhH5=VSM{$(!4%@JZLRD? z5OCULO^ni4FMrn0K&vN_r`#eFN*?RW6{q9=y8#%&zhBqv3d!rVKN;`4UHz9efBWzK z{?)*?#pDXcfWUvhYlVF|V)d8yFKsmIc5AOyOs(kDX+KoY<`3(;de(6F>&ex#fY{nL z{b!CCL3^~6e^D`Bo%T=G+&(SM;L}s4)e07D&p zI9~v6#Y2*&U0}dzO*_$5ko^i0UV0t@g;Ha#I4_;c7M^-&m|bQqY3;oB`!V3f;#^xi zh1~@gg4c}hCUaaSE1}mK2JXK6R9oxuDHj&f5*JH67vmp-^4>1r&!BJwH2)Iq?vXIh z?Rn3>d;#B%my}{70{c4azv_CP55Z@{hyH-maUZq;vCPEIXgJA#0d59^6wLU{8U`=> zaaAj2kPB|4c{wk>fP+l6>HsU1-NI&z{y4p?pY%2D3$riEtA*;HBjJ;Wiv42yp{Sc? zZ^wN#O;*Y>?L>O0=*s&tDJ>?2m1C&s`siu+6mbg9o&@rin1A*es1&d4inHy^ig5v} ztGZ_PYzmC5>L-V%I`b|3b^2PWqjiOWJ5sJO#Skkz`lAd7%h4XUIji_T5S$|H`3%#u zlX8_Xq+|ByQ;B&~MF!d>oIwp!Gl1As3bN+WvtWx~B(-wNU^oO+aZ0I{CFvy%&d=ZC z2Sdebhc+_kApH;UO&>gj41EW1$#8&Zj~7UJeti%qg&~YgT?s^kYC2lu7lqV+CQbrf zL`x>04sO#R^?1metL}r|WFx{Yota&GWGwRHaC^w?Xf=`tbM8&xH4*T-AYOSL=?3m` zmD9nR2*c`zxRGY6;x4iU(pCaiYhg-7bF}QxieVFlX!tNffo{9ZZO9Lz4d3;|YFZ8E zt6TvEHIZSgMXIIo#-q+7-5qvvg;)Z%MFb5ZoIlKSL*kmubl%Y zy^W(5g&FPxrh)t<$vOue8kiDHoVrufGOeMlT$P$4vOHfmIeb(v3Ws=(tC zuUAgiTA6U@D0N04?lO~k0P5{Y2;AE=`^@z;fZOtpn(5QPBZ~oCH`U0qu3FeA zut|nll-A~*LB?&Q3 z8fx6x%Aif9==a3qfx23gIegxn<4ytEISNkClcZR-*AHpkXC7Mk zY5iTEKZ~mBF-x1Umj=mXd=1whf+gA()rh$gt_2Ah76ApR5w^0dsjqt?<*|o z&xT0|$&^0oz{wD_8RLDnLL0bq^uf5uaJ9JsFFr#ily_)WXL|jac%1L_j`*I9%*2W~ zJjsP(zxNnEVfW76@3SeqWD1A#!iRRY6Sb{8N{&3Xfkr@A9S*pzgZEonbIP{<5;0jl zJI5Qu#D{F+uv0q@EHQ4#=C53vMXMoha>+v_v%04K|tIumF@ zbL8rS`B4K9v2F$ou(@yc zNlzIx=QX#rQ;Bh|bR8syQf__$=&0*0^=Iy@13x*S-~ol2-b`zN#59#Y_@Lnu?{Q(~ z4*6oZvrJUg%y)*^T0qAIl~xh9-Qb$crT9M5iw?~I1=!lLcH3BnR=!?2vM{ZL!#1L|}g0r z@F5#Gz0RK_G9(3RRJaiugI>LBf`V>yb8IF&H;GFI<5yan6lNrxgz^xcKIAo$K(q|N z3~C~!lp##PmXtu3x~O5GtuyWLf_r?m5rx_;i@5Tl|Db83RIQ6^;CW$ha9dWRz9ffw zJjvN(WI5-1!Wr;v8-96Vb%qns)f^o^Tli=Fyj`9E+kl>FL%)Ux4x3nOtXk9}RZ}o# zm~m2j9-V9dOA`@9f)M+AtLQ-6Jvb5oS{b&6>!A5Vt=f+Fj>CWaHyvdJ!{eN{L-Bi{ zSsUL0#wA)ujOVKNt0_GaDQ0DpZd!aT6N|dWFzzGybLoZYSad&|Q|w89{t=Uq=zKL) zuR>~4ia5;fX}0Jz(KH6$h%;ZLYa#})h}lZRz0!`05M;tKgKI1=3Le=;PcaCf+3~=x z+|WMq!CRUXIz7$ldBmCs$OBTmGaTuR(J+uQ6~Lh)_XBXj>|H zZVRSE%u3;nvA#W(a`n@aK5n^C#~RmF>eBf~79!Vnhn~NS z5FL&6$%ZlU`E47MvyCDyFEH=C;Y`IOQc;q86h zvz(;R!45&K(Fa^xDxw|K1NST+#iqKLRoFs?2kqx?2^C2I;NZ_iSz8(4x2z|?%*>KT zp@#`_O}NUm9y4WdSa-*3bS}^ykIPqQatMQLb0>vN>E_rA&m5i# zGlMI-iWHZlV?#borQ{mW2({hxL|X7*2L3DZQd>OgFS`8uhTJDT6{*AE34b~FHKa28 z?>ekNiEDIP-xG?TIC~*(qgAM2ep`XtT%&sW1V&;lp+Ww%n7k-Rfh+|{G%~AdJwbUF z|Ni2Q{Mwx%y&c6(Hf3aZ~z&E zm=0^g;Erx2T#D!IeLJHC>26()EQoCmPa&w<)sU$9grVl3`9)^vUzgCaR7cq}-=k_o z@(~HcarTH&x6xnM;c;-*%X0(KB-#&`?G%u|qKHep%b`az!$0aO8qA+yKSirQlXkPD z@+H}2sM}R2Qn@|@NDkq05sG<|REN5Zhed>|gx_lI?fTQlSEu;>0taw(rSJ7i=E@*- ztHUvcgos;Rwkh@8)T1!k-rNfwEWvT3@`a0tHkwm1x~6Q{#Nrbfps;@GxGq+{`!B`y zANA9>i#PD~r%?obD2z-SIr_hb#jBivOa(;iDK()7SAECd_Q=hL;T@N~tt@Xr$2Wfq zz8CQOHOk`>h%D>nJ^$kdZ`{cg;!u#P*tguTmg6dlvnKK{q?-AuqWDiNcTSYux<5>}u zuxaI8tBlm?gHM_s!-J2lSLB!)qa{UQw%Vgrev=5o=U}S=?uDBz&BIyQS{H+2xzvKv zg2e#fTXFnM#4kk0rzUl(>RJj*AuVU#v93(QsGy{vRFv#G?{NxmMJP=-3&Kz{F||~` z%rw${i?V^Mtqz{ji`j$TLl#r`QlwlEEn}wQyh4FKnmh9Fb1S-F?w#veR3tPow|J8D zad}|d@-7~1 zZBZ@Mlu$FU`09725V!2mO(Qv=^c29!Nv+bfSO_ETj6G6B_Uo@dkypwCOemUP02FVX zCS4DD+2t+(s%!A$xI{a3UA7;Bxm^J(;o4SHpUB6~iVjgMUDBvn8T-xqu_Ii{lUi?u zv-!F2+Wb~e&!_N4{Rf^wRiO6rXxu)*yx{Q~>?~AIgw)mr3CmZWG$t}~^gBXkQ;ck7 z>FU|X*ptem5A8VWLYQJ)>_I1GRkN<_FjK;W{Gs(?YP+m$u!?~Tg>Y=Ylw$$1bHCS8 zsEGgZX+z)b{Z-=a2_pj*=(FV{;kMCixtkkB;51#J3&dK*ePAMAY$-xL63lqETpP+72g&sgz#lO;FQqq|o;&loynd(91m5>XEB^ITKNsu- zZ5GzlNXSeIVI9W%b;_xqaHwybkaLA5c%lFL!ZI?56K_vkfNLnj__q(h1su~W@x!QD z{Ev_X=G~S7g^E1i4X(=OWO{ppL4Xtfm?+$h(AEVRDKK)v-sAIDq1+t#U#ojdce$63 z+_#)E=;9GNJk@BBLo=1_#$P>PIavObtlpD$t1WAS*hsw)m6apBLsHyzkX~h=z~5W;o|t z4c;L4=X8xST@P8<2d79*xpO>Q9n}JByM7tScmwuhdHu_?}d4he$5P5;R zas#}N=Jucxu}_jsNcU~wwcu4jKHn+HJh;H(zuu*doKDxZ>nc#dbSD@W?xQyc;XLg8$(&cJW4U(ptEiY{LJ_f4Z_mguoPAO=LW@>CsWx( zLc}rjMuWH9;y4(98!5oMyI|iv`vW@M?Hu=R8Dj7*SBR=Jn?)cuV*%hWbPd|hiQ2+7 z_(%;J!Ie#>x*_JKLQsu`5!_HgjXd?GmGCbrbY&6q_A3fi*xiSRn$FxZw$tRPCwIOu26l9b@~8{5S{OYvXByFhbUjm`68beA%u#iv z2-JGle+>T_2_QHmkIkO;r|;SI!HW}HT|cVAcKurf*h+&*JEnq{+I;cnap+QBlobQj zz8p?=oPTg(VPTmZA5I=SqsGd^nz)4%cBEMl7S5*P!uy58YVl3A68%RLcFhx`$L@>F zTihYE{NY+%$r5Y^N;k!#Eb(ecM;J&ANn=B688yj78M~dg{`j5#>&cHc3JQzY{6V{XKH`Gs&K3H3fYx^6^ zQO=$8#s>fU++;rv`ef&Y;faBh!g_erfuAZAqbqhe3Wf~S*sBxzl51XRbGmw_>S$3j|h$~U~HCM$%1iP zfb15gnm;2Ar0h;JgtlWX!;K-BmU*j__T0s7WwuR|YNg)?HRbWxm>1-xB3X=jHFFkh z!r3Z-tTl(O;|-a?Z7)D@xAz^hYIj*Io}r~00N)osfm?Ap$iTVa*?qUCT|w= z!0q9(c^_j$yYkJ^yr`0NZc8@zc8;rHcb6-M*0j+QSP()0OQavS-ZAQpdD?9jVks65)kJ;p)ED#UJv2W)I=@rGAyGi_Af#l}ekDAt`>4>;BGq^R<7vNlyeUd>GgU>Y}=ETvY z>EV`p89|DSdPu^4JoB?FHlyCGebD};UJc#`vX^8uJ#URqm?ci)AnL^!HplzYc`8lU zl2(LqA45$d%Mu6DqKrUPO}SNa97IWjI!@fORI!p)l&nHMOrFMU_Ft{^;gF594#_Bp z`h}0Nact;-7XYUYi*!oMhzWd~KR)WpH%iLNiWc@Ss)D5nCEQ@U6KwY5azGK`z+h09 zv?FUTA&~fxwKKn!jx1OmR`Q=+Rc)*QNU)qC8#~4gt$Q zq*eoC4jx&B(+~)T883>gG&)~`H4BN;?!(_7K(swp1EA+woUVIp2vpJ(!WoQ>lgyq6 z?j0@z8wfGL!PSG@XL|grX?M^R#?UJ*F-E+63UKsEYsZx>EmYqa1OGW<0JAlkIE@sh zd5l1R-sya(4tcH>XXc=5*=Ve-dl~w@#>Bek|HQmJXWgUGb#MhWs2F41mB^W}81OOB z`C=W_0-#b|3I_DtD{qTrDG_$%Li`~uE?6{YV&{3#mPIW|p7lKT$js+UDNDqUl74XdGjGb-lbeyv&5nI*0o9gwBQ7M; z0Pt_MyHvfFlT2@<4`migQLM5}vNnybf?TFL`&Z|c=b6V&%^M35^ZKEit!h8U^^JI~ zSNzm9zmVci|EioY=k{{8SwDRbSh#O}i<0G2A*_$L3KE2P0q)b6=}!V{y2P8~Oa?llj&{m*H@F(Dg1wmy^R6 zRadTAc-RGHL@V7alp;vn44A_k_EnFc*lbhumUh@7|$|h z!i^TMsI4N(8vrJ`(~@Mb;CMz(QfN5gR%79x2l5bzP9toi?kM!0J@MQf)V2>_HWiCi zzAxfwKRxs*K4t*hX9*_SYGcjsVZ2{0>wyJ+6N zHiM_@h9ER5sSWZZRPOJwFajLn{(XJSsJF1sL(^^xk6`uk%bLpc1zl_b{~QiMkz4TF zm+bveJTvG%I+a&TRI`K@K+$m&!HyCCdPP^O2pbUEz-;x&e|e+I2}OAbN*6!5F_YSU z2}Pq=iatPBiRw?4CurdcQy~vdC5J_^<{FBkdeW`SuZa5LgzKMpZVzBrLwyEi`aUdv zQx`@GLPs5?giKJQ(FjzfNh1#0!HbL;tu05&synxnkEgccPi@%cwpH>l)v>U+>50kK zy|_@`sgdq`SK?J9i?^im{atZuq&r(Dk&H*NMf%c`qO^_r7imB%d_u!EE{Ea%2}IQw z0+l=No#^89jmhnq&=HV?eZfLE|MkX{5HOu+*zPAHi;zv4Act6L>bfsyKDkkT#!I^Q zTEQiCK2ZAQG_z{rM9`o|{_&*dS-ILv&`4Ip%wvy2M#l{^lPu?P!>&|rmh4*6ZbP3F zS$)$Q&wmfS?SC2c^Zx5#7o11Ce=SPROxA3_R5de4MAKST1|EFD>4Q7_|EZIMA4D5-nNOaWQY5-=;BEI_>r>M>9*|;N5cEwFOs|Uoi5})4goqpXQUzE zKtfx0S>92YkB&~PZm|k@flQ*>r%lXCwMBCB_UueNlL^t?_}3fJemgT)@d^B5*4IH= z%@ow1H@Bvxzb4q>Q|v^bUd>5=+lMKbrGcebEdKPIkb3+NKQ#U0{2lL?syL4EEbZ9W zCp0y~F&)w=<*Jq}EiJSk-i+q=IN$%TT&RVK!+)NWsVyuQGc!;H7mNaw1=vIhK}y*d z2!brfTo6b}-hvL2qkw?^|2Ks7|K=r8K}Yg=_WwmoRig{# zji%K?NFtemN@bk7i<5zZ3{BnyA_pA`0TycL!0Zs`m*RlaF5N<}*RU?FmAteSv&n$9 z$y}C!DoO>4D#}Y~4}0#o-J`(SUQE!t#g$Bw_){i*>*;)^f!?mf@9QrA>-|LEC2?;M z1u=pR;CWDXQItBmh;=e$2t>)u!M^ZwW)n!tgZP8Y9yCiHfeBws<#tLl7&WuvR{Y6h ziOuE@-|9Dc-S1rHMVfW*#l#CE!o4f==&9YOjT@Ueu$}E|JK5L+mB_M?7-#|wg%idf z#UCOs{u#=gvzMbH)Uy+ft)hP3k+HfMI~02kcyT+P(rZ&qO=nPyRT@Q}%;$3Ir{B4l z2wpjXUY3%ClSp_6j~2NPQCQyzk$GCb z?=I4%otl|0=oU}RPQqe9j1mE_)KsLrykh>HTOi=wuTeS&H8b|N5qHc4x;gDF|1B6RTxQ&) z$T2As>0OeRj}Z^eqXR95qUGKM{f)MMhE;86`P9G6|j>E{T*U5`XMK{_wreLTn&_K_dM%m<^w1 zPyV2bhu(Y@*B=EuFKTIQciZ>p8@Lg`(3OwxC)+4n=8U^MvO*hP_298_?z8** z_E54+okFlV$V=s}SicZeFE2L!v*2}~0HJSwsN3`##7bvNuu(Nw1M|3H9TjQUcSbmK zhDDovxu&)B@#C%)N}a@C)cbx|KnMw*acX^^#3D!LnE-=2n+ zWmcU$JNifSj@)Ti>YyGy{fU(%qKETwQO}^W%BJ8JEt0pi)nagRfn&iMfD`RhQIDc! z;d|LI3(=fC*5VAE$$JSY%^iO4wpVYi_K~HerFbv(6(8BL=;@+65VK|J7ty6Mio$2X zrS{_hBzzs0$S!5Cl3pkweJo5Ec-!u9czY{Q-wlLLetf1!sM;qB>ir#EKl9R%z#xPx zz$eM$_VNULgfZitIjpA@z*G6u280KF_Uw>61j(~FiZ;#`;qt2-JT>VW6G<_MY&Y_1 z)Z9Cw_wSMoe1b9MHKQn-@2m-U4LANUo^!3tck2JuUA_@-8jd8h(oJ&Enk5HcoaQaB zz1W`Xw!zsIVq%bdH4!bgS0uPp7v#eB-u#;h)>j0>Nb3-#B@9{=;OsXBN0x$OAP>7G zMkzZswP`r)6c;tTs{9@V#fZEug~S*ri&>&LLMkgQrr+so7M?RVK1_bZ49aB6q;3zl zkm&=XevZ2T-S6UVm|JN*2Ie$PNOvxe7LR@Z>5G9;c1!nvKUlA=%SQM%27_jH_I#DR zfj4nsl6E9N;1wxA0gDB~BKG=0hB2#}EqGQ%*d!@qhiowbZ%kBXcvs_jmU2{U4%>nv zY0EW6%k%KzXF0g(J{qjE-I)A7lfxCQUcpQhMl! z5DAhoNl|4};o(J?@OQP3fA~6+X7V2AWcdR$zV`vB`uOk#+7nT9H6nuUDxQeuju*{z zv(;dm#7skRDlDPspIzRZ`HUVgRF|X?6`s=r(j)BH8Z4+{P-R1UgCSKzFqYlL4Nn1V0pQwVMQfNW>;0eG% zRObtZ+SXXZ|5MmG1!o#{Z8}cJwr$%^I<{@wd^_s6W81cE+qP|M^3P1w{8RH)&DlQK z`#F4S?{%+rUDF}T$sdWkC=-dpWSgG&8xSG}#1o_#?sv^a%-ZOD5;oiQ)J{}i7bF`2 zV4KlRJT^mWAJZN30`TTpM4$x7u!i@whwAsW3tG~BrGSUU;evH`a)if&A~sQey|Svw zJ&JO6hRb1Lo@CyPh^^(8twh8d-Pb=xsx!ts(ZyB;IGSMlZx06%9JE?)D<ju_rG1V$$R~H!sMiN+ zf(4+ks%50wB-4SZKhzEzflnw9_Z-Oc8s#D-Wf}v4^@ixGeIU`)gE}R)=$t^3Sv=bR?r%y&x3Ch*YXhlNc zbn<9nNQI*g5Sp4+C_D=#_tY{5n}RhLXJknkjrdp9>Y3METoO9rMFnUkmrlU!VBp3+ z*z+9iNfcLGjVK$h-3u54T^+qkIRX5;@bC7$d4fy~g8f_D=yhMG;1lu%x?F3>fikKwm<8y8-1P_jLy^U%Xipv|XqZM8I3;%|ae!vj=5?e+&c zd98rlIKL$2gwh$dYV^G|#?I`4a=wQLf$e)pOh8?q;G4&LLU)r-1;frba|#eh_Bx_3 z+|`ebxzeLOtnzsz7F~98$%2M~f98Ncb9@%cs=78|>yeb(P%x!wUrkmn_t>**UOO$k z>AgfwT0n}zVX$p{#>`@T18+5hg9@a~=|oc=F|KQthf!72Qn`rkUbCMwt1^o`qqOLK z%I&v;n$>72z@mjOdqk!Lt_fKA8hSByU*C~|Nq0@&O=8KI&P{}Y+|lbAwQ|KU>N=oE zVayUmxN|Ac%TKhb<>e_MG@z`W0|}p?t%QePY->KL)Ok@Q>;4T+n*L`J{EsbV5S|Td z#XoD0#&|RovYFV~SQQnOATDV2V}w#gZ4?|hzH(Gj7RMS$StYcRr9FTqIH9)T{JM|T z!u-6n>puHZ)f;p@!P@oe82{JAEJq-}N%A{oLupw9M$;~JYi({lJD)qb49lGih;QZqL#X$(w!N_YghZ!?C5UlqVvEYM?=7jG`Pf8~qLNyZKITC_wg~h6zj@8wJrv(8o@tHHW=g-TIHkXo}%-TYb68Pm5RdK0kT77GLfSrM;F$zRgS^tEjrP ztdYzqS*0k5b@@+#wmi}mXbSh`lA(A$xD63d0|)S zf^;<%2V+Ld=oC7Eaq6sd?O?G}f8;;Lp`gUZq>tGP#t69Ac)tn#O#A&@^E7_;f$IKd zq@!72@QT+J{pPDu!1czoFl3LFbiW~P+?x0S1U=i+i-Bgx@Qrx9w~3<5D9RTArNc%#ia47O0{Q9F247x(a+A;dQ&mDl6+cyE2y z^y=awm-{M$A%kln#w*Zg?Y#4mbB(8_(rWu`^BwS9jJT{~n7I3r#3$J;Pa0j0l1_5t zAt@dnnIL{_3$YU8fw_sk=OuRNh@gj{geHR3n86i%#s)zbOm)$!Dt6c;g)<9>v4yd> zx8I|pf;Y3h;QfT{2X-+gt6Y2rGx8{N+wDACUHbg~?0H_E@7{TT%@a>5MJdBdv9FaB z3JnP970OyzC)3pM*Q-=VnyBf8n?P^FrJKi$)Z}D&8eUnI14RQuEhJ5a3WU3xY3bM& z@Vj_%K|+P!5^<&z_l3HmHlZVe0)Ggu8#kYj9Wro^t<8W%g&hoe8V>1tY-zkb?9?vs z9QZbtkpnXJR2HXbQKueJ9o-HsPXIYUQSPNlE;OaCCJd< z_wfN|{Ce)?Oo3t>$P@Vhzm>U}@&xPG6#$$&`RI+9J3Pc9Q?|f^j(Qh&wK+n5pLEi} zx@}Xwc<#RZZeF)ffHr)t`yk8W`Ht#6MAQabBexA|$i5}JW(TQ4|Ro(D;$N(~q zJHNa}Y`J9hPB@n~f(h(5x)UEblrtZ1kk|eAKOw6)676;oDsH2B<@m&(Ul~58!7W60 zb?|W^Z@c?q4+0>%NZrjY{D{d7n_Xb0BKLKJ;hu+yv+t|2IcxkYRe638mkKS-9sPy2 z&%=1$#Cs^pVfC-OVtW8AX}FkDF#u^l4y-y^c?qUSO?5*VofgjOjES9%<$Y}J$~FB# z`=3S2#zl;zSjb%$)?qaaSL6^}PVA{aETa-;BV_7^QmmmW`F@?voT@z z>J-;enG&G&vlWfnbPet0_DZm3baX8O-`Vv~9WUI@bU@XPlAFNoKhC`h+~pG&Y#`0q|0@nR35Wp_zPHUq{W!POxH8TrJuZnuLud1 zrWshAIw}&xCWMF}{y26weM3XKwl;zi-IqcPjlp*}6A)+ishOtt7@WM<(Cli={FA9d zE6100m#qEdIE%$S6)`D=SW0BE0vz4rk(M)`WmPS1)zB0v?%H75EPx8KdaKpn_hXIC zjJ7x78+$wu*>CFJ&e?pK;r;`fR)naJ$3>}0Ne*p9K{8ncpi8V|o;qgIOVfA;Y0jK z{JE~A{lN890#Y$56ChFYJ2ruBJ?(;g51LkNa)^KA8@HJi#1kzwCnrd&JEZI;!3D`b zKqlHmjiLfh)8@^G%jU@KS!Q0trzn%yohZ?d|{8 z2&+?--e*Yzmup|tIU_vj2k3{@RfeM(_)GYQnDOGa0OoksI3Q(qIr`aX|MfU^*JHcf zkyb>_#b~FpQEC^sUcjIMI~x))rbES#TDUAh_KZs4loJuy?=J#nYE?4KSo?^iv$N*r z2(|(x-#Vdw0e1%TI*$459ax?P+1n@|Dh22;CsRnXE4GYO(9s&6vq;;37`e1Z7rGhO zjBWvq0dB9Y4$?K)RPD$fp|?Tqe85;!X(leFx=kXKYasvD2C$_)=0AWsBfzdw8`--( z++BKLm(bv(U3Me9&#sHmuW&GHjh+8t!N+=vNYu2cF-`$L$;%~s@3^MuNR;8QmUd<4 zYHDYDI{;*3I-i?I?$mjBRq7#w^ z05c1Gw=#V1egfZ2YF9$tRPtydgbMx|Vpwet_Q!|PrlTV*BMGtUqB{zbm7Zrf%`Gw7 z@Y=287QMCh7=#CU@YpQzFy*+3(w4tbZGG?;5Oos@<#NM!sK}XV2JmoFZnKL&e}MdY zlal=0&=w17cBbBpgAo2~Tbegna3d5?fS&_mVp`-HbH|8Lg+uI({6qWZbwfM%M6Vl( zH{H`tOCad6E+rSJgkNuA;sd6icNYblNJea|xw7`6=p}GHBMi!=F?1>t59(*7x%L%n zhlIvhMPW3!=pFHBlJ82#cBRKK-noT zv_{5o;rgeFkR<|dTzSsu`Kv2HYjf&K7%-x=XlXOj#G<4^$80jnBzCu#S>F3TSbV?6KHO&8rfwK*Z`KnN$@K(SPel9$9^hHG!hz&|T9l;~ER**{Hi=GmNT)3pYEoMGe` zq0%zWAE3UOp$uA+O^4Hpz(qa}K<<`a$3qE|o(q>-YV*3ZHNnXU%A&ROxTmzM0=;8q z)`BwXjaJn(Lrh!bI7$8#;z@J(Wqr=5}t=Spxq(6Bal=Bn*HYstg^egYc+l&wUr?4;At zKkxEibx#4Y(*=CVFyy>!&wyDqr1^<|VYYD0L(HtaWDhQ7lLl&lHiG zOdb?;rJFmD0B_)0E6qS!CXM~Z?bxhKR7FT;TBwBZbv%D_s>RF*IN0U=5NQVBfAs9uk%we-(xcHiq^wT{Pgv1Pd%v9{r*4i@Q7CO4 z{fc!HiE2Nz&`|xwqR9DE+9L8q*FEOJ#Z)rgru9|XTI*oUQL6Gk01&&b{`fy3= zx~^&Xd_sDG%~+QK=={0PRhZK5_o2~`trv>LulTudZv{(}29vd#{LG@(O#w8u#ETr3 zH8qbL>yN!QL^o^al&6%Jcid6<@u+d6nru48Xb5JgsUbk%d2}h-9G&?_=t)=_lzyVB ztF^bWz>4x)P^W(Vj3U3egOG!(*Z5|Frnv5kL4uxUAD0FI@I)a57g+wD>W0|2imDG3 z#9ovHl_OqjH;`x`fw=-ls-oZoSVgEIc`?#z{p=$G&9c}QP9#T|Cp=%y)H>F8Igqa> zrd-H8cG&$59_p2M!P@IzM!!b#tMbPYxE_&vbQrmp%2U?xV80bR_Bh_MTYYWC)i{J?XN%)tWpbb|#|eWG+)IhXYZ1&u-) ze)ox=u#zN@oKmBQx$w(CEfIr-D?geWXdFhXsl4eOX9w~F@zW&cPvSrGeb_emdtIj4G-8;0Dnxa?(^P)Gw#=e zugozOHUWz-asXlV3%`T2u})FC-f6&T5i>_LCL6B`s~IUK%NpO=ghg}eWBY}>x@WHI zM9LJM(^J=^icy2g+~SXX^m4zG;qDtl*~wP5 z4|b*2M{O^7pHzRC6oNId(TSWj)wi1R5`i=b)sBbSx_A^%#^)%5(-6JoV7>E`HNM-K zgL=-72VaxCXK!tn1LjpQzbGIBBsfu6Lb(Qb_ZYyAZawF-t{aGykci+y6Ua-97c9HB z`1xtrsgUOr08F#YjRVWW=C8=>&(x#h_L2rKx13_()b!%WS%_3n4@nmfNnlh;E%i)s zV&PztK`Q9<(Q%Y{q`h;X$Wc;dDw27L(6V`Fvi6cSwBlrvij@1L8`VnG$$!fWVv%Kq z3sk5S#|z>riM#{JzQVy8O(Z59Ed!*eZ|~xMt0{$oZK2lM2emWHl0>Nz;(nKM^NLs9eSKH?l0#A@-0nIB@Ejri zi9w7A

)bP%^^Wm98LEL$(fm{F$bPsm)vwx*&Z*>=pf08_g+*OA3#&D+-1rHh_35 z3X)4Y2k;n#|7k9~C15}_P3Kf=0f*g?GKv69XGDgRP;sEF4?FX#kVC)moR9Q98Y5s_hxUeLM z@_C5%Y6f9X)yD!)kFL!1)eQ2@&Zso&jHKY_bWiq4xM}&94_5ElJrq6qkB(2yJT zXqJaWE*tBsA4to zrN(}{^oboHX0>FT;g339N(KQ(?3_Qv*3;FK4R^DR4zux>eOVvzOn;Q875Y3paVpGb z;`ER9$rwg}P4t6HlXvri{cLxmG-ItYrp1Y}N?3r8tmRRQPXyKc&SK#nc+$ncT>zyh zP)M6j5J$|+BFbhsn@e*uZ0rPojk$bVi;@9CNUx3bBnq~VfSMv{s{ALLOye%BqjT~@GTFQyNFLjzU z<4yxUqe4HkN0B^7LloBG5d$aao8igdfk(eca+{uw(pNIdMn5!BSv$`pqISs=_o6^EKN zp}=uK&F8c=&uX~UAwOhcTXhHld<~6v(k9!tRBfs|iVViasA!}mSYeJc1M2|TG%8$p5%or*wa4e zN3MYzg_vWWcKwxeh-o%DBDaK(Onc*a$w>&7%vci4<>V^2Uh zf>lUF|60n4>y^G>zvC&6G%xRTA>VeWf3y3)fws+T9;BOc6kx8S)kybwE3TW&YvU$u z07N0WMR7HFAx2oxzMiV(dnJogisin<%3Neb$y9V?QsT@R6#yTiOkHvsU`V3@r{`Lp zaLW^D35M#hG%U0~IMf%q0s(C8h@Al_{#hUE1dO^Li; zsKX+-QR!eMqX60l+nM}>NDYhO*{NI(c@wLmg5d*aJh(JzzmOs#D0(y#e(i2^NJ|d3 zl|8>lZl_vbv+4Z5tJv_$%#jc!f-A5{r#J+$Xr(~KET~bnE;>F%;)4qMt{BBVtlU#w z#xOq7y(E1n2a-7m7wyi^1~Y0zY`pv~jwOxke26B&E4rJb1x*-`PT^vnua#C`GGfNCPF?sfYaHuD9+h&|rVj+k5@wU)_;Zl}Yz-<9_xG1@VE?L+*tLW7G6?Te#X$@qwwz z=@=j!vjO@3>}K76FIX|9w*g^+N}29ntkD#Yw|28lHy+K+`8JM=m< zp8D{vq+_bCe>-gKQ3p)m4p)x|E|FSYW?JMYzm?IL+JCo=xXSl*ERS$)a=93{(x zKNZgO^RuQKJLh#oS!2`w94_9k%X&AnF%O_Nbu-Vog5M!>AyLpNR3t5=ZX`O4kIh5B zgUy=t`l&s#dyfF2rHzT}{+m-ZY-uGh#CKI))EmB1tC7gzVYfMdX?wH1h9AhmHng&# zMZXo#qSHbwQq?Ox&biW}@nrtwx!ziEV09qeHL4tkudlrD6Jszrgke)?%2n^qDF;AR zIc8sE%a+a6)w~#3aq-E3J@?2_V}*dv_=bw zEdCdGqa1vFj+2S~@HWJ`#ty}}B2p4OX6Ni_TMadk-Dn|D7(0)FFqADTs;{AtA3Xgx zS2GV_peU__wW3A$o6)Y4pXBP~@)dCZx%N_^6k|szd0m@dwIP{SKE8?pG}qp}pnSm2QCPi7^29 zd~VO>&ig(5b5R|f)06qkMMtgXm4}5=$RbMp<7(!^mBGH9K#kDfy8WW`O=UxW|H`?+ z1D*ZO$OUJgizms#(d&ofVFp|poU$?xXq#ah0$w!Jf7S_U%O3{?X%~G`N(uYvZ;SJ|74ob>jNhE{!n;7F9 zBeThvX6bJCIlhqNW0XV{Vws=T7m`bp%zyPLXj{4c&>r6*Z&;19lmT9@Bo#%FQspcFX&l%Ki z%4k~ZrWM5g^~N=QuGlWFwVc|z5a;7?P5Bk88dqm)+e$_ijI3C4ehcn%uH4pHpWS)i zpsdU-y8r2g_rqW#OEwP$LuA^qYDr;|6cfJ|j!O>!1tt2E8xZF24-O=90B-c3PO>)t zFx$*g{ui!|-vW&D|KQr1Yc0T{2|$<`|M$lQf!Y2`*o6tl#LCPdZt3LgLde9-l&mQ7 zBl7z3cl`peaIpSg=)g`bYiAS*^v^7Prfqa;)6(teN>tS$_y}WM4Ms$T2ccTXd}<<3 zGA{qmM{l`ie3qYXN?-=Yhl!1@=Ii~9tgcNbE%iL7H5KZd+C*0iA2f$z8?r*7b81Ue zEqzIb*}qayikfuPg~VE#R5&x#zW}!uu7!mn*3`CykV=gqxOmhgtZN!rHyR|=`-HJF z?~p0Mjt;?p90^IJBKJ+unEtvJg=gXa*$e>X#-ulha;?$f40QqKKw{PnvIn3B^$|HB z7Z{0B@eO$C`h&FAXi{IsAt?!!J;vk{PlB|o#o&V3<+Q-Xt=z z>%6LHfMU6f50iz>5`&>q_Lwblb4rO^Hh$du`J@p{KLaNx;ri3Vc1z}8} z%9$j3Ch&QAG^{{BdEJ;`#~t#^j)KYvbO1C`69_1y&Qz{^kaI>pHdxp%pbTMxl}a+A zUoZ`M`aL;pU|MA*mPC6%DpDIc(TOvVFEi*sL7c-LK$R95kP?5k0k{F-e+&eJhAF#* zL?izS!y98wPTh3pCemyDNmJTY+Mk_8xpRQBOTp;_t^kW6w=y&UYO!EWxdF4CnH0j2 z&_xXWrJfJD8}dHZpNeLJxq?h{jS305Nz&HS%qXgl7AlI63`Q3Q3q&yHZ6adOv71GC zlam50SQ0|V^ceZ$FKEh@dI%WX7ZVHxVy7bgIwmDVo!KQ5?{f%6m>&T4X-OtFU}adP z0m)1|?Ik4?FIf1EQJq2+NnuP*#&;M-vl9hxw)AV%y+cHzqn;|>0{OZkhjy@_uY`|N zRE43vyVE<`e-~t12tC?fJpT)UucziW32TICU*m7m88y+V-+-cgX;vieGEr7N*PSwu z*@1sFaFzvEgum9q*52WfD+N`$KpX5Ri2J#FZ2V7*B$8zmwcLS%E=93?iwmM3iULQ_ zL%CgaVwFq&*>aAshT)I>)gTtGpm!vG<*pO0Z!?x+g$hNr$Vh`h-A`MwoG-8dZxv%D za*>h5n5f6oCk34G44hev`OSa9*RBcMP$H2HB!!Ok?~CR6k4K0WnTCbh>qwx6Vf!%z z!(@N|vqt>X9l1*%QECzeGjI(%$*>-*mo*uY67|{TkxLF`ckziWQ7I0S*@j3tA0VDh z{pU)SC%lvty0*#_I_D}SQbx#4CW30dqa=e`WS~rpObU=)c$1B3Ax0v+wT{_&SA!UZ zKq8D4llvn&1~N-jEh||8dQ3g+_g3nyGkwHTLuw)56@;L724B8XK+kZ#dSK?II9(QO zd#qn|eihM+_MN`8@2RyD>%fXNeam}pXexhowd=tvxwkLuBaPd2+a9KDDRsTQ7u|jr zQuz^4{R~)sZM3_oZkwLjh1i@-V|YJy@qo&@7$Oa$F=E%USHiAdykl3LsOvRD0l>o?rLH?LR-uMTsdY)2qWQ?gnd_gdi z;aHk5VXHCFU=TrTma^JFhE95-OZHneX2&hb00vp=zOX(Wxk)~yDe_Vliq2*mO?f`l z@~cZFLVn4Q$<#kP<^HW(t70jsEHP*_WfjBu%V|j?z_!$&+gnW^9x|FD+q+=hf~QA@ zu>=sbC;TwmfImSo4>=8S{Ajd1*jfN=B}xDTMt%?JfiWrGF*5Dp03TpQx=zq<1AcNiRh_@44(O}&q1 z)sLEP`tp@$Z%;_{Gflj^-jrq)c49wClLM-8lR6w&QxL@CY5eXJkB@*XKia$2znWLCxGnPv<`mnCACC@jp=Fd9?f+BCEOZl9N9Dc&!*cKkE=1F_Dx|o`!R&|a0 zqRRZ+3p*hzhvCRnst^}b?GEvIz76Ica(n|0xeo@1+aG{K2ub zd~)`g|AB2|fN$j^^bxSL;Yzl(pHgRe!Qm=}PJ@D+I{F3L6aLS-Tag;~+nU(5v)Zm>caku_EB|#^{;s zU8x{9ofag>j>&%FJY`ZdJof+$vM??s@y^qvMbYdM&3#ApY4mQe?y&254vYOHD7Jtd zyhU)}IZ=!d31f&v*4)QZlo`76+;n&-YxLb22=NwCX$M502TIc^mL_S(5VWO$eE!et zg2~^M_JI$uN9A@w5P8~y3u{2%03+WUgg?k#3KcdDo!-{kLIl#W8>?3(@g+*0?~!Rt zz37#ty5)Z6`_zQ)1K5Bw$eUEJ9I0*SFQz51u<=q|pWaSOUVAO^J+9E)kPEc$y0}-JO z`KY-{$tW>(3RuHPlvI?Ye4?(5E44(FtPYaRm6fg?vOm0xtl#OKGz13=*-*^8LgLYS zO&siZ!BF)UX?MXA{aPdbgNT)?**ufAjjNx%t(q@WY`aTPBDoVD6uMpdaej146=I7a zDvF@z*uHJrjAv-yJ{=(2R;*uAm&KOt5}f9C*W;@v5SM=qUyYiVCZ}fr+6PQdKogl9 zBl-7`haME60oYy?hiFG_#<7?^7pp%KVKnTyJJV`kYj!{AuZ-U&!eDwP(=V!HH;Bmf z==@$(wx|f3CQ>;4mGTBSTDMZl+0&N&_n7Jb{wke$yp5A`bp!xLNGax*IpBKf|4>iU z5=Lqeh7Bi7mCzDQBR%)Fp}`a~hOM{~Wf;esSa8t0>DN${r10|HEx;DaUs)<6w5u@{ zqrx2&GU4IJ{mSRXQ0T2FfN71mlywG_??a0 z=mEH~04W7;`&OLk^4+|g9T^{r0vG(*dQ0=WFK4}@WMzMLKgLC1oqhSr1ZpV}qwx!K zqt90dT|A=DJh`d%>z-feo*~=q=Os6+1gr`#((=7I+j9UFgt#0dbD#c^MeopBox7!I zYSYW|=evu;8>ctZ<<0%l6)%~vl1s*F(2yFFs*uq6<1S@1OwF$HHD~yGxc{ofUCG+y z#RJPV3vr`s2tUV5{)#g|?2lG_Nz#6U&A#1R7!s+!fMdjcm~fV{$>`eb_LM&Il3m)4 zs~fJ7@k0Wnn+bR?zj(V+bcj(mhuSY{#(`N=3X~X^<&Wc+_ldlgHwL<-LoO9OKfVqo zSKt3q3l2_NFm^&Vh<1jdztdSbHi4o0oTnM|VjvsrY6{5j7gGNjL~}H{?vb1Y$d;tA zt8CyaP>pV5(0#FMeSBGPy%GEnG`(jlqBV z+xP3?kOk<#hB<5F*yb8T`>_1u(No~P{51lg@p&=3 zku7<#Pd%%ok$OQ)GcvLa7qZcAiOn!-;iKXjY`NwdOz2qt z6Kx$R-d(NOTAW1i$O?d2VSq-i|* zba*rm>@>@*8hcTAye9J#Q*=C9$nFJ+rZkZtQ>XoQF#Z1CycqNHke%V|`cka+%Z;Hv zKUUG`RGuBUO*u4A#%i z#_RIQQZkHizx*s}p=#w6NjT<8M}%uu>ey{9Vw%wsL3`}~2Er^WCHxOBW#*r8!DKK@ zVMZ=yGgC7r6E=1;LoN<8V`e5pMrK1ZCNmZ$V|I29CVt-k-;QLM6fi=4KAgv(i~Q?R6)XKZSD z(`Cxd$dl}fwNL5WL@Htg=jVGTIMXDY2f}|dRaT~?5-y#HkIj))839dmh&fkFZi6k?mju(!W6J>vxYF9i)Lb#;k3-<;AwG1r+Yh%4Q95pWO+yO<&I|0?ZeIX9BNG+Jw;z}(EqCNPuE16J@bNnL-$}on2vDAL`79Y?+2m!qntzYMpoD~R za&$QdvFh<%GRg>mEib0?e_8;r>EjebmO8lSx&PRk5bwL67+?!QhNE_he;fT9eIzf; zJ!%dV9RrA`a;CxO#u6W%%kWDT^tn!T$cD@r62FHSHjI_moZ)End-5|vEur_fy(UzK zyj;P5g5XOAN*vGX_Zg3$!g%naX59o(AoMA_@1BCK^u$hA1eg!cdf#kNc^HEELNho< z>F%W@dp?c}^prQ=tYO@6s~eC!<+H*u$zzi^83V$@BxZMAf<|Pgb}OAccq07Fz5b!L zu8$4mi=|2AH3mOBN1q>jt=|a2`i!i453qe3vijhHXN&#}CdUcj!vco*9-)?BmH7wj zfsd$-Pc*P;?n=*fGR5~#IJxx>yzM`pyt}HuThp^pMEydUZcVJRw@}3NC~}XZ^?Nxu zSpbQT^r#v3Qij7+5=b_|^>U3vV=A`dA{-eFWtUcOWDh-~rM$j|?YR#uCyRggOqB5) z#120p-Yx<&+h$--c%a`p1HLv5fFxiz;Xyf4ic^02j8en4^KV_kORD&0Sz0PBGtVMn zWOy^Da#$<_N)Rtx5)&-mCie7xWYtf>qksuBGw6LZ7ye9<sH(Rp6+RH< z9tevW&DcKHDs&Q@D``$g=N%TH6$pPr9Q?}q>vv+KZYbYzBLWE?a(tAB9i_@q#sNE{ z-+l9*3D!Bgth_wSajNm%JNo-qB&u$pPGfeP!U@gmN%YVjis9x^oI18eXXTk&fbvP7pAdNk zCCwQ>fbq!^%t${lcM6YNaCJ&m&x)^!+RvmvpjYNZpBTRd`3;!Ye#5MMi=*AYV_6S1 zpK^I+>E&RY;%P}<1Ln-ugs*4cABa9<)Ml-;NdB!QAK`E1!kl<;rwz@Ix{zlI9nX8_ zWw#m@1H!Azz9NfP?>$(KjAjLfyAeJED0Yu8UZE4yzv^f{=stZjPrF4Q)PE?P*6L=Z zCj$9p3)>q^sit|~|G4$PSr$!_BEFio2Bw_9+L2MJ5ksn20gh=bUrPhm5&0tVVJri? zm*#t-j-stYtCasN+)=7ghkp^E*t-o;k?WEDFBYK0XR_baG{3Z3=kWrChtNBe@aH{T2FwEAve@Al$%! zZNQm<$qWah(P|?H{{O^H7VAZ_Thbbfiyd|IS*+)&VqgAyu@|%dXD``Za@ww?Y&l+j z{bzsn`~LU$%U^z6{`@z4vEk}NvRk_Rc<#@?thsKnY?mL8f8C4;e|Got1UJK9R8;<_E#O@+vWYY zf5IqXAJx0>FSn)Is?=Z5vM!*$rB?Uas`S3Cg~KP#@BeDA94xXe?v6v(3GFXz97Ath z8^r)St24JT)$aB`e^cjyLF;>1ArA$s@qLXvt@vI4U%tPBceRJ_l=LuGXk+7eb?;;iC4w_3$bu_-hQC>qk}J`;6m;JMckU zj>EkcoO-Jr;{NnE*q6Vk6Yb3(^ev2 zdmiPO2-srhpT?Cr(Mjxm@&^A;`~KoStb0^(h`8c3lF{>nVhQ+!!44fj;BI&9~keMa$sEQEJcE|V(Sx$ z?K}($Ba8DHjh+pQRZ30dto2FkX)x0>K&^m#mr2iYIdSCwvp;5sOa6$)&u*x#+ZB_k1bm; zK}esrlqYDqwv2A<(ZF;w1ul*?+Q&)x>WedO=Z@~;fZRoJn7%DeRCPl1O2!A>qc zsokH;xpw=LFNM$=V3{qjCY71_=X@63R zIXb7+uoW_F=%65#q|1zMPiRi{C}=Pnu7y<$)3Z^TJanQ&TU;3p@;2zHKKv%MZ{NC{ ze=5=P_~4r-CM3Bs4(d`Ext)jb?1A}!EFpDugGM(@Tyd;u$uYWj<@aDNXJciD(7_3X zPfD>)&4gtC0;7PY+0?M^CVf)0VW@ii)>Tgs-r+WShtOV%s%I7S7<7Qsc1o`VPZ+Hi zGMKi1*xCtS6rI9$UdrE~=BUQ`7x)uaf0&5s6K26~O$H9!n$h(_!Z#TSn;NaN21Z!U z4-uK^Tvy%Jy!%x4>3L5S^0;H?)jjq;I|qiGd+*oQdfkg#HZcy{!1u`B?vB2Q7~`|R zGtmSjh+$hes!Hgo#Zkm|cvuuuS0?cO#v<1yAd^ww&c4QD2<-17gioE%JPA%$f7mdM z_aH{mLBJ)fxJ}p1r?Bno-ED2k0l|b+gyl2&ws0!U2~|xQgAY@Lp0B5ZZVuTtlZOx! zTXr7;#}J-*vt&FMt#!9&Kq)7J29s60@^ZFmq=i4zgkm2(rM7y@PP4nSdO&;uC6cN7 zij%&?2_33!6^<&$RCtWQxD)z=e{X8Mfn=zL*`G0b^$*d2p@5^bHTG_uC5Lc;KyQUe zfa`7!mw`_y^ncQur|=f;Vl;S5F0~2)8Qeh%`X020s{QmLgc&7BsBJ>xD##k)L}4>( zMF7K>Puq>{qQ)eKP_0RLwqxMZo&em@RkM@BlArn%9+vP()%K88vguwDe*{t7f=*Ee z6s>iO6qd#TTZG@5CNeS86SRRDgEnThJ0I!j;!<4M20@#aoDjFFLrWupgTJN>i|J1) z;za&L!ObGk>**hz;h7Sob$NeVJH7|_?%pFT6dOFnmEj?Rq7OrqkC-XEIgawNzn1_A z%`Uwe35$)4BF2cFqCW*ie|@;g9V{^pA4V^(J@MpoATtksTAMXFVNDGMkcTkqE(c>U zXE1(VA;s`ks{U$=VU&QV75uuNVea4ozr!;y8sTNezB#rE7CW~|cSnzpPpV36K`T(9 zpW*IpLzHK8my_y7;Hd^%#%QPYo~fGL5Io@qLXsQZ4`wZ5q|v`6e-#>%oZMm;@D$I* zB`;JTt)W7t9IrEBc*xeYbry~nHo*Ca>x5+Mp%l&yec!MqF4=?7U!nYMc{-?NH7cm{ z9O>bnynPPeun$RbBq+ICKR$iST1Sqaw37F;iXKAAf7acuJ{4!Yt9ft&qgb= z+Ow#hXtmV%9rkl&e^ff~1lt_>{M^sqS1S%R%B>4(kBQ-gGAH;L$K`y~> z71zm7i!?kiCk}FqYo9x+GQS4xkli2~rAKVmvf-8zPCNn-fBl5v$?gP;E`q_%9aV#q#f47mfs#tKIZ3gC=aqiCj6o0rGD&kmzu(B*>y%T;t>_dW!kmXy0gb1|K z$q^R}BOIg_#mR`|ON*ia*+!@hRq%#yI5|j9x|5-_dy%6}r$~4iPy zf2L%;xt(R0gl6!;`)SR?BF`8q7@#b_ye_?KHy^cH-)PoxI58mG9gax-d za!}p{$xx}_Xa^`uPZN9$YI-Z7E^l5qM$mp`IN{A%mX`&Gm0@U#2yFs`UV{tBNu8z^ zpy!}Q4@)r6!J=dsa-bJg3PVjq$5kavXrAW zeawc3f4oF#A=`QuGjI+%g_An9~avbF=XA*5b! zgi@anVQAY?SVAZxme5vpRO=NyBMP68MLRoes=zsH`z|;L>rplwp%aCzpZ(^hU;B)**Oy(NPg^ zf43nW_hh4?6s{`H0(I_g6spWZ;!?hDW%C>H=UKfYu1nDnv+dKEk_*Pk3hgA zCpYU;35!CTNmu}SmA-ag7J9Rs8R{(HcAc;f>R#pqMLmJASzseoF$5Ya^Re^B^$&Zi z{BsC>y^No|lYN%<1336y`{5;gikqcbe^(o%)uOM@Im5}tJn(Ia5(r8bs_hXOPF|Z6 zrN*x^@A)Cp594lg9?GcUi57e(8;1d#gQruyGS&)ch*xHB^5|#e{KpDlaZs=x(c3X> z$wheL7}iZ=IQDlv7&2s+ECWpGM-sV2P_D~~aI8j9}J{Y;~Ay+-$ZJ1-SA`GHKj0}Yz!(&~38KosIOJ`|s6>hp#*%PM6&d(vQ8HxellGjdBi*mc_;^SvKks`dJOlE zC8kuf9kMmr5^EHWgtmCoAYBEQDZB{8MCB zYjTxDIF2rXo{!93DO1I^PU3O3YMF{}S132X`53$G1B#3jR!31-pjzmosBd#u+Gr&y z4SPSvV-E%GjlA3;+-}oo#oWrWW9!c~pY?BE*ol=eb}G5C>K>H!e?#;~c^?}_o#W)2 zk2uh}Ln}x-lDZY$C#wpZPOf{&%SCW0G>T}Ctg*@BCdE*ufUhwUlcm{smUKXEu(T2@ zgMk4aBJu#w@cQs-%QE^$XQiXOw~r*R*FXBPrpK#N-sZ*ezNFhh6dQ0F+-bkTbWm>l z;wY&!BiU6;8j+5ej>2(e`DQmVDco z^A~ySCJcr&xG_+YV`>!?hB7dv$IVnaLdr}QgTT1GdKsw~^h$AE!+n9%;&|i;MD(U4 zCBs?Ip>i4kDv{pBOIB@PIo{@kCSrkHn4AV12hdku75U~g);@EwcRs4)S*7^^R=$F$OY?F_4?-7U#V|sh@ z!&Q+#ls4a-23pU7RW7I+>f*!pk{829XDF^UD{yCWe})2O68byF$Ai;%jGzTxsh>VZ z;1TOkUNR&)9lI|C+e6Y)yrC}$6xkPBDD#18P2M0$%2U~<$^@ulOssukz~V{;#0wda zm+4M#<-EMM)*^NrZ!-udFULy#MpX!@dLZz4Bm;|foa zcg$=We@e8+9nCjbWKK_^7k0dedNew!Zgt#7UkXMb80|FW5>yZ=+;F}SiTCk$G0k(5 zu`u=E8kiYx5Ctngeb9FZ9;#b}z;w;Yac&MTP>9@b?|puWq55gBvoE5cBUn)?CP4y_ z0;@{@d0oBgSMGk}o#S(P8=i4k{aK>sJFN5Cf4lj-mYf||f2b;shlcl(r~>PFC?qzc zJRp&Vtl74!EiT)mtAIxzII!UwHdznSU@$j*;K0n^|LCgi&UXn#ORSZn@AWhm5v~_s z{!HV?2Ds0wI+NCSyz{^{Jvf#OUv%OwDVD+Z zf6*6AYJcI!6s~1efInX61nfA!6+sbtkI_>kZ|{#l?fIy{K9=_j?90mn`z62-%T!>W zUliE4T4n&&JV(UuZSx$LKKY(PrE=bu9A&U)f1A$oQV3c#hIC~%;i1be)VA{`s|Sg!%b+ut z2_RRx^dw`szH3Tm!0-(T3Ua23HBP)8;k+UZIAVn=2Z~ss3QU8nR$QZzv|be)$kFFS z;V0RCAN*Omg5GlK^7I1s>T)mVU3<=<9+`9Qy4hWZFT2LP?yIsH)gzd^zy7%V2=h<> z16Z_bahK8k0Th!QA0&S>F*YzhJ_>Vma%Ev{3V7PBTe*%jw-L_$6!QYpCW}`S7;Ip` zHsH*_pBWA`e`^#u@cSnw>td1Jv%9h_VYS=cyz2g{in{&#=5Mb3>u&{aCD(1=Ye~1C z|7!1kYd?Rz{prW;kAF!w@3!iy+xOd#S9^c&&uOo>AICo)&gFmFew=gur*je7&*5BX zzv1<942d{HX^Y`&KJloiVp8MYcbkwu2_;JUj z;YC~ask!}5`^P_j{O$JVA4HSC`M8Ccea%c$teejniV)T2g%>Yh5-%_0;qNCOYPTc$ z0fJu+0EY~jXv2Ru1!N=Qh1n38xB8}!DUnuzCw*1k!a9=n!MC0CD?!4(`Q7KpM8`z4 z8WAifIzLI%@Y`28Uw;Dk8TZEmeW+(9jEVlIyzjH3kitsUpCORQkz63Pi3bzqVW!I0 zu+@-DZ1fRTsc|g-irRO{2XI?1C7209z9Tgy(Y)CY!Uli7`mD2Ky56R40FaZ8v(%z6 zofru-nb@jth{c0%Dbc*n%_K?@SU2}Ev;P(aK}CdaETPP5G=IGmCT;E6ZbLyb?-0w@ z&eFWW-G?p33PNAt`GGaO37XkD#E4>T&nffvK8xN7;azQf5JmB^e?Q0$RITv4s7yXY zOXL4fsp@~0a@oVARN$^9-CWCFBIsB#aU>YzaCl^Pn`-zC&Z5jpq_3%#z^IY9dYg zF?Pjo59HUJ(J6Q*xn`fG<}#$ALO$E)o`TnLeN%sw!jq&ee9ME7!phqZ{FJwo6!48$ zr&xG90Uo-I^e+j0kbbwaip5GOMMt-cpa?}jB`Z@omX&DP{8s&;v|tH2mchj5_n6pS zXn71ddpfBC2P7QaXz~Ja&^x-5@547v*1mt%nV|)3hO3WZb$v#PZ+tt%EyrNQrw+>X zc(*TSfxWL>{KpQ)d!h?`5xVGJ0Kse?YYm?D6^eiuXQT4d;y(6|0H^Ku#bm~zkW7Q7 zSJxqs0BSKo4Be4xn(Pq0{3RGw*o>=jK%spZ{-MCTd#)j#zym2r6HF=w&4%ZkVOW2) zQcN?CPTn#sgxMtQ&2LmUVYtfzG90uRW$)1=(!TLHE$V z3J7n{_3TCEd#QBp1pLMCIywP)TJ(Ri?YI8~)cfGK0tE)D&kc{Jm`<+qIq0-di3tde03iJa_e zt(aMTge0fvGT|D|9kP3jbl=1ua>;Q9VppjGpgUMo1#xYH64pG7P9sXI5Jk|zsjfxm zjJ5)G7~KGy70BvRxDd{d)1Fth6*D|vc3)H>2L<1#LUzfyD?B1TB_e-LS2T`bSE`YA z_Y8K#Y&rQ|(7`X4L4-i{F}re1i(Fd#Az5!W&PZx50~6_rQ{uN|2@vy^T5je|ZA{dU z(DjTRhFZj=wFAcCh>Ko!x(#d`UVMElJ0GKtGgk!q=>1#DDkHUS2utli3$?%) zPwD{TLPpq>dSY8u({%y4GbI9;asDyXREv4KYcukcjOmE_by~&s(r9)2`|2J{VGuJ@ zn}kr_FS}6{S?UA9gtDeTqYjCl)?M59MCkBg&!yLZUp}j0X}*7;{<;f5(gH14xH&^X zR%E}#76QKhDc?>&!IYLphG*mJJIhWjkIe~89!va8&dO{FOkvNAJ2dq=G!wtH#FDlU zqp7VYmQ{AY%Y;ZCdBmX(VFrg48)k1!H@MBYnlDZFa~-jJ$IHFrkcT`?$S|I3H;jv? zFW<`}amtL^c&~r1B{A;?=Gl^0c$(5Pp8?G&YDIlIjV7M>R&>>>gJS>aZ4YRnf2?Rz z65@-E)`64XDOZ1e5`z&j?EhOy<=P_lwpl7MO(lLzm)NaADakf0_lne zSrJ_~^tru39;+M!^^bLILW@7B>PxmojfUb{a+3<2Dh*#&du!6F9d0rW z!BI9uQ3nc^Y+|%b776iN%_;FIz`0hAB^VFJ9j_H$wU`M)O)VWioegY}lPpiPl{#;V z+Mo1Q^#6aFk<^ic#~9E4aB80l^tftZ*lcNw&0xu~ znj(LicqM}eM`Pz$>VT`WaP<_2biUCo#Xqi^S`$5e+M;(SgLlg7Ph!UTj$iN6@c?0I8f4kPGT>G z7|P~jDt^iia-SGD^Xu8T(R@SbUOo$(n5TaT<64X#28>^%s7b4!9qr}nKE_@i-??S8 zuCK6|PWiNno4#Fd5Aa!xJJQbIo7)4tM_!elXj!h<*oGliXwd)g?GQtMFr9&rx7cFX z32UiJG9Xco9M>hp=#$|iy|Mer{RY#v?$-iLyv7p;tv-19t|xw)PKYr(+Q)W4v@Cz; z(_<;V`Et6qv}7-|soX!hsC+m>L3py$=A&?=G&)6@5@e5@%L&YA$DDfdofw(N8X zfc5U4%|U})vYZ^4Ub}pIFY>dCnd|LNLAUd5(DQ~Cik-Sr_DW5FZ`zFmtO~ZeMPAOq7f*syeSz>df?O?;D+m;XLiQ}8Xr|J4& zPK|&CL3eN8&fPy<@4wflzCZ4;v_^%lYIJDv`)0>O>)H3I zz>2vkPKwf?ua@>tb!pcY`8~(2y}L+PUn3)oMf|~aBRX@doc?-u9e~H3TEKs%%HN*K zsi~a5Fyb-d`gS$;s?n`5c6eAZ#jPcqSeDAxC!EyTRn6?0?&NuBzpl# z95-r2S*mSwp}pU>4onXH^-hWD>*-!=(JZ>yOSy7s-QJeN5Z?U{RYi_j_JoJ2 zMweibtcf;Kk+C6Elaw6DH{E~z!(NC70Wim?6VOF?Sh72@9@Mal*f@`6D1@t=FRE(w z?!JWzXPY|lH-tCc+^egS8SISZ|3J99NUPC6Q}pLjV3^z-M=#G+8QST&xF36QKe97G zu?H5!4i`m34t0Kt5&&Tf?Q`n)>be&SVVCgmi~QdZfB(aO0NX{bjtZBO83GcMZ7(K& zFfcbhJ_>Vma%Ev{3V7PB9J#I|xiR-wDD(!emn`0Zpum9FfU|`svmH#2MozrJM*e-1 zEEZY3B)zdQ5T3;5QLOvwx%_^S7ZLx-i-U`^;qX2T0)D3+4E8|BCVedj{FNFLSzlZo8@R`2fK9q(yQ=RbxX$1e$EQsS4ZvXu9*NZeF zLFHY96_3$h(>Avc2=NCz$__s5snY4XJbDerlS8b;g=vFsM=R1UdDWl4^q7=?@-W8Q z9+!3S4Q$GAXmGNCpuX;q6R#?vTt2~*MwQEOl_wesn)5IDjlQ`%?x)Ls3h2ov(02Qf z5Ntw{zj<>-7}7KiZTAdygcOXuwuNBa1`2#)q3m=b5D9>Mci=lINAvT%1uyhO{gC1a z#s}ilLfdpS-L#Ojt?8GjxWs9Hge@~?#I!LPT?V1(?tYeAM5Z)W?BR@q5qDPIKCJcS z`k9z#Wl!1w37TLO<9*USMiQj`;SmqC8BL6Z(i!PQ2&%xM3)1P-jGqEYMtp`1&L%c+ z(h2+|UtozBUBa1hEgC0TK{D&0X^uR@<{dtih#Q39RNP7(K|4}4S{AH-pABJzf?3t1 zBeHymGJ4WzmJ9^pjufWr;HU-0E`lx|FAKpG5v5ISVQRuC2>{)Yv3uR#3A}fvO~H)2 zb0DCImx9JpxHg?b*G<=uHo%5cr4p{3V$K3-;lHdE?EowZ67y(DKjljguYJthIG!V*uP?kuT5iBr4jdhSOmC`ld(utqlo(l z$p+y&iFBR?_{0?*G2>qqTNUx2_i{E#bYu^uZQ;^vFA4!iaua3~0T5^cAs;U|r;^^o zUpu`=u#v3#ro*%J0Ujc)kZIaiX43>|+QAkOPm?LCvL8juI$IcjgifqWNj*Sh%@D-3 z0YQ3QL)Gn18K|ffB4H6Tf)@`i=#Tgf48K_hCfsJVTU1;*d(T(q!H>xV$w%dF3r1E- zN2_#YYtuJ*-Sbj;S}sExvLmf6NY`So{@Kyb%-9RpNuLdO>g?3m+DlTBnQErP;2fm| zaa8qQB55K@`~V1lnV;(PeIf_QBSb}1ogzziiAk=&9o9Oa8i0iTuZ^0M#pnJd#{wy1 z=g33pnfz=Pl-Sx%_<_570x^<2ClH zqal5N(}N1dXB}osxCHkX6uXb$r)xHYj9f#^#$31NxU{4U7(;RMN^cFG*923OHXg{; z3`^jlUDjmmuG|}h#|vpn5i;or1uHJ#N3~yl_9L&jbN{HzooqR5(0Db2`nY+!$!+A{ zf?07EixbAktFcD2`03ddtue@F@}Wk}R+)W&8+h@tn@;IxyL2(hf^tpB;S)g$U>4KhA!C;Y1XEi<>6`M?<=DX2;w-#h^O2sAL=y|6uJA7ir&R{@W-|{a>VxENEqa z$f^IhSV=gP0}q2~w9ss9i6zU~v{8(Vjgi*VMmnJb(#ExZ@G?2+%$0ez_1Am3Rm|>0 zw#z8H<&+0GAC&D$X(HwHDh28c>7e{AsqLKSku?*XZ~XHwCX0&*!qZ(rA8{+$rGwK! zIai)jo>xZxl7h$4U8269m07_L{#U+#OHOkWPM+e{%9YK|qxshPKzGLjyWF#ttF$j( z4hX^l;I^sEF94YIoodGJ^1HfyaEuztZglJoK%$%4srA^oUUQ$mfEmmU1Ok+Q)Ddf<9PN(FZ z-cPC19$eMAjjF45#3>%K!z?p!_W8`pK?vRGfNX9-XD2o8uCr@ji#2(+KKmrP<4JU= zPy??+AkSuMWb=0OIgTn>{isU8!AL{v2kQ&)S&Y~G(^ejWIlMH|G`9t+q0J)Df~qy^ zE1Q+B8Cky^-XR;{UDW}+WYtQ4IRi5TRV%uvPw>;KYKP`~)|Hww7hly?6z`llM+nq( zW2z+5Jyp>u&B;chOO@=NQEsIB`x3$>)-oHeoSXREu>$0X&@Nr=uwcq0p*lBF6E%S9 z)XMhm${a&HP%;I^s}kE7t%ICrD{BJ8m%2*6wVU)*2C3qJtQS?>8!j||mAoq0?GogA zaOuvL9*Y`thLpmeUVa(=wen9qFs;H?)B9b0g_kCsr(?o^5S-^+S*|*FUZIj7}s-U}}ph zD(SECYgGmAutC`dN9C`7ux=heu%R;4=S>GXE)`4`c-T}-{OKAL|qv2*{m+x2) zG7f8~cz8+HWsRXDe`(`qjqxd);Kiq{aA0oZC{daxWSMQt@|iO(li%ybsGWRQ1Y|ei zfiSLnX%xKdY937aEG>-}*KzT>38w*|6#j!tq?Dh)$N}XmbJ`@1QubJvJja>}e{8(KeYk zQq4d6NdUF;Y(i8WVUpNJ=}KL;sO#6V#XZt2YlfAw9T`dXDF|onSXC`d8jHBEY~Pw~ zeCawwL<~ut>L3Cp0f&B!g`U%6Su~$cqZU}|6pE*!ceUJqa+f>flrZ~cT@%(;DfTce zGxQtGtz8dNrW&!gR+?watErC1hFx5_JU;u zP9}1sA~6DCd4Q}EAJvI0GqPsJxuztl9F1^eIGQ%~MgDS>X@Dhf#w%ya6?(m(-)>JJn<$x_aWW z1CnO)o)|bsvq3He?Pta4gRjB{HHP(GE&~a<@S-*FLLC_#r>%G6{Q@Ii zdSAJJ7MAYJu(g<>FmABnhc>8n$3shI>lTylxQKIki_XYX%G6xqFj&uw+yVOD6 zsQSnQ$hB+1ZMqg}rb5!jl;kE(!zZKrQ{XqBDa6n!UYhK|C+xNADu=9xlZ$1$a)o#7 zKvEle$&!dNfz@w$a}6)MTWHmrFtZ4)HTDh23eAdOxE8Z(A37&JDF-93z*U@@Op{?f1&Fc%zZzuAnQP*|DS*9ck zVmEeI9bl|d#qj?8zSYH71h#IdJ7n^1T9!*CVySHiKV4k%`@Tg!?cRwH3T|ej%^Tr= zls3!G12^H74BwdYEYK`U^p2P4O9@SCeMHwnO%B+7S3#3w4}^p3+>bv;ezdk1zj4QX zTr<5fFjsS=i!O3dyqq&4GKBhi#jUjU(r9Nved^7=Qs)$^ja%&lo-?L5gT|O20j!kT zMYy53Z}rK(Ut{Xkh^sw{XDiVPJe{w9up8$TOe8Zz32FjmT6!YCo)C)T1oqH52wGU; zbq`L{)Eax8s`DmZyr+gU+Iwm>hcdhuM&9$F**EN9MdWR%RbSn!?=o#L6nQ9OiG>c; zi=OO>dRuQRab3J*!jorfaCefSdJRiioTy`@q}p69e>MJ*)S%Bl~!Yx$hZULwX$B zrIzJ&7~XL9-5bs&-*R%c9$kKaVr|bGyF>3(piL9Q3v-6OUftf#63h2Ax$2;@0KP&g ziuU5p4sje|ISXQfCuuoXp#=_E-zz5GDcm-JYti`gckXcGHJbS4S?AL-PYX4HSo;CT z=I`$H0Z($yvu~O;Vpf%nZS|s5BK)$N;N`U!6z;p=(@YWD^epb}it7e{tj`8K?*q#9 z124yBPlxnd8ISp%&xydHY^WSVT9RJcXH|;o@sVF)74;R?eL0-wcicWT9nR8wX;M=5p>T{|iFRG%Ju(D+*t?HT(TbXMTJwv39NS zSG*dKPxEKxa2CY@%iWAq-8oB|^-WAoEk0EbHGWX)joMQ7;nmc!s`&euwR!zMRD_jU zhH>>tbzIi_zhnGxW^eqe&PYgu=r08sgYrvsm=EugyH&7P|C@|o|L|Xe%&q>Hkr@IL zlU_h2e>65XF+M&Db98cLVQmU{+NE4Ok0dc}&iyMUw1cUWL_LAV1P1O7aAsgL+rjoW z#*qX6|HP6gQlv`N-E+obf$nORL{WTvd}u!Y_rrde`p( zzkmGeuaCd{&3;&Wln`vpAHTlq^G|E)_gp@He?9)=W=!}t2AnbOUO!u7{QV0|eeR%s zuFw6$82iVVSnzuvQ=hp%U!Nh{Tl@9v)7x|tbJX9X!+E1ICYi73Zo7Hfo1y;VbDyWL zoEkOl7Z-0hw!IS_tsoEL zzx*wHw9_)hC#&=ZE0hp**4XdJqwO&CWw^Y@aBBTme=>E^O@rRP zHxFfw_R=1XR_U)Sah7qVl+!h+T8uKa~mBWPqlv6reL6zvp6k+Ak1^874W1h2s^4cz0iLPH(?U(diLYV z8-4?R+^M6)q{$)QU90uB!r?ss##qM{u6Fb+x z6rLIU#mqJy5h9d1^@UI>T5pI9!KTHevzM_$xu%eIZP;=c#_ z*fa_>nx*%dK7ki;hcrUd+V>lVm^^Wb-b7*91l|4kb44I7=MxZx`+~(RajS8AC$Ow3 zCy|vJM(6d+tZB8le=t~sqDEs|>4eRP*1mwSR>}am^*Evo_`Wu{d-pla!Z^L*)yEqM zETk3T$r{mp!lD^7^v%ra5Xp$(@`noh>4D%1pKkMhkQV5=(;#y8-Gw7^&tKWCJ`g(G zgh2!`irgw4hj-HC zc_J<$(un6kjD*y@3%{WItM`Cb9qr4ItC=-p!h$A|zOjtvb+r{6P15xYXNG#=w`qoH z-Ru~_;FeD@b1Irk70cO)N`)&{h$%_TE_x%cf9Q(4G^|D@;Ab#27I|&ZRFHgzBsQhP zr5KlN@Hi<99weS`lPAN(GHeALO!akHYZt-j@OqU{n%Z{SRG}RQ#r7_E?SHP ze`pCmir~;5;QYd{?j^&ydR)%I8Ld`!5K3J=Z~zh#f0s6ke(@rWJ2iuT#j523iSGjB z=f$a9CTr~x6F9FVEw;Moop=zaHZ~kptD4SinNl`HycrHtLO55toK`o|hHiD>Y`LcFIr6dY*WR=7nvYQew{B%!0%jan9^SeYpqhQNf3~p{qmrR}{hcmWST4%N#Bjyf7}#>rw{0?W43vI6 zlE09qrfG{e`=FVPn-S+|#wMeGGrFDBupIPn+nQbUT(g^JFKTs_-Q3uO`;iu6f$lp? zkts9OSJDUEp2jIduOplopX8xbyVW@09)3re*c@{_H1!}5f^1f3+1}v0f7zh^oEUHG ziSn#vKJ`(q*IjFjBA-N9O=`s$0V>Lx=>VGPr8dJ9TJhe{X%WY@T~6w`Er97Hd!>-| zow|e;G-!5tJs5E^p>M*Lr&>fRT7@fiGEbFuWbKT<$=sW{oizuaZqKGS?dhD~yA~5E z-c`qD+lS%YpB11>g+5uXe;eH3HE)=wj*i!6 z=)FgY@TyY<+zG8wxl4v7O6dBCq_7`7twfG=;hcMY!+lHYICG>L?3nguwBvNv%?x-% zUoHH+8{0)ycbwUH6o|f54yL23i1)je#7-UDCw5;U$eL2eH;|jWe@hNs)9*GyAy9yK zdU7np!<(39s}pTFYB$)o>VFgH{Q68u_GEGP6F4A3G&{nyi(Wi)*i^DT2V1`h&-c1j zo3D(>;t!vJIrZswzC7>Kl~NZ4QdDmX=Z93-+etQNPu7U(tG?8i{jRWe-B&vnNM#f2 zbKm90T9VMB-7Zhde_In_q8}(5j0nk!cV`~7{i4!i{o5On?^tJ-jgfgs$M%|dqapyR zz9|ck*r4`LQi}%Y;-Wz}qpU)Gc%ss3nE_xG->0a9tJqF-s!dbl%!{l`Hyl{lv@$Nz zox_+!9B*-*XKdcEUB?Z%I{`)6EW8b$Fmxh>$JM@v2(Djre;X5Zg!1v5TN>p%e%+w5 z8C%+)c{dqLQ|jC=x*Ao}b>07K_?W_Nda6b^L9Nb;6gWB*?3c$vN;|qJ<&1C2;fxV@ z!Dnu=cgBM}(QhIlE=3o1zA2ox9WVVj4!-i^ZxX2d_-6TV+qaWH%eK4)!AdzbTZ}RR zmNUtgN_p&`e|y36C49mmJ1h`=hx_*#Bt)NtlH;2aYWEHc^G-?x-+hu2p^0=ylo1En z4^cS5ncm`e4V5RSa#3U=G3=Q4ffjS=4&5u9Q#V0A!J9YMTd&jjUeH|a5+iFbGY_EJ zKDgk+ox}Ide{Apctv>+z~H3Qf;!3MJYB;Z;9Xw-MEv&CM(aJHPi&bg>cU8Jk?*EQYFYY zyOv;cf0C*pC}%47N2h_MWLUWh%J%ZQOO8pPt|9WZ%)6b1crWC7FVwQ4C9QXfvdnPb zDBn>MnEc$Z^~NM3S%W|QP|e^egBrrbZ=&HgD)NoQiSX0=Q|YANj( z9jy!x!&r=;o?J6#Hhv4&+@BlPfEsoZ81@iV0ibR+y zL2gEdq8z4Ozq^=9!1K zHG;b@HD&2emAP`P$HJRd2D{SoQWxnSCvCde=iM6fy6I%}2%c{y%E<|@oBvlf%oLf} zElaM+hIcwxpt!?F4_@8h!3H?pWWuR*kk|3r==xbP&PL9hL;pYOoch1Jiy z@RNj$lfIN@<5^S!h@R4kj=v)ED*uGM0>0)&ER$CuzXw|nts zblrLYMr)T2ZK^?^HO8-hN?<~^6QSE%z$I+ix?$MO>v+5PGvQSXowFG0K-2(R?4sW!#4By^L%?|jWNP%C z+;R`OOlw8#z~u1ri(|gcW%GI+8UJL1Z-3>-pFqy#osYo23cKgyEQ#15R2{x%DjLc5 zxgE?Ho&KtG-)*3vH;S?11LoyvrfJls-{@M(cMxVf2aaHs754%3f3Dcqsy=PxNY&Hpw8Ylmz&2$flV2%oF!qlm{JyDW6t)@?pwF0sn|EJ z_ez5-)Nk4h>4IIxe=>FkHa(}#N;u^p;HsE-(Pk3!5mJQNiF_B8btkc^~-=z#2=Um?o)swC^#Orl zPd)N;?3$+F%Gktm^UHLiw6~wNleeDqO|~{6W;jUJXxejG;e0OC4*!Y!sQ209YCCBH z+^k+ql~uhOa70){D-C~FJ_{EZAtJE9khZt7SzbriJ!E|;drfO82^83>=@NT+&V9i` zpmr`owRE!huVW6N5;*q>77-M0l?%p60^mWEzQRfYK(SVM5U~${w*XjP4rm@1XwTLW zzzNa~=WYnID+Yi$RRMSn;toLoP`xc~!?Yth0L0>e+i>Q@46Pvup@@O?2Eg)hfKpuR z0i2NZSXe#)=&cJLgyR)J3;=;#@gUg=uoM80j2k{A0RYPi00p|^LAn(HtpHG&2Oiid z6A%De^~3{jCBrs@NWJmE=Y+8DAQLIDm;lh>XQ-MAn+Wpp!GqLd!G!}r(!O|Lj&#@s zP`Do+*f9@a17h~a1D_$o`hW%k@WB2kux6m@K)eRgAVC0#35W|8LxWueH39J;f!Y9X zXi^Y9q^1af0RVY^!2?ef142N?!Fb@jQos=C@GBlDE^HKp7K+3}+u@IWakF(5!PK3K&D&|gV(_3 zVhMobwG67)=h^0u-Zm09a5Sfcp<0 zu-zvBA-JXEB5XU=XqWvrMqpu^zvb!x?;u8Y06PGH12SR{zybVc6iyrfL;wKK zKTg2Q4ghp;g)=}9Eb9!=1^lDt83N}7kOqK*x&bY4oC5!d`Y)0e-2fueoc}+P`2WMr z|00S1KPCQ~Bms{9OOgO57)p{HB+ni2j1AyU#G!%oh5*Qr(As$pa|7?O9yMnZ!Bm<6 zJXk>v@atdNI3am40Byj(nO6G(O9B8tMgQ&2S6D>osKi1s$xVQz1e<&SV1tw zf?&)lSZ;7}BETM68#f8S008j*mwn^n_&creA(gSP)Brdx4t}si0s!wj5*#4_EC7PG z`_IOK>yx3w0X1@L0Pt}Nl;6Dczj(d@L;D(P=73;@G-x;fFFa$?q55C@2G+}jHu4`l zkD%gz@MO=1N>Ds)!SmUGe>Tn+?2-$W{^6;c_y6#mhf4p!(<~p_%s)E^Udjg${KXRs z0H!O1egG(*VUTGIsCfow6hX^Fp&S8s76WkqL8(y!P5Nu;z%iwOe^AbVt;?W|{Rc|v za%eUb$^dB3B;vwG0l>u-(3JnOaWY_$Dya64&mVuK|K?Mt2HFIas~7-yr3Q-lfB0mt zh2G*HpO3ZB(*MI}dOcKy^4SGmZ-7ew!>36Tv>j6jAQb>5?!Fn20mYHC7TObNOmLI1 zTs+_^W;lQFO*6F8iap>3?AZ#Hpwo~BBH9js1pv6goMTWU#|y3X@5=k{3K8o7T%p7A z|8HPz#{nWzeE;bCf0mDnm+PPYKfH4B{a;?W`2R1jT>SsVE9V5@CnGHXUtocPf_(pb z?$D;gz=d^CfWeI#f*D_<&al^A{?Cd^+2iqFAV)#ko^!C6En<3XoTFrXAEc19!ce zt1L#eqR0;Oa`JSycXzn>Lop@oSCU(Y7U|}sSVPp!G2!$kdHvp|A~2>NS5<{3kSWLy zv6HU9yY@QQ66W?Bn(ct)k9#->tPnA-^#$KJK z1VQz6N2VJ^**D()#A6p)9tD)%iAN0Ir92r7d;3~K%yB95uUMZXK)rx^r&P@f@rv{X zPtoq5s7r8)dXYobofUg6)x5530Tvg-%rDhFtUExpCRU4Fv9|X8Qfw2ppy3cX)u)U2 zT8us_h&l^$a3k9!gk)B=m`@3PkDY?+t-Hh7cDvrx8gK(&EWWs}mQU=r zN_kp}6COnDW_*!Fqliz*KD9WEWho+8hyAe>Jy*EZ4esEpcrzgRdHwRaLIVkWBobSN zyodR?LMWKg*p`V_v)VXAg6SYlJcRzqFGvXZP2HlQ)#Ri>BES24cAe+Qnt&6!I}+in z_ynB)_c{mnIU2r3%`@1aw-W0w-AFQ3(?2FaNL}_665mNGAeTCgfwJ`c>i1LnT~~bF zg^%SYdgl7Kp-d<5E?*fm`Ie6{b?qxQhs^J6sIeh9A=3`@3tVd>ijaX)>(S!xbK^ju z4WVZL+e7NIr2|<(`?r>SOP2NCqZpE{pJFDb)RCVrH?dOO3`?Fp4M(&5`}4pL#@}`( zI!>+kFTJX$X?e~KQpaf;u!P9+R^;EowA}J9VJVpR`o4jXG!bAGy5v;SJ-B1Zzh2(laO@3a?Saq#*#^dCRGU*>=lx!{lzqZPhX3K>{^R^=*XDI5)9V4mX2Wb) zR$lM{Lh;Hp`J(&se6>o^)$#n0894heym6GU(HW4sf2NnR$=DU(2aOLc*7zvDJb793 z=d-uOBE1Y_jrfM+v+1e4pt>=b=ucnyB}>@l)6kG#MDbI7(FY%u*5t1Enbs$Ju{!PN z*M|*}(XEkDf}WmkG_Tg;Z(T`RrPe$ zm(!KP?=E+IN_pgs_t0RgKS!gRAv^P%=XYO3nF8LI1iQXooUnK6`B@N|o3q9b1c-T< zOVC!ZwQxBY*>P~J)pY%Vv}!U{Nb6g#sp{F~Y_-wc?yP^)Uqzz+qnuTNccIk>T1geg zU5$9=z^+Zqe4Jt8`BLg$9DkQAfo%8Kjn|}jdq;;`EZ#SLTm`IwfXv0}6dlY$GWp$* zzO#;yPsumJ|7@NVEt4h741v=cv3HSbOVJ&hQr?~)YAfECj|lQ!o;nqFIQF^m6|DJEjrCwF~y|3)nit#dYuE z8VuP6`ufahUsNC)7lLNMy>vKl>!9;n!_$U$+3?E6=5Nw+Zybwqbx^L&H)FC+yFMGw zQ&z<-v3k)ShnBHjc93pKS=wK7+?~QfBZmOv(7qu>poW;bfznLq+AUi@|`h;Bz}|e;L&}MvCbSY$8L|KvpiE%lEOnI@xlLjl6Xi7tO{aP(p1=@icKjvg!I4Bk}|9x zwUIpa*OyGu$S_9_GzvzS7pIJUMLVx;(!z@lC4)gW<8^%v-=zm z!>~oVB3*vw9$x1UD-hfbsOXd=E{L8*!v=WWw9yw54dIR5kp!T1dEnHe);LK>v}0favD+7zC95TZA{112hLL5KnnNtJ99(1% z{DgB7`8X;}z4im^%$ePAY(}c^(7C}SnW;NoLeahXXa^V>ykq>m3AoJ+OtK9bXJJ-p^Rv8}frts}${dtV)F*)hIg zLZ(gTTu5`{3V!O-NW93C#S?U3UYhK?0<=-OzqNf3q9u)fH+zxtliZ;L&Yp&dp%G`>!daTb}ux)6Ql5$G#s zvYJ@1xZ0_m`dVb?+Or;lT)s>?Px!?$Gw^0(cMaoLW@2`&xPNSNy^V9tsF0GU;gxGV z`!FMyxn4QRAvK*B+#&*|AIDn)j+OU{E#7n}90_7%5B*%kaA?+p1lpTF-~!D(xz9;c z=3Xw6Yt->W%*TF_*n9+MY3Tv+c7W|LD4bFBbDNjwG-old`!@R^Q_Ydp` z0UO+}M98kiBVp<43TE9=ZgXAhurc?AKMOOv*7^n3`H4A^*-Ov9zj!39=q47?I;t%f!s#^7fC!7e;mJdrF|F5m&v| z+Yj~v88+{_PzWMr4cB=f^lgPYKYBrOEUo4j`MD9SA&#BaKU9!uPx8>~9drW%Q8{T~ zB(CeC0iqmtQlg`H{dasS`(`*CAj?7B-F$dW!L3*eh@ZUa9#cHaS1I(i4E_&uALr#u zy`_%=h!VLfbr4lT)Nx95gyw;au%2cLv5n@tbvhFML?7C0(ilH2(oRf#PwIW*yS1i@w?{*0ZNWEz{j*F!_HKu1NZ&H{@xIVOCxpt!0~I*0S(d{5jKJYt9$iW4VA z4yiWf#Mr9+jsgiYSIcS6#i)FgaM-j&Lrj#tK2(SG=6V$!`w;<5$i^{n!^gSFaCeR% zS>{wLtz0ie3PF&oZKF1v2_;^mqwfAFCA#*p58P-xSxBn1yhZ50JrT9nIp|S~S@|Tu z`Ffp^mKFPvoU(#Wfoc-wZfSq*$-F_?&5p!bQ+X05quO2SwC>q%Z&EHz-5WbGQDc+c z%}g5xAN^pdobwcD*qnH>kJh_o+h$*~%!}#vu|{5Qfbf&mR5&;+TF((6M!}Xj;%nuD4 zN_pLUBZw@a~g20%gb^bBV%zRc6j*aiQW}?)Rg30eMAMMTm67uIFS*t)6b`_uo|W@ zGy7Kc``t0nS9w{tx~<%&{VRcg=4V{?bsWn5OB%Ef!&fn-@mgu++<#)O$kk6oeU8+NlEQjHcHE`)m5R18Q zJx@(NzB$%+z3oWpqnPeni`#dA!PpO>(+ zsoiByy}pCY-^B~IdTUh+HmIIX7kBQ(@K~SfU##7=>icJ{oqIdp$GSilxKFmh2m-AT7Di($RDxum!Bq#;G zjL!DbHQqQ}(Hn1iuWLqp$|gluZ0>w)k6qFD$<+M#4Z2MPyE&}8i>B>4-G)roa)uJ# zPOWv(#_4duRGStGb^YVJb(#yh-G)Z4}J!(gd^_S))*I-Wj0L*Bv<8C4qLJ;@y8Tk18$CFXgU zeZoS&-;K_vGgTVSdIe{JgZKcci%L}e2sjrwHH)gZK)>O&;|vRa@OoGzNarLZZ&z~q ziEzbf>1BH4kAM2B=qcsG6~%r>{fCkMH#~(Az?3Ck{AKowGYww58K4Z8pkaQG^0C11 zv_i1qY6(^zlRzpp*tLeqXDH8@{HW#r=v%F6TO@kDPM6*MdTeH4vE>Q&R?_4p2b|PbqBTSy#XQw*v3p3- z?lycJ2zMF!6PglBxIxIgsfX&L#i25*`cw;#eY%!h#%Oi)i;|6EqYU|Hyv-aauYM*9xo8eB$tjxgs_>u#~r-J z@6d4kJtpcD2%6!(hJ|;Fip z^O(I*k!toZ)7g!jGcVC6mxMAT@P%9;_SwgTmpnMf)qG5P@aBdTTfXanXOY0uahVG-WJtUpSE+;drypz3tJmyBU%5d_%vaZxVP7ebW1W? zYt5Q3qvP7L%&m=NNM%fugoK|-hakP^(5K6&qwBM$ns zWoM+j?oQZFdj-n~8CmzbBQMd1dvXjj)a1+RvS9xK{Q82IScbiGhN1CW^`@bI5mfmLltTft0E*ozwE!c;t zY<&_OX>;?548GEq5TJ1iHL_v3ZEqkaLR5j2! z7knBdY?4?jTOlY*Wq z_9SAA_PX&2SQyl8;zHP6lt#9^;KiW22#W}H(D(N)k+wJBh&?BSla-6(MUq43R5hEYXv*W;>X!i9LDC(llwdyP!}>fAEk zBb;Jfg`eYv@WOd#X6=@atW@=`HiBS*GTnheplW{kFY>OBFm zUp1dim|ScwFlDr;P}*mEN5UeR)hOLBG_a|3?U%uI&FHVnbC`C?>J43S4oPQQ+44KW z@dTB?moi>I_>+N95e+MezCMtSb`HjM-nwQjmS>W^+Eg#=9fXYjUM5L$TeM$Sd5G-G+?Ul&>a(g!N051 zJUko%|94GFE}o_>wsW$kw8aOLV4)KxQV87yYb*0h=lK29^70-C>*W=k_%X2{bvTBJ z1S@nACqJ}+fsLZf|8YMvP<>DQnQ&Ym{m(~6+1J}6ioNyCADgf3&v)?<@t5T;TThD= zO?hW=u^*3j569|(kHCxj%}li(<~@)aA9Q_c$XGs{!)xkxSgAxv`|v#cRekh{h%ZZK z;emtkF$m*mxFzO6_hoFT?V_%oQ;YGr!k_cf(UAy0&~-1`@A`Gsujq8ALq$ceb*A@d z1uF+VZrAgVw;i>$Kh_yJ&)2HL^@~?ey44p)v9i#+lTReGJ;24wGuc9B)gNhz#y79` z^oy#C`a@+`;WEGg*~z9@c8p-wj--~Glm%T=_^;evD>7oequ{2-j#;}kypl^C^Qac* zsE2b@oy&Y@$0}6cF5#nS^mGH7e)0{=+14~a{pgI73NTIY|Cu%mNB{~&Yqt`452={xdo$&!Y2YeZAiY{N3w-C zoQ*gZ=eXtHG|YmA={P^kpns}g&ZWk?{@IR^j+HcDg2dR8M}3tjJ;7JU{h_w%`y{++ z#@Z0I-O1S9zE_v)%fsUSViyhnS8Q{iv{h9&ss<`j&Q+l1sKWuNZ2lKZTjB$HOeK=I z&EK{hkc%}h-2wNoR=8Wb5p|2VBODKgoxIzPS^=R~C%@iBNfy_5HZqZX)k(~QC+^?) zIN>uaf+sT?gHOpwm2}8-lwzY2S~iVNkwiL|p02Lx6*+0x5FECfqxFf48O8Bv&S{7# z9s~J{T{tn2*BK{k^SQPfNd>DaQ%!|ptnqApjS~|dyD1Xz4dXP(3wX;?Eb~*K}O7m}6(l-TJ&d;oJacDG2 z-vp;?dFk$TfBf#yvN1l0s7>YRSJO!%Fw+mMRgMY=E;jooGd`P9l~Tu`UE}w-{k&Dv zOx1c>80WoBWbWmkgpWoS{O;eBt?AuzKtexKFhw=QS*~S&wL+(!Tlk?&k~6xdQAc==NvOpf=qMm_QK`Sce2!jNe6GG@Y}tX<;hnO&Ybf+3<_^! zYm>_=n(~S&VO*>nYG*0YrMzLY*%`iZ%z3Xinc|jA9mmt099Ab&&IlQHruFLp=)sNm zyT2#j!#^x=O%$^+xQdAR-DO*|h5cE+L#AV00rqE6&_wHWj5v!PqjE z9AfT){Z4bVmV2Ju=-Sjd3vm^GIy}t((_?jqP+8J5JiegF5yXnGM(o8g?{xOtM%6*4 zGUkWV4))Sm#zIVw>>h)STvg2_#^AdS82TL9aI)dRMz*=Fm^U0Q2QfWVtCZtGGrdNP z{=ldWmqF{WZ==~#AKWwpvfRJw%~&+sG{RDx^)GWTqFO>m@&n{>oeb@xU{SNXMIo+r zQ&{<#j<~mwaoW9B44U-FlM4SnKJm?e8x69F{2*;Dz3cgls`hN4GE?cs^D_am5 zZV;|AX@%IMx~QG-O6_L#kHDIy*QibVabT^cSD%R(s|d)xqF{J9vb)uR!UWsoyXm#+ z`M2q8O*ACx^*uUNSrA7;)Z1#V3uSTa+!@|qV;e`^_ZsK2UL$zk6(X4egG3Zs9WE^j z!;*aoW|`i_m7{z$oi}Nk?}k&Nji-3Ht11*U0y7;#);M=Bi0^QQk}6z&!~Y28ya7h@ zV&P^P#(Z<}3S&;1l~;AclDxZ^cr#*}|D3|ThGI3n^p(%6ba=N8vn!aNyn{;59Scwf`8fGa%O(UyF{ z(soaJBXNTS^in8{P4Y$l+Nq7j{so9M6C2ia;3al*!ou=3yC{Hko^-%X#E^{34tC)8 zYVNg6l=$OGh1zQC>N16@zIpZgiE4pQ)zon!;EXutCp z*Zjzc)if88fv20&EM%J{f;FeHobPgaJE3USepNsWFZnF~@ zOWwGveLDEDQqHRz{1J9t2+ej42+)4t$ef6>-mEU~yVd6EN1}EnfYtQQHnk!)D(3^o zOc}`NH9&!$iiLIGB>TZ6Fp?*U7q_G0#BNM9-z(QpW7(q|?^xE;r3MLe*KQ1;<>yW@ z$t!616W#RDGpysci*${7xYT$D+TL&a6NI^9sq{3z)X2OfGQ#kSg98R^V7yD{sjm$Nt>VghTb(g@zjM3~dtOe;su zI?mnaoOK+cMF=N;st<^~Ntzb(bAiK1AhOz^7W3cWajN>}j1- z5}Xg(t{76?EcL;av63z?$W&k?LN-JU)nUH-Q7wZk(WF~3jBUxBc7mMGirqB9QKkdN zQO^7#P%{;bt$v1?fNT2j%LI)ITnq^p=_{+lDZziWCB{ql^_y6&TqW#6l0(zaGHygwSjm)IY!N z{`%l`9x{zX9}c$&zg&S?ttwJJr370=SW9OE1ujVidb z)GefD5*cf;)z7i-ghgMDr|($)!ntLQYFq)p*{IXA;B6cNmISFpO)4di0Ba}KOZiM- zTIeVWmZhw%KCX`wpxh%C<-4OjmDtYpq+;xFU>_Yj5|R}Ix80@ZYlfYVe88o*#M0)` zg61Q%G`q6GZPT`MqbO1?zqR+tC>Y{#oy^I*T>*V4@K;V(YQHQ?7a|`%xkP&KjSybI z?BBsoxcx52k!=O9kN(kbqqWUnm=QJ>r2J~xvQS0iVy}{JcR8B~K?MjGR*0bgtxiAIk3|PAZ%vC+-Iyvlj28CIyyu$6Y>1KZ=lM9Tcyj ze~E_1V%Vd3)dKj~37-cHAvNc||E^FWd2*2rGR;nu@2w_n-S zwAx;q4U94>jS5ejz5FUfx{RoOB)wxQg+L=n_F0Qa6+e3JDQ+`kmFTF6W4Adz9EN2n zDPo0w*JhzYBI4>H(+K<4pAIWP^Ve<60==Cgemr)2GG?Koo*e)s&JVK(92xWG2$znP zE?*{CST8r1z{shUWJ85BDGQwTzQx-?Q09=87Pxy=HT16O6_arA#TX&8hK{vgfz>4r z4|9J_)y$8~7log8p!iru`>X$amNAvOC_4G*lg9IOL|rpp;iDi-5?VNtId8rAsOdf-n* zet<96p8981b~Bmroo4jif=s50UiM^?efxHc3CxrF)Z2ilq9i!g_{iqTm*WAH$i`YFOnyUI-oA2fo*j0MY2=$b{2I(DuIJCz_ilX z_3Sh4LI2$eX7f(^{xY_+5-zGsrxQ8r`MXvvAD@Qd{+YN_RJ9~#GXmeZf#+cGCzTeg%A^s|6==wQfw`CY-vs&z)xPOfNbLM<50}IFw5S)s1W?+73lHHuV zA5knp;S(jfvS#zOakv~?k;rYoBUd_kDhJ)R%Nw+w=r~zDl60k|)Eg!RUE7xoxTIro)ENyg z4MBmiWcR7jH}Wx730Bf9`H1p9!@5mtlIlDY-@@=Tn6suZjI0}*RanYAc1Pt0NzKDy zN740YW^PsV8NS0(+180Cdu~^z=ItNNv6SKpZYmbaZZ%d!=P~CePPmfr=)I$=YOjhc zL_3pVC;K;iFL8Iz51boIoOlKy%OAO1eM$$FFP@6 z0xy>EtWsNT(*6gsBs{Nd{RY@O`=8X8bn2s@Jax0;(NtDxI}#$vu7`Q&G$y5ZbstGG zF?cd*vpiR>6q8#>7NnCEi$5`|oV;y<38i_j@dZMAXqWvMHQ;f239Qm9KMzu9zG6SU z5x)4u<5tySH@HzOS@36tr!ab={cRvIHvk1vgixr}x z51JS=BuXZqZ8!wEeAFQGgTjmy##^{2ba9O@moHXHJU=c&$u>NP|cUYz<> zsA3}gP?vXV%cLbi`SI!fI6Cn}wYqrfoZqXT(8*MAuz`y~7~l7UCT-#C45bY3%|5pa zD~g!fyzytfak(bB3iuaN$7Z~gEGp88&|M=z{i>?wGsrGiD51LRm8H?H9wCia zzGHTpUOofbzM|%Z05^`u=U$IldsqkDb`8{;pPys{?(QD0z^mfU;mm?-FI{3Cx&jX8 zp2fk4P>=kv|&5Cug1X?ajzOg@HfU^Tu*UqD`-YICET6wzM^LXrFG z&QIt6W~VzXLOgfy-r$w>4ET6q%x~Y9=ueifq!3%UHy9e||Gd7abb9gftG$mcheU@)Y`nT&2 zzw*%x zHpwJxo!Mj0Avw>}FPhugpLa;kICHi_M#ad>UrR6?7eYA<>&GZnb`DlB z6OeC)B@>nEi?>qLS}4UutTy5dYY--9FAu9~zJ45~eBkbJOZn0l;KHEfKSMNd+0FJ7 z?@8Z)N%R3IF`c@JS*mQg{(0WqaoKP^sMKkDmC40V|>#QbDEsowoa7UqdHIJ>G`P<$nW- z0VA|NA!+31UlM6NtRz=0@P8El6vp=Jv|nVfCAvlGMnXS9>-Or~?UKTWOWlpiyPS1Y zLc4k+U>HXKVq4QTuD~ZRS_0B(f~Rrg(wc@0yInuKKhwKW=*4UE?nTxmwKE^YW-+@- z#MTCpo$B8{c^0PJ70`U;zx)*|;D~8Mc0W#JO81RR&2pwk(TBi+aH*(hvvDXfXd*vx zN}>Fq(JEeF*>GK7u;*@o7^Xt7B;-P#s>7vq3rNJ?@PZZC*|it+bqH$?&hOCcg#nv2 zb7u3|i+yDehQxq;LZsF1Wu}@-z2k!!wuxuIZ}(%LFUhg)vC$vYMn(#?P5d17O-l6@ z#-j=OlSl^b6zN~6jnS44>hTnPqE|k=Ta_Po880T0mS5gFP(kufJz?Y{Be_>eLv9f) zrGb8$Fy9tbipvnfq(`uthN*KzE*I-r7IcxBW=2Fht4g`y9C&HEE79nMv(Y1ZeQJ>j5{*=T{GxjK!^2$z9t5Jva?zNhg-^l{PZA zx#xZ~z1k`~*>0+Yn5+7$wy4Gbv~da)F6Ps2*cNCSVoF*d8=yxi+L=_Rywaj%Fe;$; zFbE*`#q){l++vJo@mASla^Q-FO!2G;)V@vB(rmi^)14qHEyG~oZjBHouhwefcs7{8tqq@?iGhfINjihhw^DX^2w4Y9|A4od6Awy*~_0} z8bo=IIGnwwKPWO=RHlBOS9#`?VFfR4R7c+`a?1HaMygZs-MIO3?=!nhWFOvz1?NoZ zBu=+d_C*#E_twc=W5emsx3|$=I=lj-1UYaIUD6~Orp0{MS#qju)$%3DTQvn6#H|CE zE*xZyJB?7<0}37&r@RJHd6NW$-D?rBD~z-21t1#AX5`N3IZ8I$(o zJk>J8lroKxQljOOJnf=`j6<~h_ZSxn3lOP1H7?3|Df2ztH-4NO-p}gh5Rd}r`}!&l ziE4O~-6;!CZdnE%qRpk<;Y37l&Jr#AXM_+kP+lJ+bj+Zh9QxRsb3vGkQ#}v76ZfON z-mz?SGv2Ra48t%yEbqiRzayHpE>DSEDW5DA-PZD9lc^i`PNLmk!ipaCs?5vFlgj?0 zW6{GtXv?Fr5%|N1Y|7>m^SK|`DzT{w5BI3DQw3QoWf0tu&`|YwJ3cjt81L&j^_bWb zSk2@lYUIyk3saA#ppun8v|u+TA!4#@Q4pyLnkIFb{|Q28CO^`7x(T zTG!Y#fV)=EYPn$-t)#t;Mv%VwgjR8vaHlPS6weibZ2E_iJG|`cwv{_@#p$-mjw}eB zyHidXBHD!N_e=igFkteE;2WgZRucQmyq$$yZH6!27Qg_a_^1Kygg@JMu7OG z&S8gigYPAO6%C+vIMoZ1>9FH6^X|O++UerACba`-=;wbB4tak>r*9LTmdMMx$nL!g+3S?MSN19H2+#%4-=S=TFHSP|j|8!Y+{$_~P8AqEq266`OBD-Eb z)@D3JewU}>x&fCaB1tz$o}0@fWGi*T?Wt^As?6Vu4_ zZ=+3PM)6eHtjH$#5v-y{(ucnKFF8~ckoDQhhP?xX@HGswUfK{yOZSR+HI)`8n>|XH zD}4iSDqdzGs^j+uhSB+ArO>8Cp$Aa1hAYbjzheu(o>eX|&N;5iX7KZh^@ZXBMaZy$WKH16z@HM;kgOpac#ngGJ zvwVfYJqHbWK(T9&oDFw9_H|x^uaRSqW27ZE`n%m3y4VI725Mt`v7@D0ffwZHp(I)) z?+k1{{Y((A3b4IBbQ6I^0L)(GxNGwCd}#McTHJ~!t|w{vtyE@0ETb3CRmKE3i%u$F z6jdpuFW=)!ZNhpp|B}3UzC>215MX{r@&-U)PVeee479D0C$>R-C03Yz_-@-;ZOm7S zM%QmzxNOy{z%nP(KAPeFNU!UDp(5pZj+ZmwKN#!61_)0IK^Y;rnCnk*uJ#MI41F3L zJ1tNN8LDp3p*>>d@o@JU)fitGy5aD9969d&SuJR>XYdG6QXNuZG-@{9iaoZ(T2l%> zx{?=u1lIZYzRT?-Eox5wCBjt9;=uMq;(EjGKZ~~Za=tGRziW?e+m>fOHItp^Inmer z`(5mtXCsCmiRG(8O|xA2EfS?6wI6oO!_eDN=1o5mYK7*5TDY>jbpxgon#4lnPW8D2 zdtqLR>sd50XB4DA^&WJqP}BWjiZYuSG6=g}PX`vwGZ0P{de?ahDYDs2wWFGr)kUjT zEiYaQx^SF2@=z+7?fnT7DvYbO6QYQem?&;xi2oKSI4OlL{)Mw+ljdQ@82#CMAlq8g z_+#bbjRBJd2Wt55sYjlNENPX<1kf8|VLN$Ww{uVFHAf*4?!^I8PU}#H7=~Pya&^oS z(TnKAO*oR) z9nD6s`Iah8;zjefgR0nfM4IVl4$4EV{b~6Hf6@gt_(5IMC^uoVB9jXubf+nbbxZYjaDn7Am+T!yZ_;&0q+8T$c68s%Z4|6zBsxFDxR9ycc0y zCu=70JI7Vt7jv2aChY@HOFbrc=Uz;`?uq^NO-~~uhNiBz)1jXVo@RBTmC~m}U1kMG z=QbdsHQdxYXX^7tX|ioO)mQDKOsQU^KiTN^#sN1Np(#Hn)I0lkeSW|_CfNJ z6!*1X+{I$2_|keRdNwAfPY&YgVjfVtj-s&(XCy10tMw+|GN)4~C+!r}U5d?8mLw6n zPP#W*l4TWCfxMiWr*qcsHMQwm4;6v44`@PSBYu6Al8cq*Sxw|0hPLBQol8CZgds*1 z31Wkv?JdM^-_5zy#{1aPL5gIJDhcx4I=C!UZ;ou^4VCS>?vw+~Ei)vA>al#h$0SJtT9NA*O_^9)V>JwC^f|8~|~5eE5TTai-3iN(2Y!0^?HVs>HKt4(^RRYqo0Bi~#kdHX^pDabhXK zuvg`QXk!F}evXh1af2rHC}G-1LCH-V*87$zlYZV$A1L+hI&p-~r4R0b4h?7KV^*Jx z>F1kME9|yRPL=JUZ|aEPPq5R$ReH#psyikcXVW9XYB(vSSCBQjDmW-;QmVknh(bEF zy$tz`uoE_nPCMycoGg=Q9ATlt5{mR(ZW{Iq?gg1{JsI$-`|XZ7!~@^bgRCqao> zrCG9hOnOLkNI0b3h+shsIG3SHX@ph#=pAC(YmLMf! zfL~%+L!5cDKpFClVO76XVxm62* zRE&$EKc17*gZnnvfa&fe@!JKm)-+^nbZ7I%q$Wb1jyXP4<%COp^^ z2R+#1PF!8f*UC=DNoI^t>+cwCmaD)jcj9(j7Lht=u^B%MB5d4%P|v-{qK``R*2D7A^r}=Uzt1iSiW;u_ zl7hFhJEuF>fMv^{#YIkIhD)(W9`^`QYV0mT5Vb12xu=(Fl3-tG_d4{6o%Y+XerL&o zlPNb(lQk>FSbvBO$UUkb%g}iN)qipc^>tW)!he1Yd~1y%X1jY8!0=quB5)0FV6rqg z-K)I0=fOkXUW1cJUNyh3JALJEItTa8duEjpHb3R#PoVi6GOoDj!r7LF$=YdK)rN0L zWhT282F>wXBqF6^eb(U=ABkbl7VWNk3i<@fU(wCB56q;a{*nsJ@R!#sU&g5Loxs#f zo$nVQPc~S4OzbY(3>zamBe}wdtbCsO^5H{I^5JI%wt7eN6urd2^I)eBU&=ZgzG|>i zy@XcqNMmC?#9oU(H$UI-zrreh#+m)MIt2|<`|A)S!25sVRhtZiYiNXVCjJo6hnR#r zwLK;Me^i}Am?liLq|3JL>auOywrzA-U)gqb*|u%lwry+vnY)?0%O}^lI43gV4RH!V z5HjYXX@5hDQv_t-UxEG_=XJG~&eB=%Chm<40)ffruZk_g-=8&smM@22s1wpRmnbIi z*ZW6<_jQ5C-CTj6yYItSfHUEzn~cte@z>kW#B6WMEqzJc?JS^oyA|H}hh@|JF?SX~ z@Aq@~Y2ZBSyT}*Q-ES{ECZecdQ4aayW1s1HY~f8{H@VxY(7Juj@k#v)=->nF0#bL& z4Bp{*H|wV2TaylUC|G7^9WZo%j#O}&1?t^#>ST}nTN;1AScwG?dzu*CwNv4@_y<6cgPu=%$Z&M-e3ncQ8= zd0pRic9Ldo{^sN!=;yc!`Jqyvu>YgC6>V;@#XN4Pp9NsUKVxnEK5Z>~RnuKoFFp65 z0H2n)%}LORf-a8hijN>$StNeeIS@0}ywA@@ z`npc+F$mlvJ=SyG9ReRv30&l!PbQIxSxu7#z-Mqi!S@yNn%NiZP8Rs^DcWYHr%jw~ z?PSUaE&w)`Oz4Vct0y^Im|1e!859^oOyk3fi&3_1GBGDZalZO-v5Lj<#=9>5tZZ87 z;KpE|VR!4%xidXU^0pc|}Cg zGxdn_mV5#m#>iz==F{-bon77N$z%^DmrjR_0S4q~C^f?|c?-j=BFBOi&xWX<;R5BQ zM_`>U&wD-@r69)cD2 z^0AHF%?5gF?}{bF9vh1sM7vRfWGt(dK~yjNI-(a-ZVRJrnaYRt66%7^A`V}bmN0Rk zd}%FQG^zhM&l42Q-`KhREhrmos~h-sX8@5bHoy#pd5l(#y0)gTzH^u@XGIiqHVzFc z>Y*{bvXK((3$v{3a@PE#(zmOtgmqn2knWOnriwjPAw+!+ede}+@iCa4-4yV9?O%U# zgtJNpis7#vJy(0;a&0+|8Qq!a)z$-{ix|JE^HMdczDrc~)&(AGtx>hZoYhiSQ-Ef5 zT;j=qR2?x4->R$BdD4wnaStEe6R|c`1m&Qe!nL@9J%&sPI9SsR069Ft!)|UDa?N|~+YF1O zn&gM{FyumT_1bWMsZb0wUnU7C&#xxY6PUr6Fltp&wod=KHcuMA|1Bdmv0#>5eO7M1 z8Hc?@_JeZgWb15xV#N>=id=BWYZi4nL&P^zf-Q+^P!R5XjPztLG*86o2As#`eWUQp zu5iy#>kxcW&&%V8O}601Zi@HgY!4k1S07LvDdJ(U4Ky-~L>8V%Vj0~rWIwa1c-Z-chCml^L6Xc{gUh2t}%1_GrA7qc5#}# zH)BQqo~P4j27!0C+gGaUrj&91Nhzsc3sh6YR@rzebF(^zhA7Yb9Yx-jKKNXp#97>w zGBKO|6N&$*@vp`(X65`|54<4;kkhb&0-BhF=6Xy0=(xNzAcD6~Iz@BYSmG~Y1J8`b zMv>)aIg|E~u%HuHsp|x*NOD}S zu>U>*AE=ZW0MfFN5=*h`W=&lTE>=Hm>CStcD!+qe+;#@S>xlmO9VgIBqBFnEZyl*K z=sR*;f#`N+v))~uO~K&Kh~~Ym!MMPp@WU2b3PDw{9G1k#<6D70+LkF^n%`WVZE%F;a*k@x0FLf z+cN6N6>-2`u7NU{cjV|32$k#jUas0o?;Tf9Le3_>x&HdH`Ykd=YzAArW|qq>%Ea}* zGbI@=GsPsVoj3JcrN40jUTH3kAh>bvToAb!X3`;I=c-8KI$qR4 zCfH5xKy^*5g?4i^zR}*e!}76?!5|FI=u`C5W8gfZe|B8F`zLeeTI0bjcL|@9{#!4$ zcXsi3Uu^zh_cHa_R<88*p>IQ0IPW__N5vm@0o=uD__NkFZPp)#(&@G&DjY4XW5U;4 zgy1;j>SM(pqUF~j4-E0cCh?pNke-{1o$fJW?)AKTny5pie#a^he$l?9nh_Upol)bD zt+sB0dL~Eet6?)|nw-}lGFK*&C2a5P8g*cxt!)m0)(|+0>#URdnW(m4cDC#uqs#kr z08u0tYP32S5z^|Dg7@*0`34U^y5rhBpj_RQk^b145>F?@qk=P!CR+o&Ii-EaZk$l@ z=)iJrmdzEjStVIQWs#X15VK0uZjuvBM@L1iT!MhYp9}2l`hOrbR3m=+Hdvk+R3b{P zFoPy$q+LF&XyXyOVPGrXY*FntvLeuQfENRX0R@l7PKLzs@$66%agEV=XFs(SJ}xa+ zw>n-2EF>hura1_@%X}}!&V&0+)FlHP?LJY@1lar-3q6=mI@s9t6yx}JYC%SDFeyXL zsz^3GDCRj1DF{b=>YIRfw0ce&8#Ax+0GQ*-34O?OE-WXlp+x8UZl>R`BcGTeK=e!% zb5(s+go+qkRp9wKIj%v-%9`^RFg?Y{s|R!po-?baleqymMNn@#6X+HzD{iDObLpCe z=-4*UNO#u(-V>$fX&S(*O^f7x=Q@5|z z3oQ?wFZI6DCsJ9BBiw}qI=riwzkIe7Y3`MDs;%UjzC)1Dpg!))AHr__p|HO6a`kFL z8i4^5lpWZ#$X97wH7|2(b(M>qK{S{1>=_r3vL;V$B#;4(zgS2!Kpv7b03d=t9A7n0ni!;ND{?2P(nvNbS7ylAFQ~Wh5IM;Gjmac677tomnhO@euadMhe(at;TuoTRLzf~w`n(BJZWJnrD z{-jE{>e>!Aml_z^U(83l0m6;5oh=p&qt5KpfpbqK6R%+3b_Iq)x>z$2iQnphg(-tYTr|bTzE_-K9{_}X|}x};;N>+ zrgKg_XrM=AXX&&MkAI}wvmRmer%K%|F|W? z>5;QeD+M5MjdXUBTP*TJHvQd>jd(B*Vrj0qRf1ivjA5!KiVF51tbjd_0GTHp1+Vs* zv@H9l=3s~2gm_)51ki~QWtAXbsE%{ty+E}>|4{t=oSdPZY{GH?VH$S_ZcY*IYOX?=NGp(BM{^r~7om1I&4pIx~4@sR1MP`KeF9 zxTrnQsQq=RG)05K3CE{pj?~@O_uCe&XQ;+-O(uUH$#&L%z5-#=L7qg4vYO|RB0{*pA{wR?fs$skJWmFjJ^ zCB3D90L&J8+Ui46?EG?_u|G*ZB(*BEU7{k6p&RGBS!a39h+N4nS1^!SKMijDE!s002xdz<`*hHS42QiH&W{( zO}umq7R4ccAHxXy{WDz_$=?S*DySgt)afRb1^_*FoiAsdlkwf6-)>jwZQ?<_*h@E+ zu>z{quiM{%WzUM{8(PNRH-R4>#Fuh(umSVPY9spwD$dzr@}l*SYS`9AWV$raLUpR% zpjKFmpI~BPCU@_r#e=}SeY9bni|%Yg>yiM}j%>hkQKn-{#~krxxizjzImoRTRugg0 z13Ii}y_mCBQGoTt{=z6!i%CSehX)_iWdT`S#9m}4O_dn?)>?kn@qB3}8ROoIH!&c!MOU1th{)$zabOdEeDgHf292_}7b+bO^1wym|0 zPqB7IZ-H5kswR4~kWy8k*k>_8ZA<;b41hnUB25yDN=8<_1Q@#vK~4+e6(Y7bjAztD zMxjPcnX^(}FxLQB@D1*v;{_h$3x#mR_708gmRP;id?-1d34=iG1F#vpT zJcgVuS%&w7L40nvSclawN@k9*Pa4tO@x-jJGj!1~7!x>IVn}E%V6o~Ar|nCATRo=P z60&}*eJuEsZjRXNTl21UuhCaGyCbKpJW+jhfqo1Z$x6i>7DKiI)F}$&f(Z7u3jr8w*c_ zD}b%jx&&DT*JhQ&_D%n|;nS3OiIie}^Ho8ns=ruiLLcwqLS^LQQP%DC3EkE>R`BEV z<%~wqAZQ@)^v&n?0uONyr8x>?^*0ChjwP_UX1Zh{2&fWgg4O1T?E*j^2q==ZW0~&m zQO9wIMI9Rnd3&btz3p&cP@DP9#%j_w%C1hs+#-K=L4PFQU7G1R*F@h=?8&n#WPo^~ zxzB?l4{H92Wmz>^W8h?$!#wExT8P-CQfZeV()cs5vRvV*X-tJ zMWjan=ce@O<(5=fb0e4{1?XL?gjJQh`c4NCTz~m%rqS60wjE-(-*C;~aIm9c%Slr; zCoV9aOn~&edE@Iui87TE`MsQoVhDB*0^^o{9I{ive8SNbn(2h3*L&<67_4$UuXG*m zvx(dNE#>)H?QEsyr()3aI081Q#kf`Ig*LPOMv(KYL68CK(3WIIP5X3H@s6(W!c9F3 zQ-tIL=?zP)rE*if8_1-U_fe^Z&-QWn8rS=CBA`m&viSdkbZPj%z;OT1)Wyoq_`joc z;`>u@?ajP4cY3XI|6jbh)|s6A($DvMI!*W6 zp0;64sTZz(PUILx}Gr3=G@6 z&)4Utz{SYB9b%XbE+)y-+I5&r41u=6{(a#+7hw!2)xpXEJ)1V-=B=D>A$xaxj7;Bg z^6($f#Ya_Zs;K-+H*Bn!-Bi2s&HFP7fx7i^Y#YM0{3_(*xTB+Gw;DJ(I#{J*SMnNP z067ywVU9sK=U}HgC(Ofg8(uvhedJ4SDiB=yElzAqBD}L&5k0KiWUP%kToCjRh_57mBjOv`aUk>hM`vuc#m2v_J@&vVm9A z*w>{cwPKOGt}L6P+XjuNBq%(Ih#_kftE!n$J<|Ty;t>%#gFzW zFl8|ZePgZ;rvE79oCkl*>8L;3tM($1$SZ>#ddl^~MM8hmVuDR_Lm1mjHkuJc0U`g+ z&yp#qf?YhKzoesi5tPM7v!{&@lui$X7iph-nnO|~o8et6GletoAxp}gNs#rlu-Hkb zS`&QdL=P|(47zX;S?}6NOaTu^|pl<*%iRK*?a?(^<2+_Hjj90gm1m& z&bsZQ(`W}&;0&B7wnQ!kp6Cr#0bAy0L8nnOf#tI6uj~T3-LvHtG|q>i7fiN|IdkqU ze0O3Mm>#4YaEPOppz-o(7#_m78mDkHcG7N-XdrWtE)SD50E_(CfvKR*Ai6{o)Z=qI zqyqo20wx{qW-^FFXvb-&fFxj2HIM|th0i&%e05Q9BIt;k@hqX5VnVqT0JviI%Z@jT zyrtzqg^@WK4(c&8XLU`6>c=gqCHRRTLdgI{^+-Q}KVBJNJQ zoAtBsfJ5(Q-XSh->)@zkR#WlxC>jXN8shcZzOn+gC5WFqL4hGQ!05KMJ#Z(0#zXZK z%<5d2&Sn6^_vQTvZP7mr0EuY1J>e#3I4LQWzO&GeOR*LHWV)@B5-9Yj%Td6OuF+T# z2>F26OX+Gn7rD|mz?6bqZMlu}EAw`Sv5luLI6qAb9lMJrvdL43?MD4ehhK^)3@374 z<_Bw8PgSY6_L3Z38XFT98m+Rw=u#3?c{ z62+prI@ncG52v!_=E!<2eByYN@_Mix+tK+IQn2(du&54uFrvRn#K1Xh_`g6vr+;4& zCzAkU-!MW!dE9^*KrsKb4(e4tG~CXzG)P~yMX~@+)_+wgbLCGQB+n&d-P6csCL{zF z_euf?QER6hS1Ve(h_~>EsAUDJ;&Nk?=&}w}u3aJQ`?lRtzX$Fy_8&`2%!N3Ljc^S##7$*dB_>qX=fmCu_AEJjBSdl?h*qL^L(2rx~~UTm(b? z6q_E;x+_hjfDw)MUX?EGUonZdFgTi`FGHC+S&&LWqgvZ0*EL9X^@K2!l(kUd5&H1_ zI^dMT+)?_us5@0;JXNO+rf3b?-MHcUWR^lF7w?`+EJUjTk<`vH|2R>Y?0#~%xih)k z!`QGp&dBri!6U|ZX$`e#EG4pcF>IZHy}s2aG{*Y?z+Xy@f)<^r3pT%%!?wA|oi0`s zPPTNrpLIWGk|`UgB*prUc}@gnYD4Rhr|{D7JN_oPy)LnyS%PA-AJbX5LyqtZ$Z^+O zZvYCsILgbGMnw+kNF?8vhy__!8g;40pdD|1e1*xrNNF^V>#Wq-wF?{!9UIDg)Du1L4&$6>TNVQ6djgTO_?$qv9xle~(clw|yYhMKU+LHK)f-USDTl zBz(+7^Vh@WG<7CY4Z|p)XjaAF5(*gyy7@^TAVT5KW;N%{yC0y5+?pdkdOq>gw`v0( zVS`=JYP%!r-;&zxtbQqsC}tjANo|cbL3`hq#gP@%Y@+oiBHn$ONxa#79`1T&NK_U- z@i*T%up-Zhuqu99Ufw7TT_|jNP<0GXYK#B?&L!4!gmZOXGWds zZtSP&rJQ&qb`Kz`KDUgfzI!}VU><#XWLI}dZvVYdRNLeIA{sSB& zt0K`~#cCXHhF*3^oe(Y&#y|XrsNAdpcv-Ufh1|9ZE!c(pmkzim^r*{Y@ldM`z+iF{ zpQjdh>I1nv3C1SuvglKgWYqoQ>9kepQP)Vm@k@4k*tc9VCn}7RU3q#FMOD(C* zGD@OApR8FJ|Z(ER=?-coeL zpCTbSd}(HQd@@q3{W1sRsEGn1+&DZQP-0#XUdv;Z`)E~0RHtP06$7?B+8&E6HZ2|Q zScA2h?ppN6Z7P5hZSZaRNb5WJB$He9iVj{y$7zUHHv^$xImni`qWkJ}ePi};A?%Q+ z8;f6>Cx+F$$GH+SlRj<$tnKHQv84+fGBA70Qk4bp2^EKj0ijS3qA+RD>wZ?mKi;{4 zJJ>~(@vX8qey*+If{idpkExG3Dwn3C4aJG#&^+zVDUn|FK9?`1}kLJ-#hP)q}42tsI8j&1g%)|A#j%S zyj_oHqCw(b5=ajL4Xh3DDijG5oxSX734Uhbn^dB0E8sKzTz4b@pfw zurjBQaoC-62q;a7y)h9Xu$#cBElHa}DkkcPl*PK+!eRyn*hNMeqKX@os9sgsmnaQA zP8p?|X-20Gz~psl?W7vfwx8>92Pf&!31aUUei*Xy-!x498Hs=z)X}5BOX?6u-K{w& zT-i>c?h$8R!WiY$ixsAyp29#nm)Z4j{f4mf^O6c@xja^RuihugBo*M-J*N=qWYbE$*4(a8e?sPr2Mvo zc&u%zfm$eWu!T)FVPxZtwKToJi1|QBuhOU0H#X=#&k*Qa zwokv-TaRXb+p>WUq4?czCR4akN~-d95B`{nv`K#-oHR)JVT_JfCK_SB-@IQ0tv5VE zQV3pH{fhmZYJ%rPf=&E0H@Fa}SdsOyD8yoV>Ti>m&Oozj4dJy~wLA4Tt`+wr!Fa@eGVxQ~mxDw`QZ%AErRMWA~##4vN6Lh%alUr;t#15XQ{{ zXct$B%L6@z#y1)Q0}N0igDe59~u+x3t1& zyEM$ntP(C{$p=QUPW04K@Q>1sy4=BM9&}5(pya+n?p)k9*kv%Zpf{kTg~{5>zgP3t zMb(eGrBM(bkJ2=B!_Z;Z2Eq<%`S0QZY{n8npuas|p%+BT;8u1%oja+Zs-P8eo~Djn zpXc`CJyt1Ksl^t!kt5ckp6obVRNV%_jID7A4=3g!lGh+ZSSXyaRv`%0skTaBAl~T zfpvl^0o22hoI;z5^v(E=)IJ&HMm!-2qKi~GmO<2LJ!v{2{Rgf6uL0Jqt0zuFDMR(r zkXX7U7-n?T$4=a0Y>0%vy4Y$=k3CTvplWB`_!S{^*HWrjWWa)JsuZC5pJPmhzI)h#KK33id zV|0KqIOsqq2nD)3q*W)cxaFjj1O)4A>e&SeS+^0rZq~LB4-@p&v-xlU6jHrYsBdZp zoYcmT(tz}na(xwj=_`}ER6`rNOv|THO7oQhlgd(5dL0;JY1De)zBVU1gi-W}iY5Vy z>KkoQP0t-JNoZPp(wo?hBg6-ILz~Cn^xc7LclBQ4v~Pw?g}sb>DcH!4Z}j-%$zmgT z4`H12&AGWY6N)XHvBwC2q9*G(cQ0N}KJCy#AXQnV*X>#Qp2=MW96j=s5OD|Iy=dQn z;IDHQ$DqW}0E|u2;mfzGL~re7UGHK8K2Afy7im@uJNuazvlX$Bps?>oacC&z8@VCi za}>k<)O76~N@@^N{K658@)oPqZoN@@kV9nkiM@)U(J#-@-A+z?N-|EVb=w#E?HNwz zNv?YQ3Vh^>pHqV$NPqa%i2spldO$$_lWK4Q9RC}{eXPS3N7{+}U;OR_(N(M%s>-%> zIaiPGS0~@6#^=}b_3ZAh-zQ&>R->AV9F@baz~-aFw>Rzc_2(fW2f)OK{SUTD0o$Le z+ug4E>+(0-k)|o|I2?v62IXj02Z}<051rb2!tLu*5xy{PU zR3H5vAMV%Y=${~(0S_&v-8(Cfvu)5ABbbFCt7K$w_uEI)qWj}!eo>Ymjx{1a|ZoL{FSZ%6-KU*VHhIcmbyB+JLgeMwFAFT4v9Cs6M4bH3L0 z3lT~wQ{|r9cAHb{Bk`Bse>fV^X)bcOHjGS~RKalpIivUX=Q~uma)DZ{#=?ffU zXYwJl(V>8u-IPwjOxcfEo;Be$| zcRPFwsCQR78HZ^XE(sl&BU1Y1dE%!JyM`c1*R{eYla}ECwW2ZHW}E0$f~(fjkau6E z`AsNrNMW`$WVG*lWbnoxRZRBAxQALSaI&;)^i1sqSpW;Y`-&KUfy$SI&b<<>t)73| zu3xK>+Z?;ya;YNjvgadJn4oDGuO*P3*Ho->C+pTfc#}gTsB?CY(NS~~jCjNBU1t`M ze7&7Dyx92!K?&&M@Uu!D7t=uv8Bpb04;)QQ*Fxg)On!DZ;6w`Bv^yQcYopd~#6R_~ zY16P$U;x<^8?Px&lf;REGtsQE^x4%}`-#47-^PC1a=yFD^oU_2@R;K>l8r@b`>avC z6hp_?vv5;079dR0=qdzQn2f+4l{FpEKHVYeg^K$g!FIoKy@`)8VV8tlsfh=7^Kg(s zX&O~L_l0BP)%udZvYWD0pDY6Afn8%#H@ukk_kf=t_%E4F4qtych)_HFRsbXqp`*Rv zpy20)LmLDC`NbZMayBha*V^*b;P*e&#erxOQAqPbtpc=xD9iUV%it{Et3l_S5hIdm z^wrIumi%&}|K@LdPu8GKd-8kEIzlm^D_|_G13aN+c`iqAa4i2i1E{esKQY$6H8N+yJ5+YO{gLzDIOD@iS!&arg2 zn_Pbl)1T-?Ofsy)aruJ)RmwqT#@i`BSU3rgHKnJ>4{ES*JYl64TvWA1t7~HROkjC0 z3>}OCe@DpKGHkc%IQdCSH-pBM#Dx2HKn_SU=kPzoqh7=8(Yl!b@&mod*wFn+MF2hA z+8IhLr1}i-#BHoSOODDu$5%8EAPtRC_iS%U;U|82e};cmhgO1I;D=(B|1l~t7i7Eq zH*1);dYrjAjYEUc05q9FjFA#7-da$nl*q`0u~~4g33-O~D$#x^NQ-AQ7U7jqbRrc! z4GGU>%9r@o{gQ!!HHG7B>+u*A5dd|8cx_7|3n|TSu+yqH{e?s=WCrAd8K#J@`t0Sm zm~1}Ixg!S=m;s~ZyF{z_Zw3L&!o#k0VZ830dCt z?73VWKFCr2zJiXL0wHNfP#U-2R65GUZ|yIw{!z7@y&+I%&QFmG2r1_Ju&*9*QsD2HhXE18tqBVIUIz2}-6X}b_S|&3CTT}*p zwHvHsgluLqc;I*V*`K}cnt3%+iKi)@J_@m&=KqCGnmJ}UG!p-#R9rC z7C1wCkWE*|by^KXWAHPM7#$ar(I*-Pl^oyTHRSI$iAfpIh{e+%@V=YXlpoYq>}Wmu z&hMVkZx4&rpF<{$KLDevNmm0^11zjWgHa$I8&ju^Yntc|}RSAm9C6DQB{$N(eC_!6bd>G+y)bR8|G z@Dx8zi`gNEMOmjFFMZiALu^@L&EiVNO9QMRx87WmCBTv!#{=$Ww@5bZBg=+S-O}W7 z9b_843dWK(jUAiZ`7DeHE1$k}h6CYG^yqAgzJ^|7g%ZldSLy##e!9`TX45*+?9a#kv$2>i^APgDDGl+3K&xlu9Gx8ya z85OdRB(o7u+Y!RuMNw3&br=o)i+Y#Fv;BecXy(anxMms5`^E;O<$dXk_078LCWe-X zHPoYv&*%+%ci^k?gw(~qoKhAth`}>EDHO)*2Hxm5!~kB@NDP?)rNErxX&}!x`1vOiWMTk<-;MICXNhiT;;69dqwo7 zLUbA)K8qG0NpRew$-egCi0O!ApX!I_w5NPm#WDWmZ_BrUTn)9I)9Io)Sq8B&wa>Yd zf<2kOStBW4YR<>HrV~p#BE;+~Pu;X>RpNwnBCtP}|wk2eKa~V=lGpy{Ga_!I_YZh6P;k6Itrasl% zzje{!nZ*<c=SE7R%RDe!rt3Q28KhEX_D#PXh(_l^RVDSA0EU4=a zD{6;YAh4KPpylvHK~qHBMF|iPD)Yznmj7)^Q1uEpaE7!ZwrPvj;wAMF9o8qV?XCay z#+pN~J_@LBbZq=}lZZTsGXW9CtkFZjtoG@Pbyj+XgaMN~dV@QTTg8n0LW4_hj|12u z@Z?x|FlI6^mBltArr;r+ZDUjSU!OjSxBK$L_E0LYHhL+xu(Gmb*_Z{^U@#ok@u6&? zKdZP1tdT8)%oo+@FDNMNebZ3S6U~`d&Vr;KCPR}rMW^aiQLJz;{~+#q|KYPbibmBq z!prxvaEHIzGZ8>No-^9HXxjTY?E5cd;DkQ4cg1UQ0c&(`Ax1!Uo(OCAGBi9FUQkEQ zRsBBg30tzWkHWsWslp+4^{@E=+kARnH*XOSUklnw8^xLhs8 z(q6qMf!?(18{0vWx`>QvWjJR!$iXMO|+p=sRMkg&h}=Y6l`V6#{r2MbWAQc>_< zx5r%1e$NJ!P7bNv6m-*aHf%kf4QDSBP;Xp%rpa}WxWGrHIa!P2#8H;&>Jd}Kae+bK z>k+WZJd^tqMe1Pj#Lnx|=+GUyX0RDKcg$83bZ)0T<6eyX?um$q^wZ_6i?cF+x+!K} z%l#=%$%GjydSR~ZnEwqW4k@6WOQdyA>iR3n@v~}lu6aWvrE(OnuF@~Q31Q*G0FO?wUneAH{CUVncY2;YP&F0g*31;`_*8Gf{80|V)j znx-~tTnIdL3%=zA2;BPZGvY+OjJ>v(Ai3D(`~hd{WnM{0CCmGy_Y}aP@slaecfXsb zZ7)L=D)NKbfpvT2EKpV^v#TZly~I%69LO!>Crksy|HJgqWA-A1w=MM3DjEI~sZO>_ z@Hr(dbY2*G5L7`&QgS=D==p7uM4r@dN6c5>@<0Yy_B6(+!0c;(lOC9DZa6>W9JxgAqZS;)>|7U2{GctHyk&^iK|S z%6uLnNLcVaM@v6ohjR;~!*Vc>g`+_}p27^tN*apTwbdE?owufzNk{~UvIUOSy3lZN zB>AvdQXN+#+csEZjj4HOiwZR&YMIghAy=^z>8FcOjg~56`v%C)cZvS7xjlNa;aI+p zZaNyJ<;BU9rBY->i=QO!+zAox+ESs*?EISO`u&)!QB-X`+Xzc)`-%{lVtGVde`auZ z-6*o}`9p|ES6&e#l=c@aD&*Ip7IpiM8}FWxo^-ly*{<0NaB)6}F6X$IlKG& zWFY!@n)QEcsnW<>fDzNOApU0=W@Bgj|7{D`x|6Yo9mw0KG+&_AG7WEFINeb3xI~>m zjj#=M8{1lD3BS9+RghZA_j^w%8F}P`^!!74g6A0xg6a>X*Ey}&8R~fA%uFz&I%WL$4Y1j zx6{~D|M^h2MDYu=nk{Ad=DxDy&SGrb*g6!&(2a=vO-RszCsu!n&%A+@kn%b% z3voYeej44;wBJTg0~v0e&d7B?Y_#m%ow1nLNST!eT>peVo%2@c9Htp|&mRSHfQP!~ zBLIaLX2iQ`Bbk|L5rBtO!KC;F-N_YUOJ4yl zfgk<|4}>tlT|nw_8z0YBgm|{&umDxU01?Nx9$YVK6cS3qJC|y^D~taA%;9yU0*J6F z#4FgF{f9Skx&b$d(lmZ!m`+frr4a+QD(w77{)LZdn;l)z})0Ygy0i&ro4Dt%pB@J zg}ElMxGd+X7YCGbqh@Ccf#ZNM87Ss__HxMNX1hgZ?hMvIpZu!>JT4bt9veUSv%PW^h`?n%9j zd&a4wLq#yhC{S^&17J~jINeDO72&-gvcjN?X?h^kM*{%c3&#K5crK08R=>z3QEgE^smY4jNHmbIv* ziwXgU2Z^I6ZcpKiBqRi;<&$sK_!F*pP0SaNRZ_sNRVdswt zyLhnXBp(QcB&nKf^ql~$eYG!lO`>Bxf@giZ5*s5yIU9js_74^2hq5Pe_;Ec#zPsXP z&!LoF@-X7yv*<2%{bxZyBdWE3z#YmJ8rp_s>Y zf+XV&wXZesa0whPHHb@lHk3WKB8#;4<;#B;R{6}fv4GB7_3MIXiZ+qTGsQ)P5eF;Agp02|bYe z#<5il1vHfNB3+0}xAEoC$dkCQj+i4ZSBBSqMc!ml(zb_BPMP;N${eQFE(byBz8B;47YaSCeBT9 zyKW$|Q7<;jSPTE)N5M|HZXv5^JDbUPQoNPz17N@jsqX~YI;y8fmFDV^CwiKe(bD3L zn34Btsp7kr# z;Yu|+T>>NprPl0B@R=}EqD8<{bEMNaBuhjo3PT6TRsQh(E1|MtE6)!vG(k@3>-#aT5Vo&p(elZ(hXNn`|?a66^Fk*;TBzZl!>ykG0>7{t>W4!q614ZfBh6P>&a z)x>Ql{wb?Qt1<(NXXZ_q-eNaXqZ!{1kDtv++-k?e+<9RK^{$?u{~!~ru!t?e0yt1k zGKp@WtY38l)K)2U1BQnrc+(?Wdx#NNJh&eYPj!hihF8j@yoGmYY#&o&$@D91QuR*{ z(l68}n!2B*NeNa%G!|XUlR-n~-?)V$mJ@BlJcY}cbWa~-a6n7ASlz&=zu7(>`51ah ztLLT->g@;ujdEj4bX)l-Vsr-5RNfy5xlRWhT)LAz*}Q4>d}2`Nb%c@zyF`R#gc`j?=8)eR^EUlWb&*rTn2 zsupgY7Nl4h?MjAGm?l1}3OQ@Bf1nPYvv0T|XF6onLCB<_JDTJ6UKV)~FAdJ#!S*ce zVLCd3>r~QtCYC6!(FNS}?6uM+WZqE~;+;t>a%2pTDgJMCoo zzI-*=q0FG!^f$9XXzkL1y&l+NEDHVi&C1anBQS~0tWjO+K^sz&9^Qx z&DJ&LnI`Pc&MdY|9D#2nF8fP%=xsuZ@!W6p* zA6Ih6>V)Sm#y`X|pxk_A7}sVf>7oZ^7zwy1)k))lpe4{;0mezHgx@`FF$A77>U7j<$p$0daBu z&rPC1M;vn-=J+4nr_alZcRi7!j$5SCb`RR-dvtPotWG{9Um^-|6=P2PyY7?(XS_WCP~xd z@&+|G%uwaL1q8~9jRgA{#UehQW~z3JT*jRl4t8v;QOv#F6{d0BvJgPGJT-6pvT1WT z;4un3cV|QQkk3|PvEn}#aX4K!Di3@X|3m8&vhqA|SEO1-jh2?P6)FszXPwpLoDGI{ zTHk&-S#Fu9Sn*A;`0e9esPG#>HQz`~(tgruHjz698Z^iU=;%?m4Z{1p9Bxm}{(2v1 z`()aCNz=;Bs(TSw_nzWzqFFuWy#FwC!qqvn;*%2gFNh6q8+zYtDhmT{_F@ zE4q!&69OcTgDe{E*WWJvQu9n>n-f1%q#*tAz6C|J^Lc~)e42P&?f9eQ@Xr*9>F?4H zy@~r$*Ff-jWEZ4&8yF7hHS_K(nYgLsg~36Vgr@{;&eVg&e@_2!gaqgx4Gt7BdgMRO z7k|fX^Y&@MZPE$cHy=Wzslihx=OrXkU`YPdE3C;StlIN7*Ur8X#{kHuaa3+5R|2N{`;l{Y16hy+t( z_?l8areGtmaUT4kFVzxl*r4h@yO{xz+hp;8$smLoV^)TbAU4)({dOL0MBW?Bij$Ic zM7*lVeYglKxvUHvzIEXvR3nEsFL0c;v2mdGg!{mXIMJdD?V=ddmQs)1Y%z>ZimCH8 zVG#7PqRXyW#_bwxh=XHB4vklVQ3)z!XVFY%^OR|J{%F}X5O?XljK*h0yjSg{aC%$K zS~q(;me(SvYA{}bCg$wfAxW^I-NF$fdOdjQkGD#B%qcQWmft8w2q1qLjOD9BP~T~A zqtW##V*0(tiB(V?S&F)-5JWa41~V?691Wu6@3O@>X-7TRDZZPt8`{I{uXr(Vqb)dT zJYOmgDVG?t<%zeDdx}+sk;mcE&ZKLZKIPj#4$>!HSmIcbU4>&*y&B7r_G7=NbEq}N ze%bn^Z)!_gHgzBBL9rl9^$pD-0C8qMI-ouQ-(KGao(zsiESTseOA`0)1!`5RU%zuPVgJfS`F`_+Ija&MrQuW0`%gFS zfOrH&<<@BbDK&U|n}+Ys5~3iqC05o6LC4F7^NzJV#HCX{SNO$=r88=f0%5b+@aP&` zL0OL85S0F^4yKF+aU4{vas%UCU_Swt#Du4svQyb+tKJ8uKM^?HpJ%-mV}C=%fsIDw zmMzZJG~pOw*Y&ifhgXbn43Rkod6(KB<3lVuL>@1Gzgd$BZ(;nVmAD8eGPiC)L->GK zM;9Hh!ulQkID`iB4Ni1_5M0g4VQ4rV&a zhU}3N$zOyl3S-UqbFx-(!)_mP#McGN`DE06l++)IS1FLh=bU-p%HpY}z4vpM5Kah* zzXi9gi4kjjAvubEpg0r!^=Zj(!;Kuztn*i&n{no!FrTjzz}~?sdXQ7UzIzj4V^<_M z0tmPGyZu87Eg>FxWt}J*TB`>R$$|D=--Z|qZYlfs_#vpO+K8IEAPkt+GKx>~CNv1& zscBZ{4%e5}jbE6Hyy)20gwAow8M}??NfvSJyoI-KV$GG1n6w8X~!8+ z+!^YP#*Wgd>M5fYsi3sIPU$WbZ!#Y*qhSII?ESV$go*RZz~mmjpGYekpUh$rk9Kxq z!0o0f{z0N+aTUK`o6FK>M+VQeHB~cwc|sBnsTn!@SF~`FxIl3%rP?A)A<1?v?o2`y z*`&+xaP3T*JR#kE+Sr+5tI5{yIM^*wk|k&T=-7DXa=aOB4R%17fJkd%vx+!dy z5RWZ|%$byI;OGYP{ScyUcpI9*+zpK|HC|PS*PlprTWpI3&*-6W8J2g>zfUyJ@EI+P z6EG=kN#iufoAj7*FTldiT5HH{IGjdma6EjCs`rm)x;~)Tg&}*CN($@ReLdZ}w@4iYOEQhc>#HYF0i@>9Q^g00k3v zX|i>31FZ5pTvlJA@7~%jSi&)a#)<7RGx>NILOvfQ{fHmcTu1Qs82d~zlgkUz9W}L! zHLo;<9_B%Ie&G|z-vUZPm#HvGwboC+pfrj-N-_|2wC*0HsYpM{tBkZ0?V~QJ=JG#7 zDRN>|(OIv#^R7nFSG(`?SICyw<{bSd(;>(oU@*^_{MD|#$nmvsB8*#mfXLHo#$!)5 z6UQ2UjCiDRMn%v0TF)FV)encZfGX(&Pl<@3sW>XAFy7IWL9<*a!nN@inf6vjB#amPHjqzVi32-Q-tZ zpfUFpTm4W^l`Q7yaDq{LMmiJL_MNMrgWYm=^OavFedRHs9exg?DLpVYe4Ac35TNLT ziEo|-8Lgnt&`&MuahzCM)2fJUW}v>WDl1a)o<4}2sQempQ*Qi4;M)(Tq^782j>o|d zSaz!((h|Rt7t$4Ugl3$0q%rUKvlgRaVsFR3#c9UrqD)FpOEx@(V+{p#MLw;t@3o}q z|L8R!E!QZI!oQ?gF(U65Ma3;`M8XxNs`9*@0Hs@W2dy?9aA)ZVpb&{%9tWUqfDOTn zHr-Fc-P+84y5D}*b;=((|TXjB45m(O(-6ezg^Iu*440M zuq_#g>|m69AN-(zh1H~rf_5Wm=rE)A?pTUBHRwV520C70*^{w!tVW85Qj9$RGFPhDKgKhEUre=)qQZj7{nn#~G%Bj~j?z z-jULc_<4&}(_(rbMlje7zvybnsE;{@L=ulj;K~2al_<9G+xsJG^pnjt>LJZPkD-fk zjK$xM6Gdt?53%dx4l2At`Q4&@5ezgdatV~ZH*01D#C}9>WK*G?1Z4o%D%mRZ^ZvB| zX}7B;XQ1KX9&wx~JQr_Fq~Ql`3ZQ{f3ems=Zx*dpBTa1htlR3JW5ztArPzetpQ zX&PWeaCL%9mk;1PS+a)7%0u@Kd~HM0Tdblzf7mfg>g09qcHH{rFQbGlJzv?92%FM4 z$(adxukzcu*{W(lniI?-E2Y#kV(GV$!@$y+&Yce#s)DK0fL5JOg{O;hC?Yk+%@L9 z{m#QGWN(eR4fUVBH6vg>U`=n4BEZfeya7hFyy5(jzN5U*-NhV?%b{n){(D~=y+m>v$HQ-@4oX8TR*iB zJ-%;x^{UK0eq?E#hu)~?ev|LAI6XaT>d2;?Tem2$W>J)mRF-MAjdY}wG%pKwN?h7l znSIfCZ|$>?ltgMVXHtnWR4906B0cAv80UWC?Q(-W391Y9Z=%D0?T({QT=Ogfy*&(# zqz$AVNUUPCz?gjElai9VzPorJXytcIcEd(c#t7X0;yd-bB<*L%O=9Jxnp>$+6R+b^ zZ|-_tBx>554SE>yPxSd_t6?$&mTOz;MemFunxbkIf0Xo(6+bqvU((cs-k&?|bm^aq zpe}xFlLEQfG~JUUN6 zx_x0UGTLz{Y-V0w)$pL(r3Ju3RI!43tETGW_e1M-oDj2WIhoOW=4}uTJ~HhCWzVe*lFKF;v%S@IlH;ER zA?^?U@gcZJXFV~F@#{6yKYej-_cWTm(uwOk`yg~t2kDn$JXgKlZU3&J(+!wA8mar{ zS`oaA({u0>knux|X_w6c{n^RLpweSuDA3hFJNJ&SHD9GLm%tTXJl%-w5tdqIW?1d& zckw95H}5m|U1D(oo(uKdep}60NsbX8<_a7O<-|~ZO$Tb9fvg~RYm8>UU8X{liip%D zORd2@-~JZY==MzEuB-2EJdSDcn^wB>fqxO1HU2NgcNo|xYC{1kj(gKtAq4K~a9OI2 zfm#H=@$JOPT2~{}0tHFFcm1XvOo3Y=hL(r4Z#z=+u6)hRQ&9{-fIU%BvmR^vp*CqH zPg&9m>q(Rh`vsim)cdY`DK)Py&T2nL>TwD`G^Mz$h4&ujW)+vd0aNk2V*Ere@+HM6 zlVjFzQ?E!z>M(cnTJgZzs|Ee+{kxqsi+%x~w@u+n>wTb}Ue=fjK5eu+-r%P@vIAdf zkYylFdNt4LDc7-+JJe767!yl4DhqPMF-cbV43lixpS{mQ)hr9-8PX8R}&d_`q3p%Wt2% z`^A_#^WtQ)gN+?hX|owkx7&}VooNs&f~*EY)*Q`PRWYc7r!Y30+*%AQOp62)Z$P~o@K(Y z2DBgVasRNshi^o?6`Fznatb)DG$pB4BROHIGjgzFY!9WDwY>k@gCV-3XVh*q1oE#| zO-|NLd-F3MaMz<)M94@BWzmPDWbBW!qo9LP>bc~tq zYAwT6zNUXcIj;IbQqOhz4r(Lt6G$v8wiW2=KYIk(z%?bmYNGbuvKQMdP;~cb+?Pur z;eDnD(5{aA$MqMptXF4aO!qGn8YsN0JD2*gWe3xFD7nOkMaY3j7l@rqaS z4_HIGRTCx9m}m?C5T7Aves5+@n*F?HzahK$ zN1yht>rQjtv(GBQ^Wu@;SToI$mj<5THs4VIC8C|*4+-;dan}BbS{uQpleW@$j z-vh@sAkpA9Nw_W6-mHGY8g*kFVeIk8<5#pdIE1Klqjml>D zK~(M_$E+w?IB+d%uQ|aGV`!|(B6K0F`3U=lH(F;rVf7&Moun5CgCVZ?6-sPJ2a1oo{8GL zA3@A1cU6;uhH2|pim$WchEdV-oSCR6>?f;p+%Xk=pMrm{o09|b>B3A>Ke6$U-se0_qwD?7Ako!|&lJ%ag4 zDsZFphN((-7`~PcW2>-^-pAILY3LN{QzvBIAA62X| zbx~kD1%QxOxJ@G)EiKEN7|$Z?)^(PoxTXMDHb!ng%IEoM@^zP^n%Z`Q1J-M z%Lv1BCIS^Y6m>P;O)#CN9hGXe%7sshhuT<@1WZl)w2IoiP3cU>h45DEYz++p{3rR% zL@y;Zc~EpxZnjsK+YO;)Kh(7PE2_1Fz^4b{7qwR4jp7q5rA1m@YjCH43eZ_V%S|I( z1vm3x4BHzNoL&U!ndQ**p|WzqvK;#|yqhf@_Ne&JuU~8DnKE|WtxeQ^l{=pczE3WO zMr{=j9;j!AmV1@rvu;(Cx=h>|2?#UAKy@`#qbfOyacuVSs<+q*KbeJ^kDbKyX=k{; zuoij7XFLtfrer8J#do<`R){1T&Ww%qKTsw+bl^qEMIB|VIw zJgcnB7r=!#U_e^uw8l6Cgz)^aG{08`&}Go_4+~P$;eK@QLP#l-^WkhT&xhw#vCc-$ zmdFt&XfB%cj|n2o5y!SgB#<*6-b$shx zzq(X$gXb$JnI+DP0wZF#=ViSnLGEc%%s92Z*bpBZ?w|$2;qMF_rWof6a?y{7K_Kf} ze}*8&Zr!O*7Y1ZsDvsna3r^bC%>>svImZ4oP2oX{!$Ob*?t1 z%i=WfZk$^hASC4zt$O~so8DgQ>es5{kM+XrX|i@XZFfzZTrTG?{|52Rs8^hEup4(FI&;q{2nCtgX9G39nG`hj~4?o75XvHYy{3P`{*;~ z3W}s6$J}Wj+^eI?x`J$Tg%|Kaq9kRK#b>7>>3;36PHtV>HwJY&5P#Eik2W?d!1@r} zo&TK*+d^xZF#2XL+yOsT|8DcoTI{fR@^=RuYp>=!O+L%Bx#Mr<=bJJ_5$Th33Ab4| z>Or&~hCQVsED0gtx^?J*NH5h?8tjoH>i3P3;rotefI8cqQf;<^8odX>(_Z4k?ixc@ zy6QDKPX%MrP0)X8hP)pq_kQe}7!2{y9s5}E8pcfH@${qu>1^_s~ z({VuTmdaQlEdZQI3<25S00aPNE>3o^cmjY6+>-!&1#q@>Cj!L*08TK6Ism(Y-Mb9{ zrX>a9g7aX2Enq}4AP)Ef4@lKAkqSirE47veECNT6LA2GOLx%%^T%6!sEa+&6R(|mJ z4*+)vAkeaj2vmTu*ugBwz(a5wIS{3#v>1R3SsNnWQvyhZ09;^a7HF;(Qvzt`mw{ZoEg%x;_tD8v%Bm;n> z5+FOG??(Yd;B|7KsuUMH4+T4g2Ai&`xt*nj2L%@o??1S-kB6+Lhna^Zgu?d^Ma|5^ z!_v(Og7g1_YucE7v~{vZ;^F~w^Z^xGIw^n^a^SASzYpaSf%pafrwf0!wov>mJI2`~ zE^EHr1X)ip18lHlfsn0G9tD2d(U-F~J|C<1&G)^lEc(@0L)Q;`w~I%5gvFziWe7}R z?@Ln(b>dZbm(Q=>3UcNki9WJ_>G&hX_|k6&wPpiGNS|tyyrzNdj%WHxkw5C_yZ4tM zwKQWwVe4#9g(DZ5$CG^<<*1`NkU^C3)KdOyiRbT86R@`~eJ1eCs(RD)kLyE)2zgD0 zhwt5nVRs{m#@xwv*iM%olW8my%F3zD z$qGYFfpHhx`Ch#oZQ7vsmyJ&>33tYzMmHG-zig#B85){;4y3xOUE?N+*E>wqE%z7% z7}CK{`5i>I1#+~OYx45C3fzNVRWQ}+ON{X>tj{@T3tFqUkbccTwpQ&aEZ; z6GR8Gopyu1CBky~Fk3%j)a|n^qmXjuN7ToxBn@x^t4r)1gN2gR{XfgH`{-nX;7QfC zywJ84k}4{@&9ASrqrT}_8#ptqd43gJ*S3Af6AsHWj*2xmo>pPkaigFRXSN;Wo>5Uc zS?PtgCU_Ey&=vjh%W0xe#1%Rf<6K_Cfqj)P-RLuWnxl@Ja=v&i!Coe<-+1)Ls^qdZ zi%YLICY-2GMGTg&>yBUr7O=(@_77aKEc5h9 zcz*HN9`tl%{eq)F+1A<47vWfDQn=4_YvqR*t9?Ez6jIk`iN_S_PFJIS0BQ~k`OHZ) zsxHbEBK&jtcZw@Qz-*yA@M`8)->K*LU=L9y!N3t8Uf5~Wj^jR&-H^1EVZypcPW|g! zwl%mAHhWHGq$On?6FmRf3k}cqH3MH)h~B;=oWQdfo(84}e*W6a(Y0ru{Nz3?i{?G@ zu>bJZpZ$ZtPKi0T?Gen&eo*TP4&kzEk&sM175$*@c@SVHQyXDVI;XaDo^ipX8k~jb z2)y%~*KPF@(R-eGligD3@c2FH)zFQX4lYe{ ztS8FzB$Gj;UT|V|(Gn-ZHaT@i zvAS3Sz+Xxs>MLS_bdca*`rpV7E-izs1PO8eV3`UCKPLf54hdA15dKsmkPQF^RYAb- zN`Q6%7^NB_x{(5G16Ni<)c;?I2M5*s#nXUE$)Dr@6Xe0*T8K1zI&c9fz@1zdRa2Ex1Dv;x!bdU)Ea&v$o%V4Kv z0VrE0zX8wy;I;1mtd~;>+yJlkKyUI?-`{l0 z4OVFc{(*2Pc>%OwmuBE1Bm;Ad{uSWj0B?2!|NKp*fB!=08{NQHAS5OJMZwvxkV84{ z23~{5-u?ptAt~|SDg5AkXMiL)yc?(snfHIXx&NaZa_;{qAp}CBfGeE=0{@7DkKz7B ziNLu(AlZwT`+t%bH|M{p>Hm@!LdG<&H2BWQia+Xh$$H@n8Pm!Uvp63?V|?!w(!l0+EF1gjgdM83cs52P*)aPX>+q zZvn(g0sWr>D3k&+=fB|GEIvUyK*L%MPvkw{C+t z7VY1{3>WOn0WBcG!~Q?^&BO7pHUFP|^KkrcamK^(zr`63n7t3E*doday@dmYsfMNl z%gRFIfQb1}y6<(y>VWm1?6nAoBUr+jq=-i* zle*{e;xO7I=#TTGQct|K-%T`+J?~;sW7awns4E;7i>@Q97DN`?YZ?Rnc)iwT62*eiRV5Y3wFPbLg0|chKKGvu?jIG#L>8F@$Puorcglqy&VI|GOR&I@RjKYGCqB(ZXLzE{?Ey7 z2b(ReeqiyC+RKqtGI<5J5YeOvs@=X)c3Pe)UjHFjnXRDL=m!wy560&r94HAaiL{_( zf`{SmIuW0I%~1iy;t#6(WxT5Dh1*6fAYj#@tDT3kp3NCYW^D=yZ);4iqr|mZyg3mX zlii0(hM=Q#yyEOj(mO`vIS)nvYiP*~3Q0NMr^%r>$mcd%+OqmAIn6d9367CZu%_cK z&e2azB1^y>Ugwviz1U}VEtZdIr1`SyLVVrwS{zNx-NV<$P4d!pjk?HU76w$)87P`^ zEEQH0Lk!Z7?T+WbusR2^VH#A%anRs1!3kAMW?i_UYnXpf?ODA;v21}n52;}{Q9}K` zy&8_(>9~{mYEj4|KV310&x1+VM8{EtAZD0ZYp-tppqJL99WcUXMa?Jhi1 z@!}Qv(>Zf~4x0qkGe5DscZ0YPkcYTuXNA5^T#t-)u^4?*n7f90-bQgZ!~_)}2Ilnk z`O2K~?($By1N^Z&B$~J_GF_OREV(|z3UIav7iuP_hE!zSSVwcX zfUu&<{K#rxWrr)Vyi28pM&qVMU78$m@rugZns_GO?Om@(td1dO!?dr%%KR)0Gh<-s!@E{yt=yEq$_vklh%~td!Zi?Sm`ZvB0*f6V>5lqV&G_v zbC)k)+D|3q?csfG437;`gpTMSbF?1(jg)sLyEzh6WKfUi`TT*?v7?)K_tUE4s@9Tu z-Wt_>b`+n3FOyI)Q-N6bj+2Z%4IcZXLu%FsGKcp0sN*2O@M04_eW=7KerojYTE*7b)*G}wt`KN zU^QI=z)72cMgZ~fasSWrySs)4?!%M)hYH?WS}<%=HP%2mNJqB0%9``2UBltC`oYtEs#oAlc4;-41RmliX>Zio=S8Ha&-vxGg^a-4TW;gLZopYSMLGs zh6ge{*!L2>;^Lbo(yTFxG4O~R`e|qC=*Pq2mPIa;8tD5U4CPGI`6-NS%TVu8AaC{B zq6S6yqix$ldr$9#WEdKl(Ey&m*N)W36c)qj;L(|?DAzbK+vtQitGRx3)6i097N8%g zH8XtRu9j4;PD_Iq=xl04u$cx0{}9%NKSLR}lN5e`Qy{P4^R5Z0V5)#cpe+W+kAEw~ z;ayhgcd_H&7JAz`l=OoRVGcW&CT3C|o)0DYbEkzQALA*^0Fq`1tW4tSWomW6Lvni{ zNrICYehB5V- z?B=3qG^7OD_ah`e@x=Lo2w=1Qj>NAFa$2Xaik_HRetsoFftZbprm@Vurs=8#q;wM#o@j{pfK3KV`W~M>If{oLD zbcQXQCm^hM_sW5^H%>ga3H=~sX;Ep8VY?&o%g+)b z3`q#afT$C?vhbR+aB2k&mr`xVvT*t+iIyVmD{bb(oauhX5pj{wUD*#5H4*+-aFsS6 zkxib5M1PuynxK!aR#skCz3=wrE37Ontc>4X^SKJ*7s$VNy=@LKBujm6{T}RVa8|JF z`gk>f`R9+10D%!`Y!D<$9!vl5y#IJ$&Yyitx1F6E#AJe^YEAEwm=g|daYW`(z_bWy z(c$W? zgGl)cV*u5`ZZ<XB?kTMG*&eGJyca$6nuv9 zD~{eLSQJB$m!Du~2Id8$tw#a5DQ((w$nj9$^kcZ5daCo@?6iXmhu~_LU=P98kq>WY zKmZ)u9bnr%7_VYo>xRQecY05s&6KGxlD^ zya2&6+hsKbfOAHjaRv8hrUg5jF7vWB(XYxfxhj4jE$I*)Y2{9qNW^cODG4Iu&lg7? zFF_uj#-rLMVEDp@hRu&Yo`P99V!^9!=-MoW-6bjHYWpTbY_&2zK?pwIm)gfq(C^7# zuWlIdj}|+kr_cE$n*-_;ffCK|m=lX%Wp#TjN5hw{;y)V}DD6YhB)0ZvGGpoZs80q0 z?hmv;uSTy7N6jnYC^o;0UfX20-SIu$HVeKVGK(B_1bB>Z99Os2XNHZh;%K}t!GYzK zT2ePo(@7sCT6NL0;PqGx$&JqyKDOp;hG!JZfG3heeWWa=Q3zzwA%$v}(KD^z{K)+W zfh}U;{F4WjYgCIn%~~G^6-Pg@Af_v4=KT2u=svR-)?A~LpE+Sh%KQeNOqEWQc%8xq z)|ilWJmgdQ-s8pNQ{M&B8g{LWoHezpI{&!hFrl#otFjwUI}z22j9<(wFQy0ExU29d z7Pc^Ju8$qj#L2^rs>#Oa^yyN~oK7jpi4&pp z1*z-|m_FrS*ghTJZg}W^;pRaBTV(})MRDaD=X-WH{KCD6I&&)b^7)+*RxU;l)g|Fx zRR0?qz0cahB{dz+_^myjhf?sRwU~cuj#f&VM!P~W(f6B)9s+94jJG4#vvG2OgTf2S z5O)!J)E3ip6cQG4P~_eNo>>4Xd{ds30!VF~p*cNa=>QvKR9|PwIBAjh(;^RNoWRCV z2gQNVBT)A@_Z?x6?+Nt%h;M3UErt=xyk>L4o-1`Vo}&-xwT1#Sl^|=ilG{Tg(40MD znzDx6LX%zIehDd#t~WL0%E(v0n{B%AJ)0;6d&EyLK2AdYL8#j9`EF5GTEi8bB1m?) z;^jlogoGzTtQt#&U7v2(+fM7-)2?^?oUR*5Wo<3}wyHc=L58G4qxX*yEyO)meU|qg zyQIp}^Q;tYf;(O$DMxL@{x+%1)D66I!I^s#Q#+K9Hmqb+zkOc&cIQ?^yWPhW23VRWz*WQ!Y!}_3VPYOm=2W4keHj& zC?hgJ(h5<`xAS29fW}rXl0rdyIy1&mtmXt@3cZS93PnO;hOPA&ObvFo=md%K8+7GW z2udm@a^586a3Q-eYA`9V5yjzc`|yo>Dbqf{&EQzU+96T9xiIo$XQ?gbBo=LaF&BIZ z(N$iQ3gAGBQ$7rQN5DUoClZnzEdG8|z8C5hF2z)ek>CfY#O)2JfK<6)Hu6$Cn51}5 zpvY=Au>5+4?X^OewiD2ftO(-FIYDu*jP}%gC!d*Yl<(*{%hFY8`2gfz1W@XTS>*qz zj^JQPCE;ER!?kTg>iU8^fTe-sMf-axmRTJgv>X_Cvl>!;OUdEy&Lg-|7^tBWvWng4 zsqLxVwvIx#bf!OXlwjZ{One5!*28Q|Z(B=9U$i7Z`<`(3hmde)jT7W781Z3RP8E}E zryXWfyhaIwY-dR-_lsiSb4Ii98>hmF#3$#RuII@M?zC+-#A430t|H~x_L=T>eDUSo znl^ul<+2~nU4o$2XPf&0wtMI>Kdo^y2Cx;BWE*dTs?cbnU|#kv95)NemcWyDbpb2K zGG&E{o>I3vhi7ajZ3O73&xq2l&4g@6k%{{_pUOcWV^wJFZFjErz9h^!R#{0jUlVZs z+tvX*w{X!@7GWI*zsenOPz6^_7$Ah}ig~7n2Rv5gtQ`FyH&VrVed1d^N8Tm)dit38 zoRTG1L(*B;K$?{`#aQzfgHDK{R8*nCht*J*aWkTzV@N6l*(s?Nsb%F2H;Tu?NFycF zGbB;pBcyD@NSD~ez(`-cGUjZzXu=35T`zHD!>2>o*eCXa>(g>-G^HtouSuO2qX&vt zLe3J_T>RpRC=VLJR04^GFw$;JV1@OhUu`fCG^HL~g4l`0%t@Mp3OY3Oo_u*ErOXi4 zWZn-dBT!=dl1#gs{b2{S(W5g%;AG2Pw^RGfw*Wq69UAbXeu2?w*>Z0TC$3_TM*BCj zSbrW1TNd|H;UOVRGp6!XrcI&++JI>e>FBgl{AH@(Z-&35@Rz<7%Ab)^Mnf zq-2j0=K2^h6^@&TYntR%({9Re;rBGum-ohx5bGTP#+t8Ko#sY0g3Gcam87UY97(g#kbo`EYE4O>WH@?6oM33# z;xR9j=uz<%`HYMPe+61@MAKBzKsEf)Av?}YVq)jnqdUR7==t&^ma4op%4v)I$oJR| zO(M`8Zd{Gy?U>GJc}GCOBV~(HV>Ov$2=c=cU=!G>ldIgx|M{+(yFTcFWq94Mvqt&w z_QW8&Y|20sscpa_{vI@So7zX;L&{p}b8jB6;+oMA9;)N&Dup<$fpz$f$qQlL7|nHj zp{hc#VXG#wMH5XuwZ~t8#=n}$_%?c>0t;mKi}UIl%=YH#kvn5P!0}?Fl&zs(zms6R^5LMalLzF~g z341lBNekTltglhm$_z;j<0(T-P5iyZ{QEm0Y+2qCIBPOW>6PYRmD4M?0{e-dM2JAv zHBr5r)@oLXZ55A+^9EcC6ESzJ#E$pA$*0Y+yc60z8hcCMe(;VP;x5=PQOz8cEdySo zq9we?3{UaTimDw7raxy0SrtTGH@_RSXy}N)i@_MiPx>LHY-B=G0|oxsUo9Ket>pLlCVc! zPxN+mLT&d@|3$@Br~OW=yW{F$x3Vj|_I&m1>sFT$l-QapEYuFyGW?Bl^$thq7W8s> zUc1-W!SU$Lf}YBWnmF?Z59Z~mD9xAVB=ezRgWxDb)z<9^y~n!uV+(NxdP1NHhtBwK z_uoMAP}DDjX(1zh2?cBt=4w?`HIe#~wpp)X028-|;^j1U(? zfjiyPztcMJ)47ZYweMq+7yZE-pg;BPk-jXUm3U9y`iAgx{vnxf_mik^bc`R&U~WcB zHv;G=D&BjV4f11%0XKff^$`bXq2L(bLkmeN*A78ug3^}YEh74?m?{ocKG3I>fVmVnUjzd){ZW2xjF8}HKO zu&3-)0vgyDqUsp)0zEEn7B~M0;KPYk9>=@@i;S0~re}5z(s%m0;!@VPVi7j={WG}bltjYVY zCF2SQyqRHV;iC8G_qi+NHi3pVYiA@=@#-8g^CZq|y&g;RAb2o0JsqNky~^$d)$Rv0 zzl{&9uK5DlLd@Ky?wA_(uV_wRW;1@JxT^)YAf}8{Hd#RhgHif|4rD+YyVk5gs<+s~YqsHSsrPOqYw$ytZ zLO+RirUkp`B8dLokv~&);A7w7%MSEbNLgB5%hok0?5D?%b1iv3!Jdy56OJ)ylr1;A z*FV=VgK~{780i@3y-}y%?^#{3ZOq*Jk8klMoV;{&TfKT7!hDbY3i9oXcm(ss)ENk2 z&xV5hSGc2I4X7LHCk_+4%HKHLCWGsnDw-Uk!ke_8S+UNtKp zFRDrekuP#Ey6=o4Ezot8cx4(X8g`uMZwe~Rn9?@cSxK=zlo2Z_q&+Nc=zm=B8Y*gx ztX+*$7g!)$5LgB*5H4?UT532Nj#TXZ`uEN*IdXMKSY9sj-*xG?+_N@ZrN7h)$A8cNGhmEbIoK1`x>wuuIe^kDdhzsQM#oAi zyZ;>)!yZ`oiOt($oQYU1h_4A3+H_;e7Tr*rr_2qlV4*LK{tkhI&7U7cES5i^=hiz? zq)q-F{4#iK(8YL_8o=4x`XHHm`K#W!q`y#0aI*0CUXo<=W%?yb$^n;csr8Tj6Xx4e1Mdd_s z@Fx#m!Jc#!|MB82FNf5{iSyq~T%@QOU-!<;hv>>Q))RzyHzS_Hmg*4=Sgabgmd!V< zs?zK`0bpa-Yr0Gf-N3%H-(PrIN8^-h4ZEwn9F~kgoA!1Ki#$Hct>z+yxL(<^ECh-{ z=~&Yb?%xi%9!r(ko(CDq8){@nND7?l%Ri1UB=CMl?hdQt#K%iZYHMUq9V-s4E;`$i zLv1B%85CAqCb)v50S|qmT;UE zD?-dM5cy|HLY#L_y}uC&Ki(e*EVRNS5+c4k6IYUfCdAinVaa`jX0i}yu$*@ybZ&$8fqJu!hw$9pV`L<#IoEyVOcmg&_o9VIl`9 z=kh)!Ir;gt9&Z@p*Qxu3G-gwoJ(6=Z1G`Datg4Ktj$=~B3VHBLqkpg9+7 zNL$KVREUszBXlP`Hn*ndyP8}bNV~>*!-DuOr>1+UkMu~UH?{h=!6y^hQbvC*YiG(1 z;$j1Y{^!vqsj#nJ!V9#CDQQ9OEobAjgZ)|u)A6U0=}`KoUD9Px7e_MxA64fRok_H9 z;n=p*v2EM7ZL`zyAKSKVqhmYi*tTumobzU?KZPc7Hw&(oc&skRZvJ5U+O>%8a3|iKC82pnr(b;9Oo` z$W&!Dd8}*4x*1Ju0ioep4@LUpWTtz$ygT9SK%`)8q#d=y)8T_sFNc`Q+eN5yJ17q>W;)sjo zx*Nfw^0i_!H*&Z3Yo%(wedTIhE?0g;X1i#Q?egg|k9o*^J4@`fm1|+skkhNGgYc0E zePA*7@Zou)6>uG^Fe0+4ez>W!rnhzWb#>^W6>OZ?!u$ELRoow@asT*!nuSp5=0!Wa zFQf!lJMdKM@~!WkZXZ*7dVaY!v_w`xxafwPI5Fajhjze=@QdII369`Gu5jUGq;FSlCy2Nh& z<^)4Vhyn1qaDv6-i8?D{aD{O~N}zL26NTM09z0;RlLaXu8JU)|#KxC;@`@`3#T%va zii^`2nsP$M#uG)3jP6`9-Dx8hQYxEKVVCZeFZqd}?e3J1{v66vG#ScLIG_Cji#-GI zG#O5O@;pJx3&R%FY2wzXVaxfh{9VF`x$5+j0D+PSq>;`@r+6#pWiCj&JFBl@B4|zs zIXF>RV)4|G^HC~Y5`@>vSpfZ6+sGN3h5aOKOph4qKeQ_92pK{iu@$pZi)rJ6TxMbh zWhp5X7IA?`ihknlRuw#3Ysb~sfTqu*x#wl_tZzJLub$s~KAHSq0)I{Z;3oHc19=hK z1IzuVLsOwALCITP=0LfDLD)H4Yv)0)!BXYsKq*>>RzSsoff@hPsXto&|8#1SjLkVC zQt3F8L8WyX*mcS8b>^(y5w@%Ps_w+IXy>5bJew*@OsshJ8} zO|I74l|*AB3(r%{{S&W#XZ?ktLYIvD^xv&X(c5jf%cP=18J@0|K(o`HOW$wxt#6-K ztWdzx?L~h2`t=s?g@8?srbVpeSEqc+;hlJ6iBexOlCzMyN1;Ff2rakY9 zz{FbSb=CD|_~zw=p6;q%$>}=6(iE*s1hM_d?sQZ(pF(wg_iu{Ks>T~MTfNXy$*ObT zq2hj<(85>i&xGk*5ziE8eNk!>D#aXM|SRGMNYU-6U3Is^0W~08xhlpXs4u0>V4vw6kkG6jc| zkPZ9r8eervi}ywpp27X-85V-`-DVKJq1}8ctzk>-@AZew$Y&$0eGcN7SLXPx#@3lp zR{05?+h@WqJ=mE4xU0IjI-42U!6ZgGL8VsDgZ?*ggzJCQH@yTohX6sOkn4{KzBN_| zx7WGBI7(~lDm#_OIoy6hDOVyIypQK#4-SIA(4;fdlRPXu`z0NZ^2^zdJ+g^}<)~)W zup-YT99-eBYRzs%C}P=vQeJYm8gzf-<047L$N#X06OeN_O91}rRm zxh>2n;K`2seA+1EclJKs8$5q6U01_~`jPwb`_H^A2#gzOGhn--{^XYYQaP8Foj#3s zi1At1KNbIrnv+OMOQEEDiTjDrS#7o``Hf5;ef0PGcvjj&cYytKk@t+{#IHeTK|)1_j<_Q4jQ7}9Fatm3WfFoJyvui_hSyrM z^q~9LVl_hf^#vsAH)8qUxtz1|GnYlTK-Ym&Vb(y&T19t2wLnsNH-1k0_dtJ8A;KL{ z*wprUP^wn+1JF>=)U3Uqi9T=)iUgdxvk%JF3UvzF43e6C{BxWAb5J7S)P)mJ)>h6- zP)^X)uJfO2$_*$Ka4Oj4Pj&7Vv=bPFlcP269uyo2gp(`P=?0W6wG$MK;|JXSDRQ^c zgMmfB{Xkv+9oRww+H=*!MIN!zFdKIvi>5KAm)u2H z0kPYu+^Lxsw~SUx)@Ct77cFf7>`$EO&=wifFewhnFc=#LQ$|`E%lYon^Hc!Nbu`Dd zPtUc9k-v-iG4G5^&fT|w3s_n*Iyn;5#ZuKK7TyT~E5KteD0-_q0qB`=;)L0rkppan z5iiiG%CGUd4<6llpz5C_u2EvN+A*=KC1Z_e_ze*jxujPt;JUt(5Yjf}79huI&eG4! zxP1gV`4^JB*VDwO-hTz@0{Th8U()dYSQxe|fUi@-(*siMA|@XG0$c@5o{18{r<|#trfLsUa22!y|N6 z;>BZ~;g8#zX%8(YWxwx~y~1^#qr*Ei>c!QO?b+ToaRxR{d^D`4JL!$ank}3F?IhI1?S|}1W&)uAHlW(1KD?TLp?{$JTIVO z!0d|Q7CWtZEc{T2(Lnv|3YIM`I|6R3TrXZ-_=>453_JRd8q#ZS+*r4|f_EWeOTbR3 z1MWvRh+&kfj0&a(kgBJo#fj|Os|TwR7leg?Wd`%?s}0x&>w$fZKJ0Nb>Y&xJ4VD5A z+R={<=Bt5y)|sl~Hhv6V{E1ibMc{}m2_Fwv60QG@Xp{U{%BS&Ydx5@qp14)RZevIA z>~v0Y>N4$CS3Uf+>c!<#`Y?6k)lRrtyJlawNpYZjrw<4Pe4T#wAl0G(cnuf)f}WRq z2p_(}{bxaWAi$Lh%V|BiJ|}jP1J6Szu~7IAh*$#KC#NkC*AbgE*1q|hoK~K8*#~N4 z9Pwtq`kD|`Nx(^BqqOj-Km@OmqyhtGTe#cGyu{WvCfD5E;c# zV{Wy6Gh=`Q(wuk&J_rtkw znf9$tP1bstVe7JCv+(O+HN}mSC$XL_Ra&-g$aS8Ssh2oVPv+gf5_@<%>ZT&z+jH$= zCK@Rd5LPd(%dS!gZv+}vI9yg&)TG*+u*te)*1moeBuwfjqDy4tb2LX`(4r(^vi?KT zwpo>V9qqkQO%~O+P1O8Dvanw3VpHg8cQ31ncj|OvdgA;%T)anYvQ0Z7IXvT|Z@N!e z0{yy}otMkS`#dw$@$d;9{@Vzt%RqiTp+21sz#te8+ECMHRC>;|K4Z zVrNUsLBBniTzW~{NN zG#0TS3gfrK4YBr(7kaeo^eXi#_SWeQAV-)E1RdTVhs$Qpq@Tk)YWyG{hpQ_E-b!uJ z@m@0Ymid;%6afVe};2{tXk7w2t!c92X9z2Ax^_ms?ts*v#7mVQ{EAa#k%&ph}%R&n3=(=wOSpqALB#+?e|r)^LkC| zoX$X=U1ymW7HhiHW}hr`IuwJLhXljG-=^lpyHxH^#-!zorYNq`a#Zb|*Mhwxl!$M- zz8g#tnG-KDNvliu<@SUrnKr@;pz3nEbU9lOfBpwl(kLNVeA1$RG~uR>(oMa}OU~WWdF*HE-c$fN&nzALc#$;G!_5urh3wNt% zIMdVwUq&rjoW}@7r@hu>9yr8sp9S*KzR+)~whdi#yTj4|0UeCPyGIKL9R0nFiZRY>AnrNA^T6tg2%}dKJMt(2D3)D^Xlf34eW?-g9hgdQ_h>kWI9I3nW`y@WexJwXiy{g8%>Zz$4aC(KXEGuvQjAp)fg|o!TNOg zLLp~4O6WAG)7bU8TK;hXnBZg`ydNXnAMr$KF((Wks9ebT&SdT{*lLf!>Qro_&3*AgLjcbP;7)T?F1Tx`T}m*$ zN_kW*O5|UKD52;5f`}t`mPhTO>$F<41K)Xm5+u1|KH1z^MZ=y??xX-G;$vnQ;{xU* zfE!0!qyl5@fVs0_2OQ3q$o)nbio>a8FXLuBFYo3iFK_%NBV&9V2V%4SZ6a+*6qY#u z9u`>F3Q1Ta`bFCv&{%5VDrsryBq?GfWL}dq>x_&YynP8d`rVXC_o_v zyVn#W4J^*XaSvP8ot)LdAK$nQdDihqw6-5PCkh}Gf`|z@UsVE)V;rCT!-Gn^*=9t_ z*RbBx1}d5g%a5wpPNX$APN1F3;zPf9XvI9wg~~heoa2%u);l^QbS-wmQiVt zuiK~v(|SnB%m*qU8m%5r7w-y*7BUw7V#SH5!Cy77G1Wjh0HJJ&fS}fb6Cv0@J|T80 z)bp-H2x`oq8o!@XEqqd*LLOH2&AWor+eSGyHhyFhGz;%+_dd{L=40$j@7xOH*>NMm zvsl0mHuow6V1|ariD(WIH~EPN>Wj&&itL4N0U&IG$cze38~o}gP1D|}M+b>HjjqWD zG3v#Y7?XCgQ@60HKBD{hF@N*O8FJ~GBB{0x9JFZYCl<)bsE?_rPD1sJ{R-bYYOiGq zfagV6!=coqVLlOjRXL_8BW%dw3;>k+;=)2(dV#?JWLA?MMj~vz5H+)sy4wkc@-6q{9ice{sAXWdi zS{*w<#T|%ZTEYz|;fXE@`D)nX2>br`#nTA(MZ9&=JTMz8e6+WyGy<}Gi$Nb8n2{%vxRd1(5T~#(Giavt zE9dk>35^iEXiiyi)cG3;Jtcq{!>(oT$HjOW5VrUC7)J9-LdoB3dWkn_yWE7_J!vLB zkcXG2`wWVue^MQdXohn3%+^8DAT5n6-j_ew0szqvE8ni;hrOEmYcx(FF$p=*dBD@##cV4LWRwv zVitBpGo)LkUX2s>U1CrgcTF%c?#@#zMfUm#JX9fZM3I*b%mj6M&1L|fFU~STEB4O&r3?V3CWiH-$c-Xb+-=ewW(eQ(yE}_z zgDj-WLO^5%p;c4B0K%R~^4!@&OV64Pe*67F$+v`=B$>?1SP*(HNw&eY!xdcash9VU zS@N~1iYp8dRA2KNvWZ(tL< z)-m^=-a|ojDZ}v1FJpi0QkC=5zs4qJW>HQGod#$0DYS%~pTn92A6nmcB&2knH|wJ5 zoZ+fK3fpMTUydy+tP$%%N&b^7rp08H1v1$kw?LZQQw>tE>7zjLHLqxv`8x8T0JD4C zLLV7grto-Wuw-3@jV0Zo{Wf>hXCV!@+*u;l&;eG04k8`0D=yFz?^QqTy?ptjfg18S zbmGK7jr_|;Jr7@<&EjoBIq|~rv>vJZ=C5fB@p()94F8f}e-02z^UEFgt2@*%pBocC z$QAgqK^r`<>cTkFa-22`$nVPh0Jscz<&+hEdBpYo2=bpGH??2F*ixfymiA3;ob^jG zXVlM*7LufuE`OG#EJMvEj@ofLqvLIbbsq=>sDEv*lbYd~d~EWT4=Hd6>8zYDFY!Fy zR_j!zpH1QRa&u3@TPkaPKS-+3B_mKrOr-4uPK_(Sz2kD&a}2YGsMWpW0&tk6E$@7= zM?tvRNK}f(P|DLj1`9Uks$d}Kgwi-gF9|V+NlQe*yyoV68zZ_h)`Ur9^b-%wG*mfe zw*!=rX)k71buKzh*VE!Bv=s>X{<&+8=%VrYsT*>y%aPWk?R;I~c|>sZtOKcVGR-t7 z>5|RnEpD4r%MllE3-8Ez0T_U|1=#ZvP9nsT{N($Bdb$o~oQdZa0n;uKpawu@eD9vX znm-1%{eSfPQ<3>OA-kxF>PFyCd;&X%7Vp?hWw2IC&GS)jfzm*5()M2X%y3WAYht?V zt8l((iE}%n4I{R}7-|C5G2R-5A!6zr;cw5UT$XFjvN@P3zp8B<0Xi>0-17XbG!lHH7y&pYu*%cv!=r<-uL4MEb_#G z*!Wf{eKaRP=Mb|KeSnwl%NZwDf8*5#LO(gpzxmc|9v7|{@)n4+_J`!(<fu+dh9NEyHyz--44{nJ&VRFo)fplrKGfjOAohv>E!`)7HD>z+?4+&hhof$Vi= zs1WHP;b-EdRU;%K_$uJF1C5P6*50I_TCH&G9|CActU2wZ^4#Z&&b@Iz*clOA9mV~; zr%#Q;)+b~5>i9Oi3woX0Hf6lVdRl)8_Qdp@m2YkCyKn$|Wy5)+kfr~QZI9x;y|R7U zHsh5+K0&Crws&JQ0RF2Yra*R9zK;n&a0{u+zK4DYpy|I)Sj?crnSw1tMo%eU47Lqc zE(F9BaLarL?nZ=1u|{+2O-EmQGcz(t6iyew3aV30WdC{!Fquie;{8s^9?kg9I@nG- z90pjWWPBGjfG4r>b{jY?LJvWp?YmL%*QEE!3A*{_mJD&p?*4N`fV%e znu8GP=7XJv)J1^}WW|nJqTZ*=jo>pLHy;InL&wSaR-v#*L>+?d~ZAbzP`hk6>vMfxUpZ_BsIX2m+3Y z?-#k_RY#bZ3)#d?Et8)}0g%&=8GW2oUm!t-worC?ESu2^VmaxOK~z7SOF#Wv)oKrC z^2*aHnEet1_p}P;l+({8-;*OcfezD-QYKxvnk$|wp%Tv#eDr>k*TB$Vf`Z~NEj2ce zp5Bfq;C~=ig8nob*#F)c=zC@AHwpx&Xz{XS_~&ZeY`<2V^B88s5U)=G|*bN#`!x)^nOHJJ5fLlS~G+8ke34c#II zj<;nRE~&>cCDvRguL>kxtc*1Dtr2<__OiTyarYg*C#0%{O-W~Hh+jpfue7P$^!f_F zFEsEI)M(Pg^Bnce@BX4y<)1V*`6acdo0m#fsgke|)wsd;4-mbD-SVx?Dbr$9w|weg z2;hu)lVNxC-**>o99PXvR~d+6piO(>lbcTb?030WccYxSJSH#57dd7=CjFvVp4S20 z;1#|rWTjqehtt{&U_zS=eKHU*A9cKd#^c2|YqSKjMwPgH(vUoO zsid{=MyEIF0C*aUaRlj-t-2H8GsM zYwZs?4*N6oHQSfqNY39UU`o+tsRv1Fc&)H5(6i2L$I_z;7L9<6g3sAs{X}f=SUUSW_3l0IP+lVCF zG(9&dEG$ecKetdUl-?e{ezWkjj4aryD>GF#3 zNP@B{0Vu?JxYb#Wg!4@7!Lipy8knj<7(V|IYp_WI6Pm~(IYE=FipdT zYL_?Iz&hYwE>+G^zIkV(IkvWsZ8<0lqy|+$$HT!KJMb;@ra4&Of?hEh>A6Lvjidgx94m&l6tMB+y|?0QkX8qNx)g}2i z^jFNXw~8akI*H)}Ai6@SB~ZcNQXejc;O3PqZV_3>aS8pN` zDy7cHalYMJTx51%Z@WeP<{hb3^11r>OPG~E4@53gxNL5r1sT|0iOF0}m6By|1p_!2 z=zy)J`zMG$#@D7dks3x88?4*pk%vVkI;x&*P+>SA>JqUsb1tBor%0Bp&&y{!WN94Z)gBb}7+M#EdLK^g-5EjO znU5d#QGykvP*D}4I@Nn7!6N@A6wvE`VH)iYxq`7j!3l9C9htWbrca&JI6Kn2iE<`? z*%&G&(Wi~7vIp8?-#U6*ftRzth3OnqJFePEOeHmz2l%LmQjh_%@*WzKS^{7e9Uq-x zy~L(Q-oW{m0vkMdzbI;BLUUw-!HcJD0yjo~LS|w*>kKKST#=%Q5kWi9aCYU@A;c8Q zRQOS+(rEx`fenSe+an1?-8Mb+`vOax7S42d8s!W!D%Nca%a02RNpTl0UV|62;%T%& z$}!ZbI;11}QU+6zw~1Rnp97*9Ns6KlWM1GeX!r?R!Zoo@E*n93sm=4C5jgU9G3t2e zWQ%g}Ed|7l=!hieIm*O?w%+;ccg-KCY&P{qGkHVtm4LlIy@x3M5MqJ9&lGOFm<+rf zrm}YrC%d{Z`9CB6ay)Ld#oyx}Uo(mR@B(mb6n_FZkAlkSdi?GAX#nY2Wot-h;}BOX z)o?00l`8#e<*pIjYT2S=4T4NAny$4@uUM?^T5*-@0eg|YwmunSj(*lqRIo75v4653 zKBX!BT4D)4=N1(@KM&Y-pVnu{I}&?7A8_48yAQHDF@2yZigPT29-hbbjr& z#IjeL{UdLJOda#1&f0@QZV#j2K|%d9-A{VN!}U2jMFcH?`UL$Ilriau~(s=O5|U#_kyn+oWlL zZZAzq4C@RdwDc5D&`}YN)s(m2|uGblzfi{Al?6-W^CBLRrRr z?`RrS@intPgD60|rduDT(*qB6`4uI0d|dHperPQ!01hL>WgCt4s`^!wB`g%QJ$^&f zmyce5)XbI7K|>6zoaDqR1(&tr5m`yKK{`KxA}}w=03JHYJ*-q5 zcN79iWBw5Ml@`zc7eQ{h+(ha9yo<9(z$braqKcAtSv0`l(u_a1%9lAf9Y@xUjA3A` zn&EJh8UmsW^z_Q-T~n`xA|ioCg_b(z@R|&`MPq#ujU}Px{VGYyD)@;1Ht5!URhzT zLJtwDyM;A4>uVv!;Rz<)a7c*X>*e*5$#E0rqmTYtO7luIYo zra5!DsY?i-F(Vcy%I<;IEU%aE*z6wxI8Iug%W;63iDACoWikr6*T>FigJ)psDcs;A zK2TYu!IV)r71BuWT?~R(}!Z}Ua)7DYrG2l+q;=$Fq1;g%*S|M^)*y7!}mtH-R)2kH4{Z)EA(Tlmk z3pzl)v+JprCj>$uzaxxvUo%X$uoRiJ@y3FlPpeMtk2iActKS%X?0lRweQqp;bC*j~ z3weMTnr_>5uRm(CTA(caRPYGxC%c;EFG3Xm%;pR*efPO;-Q;W}nIUqrqOtU5m*cM} zienHvay88~l@g;-Z+#8faEYr6e$l&f@k@aF&&T&555B+k@vsoCTa|{`UB6AGYzFlk zBrGN=rlf26Yj9G2$<>|3WNP7!t(I?9mb82ftP)F6LkC0abDnl{>}TV(^SA#$yjOB# z5lc}`_WE{8Y3iSiAqTV>(c`ny?sy7^!4rAo(WerB@m!H_&e(0BZ>3sjo5F(mPapw! zKUzUJ14)?2_(mYDD|sHv2UgF9J5$$^vG9RCWji8VR^?dD1HZ>XZN05B=1`E5l(m{R zF$+q$$Dq|uG*HEZYefZe(w7b`Ap^Lt_Yu}Q6_ie_Ljo9sj$C5vJh7k@nvL`fJDQT3 zk{GhGlyII*tP)bWf4vm-PIVGlriv4U!E$lp-lIW(MM9*5`;L5fz+LIaf&>OXCq zDHJEr_Eh&20?URnVy-ym>`c*ytGP&v%XggVf?ZRk<2?DG(s4iT;D|AxL%3sm4|>=r zT|i)7#yQmizh6vB$CGr7tJo6j8Cd}NG{nrT&K{)zxG2jB24MO0Lk}=UYX=D8v!q1H zUz_H4XEqR@D|?~Uw^*TXNKy^-$P;p`!#8h*<>GLc@#eU!**>0(?J&n|pO1+Y#^=@i zHnVaOD8eXGu+N#9vdqoUgUBkux(_LVG7>?<8*swVH(iZJc-?Z$xA3Ixh zf#U8R{}*D-T@c(W)i>Ux=p3+CAd%Y8E^qjSLomK^Y0qwf1*W(R@q&pcUNvry@)AVNFJN$!QHeA+=v$C^ISzX9|7juR%heL=jb}u5rupKRvQ(6Bo^P}7*#$= zlO#q>8a05bo!r)exlk5pEF6vapM4mr!d0`bfD%6V7VF!u)mg)Q(i;G%zT=(ka=D*9 z2o-6Z9y39udO2{hdbuEP4+j!ve6|81tKRBz9GcRwd)Yx>{X$Mf&|E}Bfe$7}Gp(y& zS1U2_y}&7FFlR>cPG3O;z)GkGL&_+@HT+;*oZmwxPXNk*t47hL38tlY8+7!>IWS0qGCJ7!3E@*6+q@LK|@zn`0lKVIYhmx>3Ifxllz}p#VNsg*=vp`~}wg@b%2y zgA=L0B6DtDzZ)oYZwWp74|*l=4yLiY3NauSoPXZArE{7m!( zEy{OTKy?l4keDc()F$-}th^I+r*G&hAq3UpPmp5xo>5b(FL$slYQK?w?px?%!DyGN|jO+*syvjj^h9vV$XQwhp zXP3cl)spEgmC<#Y!oHbF?^@VohPjIEuQZEXJlfVfc|YaF{I)rHnd!kn1@)#9n*O|I zF#o{5J@se$HTU~9_dYl*8dj^zkXa&a&o=v)kf9DAQ!RI-Uiqc zTe^#ignkm&a*zE}W+q4Q%vPP-P=(yO`qWs0DKv_xv5>B}x~sGGQC2R`%SaDL_@LEn zIn;Y_qc5~bj&P1hj)XZ;V|@E4?5MXX@*PS8ng_NAya&Ek2(82xKv4Y`O*}{Z;wX+O zegna(pRH)iFP(ThNzMlRRsRdoE%7H}B}rD4K+<1QlFSJaMMT7;pvWXzk_ZX0qjAF1 zXw!(g>^~qDifOn2V_`h=G)vYW2|L{1@jQ_<2Q5)sf%Iay?o=k6Ch}l*xIBL@qMZ^) zl=dm2Ipot=ch4nI0EF!r8&75L^*>AhnZayS9df$#ht#-aqMEp`k?mcEqi4Q+&;I9w zeu`^Zd3b!nwdk%ekJ~0HmC~%NH=lL1{>SR_y^bV7 z9HCVlSJ9-{P|pX?M6^_EHD|W54XiKTvqwt11bWq)KaWr;tnaAw7a}HRYs;38W{sfLV>a&F}Ihj9{zi|+lDesg`_#z z!cDq6!3J1uexEZU-}=Do6H4k+Z|MsHnIih> z7|_$IFcJq0iE4Y@(CB~u;&-;i!kO=Z)&xL9aym|WG4A`&cVW8S16I7hwskva*k@A=sUM>Qy94ev2(j~;0Pw4 z-pswYZrU?-x0SXly+y+NlmmdfM1+9QDN|4ZxG9CHg)QnA{pncZqoyoklY4|YAz)E{ zcxld~`9L_$FqBjY9=|DAoIZ^|ioq$2R@yyZ`q?-mK91%%Ky!DtuK~V9?FuACjKQF3 z+2kT-(8xDn7jI451OA0Iy(hT3+@}cJRj$;{o?XIHlluGynXwL-&$--NFUuh!2Sc9# z3GO#1iyr!ttc_v_C7?v^;^yezIW+@H+2`paSp2BNvbj#4mxR0f%iqe=Ti@3^UX*XX zD569XU|d8Tc`v-%B-O~!Ds#$DYKI(6UhCRnGdqMY&2!es$PHv%YH1m2@#_m$c8?AI zZ#<2(koMm>12xCaMoYhX#g0cU=)6jve2@4*Ug(ss(Jt=!rUDGZU z+Ob#q%jJgihAos!;Oik0;;<`c7B6`cb!p946M@Y3~v@%^&f06Sgj%6wx!wkxWbuHB_RZ zKRzzvi{2PCb}1ee=eEmex}EH;Hys%hv!$bmi)9oznh|wiQgmPfaA(xYmpevr#!FI- zv@z0&iM_hzbI4&LB4HRJMl9%F@!|N?AThlm+I#v`Eb=jYpya)8r&+O{rb0opr%X#W z6Hf-K#o3`pMI3RpZJ4+b_UI{eShcVk?1RtwT+?!&G_A2({SmnyxSNWyqHXdhG0dpu zn?)OG3#ni%#92%LY1orZVu6*l1efP@bYw32)N94l$Yh~c;nCpW?ncy(=5dQHa~aZ; zY8P9nsnn5?my@_}=OaDuD{_)XU^)yOT{tlg%RU#p3m_v)a0wJ=h);j^EiM?Sq@+V2 z4hFdxY@Z7kCmySf4})M1?7`W2fLaoqLH1lR*Q^^?r}xK@0OSsd)mXi z{em5`s<^#!Lt*`E1Z7MqSHH50n~;4vKYdBS4)ULK`V3tfo=41a?d;zJuz;KE0KGVAXfj)?*{ zncA|VqoSHc^r?xF*e+~Mii?aegOAB#Ux}$C4UMz8%QKB8>ikw0zg}AdvbE& zg=NwVz?hPZWcL)JQ|ef%l9)`TpdmHdGz86}sMV>S6FIavgx9eNkxN4)3>NK*(o@o# zNRCzl(@5i-p`+7PT{V(3{iOSmlH$SD+Oz*A{)%z)Sx>BtQhqSXSJLI~G`m@#I9}-r zqKAHJ^qSW@1F4PO2L?}(H$m!94KreraORZ^DB7|FQ}3LZ*WZpiAJ1>`TUO5v>vDY0 ztTgbFve0BO4Y_xi-Pin=q?%+DXyvnReqorS z1}+O80t4&R-F>{l)VY3T<~$hwMpf7iSG5j>9~hj z^u7I1HTJD8-lUm%A-+es5$R@4e_%TVP!*IJgRmaZS!Rf~Iq#zR5&S?GF?;*xic7|5Ys3&DF5YuQHYC7+YTx@&NtREgx4D?KWFr zOEMuhq$>YqNVyq=<8Be!BXS=N5LAolm{czfE*g2`yf8CgyLrt`bg*bGudGZx>y+ts zJd#v7fxUI9y?z^Ty#BXz##iZfxtS@ht1jL~ay!+bvL!{1Sec+xa5K13#Y#mLPPYM8 zhkVEAJ~^07{cOar1vGa^g54a&6?3-W2_HOW2fkRGtlpY@;tzX`Qh(0))Bj6g_*Pdu zzp)8{cwm`Tf2c8M{QWl#j87adfGcL}7q{mG$9E00RDMvusS0(Ckw0bY`Q-sqHXR2f z%~H9mbO1J6XU?juMgh8Y7ZIcd`4<+?IdmSFUjZ5^i$|LhMvmgk^uS1rLaOlYja^f( z?Z4#RX%uyo1VzEz>NZEf+=49>TQUyym|>r!_lcnQ$vL79Zfa+jma^oB!0-VjqbfV( zV)CQG=L<*Ve(NfH_s}CA=ROVh7sT?iti$fW(98!UOL(}sF*dkLFE2m`sI-@l5!hUW zt^NcEAViG%*My+Yad~RH0WP=o(uLDE4-TfL*YNJ?wpO4s{CgJ=I0fgXK?(NuZ;nQH z4W#Qsz^=XXMqlOi;Wc`B*CY;cVCBhEjYEL?-D=mInAgYRz@C@B@4YV`V$_aLDCd%Q zVJ@1`eyt-Ij#Whu{-!f^-)k>c*MNI-oKPDdiwEzA*&h&?6|l$$4tDWyL>c_XjUmHi zy>o#zAHF8rOK<~7vF}fkZjt?@kF(x_UyslF*rv{0-1KXa>~4MVr?zb2b6vzp%gR&8 zql=|Pd|S~j3Ra-R!#Qv&r&kp1J>-AQRWRFBnVK|~mp5-Bifah!jkwqKE@{k=9S)P`=?n*9!V*vy?I$HGiZy>AWBMacRc z1bdE!PUwe{9Jun;em9H`C(k7p>rag(mXwSQM?P7>R(?2{RxG;vfbJY*P4{1Nzi#3O`_|G(iI8#`{KQYyyxSzK!shiWW5MvSq*bHh_YUwhQ(qw zaf2m3B#I1R#+JEAs|ae~+)bM?T#*Bb39#v6{D7oow7Efq;oLC5V}3FFjs^A`C9^b2 z{&qG8|ITTw5^wAi?k#uS?vgHGRlXY}jZd&mdB>`g_X&HZ=A!2iTEA|IPZprQSYKvcu7^J~ z>1hEfZ%=vMqML`zn{Y&;m(+KL8N6Z58<`+u@Lm>64{1Lv8uY57e3JmS*(`@sgL?^~T9M7G=A`uf&I_g|p&ty6E~s&8iBiteWzd#o)DBu4$h4zRZ%1rSJL{=C z`s_FOb!ppGXPe8?VO?S+T5>2Z6?!Ic()55Chsz=U?}G{4)=YJV?NSY^M(8fMuI+Gw zz=eOXa^k?mL=tI?s=s=Yb^6JlN-d9yf-@k07#f)qJ^b;CmH`+zKF3pZl+ z>qU!W2A!5~`RexB15!AQ1ueJ*F>Zlv3C7zp`QlPaS-U0eDu}`XaZVXrZTyRn=bLSuU>2}wkp}VX^$2l)uPik*}!2Mzc77tMa>DpV}-wc^$oF8i@ zdEqiOb4tc`vYwiT{3eHAC0=wa>_@L*%k<+J3xh&pJ;prblkTsl7u83F%6nzlcwlKA z|B@5Q#$z{Jsc)Ze+Uc=)d&&Pvj;jOaTiYJ$7_zTvTWFH3EArB?_+OsUDRr6`eDk#~ z>a=7J-2(VuTG1)jD>Q*Hc%eF3_YZr~yKi+3z_W)P{42EtK<+oEtylUc%*>K9Hu|f? zSBb0598_Us$;f(5iKD~=`(}~AO5xcT1^Kt8E5Wssr8TNl-b2|hEtQ?>je!C5hldlO zBF-!x4{fI>^{5L<-zz0+9q5e?!200;Y8% zd_xfkIYgBievh7*=~HCg_Vxnwsz!|h@keL{F{5o)Ykg{O)fxG!Xcwi;R%%*SOn$73 zM5pS?d=egY4i+I7WVKedV;sE)j->`*#$Dz4` z$-;5fy=l-Ql@0?)7;Xx^O3{PM4`sP!nC5qz1$@u8uTu0rXB@Hk{C2(+mNevKN#TkLY=zLe2K|lLlD`IwDcsH7@DlNm(-7Kcf0j7JT~6Fyc&=aAxul6$ ze|BMXxft?;I>=Zh ztp9xVWzo&=6D}z`-9MT zZ4w&#(`c0Uk?H#dWLxgIdki0hfiNSnq@pyb=}{RHhp8I$2izB5N$c9_$@-Er2#0CH+zNkev7zYediq|Q*_b=I$$$5g0}E?Y7kCg(ox zL9H(Fzx>#MZ!%xTj_4(v4hO=yiCh_({^%k$ZsdV^0&%(H{nGNV3l}!O`uyWLsZAMd zEdrZ5^zCKB8U~C#m1mRuFTu7=fScL&PpOD`{DXjGvR!?<)9Zw%P3q_Vs)&aM2prXBL*oTk5Q>6hk3 zHY3x`i}y5;1^rVy@D;YhcN>t!Ea@?R#eamIa#B0pXZ+w?M0GpSW%Ad;=<>Q?dbQR{ z#>e}m4s`@h=1?~Tt5^^N=*CRps@48QziiTCMuV`L2a0(>b z&$sVK*XzqAhllmh@UfWrGFa@2(5!6wgu36WYt*0J9QP1%`L z7CW=#c@VOTbKM~0$G(7C=LJnkj5t5!Jao6xdf_S=K&p%pQoTKZfRd*8s}ohnK~BNW z9|$juaci>iUQ{KTol5i0c*Y3><_RP6-n4WW}uKi%Lxr-W`)zSgVFOaIVm)?9*Iok#t8 zXF6oceNqGblqOA?OJoX3Wx4dv9qVfmmvlH1og90hhcI@-sTP9V^`pS>z^Pt^%y>T{ z>#*DPnu1Yg;-BZtGRWsfbH4EXT|07ahju-7MZaK!O80u@J2iHnyR)T9zWAXWKWORI zds_< zePzw@%mV1@Ak#cOANT2cwK;D&SoIut9sl4HfCTo=72u)$pQ_K+E60_p5fDOEzFL%T zgU3U7TLxCEYdE*j27Z51IRC>>znj#Y-T`A8&@{Pq_UxTQ_E#s!g`NAwAgpe|aue4=Wh4&K4Ou`MIUt-# zruc8~sYyFH$MI>~3uRTRS=L{hM)L5gz8L55X{Y8Tp3J3zxIe0uOR%oA#q;=hDU@?# zPZAKGqEb4hu34Oe72dT+l*NP9$xz{9fWHk0LAs?gW>p3)D#@8&O6I>-%Wa$0)2lI& zoyi<{LK=CJ=R@fuZr~8KaDtSQ%j2m_!PjCzOakJC$gkN|-XfA^;G-+LaDxb1vEfqO z5?or~2W3MzLeadERVnhUIbFiC_*qe2i6vd|vLsOsii)k$)wfTHVueaa-#!=S0@u)q zBI|9h9!VbaM;5MS{|}Idu|PG)%P53uDv+f4xaTy=ZoGOjzWF`!+XoYUjqn!h6<|d=miT= zc4FROyPprSwSzm#bBY@_*nUVL1Llqv*lbN8lZexZ^0G4qcz)r1;@f04N+S~SSL<8# zw@J$Wz>-$x^zo7BwktGQjyJu$KJGC0(C_1lQ3GbRYw#B1ksn4_Npm#OF)q``s2pu54iN)gR)xhSkzr_@?8|svVx99mG2U1z&#O#7NgMG zrB#`2)kVzwreA*O=n7TYF+0FU+OQcr`z=E|%?PD>L?rN8%9OUEUm~@WW+G-qbN#2O z@Fh>KJ}x$4#K_$=LWVtYPy2*eqE^u*(sx<~BK#hTB*+=+MDL-jxLI!W3uU`j1kNFz z^vcSL3GJ8sLKjHOFCq$EfFfJL08D6wTRoQgWMzljq;%ca3U`oMGZc48MoG~)uMv7IM-XQT zF|wch3Iw5&ale$$5-KaZpKmeNYBb8fUO5Pc3eP*EB-hF~N*YC10LA6rh7Pn*lw|*` zw&wa@(#b%W3_VD}xo4rj;9fP*g3Zc?+hLjC4nl0Ri4P^^Oxc+DqN$D=^(_vew$$KFRtVc% zZtnSAi{9DnJ-2tRfR3pl*cGxB6nmGAE*`_&oxGizfTvjYNa><}qGn$4FvnY(Urr(% z3|Q2gpH}c!*Zkf=ee_a7reyt7qj=xW^BwK)@d?k5#p0P11+qj5YuzU9w$xoN?Twx> z`*gIT3h|q$v-&x^n&A;oJLRa7xps!zYcX3SoJe_#zA+-g1MU~7Fbul6Z9JDWlWm=p z6KKpId!|f(e_>=_WW&Ng>Um5?H1=V@ALUT%*m$v!wRykx4E6|Z}&r%E{3O>E8)wA1x zq#`z<3yt&Y0-kiXOMkJOQ~4*(iH8+(C*3iyQxIN03%VHW{@kQ`rOl5%V~evK9v>Z zfOz%FeokqIqnl(jQQ}#gS!2 zFmjD4K#eYwx4nHo*l*ApoU+o;)nd1G{9;S89^y+bhJou~`g9%R%B0EI*OysRx!_N2 zrb^n&T&M83w@5b6^ymsp8C3&Asc#Hf1d>nY`LG5FtIyEvBamUq{a$KJXn zsfFEE>M~H=IbXTcil>*XiZBd$H7s%Eyx~}D0GXl0E2=}g&t(;*CG2XkV7qx#b)ovh zN+v!~N*LnOoPuuowBn5v*>HgWDSnko?WoY+%omcvTcX)x?UT1IdQ^tnL_)A~&-66Bpd}Xs1g8B@j zVg?3mE{OedZLLnDs-fh8G*1x=mv0IIfShrdiM{?E5Umd}=w2f>?n@bJRlH(n>tW+= ze`R0bW(G*x6iWKh7OODMX+}4>{F=+`@FlKB#hCb%Z%%b9W)hM(KF!;ajEEA$izhe^ zK?gsI{C>a0X?*jIHq@OZ%LPHGzF__x5(4yAK-&F- z%>WO7gK(kR+v(wX9y|owc&P&y9a~Reb)EK@^fZ@V1GieZkqLuS0Ek&f-hnHpgLg8u z@OR;^C~&s#&goW!>+hBiJUval8#0l-{5&IU^#!|;;oUl(gr11gR&lR$)r)8slJN;5 z8>lrkWuYr~_u!7MzxHBip|O1d=y`2olID&)VK?C({eo;ESPURSj5^VLJZ0I^){~1S zOTI}5X!Timh6KCOds0E?WTQDewG*UIYtbfmtF*$eB*;|kmKK_ZHK%l`EoZ1X!)wXl zniP%&Tg$^G^w9a*3-Jf=Y!4yp`FuJAc_) zn67g&`>hmTL}eWtOD~gM{am-PgNcTAoRG>xsnltrG~Vu6z>M1P_OW;5+K%8@Fns9n z;O(^^hO7{^PglMQ^6c2%*=>qQSMGfpeU_iUQ}24D3Ob9s9eLd2-CQEbDHC0qi`>g( zvw$_}Q9SLBzjyQTz^|nQ`auIDp)965^t>C$UBza?8MMaQFMltN7Nj0X+`aRkmBx

*ko@i5se4gX+z**JHEcj^;YwkC zLM1l!b=15Pj!v8QN6EVxD8+~Vw=E&W>vPyhogCp*AP0fCc zBW6t2L_T}|nV(zoCDv0gDC`#ul`hqo)!4;pm92G^sNIEfN=fI~FTw(aI>9~aA_!RM z8}wh024UE)w5sylz>{SY?Qx1iBmTdyySZ5N0(004KGGk2(iJ*8Bi}6$)~Ts2KNRDs zFW%oXX?F>HZTR`$7|fCCW>CGJ069Tg5PRnjZ~J8lKDJ+*36FrRrX>*K;_&m-^n=D2 zr8u>%Mva60-@%OhYMJxZhdrlLbi)4;Nq0J3x9vMfQ^~wMY>K|8~>=&*}F1ZanUG6*Kl~cY5tt`8u(N` z1q0dTpGff!(CjP7kg)?53%Njv)`EnC!MWX;AtenR_DU$2(&G2xN*V-*HBUpvEQU5A zpHAM5sqcG0gnFT%&@PqKHocO#{9!3{ZBHzBV9$1O@YF}yu$@n1EG#wh)H%+%%+I=euC$i@I36Ry*pI)6&bWQ2uJY^wq~8fgwhoh7 z-#%eNakBgefknmjzoQc?L&MD<%_*#5;$jyfaq0e0P-L4B5Yy-|$vv5n|2nb$Use`x z6S+Aw1Lps*vYgzZx&E7#)pXzv9Zm9yl{4AT3>utVnF`JFe>{v+q4AN}xtS%tJ2*O# zva)jjk3kYVzzO31-{W`a=s02yV*K+*T5N(!ygpxLE>`)5&kH@sW!EhvG!Lsf)!&EO4;1iv#6Na;6Y)6K_OiPOMOX198^^U8OiJ~iHSgaB^Dt< z@K01dO?ajP5SSQZ_SW|Bq%j9_*WFQi-=W$z4U`AjNQci zluG19zLbqb!n%rZ={Fy@YD^&!+llimMeu{7-_G<6P9SO&)xMD8#l-k%3j)@xFa-7s z12zAs%)_L9o`o+i)_H#50t;Z@F9rLSK-!d-C6p%P+v&~8 z4{#YVw1GLpUXq2mbp+9EmKG~3NxA=0iDePX-vnX9j717ep$o8f`Er99atb$jb$xR` z6M({ukVqwsxgNFa@E}q+(h?*D5toGRHhl+)BLkL~`+89b@F2OkECIu>{C@2cWM98{ znOLgBCU5q#pCWB-^sN3t)NwH16pFYIUvpt4$q7N6r|u7e`6_nl+P-^T4BZD z%|1)Fh{!0~LrS^Jk>gJlAc#OuZIZ!zkk(_t_#&}7?8^8?ObW@+igjq@(8(K7&;(n~ z6O+keF_&*_big_up364>bYLfb01nZ8)y2&9jNh@R8pGBH*i3;io`mK7(CI`vQAgo~ z7W?rOHdD*c)#?4u7@R|D55>L&+1UB*oej1!U`M6vyGT zdxg0392&)ZEkfy%SS1st66j>>bL+M8RLq82cpg<6Ew`BLg5>SUgxJL3!Fs`mdFNb( zvh|%esN|#H`aS`qE z>D|R`7!`s{3LfR-FFL#VY@ABLlYSb>mTKz@n$QV0W~3esreyWf385!Q$Psz!3_meg z3)S$+Dalgvd-#}kLlIK@EkdbA)bE$SeyXRUu262p;x+=nqvsAFEA22Ko`&>i6uo*) z;Rt*=sUizLxxPu^@b3YYDv<1V;JK8i>L7xfJc*az<5m>)LPOW+OcL4EU2CG4vo69T zN*uqA=fg`+gZXpvZI>3m)}_ZUILVpePY*Z5>+PUco`2-T(RJ`C?`gtvY^L$ibJL$e zy}^*mofcqr(X4&G@~ZsY(WzP3wT?I$FDoh-FMf20edbdV0b}FeSaA79OrJH8V)^SD z-cU}H&h*8I@Y9!&3Jjz8J{QJsXuOW0u1})Y2ovu);l6rXeR=_BE*3?<)0TcbftBImm_$b_RT)&6O0>Xep*ZobGTg#vBpvvD7=HU8l$chKb2QStLq|F5kINfh9ZMMI6JX+L0YI^nmtkZhiGP*P7AX?X*t%J=D2BND% zNU8*|0Y0?C*PX3CkR2%V3o6cCzz#s~2R6jg7uyNsMcX!byf8}Jd2;6GZqFG%!dB(S zXMC=hSaQU-Q~^qY>h5MNjlENbtaysdM-7TkD~>3|T_@#d zfbBy8o45G+>MHp8;{}(;SlG1i!$rX06cex8;0AX9L3*h+2U~|BR6yXIW3YAO3|)nW zBRu>Y_M?6dvh2>w@_k0FcfJ-zyd3_{g-ZKlvNt@;g=8`-@6S7HYjTp&e06 zdP4Y;%y$n+uR^CWZ>dTZILnYEY@^cnVEvnWrc!ed=$2oW2d=XYO=RY?!PiQ@W2EC_ zIYs&fSCk!W`mg-7{jM>|_q8Bv8h}tGSOR*{uhHd5dedj>&Pgi$FOLwD(NWf2u6{uX zZf7+M+jDcm9My#<$uog@FHdWWb>ofh2HbjB9*MPTkN&{%dg?3;Rep!)GP`X`Y}OEB zJy<&Wk=6Phfu?-AK1J4KOST%r*j)Ks)io{MWqY`U58elz413a*J8JQt-}z1zvMYwK z);7Pa;Bn`dPSfv0ayTd=P39zeg9YPp$oA%3Gy^}?dy7+zBvK2C>ptbi$tR&ERmNMH zEau=-FmSNi`JqktY=R^>C2H6#^_cNN_edA?G!P+nXp2U8beMUjm=sAuqR585@_CRa zs|Ap(&S6I=#*Y7RKpd-ur+}4zNN&v`As1X|K3r~x!RvHF9Fct6z+t^ZjRoloD8VZo zFOLlwTFhK{kZ~4sNWuO<*@q{5&X-bG2)!>q!?>E<<2#!_*N(q}GWnYni7_VSZzGL7 zF?N|NWmspa*5O?dX$aM36lt2X_?JgK2|F<^3$EHo*;s2D2bx;59RMupDFM08HkzCv zi~$-K8nucg7>L|z>*Yw5r?gPU)PPQJ)lMd$irH$GcHNar%#ePfpOOmf_g@OE*@pA9 z+tE;5L|I)f(wk89a4er@A&M`M_i^!x>3`Ca>=l`0<4(9?Bu7k*lL{0RK&#CtX3Qj6 za*7z9RH`bXU=@Q90>X25frSjJe;&5tIbT&CkB3CJK7_O=?M24zBd8>vZ0ZE5b=C$dTlgfHn3;e znd0qZ$CdG}C0LO)5q$R()Z%5w*1zpj|M-eFqiYa-La%)iAsOvBY>`R2C59yi!MPq!0o4plhiTr@+Qp(O7(@yIGx%9#INpVJe;0boq1kg zuFot(gWBW$-1&<>Z9+uUZ~Oj`kf!ccA$L`y)|888LZMFHX+@ihzuLs0tWljtu>0Vv zBAbfnG3BWSF#ZY(JH7M$02Ksz{s|QGVIx=|DC5v!XZaB+LxWvIm{z@79|52wnG3pU zzwxyIDJWavWp@ntKe>FNRjV#lw44OI`SEjI$BH*yhiZuEqBpYJxji>9U`m?tQOzHG zlj>)tL}ZtRAGPbb(xDL7&8%xSk==w=bmEfOo#eWR7?>h3Jz*7U@2!LjmyonYD3AGb z;&!^2wO_64Q{JW=u3zH&nyNLXLtn%qmc$}}7pi(w)d=wKNE}9AocBNEMEZb6z3JJ4 zM8sL%z(u?!+v|h4YKLvh32dE+`o!nIvRo#G-B(w` z6Sf=gZcg{dnONy5m)D5Z$qgZTld?jg*k8m3<6);0jNwu1Zp(2QSq$I2(1^SXg}_w6 zypXkEi@CuUo^EB-GAjr#o`SGrE&hrsn?Dg#caM(F{GF)=kyM;F$8qN0Za za!jk5sN*+LhyD99!uK=i+{|A*i_R{OY4zFcdjy=F+*z7if6o2VkzTy*^ZOZB0k_|R zHYkeO#2tcqd5^@&aJQMupKW?M3sX{LSud?&Ipb82&UmIna;Ci}u;T0V3F$f!6TRu8 zkzP7-9e4-aS_HqXf9ZPKdo@FBeo!s;Xi;{br54buqJ#_Z=b5oCWm?UGyV%BJ7VvCeqry5RH`)RCz+&q>njBl5D4u=zl<`iPqN? z?+gw~DQ9iH#8pA(R+so?8L`)b*vND2R0sU zbjqZA%&W#X;{~6ON{gA}zQc>w?X(KB?nW8jmFlOAAK$xfJ7ft%tZt5Q5CHzyz1v%& zWS6VPr`?XkE*_nC?PuK$E=7Uc^HZR#>P`L5rg20Mw~2A55qo{Dxz$S7Q8K1{{=Cv5A3}P`d1Dk?DoM0mmE0|k=@Bh~#nYR?06cUuITMA7D zaQ#1_GI+uLo7x}|?Bz?_4Hj}qiB!TIvKWs2hLWGH%fadENj=H)!k1%ysd9#PdsI9jHW-FcQ{(?;|Ckv z-lqjG8CT?oIYF*TG))3O)5Kk;h>DE32!cx-OM`o4Q0RMgg>L#;f^+8dIkn&m2kMoO ziql)we%HpF=eB+4X2kA0@jL%+VuoIshiweKdpPe?5g3LZmq-7mrPcfT)!UzE64JQh zY1Ic?qF;Xv$ia&RrrlYb3lxmScylan-SJdmO!S;4gpK><=iC82B9Y92(2bxhc!S>l zc)PU_mQviR>`$r=M%pyqBsN?qGasMWAN5kt?cyWVyT_vHF>Ks%ZM@OV?K954-+HFJ zL9W_kpO=Mji>w%(Ul^^+Yt|s(_&`T&A7X3jLd)uIX1QsOWEGP1u{9=DPr4eG;9(!} zv`uf#ivKD@ZxsMUejNkHd@8@vay@XsVa?SI(NyOKLeXAp7Ff>t=j2)LtpK>vXf4`9 z74?{Q+aSXJlh+-WKDpl>Pjr6%5ne3mzg2wc9WJ$#Vb6Hx$*}3h?+ReQax$GSw!7Eo zT0jg&$^qN0D}}c)-jH;|e6iRBFC+4;wc=oo$5cq5_5&?~!v=E-^v?QcG6tsYty%A^kO-)@op(#3yHi1z`Odz5*& zDux6peZ4~0t|LC0Cl!b-97R0D;tBg4u`8Yn5DzkJuF){eXJ;~~%og1s@gyo>(sov# z%!s>K90GhH;{|nMGTJQK-vQ0wCZQ4cY|qg1jmX@lg68sXz|{tkV|NF zDDGi=x9iV*YSq*|k8yI)pT(Vnp%o<`3;|U4p^F*uEd*wgi zQXIbs!M*Q^X!lI!I$-APp^v=$0d=}B*QpL$eUs@Cb@S@3KY%#*tb;Qx?(};IxkB$+ ztH=g7m5MNMwu^i~RFK3VV?UP37 z^Q;rFv*MUMeqxCsUvx6y^^AJ>VYXqD-pl)Qt&ky2LG^oD3My;U zSjdNeG)`Z&!{$Ot;XbE;M+Wk}m+XN&D8L5mMyG<^T#@hlMAN7?iHtbS|`i)ac{Yvw~ z^$w9)Ls25Uu(b)nWs2wY$0ILzEXR#x?h^X&8n1O5w~RXA$eN1{J}RxCnqM_x#UZn4 zYRoBu26dEY&reOn{XVm;G5p#?2RJy(YX>AwEJtC%pKs(N_o|6PR09bnHa2gFZqNKM zYjkJ@Ptkvn&tvXI--*Fo>ciZ3Xf4h1u%2UK2hOR9Kg?9lyHBSnn}M>m4~BJ>?YqQ3 z-1u-4?MSHZ4=47+WC$u8Ss8PJ^5SrF4?HSH>70q=c%eBB&D4)_6DI5_02i0Uu%$u{ z-WkQs`Kg4Bl%Sb}<<7WHm^za~U^B|88Iw0+ImlIO&yZQ^--~fbOWE~RQMo@$c)Ty& zss8}kEoaTwEQbvK#VfTXDb1*dZ)##Q6NZ!)=4uI@|0JGsD^5&>ZQsL62u^v+)Ri;% z)Yep64~UiTB%PRW=epabfFb%Ht^ab3^>ON9Klas}?lW)8jvVU)CZ>2kbc(%CzS)|b z=$oH-*<=nA9YXNLXj9+p)CXfr9*e;EHX;7?jsJ1L$Jq2}3+M$qL%sfeWEAA2BIK_LbaZBtSzg}!RWiTV!PHvUdDaJe* zaf{I*XHXP!#Q2N=cqevB*v@bs>Ujhb7oHhg9Zr8p?oUCtri>TQ!}s%~icLuoDDH@@ zW&PqjiS)Z1r@vfR;5Uye;cO{x?<3mPJIB2+8q~Om4j`F?&1?l~OHfAZC=K463q&=J9(&{=5UvOwLew{TcV-MLE>W znSuS43dc_Vhr0;sK1k>h6ftCcrYzpd@(TBfsJ yd_K%yRN>~hleMj)8rU>#J@OZq&vsr(TEuvClhuDi6C<*4bAk{kDa92f5dI4Zb3!cu