From aaa146863f3522225906d7cc17e650b81da3075a Mon Sep 17 00:00:00 2001 From: Giorgio Gilestro Date: Mon, 7 Sep 2026 09:13:30 +0100 Subject: [PATCH] typography: remove bold-for-emphasis from the main text (PNAS convention) All 48 in-text bold spans converted: term first-introductions to italics (model merging, genetic drift, continual learning, Fisher-Muller effect, outbreeding depression, directed sex, grounded inheritance/evaluation, and kin), pure-emphasis spans to plain text (the organising-claim sentence, the emergent-null statements, the supported-conclusion sentence). Bold retained only where PNAS uses it: run-in paragraph headings (Design rules. / Analytic tier. / etc.), table labels, and reference volume numbers. Repeat mentions of first-use-italicised terms de-italicised (italics at first use only). 18-pp rebuild clean. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_01BkRLcc18rwT2Lysu6PbG7v --- paper/pnas/body.tex | 36 ++++++++-------- paper/pnas/main.md | 102 ++++++++++++++++++++++---------------------- paper/pnas/main.pdf | Bin 303556 -> 302387 bytes 3 files changed, 69 insertions(+), 69 deletions(-) diff --git a/paper/pnas/body.tex b/paper/pnas/body.tex index e451ade..f33e2f2 100644 --- a/paper/pnas/body.tex +++ b/paper/pnas/body.tex @@ -10,25 +10,25 @@ AI development increasingly resembles a population process. Models are specialis \section*{Introduction} -Machine learning has become a population-scale phenomenon. Public repositories host millions of models (Hugging Face alone grew past three million by 2026), and these are not independent creations: the overwhelming majority are fine-tunes, distillations, or merges of a small number of foundation models, forming large family trees whose lineage structure, inherited traits, and mutation dynamics are already being mapped with explicitly phylogenetic methods (1--3). This population also reproduces. Weight-space \textbf{model merging}, the direct combination of trained parents into a new model, is mainstream community practice with standard tooling and thousands of hybrid checkpoints, including leaderboard-topping ones (4--7), and the engineering literature describes it in evolutionary vocabulary: ``crossover,'' ``mutation,'' ``mate choice,'' populations of merging models that climb benchmarks (5, 8--10). +Machine learning has become a population-scale phenomenon. Public repositories host millions of models (Hugging Face alone grew past three million by 2026), and these are not independent creations: the overwhelming majority are fine-tunes, distillations, or merges of a small number of foundation models, forming large family trees whose lineage structure, inherited traits, and mutation dynamics are already being mapped with explicitly phylogenetic methods (1--3). This population also reproduces. Weight-space \emph{model merging}, the direct combination of trained parents into a new model, is mainstream community practice with standard tooling and thousands of hybrid checkpoints, including leaderboard-topping ones (4--7), and the engineering literature describes it in evolutionary vocabulary: ``crossover,'' ``mutation,'' ``mate choice,'' populations of merging models that climb benchmarks (5, 8--10). -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. The claim of this paper is that the vocabulary should be given its mathematics: \textbf{multigenerational model populations are systems whose inheritance, diversity, and compatibility must be managed, not merely collections of models to optimise, and the branch of biology that studies exactly this problem, the population genetics of the evolution of sex, transfers as a quantitative framework.} +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. The claim of this paper is that the vocabulary should be given its mathematics: multigenerational model populations are systems whose inheritance, diversity, and compatibility must be managed, not merely collections of models to optimise, and the branch of biology that studies exactly this problem, the population genetics of the evolution of sex, transfers as a quantitative framework. -The frame's entry point is the diagnosis. 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 this is the mathematics of \textbf{genetic drift} in a finite population is a conclusion we reached independently in building the present framework, and one that has been derived in parallel from several other directions (22--24), including a closed-form first-extinction law placing collapse onset at the Wright--Fisher first-extinction time (23), and that was anticipated, before deep learning, in an analysis of sequential inference chains as generalised genetic drift (25). We cite these works for priority of publication on the diagnosis and read the convergence, independent arrivals at the same population-genetic account by different routes and in different decades, as corroboration that the frame is the natural one. What none of that parallel work develops, and what this paper is about, is the structure the diagnosis opens: the full arc from drift through its remedies (immigration, recombination, selection, population structure) to its limit (reproductive isolation), carried as one framework from closed forms to trained networks to language models. +The frame's entry point is the diagnosis. 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 this is the mathematics of \emph{genetic drift} in a finite population is a conclusion we reached independently in building the present framework, and one that has been derived in parallel from several other directions (22--24), including a closed-form first-extinction law placing collapse onset at the Wright--Fisher first-extinction time (23), and that was anticipated, before deep learning, in an analysis of sequential inference chains as generalised genetic drift (25). We cite these works for priority of publication on the diagnosis and read the convergence, independent arrivals at the same population-genetic account by different routes and in different decades, as corroboration that the frame is the natural one. What none of that parallel work develops, and what this paper is about, is the structure the diagnosis opens: the full arc from drift through its remedies (immigration, recombination, selection, population structure) to its limit (reproductive isolation), carried as one framework from closed forms to trained networks to language models. -In machine learning's own terms, the problem this frame addresses is the field's oldest, \textbf{continual learning}, reappearing one level up. Within a single network, sequential learning overwrites prior knowledge (catastrophic forgetting; 26, 27), and the discipline's remedies are, one by one, the population operators of this paper in single-model form: \textbf{rehearsal and replay} of past data is grounding's within-lineage counterpart, and the field's empirically settled replay fractions, on the order of 1\% for instruction tuning (28) and 5\% to 25\% by distribution-shift strength in continual pretraining (29), sit where the minimal model's operational grounding threshold lies, a correspondence for which the framework supplies the missing theory (equilibrium diversity, and a per-capability survival law). \textbf{Pseudo-rehearsal}, the replay of the network's own generated samples, proposed as a cure in 1995 (30) and revived as generative replay (31), is this paper's ungrounded null: immigration from a drifting source, benign for one hop and compounding into collapse over generations; verifier-filtering (32, 33) converts it back into grounding. \textbf{Parameter isolation} (34, and frozen-base adapters, which forget far less; 35) is the engineered decorrelation our specialists use; \textbf{complementary-learning-systems consolidation} (36--38) is our periodic adapter-into-base merge; the recent turn to \textbf{merging as a continual-learning mechanism} (39--42) applies recombination within one lineage over time, where we apply it across lineages; and the observation that \textbf{rare examples and long-tail knowledge are forgotten first} (43--45) is tail-allele extinction observed one model at a time. One distinction is kept explicit throughout: catastrophic forgetting is largely deterministic interference from shifted training, whereas collapse is stochastic sampling drift; the two phenomena share their victims, the rare, and their remedies, but not their mechanism. To our knowledge, no prior work carries population-genetic formalism into continual learning itself; that bridge (replay as immigration with a survival law, merging as recombination with a compatibility criterion, consolidation as the slow store of a two-speed memory) is where this framework may matter most. +In machine learning's own terms, the problem this frame addresses is the field's oldest, \emph{continual learning}, reappearing one level up. Within a single network, sequential learning overwrites prior knowledge (catastrophic forgetting; 26, 27), and the discipline's remedies are, one by one, the population operators of this paper in single-model form: \emph{rehearsal and replay} of past data is grounding's within-lineage counterpart, and the field's empirically settled replay fractions, on the order of 1\% for instruction tuning (28) and 5\% to 25\% by distribution-shift strength in continual pretraining (29), sit where the minimal model's operational grounding threshold lies, a correspondence for which the framework supplies the missing theory (equilibrium diversity, and a per-capability survival law). \emph{Pseudo-rehearsal}, the replay of the network's own generated samples, proposed as a cure in 1995 (30) and revived as generative replay (31), is this paper's ungrounded null: immigration from a drifting source, benign for one hop and compounding into collapse over generations; verifier-filtering (32, 33) converts it back into grounding. \emph{Parameter isolation} (34, and frozen-base adapters, which forget far less; 35) is the engineered decorrelation our specialists use; \emph{complementary-learning-systems consolidation} (36--38) is our periodic adapter-into-base merge; the recent turn to \emph{merging as a continual-learning mechanism} (39--42) applies recombination within one lineage over time, where we apply it across lineages; and the observation that rare examples and long-tail knowledge are forgotten first (43--45) is tail-allele extinction observed one model at a time. One distinction is kept explicit throughout: catastrophic forgetting is largely deterministic interference from shifted training, whereas collapse is stochastic sampling drift; the two phenomena share their victims, the rare, and their remedies, but not their mechanism. To our knowledge, no prior work carries population-genetic formalism into continual learning itself; that bridge (replay as immigration with a survival law, merging as recombination with a compatibility criterion, consolidation as the slow store of a two-speed memory) is where this framework may matter most. -We are explicit about what kind of contribution each claim is, distinguishing \textbf{interpretation} (an existing result understood in population-genetic terms), \textbf{explanation} (the transferred mechanism accounts for observations existing accounts leave open), and \textbf{prediction} (the framework forecasts an unmeasured outcome). The paper is strongest on the first; makes concrete progress on the second (separating merge failures that are coordinate artefacts from those that are functional); and reports a first, bounded step on the third: a controlled predictive test in which pre-merge functional-disagreement measures, chosen by the framework, predicted merge damage on a constructed task grid while the tested weight-geometry baselines showed no detectable association. +We are explicit about what kind of contribution each claim is, distinguishing \emph{interpretation} (an existing result understood in population-genetic terms), \emph{explanation} (the transferred mechanism accounts for observations existing accounts leave open), and \emph{prediction} (the framework forecasts an unmeasured outcome). The paper is strongest on the first; makes concrete progress on the second (separating merge failures that are coordinate artefacts from those that are functional); and reports a first, bounded step on the third: a controlled predictive test in which pre-merge functional-disagreement measures, chosen by the framework, predicted merge damage on a constructed task grid while the tested weight-geometry baselines showed no detectable association. -The correspondences we develop, summarised in Table 1: single-teacher retraining is \textbf{asexual reproduction}, and the irreversible arm of its decay shares the defining consequence of \textbf{Muller's ratchet} (46): once every copy of a rare capability is gone from all parents and sources, no recombination can rebuild it, which is why remedies must act before fixation-by-loss (a consequence- level correspondence: the minimal model lacks the ratchet's recurrent deleterious-mutation mechanism, so irreversible loss alone does not identify that specific mechanism). Injecting verified real data is \textbf{immigration} from a non-drifting source (32, 47, 48). Model merging is \textbf{recombination}, and its central payoff, a merged model exceeding every parent, is the \textbf{Fisher--Muller effect} (49, 50). Merging entangled skills courts \textbf{outbreeding depression}; screening many candidate merges is engineered recombination with unusually flexible parent choice and pre-deployment screening (we use the shorthand \textbf{directed sex}); restricting who merges with whom is \textbf{population structure}. Merging's hard limit, models too diverged in function to combine, is \textbf{reproductive isolation}, for which the Bateson--Dobzhansky--Muller theory of incompatibilities (51, 52) supplies the structure. The nearest precursor to this programme reads sex as an algorithm for mixability in the theory of computation (53), pre-dating model merging; the model-merging literature itself has strong empirical operators (4, 54, 55) and emerging merge-success predictors (56, 57), to which our delta is mechanism: \emph{when and why} failure is coordinate versus functional, and what moves the boundary. +The correspondences we develop, summarised in Table 1: single-teacher retraining is \emph{asexual reproduction}, and the irreversible arm of its decay shares the defining consequence of \emph{Muller's ratchet} (46): once every copy of a rare capability is gone from all parents and sources, no recombination can rebuild it, which is why remedies must act before fixation-by-loss (a consequence- level correspondence: the minimal model lacks the ratchet's recurrent deleterious-mutation mechanism, so irreversible loss alone does not identify that specific mechanism). Injecting verified real data is \emph{immigration} from a non-drifting source (32, 47, 48). Model merging is \emph{recombination}, and its central payoff, a merged model exceeding every parent, is the \emph{Fisher--Muller effect} (49, 50). Merging entangled skills courts \emph{outbreeding depression}; screening many candidate merges is engineered recombination with unusually flexible parent choice and pre-deployment screening (we use the shorthand \emph{directed sex}); restricting who merges with whom is \emph{population structure}. Merging's hard limit, models too diverged in function to combine, is \emph{reproductive isolation}, for which the Bateson--Dobzhansky--Muller theory of incompatibilities (51, 52) supplies the structure. The nearest precursor to this programme reads sex as an algorithm for mixability in the theory of computation (53), pre-dating model merging; the model-merging literature itself has strong empirical operators (4, 54, 55) and emerging merge-success predictors (56, 57), to which our delta is mechanism: \emph{when and why} failure is coordinate versus functional, and what moves the boundary. -We support the framework at three tiers of evidence, in ascending realism and descending exactness: a \textbf{minimal analytic model} validated against closed forms to a fraction of a percent; \textbf{small trained networks} (MLPs, recurrent networks, an MNIST image generator) where the operators are measured in real weights; and \textbf{language models} (LoRA-specialised Qwen models, 0.5B locally and 7B on a compute cluster) where the claims are tested as signs under seed replication. Negative results are reported with the same prominence as confirmations; they include the failure of an internal pre-registered prediction, a null on emergent speciation that bounds the analogy, and the sensitivity analyses on the predictive test. +We support the framework at three tiers of evidence, in ascending realism and descending exactness: a \emph{minimal analytic model} validated against closed forms to a fraction of a percent; \emph{small trained networks} (MLPs, recurrent networks, an MNIST image generator) where the operators are measured in real weights; and \emph{language models} (LoRA-specialised Qwen models, 0.5B locally and 7B on a compute cluster) where the claims are tested as signs under seed replication. Negative results are reported with the same prominence as confirmations; they include the failure of an internal pre-registered prediction, a null on emergent speciation that bounds the analogy, and the sensitivity analyses on the predictive test. \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*. In this minimal inheritance model the resampling step \textbf{is} the Wright--Fisher process: the same equations, which we exploit as an engineering gate: our 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: the same equations, which we exploit as an engineering gate: our 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 we 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 \textbf{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 we 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). +The boundary of the exactness matters, and we 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 we 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). \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. @@ -60,11 +60,11 @@ In the minimal model, grounding from a fixed real source is immigration into a d \subsection*{Recombination: a conservation law, its operators, and offspring that exceed every parent} -The largest returns from the transfer concern merging. We begin with a result about the most common operator, stated with its assumptions. \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, \textbf{weight averaging} (a nonlinear network's weight-mean does not compute its parents' output-mean) and \textbf{routing among intact specialists} (58) (different storage and inference budgets from a single child), are its empirical cousins, and the measured bridge is a \textbf{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. We begin with a result about the most common operator, stated with its assumptions. \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} (58) (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 \textbf{Fisher--Muller effect}: 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 (59) 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 Fisher--Muller effect: 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 (59) 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 (60) (Fig. 3). When skills are entangled, blind recombination produces offspring \emph{below} their parents (\textbf{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 \textbf{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 rugged (epistatic) NK landscapes (60) (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. \begin{figure*}[p]\centering % fig2 \includegraphics[width=\textwidth]{figs/fig2.pdf} @@ -78,7 +78,7 @@ 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 \textbf{grounded inheritance}: external samples added to the reproduction process (the data channel). In the society model, grounding is \textbf{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 we 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 (61)) 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 we 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 (61)) 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} @@ -87,11 +87,11 @@ 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 (\textbf{reproductive isolation}) through Bateson--Dobzhansky--Muller incompatibilities: changes harmless on their own background but deleterious in combination. A merged model is exactly the exposed hybrid. We 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 (52). We 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 (reproductive isolation) through Bateson--Dobzhansky--Muller incompatibilities: changes harmless on their own background but deleterious in combination. A merged model is exactly the exposed hybrid. We 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 (52). We 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 (62); richer symmetry groups remove more (63), with known failures beyond the shared-data regime (64). We 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 strongest constraint comes from the pre-registered \textbf{emergent test}: true BDM incompatibilities are emergent (each lineage's changes harmless alone), so we let children diverge with \emph{no conflicting signal anywhere}, using complementary class specialists and divergent input conventions, to 6.4\(\times\) the base training. \textbf{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 \textbf{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; 65) produces no isolation at all --- the merge improves. Across every tier tested, \textbf{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 strongest constraint comes from the pre-registered \emph{emergent test}: true BDM incompatibilities are emergent (each lineage's changes harmless alone), so we 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; 65) 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} @@ -100,9 +100,9 @@ The strongest constraint comes from the pre-registered \textbf{emergent test}: t \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 we 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: \textbf{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 (56), 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 we 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 (56), 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 supported conclusion, stated conditionally: \textbf{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 \textbf{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. +The supported conclusion, stated conditionally: 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. \begin{figure*}[p]\centering % fig6 \includegraphics[width=\textwidth]{figs/fig6.pdf} @@ -129,7 +129,7 @@ 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. \emph{Ground every generation} in verified reality --- a few percent retained most diversity in our 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 our 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 we tested, what broke merging was conflicting conventions on shared circuitry, which is the thing to detect. -\textbf{What this offers continual learning.} Read into the field where these results most directly land: (i) a first-principles account of the \textbf{replay ratio}: the field's constants (\(\approx\)1\%, 5\%, 25\%; 28, 29) acquire an equilibrium theory and a 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, which is testable against published replay sweeps; (ii) a \textbf{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 (30--33); (iii) \textbf{pre-merge interference prediction with a mechanism}: where the current state of the art fits regressions over candidate metrics (56), 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 \textbf{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) \textbf{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 we note the standing objection that apparent forgetting can be skewed task-inference over latent capability rather than erasure (66); our irreversibility results 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 this offers continual learning.} Read into the field where these results most directly land: (i) a first-principles account of the \emph{replay ratio}: the field's constants (\(\approx\)1\%, 5\%, 25\%; 28, 29) acquire an equilibrium theory and a 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, which is 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 (30--33); (iii) \emph{pre-merge interference prediction with a mechanism}: where the current state of the art fits regressions over candidate metrics (56), 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 we note the standing objection that apparent forgetting can be skewed task-inference over latent capability rather than erasure (66); our irreversibility results 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 ours.} The diagnosis --- collapse as drift --- was published first by others and we cite it so (21--24), while noting the derivations are independent and convergent; prior art in the strict sense 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, 54, 55), that model populations can climb (5, 8--10), and that merge success admits ML-native predictors (56, 57). Ours 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. We claim the framework generated these measurements and experiments; we do not claim their outcomes validate a uniquely population-genetic mechanism, and one refinement it proposed was not supported. diff --git a/paper/pnas/main.md b/paper/pnas/main.md index 019ba8b..ce4d823 100644 --- a/paper/pnas/main.md +++ b/paper/pnas/main.md @@ -47,7 +47,7 @@ millions of models (Hugging Face alone grew past three million by 2026), and the independent creations: the overwhelming majority are fine-tunes, distillations, or merges of a small number of foundation models, forming large family trees whose lineage structure, inherited traits, and mutation dynamics are already being mapped with explicitly phylogenetic methods (1–3). -This population also reproduces. Weight-space **model merging**, the direct combination of trained +This population also reproduces. Weight-space *model merging*, the direct combination of trained parents into a new model, is mainstream community practice with standard tooling and thousands of hybrid checkpoints, including leaderboard-topping ones (4–7), and the engineering literature describes it in evolutionary vocabulary: "crossover," "mutation," "mate choice," populations of @@ -62,14 +62,14 @@ exhausted by frontier training within this decade (16). Meanwhile persistent mul 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. The claim of this paper is that the vocabulary should be -given its mathematics: **multigenerational model populations are systems whose inheritance, +given its mathematics: multigenerational model populations are systems whose inheritance, diversity, and compatibility must be managed, not merely collections of models to optimise, and the branch of biology that studies exactly this problem, the population genetics of the evolution of sex, -transfers as a quantitative framework.** +transfers as a quantitative framework. The frame's entry point is the diagnosis. 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 this is the mathematics of **genetic drift** in +drifts toward its own most common behaviour (21). That this is the mathematics of *genetic drift* in a finite population is a conclusion we reached independently in building the present framework, and one that has been derived in parallel from several other directions (22–24), including a closed-form first-extinction law placing collapse onset at the Wright–Fisher first-extinction time (23), and that @@ -82,22 +82,22 @@ recombination, selection, population structure) to its limit (reproductive isola framework from closed forms to trained networks to language models. In machine learning's own terms, the problem this frame addresses is the field's oldest, -**continual learning**, reappearing one level up. Within a single network, sequential learning +*continual learning*, reappearing one level up. Within a single network, sequential learning overwrites prior knowledge (catastrophic forgetting; 26, 27), and the discipline's remedies are, one -by one, the population operators of this paper in single-model form: **rehearsal and replay** of past +by one, the population operators of this paper in single-model form: *rehearsal and replay* of past data is grounding's within-lineage counterpart, and the field's empirically settled replay fractions, on the order of 1% for instruction tuning (28) and 5% to 25% by distribution-shift strength in continual pretraining (29), sit where the minimal model's operational grounding threshold lies, a correspondence for which the framework supplies the missing theory (equilibrium diversity, and a -per-capability survival law). **Pseudo-rehearsal**, the replay of the network's own generated +per-capability survival law). *Pseudo-rehearsal*, the replay of the network's own generated samples, proposed as a cure in 1995 (30) and revived as generative replay (31), is this paper's ungrounded null: immigration from a drifting source, benign for one hop and compounding into collapse over generations; verifier-filtering (32, 33) converts it back into grounding. -**Parameter isolation** (34, and frozen-base adapters, which forget far less; 35) is the engineered -decorrelation our specialists use; **complementary-learning-systems consolidation** (36–38) is our -periodic adapter-into-base merge; the recent turn to **merging as a continual-learning mechanism** +*Parameter isolation* (34, and frozen-base adapters, which forget far less; 35) is the engineered +decorrelation our specialists use; *complementary-learning-systems consolidation* (36–38) is our +periodic adapter-into-base merge; the recent turn to *merging as a continual-learning mechanism* (39–42) applies recombination within one lineage over time, where we apply it across lineages; and -the observation that **rare examples and long-tail knowledge are forgotten first** (43–45) is +the observation that rare examples and long-tail knowledge are forgotten first (43–45) is tail-allele extinction observed one model at a time. One distinction is kept explicit throughout: catastrophic forgetting is largely deterministic interference from shifted training, whereas collapse is stochastic sampling drift; the two phenomena share their victims, the rare, and their remedies, but not their mechanism. To our knowledge, no prior work carries population-genetic formalism into continual @@ -105,36 +105,36 @@ learning itself; that bridge (replay as immigration with a survival law, merging compatibility criterion, consolidation as the slow store of a two-speed memory) is where this framework may matter most. -We are explicit about what kind of contribution each claim is, distinguishing **interpretation** (an existing result understood in population-genetic terms), -**explanation** (the transferred mechanism accounts for observations existing accounts leave open), -and **prediction** (the framework forecasts an unmeasured outcome). The paper is strongest on the +We are explicit about what kind of contribution each claim is, distinguishing *interpretation* (an existing result understood in population-genetic terms), +*explanation* (the transferred mechanism accounts for observations existing accounts leave open), +and *prediction* (the framework forecasts an unmeasured outcome). The paper is strongest on the first; makes concrete progress on the second (separating merge failures that are coordinate artefacts from those that are functional); and reports a first, bounded step on the third: a controlled predictive test in which pre-merge functional-disagreement measures, chosen by the framework, predicted merge damage on a constructed task grid while the tested weight-geometry baselines showed no detectable association. -The correspondences we develop, summarised in Table 1: single-teacher retraining is **asexual -reproduction**, and the irreversible arm of its decay shares the defining consequence of **Muller's -ratchet** (46): once every copy of a rare capability is gone from all parents and sources, no +The correspondences we develop, summarised in Table 1: single-teacher retraining is *asexual +reproduction*, and the irreversible arm of its decay shares the defining consequence of *Muller's +ratchet* (46): once every copy of a rare capability is gone from all parents and sources, no recombination can rebuild it, which is why remedies must act before fixation-by-loss (a consequence- level correspondence: the minimal model lacks the ratchet's recurrent deleterious-mutation mechanism, so irreversible loss alone does not identify that specific mechanism). Injecting verified real data is -**immigration** from a non-drifting source (32, 47, 48). Model merging is **recombination**, and its -central payoff, a merged model exceeding every parent, is the **Fisher–Muller effect** (49, 50). -Merging entangled skills courts **outbreeding depression**; screening many candidate merges is +*immigration* from a non-drifting source (32, 47, 48). Model merging is *recombination*, and its +central payoff, a merged model exceeding every parent, is the *Fisher–Muller effect* (49, 50). +Merging entangled skills courts *outbreeding depression*; screening many candidate merges is engineered recombination with unusually flexible parent choice and pre-deployment screening (we use -the shorthand **directed sex**); restricting who merges with whom is **population structure**. Merging's hard limit, models too diverged in function to combine, is **reproductive -isolation**, for which the Bateson–Dobzhansky–Muller theory of incompatibilities (51, 52) supplies the +the shorthand *directed sex*); restricting who merges with whom is *population structure*. Merging's hard limit, models too diverged in function to combine, is *reproductive +isolation*, for which the Bateson–Dobzhansky–Muller theory of incompatibilities (51, 52) supplies the structure. The nearest precursor to this programme reads sex as an algorithm for mixability in the theory of computation (53), pre-dating model merging; the model-merging literature itself has strong empirical operators (4, 54, 55) and emerging merge-success predictors (56, 57), to which our delta is mechanism: *when and why* failure is coordinate versus functional, and what moves the boundary. We support the framework at three tiers of evidence, in ascending realism and descending exactness: a -**minimal analytic model** validated against closed forms to a fraction of a percent; **small trained -networks** (MLPs, recurrent networks, an MNIST image generator) where the operators are measured in -real weights; and **language models** (LoRA-specialised Qwen models, 0.5B locally and 7B on a compute +*minimal analytic model* validated against closed forms to a fraction of a percent; *small trained +networks* (MLPs, recurrent networks, an MNIST image generator) where the operators are measured in +real weights; and *language models* (LoRA-specialised Qwen models, 0.5B locally and 7B on a compute cluster) where the claims are tested as signs under seed replication. Negative results are reported with the same prominence as confirmations; they include the failure of an internal pre-registered prediction, a null on emergent speciation that bounds the analogy, and the sensitivity analyses on the predictive test. @@ -145,7 +145,7 @@ 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 we exploit as an +resampling step *is* the Wright–Fisher process: the same equations, which we exploit as an engineering gate: our 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. @@ -154,7 +154,7 @@ The boundary of the exactness matters, and we measured it rather than assumed it approximation, optimisation noise, and inductive bias, and when trained networks are fit against the exact drift null they deviate in *opposite, architecture-specific* directions: a smoothing recurrent network resists collapse (keeping spurious variants alive), while a sharpening image generator -accelerates it. A one-parameter **learning kernel** (a smoothing knob and a sharpening knob on the +accelerates it. A one-parameter *learning kernel* (a smoothing knob and a sharpening knob on the refit) reproduces both. Throughout, a real learner is therefore treated as Wright–Fisher *plus a signed, measurable estimator bias*, and the drift signs (rare-first loss; the grounding response) survived that bias in every architecture we tested, including a convolutional VAE retrained on its own @@ -222,16 +222,16 @@ cancellation is a first-order result about rare items, not a universal impossibi 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, **weight averaging** (a nonlinear network's weight-mean does not compute its parents' -output-mean) and **routing among intact specialists** (58) (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, +operators, *weight averaging* (a nonlinear network's weight-mean does not compute its parents' +output-mean) and *routing among intact specialists* (58) (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 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 **Fisher–Muller effect**: recombination assembles, in one offspring, +The generative payoff is the Fisher–Muller effect: 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 @@ -241,10 +241,10 @@ worst-family metric the merged models are the only ones competent everywhere, in Sex has risks and, for AI, an unfair advantage, both quantified on rugged (epistatic) NK landscapes (60) (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 +(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 *screen many candidate offspring against a verifier before -keeping one*. This **directed sex** converts the outbreeding catastrophe into a reliable gain in the +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, @@ -261,8 +261,8 @@ structured-population search, mapped onto merging populations. ### 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 **grounded inheritance**: external samples added to the reproduction process (the -data channel). In the society model, grounding is **grounded evaluation**: selection weights true +model, grounding is *grounded inheritance*: external samples added to the reproduction process (the +data channel). In the society model, grounding is *grounded evaluation*: selection weights true fitness against conformity to the population's own consensus, `g`·true-fitness + (1−g)·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, @@ -281,7 +281,7 @@ 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: +separate species (reproductive isolation) through Bateson–Dobzhansky–Muller incompatibilities: changes harmless on their own background but deleterious in combination. A merged model is exactly the exposed hybrid. We 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 @@ -307,17 +307,17 @@ information-theoretic (no single model can satisfy contradictory conventions; SI Proposition S2), with the framework's role being the *structure around it*: which divergences generate conflict, and what moves the cliff. -The strongest constraint comes from the pre-registered **emergent test**: true BDM incompatibilities are +The strongest constraint comes from the pre-registered *emergent test*: true BDM incompatibilities are emergent (each lineage's changes harmless alone), so we let children diverge with *no conflicting signal anywhere*, using complementary class specialists and divergent input conventions, to 6.4× the base -training. **No isolation emerged** (residual 0.000 throughout); instead the merge *rescued* the two +training. No isolation emerged (residual 0.000 throughout); instead the merge *rescued* the two catastrophically-forgetting specialists (parents ≈ 0.50, merge ≈ 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 **function-specific** hybrid breakdown (the merge scores below both parents on the conflicted +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; 65) 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** +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. @@ -331,8 +331,8 @@ so inference is condition-clustered, and because shared seeds also couple rows * report per-seed and leave-one-seed-out sensitivity alongside) span three axes decorrelated by construction: *conflict* (contradictory conventions on shared prompts, private budgets fixed), *compatible overlap* (the same shared prompts under the same convention — overlap and volume without conflict), and *duration* (weight -divergence with zero conflict). Before merging, six predictors are computed: **confidence-weighted -functional conflict** (bilateral confident disagreement on probes drawn blind to where conflict lives — +divergence with zero conflict). Before merging, six predictors are computed: *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 (56), LoRA-delta cosine and distance, and a cross-task performance @@ -340,10 +340,10 @@ baseline. The pre-registered outcome is the merge penalty against oracle parent hybrid-load analogue), also reported against best- and mean-parent references because the predictor ordering is sensitive to that choice. -The supported conclusion, stated conditionally: **across this controlled grid, pre-merge functional +The supported conclusion, stated conditionally: across this controlled grid, pre-merge functional disagreement predicted merge penalties (clustered bootstrap CIs excluding zero; held-out leave-one-condition-out ρ ≈ 0.35–0.40), whereas LoRA-delta cosine and L2 showed no statistically -detectable association; gradient alignment carried intermediate signal.** Head-to-head predictor +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 ρ stable at +0.37 to +0.53 in each seed alone, @@ -351,7 +351,7 @@ geometry ≈ 0 in every seed, gradient alignment seed-unstable at −0.11 to − bound the claim: the initial two-axis grid's best predictor was delta-cosine (ρ = +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 **failed** (they are statistically indistinguishable as rank +weighting would beat raw disagreement *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 @@ -393,19 +393,19 @@ evidence of incompatibility* — in every regime we tested, what broke merging w conventions on shared circuitry, which is the thing to detect. **What this offers continual learning.** Read into the field where these results most directly land: -(i) a first-principles account of the **replay ratio**: the field's constants (≈1%, 5%, 25%; 28, 29) +(i) a first-principles account of the *replay ratio*: the field's constants (≈1%, 5%, 25%; 28, 29) acquire an equilibrium theory and a sharper prediction, that the required fraction is set by the rarest capability one refuses to lose (the `1 − e^{−m·p}` law) rather than by average loss, which is -testable against published replay sweeps; (ii) a **failure theory for generative replay**: +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 (30–33); (iii) **pre-merge interference -prediction with a mechanism**: where the current state of the art fits regressions over candidate +unless verifier-filtered back into grounding (30–33); (iii) *pre-merge interference +prediction with a mechanism*: where the current state of the art fits regressions over candidate metrics (56), 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 +(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 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 +*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 we note the standing objection that apparent forgetting can be skewed task-inference over latent capability rather than erasure (66); our irreversibility results diff --git a/paper/pnas/main.pdf b/paper/pnas/main.pdf index d310b17543c1eb493f19036d01e99527323d3de9..d80b772c8213a1d30990bf5d769c3d2c1788deb7 100644 GIT binary patch delta 70412 zcmY(KV|SPhuxR7PX>2!cY}8lJZQDj;+dA($_scy$V%Ds^XV2gON<%wJ z?GZdEV;u4?fQ5~njgJr3$=T7wzy{WRJxiN2=Aa#U`jh4j$vT6bDwok8biiAZW9(ADfNf$@0UHt4d0pk zK+osl@#PHAz>a&x_p765W9InvBb4=-Hu>`D{$*tVNY$Uo*`LU`e{Rg;A5Jyz@ooHZ z<-x}>{q5$Ct?jeM_3?hpFL!pUR+cuEIA7M`*fLb?cK~1;0TpQ7+I-(n^0_E-<=Z}= z_FCmtQlD^|%P@;Ai-*ji=ONo% z!};As=4vfEH@g}i8F5fPxewjsi0hLN=_F>9-l+)%R#zD9rTB%<3GUWGWfhgVD{l!P z;{_i$mgP_JDu8$r-{Nx0<8HP=^|IK;$Z-JfGn^8^?_TEOTo|mhan#UY@329Q!eRGO z^|yEyKJO#Mn6*=XPxcG82_-C zvLc+!Q?;X>0y~>{+Ru~CFOj{3IvE`f*ZQ|7Y{d{lc~Cju#So?&MxDw_8f*s$lOyv? zYAIjJI1e||zCm)eX{fER*c+%gSgiB0f2yjmx4l=-RPdSzPf7eW^Ue)E-&)73P#eXt zCdyXcNT)ClW%CxtHQ#H`f`~MUg3>VS@Tz4CampHvwjoDuy@ZRYwIpi}(s^wZ3oOz# zY;y+VM3v1`EO&3MM|c8%3RVH?4Sp|A;&qZ4LutB~fS9wWWn7 zQ0eYK5FCt0NrCDzPSyOD1$_RS1ajV6adQe@`I5we)UnauwK{# zi90@JH8}|h{Ls=E4m9mWA5qlR@KhLo0si_m!zYr92=;KIJk|~KK|CJ74f;{s$R(?{ zUkoi!j3CpTXXlT1g|G0~JeH`yn^dQ8&F-_KDX{~dipoY)SdGmj=W7`5>&qrt+w3-F z&#;q#Z38alHVoq=^45+xoH3UAV@X-LGrSUz$lRp;%~M6`eZI~JJ-ALz>qf9nbF-2i z8^7ITE=j{huduFW%whv%(dU*6Oh9XGct{#8C!2ve6n|9gWk0E-3f>PLugIG~__Zz} z)Us$C=zhM;%{cioAQn3<+ih*HmK=K|>>pMWH*!z&`4YXsnJT&sgx5Nd{ zK`}#Wl0$3o4lGv8cyN|*32+^Gs^^7@co9?N?VXnq zT?xHI`4kpJ&~nEL;PgwE4ogLf?q}lUcq2RO2q7wWtP~8z&4VoxK~T3D7*Bvm2K-y1zU-iI70Pn((t(BJU0GZ zQ5OtojnQ1R6IoE<)f5xwC{h0Vt1>GRpiEu*Dx2q1<|kT91+A_1s8148E4LVLVmKga zP!4C7@^LiHwikQL4C|7)jRn@hASw>^Tqo(lX5^lCO8^1rwtKq9|K3|F=~V|yD;Le1 z*bpn@hS$Y!$-PI-mLsIm#fG*<_O)9d9h7Mh(@L3(Fecx4Wf1=*-A(7bk+v^+P)=YY z&|TgYMntinfb)JvY3Si8Q86Y7^>8Nz8IOZX{;}SX5)JZ@x|sgg5X0SD_DstQ+o`ve z_Mt7BbMy-cqtBZOQ#UrZHAc?eTQegAny&2auZY!a!<&2Ie^~}LXaD?Be|ykZHFL@t zf-#PErh2t0mhXUFl^ouqVF=C1jg2eD+7SH?Ofw7M8oA=7`-*F|MPzOKS~YkR9hFrL z#LF6D=9y~>18t!Y6$@E7*sFA7ZR<1WhYWfco&yCc9rK1jHzK7^UM3_~NIl!d=HN^QGyA z(Mk6Apf9MOzeG8OLnC@;I*J|?D_uXPVe$n8U_21-jMDQ9BuTgN7P|Jp%*F*sP2q&> z3xZw5f?QLQjgDO&mqX=dSx2ufM8$r-NnDlqosNq-ix8AC~PxS^lV3Y!4Xq4P6s zbj>B`Y6Xp=stL!#3Q9X%t;bsyDioTX)9{924%vMzUpYYi#>)NE_oEE2ZB0IDLgH3` zz)c)1|7B_*$VlrKw~c;sqCELW)Q~zQ$0i^Wp_m`VD`z8TFuycUKGEkC~&KE2tpd5}^hrpfK z%J}z>q6wzN30$8lUf!N>N|LCHO@|_!_=qG}$$FYa`=YJ=BmllznPU}*NTpQOk{MR^ zDw;h{T%vSZ6AW7OaqS96h~mCZVj}01fu8TF4h~>~O1W-6i&kYy5_;4C>5*OKkM!!e zo8|PD7Y%FAfn_3iVYz6ZN6S7NhuAZ*-AFA3-wzSI@xMoqD|zdt1-o3`k?n({6@yYz z)kkp+9V0<;HR8TL2;c(-hB_G&;j=vWR~`AA9GnLu>8W$q>y`_p=rsGVhl3r0C~R8QEJt$FH}LRlNO zjIBjlm%QcjPQME z?o1?-?$i^-rS)^@V;(vyn>~ zwf_EZLe%t&bw)Dz(e&;&&MQ|a4*C1M;=_o_u~xGVJPCE8B6u4SPHCws`HdSXPT`HY zIM=W#A7Yd)pB#W?gS{Apqn4I<)+GW`L$uGsp2kuLMQ}+^*%Y`&ryI$2Rfl2S4fSY8 zZP72jE#?$5zLE;kODc!IqkDp&h2gQsoj*C2wq@V!orI^rq((cuRZ z$T>pG8B?hf*u3Q$Xs&D6XWt(-fdOZ_omw`pSU~*ALy09O6s4*>sLBjR=Ks^w*LsN zD*7r+b^D&o+n|GJtF0dHtQ<2wO)TAEFzKAqcjG(8ik=BQf>)JFv_GMVUQT(2V}r1z zpAN7qQnLZZ8c^tf=)<9`yji_J%UCLXZI*31^nqt4Zm$m;E1N9ha7*J7E*iaoN!s4> z!6{NN0q&9+ryo{{2X!@-+y#TW%)1ssf{KE1FSi`GOg|=@dh&j0;9{DvvJsFDP8=+C zMf8F>=j^RP&}nsGP3xB?)J?CJbCkpQHa0heE#As!UnmVmzybJ+s;Taej>7kKSL)z%?M_ zFS~F#gqmbbyld}=skEv2(pNN_DtZ*x>eKfh`qtLXAvf#+mxuUIJL0%~= zNfW+qPk0vj{8nB;%*GScZCw<-R?s}$*|Y)9T(pDOTv+TxytEmI7OAYZ$*A!MOeWlt z-mdTAK-*?RS{^;z&kv%c`x&iq)rcj#C;lBk?Qf-=rZ*taqiU}jhLf+`8~p7WGPl2< z%Wi++8hfcjh>82h)8ZW_Z8YQYiI;kp&ozmkGB@hZuk$^a=YVdGiEd?9(XqvbSrZRD zP5lll?2^_dL;6MEy35?o-2%2HGQ^qGkb~bmI_cK7h)j9))&8t6S%H5PvOU zrpj{?P_h`~Re6mNrpW(OsqT;S{0{dJh0pt#I@6o2w%#G(wTMwS)u9)|^1C<{o4--*J%+Tl>XOveO#NRQ%i!1_lrdt8*hp8f%UZY32?SZB)-d^l zDGv5+$p_nb$>UIZgi+|J*ct;3moM-?UXCwrvTcJ(jaq?fyvJsDuI-)PoYh2Z4PJ!Ho|{bPG{z0#tKV@`GkB-ed3 z#szY---R~Qg52MD7||B#r#8{skg(4OSLFPHN5ziab*dGVZP?h&uVb_WMDp_D81t zJy}1k&QhHI=@6E2FV2NdMqCREbWp%|NblClNoq$htehX6X3)-8j-sfTPdr1x>fVjL z$ccgNeY(*d+-M(MI#my*z-+o^Q0*y`;WH!z+*89g`+xF6~Fl=UD?2zs&DX2bhz zho%htK3UUnq<*JjH^bv2mRy}adiO>0?E&aQl0?0Y}s5AyXSiJjHI5X@`Fjc-|K%w z(o)d)>;qb7YcV}6R|tk*cOCU>DkYx(89G;N#8C3z3Gog=|NCZD?I53Ru51D3eb=q>n$3>ry zKJp-^u_qezY1pxM!+asXRsW7s=8mE~N@O`bPo-<3qfc(FfKl)$K~sG^9$)4FZ~P;_ z8&Y6RY>od{Ii*M?gTjKc{I7}PWcWWOs!F>p`WI*X(lO25kNMDNPza?z`ey!17?d`3 zLZYI_5~P#gH0OnLbnyhpti0xETmwk+5*<0%pZht4PXn3XYcUsmj^*N1tRpdc-yAic zpC8Y9gx|OO`!zrW;WFNmit_kK%=P`7**m`U=gaL2u#>@fY9Pb${rdje!bq;!ebha3 zUj6xfzHsPlrQV}G+yeXI)YHuu7Wba5Q2G1}RQFqaLX6+<*|UU~ zp|(W_bS)AF9EY6_o3F}sh5Ne(wyG|pYQEoZce3=p!t&bt=y7~6d`XF*Rs+RKk?LlC zkeb|!PT0cT6L0G8wIbV_jgVq7XK^1F;2Ry{_RM8Z8U`hVeRU`z4Xq>M^bkm?ya?(k%lk z$IN1+>@zwWF9-LS9fPmZn0C-G*0u^-vwlBTfeq^4e0^U>$L0%xP@LOQtQu1&1Y=~t z20mD0eXA~G|COOsI$D(Vg-c`%^Cy`*ZZk6A0^8$&pt}kQU|zU(BC-VD@L>;gD@V!` zCxAJ!;dPUvnhj4w&ABo6n!?vOAN?@s-XyA%SNbC)fy<q%1Y>Ge z_GEctN%2NE_$Zh8{je-Ua&6ht5^^|Jk+zP?LGiYDMY+fOBv&EAsf-Aq$YlUiDxlIy z0jx3lG~v=4_GTg9&s+Mx1BgC=f=K6G+S`FOKXRaC)*zoMifXTL4M{3g~R)o)xIn)?{fLpfMI zGV^iW2`mIQRxq7LC@vOi`=Nmd43U?^Kup)1DyoGrK}g*+H!_OMEPNimt4gJ*V9;LCrl8d5iIZ zKdiUtRhPAkKJ2iN(lFU1(OlViUhMxmdTgjNyWe#u269NGu9Q04V+fvwF4ideUC z8n|n+eY#?+l!>^t39eibUzetOg9IdSGRdH+RLBCXXbS2K6iGsy0Ne16Yc1Kjn{Kx5 z1OrWEl1t8Gjs`H++&?*M)y-2Kx9iroXt+MDeh}@f&rZfM>a#$-I?#N<6PPW66nEuGAEVtn56%ctyspFkQWRZnit$c+HWUv6BeFPqe#rrE9K z1ui$ZU*rJg2m(awy^KMHd|o!(bf`RMP~AW4K3}i6&WI)s3d8FP{drUrNd{i=7!C>7 z1lm)2mzCWhEoZqqcYWxJPtx~&PAS#9l^qdW$~>t*#uPna?&r}?KPa?2@1+~7_v?S* z6@2Sj4~Eygs32KsPs7(jiy$6C^K64?XF|8^^iBYLKAalXMvl_kxN{lo7Igs)?}vNJ z=Hs)aLjHbNazo_P=%OH0DAe9DvyortJ7-Zp43sx_Xa)0UEdqU#JZ-gW{oR?R`G$qA z??{pd4h_jql!7B&qrZ@!(c3~GYDbQ2z*r$FeI%m3hpzD!A>y+74Iwdv-CvB9*hA&V zsc`}AS6&v;s7I+c-vl;9FVn;dUIW|wCkpAm!sOGbX0Y@0EUQHYtIXI&#nd*CAPI~I z>|{d(%R?U`O6Ko7_C!QKICI2$7K{9du4-P|vK3tRx0TK?_Nx za0qA2~DwlJ>McP_(_JW!|1z%|8(5kim-S3&5Pm}lU+>k(z4+6e^K zKHy*CF_0Xl$sfChu@|)4Ab*C<7h4$Fd-RbH;=RK)bUq1H$+yT7v9$#DD%1z8mc0UG8cHLMN1kcUJVu1}MwKs38A!mw+3`s?L|AY&SRdAnL+YJ@4v=femO= z2t#3;w;y`ccZU9(1fMCYZrT>~qk#u#XhJJq#x5|zCOj&Lae;+AMlliGkn-N>+wFXX zoS`&7HM)LA$_rJ~?n&YYL4*m3EGV)1T2RbH7kyVp6Cm=KdYn9luy z?`fx@c32<)d1l<}PyN6);1Pjc$duW~5izock#Bcqga_I;9Z+eeb}+Eyv?viXqof_n(({dvqbX^mb`SZP|} zS*l*-Kd;PqIvTnN4i8IpGap}Z+(F1z)+WXXfP6nsh_rK3g{p_Je73O z=E|@qjGTJS8~y@vSnr>`xkfP*m4X!P+RaxQ9?f-!JSn9Vt?*yAiMBPHE}i>v8s=PvCC_1X+ZT967Y?i$HoyYufqHx8ERgl~CYK@@l48W} zRHY0lmA+qp??nURs^9V<(Nom6$Q6wQ<_jap;A0}yMb%WhBfgfXDz7vJBNop<`t90A_Ds4OQF^XZb?Zb zLJnugO^d+$f$ceBzWrEcbEpjrdMfM?IC~JhM8Nfd>NzvF>rrK1GUeItdYL`-sdP9l z?aXnZ-Q<8OqmgHsy-G>5bW0j5?HbP6K4f}hR~)yj5XP@rRtfYxXPlZQf{|3k z|5jY!tqEIZmQ)m@-ovfWIRj9hPaKtSj_S4Vd}#CyXH+CwFj2VYdVDGB({D+8D0xgxkl3Iay4IURpiOuc`bV?-MgR~Z3+s7VAbF5Sy` z=AZ#zT~=G!0tN8hl8G<-Y0MUWMtAHme5z6|hh9`R_$-}fxyIyV=MJ$U@f~s|qmCg~ zJWh$l8cav8g|nu;W9QKU0|8KTm#cIlB%jR^?X5*CdHPTtPdOg^XE@wvx&j!xPHQQ~ z8fl*U!tIw0CnvGX-r{Zubsr8H7+nsis5oH$Rs`^D?BvY)rGpY!s+Y=kR-b+$_Np#l zsphu&Ggx0>p*^QVOvVGMGj-_fVaid?lR|J8d^)Fl3{Xk zV|fN4!4Eo|kAp{xe%@fD@Eg^~nV*@lw|l0rIoa${Jn>sEo6%S#q?0dbGf~&^umFC% z5J^MZQg`*f-#I5YSG@$4BT8KeKR7bG=tQAaLG-nG&j|^-40LFwje2{fmoD>YkGA%Q zVn&>JAoqlGkeU=TdtgrWo>@V#!unD7fWN6}42A6HVJ;0t753jlNUk(mVEG21M9nFZ z>`CzB(DjtX&B*qu2$##FHhz|=F9ABIZczR$PUi^DO9b{wZtE>YzSD%N`ab4KjBaZ@ zJ6E1Aa!@rF8Hpzz6mnA>_$m5#@bYx{*6}j1W9FJN8+ZVDfptw}f!9rP{};Nyx&=N@ z6I@j758SDUN*rje92Vp-RhI^Hq?82tj;m$58e|jJyefF5fF)44`-=*wYGC>Gp_S3; zf{}p*>u+TeV^3h^S36JSU5}MZokpCd z<{hhDrjF|v1@4G}D&LkODEz{SE9hdMuig9wbSZw{6<@mcLw2)#{45LQsrD}oD|A54 z0PpWx*77?VCVMxFzhZKUBT)aqeGIFFCjRHqJd=5xyH7G?4?jErh187spqhVw-zj#F zb!ji*&94v26jTZ=n%9M)hSg`O>(&s8foihgvdn1BY?PPP=)#p_IeLcR@{yyhfI}lh*TE`8FKI-uT>oiBmf1GT;$ZWO{` zciTra{)si=M{zh7eBtb+BjQg3{d?eVVyz(>A>$f z!ywrV>d274lkUr7G1q0>6M_t{(-j3N_h|-cU)G)?{7!a>=F7v}t}+{9x3Tm{b;O@0 zYbkf-#Cmq~N$pp@86ht?*LllFmym7+%WbCUu9RW1-^D!YrB)RNMp9C6RbdeRY z3znBqnx%K|U`0i&e3*R3BqyX#9vrf28taO5)QduqFqKQpFza3iLM>&swDQHZXEaOS%m7eH-b zysGfsy&ou25dSFf&GpUx+{=>6E7fgv? ziyP2FxV}F*ZU~-P>GoGl3GLMO1%A&KugeA9=6??DU((K~Q;9@B;cI`suhqcWWVlZH zQo-F7x2u+-MF5Q}&(0uDgNd1*{xx=dw`_yy5ud&I`Vmog%}DPyJ=HNpi*{6_G#x{@4=Aea2x0hYLQhPaDa&)Wi6RCSrdnXMs`h3Y;dVH8ypj zf@}9%`TEI=^Gd5`BArL?!rgpo13NA-HIY}w_|_uL%h$wDmo~<2`w{GQb8G(^E`@+ltNwjw1{;@~7WMa97Xuc{r z2*6;2nBZq8zOOcXqD4Cd1>9_Cn&+U` z5>sv6va&*0;6C-EYf?bf&dos_RiAQ>T`;#;(smWvFIE6lKQ z=M@m4+^=WZ9axN2rlD@0pCikp9c3d_G_d}z2*nF&x`LRB{qFnIj&7RtXsHg*-j;m zMkQg7AP{SzoU8)+>2Va+32}Vp#{JmH>X_t z@Af_G-9b>|B=1b{nYSB}Uy$iH3v4ZtQXN`n7}UkA&Cv}pD!tt4?_li(io-x578pBYch&{Hd^J-5P)os;psP2heFsiDHMEBy<7Vn%+#e)k7 zX^S9aHz^3ocZXfnOP}Q($$SdLhQNHc#!pj#cI#5mo@SS7Utnw)%1#8&8hCV-eN8^=7XHA#{!~CVa(`E_zeZZP*Ru2&s(N$#W{; zb_}wa!J1nbS4G%Qq17W~|G^oEErRM!K)WLwF(Sobh~^W)H-yJTcMJLA9)z=`k)&#L zORCKm7(tA|C$XGI>JFix=XqpS>z33(@__M+`h+~dZT_AUBzQ#is<^!`T;z^Z-VCaD zP3SP3QkeYld`^Zjly)qu2`d>5SHlyyylP+Q6FG0ZcQW=n(iY0d>LIJ-QD%y+HOBun z1cwjd2<0yu#-ThWvaCurYsSK}^J^!KW*0#e2mJ?LJ5SK36H6 zS0D|1KW@1AgIO4)L|Zc{C$3LK+UkkRcNk_pTPi-Pb0^{g^9Kfa0SlD<2hm>ukvWuU zTu-hWQu1Y1*CPOlnu^s>_|mL~h5Lnc7=+Y+?i8D)RueyAqHZ5)43dW|WXIAFHuPTC zM%I<+krlV9(TPgwt|Kfyws< zC`A`jK})xK0!j>Y_+GYLa$6`MrTNU9%jGdPBwrGzl-lx4T%m~v!S7K~AwKar63!(> z+-`0+Mf{Dn2z1@_*#zGqksCeX#8p%@EEkt?_&F%CTM?w1>fSMSI?Y5qPpY~3)4`dl zT0%BdBm~H_o9W)sA)7|JN@@G?6PLib{rQNaCKmUAn=zAX?RZfQk6Q+~dFGGlSctw} zs!yB@f_|%cB0k8Ux03w`AY6e%StCH-@n{L0B_)D+zCQT7G1=-aGv|!}HhMeijBddY zJx>>Y5v`XG6E5&CP1wg-BwYzrxBdbx!UD7XC`8nV9J=T)(};pE>bs+ufBA+k>GMeJ zH!ty6tT+N=w$mK?Nc9I`vz;Owf*^DT_-ICqFQAnbVS2aDxGdAG%EQ5<;;%S*1J&i0 zOxH<)IV)!K_H0h&18d8{ zAT4UM@9D4BhPc4->1^hi5;q^&_-AT=HTn2M{IQ|voLdj#MWK6yFYQ}FbdhqpTIUZ_ zKT_!>HY0p5xvZ~svv8B`Oq#2Li$HIwk_#tqWdc0r=1-y2#V4>bLek_!%NL6}hcvgp z-H$J~vS6h%CuPVTDllc0B)J2bY5mE$(juA+FL=J}{Aa+M%SjCladAlrU)rx#o>Owd zUo?|TZX%FVbH)^}ufz^!{5#7*9pQ=z${wG4LqFtj+vMvE2#LAsWhKVv^nYy1-Qp$v zmq}_>7*+C88bAr-5;OYd96b%gYaedDkfeU?+e1!sx#ZVy__2<3Sato;9#-aVw=tSl z-W1ins}Kchjwp7JH*D2hJ%1mGvxB|a!pB}#GlL;K;7H9fuFsE4f1Jl`?ARqKaBhCJ z{B4&$gf1FJY3bSWt_`IpJDdA%`#ip+@7iG6zEd=GhUuUkP*j3#lbb%hnG?Tihre>V zGS#52QTB=r#b2i$r$t7r>uYQ4NsY zs;l!rIfI-XU_I~3Ma?UNqPc%vMqqfPVs|)0pAUu|uuR-de@F@QSFj`gG3Z4*t4K~a zubu%UkD6Ymu(FKg!`rSS3sTB{q|yMQw2Q$DS2A+WfrtaYf6;8(gJ| zF*ov(lk<|iTCLr-W!mbh7;Q|h5GnIhVuafOecvkQwR5n1RbFFgIhT377;%9gV-4ml${|cY6w) z9f(&f$pBky3D~s57@$bU;|uh4Afi^H;w5&3bMcTYG=U?tWPI~?QW@auD^46$Zpo?Y6wq)ESvF;UI1hBSIL$WpM0QCSXdfw3lDgq=+kU)i>kQVx9U&jSn1Ofm&42UuwU= zaHkd%rmO^D_tA#z&5UY8-;AdO_fn|^feGr4IyLA)e0otAr6WV=>q!&-cQ{{2*fH^B zgR%*$6^kB#KNEY!gckFPCk6)9?*4S(sbzp}t<386ppQSM7<{2Ga{~qCa=$^KydNS! zeoef-sI^DpVv=4i)xN*3VkK361&ZIQJKKtlP$E3gF>RKB)WUs3#ivg&UN!lf_;Em2 z`YR>pM1N=m=l*;yLV${F1+>g94&zXc;>em?)e`s9mvsdG5sA9-~XAHj@22SqpQPccuo8rV&7~{a)J4? zu|>y}?#pxa(N)5pwrk(_L(0mhgBb@XeL;E)#?vF?emDBkuxn$}mZAGz9gsjb#~UvG zCts|8J8E!;(}Rm1TFVx_;Hk9VDiO=FoOs(%IKY|F9~~M_LDA!@fv=p_J;cx|`8z1e z>W=?Sp@)pH`t&-dqT4nLpoYsA;=6Vvkf0l%;J z@o|o(&e+hy?fxx4J5aRQq6C$AcX%yKg?ryq_R;;l6Vv1Sv4W&2c5ET{zM8^#>PSwt zvclMN#7SLK^3L^2f6zqR_?=PnUIKHqoPDsgM*qI_Erh+YJlDg@G2T#bs&n1)UZPAr z6HwDZ^_`CBgMDf3`$c)a{y{FVLzSf1n(>A3jp<8&ex&!d4EU>|@9BO0dRB;?zL-bL*U{mGSit1 zVw|5}2)Fpu8!ZU5@#>Hi7Vl8N4|-S@EKvFy*=W$+brPNy7rZ9$=(V zRE;$(1%Ax5{om_SXT2b&`-WUiZA)9G?CcDzhUbJFdhIZV#O}wh^;Z=m-dLvlikj7! ziVrwb5bYhGJsfv!LDs~I)H1Py;q>Ql;TsbuXV>c3`RXJFo2Hs(GhPDyk+od7(III6 zw%dYVoghfngaaZubTIe?2b*_LZ}7sffw>{W`XXmy zju1O!@#k&oL8IdIt&V~x#lh`LfuvfF1*g`nhU!E`-$%B^Bs$n%hcsn-5gyqTz&N=M z&R`3I1<=Y#y$nGz<-1=|iWj`2Yzj$4cg#!tv!%EbhtbMs>;0uaf3x%io}fAj5*7lx zY}YuTCyVVe736-w60?aX3M9<| z+cjxET(i0OwAkTS(U*Ft&zkW4hY7dfh=#3;e?!(`IBwJPMTBN8+ISJM*Fd-+`aw5y zfR@?82{GI>sn0HfY)rc3$^9dfeQ!fm7vmAi1&Z{Wn$(ESp_%lfSld9h26K(!iOmyM z|KKbNP2xXG=-E&s5ILYzUMBQ~U_aN*&xOPq0i$~HmYa%qHRL{(h)D+#bq^ugx|TaY zpud@G_V+s z=o|qpiLemqRhCCka{B|5yj+?Fm&XG{;|?)fNNICLkRdpGsEgMihO6llCsoR2?!`n6 zBa7S=^NZjnrTPTajF2AR8;>sa(-Y}D-accND#INfkW-7X38Vy^1#V2AGLE-c@G=E$(ye!#P+4- zeFEBzjdQ_gP9a1LIqF1&!(ZzQ}NrOKs# zg0#%_3_Z+hW1+Ktu@GeN=jN}HAPHO;!`LAS-btZC;jzzeJrT|!s|fpqB|2K(E*?==n5 zQ!1kGbc9dk2wjd+$S=aK}AEz3g zgd}PSKn+zY2{Crut+qW!HE3CmGgXY~gu>SU6GiMUE>)*&vWijs@UELykngV6Ohj`t zjBSeRRYip35lw>(z(B9j zMwO;B3n8LT1VgJ`PK|^g-vtM-|7QW?_Zr?m zU>;=PIBOEXbvS7x9j^_?23+9^)P=PCjSc zxKtc6wEMXXBj!40ZN%;wXgg`}*)pmBKLBe$l)oJ*%^<0d7agckmljl*I$2q~QF}z6 z#~Z!e7_mlRbvR_QdiK!6?a-2<-3=8)+c|V^!`6R53w@c|sigxm3PvA~dAKZX5z*U1 zi-cd%16&KgR_~&`HW&F1J)C~wE8?Ec5hS{T&l=*)=yyEuxYTqL&-rPEI8&7ayo$*W3!8+SJJLHDXNr#Ugp1 zZf_~GXK7ecMnP?Zf;5R%dqOIWT7SB9Yytv@{ZiUaahoxh(h)|m&+~y*a)AZj+vdrd zZmZoPh6Ufdc2G*U)hFMw>w(e%4_U^w#%+JnjM%d=5SYWTDYa$|&*c*5eTkgtz7VA@ zpCXJLm<_Bqrs@Xku<3rIB}Tp*X#CVR^Q51K1rU*|qkU%^z8$v!mQ2yOp}-T54S2^a zn8*>$4Szl>&Kya{mzB(PMeSHa9uKewxX;`4nCBbaCO>Cf*09z`y$i+Zpbi$Crx1TF zb!Ctu{6-2=OunZ`3|G>@u(J2#L=fX4d68XN3x}xMsRs(?T=eW64aILJ54lyG&Wt&f z{*!hx13WQp%x!8G`sNVW+sTKMP`=DXLCR#cF*#8ME{Q@>c5#8bYHpL}Dz_jmfCP(X zA34CgzMay5rAenqsI)7ihqo4FWBh*u9e<}3qc(84n9S$_s8+SM;TYX~8lg4NGp696 z=6_~6?@y(S^Re|8mL|4ZXlgHSqj!~?wyFqe`ja)k)mMSFVa;-fsM_|q0eV1RdtrRo z9RDEaCWSBU%0@X>{@R<4lRC&&2^2d$QV0LOPK;K$v4<{YfchhijtqM$3i*Glbg!hB z9T~f>lsxfvNSnG%;p-`ZS$xKtE~%HHCQ_W|sVIlHzc7s8;U%WBNME|Z`%qnYqJ66c zaAA=ZMB1{@;+Yuk)mrAkaQUn-;;uHl(m*6mmSeH1bjn0|b;rw2$H?xI0K6^U*cKSJ zjf_}4tlkd*zY4R>|Ij?D4d{P28XjduK#RiJ@Dv+{=cdb=%TzDSAn;otE8KYfhggXA z8g!M+*xSt1rJeE1VXsVzonC?6{b`Lpq~3y?vq?a`O?wU1-`3#hdy~rw{tdrX?z^Ur z2nMe1n^U>9mt7IC^i0Ff-_V)hdV`R#*cX!|VX#gpvpoa(cVv}eK&O8k|FzbeX0xl& zi+8%oL`E;JB^qW&d9SJH(j_7C!$DyJjoq7bLZn&FQv}pHX5*WBxPeaWmlXD@#q=aK z+)e>!SD)~~yR4+pO+BbFvF>Gl+jqQ~^>nK;u$D>nEMW<)Cda$ySe;F@1$t4ff6}?s zypg4w+54YVd~F8n7&w2`?$u>fVjZBYLXX9Dy!wE(b8|5F*=4fY4Yy#_wCC=)8E3AH z=Id>u3gjKn?R&h{Jyuxb3znny;YHNa*N3%FkgB*D$p9|+H7t0d4-GF%5c@cIVw+C4 z{J`<+t!<4;g8KYj*QU)oCCA!Y$K=?U%<}{cFy4`dlI7E^jVFK5xf$KvK_gA#G%JnC zswyzmcI^4a^GIuE;JLU5eQ3laQoyEi8%{m~1but6uF;1&*&9HKprBaCzv#*jdOZ`m zKhoavz>~57Q&M@1Jl>s%bNAptafk%Jkk5)Hb)#pux<+VB>_ACmUgn}&rB6ah#uwguF*Z#-uHkojmqC1b z!XVC*0(k0to!ctka>uM!KXHZ8)()9DdEVA20ZMyMcl@ad1{^5xZ0K+b-DY?NPg6XMlm`wLRfwC>~3hF-O#i;ALc2Gn+j{X0MRkK1zKne(Rbc zsnrs%d{KY845NY7Q_aMn%eiw8x(0riP|t7d)a^-kH#A+|wv#*j#XS5t*^^Uh*1@G& z2ER7$6oECI0C-)xs{UVP2RTnV^r&pQTQFLLHU#ZYh%#Ybokc2i%g;Y}$GyPt6Ra*R zzu;e8!?X^|vDVqMYu(@OODm(AW23oAr?bryZ3KUgN+6h`1KF%W%GL3UHU1%?qaK4e zUkd7&a7Esrc+oo2Amblk;1?PgCiE-QZ7&O@&BKZ^SqFvGU#mkko-LpB(KmJ|5u+t^ z`cDy^kC(BAeI=-C?N4jd6W`A0-KUtTk(dKz72l ziKu@Xp4m@2W23h|`p5u^hEb5Ij?q=3-#36bT(mcgPL%oba(nWQ(I`m*=~3SKU1T$g z5TE-zQ?&66whl%O8(8E&ICJ?-UojUlitoO?mk&iEecaT^)m!&A4ij-Q?g5X zdx_?Q#M|1!y?)tAbdKl?f;Vd_2^*){2QkFUC#|ltyC+Q<(`?bfB1ONZT zB~qkFm8!b?8H>ekSLvW6o*$|D`0t1PF#8|-k=;j5bt|Rj`0?w1_Gkaz|NiCUk3T+s z`E&fRW~-)}Qa*ls@6SKm?^x=`kH_C{f5!N4W9+ZSBx8*G{`&Mb?N3wv`n0$H#u*b* z`?tN#@D}@deqw+9a_!%2zFcXeIvla}_jhX9Ms|C@&_*nKi$@>J$?e# z^ccp*_eos7QUBC8zxBmqfp?~nJ2q~N{`JovfBpE}e04iFd|JQ7P+dRbEa-r3_8tkx zWvBMU(Z<6&>fvaq@HE^7#dI7GdK=^ZG~v_sHkPl?5JIDRe`D{)w!W?%Sc12WVw*Sb zsVZ4VtAtZ(yO)1bi_jU*)%kGo39YA{ zB))-pVI|=wo&#`Mp?BP~QS@{0WEdga!+ur?l$(Y^M5-*GoReK z7U~iV{2GY=N|@!OfE;xWe`Zui-ZMJO}KIx?K*y zCpm36wrE@HGyS4Dh-X_rUVRyI2W`G-g>M)dxjJ_jj3mSqTX@=Nd%m?~I2x&~XJ7pK zFl}eXs%L~_K{5Ue%!48NcX$w+W3%?lJ+Wcf!pjr=CKAZT_-}30f3c6;1e|qgWbxIR zk;S*y1lq~S_7NA;3zhZ|Ec8tzzu^BYPmC9aTi+&b`$Ky>^CSG9ckt4a!E;_2su+C`Tqg&}Dc5t&7`Uro~PS8xC?=y^&yGS?9e&^op zNE;>pY6r-pysMwEn_?lhsPJlbg4t&0XM*K3@UJ}=3G|t`Wz<@kNDZ?aUWfv+Za5n# z&VatTwZBa~KwqUie=7=TPhJ`RQXGv>oDQ$l z`2ZW@v<{cZ#;BiCGoCF1f{8`+)3yM~XEv+DadP)_f>vxfmUAt&&}5wW`6MI_Osaa? zC1I@$l@}@ZsxFueLY6g?4&x$SJaE{&atZwCtW(`(k(_R(y3j8 z>7p;8|Bb?3ZiA=0#Ym$4|E)h~axM#Ea4TZyJFysK!Ge+1eVhnRpdSGSU5%DYxwF%ybs?8hy!8gToKR3__k%Dlh&)GE3fK zh_j~KP{P>$6+x0%M?l?w)td%lThaEoBFa=;>~~P zP(p80x($VfPlyqlwBoPq6zFmi%ZQP?v^9dkf0lxR9<#`Jvjbz^zo=#47=`7c7Clxa znVSPO&x+_=cDs{cF?8=P!MDIK6uiW1`c2T6sS;AH6R>2X7O=J2^VMx31s{giBD~R% z>iddT!r15liG%3K(0dy5uZ zf1B>46QfHf0X^tC>3QM*^o0{;O~BiBP)l+X24zKqqiZ{n-R5ht!%%=2*CIgczy_@d z$WijP1embZ0J*r|ouzacNU^y&lp>aB_khJ|G`M!)PPhDsU&jF!UUV?MwykE9T+c zI38lw+0DRl6njRua0P==8!P?;bQBtCD79fXKuUXbhpn)b0$Z@i;)?Dff2y=Yoe}KF zZUj@>7>Tob<)--@AZbqHfVt;X5y?mpATMPE*VB-@4iSS{%PWm0C5B)gU+oib6I;9$ z{Hh?DOC+ijwB-5Fs})DgZ^b@vil83Y1D?DY)LCJqMqp=8@@}L3-~>9rfg&u_+1l8B zOTT7{eJ95%_$agTlbuS8e^j9B+fBe=WvlodFW^OHs5*v#gLo}PfhvaL+Pl@nVSf3dt|k6qo*0kC0v zC69TI+*SU$vu#d80D6%Q%^f7RJ2ayD%`*#w!{AX+bnS1}smn_fs=eU%w2PP7Q!XA9 zc!@h84Q))FpB%Xqy9g1v^I2&}@B2F4jJdVV`qK}rj7c72Og_qT6L8rR52JE)t3a z;VCP$s7Y|lRVr&cwts-_+)1HBbKgyZ>5f8rW1rQ9A;9%qvUo5j*ydpLz1f^A zX#@;j0H{Ye8F58)ILvb0<7CF#u0xhux01rZ6rq%)7=bb=8!FG)kWNxQN)g;-WLThT zmy8uUcl;VDf1J78<}`=5i1I#n0p3jYzOIT6^+ z7xDSbo@q-q`=t0YD>p#m7KAnjo3TH00x!RJ? z5uUtsb0ekKy>^Q4=M)!CH;fSt%WJ$#vQ(MSou3NumzJWyU4f$m_5ihmIAPQWWC>mA4&|O9iUb+FMYnKXGaf;FdRT$rbDKlWbJ!j>dF?Ri^6ApW6=@d z_z*?dTwA!rtSr~Np1852z(!Q$j#|-ke^+zwVbKNXvplPm@c&mOBKF4F+*_ctPRDwQ z*c)`Jl@7Io4JcW2n$yG5)6M1N)5`Tp=>DKlOjH0;A8nbowlblbv*+uWZsO7M%eE0bM6%&`ttcmZ+FyLM+~8iB5{ zNgkB-uyE`?zIN=Fw|(9&a!z3a+_gDsEfVuK{TjW=$y;PpW#d&LZCiBoukt!h^+vjl zdc47>xEa$@gQo9zxl2??uv-YIe^kP#kS#rXvH>s_Cc;9W;ox~&Vcj)~y1=Zwz`|+daqU-JsggHm%vJEo1`X$eJM&s#`z3Om?(vK0k#R5{MxB*u{A89cN2D zTJc|fxm3$AwlijtLWvcTe|3Tb@SZgFKkh1zEHP^gyMiu*k*bx@?~K;_E-pdCiq}<^ zy0kHv^J?1D{Ie_>86OfM;s-1Gv+IAO;9jPTm93<3->Vx0;Q;k=3w3K?WZ-C{hhUNo z>ys*KwI3$0#Lm#e0N9vz`4tsa0nCuPf zE7i*@cTBL@M1EJJ&vJf&h6^rxrPk%QNoYKZ1rAyp6N_2IgT>ya&pizSTon z9x$o@8XD?2&{LB!}B%x=nti)KVma%Ev{3V7PB9J`Jrxh?lsjI(nvbrSUk z8WR}s8E|G`GoOR$?QKU6{Qrq1QKTr9s;XygEEY3Ur3Xb)e6!yE^QPZa{-@spyak`O z7*lX>-~XMT{VV_f>)W5dz5V)|d(&!5DybOXzCH8v_wu)k>FwL&A2+3(e^p9A5Mtm-pX4-h)kfCfq+DMgo02*KRC@_~j2qNndlCo!6**_3Ec*+^&sZX>r(0&A8YCjdi zKV|KG|Jd}hk)=}PGFL6 zZ!xN$ZtGo-Akko+52MwM^bd=9?oNb#_41k0KsoyB#8&(^>3P|OQ-avV) zKO)>{c}W4{wL*y_P>8$g&xm`K@wy!})BQE9IG3OJ#?W^H&=kO8e<4JE;*rH04h|2u zCRuni+&)GoH;xWgR+w28N4WXubq-%o)z0$3d37xwRjxysXPJTNO=ru4tKQfhG9w!g z^W`f~UAPG@FwpNw2cT>bKIt!_SLKpA9q#?;9{L7-1u$ZRvUtDhYwQHM(Ouv>HCD~r zc~edm$GPjv)@D~+f5p|6&EQoi;NFE}IWq8(>yC{>f@;FC9TMl3zgQSXAy^fyBhT3O zV_Njr57lFBIJD~hMq|I|J{*pR;9**GY(h3 za<1TBl61(`e4M)^6GO+$pwCeiO=OxrV% zZm?DN2pSeWRCR!ZNuP8xkLK;+TPtXxoa?9~<-DX^=RW*Dh|98ZRomUN>w*#AbS|MM zqvBfHqKx@CY}h6tkw;y`$v}qmBSFmB$X4xf7j6JAMy>{MnzWTixFZI8X<7y zh7V~ZkQAbuk)f>8z#*Qs-;(h*Wpszl_+Tw5z^3F(GLxwyuQ}R%MxHsbn(kJjBEtrm z%Bme(*EQ&Z@GjMkvq#uTAo!x}5q8R!M%7!5EBS@sfAn!=eG2h5S?p3K7t_#ryR^<} zN75S7e;lBZzMoy%J2Mj5$a(q+Et35QK_y=dLZj&;8`A})ZPG7Ox{OvgX{^$d zA}s6bcj-kj9BcIt5H_W=Yqe1SA5P=Xs-+bENg7mAufX6SNb{`GnP4Iz>0xb0B)g^1 zLCAdLjJgSP{4O{vLqeutH?0ZJQjZ~Ccnu|Ne}(3l*oUA7`iJiZ^p;{l$*km_-Q@yl zxqBr<{#Vf7~DCMh5fA21Hy_TMKn4Rz{Zh_ZWBe5|s zjbgS@@+rgq*?hJKS^7Hzt3q-p23(RXN{-E=L)Y;&T-kicx)5oNDjD%u7p}%^XP8F1 zf5Hdw?A%*rnAQSOl{0`4UfzO0g-P8Kv?k1!pw^Rf$E$`3!CFd2$&su9>WHQ8)i6u? z`ebwsrhtON7L1`qrY~BsMHc^?X($9Ohj)bQ%`qnN2|g8vFlI@y;<7HM>v7}s`LB{q{AS@vs=c8%t$ptgd1v z@+zP7+X!a0jDPen6hSd-UUt1-)Xn`gN4g$I6j?xwQLM3!I4fL>eK2Yc1;{Hcf0^6E zIvVN^WF?1QjN41rMbJQRv5;tR`^ok&2Rc4`{~i*ESV8K}g7vgHitN3X0%M!OiQ`Tg zWH<&n-s4k>+~U)vVQ^od-`GoyO6ywbWWd5egnjU$&zeW_p2Nthl!TNc`^2-C*ofgL z3Ocu0qaYzQe)(M){ih|_wBGV6f4EhFI81GdsDso~NlF)SF?w^FpJBNMsVPHGOnAV5_hJbH9v zzPRi_728qyBVm=`E5ZsLy%yj<5Ljp8Vl5o96Q!(kb{SACD~BIFct__Hf5yR%1Dz>3 zH{D$nQj=3GyNG`)j1aPdtv-`0IH$-J9Cs#HaD?RL_GA=)yyj~%r9mA@E3f5AAH|G& zA0x-`6tZ-KXpxmuRnKhhP49AUEt5lhSb~H;VmiGYfbt zT#5iomtvQWs80?vcKmjof19eWU%hD@PWCc8K!NBQ2Da4=i8FX(AZiBhRjx4pRZY`I z?$9)HWsk!%Wfn48>SJr+>OSxYPV!pw3NPE)*2Z~kF|MS98bgbn#P_aBCAwb$-}L&ZCTUVJPO1#Q^?;BKW z;;vzA{wb;qjRy=O1YO;Kd7S`G1tN_Yiyeeg05dBEUU5j4fe|4C9CK2R(MGHbFq$q0 zE3`do`U59K>CJ?wTIBb{lCTh5T6!QgFxfFHl`)u?Fj$^We>OZ#K1qST8||_a?7Ux> zIFxKm5}x`hl{uRplxC#JVUHG$X+B)Jm~Ji9+MWjx_4I8!PP{&>{>Ztuzu0!duf%O8wOE76*=^Tx)cOq%i{e zm6#5AvuW$i7}Oz(dSSU>)E<4hy1&z$`q`I`R8!3Lf`Wqt5Nz5RQG}AUg0H7N%8?^3 zUN7k2;8&?~+8`xsb_pIai58MMF3bj>8L3tN9ptnFe@t~C#@sFpd$Wu38?%$)*E#1w zDpt2ol9tY5?o<9;yXmxYmIzaKmqnW?2f-)NZV~ltunbuYcN@)CX=71d#!q#l!q!^x zbygqID`4I-AYI7D$Aa`ooUkt`SaNQ@Hr}~t2iCR5-p0$*PX~apXZuMQxGd~yj+?P= zQ(+NZf4i%$1fIEzJl{IHj->xc^+FC5XniY3Z(2VO{6*Q9x zOm<`6nPl-~*`k3By@a8|Z78?1gO4yu#b=cB#kuFwhel1>V9lo1q|F5-qg<%lF|KOW zmR-rT_at5ptz~B?3LAl3ukM^BA+?I-7hp(8f0MXhn`osFP-$64P!q*T)Rqjecvuw8 z5-C>hAhAMEddsR7E&%rC+8Y{$2u_1pWFvr=+J&qX&VN(=qZR(MH<#9qh|WORQuaH{ zDm*irLTGPOny;{V`^!I;p*=MxqGod7)KrnF7#5|Ysry@uQ{g?l01ibj%nPw(-0&Pg&mcs7DkQPLx=Oa z?MA)(xFzj$xb99k%E>|2{u%I2seWgv%oOgdj2u>ouCH^JMJA;fmQ=UO9i$)d4cujI z#opE^SHnDy>UUd#rL2{Z>Ge6#rrh~`e{anTi#v}~3yWOT{x2^Oc)x{UlD^KLWx8l# zeazAJ>NS8lBlUcF)0Uz!@u#;gKDlWiTxRZS6f-JW5(ra~uPkM&BPlIEv>@aOb;f7& z4WiJ!iuahpXa$ z?nzW4VMB6{WDlzlwe-x1Zx2eF(iq4#)$3dqM1;@^B8;t>Wlg^g+791&(U()7oW|p& zI-f>=#h+*BdwjVg!uMkDBkNvo*+CB7{JzTXP@Q(BD~U&^MH(Ss?reFXqT0X(BQ`sW zl=b9@>fBQcz@Li(IVRz(8I09ie{>my7gR%*YEsq@0VIBV+5(i(vkPKU6r;|*>XgC- zI7*^+nI5GtIE!V8(l5}n%1<2+y0dVND#O~@8&f?;#WH4H#%LCqA|k9$XnKyuMY#)@ z-#;%&l*GrTj9sEL$bz7{zHN}qPIsV+E$f8`$Za&?sE9fQku9FRElPPfe~%}shGv`# zsc?;>%nOAxkIg`W{%hxa(X?-64M0=ffjTwGY}QAUE~3!&58btCseAE!grO{4FvPucL;yfgdy;&S<`unSFg!>)cWss%NNVi#6Ab_$Y^NhWG;g(R{L0ef9t1mp2lAYehDFkyW3x{^b!tAy8nfDq&?A^F?WsaCYJRE z!JB39pBrA(QD~2JGhkaX$kt7YIiqHzMwQtEt!^-9(f78%x(1=$WBanBRo5cffJ)6s zXVXqC76kIq@$16AqeALQ(d~1RG@ldZ48(~-seT!@R8i9XRD`JWf8Nz4ro353qo_KA zvmsA|$hBcnk*to?}aui>WkdDPR!!^Fr8EY180=s58^j-e)4Zn|}OCf1;PJ8-3hBIli zI!_`;vj|~qTcS9%8e^0hgE{foPGu-vg)?RWS)=wgd_nHau&%Q~tt9g<{6`oY zU+O7>TI4;J(Jih@I6L!D=bf?>c;Ygk@kv!Cl$$mumE7-}e@RfgGeLx( zNO0n;D{4wtZe6EKE3G;oAsUeg;AV-w^Mi~9sp_B=FVSyn$t5TnS3Rk1mzATA@}QW2 zpFP~sR*9lyUCE~Yh68C-1AS7U+)sTwtaC1uB|1)asO<=&U0oDgC$|M{e*mt7(0a4hh!k7jpQH`33V&4rYn#>H*TSnd+UrOutWu|!G&|9d5P%;6*ldH@<&B#=afCpw8_Gkv6-bSl0MRN zfz~VglDE!1gv*4mgl#Ad8*9moh(r5F&tvq;x|QY?e?@LH&WFTgFRWA32L^h^i&xvT z>p`**VAG}~_?76wLK2G#3^;DMeeX;B6x~bziJJ4NkVI5(?n!TWXl7#8y&327huiXWWsU46z=cz4+t3 zqew5hf9nu`2k|D+Gc=b+1FC)DcC{x51t0z>{`UQ-4@CPdI34thjUWf9sJJNzsV9~x zO7b&xXAu6EI5pyjBMTglT8f9@2RC^93xx-xwBiyhzhHZ zWf2=;NRP8Vi!7qasy6tvrWrPJ_HwV?2d<>|e-f0q;E@t42jyRC%kF?kRZW5L)?Pek zd0Mmek#nTgDGhl3vcX#IN|W>w0`*fI)n=k2uhm%;wWF?$6f&_@4A?E35oRaK$tEVh z$v*_(HeHqdyDR^OyLWoc`s1V$OIx>jpJHzb>H$S$vPEQN$u8DxQ9Bn0c99j;WLcN| ze+(CB*z{FDcY&jF+WLK|DB3LatJtcRevN|y9DEhPoridP##YC)W#=oxD$9W7uGGA8 z=1{SEF4>ZuZH)8hLY=uW;#ddRuq6-Rk9ELf{PbFRuZ21fFjJHTWBVRSi3yo6`0cJ$ zDSTZQ=`5xqc<)AKBnt0_J`M4Oq6v+=e{+jwK%eG>2vz@ORTxSmH2Qbf#*AgNXw6vq*Sj&;b58m;SCf&2tuHF{aRk@8&L8_r9K3H` z8>?d-e|lgV+Ky+Nt-v4Ep`L)&rLwA(RHLoc0us)OPg%w8b_3MH3}><#Ueec)4=4s^ ze^Aiu?|=9&=PYw|mvIyV6SseZ0;_3%H8n9lJ_>Vma%Ev{3V7PBT04(xxee~~D_kbU zoYngQ7*hnu21plRy4@|V*XvX%^8cHNq$p93q?Dt-{Qlqa?my-4pO?RXTz>l5U$i>1$||MH$D_P|gJ+-e<>UU#RVinG%HRFn z#PT_0rEGZ)N-6!jQszeQfW~W}E6?S#xxc?z=V0}XzG14U-o3xlYUXAhKyAufV@Xa7 z{03mSI|Mu}OJTH5AmrXq{khj`_&cx|(?Qg)Qh540a zuvW{?a=LcOjUv%0(Rz>rRsycL|LPyo1|+|oQj^{GTGqxth!^{n4^r>tv~vE|X-Qp} z8ojx4S}l;P*05{2*R3ZsYIveO3NPNDnHfE!MTL!5@^-RWjRSn^eR(Q>`UKzYWG%>c zku2(gBWS8SPSY7o4n}JQ9y9$!a;>Q6O z#MWf-u3Iu=iGRX>?=hN(-?u`RYJ8QX>>T^unwZhXm_ed|J9Rxv={hP=ayg#+ z-sUX3%X*{>cr<~mYB!4ZH*r!6C|Yd*$78<<_)IdNxPGevC}W$!lYO4SJ01z0+^?>( zEj?)nY6ycNHS;>o_oN1{`BAuQDHC``H_GMh*4V9-F?CIUP*Ra^5d&F)Ksb%}!g!vn zDecvAP$IlpN=YPJQO*nQzmfaX&pM7>kkeYab?9uY?u5D$ zPK{OB8Mv(>Lf8x>Cotv@1acYi_u`VsXR^mBTKZ=HL% z@e_mA;#mxT$$Roxi0ktR#*pr6*VRI3dW)ULNJzb zAzV8wO5$1Eb>Femdu zj}xe)c{^-2(afg>F&kxH1$>Q>{v|`|Q8kTf&JV`b{KP5OwzsT%uc+-C(`zj9uI;09 zXz%-f9d%vR)#iwzpge!f(Bsty%K+%KhK-L!M> zyls#0(9DsF^f?)74LMk4VO!oLZUS@fG?6jbwbZzr8f6@$%%xfxsWh$nCL znDN^gJ|q2P_rqpngH!EO6&Njp0!(A$ruP+p-+F%}om5yWzF#Z8N@cwg)0=IjQNtvN zl+0t{4^A4XR3be00%4Dm8e&|st5}Z1pC4tvG56X~cCTcXZp{tQrHuy=eX9ooP(dV` ztVC$^xkZPL#UJ*#D5L_63X=F780o`Xr@a71=bP$1UK79F*sT6%1)Jo(TCn#L(bw*O zvfp2QqMN%()mimJ^M3wYuSY+9bYQ4{KQ+|qL{m1&6+1j`s9D1#P+ z)K0XlNIf;6`Vr)}o7oI2^3tFKnSdgH9cJ#4(u&K+CT1PT(!A^>fdzX+6|MU;f91nu zlY9EIuJM-1$#Vw+-}+j0@DrD<`qc2hXwHJ_N9-H_hHi78OZ96c=C_V%ma&udn@bW| zoO{S|!}qF=&*vFT#kh&Puf9J_J&&Q-yv5S`l{fkaTyz#o{XJ03+$4A6Z!sz zX9RKxL)#Sa5h=S0X>tJgA1V~25(0f;D}5;L#O0gAa$0@bD@tB|UPmh7^1(@~Un_krnqQ=Bvj5(sV zL8OK#vNTmNj~Sd%)RbCGK1$$!#a_zC>3+S~Z8?NHj=H;(*}${bpH|L&pIqhceb&$4 zu^Y(3Oq@i%ely-2N;|hSP5pM0?x>v4FBOM6ZQzkk=5a2@UswS9!Zqv$4M=6DW_R{f z;{Fm~_g{kNK7Vt3TID?Jx#)-yxRAY;ra!!Lw`$-{xlS~>8yMOIp88OK&ksOd3)~a$ zV#K()N(Ph_5IoAu0OEmWRDBMB{ABQ}>Oe^S>Q5y$y`46nbV=JI(0ZBj=`H8aX8eIC z7_NdbniM^HQVi<+;yY9)>oXs|3k}t|+JMduUrAq}H_~K`+Np#y!d>7OK-q3ii4ClQ!U@{!1PwHuQkFEEui}Z<9GZ0LiIeolVjKw z&fpbxz`EZ0JH^8)zB{m@k#yo3pDxM!&YRmS-K;z{`Mj*n7fINYi#_@QNS^&)nR&`S z8IvgLJfuFoGC1l6k=K&-i8D;>nZk)l9#u-9woiX9^CZG*yathf&gU5lU51W7*^u|% zt>J1{spHQytoVU|Grio=?VepKzFjcA-C^M2mUvCSy?3)_=N+y|=UhDJv3kW@(2$!4 z7^j!M@0pU!yS~m7A3V^TU^^ismye=>?Fea^yQit^!<_1GVuC#l3}?*LgOFE0?cov7 z&6C@@?qr5dYnB;*HlH@l9KJrirOQX;u=pZMeD*sUTY>3VwVk+I3gK`rB=_0fFpUBN z|HLYGo6*1_UQgzXUvyrUvW|(Q4ucg?cFPkRTKVN>kD4UbN{HI)9mB|_W<=2Y~pmGWx`n`W&{&x9)>*bH%EF3f{o$w?epSQj?t&fH7Ol(ynK87{ie15)Y^j90e>D^yHtMXw|{>7%f%S4qATy0 zC24RlZE&U^SN?Y8C3}CnT4!-Y;J@7CWD>7wC7b~6<+@Dj?l>{O3T;pELDM@J%hl0m zQ`_(KqWPdn@GyOkP+$6g)3=$F#whPYScuv0eu11OD+kfEEx#I0Zy4Xe>*dr}2-)G#JDrJNiD3+dGI2KfHttl8A<^yI)quveLZfJ% z;L&zCm^oATYP8BeWfp=gBnAZmmhsDQ)26<^LAnhgZyBW( z8RT-8+p;%b_G^)UL9wn;o&=S$E*j*lO4bKp(!A3Gm1B>s(6I&4!^X^8_<{j&?_)~l zCMijwcGNez<4w8B!q8N*TfFJfKp_zd%Y5dScu(F5n{q;axX-d3x>@tCE!()a-uA31 zgXqr`%N7Clw;`dp@WEbW(c&H;3N zCK7d)MEmt&Vd|od&H5I1lP7nRQF%QH`cMDR*8HIjo4rzTT-kfbT_ada@wY7}@YcTO zz;*{dsTjO}4t)>c_j`!Sgjl?2SIY0Rkw{6|P&Dj7CVzA;T_mi#4IgmCA4u4y!_l;n zld4y_seHr9d>0PlhSJd&NVMAXJ6pA=pg@KqUr^|4VlvX+ApHzEx{x}CxevaxyF>Fq zN8nNBaWH$$O$CB1SL2*(u%an7c`w}@#OIqa3<<1%A-@tc&S*3KgKu7vr|~uBA4g{;B;ed6$CqZjrMe7mG`Q|lUMSljXG@@v`&0=nS}|dj8vIy za5Y{ThxvwR=`V>mA`l-c3H4-(4tsWZ^w`b4VGh16&H$R3vqBlq4AKkG1>f5?%m3ix z?V3D)P9EOT4(zqtI+Y=q=1PD1{a&@R0^@BOSRSF8JN2tgsIGwo1e!T&cECAhobl$U zcCWAk9m@IG4~2Z{2W*tK!n-%4_C%G7Ng{Uz5@Kmn&ExxkXD+P{ycrU{itXo}d_TB^ z(uTIz0th3e4n5 zBc2K!o|t8p<+vjI8gg`GU$4MQugmdWLk*Klh6cuEctkI}wh%SU6ebBFNEF_Gr-@>v zh6*$7ziE885Ks1imZHvb9}o*wQCKTyTuQ683Q0t4X#9fQYbK*~ZhAJGgkc%Rfat=#tx4UC$eYPsjQ{Grr(pZ&U z{lK!J3m~74!bia2;#-hf5#nY@$R5(XWF)a|x4!7oDAcv~PmAePH%t|{> zk3&Id3Ve#u(uO~tabhHijxi$hzg@uyJUDDGFb{eTL{rqCPjsDZ)l<`kS+cV&`Ba_F z1OIX-;z&=}QP34SD{@*VWH^XaOW_#JcBtjGihH+99q#uCon#4?zBo ze|;T3*Rz2&%4G{-1V0T50GbLt2goev8XsBR$j{2l>hd(#BccW-brA8Y(ZHc}1@C!a zm-gtW&mtYw!^5-X7gXx=F_4jeetXtp`NX!2<|gs<6yKxBu&1>mX~y#J0_GC^?daG% zV^a+>Ttk5Meen3q!*`3$F4A>;??uW0WLktVzVrs{x|^eUhEL5nr9@bp6kVe}ho2|u z%aiO`SB7F%Vr+3_s>&i2fu3_V1RZC)@=TbMJVsEV9MJVFf`)ylaqT02HztbNJBZ4L zPG&Qtvq4s`0KKye_Ipjm>lYTpAB~BSrNqwcRMOG-ZKimbDGr;O>eF>>@|(`7UMe~Z6o5EbVQUL=Ln8| z<6%?W9`?0q9SNBx_&Pp+J@z!p?}aORl|1tERZ4G7r!he|tyUsSN~>7~Av$xytL-MR z&K0@|_#ZaGHmuvPP~_E`s6)qJT=3=n88>3cHNwG2m}U4Jtta60c%^>=NtLjOXOYWpNPm z@2%j~hW79&nk8@dtf;{m0++)Jo#zp2>UC4i>528su~Ky|Jn0MAWA8E_Fl&6Jzr_@2 zIb;oOF=u3St+n!bsx8{Se3Hr=scZ(I@09(0nb+rP4JWkjr&R`$I?%J1$x1tChj`jE zzgZ?zsa(K8+VYHlTks$lfUNh?g(D6&z8Ww(A(oDZ)pepan_3Dz7TlY5IH|hcl$z_s z@_)@{>-|BUE(Ro;-9Z=ctaOI-g_4NI*d6@Qc1?R@$LEbilcc{stHNz?JN!#tr_54` zRZ(zi_^aiQEM=VBKCGtq*jmN+SrA!NM2vA1%4yAQmxHx`g{nRm>o&DHjN7u$;bL4l zZZNr$58sYaY2yJzcG)eiF$gRoKxUcFRTHMOPN~fYNttBIs$)6{WhfTF#PXE6NMn>| zCGqpz2~=@Cb|6};ilC))5rk!Av%yLBV{!uDu$+p1%sm;BC8#E(C0rA%7Ba3C9a-BC zhY^h3;Lh)VgTU2e>`VqZt|xsmkEfmTRWEd=O`(%xP@0a5#Q6ugUp`wfjScoLIWJ2&0Z?hmqQ4pt5 zEG|*;{+9E1TxI2lLaeTAa?IcUE!{#6aX*q$yP9mBP| z-PkS1mK_6(SZt26#FBdo{?SS>Io;07#_j2ovT==Ka&Clc`?H+>=qtLx7}ip;^5=9% z&&X&w`+W#>e~p0b+=VeUZSmgvjtAv-QD0Y&$~mJT1{+q!ao+=2K^RSKX#RGRYx7YM zf#2S5zsqCOO;}XHBn+zaTtY-GGbu|=G~e+~@9p~emoQJ{O8iv3NE~>*`~PK0tF$fX zV65o&TAEd;PF{?YI)Y?KGN#THu~!NHzJW^%8+AknOK z8~Ye3pymkmnhPf8bQyh|_?;7XJLYOKJ>?qYszcrT`F>E%+p6LTsq=ic0>W=kn|Acd zvECS_e}y5&^>n_iS@RL%ncJL+jf>z^Z==&83~X1je>&KZ6y&4lb{nR)RCftR@}g|km`zEJ zd17m7ot51hW`XP_<%Zp<*I0|_LH3}(px3knJ3P5j0Q(vfPtpdoQ-9O-EmX; zf68vB5>1?Tout6a4$_GqF^j@v!y@h7V^UV^s{OSVEc;%P-sS!4-5&W&yR5O6o(unD zFNb1bx~)CteM?dxBI#nYazWbZ7nbg-f_L5>RPkLx34#&S2QSy9UicQmHS092&Kkve zV}v)OcUkad%m3&pkhZ=9|6>eO0)b>*e@>I-iq`CN>cwwGA=X{Nn^4fxgFUWyscyGY zm!HnQHVCm?Grp6GTg)eg5gQX%^{5-t$7sSB z*iOa=^%cGNN@_(J%Jl*YO;L)KXk?AmBtnrv;W3nM!r_cOTPw=({rHeQZS9+xyU5gl ze2?Ql=LQyenSc(+2bYWKTq_Xj^U?(9+` zo{2c}cK3P9aoVyQxRPJi!)=J*NPNY{?~as#ND&5^KSC#BT%Sc$qXkrVwa`e?dl=)ks^f%J)h-GpOc%ezECl0 zql+1y#Cg-mvI$6P|3c5lQ2x;8#j3N#MTWv-WD#Yp?xc|bM`yD-8=qws)aN;^N2$jQ zsu+djb_&cNL&V#H8AXdWe|fd=Gy{ku;&PO;`ZO!OYYo9iMC*>VfN^J|j1)v;17qAY zFecH!xGKe5iQqlW7mjhHS>9`Max4 zR+JWGPL6k}54)`8mt3K8X!O%R{QH`es%g|hZcxz_b7($ydXd~XjZ zLf%ibtb_zREqG-!bvsjr#$}Yd#aECr?V!G-55-LI0j>>RVt<5alK?rT~eY;Sh2yAryVOWw=$A zTAqsFL(uWpu6LD*Y*M-cxcPRl_-j&684;i}A% zeXAHc!0oJHf4mnV@05cS+>~b_f=&9_$uwl6WyAZ^wInovgUrx(QuuwP&6F#fQh;Nt z&DwvVTyiN+-FZGXYgBhc%~O1*%KLrXt=%&9TfutQ_HRTA8M8_ejtvMn5WBKah8!5d zclnKbM>qhoHv-HLeh=>e76t1cgih;smW*PAMY#MqzCNfmL4G z*7fng(#tfG#OMF40L|4aBhzP%4c{q<<-Y@Rw<5aChrl(Pr}cC5rNY{}g2ZMvOCoS| z6{dQ-OrO%#{7;bZ-Tt}yWG|N5GuYw}%7Xu&_3oU#9(IPz7U#p7S{Bq?!Od=EtfGD0uz_~SpzA5I5s{$3UhRFWnpa!c-p00$&O>Y4c+G}>IceC z^8gsN2#^;by8yilvS_#Ctg^`eZ*oP76e;Q2t`{^4ZQrYr;&3>~AOHQZALjUHKeGGC zsa{H{Iez^5&++Wv$KPK+{`}+Px8LK3HJ56tDdpqG<9PnrzQmAI#3{3v~8cBc7daXMB{I11AV;Qv7l?9z1~_&-zXoC7C(=5 zRAXW1Kk+Xf-IG6BCD&U)-~Ja9(Brq?tl)qs@Nt za^c8wzK%sIlH}2YN4f)1a0KXGmih24Z5+3Uw zMm^jJPSyr*LP0DaIzNP4r`06&$zM#V$up(gF>ErfuZAQV8RD4Gx+|@Zp7=g(J7SPd zLzE$=Xsv9Q4Bc#nnXwkikIv&h-3 zl#b{5Uy|!+G+Ht2#f`HY^D)IqWk@xmTh;N6K8+RZC@fVTJ|AO$&}p?6V_%92hIBVQ zMQIx{+hzya1X7BvrJF5*Ug>S#+a!%@guNne3+9|i*)(_cS8WAI zbZ{I^m1bpFjJvph1v81E(G;V$Alfiv6t*H8wi9NjLA?*}iTd8s#BwRlrJ+3^S`~*y z|1^_VI!V%9@MkhJ_K9`Y1PRB*@rOPW5M)_+H}Ga05zD(9#3Zep)3A-l@w45cm7-Rl zwZ*RWS<`B!-5MNM^Tp&~GSiH9$hw4&?~SBHd+F@W>ZP?p}IQoy&+_aSg%l?NlYc5n3~2Qv;Lx)vzhRn5AP|)ab6c+-Ql=Q zhqXf^DC*LG>i}LgsJa-_gQ38tO(TRMmdGZk_s~$2nO%ey`tKK!W?q7}gqVjX5T808 zP8fr`#31j;`MOk``h0n*vpHj+NgL3q=pMwbaQq(5YkyqQ&;~~kNND3#mj-))_T;Qm z3^0yE2oNHnUFO7Tv6Ayb41x^!O#w@I($#M?LN@h(C|)=N1Hz9U!nSV}o~`Z@4-u;x zVmr5iNpzG9izZ-49vp1;`5-m`&+fiw7Cf6-1>@7VY1C0=a5JR5St@J$#7n@<+=OHl-7O>mC{GsA|>aA z?w*(GDun^0t)9aUq)_3AI~&I))E=;mL3YHkW}-U)k&HhaJq5KQm1vmUPHd7LEu}rN z3eVD0Kk-Iih=4}B=mqv`BQb?S7*v{t?b+^sNDQRO_uK6Zj3~sgO638I+aDwY-GeQH z*GJnP5&Dn_rqPcNzd1s|3GIxnyMChSvyLN?Q#R%})eA5-A9Y=c(3kyEbC8trdL@Q> z$Qi?QC`g9q^~}_!b+xC?hTyhAm_7v@>a*g)fH(LpBpa zD!3&l)B-u9NCdX6gT(Z0LG<`a(qI99{J{-AQW6c_vJCPQX6Ctw1y(zmgofuxbLpB2 z-f9&G4T%i1k3K|_cas&oJ2#DNPafo-k$kMUvV-|TNH|Q8F1m7i>iiz?f$SZu7XrOrxhE=l?$Ni_{;Tnh1n#xYz^K#KnP~ zuv|z1Pm2XSB(L~LM+UIH@~cvq&<&2jOie-R#MTGqC1$vqJ+qd1GuNJ8DoTtpuX4PGba zsdoNe6dM76Qz2fqd!|{dIAt?jNF#~X-zy*r@6#^3P`SMfffDztnZ!tHYzm5CIATDQ zpo{|bZ4x70tK{YtGp2>7Ye3Xun+Bj%BUb?S<~Du zn3rofTKReshr|AAdxY4jDQqCl+88^o)FsZP@QAcqxQtiMyTUxU@G?iDXeLVQ;%erE z@0uN*K|clYJB;n_gyKzKuTo0vHfeA!sb`#qHXO2Vm7L`|0-ya!p(kaiJkJuX-T!=j ziKvuIN_IoC)vVq+Cb`XjAd5ga0w<(6ns-cbpayHJ7W?q1!lQ$9`m`Tq>Bk9fcen(T z87`)Nvg}GMKoRWuZHdLN=l1jX%N?gqAu5H%vYC=*b2eQ>k&o7P%DG}Q?X(xt^8 zB4#p-#4>FJr;=rM4jpBgjRdORhO?Xn`eN6(45yDRRN=8*&%H>0xq1xhj+Gcr13F;< z{bW#8(4wylXqkuz|^rob@>Y%6yhn?~@ z9JQCQw>jiMMrF=_k%!{1N{aiFw=&(>R69tb42Xl+BJjYUQJWaUy2Nr~o$ z^ulbN47cq7s)PFdVp|B90NX?#f_QcwjvZF zH%7#QfO902I=Yq43>}>l7I~f9LWZX5;D}^fX@E&t3v>j3%iF(eP0Qi$P$d#^Q-vb* zeV(bNamtqT6KX!sdAZw$mn7S*S@|^pC^Pf#)%-Nvm^Pqxn_>)Z6PAwUs^x@~Aih}z zuVWEHW;wnO^7+>^}H6XD1~J9!ceMUA;Wcw}4oyX>e~AHir;4FbvE<2-oNF@SXMr(-{`?%2|eQ z1Wbnp4zAZ#yJ(et)tAO>qE|S z*@@8Il+49e-*U8q*_=FA@X(NKT)K+e*V3ZFh)YRfUzVjAcsrEwxsSPb>MZ0WBXxa~ zLyEQ)BV*z@a*0$G=MT1Pdq~8?G0d|VDHwkwXbt|ITl~R;d}bnF#A?pIQ4UmxJ2jTH zP+<6f_96rw>!rA4$~FW&goZ<50>+aTJB>|-FvliCP*{lEpA)#O_UtuoCK$ON&s79LB=U$YNH8(P5H|(9v-6v!^XrzPTqo<_ zWDO|EkSy~f5TO8{^D0dimZ}f`|0mA;b@DKO4Ur&S1)g<&;};S`C4uV3;=__!KR0<~ z?xa^eB6ne9x*cO`$NGK*;RiA&u4_E0W6X0&;t)d~a}&4IvF{Px43oMXapNuF`)dgq zojf-lvSs6cO4B%Ceoyi=KedPa9A!NMeCE(l-?&UxvH4U0yRI5}pQwMwcf{Nw1Itl= z-zjUbeOR*Xl!NWBMwT1P0qBjcvi<+WfNNEC{38 zE(sv7p*bc^P=u0h8i#&-99L4l0c=zILLO&tLp1*N$CXXCC0Xj7vI@* zR#j{b39z%;1}2E7`(_I5->Lt$ASB6xhBFf^XMxx{h_CC~DW~C(vAIu9sAsi*A2<0y z{XGbBHp|Tj|9(846bOj+0gZz3bxGHI5usF(FA#uwRsz#so#ND%9F{d3*j^=Nvk}+Z z!d<6Wu>JGEkeZ{4^l-tRpr5R}PH)bN7R6>ohF>rjOM2dwj$r?w+B~l&>vnnxMq}Ss z1v!v#xle`dWsJ>80k1AQs>GFlapb<%YL}`|yNid>7Fco38QJ9)^4ZW*p^m59h@N+{ zewETf*x{Y{H?|~W&!K?3EU-1CU8`L*_L{`O_G@{o)yd7LHH`B9#YnV`^p{%_aZacd zuJz7}cZ4rL?wgq@bTX2BbO6K?m^fQJ6_tRKL1ngrXIC_ZQfAeyQj83L(%#sI<+~O8 zd){XXdZYv2Q|qE!=8be9-kVsWC`Bi*NESoneG1Dh?qBR7Xh+mZUemzK9)5e-;|=`% zJXwqDe5T!TY88@;OPM-{5~IJ$3RhYu$M512zl$RW)!AJfd$YO4__a{*q;gR(L>bvo z6y+iqgR3|;_kgh_*DqCn`Fju&4vx2M3TkEbQYDZdqn`SK^_lJl`}W5wF>qkIjo#O_ zxpX57^RU(KY8bxRm3kwM;mNI6aX59-%#N@H>Ut)5GL9%CG^`glg=**#%^^7NW$Jyd z=fk(p0Va*gI@{apkd?EmhI$RY$tQT4V~x zlHRG(8^y!<=eW4rY?-~z2*s&*L8q!H8OS`QEFzX_OI7hzf6CivM)WwCHAMVEiV} zy3Fb>?U$#2vRbA)yAW(I2h59Mv&8j!Lg&7h{#=h>i##L3Mfa`Wml75K#ZA|l=92nB zrr8kDxzeZYt>w-v7JdQ8_;6ac263%))mIWF4uep1qB`H?Cfq3RQ>Ys_X-G4zY`H(M zu0oSo)Yqc=oD={X8N;{jKe@Cwk!Ynnozhbfh+R~Fo+-(?TbAx!Ks_yLANq#aiEu}= z9AqsrQvVlgp4Au@UF#zwh}yi7y-wsy(r-zeU3vA9>fH2ir=4k0IlkH^lt|^f{@nmd zaeTv1prPK&<-~Xu2d`%6{#wM}x4MZ0Y@9DV$av=RgCQ{~03yyhMw#w_Z=9|gHI#6q z&Y43RZ~DV}9H``Ca^>uOt0VW75Xj&E^go7&7`2muY!kN{Yy&v~0lK$TZ39060!&!9 zm2Lya2?5NvU3dd51O%`YQEf7Z;0;;dU@as}IrC_&rqH*cC+zIppf z>*6k6oPounv1zM4O>-D#WJAYERvl{cWaU)k$$TOMC466VZ*imb_tk7A%;927A2=5C zOX(Jw2kI}VnPdJ)x8jmwzyfFqmvYNFS%06I7@cru9CK~emiBwnv2}=hovGZbI)omS zf1aV+(=%)$)1;8Chp5CrYOTpKmw20HC)~yyxvSrAIlPq3sRV|L7z1YHTBm3OR@Z<$^K)+QTSh;ht|JqtNQT=lp9S_6k z)O*44`|h-2*s}N5BUq&wp0~H`1=FXhe;uwY0>0D#b}idsJZ~7taTu~qv_VZm(vXZs z$N;rHGF#GCx0>2H9$93wZDAJq9ZU;*Kh*-)&8!;$(C@NE)S@I32%^zRRx<=e5xyqY z3>G0ANgNIL8HpPkI7wWdOw*BSdbHix&X?(WcFJNO%N({p~iQ$#5e*mBp zAjJj*8mjbw)2xSI*q|}BsYiwKHcJz=KPxu?_Ke)HYV61j4fh$jp&_5h4UhI!L}Z-h zFz1!r3TmAdkrz?C?%`8Owmf-c#al}bFu$z#{R+IJGv!!;JAi<-?6qw>R zOB@`FXnnJ8NgaD*_ju%wJU4Fjf5dZ>}3a2D1s&MFTJ&ljuRUbe~m zblRrErhTDbThH>CRKGY3WKG&r$E>FgPfsI*LfNI(V&c}*y8(oKY)ANzXVWNgW?O+z zGK>*-d3UjkOUCT%A@6ST`{k6jDXUUi>90`?-isZ!1>bCMuB`?mkoZO=H{h z2!j__Lq&H5E@f8~a>$uS8p6nqTA@(uyS6aBF{Hk76((BA zy%u>Y(|;9rOZp#yWw{$vXrRumDjz-RVK8c^V2Za?g$72X?kN!Tm^4V-P~7v@{gthH zk`{dm&{>h~Z?AI%3sv~tG0|v~rJi(LlA6jRc`5Qiw!h&6Rsw(vI3N6(Am2-jMq_K1 zyLLYk$4{!n$;RScf1Wh-kHTAM98o%VdlJbDxg143loO!Q!jhPw7-0C)?xPBwx5x!h zUrg>#AxX}QD+d&`Z{Rh}@JQZ3qZ;0E#LvO7M;$S&axL4MIhbB7j^-V^87FE}2I~o) z0DJLswiYQd$c>5Ap8Z51`h-m>4fD>7}); zzpLEYqSZv8Xyly)*yy{g;8bjbsfG@`p7dYXtKP2vkY1DtKxq|EApv<8TO@FJ5R+HF z$`%ReogYF1Zx^4vjlURu5Z_{=4-5i5;CfE!5q*&wn@}sgn_@;$Cno`BL*J;smaXnz z;N)NI1_;{Se{rNo1RhvybjV-lppR`+A-pMg%>8_I8e8R$zH{g6lsjUz6U$5|AVax% zsO%Q>IggQo+n2epK!i95>u|JuS*(4Xq`@|Y6DY44WRz`t;su{Ncb^vHqt4~@0)i=5 zN-@4*^_4(#*D;PbL$Eq#*xEkC3m=B~B9zsav?4^=wktFRNCAtA6gtWBw=mH+820h zDXW?%Qnm}$t&NM1)ic!o!x*f`X9Wnv&h{A(zCNm@4x4PDA#I?+B`0w#;(##hhfcj# zkB^ejgS)v`o$}?SnNc7TVA`;op7TTIwX!_(e>^~x6fUk~m}6)Ni?Jr<|C@6)ytgV` zokh|2+1L~qGUGr-i?9O{u6b^+A82&M;)wiuw5
s-b}hWk??-*Os@O(BbYmc`E* z<&nxO0GdQ_-8w5#G?J~ulNmkn>f~HSmtE5D?NwV)v=U*CO317vs1V?%CG^sl6clbp ze;BR$F_Ank+-d^A(S5ZY%1b9k(|_Dh2w5@o5uK=Rr~TW)aUg$+VWU zPO`IhDfpIlJrbH(r%wBxMzQvGi8YS@f575Aa~(KhN=qtiDed?k#akiYL-`#o7+ufi z`;pJFiIaUOtA)nK5Fsg9o$C|2o6*$YizldxF$th*35@^&G0F+hZaRK3Wp81>j*?nB zn66aBW-e!lPLHU;E&%api6u+n&Cdh87O?wzT-QHxUERkgid=kAAV`qB7NYBJf02s= z<;A<(>|qXn)vnSWCEQ_0N(^D!9=>nei_pcSY@FG?C)JTOFBzwk$vpTt`_VyJVF`X3 zErKd}wk;4m^OhZo?eEME zbV~4?D!90yivg4>VBY--#L|_Jf5dWV{o4r+*g14cwjI^Hb6jjZ7cxQ5r8)=|NVok& z6`+&+BF+-LKB4R%a`my?Ac?9`W;?TISSeGkwWmnWGC zQy;;1y^0ZNeU?PZHX10=I-1j#|FyV9dP+R?MjX>KlWH6n6hE82W&TVdFD_$x$O!Jb z%H;1pPfwIq?Hj54D8nMaqt>WUbu*bqmIrq67NY&tK2*rM?s{Y*q#oiEnM|L1s6y3hhT2DRUeQC4lJDtTlAh#xtSLjfay9Cx@UzJtf8=D3Xw9UX?cSc# z)e8#GJ6!>BF;M?heA(~uW#=TrT93Qs4U$T~hbd5+ke5TiCDAD7lGZM#cYb*B`c1gb zp0h5K6~UF1Agp}^*w6Dkt$(5QiLs}-DKp+<2s?$jqfD*OpxfHct*4@O7e)X&M2^vs zkA3gn{gE`Kx)EIln7#Ju<-$D8aEyShCS|M2bIIg=;VWW&}! z(>&{!1iHk4EXR>0EPGo1J;}8)6YqJ}7OG8J^(2=f1^1s(y^t4^twpVRQ*PbM2Xk2~ zB3)r1qcn>$6Cp7>27f?>xc@lY@J<}_Msl)IYC~5)qBh*mVo9|*PIwe)EcC$-dLzk0 zO_s!aZ$`@w6k>^7hZKe&fhSmn+#Kcji}1y)D&4~MH0%>^go4FrbABiXWLz&oHbn7y zB#QEzzHq{&m~dljM0l}SBaFI=OXT&RfB7GQk+_YMfov1EHi`po0e{`RK5v+(T(6p_ zaIfH7uaD49wZ2#A%WWs`qnsNrzRr>KNn5s;6A#F*(`M(VNzRDYRQTN6=(B8E+bp0l z@Q1u)S@lia)}F_By1!=rR%ux6!ba%9mkA-pGIk-b%%{TN?GvKR$+C1GM@MT#o&(z2 zBY)s7J4JKz-y{eNKYtIvGqh*EFv5K156v9MfhW)7q?z06s#-grWqgqPb*J$sf6F`n z<@-6@F;2P45E{I<5g6|cc}{lR6Z+9HP8qWw@3hl*5n{!B0^R@8`h{;sH)L(&15lv(z-LFN>?MJydynr z2&c1H1mX0=;)G~80bW(Bj7#d@S&PdJo^^xW*KA=gZy{$!v)smcJsLlYwj#xA&I= ziUEIXW^J@pZIiJXHMzgSnO6hXj#SV)9MY$cVASML{nXRek6MV%sG(POQ1W6*NO=q# zs@lgYG8OtCEh5A|gux-)P2L@UtaN~H8tI2ax9J1vIR4$%_(IPE&S>YSuisZf`XwA-H3ukf8v{r1WP2tO|?9`fcl1K5~&ye=XHhR|4(K7h%Ay8b~ ziVbCFO^R@!gb?bLEk%^jsaeb#-uzS-<7YA6FWI;VqEXm0Y^$GzYPa>D=)Ek+&yRn% zW}4pEW;{UZf%p5X2fFKj_6>)js?ThE%uB4Vl-0(gB-!|I(;IIu`pR{$*gV~Ba`@Y; zo5KRW4$ChZvp7%}^5JSLgU1PkH1}^$*}~9SQVVvwB;9|CN|4*#lxjuSSaRF#S!Xx* z5=>ldC}s8~=*WgTnpvyG0@`*uX~BPS6y&Me=u}?X=wg-_qUY$qK1Md${nT6Q-E?&p zlj2~>k)2vJ|6nOzx7{QKx$N#HD^$k|!-7XtECd7ki}N#f}e`LwvmKbv>V<&9a zA0+WX}P}o|_#zv1!~;=k{zGo&5`2ZR-S5Z5)5qUI*CP zYiN^VqXrt>bU4DuEx47XQm+?$E{on*hYtBjFe9b0*LTXn5n%1wp#nHp{M~%rJA=Q+ zkqDfLDL_Odn?lny0G{hW$^g@fGX9+!qa}4AAWb0iqcF|E@EMBZ(-hVJWom*Vl zR_VOtip=tXN+VuqXe6spE;Wn?epSHH7(KXc-iz~Cg^Zw;)Z5BYgKMT zZDQ`o=Bhk(mOZ0>W*xg)-)A+9<~Cd!%}rYqCsiqv%DJvollyE!IQsH%dq&}#lG>gM zpl;~&AU{YC>WX8ppCl92IQ5rj{M~Yw*k)Td|1dAvjwK_~{AN2kvLS!4ON5GY2ez{v zQSa~8jscYgtjQ8v2)BPZ9wdSP!+281Vy6H;xmu(LVH4m6Pz6sNVWsU1YD=dp4E6YI%&jl4qHO5A;>@(Q=iQd8pJL zl@rHo$`5fJ5n6wpw`zan*x!aDKzYTsp#(N2m9?OLug4s#oW+56hrMSyGYVRp2v29r z2ke?eHjA?m%YDv$p&!2r0j8v2R0Fl=9RtaBmCN_6=Cas!$ECZ0v={I zM1m7N$2+t8d)4Qk&K1!wBU4IfihHZ;a<2CYakk73=B0JfKnzm+&M*? zBLf(O-2sYn&DxPJitO1jgaNTO%4E6~H>0|UHzFAimxqR7_vE_c8v`YQ4{o$(r$GVk zL#DKw!??28s@#7Hs5!}6k>VO8XfsdW`BQ%#`AoeV_NX-CThBr%_Q~**PO@PRY?Urd z4rOi1p=Mnc%2Qk#N~n0fW-E1v*SPi@5Nv}_0U_8p^=1J3tgAty#g}L&ys*WwyOxzQ z!^W>f=`iW!U}Vut=BRje>bvIET~?&%J(EtIq6)~e)^&e!&XCWy4fSFS6r&fXGQmZ9 z1Lz74+Zh-{wKKLXJvU#QOx*?2aPYGtcGCUAgNCK?Tux97ER51AaFqJuy?2l6t8s=zn2f*(M z`Mv}4W{H2!Kz_~v`LyMLeBN_F?a8JCq9-9xgnW&LK>h&q@liDO$-+7PNp6)S);(LEu`do%$tD0Ar#6H1omKAYqH+hab z>~sfWjyzD)Ma#7~JvQfBM4xht{R--Mte0di3xWcC?9endEd{ZAHRk@1XIh)5+>1S9 zIiYaL@XlhxcPzi*tVR~D47PzBH^Tc<##vEQ-E5P!aeAGZGvFsr59it7ASPX151ZBX zm^*&}Z+U8K<3Q3!4~o~ks48%4$@Ud{R!dRru09;>PF}sHCTHn`%=bb52Ud2GePzBZFz9 zeJ6v_(ToVLUQcE48AWYb{;nhM(}w%#q?v17d|@<5f`^_l-2=J? z$?_jEvK;MN>%+1Q+1W8m3`8>TkXgTj4NYf0zKsp>*^VhC)kx7lf93=H5{h$SV(5RS zV#+{gpe+hO@lyoLyiP*p*pN;_34o;zfNq@&`eb-SPuOaDLT4ltcr59|IRA80uC;6j z3Yy1AF6kX~5uS};da}ouxg#+3pzh$JS#sSTAvo_z8n;W*xaTsVH|Jq7*2cS(^)~w2 zffzqV+~vrmrn8w?1;6-I4u~C*^>qyYS>laa4MHY+2!$Il{z+1Bth~ znnG7T^oiDWPUm|j$ICei>67L|i>tPoIU))&I@mN~_12kO3!R+w)^C;Jjo>jFcds%+ zPuZmUTW*-+I)Kp(m=PKwAnN(C+4dP2S}Avt`69yX>ykC@d>Blk0-bZ}b!dOJX(>5) zsTK|g*v=B|ojsD1C7=&<=7ZBEDJ?cYvU54Ur4e5)$5#=pVA5%nD`Qw zs`1e&3*eI_E7;|-(qFE|P0Xb?-72Vk=@okE9j!Qbe^rP+lHSux7gbi!I>C-~VgYEB zR>baV;QbNy5+2kC|I%^BtjvFi)Bd5*H=c?v2lhUoXg;Is1l+a^E8Sl8cs2O8e#dI_i4W`DPI#Pv0#g zzF1ce&Mqz&i27{UWh*W~E=uR27|zH+j#03e)XYDsNXy|{>cBQe{SkkNjdUZb2+$Qt z`i)r>o>NqDr7M02s(%RDB0^mUok8D&`SKzte zbjwX;;a4tn;1glRX6vB7)7=ps#?yP-;YTBF*M+eu)~zXTekD!p$uN6H++VSi(|v4B ztLHh1_eRkO3_z7xjZ+0w8BjF$6V^Kf!dwmPmR#o4GJ+1O^J{@jf@v9?L85)qw+heOsmqJa!6t2|^HG)0ZZDKU1p%yy zbv?b$?@V737!?z#b0f{ue1Mt5hV&VT7Ycuz$w}AhYZUOD5Avxs>LaI7@J@PI_*^Gv z{m$xpH`;YAai1J)^owEGr$}@u*Jgd}v<)bQvAcpXcffzY`0Gs_^PV8s=Q=L~>hT%x zrn%2ozqz=qliA)U_22pb(RYrrlAQA4vsa>Ta!blC+vx+Q%}+cD>nW>dLm1H)!c-3k z`uN&_y7MLojW<5OATsc>J^?aV_nSHhri#J(y91d`{9GTomZJ4U6;6jjf(psHh|B?u zzio0i_2pzm>!kss(_yb{jWIrK9bkTtS`WCIjZy?WDn&?W<#0MW>4JZja{Nf^0iHTY zcJz@&GD}U@%l}56`#qz*K7`7CAt?l)iW+a$;%D_K`cr*@#d{x%;?z3BT)oF%|L||V z9UUN-VMzfKm*AxXCX>N>6qYRn34b;?GCn>Eb98cLVQmU{+T~qsa~n63{=UCrK1O$C z&H#+3>D%j-?5><02D~ z**iRpp3b9(3Z4nv0Z%S6t5H|J33xdA$}>|cLg@+u>IQ^KgsMDdtyoC!co|{@0ljOe z$j3zZDzMx-EtU`GR^BTX{~E=+#lcDp>&3E}7!?zRd#nxX1MaalY;bJr+#11) zVJV#gZ-)1?W1e{>l+%h0pDCLt^l&a<5AbS18F&sGS$PY+8-LiV^3n6ghOUBwTqdwh zu)P`HjY0EstnpxZ+eJ2IYrz4_2HyYu|2g&q3%Z4>!%EJ}#a{+yA!=pqv9t#76 zu|qGGw`$}K#G_+NIKGFzkzk;fcWX4*8W!G3MawRS*DZR@6h9YW5tgm7X?Qkh9Ta4> z(5r4(8h;;=I{rejCRo{+tsg>`~ruZuB3VfVm5n1VhX zYzSKcF3Y+^S-zLND7#nBePOb=r)WN7hVb!^XDVxU@xQt zdrHMz1;hGdUIP4xQ_OQrdghTM_9^=$P*^>eG!$WBj@=VAi5JHdtk}VdL3w@-J+PB` z6(UoR!GAUU9PdGXnSo+ItoS*sA?_)r(BhasV*lcFx)+%P$V+GjMNC=NkYOrt59Won z>om{7yUC^=hAk}h;2rUgf7Czzc{Qr{FE1yv`t;`YjIW-IFMq8MCet^g=}UegH?Qi4 z^`rU_7clEr^**|t*3W-<#6RC$49DlQ$=~0NCx27?{Lj1b`RIB!m`?V-{kHyLbT)g5 z1s{aKEZYHlvPKVJ^DOhgs_^RDZ~wLnHr~%>SJ!{9>+|93!8#A5+<-wv%Y#<~iYqJ} zPcCQt3$6sS{ZJoctulc(RVLU&XA1r&J%tILO($okquI;)+0k+RH#*F} z4uA5mM}3pb$4>6%uMD}nkvf=z?f8|^LBifz5=vE{Ng+$qFP30F`|cE0u+cH)ZRbfg z&66FCuFt09tJ!3lF25gMjOxRGe)sdq{(m2h{yltgb2_}dzIQM=e^Vcy58qt_DWnGv z4kjO7-b*0%UP6bF@daLsF#hYa(dDd)I)BzYI@9+wx%zN;^>8$P_kLD2_0Mza(QJ4= zKHI;1cRs?^>1=fI0>4j&AATCYnZ2*PH<{!8%4Cy2z^gcm{I{2A1)9^ZmPD!#hS#G+ z^a(PaUe6BS52t{i^^;*syA_2nc;^Y*P0tfhy-`D(6~oOZ!`XEF;pL!d*d&{(C4XMk z%mlCDHAf3P@BPcO$(!-zyH>XLx;`4eeLI?tF3(2SFCoExeSlT1lfJGw#Nf_9PM%+m z@%#u-LbUa$9Cl(;$5F) zdWRQ$A})Ya5^;&w``i}zHPbX-=YNN6=lz(19@A%^8#F#QQLh&!iU@AiMy=|z(G7!d zUd3(A)avcc6jD6o5EyHSP^P}9FAI~ke#K;M*wSQ=zB_&T^rVon+erbM+~ep9Slw_O zOP|YBhj@!8S+HWG>G=9rtbV$FjTd?%&fs$r@dgej5hKL4nMjklynjc(-G2AM4ZE+$`?7vkZ z&pt(JoWYXLlyJn*MH~3YbZBQe|?zoY4dBlXsUaCp78gYAN*n13wjK>1k$=;waKc^`(~eLu-d z6tw)b}1(Y zc~2zE<-N%!pZ7p=2gyIA9b{d0CwW&pCH^G(dRV`%&+0cdderag_izoW3H3R;A;D_hmT6D>I z&CLFbH{}d!y?#T4Kz70$iMU=}dDTiX2eSzVQbt)YYWXOk0a)P09QX@abcu-Ok zj0Yt-agxqgGs%Akthw(ApNOizj1v10;LlHA|9fAlKzi$`yiP;TN z!`fzwy_>0T%;n}+){BhC92}KInjgD!|#7z{CH>5_|Ce_RGnjxCPA~d z$F^;I$F^h;x2aR%K-OpRVqX?#jyRel%gA%=nPM zcvtrW{fRE|Y`87Zz|iODJ}ZC^zM-J5&DJ>#?R!5kUUHK!L942rJoP&GE#T?&KG~Fi zqwH8j?sAKxnzAceB_4!I?-hWI8H&~)nN}hwKVdlzTLL6@G9`$h*zAl_V0lC?I7gEL z9ECBlMj2`T1%-1H_y&EC68ciH~BRG zGDovXvKQ~6`OK4<(YQ7Q{bnt(mi1M;cyhvR0WVM2@cYe$vqMBDxiYyoRc&S_%!5hX zF|gw7HWbW)Y{!FeS`>TIFhg(LH( zmw@$;jjU4LjwBIz`Iu2Jt@yOTK$F~wsL2dT8FAQOiD@Yq#w7G_bVH)r^I8#+&4-Qh zwx#5mZu^4f;eda2?MaS^MC5w}LfV-{QfgP3>be#TtSdQ_A-Wk<;q#Qbhh!)JY|^QK zfn+O}+U;xlftjoaWzxNN(GsYlbP+K=z*PcnF zbU~Ff93ce!**9dMN%A~JNl9^Yth~V6N*cn~C`en8gmSz>=UcM3Tu?=_ zie$1BJ`nO@RP=GFuU}L<$>|b*B|aCjn@^R5DEXgk!u3qO7n=P@Zy*5qNlN)i`E@rd zqeo$SRn6r!GFOCMoSy#9;fgeuBdr!3WJ~C?f@U_@LC!%&Qbm7R6t%va#)rWLkiqAY zLJEsh2z80HlLQ~>skHy~lB~qqNt%y(VknV43IJn0Q#UDLr7jIJq_5PUZ7o5@H}+nc zp6Q#UMR8Ek2`Mv>ZoqyCkasD&cj7$xjfTn}vVPU*81fOktfrvMAi2z`yDB%QVp#RaEn{`s5-v>@MWipW_RaziIn#D^D|f zG`k~*&$l4ZA?M~sz^>QhwlUCEes&yPWB!nK8GC3u7&l`66V6x~?^5iT@Nb2I$5T)^ zimU@%4gOsvv7zFw#h_8yY*l}uaSlMe<*Js^kk0sCT`P{tf62VKVw&!*O6fH~Qq z?bx9Z>~7ozjM)D+mH*$P#p3GxYa9Rk|Flk3R(o3=R(rDcS9H~nw#boIK#ryD$I8C} z`oRc2%}A^b?MSRn4Hk!ejqzrCX;-V0o&H8Ud!3zb<_f!g;tG51lnnrP=eQ-7@nL<0 z@qT@z&CBj!qpLO1^8is%zP>AwlhnxTU=Xwa-hg{>>!Ezk189=*QKE-qqq}F+$FW=M zA!-ru-O7J%50qT!N$L2dEX|9gyHIZOUZUe!H*X{Js^@w2*EaeJ!0r9{cq(r;7HbL~ za1~nu=G?d`sciU8@C_h-B~MXM78Gm6JgbxWqb3xVcqoQBZ}9JB8AeN;bd~tM;X*wb z;GEV2Xm|#WslSJe=AK#@P!{v3URe2)R^Nj+#k`y@nbhEqE&M%963$=2cZ(h8(}`_Q zS|8EF*r9KD$mRTfc(c@e zU-3GxxFXv5BRzlep7b{t=W+bin77maZF=tTl7PM6we@A1)N?Yhs9R(Fg05xFcJh8K zZ#vx1dF^t|7m0FZQ+~K*yXo!Ki9dLH)SkBbP@Od96r9j5csh(|1!JYK-Mu|-xMinF z)fwMWuMz)+4jo_{Jc&2y=2V)UM%y`np=}#yeuy)FqA6bMwHagwe^zOzBSJ^BIs(Q% zikj=oxfi@hLx{AKvM=bXQXLfbTBZpOe<#<^vC0g-0oEYf&wkv*h4mKkMK(eZzoXmR z{6@y+2mb48reEnD#5mrwWJBpQ%G0u;Rg({-w-qV9KpdbGA){DKKSW+Jnzk>@#|3hJ zND4(5rv4UyW7psd#- z(B{o;7dZ{| z+i^QvvJ&+VX01d%bm|C2C$3%}ynyzZxBKU?Z5J+q&Fuod^j|Z^FzJCaFqHHEN*@M3 zY|ld39Gtn7Hn$yp7|NqdC-Ww$O*(9!b!TuV+1kSQv%F%5*)-K~;Vt_$*#~3d7TUDm z+9t8D9&?wUl$gjfV4O;zTWD{1r%gw8vHK;R|Y$xES1Rbk{0YmK(c z)C;aQALXb>Su+NanuR`Bpblu7TTXkG6Te^{lFf&e?>%C>mhGS(o@U=yPhD+ElQ^;V zwmRZGJ^_3U>?n|uoOb@{;t~L~ELRlwcNW;C);7?C}h-^C@P()A#fI)Cz##72J?SAwB6h z0Ln|ytl!?2V}W|;i2Xd$rR7M#PYiO%cv1g-d8LQ9TROiWS@;)goN%HAhox+HeIhZR z6$iX5%B@ibkB2S7Z$Ck1=io{aLh+wS<-lWTb#vw169P3;b_H2^YqK^t)bA zv^JkEP+|lC=l@}=>ZQooTreSpzI>x=KPyMkOT=RY5x1ByZSq@Y4eN0?CgE7GQO%J2 zd{1zTz_5|sT8(g9u;P@Psj$`NYrS7fG&QmCJl8%v^XhjsTpB8L%XrTG*`5-;+kv}E zDLIni`S)XNcGi34|D(SB@v+2*|zu&LFwh?o5Cl21Dtl56SAv*2O9 z=4^c8Hlq82^2xUVn0aK|^R5a^u4ms=-)u#0T}|rguIZJXZ6GX9)5%1U*pKeb#N_fR z)HL+CQD#>+-J;p*g_lcKU-%D~4%mbjzgd4J&E$)Cr$OtB(vVUs=8>ias{!V2q7^>haqR=o%m&VW%)$uwdksi?!g=GU-jYUgpg-Ya`#V(OWy@?Y0Wxz?ED3K~r z3$+7rLIh)=W)@kXX28y28-Iz5@=L}0qU@y%VT7nC6G7(}LG+iC1T`g~K+muyiuzTk z4_L%Sk>wnf05Rg^rg4%-ZUbm4Y)C3(9@Y0!SdppwGJ}Fd9R`I)WJy$M=X64?*^n!X z+p-oTB@w%qRO!zb6jv$~98SYF?IUab)u}8#noxL#4q|6n2rl+oK=_9D3aPb5EU|w! z9i!=MW@_6n66&#kbB1;r=#1lt5)|U2UiGG>cv6u2byel zW{QV}=b)_fNq!~Qu~#;kuoBg*23F*yjDsr@R;|Uo1Vt>@BkeVRr*YY7L1fwAlmq43 z`6uoB(qIQMT%2P*+aI8^WWd70m*2{a0-ox~&!>$tac}SEyUFwC%5^PbcmR0-f8gB5 zg21?uE(^9h#v{LsNaaFWcIGV6A>Z!L*lNnwMpQB6nGWSPa?0Mwh&a3YyWj=OiC=@>f|QyZ9dT9O z8SkmRXcm6j#{?@ZbdT>?4X>?i`BC?&)oPUL`x{6!VAL`lcmWhGeRAgC6tOb@AGC~K zvQB)^zo&KQ4c*VRY0Z(kISUH~d@JLUetR2PJY5u~@^!b@;575Ko-D)H)q5tV;i*Zk z=a(DfC*ejK&-ri%dupkUPaAn8s-E>hXkx{AqlI&?nTl&(z8}Vx#$3W}{M(&huiw|} zW~-dhz^)x&O~0yb&>x>hX%4&eLqdYYc>Az(w5HtLWTqx#mXG-Js=)W+N<2ylLKPV{`l1)()ls`-)t9i|C!7m$dxkxE0OKI>R>7Yz%Sw$9l#c`T3hBaU-Hj276 z{45-9E7Vu3>riNHR%XlZ(FFBDg~~g((fHXlPo&X{hc3ROt@S<5PM%|i3C}Vcky)G7 zmaX=GfN3Y|BZB`s2Zz^I`DHcEKw{xh9DY642 zJAwkQaZsZn$-L0psm#;cWwBeeX8THIb)TiNZ)G#M7B`z=u3-l%%^{bLwf9XuO#3ju zZ%tigdvQ=hy{m+0zN{N8JhJahd(6D$|Gectga*eD#5w{vb0Trsd&s4u^8?xDpG}1}Yv>mt#C3Pi5yaA`LPO znAM&A*TvVP;&4g4mM$K+No%UThC2v5Qp@_4sueW)Z;j?WZ5evV5LP~0C zgX{HX_2m#3iPQcj*fU$Ei;9GP3fD@n{d0CUN9gQ!z5DQQxefK{@g!4d6j5U#U0-!q zXY1pfe4f|QUXI8itGP<3kI*K6XpubOJdr#JbK<7NjxpFVUsL3Jltwf!Y%h2(e627# ziGMh4bt2kCj>M%g98>&8g0%o!(N-dzL_0~&M*OwFOVMrdXA&hzR+L~eA}LAcq^J^N z5;9O^QY}e@r1-G};Tg0UL|t~g%M}#UNCBqeMCKWm91jUQT(?A?XxhWpnC)N&F z!uROT9^>%~U!nKF%i#d!jjTL8KH++7cZAnnGqp;2PR={=oa}NkCfGrkNU#8ie*1>Z z&G=EyGvB~dP33-PiXe{A8qPn8Y<#%)qjxe|y0w}!+xRBdH}Clqm0c2pYVFDia14MM zjVJQmTqU?I)w+M0Vb`4|a9;2?e55#|IY;_BZ6x9@8Je`Gyg7ANwiX&WBS=z_vLOk# zkE|a4N2U9wGR$vDbMnR8OizMMu)4y2XGFe@!MA6W^yj|vR|IlJ^s}cmYvXL$Mq=z8 zNSw*)f`N5W{h%jHDQu>!hAiJE;R=BC@@XW0WW8_%i>^5yOLS(BrQy=j7o-isOG0OE zE*HM1g`^Q0ptZ~QJ{XQN^(Db zb+G?Q;Ff+I;s@zU8(7rq$_iX{Jt|ZCT`?nnKmV$8;2zQxc%JHvwmGQ111VsvJV2h7 z6x+8QQcnGMYSmqzz*Rv|$ch&eTwI{ZbwcGW1z^r*mWekgbBYI2?$o~%y?M_O&ta^A zJ9TrycL;EORGy&(_JGBRo>7gFJZMN<*XNGLaIn?b)gBLLz7JX(1P#gQIOW51;K$I7 z>HYv%_4Rb|b&g}oYzNs;3 zsVYz#?>D3)RDUDIg_-QR+-e1RrDzReVv5zQM@Fky*Ag^)u!;~^)qsxNpBLmE*FH0J z?cAj?aR+1P_vFD5Og+Dw`*7WMWb5uI?Ns@SMD{BO0e6cC0ijc+p#pHziqnf*)h`D! zvBbwrS;VIH3G>3hVgm3ooW}}*a9UufsFS?hC|R7oj6X}kDUDV;yk7^{IHNv~=Q%+0 zcXw_8{=^*$q$NzDpc%OoB4*IYw_um=%{zmE#kGB>xVqeD2s_oTG%eoU!ctTE{07<(ciDn_VBOcYhR7VhJ!VVvd4W-W}2!WN4Lng` zaxS%uEVaaq#cR8##=v);COSy_pS;1^6KA7kqF%XWq8N0_QUI(CI@=QZ3<`_V?RxE# zk8yC6X0^?)iIPV3eu+I&M_y!Kfib9T)SBKb)T1JSknKagMrjt+cMvmfXjf?%T4$i>1=DI8;_Xmu4|rO7kac0f zql(B>6sc!rU+NpW#bP`5szAB?NZyFW@|vl?=*pQT`o)aqH+x{@W?@M*q)MegFiN9%kfs9R# z2YO*ywFV&@+;2CXop@i%;*HrW_dVl&5ueTIy3NL*&|4RS8RcaxRr;LGrwXJ%(LwCr ztme>?#Lk|MPCFhHklQNRiLwwfS`rA}djUJKubq&a<}&v;BvaB#|B z?cHeYGJxp76xG{pAa)ICrnHpglS1t7nNCZ_v9T5D7G>q~c^iqY5%cs7-u*X};I)`H zQq79;d)A=|-b&A3-!wNMbloR!4PAu1l@8|0C^XTFGIJmLFM|T>2R-64U5?L$*1ULk zNW2_Fr5^9XuYSK;ez6Qsp^np#gky)+K0R16zycPW@9}P8_Zq}-(SM!i{53T*5+$ik z$e1`Xjqa$f#d$d7DlvAlS2i_}Kgre4sNb1L!VmVJsd5dVX$?C{uDjRXBWYeZ=ItSD zTRJUaV3s4DoEmPdLPLLgTEZ8-HE8NqJTA@ekkNEM-QQ?FHYQ=qL=hLuDsnU2I;{tX|FPYn{&7owxLU&W#j!w*W{ z_ilz2>v=jHGO8if)zKVwe<5H}SR*RE^jQ?$V z;!P6?2>bo=zwgwJ2;YEEhn#9|pZsvxz*<2WQz{m7`(c;IBeSPC2pxPazb?!;{!l-H z>$PqQYHIq_SEOeFGDa4bFoz=VT!20TQ%857-obCztO)-0&C7;$SW1|*APd_}y6FzTPO#iFnbQ+;+ae$#4h#@|- z5hX5s^cS=Kf#J9)aI>i`D>^EwSyaE87^&^z_LR8DC^PuDEcUgSO3LsAo2UG*v1A=L zbphO`=Vv9p_SDpv?|m%Cs~1~0zt6+nPU8If4ZpuFXA?0eZTr%hg$?~n$LC~>eJe^5 zO>@U1+p%>D8M#wai?1wGW&ozNR3y9SFr9M8a+Ty%Y6T6cvF2guf2o5m^}OierD43z zEr@(tVqvgYSCrnezGMotGMFY>=PVta?wab+yqRa+&$KizuD0HTckwrj+ph)^Wt7Up zF}|{HPp7%9BE^X+R}ek)Go!bHzFA0Z?0ztK%7RHUhZ>kslcaN>R6xnLC762Gw3xewWxj-`fEzbqOhd5)O9^Y4w|7 z7T1=gmSu|IOx6{kVA#&)-%i1~74%h=R*w+5{L6ffU>+M$K-e|&WYi79ohAJXs!K}= zD+C>+G~Dc&o;EeASt&m|kJaNp>f+6s*_Yz`RGZQ6)(nTX!vIx5nQxfOWn=;pySQo<#WC&_p7aJab0!scGA1)PL*vb3dE`; zoub>J&1zO^>PY%cuzKWsCeNv%RGJqf#%-YaBU0>^7_PYUMQ`}faXav((p2@f)YCxN z8cMXg$93OxyX4^Nn_aw(pEwfZ%$bhK| zO|4NNRolhY;jdhJ4oKSNN>}M1Y__hvHCc@!bn9+nNDGQ@ES?MK0L^Kyg84P=j)3_^TPU_v9GY>%eo5a`LEqC0L>=7p zu5K-5$xnfiLn77Qn{I)9>PX7WpnA$$W`{%nlfv(7( zJwWg@oVx}U*!$mkTHSS!?oR=`j;>pMmAA*Y*p)q#1jNDBXKytQ0h$l1J#!LXKZ`?q zUWWdU{zQl|J3gVjE8fNVSVH^t&QLg36+QUduJi++{dipi?yU(zZG0>qJU~Rlpun7f zMImsgiCrNEG_ECif&2pH%cO$1E+H4?~?t;!teRt%r@1gCQX%tjOg=yq+nAArZL z{3!}7%;fZkwjf=E9JdhI3oLZP0F>0=)$fk`5p+0tF1h$X8Z5DtRBSkkso!jsM^hQ4 zva7=Ebr~&{vh6?{#yg5F(V8kdZ7pzZDJhx+U1zbRy51<=(SSd%zT3$Aeol+j*tN^n zOF))2F!w_!8%1hZEM}9pSmMK?$N*++nahmdA&s1S8MB6~av*U*Hr-60kaSEow}>#D zn+AByuVz2-zyV|AmPV;R&gS4hc}>;gP5r`sm99J8(nYMw_d{ffDV7ZDD|gQd>tEY5 zVIvgGFKVD;eqscrE43?-=#3T7x575zl{Iz(OE^&A9e4z9m=Hh~jeK<0@c^IQUpHjj zh|cT1FqMzw{)27jbfKZ%a>!z~lM1M1`&OzZHo=uMi-K9!(~vgx$x@K+jmE6DKc(4( z#`!(|nVFJIsl8$M2_^XlU89RjUz41@P_&8>w+ML2wFyeYjdmrz@#+-)!rtlm=y`%_1zIhUs&^|UyyNluS?}ebe|TD zdet%hDFEACmZRw*y*1_icHWuXdj}=0=$18eGKK-?Mbp#pPew-<)C9F;DQD>lXv+#3 z2Q3a{y0Pc?V>YMVjr3i8_FMe=jGgN9trh8rZm}{gITV-QdM0qv41igOt6~11!%5t> zY<0$+at*5{=x(^~ok)V<21M5)oNj%6Sy6kDop>FqvEjAq{dxft5?&s8vjmKZVulrh zG&z2C!(y6;pd4lkcM|rSC5sbAoz@@unvS_cGB}JyEx1K7Zh;*Mrn?IH(sC+UyJhWa zh~hzUP8nQn{Mv%6TmUA_>SHcg2U2kI=Nsuumk0bzhil02eNM9DypOIojc*{}Vd*y( z4{;;e`g{GKESVJ?9&{&p;R-c#Dy9zd-rB~(W(T4&A9@z{S{R(V6u#C>XsmbIM@tdwRcQ3ac3|PE<6n`Zr)c?tuwLjJ~=HAe?(xzDd zF37;*e|P1+JYctx>J= z5zc;PsqEBX3=Cj6I+_F(ac1#)Y(G0~KwVV&SuI-n0aZHp_Q65>gz*q$v zIzT;714SU@5L03JGj?jGPnmPq*9Xw69y1EYAEgt-jI~{>^Q*g4 zXX2}-Taq?gt!-U3`Ehdk$q!hW63X63S-P+cVH(#-u`YG z-L!sA02G%0z!nFbgy##U3MbU`Wk8EmISe9UxGVH2#SX1JR^(S;n%{2~@x9o-Niq1H zbHwBG2LQS&=8T!LnaUe^Ipk7n7m*#9zOr}=Xc^$prREo&_Y;R9H)dM>)BusqOBs`! ztBGh1i#cTGVekIs=PhjC<2#kgAI#|3jhxOqzx^gj>qE=u74RG!tSU)cHJ&QB1hD%d z041|`^v2y-H$G_>#eT4Ff+s1cnb;|u^usA}Re-gMNxQx>O4>)ae^*d3uSyMhDsXQz z*A;%W00d~O8f*_d9b?f~YZKYu#QIaK=$Nr5<|zbKF3)T0_kZ}RCG9zOi$0eN7y+998i-2wXm8Ptk?<4^2Y$T?TF%VYW)g~?dK#fxpc2WX|(1-&sO++P)l`1O+B({U!Rx*`?YC zCM*Y>lA))=ZDS-@|C}RTvk~B5sLi}_nQ20Y)8Z>zbwDSc4$M7h$6BuWza1s^>$p#= zf*rYi%OW{JcO>@m{()El8OVlw`wSQKetl(Y`RYCtk>RIG7E$5CWdLHxruU-! z-!igTA0*UDeUoz+l%>O;fnS_mpYX#@~q(vSKQr zqlj6yB4%^5afHv0GV8sU;R^lr;`?}?NTF-MGQ#?z{%aW)hK%{+N2z&?z53g@uFdC~ zYrvb!h-6QuW2Rge^#EHbqKu_fhVV?L+n~bf!45GA=QEMN(=W6@#-12;g20D?q$u7P zb(_FB-)F?#j)%T;;7Sa9%N%(I#au{kHvtWe6W5MN_tST@EB1&qpLYRs)6azmdzvJx zKT64?mc9dzl~6%l`#*mpY?NEW_4h(DM_>@+u{gMOS&is2m}s_Ht`9xqBbQ!?pP2Up%NzNx9ZzP+rC9IY9x4UEh-L6S(+7>19A6f7kV)ff!Y7+pXV zT>)ypSh{|+JYm<8-MjSMy40%g@YP!w{_*)EdiDJB?U6@GInHG zmLN$6cm}QlLER_WKk@Q;rMQe}bbo_o7(g?fgrOsLL=j?>cT9zuVPBnA_g-5zr-McO zGxnutyKkxjC7fjLVE1zj_1vm5Y%mjUTdu>i-IOboZ~z5eNJN1w9YoPXE-1|FH#T1p zA~}7bl=YgDkuvp`x4jLUPEKY{F)C`KqWaJc=*FCCX=SYR}rEtYP}qKHMCmkIqNvLnya(0b&-8qIgKW6u}Uc= z+34?~8`Eh7xe$5OQt8)CN&JK%yiW^0+8YqLitWSi!Ij!><*Xj(>pQAbiR< z&zy($S0|5>ixLi7SN6e1()&o~RNwF}m2YV5$^x{zsE>p6SdN$kgHZajq`6-CV~gZ}lDeuC zHq(=0TCxM%K~sEq?D1bjlXJK*X-rnuTrq89-aEo!CU^=(ZxvY#wcPN@rhkcgW426P zmYii~JuCyxI)mZ#@OprKQaD8roKF+hpdHRS4?Hzpuz3IEt{d`-2IeTkmS?BRb^c{= zZ^2gMm_w4;_Pf&KrA6Af(nDHK1Nu(^1$BuJl+;nnGl7wUA%0|7tS$m?i!DoO9q zb9xM}8c76?fKrY{icfUOE_># zc<4EqcCs6CKOkdv)l^!O0A=`905wGeh``I0?rjJrLU2(^Q}BhS6;4P%+aZf49k@*R zecdn2W&7Mqm#| zRiLJRf>xK+t?Hp9z-GcKef3Wn_*xbA zCB-H<$D5CV=_|aGUZL4{-lNItfu|4+$ZS=34QC(!9I!gfj>7S{nvWtQ<)^4JXm2oH z9i_aunv%mZ;63GUANTV(@_pu&YGRfoVz3kbwD=70-dE^w{#O z*22Qn10_(*7Ol{g#{s0@4YEax-Su^Ccl(DPJmvF3(RUO)pQ;C#C0E&1$D+ zA@SRGa8>junqV!k31oJ&%gestf(TY6ZA_kwP-a6^V0&O=7i6Kc7bJYfwH#JZ0@x3; zOgUH>-c}t1@HWBWY3>_AtUf)^WV%-{J;@{O)KjCF$pd zVzFk&jGK{5#h5)vIWF5=)$pgHy*h<&$)~3%zatuh?q2MARUNbbMi@*&?~=jV;ktY_ z^k(I#dZ+t6G@CI3xb^mdFc04L|7?Y-aoBm#(P6k3z+nR zB?pCKfJ+p|^KfG)T#ps7>uEh=MgPrlsZBU+Z7t`7vI7MHjCSxblT>zMM*=ME^hISc zZPZJ8N<)fW(e)y>2`uPBxP$&da{M9QE&`z?)o)hXn9z|e@2sqZ4BhOtb#0xUsr}WC z33|(6khDJ;?Lggtdou5z8T@8rSjff(sfp7HfSV_C>z3#F;!#yaSm@n7LPDWh4i?{O z+bS?^aaeRe3(4;Ir*j1fKzrqapVGC4B9*P5!}kP>k9as%PTkb>CK*3}sFeaR|~MWLrKzn4=Xm$QpRgvkyEu!Q=_qItinP7iGT{SS#NJo+s@N7-QBRD%0`fE3%&P#LbZKLMj}6u zy;QrWiwJgh^n1!Rkr7MZ9%>M!IT6}L1Oxtsc0t#t5&h2nVhw$tBCenMZnGFp%I~&3 zc|CgcjqCPoCSMT!?(pDJ82+p46xt9GAo5Y}Q@2yk75PQ6HCbeY0bvgvaTGvQIa@#L z0Oxw7u96*C^_Ij**n5u3^Kx>~sauM4d{IEfyz--powXOfzdLZ{k>no-%o_noQjf=u`HB z$VDb+8n@;;GCAuA@W!^Dy=xl#J9{@yhMtgV#xQZXPfy`ON{>etI?%vqs%l~%y5<^S zC2}N}bBnxY$xhf5)gGT?e3F!<=8D{EQ`;fQx0 zCefXQy9;Gw7?p}6pwSaw0CRExsAMr1VKoQb0)H^{JZd9mOTSW%o>=K~c8$G>hLB=J0&gfyH57@tTV35P(- z8o>bH9Z9*CJyZEqe|ruhxqN!IwkFy{2=&KTsz1u`{lrOHbTBE&(kFXWqlk}*t0tyc zi2ArU1=h(+Sjql*FZk@M@AyXHP!HxxoR{Yn7Mc1>YRXAG;+_rVtt9}*C)N2Elwe*d zHlZrSiRA~jcPftWM}{f_aDMKS2m{$85k?R<3YHQ`2#rToJLrkvBkK>}9d%Nl5k_oy z$b?I;Wp0LmHxO*Yv3czBy4q8o1vx4Dupa$KLWmcge|u;DjGvf?S!Ir#ho zCfcwhwVqTXi}x)0npp#hifHcSw@&aCn;T{SCDFNE;@744+h%@DkU>X9Gs2yjyF_8H zK4eO;^Flfr@b?=cjk$Sw!dM|!3^EhpR&>`~jThZY7?;yMs-g~5qLPXCy&5HaQll4KE+~k$Dw_32~od(e=SN{b@54br%%lh;Mw4Vgs1J33lSK)RFtJ2X^T<_ zmbw9zH>6Cf9Q-#4sx}YIYf$jaKh1%E=Zjq=x01(N(rUj5mId|TG@~qJd;54fj}}=& zuL0Ko%&Y4r^yQ)CXXsF!h=QZIUN;lMd#yJwz1*nV(>ftl!xv&96W4^JAslPuM9j@O zvV*KO?7^7G2Z}?0!bp3Vgs!fhyMapkCp^5swLF|icX(6zN^769a?}zZdME@woEvgWXE(0lRQpD-yHn6NhQI42p*ZGdIipOMp|(M^*s;Na#ZOA~b;YOPk+RJNQ&$~wMYRX(;!o#JG- zzy2^ig|fxGtDGEKXrIF$c9rP9^e@tm20Z>Vhv?gQ^2nj53iE*-vO}c<2O5zn-PFm@`g9PrWoqup(Q1Fd(f%zDfQ$~e zDV1D9t=PJ(bzokxzkiZw%vw51X2Jv<(CD)T(r~xsU@2K$iY9_s}jrltT>{Vn8`kdZC zg|p#C%>mKSC+;jZHr^xCvpycEQ`Xu?=J3UT<7M=cWbU< zM`wIpYe#|`DyC)wEr>|JzQ+*Hb)pJwn=YIl9zE6BJxIFauGa-c72I-WHcWjgC4C<-zXM|-=YzTRLK zV>}8FO&7%Of8uR^)?n^LAitQ~o}$1sfdN7Ve+NJ+=s?A}Ikm!0X~2}kp>~PLiI`CA zOc{NFBcw@Hr2|nA?i68IxM>3*v)M@P4a0!Dm~ShAdy0fH)M+63xkN>|sBDqFkM@*l z1MsjUUp2*;rHo-{j7sa8|KrVfW7n?Sf$Nyb}HfL088q7NUDyPQXWK&5ju7Jt*nJ{)4#L+)e~K_IE<6 zKa(jkDByiyC9io%0ICgk=nzUNFjm@Gep93YhR!Vuud*vViU93NiOIj9svH8uw*}4{ zoo=x|rk`|9@;SkKghsj^ODK|NpcW7T%MA7%N1$wK3s=7>BIAi{uwY_&CgPmsf+Cp^ zFlB@MT1T=++O>;}LyZljm?=BDNPzQihl5W4P0pdmZt9~cbva0~WdgCXEOrp>%(_Lj z*wdev$?$kb$W06zD!!iXE<9fk2`XQBeg_N^OdFLXE1kic>`;nT962FN<5&QBDUoVV zS}7Ou98R=~kfj~G0eXn`(%AB}aEXYWbQx9Mg=Uc2`bwcGOyeJQa!o5WdYLk!Z8sa~ zg)}eB3nUrgSZX^dD%Bbp;gelZ6%=cwjU=QC<11GwOOdN+;VbJa8DV1Uf7rvSiT zU^Xc}@$!)NWD-&ul?KXuQjsE2m5lI(t%ct?Or@O^@j-o3iZ~9vjm|v*fvFoEOzvqy zE3Il79@^y!H}wv>ww}7Al!3Y=M2Q;}bZnKM5|tS}<^ZApDL#8sHUOL8JJLcS&v#`s z8CGMLID7HPjU{_iel|;sEu8zydFxRhp7C|I8GEN>8UY+K`40Zt!fz;`*yrQmj;w!_ z%NJ6Rbg(|J{$S(`XiV-_xi_#5j!gNHZcE}|(L}*<0%gDCC#P+dBt$-<%ZkS2mPNnI5>`pW2 zi&?xuCP*!^iEe625dwY#S4^u;qIKbZp_>veLil`n{uk;tPu&N!VaZFchB0 zy`42LZ*ML~W0?2z@T!@mg9G+UCNCY9yl^nQ8D35<_|Rnuhhnu7>2Gzl`R{H|Yz$Z8 z&dq(J|G$5*ak(A&N}Rv!6Vis`@BA6>sAbE8R&4Mf`uMor9IJkRcSrEZfYYfsAPM@x z@_5b18e)*n1K!bz+*9|2TpSsk7@9fiX1cW=o`(z ztA4sO*x&IuPKSZ_X^tTK92!izi?$!IOE6AwCY0@FAcg7s!0mIubxhr)R0&o}(+|K! z@jAZ`xmBv~tV5fB3`tsN3q2mdCJLHnG%WOeOB+PGvrq!Fs~Bk&kJH%InZ8zT;2g%Q zWRUbUDBVF{$#b7{+(~0NBESzkh~zgR^^2~rd1Q%wt`5*kSl%al*o$^};RgyMqmK(+ zZH~V8VW8nA>lU~9s0s!efsUc#$15Ilm}+_NOzr}gVQPvh7GSiK=WZecrmq8!cwnr` z(_$`esu=lGG-E z3uJtXLt4goV?;6rt?Orp#=c=a2ZgO54?+r9;A1l}O3rh{k~C~UlhkAxxSpCkAm%PH z-$Cuv&KY1dpJM|}dhj$I+8{t+95X}7eYX^Ou}U_sKGoK$ayVt zqf%qiU1jGho65mr6PE&;yIPl-&%^O_Y5Nhxh|E}YVt+2r839X#OJ5soBJfG^#vt(5ctidg%hg&er(~Re z1F$$?^<7M=a|P#1AH%5mjn%FF?B${2V`QT4p0K>?;%TGP-Dmiy&r*wy#m-2Qfxwv9)Jyr)3) zo(^YCk{&(ZKl=FY`xhs#Cs)InXV^3UB_(|ZH0-tlb+_QBWBo6*Im+-d)hi}#~*?A;wI z17bbpo_`PfvTBcSKV^Oqh~=t(cZB`(=xl#{aoJR&_v~_bG2=TiX$8=~TsF|?)o?tS zH9v^Oe?fHqdos6>d-Z|~DSkG(Y&2}o#TT90+@xpfUCWnY+9p1G3%Bn*@BKJFo17yQ zz1DZxVdq0=FiAySu7pk}Z^t9_ZcrTA`*he$m8Tz%-kp4GsRCBqUJj>!55Sh7*4bjd zhAqyP$b!dLn=E+ffh^cxa|eT^u0R%7w*o zetEY4{_Rp5@sMnB>ud>M#g^jIVohwbMGLSc{Te{}Z_bvY|JrQvUADws7M243u55{S z+2a374cY0QpA5;vv-G=vfqXtPo!uOK8cr+Odof&GA{a{p4iArheEGfpF!zFb1Z8&|rVo}FQyCv}HzfYgPegT4kRvqU>aE)H`Hv z0e;uym3MarPVT;c)_?f%?c0|Jt+gck@gb;U)~VvYW~sNXywrDgkyxbiQqSTWOT8W3 zT_CiRQ3tzP9sbU;Uvg!eH@3?gzXNXy@VoLxiMzZ}yYpuM`1!%d;{}T!a*Ghx$`3EZ zS1mEIw8S8`9o|STF+6DGZ@HxVmP<^@du^(8Di1GqsXPjQ@HXHWd22-Zs>(x3Q>;_9YBMFUPUUa&^684!-@H!U?7|gWywF{?>{5Je1mBe_ zdc(3}cIV2${UZ@GUEklKa|B z@m;3uQg{@9;CJSVy2};2Q{mxUr|@uot?*E;U3uu01<#IO{q*K%Y^qO=HdP+-Z&V)G z*mMa|EI+*3RDNh4T-fxp`1Ho4TYiLY`7!-8nU^5F?e!)vs&DN@v3hndvZIwzV&Ol3aCzN-UUvQW-)1`L{XY7m>7|;i z<-oQ{Tl>$euK)Bp4_EHfEf)T?|F7fY52>_^cipESto+r*@)w(AJnvk3&)xB#>H2S3 z?-r}vRdL&LZrkAPoolYMn0p{PJY~^~;nB8@y{CHpNZJ#hIdfr?XB`e<#IT zo@*Nn1FbbwQQ!R!3*a%2lYx8_mqD2X8bV_@VK!ztW;8TpIb>yFHe)a{Ic73sFkxdh zW??yDG-Wn?pW;tdwG-NqsWnngBFfuu2GGs7eV>V`CIbk$qmz|jeM}IgpK0XR_ zbaG{3Z3=kWHP&}jS4S2H@ONi^C{>hL00k7R7>yE)vBbmz)~KK$8oN>yq=^kHA|gU4 ziWRYAFK9$XL=a+2O!jQtJ!Wf)o82`hd(LLFyBc-F^8LQypU=7H&i&2%z4vDB8_Tk+ zrY2>n9BUelq(kT;x`4*f#eZ}OjaQcCW?7;urdSp;OPLYI$uvnU%M+u`dBx07nnqK^ za{tbZYG#Is6_8>w4rYea44O_?(G@h4t`;lgEAuNSW=4oL>U)!E#mu{6MYYtlX68M{ z-qejcQx~z~e$#wkR(yu3cNHt)Qxkhqcj_V5lIN&gx)&y;!(=B@*?&f?lpqtoM%&VM zVx=xXW!mpy`iF4&9hmV?xH1r~S_Ib&p!F~_AFllzuH&^fxYPf@jj8k%+>{2hZoH4sgD`J3{TJN64d&m51t($Qa9DI77LTRnup|NQbb_Vh;4W6~dH{EG zbk8vJ1iZxB8)f27w0}MAAl80olkX^2+3!s3BUZV?tgCF1P+V6+AHzZMrpA|crd>qd ziZk_V>3Zh-(5}>v`iqRl{mK|0Gf#j>Sel8u({40SB>WI6$N9kVk#NEWIPnsk^b?$n zXC+hc9%bsUVZ=%}tsQ*{r>}xh4w?z0@4^|JXC|JT%$iI83x8+#hjU!$U*TM=8JUOY zEc5>i7c7GdU%*9pv$EI^#@4{NCK&&3n6McxX?gxZB1>C#S#R2d_7q9-F!^30%K}Us zERsAPGosbyA-NZ*PJPe0q&s^B=E=chBlMh1 z(b-g6Tv2(2)8E4O3u!g%kOMmfLGO6z!`XbkhrYGuE+&ZT{Mf`(=p_2SD8D3B`cE~Z zlbhquO&mcZ=`=cBQ~-9mT5`_RL~*Z;5|!HBWPhTW31hs7&Z0ADjHqnvYnA;J=8S|} z@ikGo+u^pBd1f}pEjw;5{eaFEm4`2v+KxS?^8W-29>PLCr6Q~%ReTnf41qg%k2|qD z)Gn+jwVVH8PXXLJ5bom)`>_XASqQCz2Vy9`gX*9!JcMtMI^2u0=EyU6l<)l*-|um( z5PwyHo~%w_AE{HJ6nj&ho&nF`ORCQCAD-_|v3JykS@2>%bMJFRRgN|BLOPGm7j@Y} z-@LX==|yi|3eM>wz`S$i>k$US>5^( z))m5fd~4KgKC6a&GjlABqw%5|dt>x(4u3P6z%>vq>W7CWW1r&e-M5l1qxLB-QPeY6 z%TsjZilC;y;$=ac7LmG ze;jtmqBtGtj!&RBx|a5FhQ7Nf=jnVM`k||9|NSrk{X}>5qxiP!ZiVLFQ$+`!GVuzU zPM3@BJ_nUOu$y#GFIsFWGeq}#Wa3q#`|xQ5^Nxa_n*3_IhGvTH%exqvYewx8&N#Y} zW{D0jH>2yAnIL**jLEELW}@iCJb#nfz|17k>srp|_B`*4&Z{t)P0UOdy(i9Ovgk#+ zhu)?4SZSZWgqh8Z?K8i9wztpo_W9gClecE&UYgB}Gj*Y^)Q!4R4_0nr+=hD6wzM5b z_cQjQ_NhCEZe^xD?La%yPSl(FurimiFSXB>2N-vue$<}^aC95vuCyBsq<=xQJGIYO zdGsJNJ!vo6oA#l>)ILpZr}@mhPWw^&G-aQi4l^FWxPb9M8baTsp>z=LzFVV~NBU(#KXdSJmw`l`yq#x5ebO*gc zPtcR}6g^F==uUc-UZdA(HGjQ9Z_-<|lxm%g6?AAAKL;fn@o)8Z^BKyQ1p|*CY~w!rx7NOZH~)KJWCwbd|MMIG{^NO{y-c( z-Z1g9=D4MPwmAA8GjVEj{67=V5l2{~iI*tUKS~%r9haM-1Re`8F*r5~B_%~qMhdm2Ub6rI delta 71569 zcmY(~Q*)pVlmy_|wllG9+qTU~CN|&LwllFgv29I~iEZ25uWE00f5NHii|(iD{OLd& z8bqrsy=oh?l4;5^q#^gH9|dEy?Q>CT{OwF*2yB3Yp_aE-em zYg@JVC9RIN&8EKj&`F2Wi^nF`B#P#^$R4mHh=^2gZ|mYO2Xnveh=+lcu5Fu5xwn`7 z{Nq1B4gh?=OxDGJf2=<>4C_UBt>2H<9lw6KxU>rV``ibX_kpXK;mZ#nUr3KHzU$+B z7e+gy{F^?ZS?KutZ|I(wp6(F1@AvcGnWY6O+v&^NrLB&-)*5>r)%)Ju^)G7%-&f}^ zB&A;zc{NbqkEhQYvj3joRRd3T8^Fp z+t899-BK7-FX5Lxqy_96bGhr*&m}wN`T&iNB9UCT^&<`$m{tsr?bga}j zLXK`ZT9`ud6|{mr(_^n%&HV!qJWijn{46327bmmtd?D}|RruSTOkZ{DI^NeXNa|zb zu=35^#T||SE>OkzJW~a!7s(cJ=Y`fdYdKfu`3B*$jXmJ!^j*`vm-<_)x!5nd2-&Y- zHwWm?Da2fM8S5JPJ8EKaBjc3-@sQm5yzrWkYS)QECS&$M4(%S#M1|Z8PskvJUZw`W z&vW3o+jb&tyvixzizKD?GfewbVVzogPs)hiqrpVirFl1`k`JNT4M&#vO03uhUdXsBJ&48K%tC;;cjgMRuhOA)w<`(5t;dPZ$kxeyRsIgz7tSkZH6cyQKa0ZeIC}osg!ehKHKq0;07_|TVq%_Q zCUaMwH@7sg66$&MIq9mMS*!@PNH_^vcAbB&^v^VT9u2a+*LkQxw8c4-Zsg9hC}oQ; zMgCm=_?UPWb{YzeYNrDYf^H4?cR&%UU(a}>V=^mJd zFg7_x;0)S3WTY(fIP7!^^>9SlOusP4Os+2Y!N8kQm2=ghi4Af0;)_Ecbk6!DrB+bJ z?lAjf-x7PuDEQ|$_j%48R@Pd`+ICU68v{*+=DqNNjFmJJZ)=15JK(>EHE0jv0wd;W zr=T7w2cShKn46A1q{QK~k5d+P=Qhlm{lG(#pfNv-VirHi!tI0wjaR-q0ijiII9w-k z-k43LYgOAaC2)k)7Sy_ePuQos!#taFS9zhU7bl|L61tLFXc&znfq*rO$suyIk3vz? zLEt}A;7d4UeTz-bA-FL05*ZFwUDYD2F&@JY;nUi}qK1cVSCR$RhIA4@30TnXjXUJQ zF%VR8_)em1w-t+WN-S%ltXKEz$q0-Y2(&|y?)Udg{H2t0Bgap)xd6VHMmpT%Tbn_c6NK4Fnv}jKBPo*;#<_0CLP|M z!L0Fkdk8lvwGA=o0Z zTrTfw3(a7s*(O(Y8I8Q^yytrEnZ<@79Y( zp#yzqVEb_xmH?4S%71b1XRx$uaO0VV%I*y9AI(2mS?101@6ashMf3es>PK*ga@~&A zk_4mFn}ch*aW*8#wT@_lDx{2sYJ)-5f6v75(Qr>&zapdzy)6Tmrb1gAdrb3DY{m~t zG;k?XO=Y4ttAt|Y?r{6Bw{c!RBgVub+R68EfvK?mf2N4>12VDt_ZM}6m2uA|JZM}o zajRJUx}}b-u6vk)I9uk|`)TUYa_|M(S8a!#h=U1=XAW2@!WSaZd<%tiR*sWv_V!bp zLMO{nd(`$YqjtBWLrFM#mq;o85$dtHj8g;^TWQTv8?3z9pm6%3b0pid%xXF1~ZJFG{!OMQ<{zX%_` z`bTGLXM@+2A6aW>i}Et+=}x9c1Lv~t6DLn`#%NR6N$KwnI8?JL{@lv8n*As;HWSbE z+E7WYALX?FM7&{O2*B#Bm-mh+4+J4wfBQtSjBPy=hVk|zaM7L*skt+K)Cb>@10SMn zk7b50u1T!px-f9z_#o<93{q-JAE38SC&hhukU|vkFZqOOM0*y@8BhlbZxY(>1dE`Z zmY;g3HFWkX<<5xv)~5*%FwY$9*7YNOB-Aq1G@pe)iQGDr)qBe_)kl+a?b zS;$Ws&-M>5zIojEPn&9bjsi}qDKDkH&VgdUj|f;<5#ib0wXxAOorm#?>e3< zE=1OAbLw488iOV(dHG~mb7%aR$A)3?l>ub5j5CI|vxSVT1m@uf+s;XEfP~N49$oGD z^5=LIY4EOhDm;s{Z*8ltQd|EfMbw(fRT}>I?(aSVDm{7ke>S?+ZbZw)_h3b(ThZ5< zcIM9(H3(%~6e6G0*QCNO(8=bCrkz3Kk(F&)Qr@k*`1eQU@s!31U#tHCB= z6%1xBrri2!Oh8mPuSeaC5gfSEHP@`JeVakmj1*buQal||dsvO#0Da>++S<^Vj9+!* zYo7;*cp5KjBk!y$o**V}>OWY91J(Pf(0xP7cq57OSMH^w6X!Q6>tY0H+zB7RO1aIo zW&Eq>Z)g}Z3Ixp^ypZf>>lxQ;N%Yss8s+)*+Aw8oQ}5%wTI$)dzgdqDHPDn1AVAC_ zqNY%`RU7?(aWej^2Q-V)%{hSx+xRlo;biiaNV7s$usei{a;=$FV@DsFTphv)vVH5+ z57to#s}@*uy<)gN!H{Z(inV=X2HF8H{rqU=@od|p=s2z$NHot0)xtuLo~xoV!-*(4 zT3&&m>)I4&BZxE&%bKUV=E2QZ5EtWAkn!KReTlO4kL$Laz@mhP6yCX$wG5&@5*I36 z%w@F5L>sfAYTP(+R#PKa^;ECx!~4HPdMwmStb2G|Al24!W`lUar1 z%v==D=RY0uE(-F$UK+DgtxI$ve=~y9rVLX&tD9HGFdcv;uq+%tdnS!`@E*sQ`55=z zhXf@BlAd1r&s&^mS%558-|rEpXdaU0(Fo4QCrrz;Y#(1U zT%2&VE}h1h>}7vG^Zggy9G{MLy*TDVfu95RMqtYH_k@BQ3#y4KRZSd}f~u*%pBd!T zo)DIp{5dKkQ8~rI#L#7N7c6MR-n??f>SEx5LjB&<_aV8cpQr59Jjzu4Sx%ny(N zB|)+%_wn#ZcdRzQ`|4n&o3}BZtXu<@9SngYlX2;*cZN3Y%5IL6`+)g{48IvaoV}Vj z_Vk7QM$Z79A^%9QI^xARE6vP6GMSDJvfti*`_z@=NpxS`=X_IV!#2R(nWF5~8|$|x zw(i5n1)TGqFH!RGV+lryDEGKQXV4um48`O2@QEN3zYQgUyJ%pX-M7rP?BS7K$6ikG zIJgxA@0PvCu#Vxd8de8qO=KA{SkAr1w`0~7TW)Z^tMfo+dKA%b;rzsg5gyzp_Yk91?~z7EI!s98rfC#lr~>CH zJKRA>^yA%>t>BWgc%j)eD&B!^WfF7=QYwo^wqGV;kT!x1mzZ%PElYh7{dCK6HG0&_ zw_qkAI)9FSvDS(wD5zHeouerX`qR!Ds-1^KXaK)A`Cr+l+pR@<+JR0jy!!Wurr40; z^yBY6d3;VR8H?Xle;|NQrpQ&%sb4!Xw!o*pymFOe)M=V9?!U>-c$CRaZlw9 z4C#bj@ybJZ63di0X~YUUwS@Z%`TJ8XHoi2;Wuy98hmg8x<=c?HYLLvkRX*9b?>mld zKI6!9{u;xUSyws!XIqXjj|2a4)FFz%J2rO(+*%}ETvM9V$Cr6#$3y`Asge9}M|{f2 zPJ8CnuaJA#2e>^P$kAQ1h71qL!(B_Vb+D*DWh%itl75(ng?J2K8uGP2a3OKYPK*hC%7I^bmojhh?D;RiqTlU#|r% zuAJihjN&|5gY|&QcH*8FJh_=xN8XX7JBV$dsN1K!s`QEhrSYkDz>W_vKR38k+u;c3 z(8XN}8eb={-sqD^llMp zzs_5QoH`4t*_h>8>i>(`SK7RLWM9Jidlpz|CftlK#+{9acgD_q(j6ho>lAJYn}LBpfo0=zXEyXdobut@)qe}ECQ(Nf7oX)dEVpsDEogWv zV@UY>mT!GMPxoZr3%#f9Fbq2%=&O3Ze+nJ{`4;^C99=i^@BVys*qzQQ$}n~Ie}8ln z1nPRl7#(_vfbTY3^jH?87}zbG{Md$Gp&z-kY9F2+BY8lmW6K-=w~=1%^sdVG!F{Yy zzIb`>?>P0D#Ga!vH~sg{ukU?An=;;2o~X~L%KQBb&o37bavE)aD5Dklp7LhkM`De{ z2ypYhzu8)r3Ibez2aHymtCODp;q3wDJQ;7RzZ8`KCC!IkTF6H~e1=051Z=FMh~ z4Zh7_lN|_j8!J+!^?t zb#yYNvLvd)Y!1AUId-cMIGM&_PF~s3)FR^k{d4w-TK-7v!q0C#&D%=SwZfXewbd)% zapMHAncWzbqnE3(Kr6}#Y1&S0q)aQ!H6(%c=z8|Mv#ZHBIM{F88nu_;mP2&D4tv?l#CrFkESd39BcxQBS&{e_Lmi{En%?Q)UB zm*zC*HOfT=dz{-8+x)~_hGCW0dcU4J<8TA~=^OARRl|ZFK@keA27O?m6$&Gd@{4zc zBU$VQ*HmvJzOudXA`niSYe^4gJiryW&Fz2@l_hbfac-AUIh^U7ejPiu8L(f-zB!=^AH`ID&n=BhJF1Bn{BC*3X*%E0!w$yCG{);hyDBp?^x0`)$z5m} zNB?Vndj3vDHnjiiXx^FGl(_{_HFVSs6QepM3FO|e`T0-iCHq5oAK12(r@L|2YDHo6 ze)!^MDF4>zQa;?!4D03^a{FQTC0i_j6{_$v`S0z6{-R;AqMdU9a-4(>rJOE^gY+{y z=t-~R!mNuA2DWJymjMnK6{5zcg0Lk)T)MMYZKb%#Q9GZV`tz`CV%!R>{F5x_2w8r| zBtzrDx3Es-L<=4_yVdQ8k7C)>f(fJz$#dCYVT+L4bL3YI%_DXTC#7>s;Pi&%eOOoW!bet|j}F#z%ko{?-n8du zhthsG%gyCADg8jOv645lFV2OoV!nW>qKNIC`<~$IgB#8q$;jNnkD5&AB|NcU^Oki4 zw}|_B#GgT2`NN!Fq2&l5h(?Wv-08x)2AL@2VTK+&crfyeGDM?sa@(1J&O#p4Ae)%< zfMH^de~Ti_JpS+(RUZ+Z__?G*gpcYoYW1$Bt2+6CebOVvFNah6GMP=*xxZxZD+dz^ zK&dg6%Ae5es_1qDGpsv5E13#7xb=lJnsvshohcJ#HktF;v7G^U1-$rq!v=Z=Mb3l; zd3}=QtdgWXxIu|P%+&?}zk$|`^O}1}uGNhmLabzV}Arcsu543 zqe&9&QlWA74lhK6lT9vq$paFLIGdk;D6dxE1Y>-idqGgEz1N8Rcfb90S$DPBVLU<; zogD(=0yu3c7VHUS_4UzA2qMlON!a@9PUvNiyj#JU!}q|xTbcgF_J&zyP02Ch-P9$m z(8x86c)D*?yCt+4R4K6Vl#ZEgEvotWv`%IibVhI%Fo4Sj7f_4Wg2}NGr_$8kSTxZY zL`3%_^(J7=c?bi9e^}okN&g)l-U-Dn&|>mSMoI$EmVXJbBUs%C2@BRoOIYa^j#-}K z_h~--#l1fa4IiEv=^bs*)?yDwIgmCB(0laz&vv;!mMTxtW&k z?`+ldE?$~i*-o+$K@s@X4=WPIsxeU20CA~b$U@iqrX39k>?+GFfP`d>QSO5b)<@u| z#a64wM1vn*V7%h{14a(;ihX2Ad>6DsNIP=OF4hRm z!pFJvGNDG#=;fZ-%7bpuPDAwi#iRZ?t?ex+DqA;XYNaa8l9No#h+S`+;n>VmORqwX z5}*rEia~IlVCOoJnml;Px96BTs1@(l=5|cnpLvU~J33&^Xf`ZodUy-uY?wOO)(5da zDyvH*k*BF0{dAkF2DfG&fcOtURgLGW{L@9K2TTcitbN?tz+kbsFEd>|@?C$qwq$~}3}vi)&$9#|hA#1_bYy6NcmZ z=Q7UxGMEdLo@5%G*O~Zlw(b@(F#Bcz1)R?qj_3VC486&1Weo<|go~$!$_R#yO0L!6 zA|bT2Vj#pC4yiu#XA)%>7UIj04HjgPKByP~9U6COFA*NHVP74dVz6q(N5OBbC(D_lIo<$+$DZ9{ z-vilmN{8j*uU36}@N@s!D#z#IaGi^a-3{%gs;fn$Ptq*ot|O-+rVm3dVMn$HQD!l+ z5JUNH*k&4M8kJ6jw++fxOmT}d2z&ls=}RB7zW8{jQx}tI=ZYBw4ypsN<<`-;AZTJx z7>c8CN&ewc|2E&YG>+&`T-{wxmVF)u8*6+y!c2md8rugg6jSU&$0qzDe)Nx;SNdb9 z-T?(VrlMmYK$-&Dt-WR?U2^pP*H*^s4CZ!-niJBKZh1{*7hojYK`HMqVi38hqT zO@^4l>LJABpUPSb){?bgXY9&Cq7L$&e{aT?f*x*l8Q!Pd`wJWM|(~#GqmZkYR zV1)K5lL}Y-*_L~%v2-+1fdHR7AWN@tpK&~evYKYuyJ_wow$m60_0bj1=_pz*TXnoX zo3c~85x`1g-j^#|u_fqUJGa7yMH!SH9`ic%T#(_fV$9dy>(B)?`~{;~#?q6%u$gsk z9*-lusLVu(KrOPyGH*8T%v9UrO5enx0zN38PVH+QSY(4BP%t>3Z>+$U9wNdn6|veu zIJBCcr}IpE`D+uX$}Mk)1t-n^BO}N6zLs&tML|fB4*P?TZ`mD=m4=&x{yvB{W^7Ea zN1LPibnfj`ZPsJBewSP4xoNJho3b@nIyo$>{VKmtNT#Wh+((Gy?{0J`g*MAQ$hkLp znu`o79-4(hNqPcb_SfbnB@wx}ZquGAG#1wrJHzkE}+qN5; zKw{P79NtGdO8wbI98YyMXh^f+ZdLhlAg={$UNXR3dYB1ES5l~*X`jdgS2pPQC&u-p zVxKq_vkXF*UdzPKtM};N?)?`GRbtm_L%)bA_*CG)czke%dF5WeBc`c(y$-}tk84Ek zsJ6ZM&H61s_q`K_{PbxhXYB~-0;@oLUr?u^x*?%S%7#5p%=<{K3>!vajn0YR;gKU< zfBd@&T15(Px>}U%m*_OPQHnKDLwh;+pw9FM;pxeJC`~D5Y{x?rLqN>$!cfxByT{xZ znwO$1j9E?zbL0(5-VMJFRJM+(3az-vN+{X-V&-Llb_^^DQ{@pZT6gDiz@iq!At8-w zbP)1S)jgbK?7WLge^-cpLcEJL)kEsFQ{8eiDjA9&>ci1Se<~)oLUydCb|4(_qzfU_ zl_XgL8VM~<7Y@m3=hLLY&GM5zQm1;#j#*OGY6Se$T;1^MpjZ0*Fy@krzGvo2k*0if z9Qafu58L2>#8;UfldwxSf?{WOY6u?k^KE2mnP&bR2J%q?pH zvRRa?b-fucDybR&>2bX6Y-uf#q_gWPtVSIm41QnAygtN6Y&gAx_gHxDYQiI=yGee+ zsyMlY4G=L%){vmna(t4^ITN)!6h4B|EE3oXl8LYfW? zSH9z9!z%f08`|$*CJDsO#(*6k*V#{mWediCbcYS2sh-=32qX@8Se1YFfJJV_nVQX( zk487rS!e%RDrKL<@Q+Wk>DR^{p$MkttTs*=3u z-P0^&KZX$m*tujS(#AIG=g7T~3Cnh)8=jsah2xJN;x-TUOF?5WD>Fo3G6PnLT zV8o7v*JlNIJ&r)ANefHrgf9V8egQo62y&0*Myc@clOadg&rf<^DAeXMwHeaRS~4L^`;t1 z8GrqALEwDAz#m6W&*O{z-<3UPPbS#@^;889{j|xV$yE*E6I+xN3W6}oL*q3gwzMM&ehDMkS%==#?4X=%|Y_~eW ziefKpL8>iw!I9$vqAvhN_Psp(-fLLkvIW0T1f+8ROhv7hCTFlOV@|CPn+}d2QO#VW zf(kRFruKuL%R6Zz2HD_3rZB#yWf^WA|L`M*=W&zc?M5wATsNkkj)O4NeuFRQBPHMa zkN$Cz-_Y=cAIXWUn6@{6_Swwy;yi)ICbw}>Q`N`Q8O&LgwKxD6z9boGtk_aci|F@a z+KTDER&h@$2pU z@)~#(YD@IPPDjNCLr>mer3N_uFSpg`ch-`%qfUQfz6Gw^V{Hfg?hCN%TOCNoMUpJ~ zOSxrC_T4j1tJyGSDd%B%8Op5ni<9_~|1p87*J#Ru={r*{L=M4Ajz$W2ebU2yZ}$E2 zf1OO9%>TAz9I;!1OddR({1XFY_)Q8PPNt9c*1i+X13l|juYH$;l17pn-M+tli7vPA zHht;7DFp_z@1|3j0+^OZ53iw>`mu=~Y2w9qVkb)9>rDv956Q^f6kVI(Hco#UAZ~|x zQCJ6i!29vk<0s-0}b)8y#-olW--n(Yun?k(P59e7*ETQgRy`E~xm1_ng) zuCG19K&}lQ>Ra>?e*AJPfe{5d&zVR4;k8sJMW+UR?)c3zNz6IJZ zhDJsUpic}#*8kvrKR?4-NtSXhpxwcf6%tx=jU1A>Kwu1yPS;7o&@l3hp8ZEdN4Fos{5-R~K} zw`W_P?UHW>P9&FRB+YGKuf*23Akz4HJTdXjbd+PNW{HcTNi|<`Bq;f{WjDW-%I}WE zeaH@EyOoD{jvyS^FI9gh|j4R2bHpK2EE88Bp2Yrj+y2mFpJchZ^6 zm)$AU97oy<^I5`I+DYUE&#dm2#=1IaoOcyBEIUH|6IVi7iC$)kz_A@env*%YazKy` zEVHb3Fh7yu>1==5&_;&YXNeF+39`Vb0NG5@ipj6)A0${tu>He|L+{5C+6vZP?t=TJ z$ba=Q2(1RN8Wk+D2urbjbdKlod$`lbS+mM0M9jX_&l<$cnGEx?%fx8&VB#P3Kw)~e z*ReQ#%v0Syu(J+@p>R_fS_eZpV!?j3f7PugS$^oppo{;0XC~YWW^ah4T$$wszKA83n1;OK3S|V%_Ravkw)}TqP6O^kwSfnlBJu(Gm#y+O%|7CF}S?uqG zFeY)Zd==5!dA^O!)v~(fb3T!&b5|UP(p7URo=If*O$@sDS7q5zj>w_|3OJc7GvJv= z;I)y8YSQD7CpB$9x6LK|%Y+i?bveSd$Bx_Z8F01c9U>dnF2!e~5IHbG(uvE!C(^vW zjhmHz{c5fI>NVvGk9-*^!@3ThX~f=rXmk&DmhGGbgW6FV9;DjM#g=ZW8>aUWVn@0z z^TZC{Eox!a7UV)!$8K9930$4ru9gboNg3s&QrM)#v$ib68L(yvLUo1?uyC)LEd}9E z5^R^(4&RBfpMFVBr7JC8*){v!V)})APRI^#OV%?ystgqw$DrfknKNqha2)rxp)Oh^ty(UlF(d>&9KK*!|3acpjlaC>B#`+1qDs@oWZHq2!skbm`G+n?ZV4m(CyoME@fd&Jki1m9h9GqGycttxC` zw|~|OU3;F(TYqok8(n8NtH0!Kd=PsvKh?U49LhkB`+ngEbbG3>o>LvowH~J5Fw8<= z)Iv@Mj^t3R)X`#xua8#qAaT%jaUMKnSTESoK^N6#yS!X9DriL=xJQ;)gtuj8k3G?^ ztSYkhpy1JDMw>RaKxfcsvBgi(D~M?trhQFEaW3b?yBp32Ou}0bpMjvCsMziaaho6ZJV;P28-$Xb@W#<>C3NU(J%ci7(;HxwR}$uB4LTVLN## zNRL;EKsZv;{K_abWx!VIrfe~LN6twAf%;oTc<351oEI{TqTnP)Q?;R^>hEThK#qFR zrc;zGh#YyAE}OA;iDK zID27oVatw0#+s;5gfAqe!QuslMhYwsZ6C0nr>yjU5I$wDNP z{7-n)Us@gzT0FBmw6ns*tT_RRoF0WDR?7c?hqTe}xk}|rO<&iS*`MTkioe{!!`YMK z{BG42`$EC<3fGp`yBu?2lP57Am#G8z9~bBbvJbotMwOBgNKUu05e|~|& zyrL*rm6&4y$LA@}VW<&J9)<(Fe#?28<4hTpuxg6ajVg_%HuI<*q!QL6fuwf_3D}2J zbiz_fHebkp#1o|ii?;2K{-}i&S$WR`=a&4e>atHHV!D=nla)M#iyl_Nr%OZL2l)Pv zr`5tvdF|~Mq6u6sga#15y*-sD6`{6=r&E%j3tMKu10jbiH86@*MIc0t zx4QXW(?>LDMY`kE)9T?A=D2c4@#g}hW9`X7_?E}lvretG>5fn5a&}eu1j*ZgPn<%A z2QP92{i3nAtK-|86zjjUkgcf=o?|mnvwcGj#ud7hm11%w3r6P<4O94PLw4%?N-^QZ zE}mMHDY%ohKc+1BL>8WjL}jPn>$Q8QIzP62KW-?m7c84%A~Mgy)D`;K>VgNU^nxf+ zor1Q_77eC*?3N39XF@Xah&oJw!XH!RdR__D%u6e%@Wv-56b#!GS0|h2S&kYfEV`*P z*IOM$fAa1y@Sk>rDb|Yv1nXkPZvMjSE(&sC4{Q4<54gwJgbcm-Wvs6;Ii{NVMEAYM zPg`=IEe7`NfgCMfRrZ@Y<04<4q+N0$J7NEnt+MRS!uORb@5Eg_70e_CHm{DeSbpo; zn?c~6h@k|ImdnvG-7YhwZ!?7|j~jjINoVx@W>+fvIIyjikPN+K)W(Cxb_zCOjv{S%Y#cZFiCY`CguKiNV znHCxs_A zs1euCL`u`LDP70ZsXws!m|e43DH^@V6F=R%Bze{NvxT^#srwiJ+2%$ZPaM6K{?at_ zmE0v)FgpA0am>Zm75exgu7re;)X;xDWChl=^OD&NlUTQO9I#>SWf3oYE~}za_9r3# z)Dtr2Gc4U>g^vhQ5v9j5Kj-s4?RAuC7}Na_ucLkAq;--DS1JBz8w&1le(sci58J|cS{&=)e5TQF1AZ8T%`>mFVti7!s6*9ods2;!6y79# zC0%m1V288}u+gKQ4CSrLyqU2VeXtQ1vQ_^Ps!A>MRZG5|+h-10vn=lPg;JLAvSzF@ zFo>=N^3A!SqyC^7=X)Qxx?SDE@RMI*Y5I60b)IA*;~F3ZM3_0Rsq17dDu6+~l_ie)5%KZjYq`XLjYOlod^RKGm^tgK?e8$)0Iu z@u~F9WmQ?ic{X5`ac;k77cc#fMz;E>Do8KYfUQpm8l8K=88fj-Jj*hq-$9ZVD z0*&BPY0)GAYu7&OA0@hKy8!M0+0Qy&i#ISc8`^r7I2J>y&zc_sw3mc;mc>Jg&g7Nq zNs?Kv$}JEfBcY;%PFian4V$2%bI6Jb>%)15R!<^ z6X{;FG@2n`hKAB+@kSt@RWv18pb*CSa>UC>6bs2)7T#D;HPQ8wahnKFrLXBQl0Hd{ zLy~uZb`^nUF6i{~oRJX5M02{yN6kMzpvXpn9)z!LhY~a*?Ebdd8Uu^5v{ks;pL|J* zq6#=x`KZYu{L3N>-W`qL@fW;$pf|2K%voNJGzU)`Dk8TJkDB7uZ)}zpBD9Ukj@fwV zb5K-{?m2}yUcO=1`#88`jfx)Yt~FkO!;%6_j~F*N?=fhg`n+sW>51gjSs& z;VI=KOV9{*&g36oA-=r!<+4+t2fOYdTX^DzwH)!VrP0l z2qyo-UEfp)clwcuD^uBk3k(sIh(R@4v( zXPbu1IHGmcYYTC=9$z%zz&jxRmusUIo8T6|ssi_*`MI9)smMFrjKuS=+w*)z_(B0?V$>vAW=ZPX)y9d-Z^Ym9%t zewTs#4rv`GG87~ostL)cUeG*4=>0ea3boY2{&)7Di&6pq|EsG;pRdlW9r5a(&OQ^v zMYC`#z$qF;CcqW6R8F;e*+;Ep?Aw7fl2oGm*DNItf~9XYStLe2F%9r>EC!T+NC6Np zFoFgwr8|QDuO9wSbsx`6#K6bb?Lz)zcuRVNA=UToC!NrD>AzHcdE#wipqB`7t$ zox%hV%dTd8D;(Y*3-x|K?kjD3;pG~+tc`NU%A~W&v~$M4^7Izn`F2bve~ad=gMDpp z2=!*}*uRj^G77Hxyu(`%%oG9rekoQKGrZeO`P(_FuMBNK)?MAtPNt&V=~-e6!|Yzz zIZas$jG9uQ;5*c4vgt!?M~iilq{AVG-dlWYX6qdfs`)RC%b$=j2-T!4eTVY%YU_jdc&=Bv z)H+Ow88yTNybI&x5&+)U2Vq=9`6zhnwBP6UZ}#x97HPa@Cv0w6lYplc{`~~l;SNKd zA*zE5PxkLz%*Y4LoU;*S`iTBlnWp=S5$_*|(>~b^gj%oI^4pG|SBfrkPLQGGS{!O) z<`~nnvSK%fh}L`h9urlXBRju5^#2ecN9+%6LMEB61-6eDnRf&r)e;U4<(I#bPjR=v-34f`XMb5cFE>7({-Nx(y8g{hY(o-Q=Iit`DNdR(si^{ThH#l?lqdDz> z!E0W3a3reOaZ!6ydbDdOff#8a4~FKbM?szV#!fcdTHEtwxtVDkvbAy!qKKfS9=g!V6If3TzO!bVjDfa>6CQ*{*u!wRy5C zMW=_}M`|o2f)`EYTMvBiG2WpGon?u2xNS{UT7=Tem4(B>cS<*18FOg4;w7PZLw#(~ z0W5d{{djg_W71!By%#j}U9Bs$^PK+*U6STB^2by!x<4G|>T3|$;rVtsby;xPV0lAd zvT4`}0UcS*4qeiJ`rF1b(ZT)`u3thDlRpOe;RjXml4JI+^gYm>u#@&g;yUE!`H9%4 zINcFO-%ndts&h$suDTPEZHlD(&{z&6pWHR`DW>o3(a~3)ER`&#I8zH!$6SoPO~Nro zHkLB57Np#bQ*YF-baVu52ncAQFxbcR zd`b!X@>z;8(vl_{`IMW>Rrd+M!oS6JYkEb-n_$e?a6jlhG<=5qU z(-3a_^hGr7f)W_58m87Fn(gQ8_^uENxI&)02|-!!rcFCl{iRXdD>LI{K_X!JTD}uk zW`ztHFwj-FUg*mra)Y zFSYW}dJxp7jlobqBFO}<|g?90Y^ZI3-UeXuhD@G9S}bXURIW4YqG``q~`N{K9DLeJ+1QnRrZ|g zkhho*j!MW#c%P3{u8|l6NHrOg=*xL;^Yu6iTaOgoLvPy^ze4KtUic!y8)ZWVj|UYD z#+ERuwy}En-CRlBb%j=nOgfh1@k_9_tN|%sXOv3s? zO}`7XafAIQoX5=g{>q@v!ju5HxRGSEV$zt8lqgUE^AC$gWoWMmkXT;q#T8?ceaYUG zP7Wulag|LdblAuLLdo(6$E-bRB%eZ4@tY|M56LyqpWEo8*d)c*@wMXEkMS3+ODuZ` zS@`inK+1hiF-kdQF7QoziDyht!MXjD7X%x7=d3wH?9%JJF+=ZzUCTxkb<9?C7t`h(4e%vyepdi+$p`-nsn;YvFz(Ox#A+Q|B%EY*T}Z%Je-=a72(u!Sa-JK0jo^mrt z+=7!!1@x4CyavCaFh~punwd|6~ zd87H22Bk!Pk5}D%p`@e>X{MN~b{5)vqCL(xOgZE| zh@#|WGm5R%O^gypLgKVke=;Jy#DI&#rQ!j*BPzZib0RrbQEa>-p+;5Nl(qydQ^4=4 zUVg-cQ$8v@c#A(_EJP-lZ0N_ME~{%mex-oUx2c!H^y2o{FS!omhK*m`%*H2&#SSI8 z%1Kd(8?`je!X@l%&Ko|lR@c80pHH7sn0$Wxt7n3>&Oj@aYPx~RV733WTC0n>};7#M7OWlBo-&#?r>KOT|Y~#sXPC@8!)Un)HGxV7>b{* zo;KEGT z+Mr;~X}x1(+jlusf}_z(8VubFz6-M^0w9WG+T%$`o{;GI)1M+I3I8G zE1y-kI2FRIuFF;EDX;xz;7uKIr$3)Yv=O~*kcDj{qlq?X_ioj2K+n$eun#1YcGy_$ zG$~?#5nsTaTihuXV#n^WxVg91&a`mPD8?V;KtOD7|(y^+PA>q1SkY@Q~(Lty3P-w|Cawj9`% z-I`A6lbu{_H@l|#jl)3IemjlpfKU7pLC{!Lk8P-6FNZIvU|ZW^#zLb6Q{a^xm8YvU#WI*MxvN+tm7XTy8Z=VT@@y*;6!{^u z1Rk2ld_Th-t7U30745Nn@Bsp6NF}$$M)oCMPrYDHj<1v{;>yfxy#-)dx2j|gPmy|r zuZ5@_9lhdGvcO#f`&6s~=Gua1N9HCkJ`r*FeMMa1Gp(P2GiK@t1VyF1l09xczudJp zuGV)vX01za%^90D7mp3u23GO=$L2iPgRFrkzRj~Mmde+V-cmo$gr>$dY zR59inp~<+P=Afr)t&r@_wFMYTt@ zpa4l)nOIsEr7DyjGLltjFO)7!3zdliJ<{&UhtPnB1RhYQFJIeZ#Y!jY5RZRMcyCwe z;u-GvY@8BOij^OJu}@MdR>O-&C#4RX&A4)YDZd990D?jOZ?7a2Ci49fZshJ@<5dRL zONS4jA&56;_H(9t&KOQUoaoG>0(X+>q5@v~l!hGj&@4xR9_DEsJf})lSEtW+rr zo;NQ3Nz+RE4qEt*n<+q+hu9@Q3mJkd; zX&vU4J-HkqSpgiZHmJ@m=(jKfs?s->$FQLO-#$}chj7#&?~;VVdqtRUUjs_yRB;6F zTK2|>j`f?x>q$o|3>xYr7Nrwl>|D4#da*B-)84^xL$x?>%dI3zKFGgN#e4Z$jl|L@ z6t1HF`u9P|5qWi!5{1TSSm{%(9@GaA`lX1K5VQ3M=K7r*hu@Wjaq16yQMBAwA?^$}PT^L&oqU&}YeEiK^wtG=hrNzX#>qTOUdWHM`aWemfNMtS zbT8vzCod+|nrp|9(!R^=MZ-7_!PyO4L|2sK&m{2{5v z+|zo>hVQ;3iuXcJNFQH3aY;xLq-I;y=HeeA)sc1r4j;mmza{97-`(WYYiP=HltVHd z=Vxlifm&9XPUx}anvVQWX)6^P3+ILX8Z5r&YbTz=ZU1Kjnu@Qk{{yn9kpnC&9RDxM zQl!gOL)!N9zvh^}B~*?2GbozAVXsYZ+V4?YsQLBvSiF~S3i^a!V)$yCpV57W#TBlK zDARGOejm^OMOyoDU*Sh2p`ArZ1>Y~10>cdeA@KcndrU~!^KmB7xK=+#T@*(cxBqJK zda3gD;`BXS4J38Y95yKv)0hzsI|NN%TzZ=K@K;IkG&?@u(Oj&^f4g^TNT)&Z{d_%F zxqg(xCQAa=buKJkBfc)j2lls1yA2U+90%rlCSvJ6^^ZzjiwXw2V$Ccc!cCCv>Ig&L zn!Mobw*36=3X9ok^rTfRYL&E*+?mu_8RR;v8}CaG0XZ$HLUoUPe|;|wT>(_vOwK@Y zZ|s$h=36%uPydJS<+z@{-L=UM+q2#N$g9Wr%FQ|oJde#GcFd;mK(G}IE2H9 zJi2RdENxO=Hx3~h{DG`3Z-DpbOJnEb>G?dD1|{#-Nyc1J4V47?j-6Hxf=+Yap-Gxk z12MM-u#By|zGo8UZk*fB-ly4kMe&*Kdrm~7$B zCw&o3bLdvqUG^Sg)J^=`lPdr1&hj81lRP;K{5x%J0k2uFIAyd1cXEjw6I>_PGDKwN*>rN`jcA?|9Z1p)wqkU$THn zI=&#*kGmC2gpQ4WC>%?pRu~ozWGJJpXG&4*Y^aqbNx@R2=`d_b3M1=~%0)tJgBX7S z<%C-Nxn7+d=+m&8RgbY7WU9f|4jp6O1En%q>P>-=t0v%u6v^e!L$Rd7mFs9Uwz^1? zh-K_VAS>eb1XNz&_ibHFey%InbRl>F3N9{A+ zjof<6;}fK&%~gc#+6IKAwyqC8EYcys4KsoRWG9`%Z(Bt;V)kKk z-rY7aUhUQ{Ds9eLs05_)&HA`LvGByv*kf{B+?=!vOqh%42V-vmOD)^+3n$|jgPSVB zn*08A)DhUV$q(pniz5G1Otr2a#)rP&6)qOA7^b|hScS(^sS|x0)itI9vm66zV3Fxx4`P8iY;>H7}Du=ABIg+uV83 zeXzctt?Sgic@eRGREd660qZN5m$T;iSuysqQD@!s4)djo}!8>Un4S~ zSEvskp)n_HNUL3%(A(vlMrt;*GV;_U9xt-GD0dFg%X9N=Q`42C+6pFl=V)`GvZfu} zB*h~8jb`qxZeCvjH*yKE;8fHYuyYP)GUH;!gKv%gG6ug(H#=6uY7wA)o5*x=Nx4+odgFQu%`Aln#!(KB~b}pC{2+Tw6 z<>=g5bGHYaZv4YfA z-P}QFuC;O=BliCrgzv77Ji}T)FN;3Afu!1mAfeFK%eJl7Z8*)I%AG zA}D2hL}^~Jr}7DWQWFkdc!u-*yp7K!)IfFzo!a!yEnta9OpklSxqj`6lY6s+^-+n~ z8!ZrzRTl^PgH#$iBNHk*F#`K=f+8In9u&DfccWYRd2BhwjowSee)>xhgpFL!TR*D4 z(Wux!4XBsw5rZH{Xf?6%@G-XO~v9PkWEq&uVh&kJ#2vB&Ch5)E%9OuOjm#`U_A7c;uOemfUA0| z2NfX^sOovW_J)pJNaVlPkIp#4RpvEE2$ap@_rd9F&__@j&(iF$dSX}YguBBlPuM&o+pw{f zu`dqY#wmKbQB;wYo%5E~%~MUQZrL&iB0BeQBNzaX`IsK)(o81@PN*Hu455fc3I=dt zacyIq7>QF5O7Af$h?94;?|%N{Tb8xr0EbG}W~?I*ZY8P5b@|27+HeZepT*Az4%~H> zG$Rxp5nP@tFac#Q-E z9>oiz4lARPG%I_1Na8wWgcQM0C=CXCkmKh;S<3ELW;%|I(P5(dvg$pqTPrN=XOKT7 z5_uTHo0Inze#kdiG&UAiM;Gb%QdD|HY&QcJ8wbC*=6NOgL%}e8n$>oq1B^yB}eA9Ncx36+hxIz(oj!8a35+FEo@Al z1*15F126ZAaq7D#!amw z8-&$LXPXpr)8PTJ-}YT48-qDa{92^3leSYK{4|Q%&FFU_3Yg6^6l!&b_bR#0Omb9b zOjp-)yUBJYmSBo-;g^3a|VH0T~)xc%j_DlujVkHJ(LbRald?t-qB1t zFX3EH((jl7)x1JZdpy7wCEjHlbh)2U>I*e`fNO3Fd zso7!vW2np}-pfC&WO$HLW$lI4)on(I^a&@4CPv*;{|ClU6qUuZ9meXJ32UE6Fmr(h zf1|oVcIw~K$5J52sra4*vF6;$6b`b4?KF;bg3ipY0QY=1e?4a*se!A`T%K1`eMDSv zGjmkVWVNAf)GM_~{V}i2xCELwi8Tqd%Lx{JyGCd1 zi2`g$#bS`LaE8WUusWJ(FeBOn$a^Sfwn;d{7ET|sfh#ag6;J4*#i_?4&PrQzOFnvb zmSF#(n%mi!gfNI6z!d4_moEqaS{U2ep?N1Q7ORFp@s$`tI)S7|zy zfTRs((hT4>jg?y3ST3rz6_T^y@96*|3nF^4V)Y7RtImk%3r#*po%s|h zz$*1Wx#LcJnj^AQ*87=MP{L=Ut8X*6d)h>fV*;c}3R3qcqWyi+8;e=K+wy~LucPxS zij;daRiEcSzH;o_WZUE78o8$)vKZvu3_<7dg4!;FeF&Py9Te{9cW9?G+~j|9Ko8MB zNZQxKkl;sD^91#mtyQ-$!}?u4Ek$rJpjtX?|7kPat)8-7Rg7RSxa8xasts{u_qOPr zm#6=+q@Qxz_1KLts~AmlKN(SUJVjI0Q}2`w5`r~NU)nUV zR#8;W;wrCcF7h%MGJk`%TT#5*N^X0R2HzSBicYOyk#QdAQxcC0k z?ZACgb7S+5hU>v5>g(u_i3RLf^0k6xq4tXiS~8I%M@6#Sju6N-_HV-W3z^m2y-pmIsV9BJ@q8~)y^G~ z;t{q#5~$^i`D?kf(*kyKPwX57oPn8qH`dLpX7ns%65GGlGQjH zr*a*=|6Hq*%@UB&fvVxa^waeFW!{@&Z`}2P4jw>NK*;|EVxV$r|5t}^NlD_gO{eQ- z!(fzmVpQk*#ry&DzPmNejnEp2v}d=dPR+Zcqn%P_Zk|R%skXduJd51VZ#op9p|O!# zI`mMoTyjtELK zh^NZJ8t8a9YEa$Y+1EbDp5;C^&;p2|GH_rGrtR`f`e z)L9I0w4f#PuqAWQmL+pA4;X-r@&8Rs{)2%2k85XW)s4}O?Cd8ixn{4vVFRRm^SR;M z1HJ^p$9H0~q*6Y?qMplAYwO?tE!!C!mMGu)dMw^n{ZiDMGc#NXoN zzqL0tC*$t=?(2BCLykb=H6Z>wodKth|E}kH<8FVmm2%5l>^|Aai`F{XJ~MW5mEXFox1x za-@)#G&=3?NjRdubknokdbZJZq8GRY)N-$7bE$#{>yck0D3EJ-IrTRD$NLNJo|xdc z7%HO~qo|u4@8{|ic(Ti@4~0G*E0@ag!se#gL60~X;=Iv4C<8By_N*Y6lNqqNXRLJo(sB<<>OH|A$wHN7kRTd%+#DNIs|NiCXPGm zt5{od#IgV%jaakMnUrW2br_g(>fv_DGaY|6GprC}HQrcVldPIcF#&&z%g-vBtnau# z&#lZC948i1 zNY*ypi-=c-+!-mFtwy|M%CjP{vUep4PH@OTyxzI-zJiKQ_XIzqW#9cm`>e}WBfo;lfH?Q+V&7`l-vifPl5+qs5Dwf}@lCE9 z2=?q7igQZ}K%j?zu?ZJu+mKX!6lNLvSyY5T=cpBIi)CTt?JzV~5WSm+#o8WZtsFor86et~MW=g-jT6G(gbZAeyL2c&aWv(b&Wx@4frPJ5SFhM+VOHTeS!mCb6bo)O;t+z=NDeH(;sf^j3UJQzbfC9TESOH3#ewc81w zULzd!%wX>LXXIO(@{`DCXH>R2lK1G~8N2OhVZ!wp1)YbTR8^hMIL71#)T$wu&|$g+j|6Z&gEIUbVP{QTTc)Ax*hKK zmUAr!@^Ls)PniY|y5zWtBydg4LLf8EG9koSu48F(k){dxSBI@m>CUR)vth^<@#z5* z3XY0P4IHbJ1NpG`XtwP911Hv2pA=z09h{AX?p&;t=^Yy=w%1Dle>t0r;+D1PqfCM~ z3{7nz(lkad>`MuESO?PnaSM$_>H=UI{a% zj_`G`tuhMGoHEN$nD-4i@2nh^6j;MuV$%xvQ1G`FIE`T01O;JDI|yKtUPkZ3B9@tM zXmJdDwY>ibsUktztMC~r=^tU1*BdhA;BC4p65|C;s}yHLU8l_Xs7a80i2godL{nkX z*~0KPXq@z8vuMuNM$_vJRO+mo;-nFdLXS?*1Mvb1iyVDc3i7^0HP}xETh$uu1;PCW z>cvD>rn|$q>TbE-p5p}#AosT~ht$Io*Z9!mjj-`N;}o#lib`~bce;&w z>KN##<7~|0L2;pw)o=2j({-A;!cr`gf>TVs2U;lW>_cZ3L&2LBqjMyQA{aNylI%BaoQKUfcHHLB^H`*1_DPdg5!_p z6b#4nfU0O|B>F0xe}Ps{>gDWiN53?-r6zCSkb_}&XJ=zxcbY-UUt-5{s45_~mDJHr z)B?UmjrdeBeGcP;D9y4ObOh_7@xD%W@M{1SPN*mdZk0QS_Ap9YathjxQh+8C=_#?R(W6`9?h$4YZkf@K0ZBso`pXFzFw#++}m7++}xYF{nC0AzpjD9CK&49SVrVv@0t^g;`ZS^4)wOFQ9 zveWa#EaB~u!1EZ5c&x?N2EmV43JR#j$v6Mj3D&UmYHZg<{OllGEAcwHK-eny&%5 zo}L^$K%MRqQ8kM#Uk+(&jD4qwwk5LXt0eQ&W#TU{<}EMblEzO6>}lEXv6Bt~J>)Kc3J35hlI`;HTo zS1o+QnAROh9CX4KZ7p1+9zHq%zH~CtIeN|5AVA5EOVj9bbS?*iSelSrR4rkfB0D%_ zXO10A$gti;9mj+*l~XfYo*&@#@c#qm$FX)uwBpsn;6%Clm-N@89<_@*ALr!Ah>--` zFe_nsX~IaFg*>Cdh<|f5vy_3nu#5lL??SwgZOomy zeVX#7B_lC|?$VXJLkbEJ<}QwwXIio2)MUZsyf^q582iINnYsjAx!zprCS0C3e!@wK z{Ut3i^6i?m1mCb6oC!F3>aXK~8*`Y*|7UQV2~}k|-`LcN;gWNA+?wp%RG%79<}+eE zpA>Y`WMLiTJvU!BLA)v6qV}elnR+X-X$d*k(UWYZGg+?}nHhbC6!_Z@D=pQ8b@))a z#tc&D^ygJAQ%7>p%NmhdKHw;X?~D!&H0OMnYHSRGW;mVRi~)bi7(Tna8&JkxxZEy2 zUF=1_9!pN?srjJH}gYVw81b#`(^&QvA4Rq@|J3+ zF>cO_;~Ims%zSP_$3B6q_n{Nb@cav5bGmLyC9moNv(n(}22z>sS5Er*qkE{SgmK7# zPWWEg4gHy+Hy%hf&jR*J&~xR}SV`|B&vCa{uHqCO0=y8~Wb= zBE?~ox@JGS^c)?itx)zJxqv^b)8~7w_T!a(7|VtX-hK>b^pCnYDR9BONi`|8yY?J* zB)#T-s@Xw%pT2?$CN`VjD`VWD=~b9MOE0uI+lwfkN&=E6o%9v}b*1Gafq_KcM<^ zW&BqzR6csB_2(cR$~U8^kh<`lea#OHLlMhyI4tcPbo1Jn@IRC+1StrN>y z*p5~6NdOrpvy6b1*xqal+Ye-I(!;^W-S>gG7Ka@O&4Xwig-|c8IT1c zpz!csd12zt)YQW(I_fm>$v=@OH=7Y{?^m0qo*Lfabt~>H+LMTT{3;F!tu5+nniuE07xn&*opo)b^UvU(NP?xuFH^G4XxTei6)@7(*&Ghj-Th&0g1xy8g1})!U;{@jL4Q5)g(n z`BySf?9#x}Pl`Bf>~T5yJWNUIe8YbhyG)_cKGSFNvOMJxv@gT1J9jfd$->f*j!SSx z<$lvbOMfk%Jyz-16FtZrlOt9TXiuSI9q?_+=Po5Nr>Os@ud!mo25lNftt2Q~`%s~Y z3SK1&H6$kyPG2+acw7HSxJeB}079i?7{e#ygM?Mzs8NZ7iN%|}Z-%QY1}J9ej$^RcBqL2P@h+rL;Pxl}4eHdWw%zq@^-|{Bx^bag z1d$HYvmY-pkStG^>7WAd^Rr_#h+f-j7S(4#uUgBWo+wh6lSjITmpgoOE$>GwUxq79u!NLf(Ir%cqi69G$xnxlcz?}&b&_HUc^pes`Rp{{b5~^SWPzP(73j7X|X?qJ$LeC$nnvy~X1ZFyY?WIYQm5X-HkX49wX4YV;aZ z`X)Uiig6PW8s|HofW1z&6i{1nmeU?83;w;YY6gihK}TX@Ev~hk~1+rCJB`? zME$%8Y2++)R@N|S^1quuQn9w-4Nx)|UcK1#mC`HfPN%MJk`r!y^yBUW3qU*Gy}U0; z@#8I+Ft4*RB`qeZqY&&zz`v~1bUuHq)w~_-W3I2c-Bc-D%^?xT8Ejgf{kciXMz_h>mG;H4Rj*1D7MT?z)>5{#ZtepuD8=yX>O zg7W(l$wUMZa8bDEd!$6Sj&k;$>RR%XMOywY=|G;G-?aSn;ON+QPTPlp+)k~s&Pe3z zWS3WXk3k3yDM7M3wzT6apYS}7Wr9-pBEAy;4LEK00>Zw?2aRV09IGo0ak?)er)XkE zW6PRDt$l{gd&Vs5x{U>`Bg*#fH8Lq-QNosk+c55S)A#Udg(K^fS00x3?~yue%=A>} ze`MUFRDL+#C4)i8Pbo!}YgHnfiGlW3JKt?E z4-|>J1$-Qcl*_$1V)Q*4QjOVHE_nj-r5GY-pCvQhC7h{a%&4ly;(98~Kk}O7&n;}o zd1!mmO#dsZpeem9^OKV!%X~Eoy$4TKn=)FaEyH=wMeetFMQ%X-;D}gY+SY<$GRY#g z&SE4QxdeCIr{G+BQK*Jc+*a4+@zX=v{u66UNaCcE>h@F>RIdf%NLr=)RA0&8Id;Y6 zutXTG_xYUjh3G)(k$MXcj@lv_1XgJUu(Naie|eeq|0RsLAAZ!RL%D&*Ai#V6N{7KLL1!Q&ujm9&S#+ho=-KT?<*l0Lfbfe2kuq~ zU%$8Mf!EHDw~P);H_~L)=+tQmf^W)c`r`59*Xasn1R&{-`g@TB_0DBvx^J7GsOZB~ z#7o6|gFkc53*yRaVFzd3|CsDW*I(iTcVV)P*8rF{Ii4K&Sf|K``ynBMe~C73>oQ!& zTypW#!hfn~GR+HQ-m_GK%!JWpN5oiJY4= z$a0r!4ag2{^CSwX7x8~rGRJsL*eJh{%%)9S?-BUccQZ$X8fDGG=KkiAo;}{IWKft$ z7A`MDcd%4Pb1mo1o7R3^kpg8i7HNG#9&d!>+cV@=$zak{Zph3pcv{N!g;i_pwU-${ zy}Fi?aJwd*PF@b(PQcJgzW;Nrr9s)Hq|uPA2zc;do3-os&AwT0#ytH>G#t^{^L&d@ z3txRPrKqqcF>i#pgTvj4e2R5Ww9$1(&fvO{AE8d+n5peWkMwx(zCURfp&{nBEeP$* zat9TM+#8Z-rtqm?Sv%z+JYREK4pcjs?^ddWPJZ+1yJkDwc~Z)Y{K?{aPZNo?I<`*U z1ccXDAR%T^4XlSPL(2N1NtfHe1XvJNone9{R4d2DR^Y;`-lJY^oPxLcrsvgk8Gv-T zkOwQp3NNME#Ud{bxnAlkTl|e(UF7B|0NX{j8Poa1q0>K)=>~)G&VeAq;s9dOVz3O8kR{_c64gh8v!ORti*68`91jEqo|c*5 z8n)dJT?BW_+l@G=GZ&^|azaAQ2)qo(zS73-TX?3igE%hMhNtFiW&(D6b$wx_u9QC# zU6%jx^nmXCBuXjp#TxOhUs@&W1fZk`pGIkR;}?(YoSA~(-5cwbdz`@dE1RboyRP&n zO0;hyPx%;b{n~ZpEODdHng_dRZqEqBds(UqpcPjIoqWd^nl>8L5gD(d96O=VPd9U^ z{|-+X9TMFqTj-6}#M2A;BIY$_S5h3Kt3hIew@)NZxa(j%KtiRtTTdaI1IM)z6JLb0 zTwfnGor4UVi*YRXL6Jm9rv^`7#ghrL&gq+UXfM>ppX#W5dJ94(T&nvHkp^m2eXe~8 zb5y(bVDHA((AJbZsxGw0F1OaU>1;g5vAGL)6K9gETid4*^;@>VrF`>EVG$}n11#W% zG#k>I#%_h9&?>kGX6^AT0OdGq(K1rBq+{1DDhe|SrswP$+7JgOy<~k%WPA!`Zv#=; z1}^xDV|wx4RXAv+0TvLK)~-0L#R6$9>=5G!o~Z(3m*_H^JUd<@VM*J;r)T@HgOMco zq;1ohq!z8j?o4sY-*da`nWykt@)?^AvklSWh)3$wD<>L?E}*Ef8IN**IN0;w(pN_*JU+$Ds#uw<5jQ7R$T+cz19> zLGlj@ziZ+)oU;M(Dw77kiTZeSe}Ou-K=#GFvdE9j3896*CIDQA0VQ_1KNos-VBp&C zT=Z@zjOzSH_t;4m0HugS0Y}4)P3K-&^(H#SSy`nr^+3V8vR?X}r$&MAwMuOkUyQ%I z>7V~T)#my@X0{I#EZZynD;9^_f?ZO5zRDUMp$UDM+GL`yeNOa-neT;9y;hKe)#99W zDH406Q4Sd?r)%}zTso<2ZCb2u^KY}6q{yT?m^9;HL5%so0QJkCI-&T-Y37=znC{t( zBXB$PASMm*!D)m|=BQ^y*GSru z$SVy;J5+eNz5e#p&Foqb*$)B&W*&ig3J6Jo8H#~b7Lvs|`BjOP0~&dlj8X+HcfI9m z$(-W^xypN6;N3uZZ0-kKS-VfC*!&G&b0DY}8Pvht)2vT58E*?}69J(_Km$d3cAGY1 z%DgzobAQ12m~bf5?Nstb5Jjq;r$=TN5)wbtXmd7T!A z*O`i%Laq_!K@Iui^ON-UIJc=ksr%JeT571f&_WdN07mUy&{LcKwG{!90yqH|waVHS zmZ$4AEsKITJa}|d-Ey8^mTie45^-CTFd2zMJ;DTw&WRjROJm)~QNNh@$(64vZMVhd zTAct%bzm#D@Ww>>kWwIyvKOS`5y85yKH8zl`zW+-v;S1v&4?+$E6fMETG0nS%D*!f z*cfmJw1^f3Qj_gBu-kfIQhUxxe6<~Q+}`YAsIdDv9?`@lou=K|o5)L=v@UXoq(+`Y z>oPQ$MS;)AVx|ZxZ{Z^4RQfxfBq%pSpD1q#@hB0D|Rk8emumg(*LX7DR(Ip-Ol## zL42#aobl!%SB_QItyIDYZ&H5Da2;5%;owD;VywSqF?Lt}NcLb9*y117z3yKj zBVZ9_cQ}p%$EKI~7Z%u>R_igiT0aY?!knu*m7@Aqgj3|9Y(yzUwx$hJc_85Mm?0Ym zoPQko1WOl0sy*WL%q3>m%;es@7<`U_S@ejkX@tzmLk>-{bvk*oE0t`A_B87V)gqNt zi?YEDCY2p@jbkXV$ZVX>W)_v`9VVx-#wCU!8X7Z%+L$IdhWrT{`>y&OUrlMm0 zxX}={tS91K=Qz~3ESeav%TN>#PwBhpo)d_>oA^32?Xy~kT_b`rIOhN6&KtFLL9x-ury0@kW48uWd zY5v@H59bbdhJ*zkwS!a)U>~!j8BRCoFY&9BG+xlza4u*evi*qBAk1SdzYesIUoUUz zc2Di!xDA9B&d=ZP`FRRST!yqPhTk9XgR`#H z0x+s4K6@bcO_>E#E!7au+;)7Q#mvtCpyLdcib&Wv-x}*Zxtp;84N+YPEzIXNIjO*Z zE{Cj2WA4=37cjII_-C+o1%T=QL*FO=JLKAOAy2)c%QZt>HjO)!v4NsQiH5dfBgrak z6B`A^yu7zwP?m&?0*26Ej$W5g?1^xOOl`iEG4 z?YfC`#KYq)!`lwp?+R6PmiRZvc}+5h9{%sgEj_x?x76)(J+iAfP$Drd?@osf8r-Yz z>x0C)Hr-4PoplYG? zm+_S;Wa`%%$iWM=bH*;aIZ=?w#wU1D69YU=t|M{;Hrvh3(HoHv&TaR0S{V%IO_D&a z<}sx~ptKeyjPrm0X2EY}J6m9=2MKhFO@8Kw#ipg+Ktk?p4adt)l9L`b7IZgpY$cV@ zqRH{OM=|hGQ}PB2Gv~CS){OGxsmR9?Ua>fYRWEfGSVefB+f4sqM)*uKTFY*?j6|Cihqwm}wHga)kFgF9T>z}jl;Bt`;#=QGP7 z{ym|kvf+b~?s*^lG!aQjKgIeHh7aO~iW-95T8zMBzn}~Xb{Q~tCrvapW2V`N73g%g5# z8D9TJ*++PIu_%ut;dl+Rka^sB7-FiQyvq|#t>9;1a(H8DdC@da=)sq0>3V8`;(n^& zo77s|RiCK~eeOHZ)KN@pt;-wpCK~J_8gAv(cfn|b5PGaQ*sA=zL*FnHyJ@0=^am^_ zziDAN&LRnmXqHLb^lctBx;NTJ9O=e5eU2y!8yvCH_)vTF zsf*!C1+MJNlksDk-y90A-#BYiig^$y78VH2FGjZ&CdN<%pWHIl&yFM8w>32#({2ui zG0v;PSbd0XU;*XA)1b`vZ9@v`^S2E*^{u>v-l7Y=68NH>)~t?{dRfoBzL;;WNfw|I zk2g2AZ-W5M<15yPWCx#2-G?#E>;{N}!l#cZc5|b8fceBXJyFVugT%k1RgOiXe{8O` z-H13jS})bJtD@~!+diP!0l{J`!>LOFava#{7GGu99Z3XBKk^Iak8kcFx_O1B+ZoZC zP)G(7%3!0=f}gTxkytYy4qgrA9iwxFS98xP149A7Qm0Eg4(iBJ9)x7{EK!!iIw9p6 z1aAuLi{Rb%?q1F zZV&+5*cU&?-pBgzc>=vln&9*mc1~xPl^-c|QQCCXkgbA9)I#2}Ne|+vO;`i;``<6k zFt<;1*dA?_BKP{s|D1Ko)GI7(ToO%L5=3U_=0-jeboQ8ZQ;C^G6k!F>;^-cd8$-1+9#Go5(>qDEvv9a}B zsycJZqnfi6ayvoI8o=MSX0}sX`7La)X4-Siv2m7vXIIIpuW|9L%xAAoGXIr61VsQg zQY?BPIRZgP-U#G4-};6$AeY?3%l7l!vPLG!17}QbedebW%8NP3n7jUd)mS!fIV_y! z3+Audxz@SkJe0=E8=cUecP#*L@Og!c=ttoF!aaS ziezr^SK9wC0Chl$zx`NS#rK&JnN>uLab(JA&VOx{gSF*UeJrm9i!650|<24Ev7LDUqm>WWjI|`n9e#S8V`~($&^{gaMG1&u>b|I8GMxFyvhD$)Z{JE;B21*gs9_ z^wPZwRiBgbhqd}chU6|~fntm|oI2z98!I>JN|_4`%xq8;SJ8l&S8FT0M=7UcVWBk}EZWr}+@u-wD3SzKfB^=j1fE5R$TN#?aonUP~ibLSu`{j3j zB-wO}D42vnb)Hj*C}k#PDT(GgzUj4HKmQWufn157iWP~&U+?yRnbIn4i+3<)bbBey zDpVsc+DRF~V@NWl&MjiE9Q?^&Wq-nQ^vmLlgk|MnOM#~zFu<*;^hlIERKPNXCHRr3 zz_N-gxbr@jT5-&8Fg7j9;=nu47u3ycls3g~jUz9NfsoNoV$99F_Jjm!Yij!#5D*zc zy<)+QIGslyCwAw=+>Wt`rzfmI&N{@s@9zi2oMshYnmEsUD^B?BY0-{eDSy@r!?aMu zIG@h96*V6rp1I9PY+M+pS{t1XVIcJsJtlAMXs5jjp%}*rfs_W;3O~!${1be3!WoT` z-De=T$LVeUf;-p1S0F=`^3F`$DJBLU(ea7Zw#3EV`d0;9`m{ire+#X(=om1d0%RC2 zDn_Ng8M!dbVV4WdieS=4=YL7@0Xo=_6y&4jb{nD+)Lnv+ya<~$W|NX*n%D}hv#?vk zB#^bFT(CRU8fy_f$hy-9&p6ybffs$sqzlitLodU$zM1R;IE+%a*RdvR-bP=b{CJaJ zG36CEwT~jLNN37y)ON?ov^KL@YO6#!^F+DsqruS@V4&$Vqp|E_cYnVrePy*%iY890 zPLkne1?fbOm_=fed=ct<1ZBml=&vrg7V(7Cb?F#qt+C+fvictFPJ&1w+wLNjh%RY&AZe|Qd>_*sa0RVJM8X8vp@G$G z)q3N@F=@S+t!9wTZEvzya#vkp?dn$b{?TpBO|`Llm5J7AxPQTme<-%&L={5!aYT7O z1v>ZHM;pgx7SqzLFBSFj+M0A0-ruUw?gRGxJE_F(6R|RJRgStMeT*WE;oHghpmx!V zt!$+zL%3eJLQ|Atr8lz1YSKfILE$l!CgyNP?v0Xi{60QpO&i^rxeKHQWIvAooEyH- zoXY80wWI%|Lw||Jidi071R0u&9~PFYQ%g7Bd`))P|1m0L>-^4)GQwmD5?V}Fd+U}G z{}Cwc)`^`2o&lqnPCJ+?6Q=*T|I=}Qb35%@horpm|S%Ag^UrkE^2sU=S?BY zCLpQx3muPX_@U2>R%eZi1ck>y5oNCKq>unxXR{g`pCt}Ao%Ffx??V2yx9n23!<@rF>VSNlPF+Zg<{S` z@S0{HdA$J3p?_TF9IFqUPNI};^%R_+iMN*+R~RXm#}Ul8P#v{_Mb|UJtk%V^v5pz~0|q zm49;$l)cxGe9kF^W+km)QBK(eaXp?$)1#M$+Vja#)vti7K?eBV9=r(oX1N@$cN9cf zw^pLNotAWR!`v=OVG~rg!C{?QEQ}c)=Z2$MlAGQvtExiQaW|PiA72E0thq@lzFD}< zg>#K{*@SeDZYd``%Rqz1ia+f{eq*|SfPZlgY(ThrcG(;fN_^=@++!+l?HQnwVZ}3H z`tl?tZMzdQC+zIz0x*D4(T1GPvVvVy*r5)36HHa;Q}^3}t@w%BR(CL*1t)7{&0;Y? zY}!;X>!rbdsjTNYDKo;FxAhTZ*t0A^Rv4lUJY=mCFI%YGR@J4kx7}^NJr@P1On>lU z28mtHD|__w8SFuTIPf{HS)t)ONRXae_hQAcX6;HWm5hzYYMZqLL%HNqoC@@OQHjsO-4mUi*LO!v{a-RyPz*3v(qQ16-G8!XK7{en3YTwZf$83E@+c|>dfLs#P4 z6?P$G{Z43@N9W7zp4gdsF8ckDWq)i#yz)mSO4R=KiZ!%s<4As)Z{HseM=qFwyVHMJ z!1-Z5+70|z{@Y~?DXc91JVp6Ez=X82bZc{#LZ=a4=GN93q`8iPiNA?t;C&BF!OpVMT3KWOt_CDuYJ16=c|2{_F_5N@b zC!hDWVH{0Q_RFAbT(ltVm#i?9Y%E(}s?ypRsbCaj69?}-14>}2XjxQS{VJ3D^ybBt z^+IdQkr?Fk+)(aQ52^5J&>mo2 zEb{-GlthstC6%u`4~@pPt29y^4hQ+;zaRF)od4`cf{zr-k#kAmo)J$zVm6jMyjAK#zn^UwM{-VTPuU3*3UjuOJB!<`pw_Q`#PfoYE!=jTSt)X>e%?Su7KN$ zr*-U<@q1l!oVIciVUv^U1t^>`seq*wZn5q@CE4b`kLF}!5IDge7;Qv*G(fgGKrb$QK#OU zj+PFz5)4NHN4vn$e55x!+BjSKzIexiuAcUKuVdIs`Ea)Qajv5n+jjl~|A(`CbhTCJ ztyX?uEfNn~me1JqWo%WUmCGzRsrba3f-<>Rj%$iI2gt_*6ol-cG5#s!rU&R`o$s?FQH%IhM5&lFTD*Awt$!Ou7vBQ2C~psUV>a~ z4rG0^fJT!>p!2{UUVu7c)5=ud+R4_s9J@$#b$g*YT{?2IFjObu#l|-k26*Ce=E7-d zlEXZ^TFumZ1~yj^A-s+%QNia8M55zkNHl>uxyrWB zhgxYvnl8j61s~>vb_n1CF+_9h9v5tr{Z8xV@UF3E*0{viS=v|9ZBFDQBDta_YFo%Zx456Sc z(H$L>-V$z9YWKzGL`_XgVQ-BgtlWae0Acc0w_yrgf)y09}g zVM_BqyVlr$pnpUmh(#)uuFi~hZTZPKWL^?z{4yxn-6vy(e#X{aQ_+Mk;v+cim4?6- z@S&HhZGWvpNZxo?5+H~#3Fy+3Ti4N3(^XYW8gXooZXSI|Mkkq2hDvHe`idPx+(jos5{606$i8?;zskmK zCe{;bWpXdlBun3|9Sm8|hN#*QQJ@+Qnc(0YC`dBS(Y`I%eZ0b5_t?q|?>!7&5@c)` zX7A4@8Gr95B<++;m7RHV!Uk~AId^Vt&aF~?-MPH**N1RYXYHLUP0BXr$zG_r!kHb^ z=POv84hIEl{P6q~llMdwI~hwOLp%5lTCn^kd@1ycniGGL2KPb9n^I=v-Udl9Ph^*g zq*SZzOrY47bik<`(*XH-7+;81NGfqL8*LB?qkr16MZ$WYuktR@3TI%sxV+)L>6lt7&{rKgmHX46G<1eDrBbgbd^al7iQcnH7BpAMryL3b0_A=nf zS;sJu43p(*MUCsc~<6_6N1DIaGO`y)^Q zaRaR{fq`^yR8%qVa@^+2HJq(nxroPMe}A<-LhIDDZ6K5-Rc*G^CC;T#iL_L>jIUgG zDW-kld5%cYOe5dD%l3Yq9;H}*hp>Qu6>qi`3@NeO+`+Y^UU8ZOibGX(){rBV5diH^ zGCj#ds$whGeTHJqc-&hw+4Ljqh1^5V>&FV0G1E9}ZP`fPa=t zpZ23H{W!tx4ws-Z5O9qBWZ9_zxetarcNKi!pWBb~e=j!iQ-DgLuVlVRudB#$vQxIT zQ6LkwqDQ*am_)#=oRL_jjsOzxBs+%;+7rV6jjDI;Hgk9~>TsTv$Hp=^HuS8El&j~U z?pO(B1IG%R6i)_K1v02!pwYkxw0~9C5;xALa5i%T6c?_s;-+GkFtS6UU-u@K6`zQi*!dLjb^*h(jVdpI{P+@SaVnA0Fv;D+{%x`86_ z^i=s8ID?3%Z9-Ig345DEPGpo}hM1X4wz`qSC+{V^^Xedp8XyjAiwsE19Dj#7>xt`H zy-Fi-Zg;`kobQw!Vau8WcsiX zH)f9eicpB$P29W}aE^phMYa%@?hGBBlMD8xEo5k_4vt8+&tOHV1w4nbar+OgX*q1| zi$o%zp&Rynoua03@|N@yYJa}2L4f;)mn7S5^72dYPv+&{s{yLGF>XNZHr43Y{QTKH z-?W^N62v#F;B_oQ$ZUse*6NM5?uFA`5{r(oQ`VJ1n*fq*O^7+9^}M!0-5q9B2?dK4 zFk`P6X`nX?a^XMd^})VKsP0#cZHQc zk*@bC&E+RTcT<8FTYvqMqqVJP=Sm(LQWV!BQMbjlXfT3Z&NC6m;S9VT%Jw|Q+&gvV zQrFB&T?!cv{BYsq2X<+2t(_$A!KQyh!O0TaaL4h#_nW zdS~Z1N#_?}2};O1I9UoxG9>HzWL?gG#o;Cn1)CYZvxMP+xUgV zP*I?|v3S2E*?+H19+^AoRXcUBWH#Nvm>O8$k0AU&=EQZCCv}W@E=?R_$YXBeb~^U0 zFXBjbIpW4!!uQt_GJ-re9PY~ygN`;4HY3$h8qh5)YS_9_J7C3fNNHDCb^@$=1xiK>1x`y zP^ojPYd2Nr5*J{~T@pavK?5d@SA>#o+PDuA#vPSk05+@(D~vJa1=5rW#GNAyF^;f% zH@C1>om-o|$f`a@8O}~K^rG#pfehzh-pabg7Z(cp>Mm57M8cRpjkDNee~}$;vK`Z^ z`g&VZI)7d|%8mOosk7Gm9UD#xEAi+r@-i31|FkhyKvDb>1L%U+Rw6=+BuI4HaA&y) z7?Rx|OiBj?p93Gv?6s;AwUz*TefxmQlk$xAAJl(W8sefXXt*-La+Vfb#|nwAol+bQ z8JqhELOrYfz~l$@_f3$KnT>jM@5kdwfi%%Rp?{G#uB;`rB0{MmUrhkYRWVF|b%|4J za#+@EV0)F6ZSU#6n}6sO3$}k97*cbT^=5C=PZs7ozi=^@*sRF#3+7@;PrK3)?C({Z z*L7vxPA|b|>?^|J`+#>OTPaG~^B9|vLM|fl5>*_zueHpP7<{t$>9hq_TysV?FL~XI zp?{@99Z$CrJtrzWpio)}JG>MB#+GF4ITY{^Y~GVHyf~$?*CY3={U`ejQUMaepWMY8Av?^9TAEX8!!4narM zk(+Vp;&SmL`1kPl>trpi^ObhTsZ~fWz-=43VIpy@E3K2`cVUU&g^`2m>@JSI+T7d< zo2Np-eG!Zk43S4R6h%1;#^5TB&Ani3$@NP`l_m=~INq`;=(UIZ{J|ERM!oa{%YQ4~ z-R#@{SBZfG)AjVeuFXXpS(t~dc2~sk&92NFY22Rl`^Y2`{iK;4+ZMPE|3!`ZC{i0Z znf2nPPz^nzIRxjNOuf(be2#B|zP{e7s#2Z34OzKLjLR^>SF7T{@N}Sh3{Lkud->}I zOQwad$Q-B3Z!LuYX@%f&V;wEcWBv$6>RL6#2Jz4SX6AM9O9O zaWam26~s?<@Cb4IqKlhq7Fsmf|Lh!k#};kU`rnElif|n#wjPv;ZtzVqBUTMr#1qw|}5IY5jDi)Y~JUZt}Py>%&D1=SXn=8|^mbF}Q$Vzjeb) z57!P>zX4a%00(4~+|Ei#4{V^m!C2Goy*VC1@{hqX|2q&nXS z6K<6EDbx*|IHZ|Yw%i|BSD{HP>T6MbP6~jHjNyCspIq7-Nwi{`PU$5G#4ak&6lL8l zOZP6I9v8LW`ufm`@P9zF>}4%7QvWAwp4Av;UF#zwh}yjI^*!<>y?K4z)di|^(?6Vc zrA6iVYMZS@Dn0aX2vCaS7yJYo>b+b}7`M*n8M?m~tRqWF3D_`Sc#!eT<%dIB>j6N- zS%)Cg9qk*Zt40lFJ5m61>P>fAj}w&~CKb-!*E(`n2!Z_lPX_-3+@>2%lM!zdw~A{6 zIROENx5R7%KLG-aS+@ml1IGyg+_%kl11tmtDLdq0lM!zdw-9^-Hv)ea=0kDTEit)* zZAO%!ZTFiuO)cNNeWi7A7cb7hV$s;N)t;s~3^TH!<0PvNHF>gfD)MALk%1DvuerCl z(fa#pwi4!Wv84|ji}|H=i_8P{7u3u#f23P+Nikpnw1i8!<(#a)&rFO?xHFEqwrWfJ zJ?Ypw#J$c`?o}N^4@!T}(Cz6NHj!yk$ksztVj#8FWSL96&9W12V-DRQz6(y&nG38} zndnCDsglui89&YA!vKIH6}-B7_D3_SH?J%A< z4CFWr*(TbcCLw7^Mk8c^+8&uLX{%dJ?HrFRve~vUi~J6zg}t9@f$L`04FKqO*&=FD z5(xy+=p?Hdf}#jt6Ke*G5RN2{hWm`fjSZY6E>EWENHxBM81ilhKk&I>wEHwZTkjXJ zF()O4VV%VAN>_gX&*?J9 z!alYme8{tDlsL1kz$Y2Th`YSI*u^DdcJ`2WH~IZ?O52oGDXsL^D2DIM`Ha#+Iq!n% z2uf+$EHHnr3wc0nK$-|@rvp(^)|3+`vWslcyCgO8mUtACjp>RhbJ32pD4LEvk;*diG;GrtCFnEB zRqwlb&BCajb8*L$47WG?I{PmbRAjyOo&1Lwx}kq^L)zq+gck}v24Q)J0AdP)MCXSK zi|XJs0cc8476_PrDg^KWSUqa(*Tn;;Jf{VVjy!S7zpO<+@;&M^4qKf92bY%=YV1pa zgTz2wZZ=$&Z`2_?go16Fyb8M)3T|!Ob@oBqCR$Ck2nD#0m37JNSm3rVHx=GBU9Rk7 zKu&+V)wCE=U%3hst>j*dJeBFcin}HKkHE6r4JtHH=T?=Ep7bynwNo&~TdG0>BU1Mi z2zpE!ByK3~dF%elRy|3JJ_YEk$o99_If8{M{O*`&w8>IWIxa~~<&nG;`5@ch@Bu3U zzy+KS{!EbXB}SvMHOpPQABp29RpMl0@h*Q)8u~}!Ei{fOox44WOS;VA!LM7*@HKZOt4^ zFBV7hj@^tCwJC%31W$my_&HmPlo;g3L~74|A`pGTrjrqFBlLvkqZ3AS1xhjN|Gs}B zinZFFtz&kL7VPxW+ScDy?rhO&B2YB)P6BN7T~=@^w!u_G2VPJ5FYHxs*MCSa$^@Xa zil>l(yo)UoI6R2SD_>=c1oX}iA%VAx&)&vgj6R5OG0_JGfgW%@r}T)vNR3UXmEKJ; zqo|XU0JEWQ)L+Y1_b+hrFLnb2?e2d#(jx*7EH*miuXE7Hwy6-_6g=jBzB-Mq@<-pf z^L5G{vD%4crW25%%sfk+I_k*k15OtQVI_tqG^8H@pwuPuL=Pg&Mj$<&0gI1&m8; zNe0rU(Hjvoco(7@g92}X_XdC7BdNO&;>j*$44;0r0MS(k_y>za0rT~|G8go*05&RZ z?v)QMjvA7%He~G!ytR~7%@Zlx1?$$v#mDLyYX4yj*5k7R1Y&3Vj0ayI)l!E|w$P9^ z(BP7jI2Lh081_S_UaQAPN$A1d+^bIc^3u#G5D74C*iFy*A@f>Uo_T*BAW8}s*D=g7 zG=#-ilk)$~xf&eX(^eV>G$@kEht)vFh?b1 zRuWVQ@Y51{=}QUQ7q$u__#%%@@pavDS&45#Gq-X%?zq2u|AM>e+=bKk&8$8Y=h(!_K9to$V`GSrl&sG63EjhHx9RK=JCP_=|cfPfg~ z1ZX!MznHSOuwO?>Egei(Dq=I2GeoCH)L<8Yc(lZlrSRtG0bUE(eLb%0AGxmX;}b=W#ewqT-EHJ-4kyUy*hlaFuOhWQbL$>Mlk~o-f_!OE39IJB z55)C|@BVqq4#oC&W(PVYcuo~uT+qb;N)<5geg$IbN=ScVIkf)m1PAOKIwjkVYTh|6 zHl7QapyyH@gbJkFexeG{Nq!OM2{p&+I@g@h7&y(L`X#7B{-ZzFbCG)m?be;37(4{y zK3z@uJ$qFDqv@^qlR|NSxG1+sa?INOY>CUBD824i=Tob*70zBfH-3-!`A#i!>a&@? zWH9obkGg-kjjZ*QOWLWA;JaSM2(&&+B4ry5lxQ8zY0Ljw+#)?Co_ZsW>6uA2jth#P z&E7J9rjQqxF+F4icU@)jcb}&xN~`vb)P0m;5#Uj4RH(X{%p=PKyLb!H{%Rj8WL*xBow&f()Om%D zqR`s_eJ?giry^fkG0v%MSgf&ej(|0GRd?0aU|jm}OCZVj^estGay{0RAzir|byWD- zWDkFGvPZOL(#>{n&*|y~h3B2FfVdc_e=5H0_xQ4Nl3}gK-SP%WrQgF8C{4)AA>fi| zlygaIm(x2xym_U@d?lWMYI>z`?!^-BU>VnCMT$P$)4E&rb6TA7LWJZlTpCarpsOOb;6&!}F= zi^2FpyE2MVX0^m>q*ZphDb#oNag~j(H$Ly*7|tU_*%a{NX3Vpf%I;d&bOi8n&Q zVzfCwlmjxZ7a<#>cs&wD`AuIq;ZjVvu{9#RSga97UBxBx`p>`o4^$(#lamo|6St;` z18xC-9jwnA<|)^!CMw)3_}1$qv{S9`75Z}9$@?hh#*43WBz@AB?d8M+^6RwO`Dv0f zqBRvh_cr=0o7Of9Xbk)zFIiT76SuYJF`n+PnZH#UR=cnfdhlgJh_Q@a2rTod@OS%! zD08wb-N(_G4NM)bh+hP* ztq-;PODL`qTK4E!*F^!8N%!;GTT6(sa+z|(rR z{ZfEe)hgqX`gc}j#}0QxyXZXO+c41nzP4xGVD~j!*vnhUnb9n_abC~IJP~yt;sllH zgf2|hzA-BhJGBg}R9xP$o-x6^IkSpPhkjVti2G7q*O4Ikv5p(Hdq?n7HLaz$5RC&s z3Iz0#(^ZoZZxgp?mjj9ce+#oV+N!q6*o>OoU*XKFfon%9=p7E}(?>9Ba;SdlY3oNV zL}%2{D?2E8F(srt1`bv2V-=YS{f`zAVjsfb5bh@Ljz3m9z&DNb!=c;sfpi@IZfks@ z=K*L0e1;>qs){btuZzF|#^9{*jc`SBUlFus=dgvdIA>ZbHq@r@Wma}-O*+Y=`0i&& z`(ztE>*#11eD@G2u5HDJva=>dI8Z_e^~#naO6b%qW({wCs*5qR81I*CTm;c5>>0Mz z&qB4^`cL#;7Ubv0e_Jz6Z)`IjAoale{nZ2A^*{TD!%)>{Ha_Mh)>q1E<57}qe7Nb2 zw-szT?lw96?bXd;0bhsZ7mZmQs0;aUwUxo+1VWnox2J4j=q#xPyIqp*KSd?T z?QTl7qH8Re*ARBd!BFKu)&OAOI- zbYLGN8|{ATt@UoYI*Unhu;j>2Et-F@6tCNEl7d`zcas&Wi( zE|Nd8WN=H2xsD-Z0$9)NwHA_4Q@IdVdNIv%2KJ<3qF@c@2f+Hd?c8W(%9=e<=_ag zcI{9BoGborzV4mD-{VLG&fZOeqdlS?a1!?#CUP2L4nTULMbWc_9>Y0?@qM!Ot{A)M zC@&T-)-dls7@p28u57DxUUEfd`9P%+FEligRj6_-e|k%`@z-?o1=aR>ai*GDuq3hIdd}WK5EC;GXavwCNki3EF5MdaBAvcuQ1oeXlOESL{8TN?)}+MqbIY zOu`5Hs`+TS%E~-cYLCi^V>ac7xQ+;|zs_4Ve{$?^!x5moV%tyx8j#bX$ zz`MiVGo2X)txbfdv*iPJO(L7cS%~F6=f2R7UxffuQZTB4+VhTqWV_1cd)D%*fgmWd zl$bQbAV{=j5R_rPWDrmnIvAUhHsT2s`@uTkR;BQf?Yk(gCx!+z_ke9+Y6B`P9kz+O ze+~AC*O}p=?p;SQ}+B-HMx0UBnxajEBoZ!?1gD z-SLfqlE4QyTC>xj0QVtN+Rb5HS!`8qe+AT>WUWYX4HC4Or|<(FbB3u7bb_YHsw&WE(_%;E)6AAyk4`Fy2EQ+`wa-T!KZ)_Y@B*CfPL20 zAkpGWv=d&~;@DlwN||BfSE6*7baF7V=p}PhygKz=^Xe`uQuLllr%q7?WLfJve>rE! z=i7#QF$Rj!i&L54BE1201&3{pu$fJokTB)rE53S!GCvrZ-Y;Dx?eRuf9lnRz8OuDf zWNN}7<(O4(OVj*H&j|H;U`!u;k_OqlQh41EMzf~`GhP`U2rDE3R%lqW^IR$z{C$|p z97*=aDG3d8G`R!d_k?`k0eQ1Te`g>+=YV|LazH-sIiU7r(*e9ngwR z`PL5TB2I@LkW;dH?)!`bayHesOnw6YVGHXxz%Y{1tdz}>Q1%GiE~s`5^jqC-Fvlcs zg?W0Iwpv!}q0N+I^f^xx_}TUPBCJf*NFeq4l=ew&60GmMM{G~`m_qSAe=U73L$OuO zt4m^^;5N&OIJTQS#~pUM12IP)sOh5RTAUu6b1kAzxy613bv)KfGM5EG0X}wUnwpk^ z*u5Haf5Uzu_e}K07xh5YhF|pxV2>aiao2PD0WvL4t6K6UQ?5^ z^g-r(A)z$mFlE?*`B}LdX>X^Y5Djo%%~|_Ud#M3zA%fQRB3X@nzq5-+Rrmb^2^`$~ zj=}WA{sn{SG}4j5w9&qk!RTm41Xr)8JYTPNIWML7q{~@cuI}*^q#B@`Z$a9$s~~g`WOREOOCk|UC!qwuQU^e{&INrkJfbITH9es- z5(+$)^kJNTx+&LMwgUytVMle0uW6azUn0iomaM3KeZjTV0_au$mC28Dq znb4c_uo!FOUCMeJeeFPuA0zH^WKz@FOss-md@2XTj>@aPf04m(f68?v=-~B>D6Jxk z#p2;0bp~KQL&1T>TU|||DpG|NJ(J_*9EJ2r^P$C6+sqsh1sNS|8nJroOs<7a zPI~LN%J4?;7>&DE8KI|aQvEGA%yAvSXa>v(jSvv^{Mc;!3=FN5yU2VIVfJ;&8h1Vn zCQ*UTIrTcUf7-N^9K2KuhXZV9iT2JO$;lGX2Rie?>5`Nd8z9-a9N*H2FPG!1h*mIZ zHj}r??~Oq4FBMFD2}{-Z=#&NU$&wZ9a#`swSK}t;(wlA-)V}lzz4VS&oV&j&L?21- z>7|P*D`=fyM>??pG)gOCcQx?-2zv<+>VtpjIAd03f5d73Q0N;^MVAA6A5b)((RBiD z;Swv`SmvuHBp;f?h*?FB7L)Rfpl!uF?Ak}oS&Q?*?pD*KWT=aZx>UJu5q`-<%O<7$ za(x|jz3P0kh>@r7mJwg9s|RNnmkUIFHtez$7a$j<^H29XRf5b+*5mf}}iX{EUEDFyls<_e>KLpjk`S#Htj~aRA7z!T_W%eC5xM`}sgF4H* zNi!Amh@;898HC;{^&cr-q*e1qdKJ|W41c5ru@GzcaSbWqa@jDRc((a-6 zC+P5Aq!zgMb*yJAlys2b0PLN=B_($efH(MZf9A^1gYbUJsAH_*KWVHdXJz<=VY}*y z4LO=ohHs^i#QWFi?RL85rn2xW7dr5Xuwt`yP~Yk92oK}wz3uR$k+$o?*c9v5lsCVU zCiY~QJtOX~*vaWWwx-qdoWy&hXaokJ%B;qz0;&usn)?as9Rgvl26js>b7~nu2i5sC zfB#Q%&*>A}{achOj-t3w)lpAh{{T3iMNFVQa^t7b2_ee*1dZziFG zw$u5j%4fG1%AkS(*2KD=-sg9wFA0o_iPX7~=4n2_%wa?N48#kCKhET&YxOk>c+Lm; z)Ef1X(FBZH$FZJb98cLVQmU{ z+SOcbcN0gl{=UCrKF0SRG}B%Ex@S+$^$P|94s0AqHXJ?(BS0J5a%4Fn`|Iyh-8E86 znh`P);N>8zW_r52-n!SMgh4G@Ttge*xuCl&T4x?L4<-M(3Q|TtS3d3t{BRXqL z*WiL!QRDc8t;iY-)9}>FfCwOno4}sO%!auk$1{Qmm`b_IC_*PpWpru^Ab1$a1OjI? zI)Va$CSx$32H%F6Ku_Z>L?VnF(XL#mEJ0R94P*c-f66N`&1bx>9AO^i4G6-;(1S}C zYcx4#1Cl$((ttVVH4_9P;^Kf_0h&_5Gl4S^BS&Txdi6@-;^;v*3>cww9z!=EOd?d} zF$V|&;kkPRfl)QM#0*UMDnR=%CyZq=yoORkwCH7g6%CFw3>PiBYn+YFaFYW8yQmPW z7#%mUe~}yZ4af{jLI2tlQ~0aBNffSQE*c&9T(S}yH}G{P!l&RgFe*%?VRsl4BJ05C zVt|jqP6}#=WWll>!9Qs>TN7Y*oQG1`7hye6k*S6)2R&$?tO|As9;>v&Ut;5F2RjmL z9gN}}aSso#A)-X_Iqn-EB^hpwkdW90q~NN=e{Kyn;EL>b#EwMCwXI?I1XplP^o7Ng zVb>sbaAPq(d%rRf(h#IqR#~_k%%?2dj-p{Ijf^?0ZMc;e3okXpZUuLXptv=_{S7+= zD4b($*pM6=bHK20Tb9x_u()9Rb>LZhvw^8JW50h`v(>RS}cd8s2e`#n2LOk7X^V|tNEP4+Q9Mz@f@_S z5oZD@jG_p>=D?p;Y)AG%SQolNL;!^~Upd3q1+1~O&cFh$!_owmA|;1HMI3Or6clNX=g8IHKyLtH6B?xOBX75e*bZwoGy8f8r>h zE$a{A1{8R<2v9r);!tnZ;mxY4Lr=jVxST)j;X*g z7(Iw@7Y^D4m8=k%LcnwlU&j&Ee_?cu)y-V-bsB5hK{ueFMaTRR#}~(zA&w)C;}a1) ziS+;~u!g`rLI(F>URb+M^Be~E1DkrV&)d|)LWY0*qyF*Ft5Lmwc{!QYr#G)>eDq{| z`D=YJnZ6lKU-BKfc~w8GAJvCAfLp(+_tEXNe*VKF{(N&W9G}l7e}6Zge@yZDzjx#F z(e-RFo$P)4ZT-XOZ1xhS9|XR^YCGUNXrl+3{n9i8tHP^qzx~@T*mysiU0wgZuFr?B z2kSgQ-vdS!Ee~D|*e#Y=IG$Y2_zPhOv;9yXLywujttt~7p)-Z}ldi%9&!&^J)6wi@ z{p{$t{&DnSRzJEJz8f9pf3Ji5^{8)>`PeDk2+6DSuZ;d#2JK)O{ucWu0tr>`dd_0> z#qCa33lk4D#L)A7}8GEIly4=+ab;Xl9o`DFipk4FC<%0LUcb68HbPFYx`Ze`&tXhXZ`a`Iyoo)90TX zG(LAxuNN+g9BH^XzJ!s*~`<(o{5e;*+U&h+Gly+c-p&+{N)=6T-#dG&sDS@}HK z7*R42=Lo2>gnW=Co@|zB<3Tu0o#uw<_e>r%1&Uf$);2z8-(5e`22e`h0i|<>u+HuZCx%D(0zS zxIE+R{d9!0dA`{n69lP5-oPe__`QmGCXAYA`jhkcMEz$po!~&V5rp&awH0g^G{yX} zpabP=$>qGAxH6P5cF!k8iGr3-JI5GOPi+LUesf>bJtwurD^hLprK{p46ZKI!4+ebfepWT=Ie2ti9 zv7(BlG>;s4y%;MiSr$`X6UlRVZL-VfHITwViVtZ8S(n2}Ue!)XJc+&@*01Zc`b~`< z^}G5#LIaY-`kX^Dmc3F`|5i`y>v~q-VDTv3f7Bo9KZ{_^Mat!9&85!m!u8?f)03Z` zfAw&UH7DWv>S{XqkW~Z3*2V#}_ae%53Fbc9EL?Q$HqDjXJ}kLDWJu#_HJ-_-97p3#MIceC}GRbFiI6vX?wOw@BS97L6 zmw~PW6@PpMzW5sIyhh0wX`;oS|H{lr={7Y|`U~Yo{3zk8BuD({;?uJue%`U-&U)MY z=E5_;J7r*i?26Z7OR@ ze5oWE1lKtm1ejEkQ(TXJXI67Uk?xF2VpF~$D?tu3eq2x>N+md+A*B)we^V&|@xSxq zgBDB*jL%@w`AuN5{Ip70PcC9wFh&s_qKy9ASxQul#L0hGPrZGEwB3nmry=F}ad z?tkcBP5KkyWDka1LyeUQABBv0sg$VkV_=b&t1?SqQ7uD7w-#*O*2Sp%zwH0?=&J`? z7h-1i=5l;|z)voH3wB?kt+%kV&i!^U(f5~g)`#4Y>v??}m?(-U%`7E{!py$FaK}1r zE)ZnMdkEwahss5beAP+X6EVkwQk-BsD1YUN6Mj^EQ;=XWxAnAbbEc;`ZQHh{ZQJ%~ z_q1)>wrykDwr$?|zxzMjuPRx|sZ^4ORh2rGy|Wf*`dijD@$ZLZ1%vOP?=L53K=hQN z$fv(TncfPV&pMTZbHe#cu4nzDxbOF#@=UDb=%?@JU)LWSzPWyB9Po*!6CiwvvI|5d zUisX+NApxO4Yz-ej<=e_UI)&DOZ2T71~&&6*lweKSS&OLl|1NsE76v=XrLPq`tGZ$uwT&=l0+U-Ly}a)M#kZSnrwSQ5Z-ZS|zT8Y{LS`k9_cP?tr8`e%dj! zQn&(ek)ttwS@{NsxSZ?Fpo|XImjIs%CM+qTu18mca2ob?&S5(JHVf|GoNrzKmXX+J7nqG3J2-7la-s#} zrTIlr6q7((Or@d7DXZE*;>?zq6mj_^DaH9jPSGi@+K{VVK=r-S`>BBH3qWYRdYSK> zPOoyJ-m;#@pfT=>GFxIewV}_S8l0O>sIpSu*O^x1JX?RM-K&hpvQ|;3WOZ2^jzhZ2 z=H_k_a=tuqx(x{=n@{qhq`^oDzd1nIPsEi@KQ=25_h2U%bnkrZZ$vC@k0aD@)cb|u z%+sm<-9ZeITBMXFKC2*41z-u{%pICDK41My7v0`w()vPiNO0ayv2!pG6O4!+@l-BoAnmK{FoB0QxD`wdFY9Q z{v!UoUj*nplxAR*KUN*-7&|f6x?)|ylt)YCe*Il~I@@!&sIqP=u5{htbF;sG-$w*d zd5*OHz87QF;08(QIgXrw#x4~XljS#dn=M02`y58GokT)RP^h>Ax*oS*U@I{DuO%sj zS&u6yyyEBhbsbkwptwysDrp}cn{mGs=B0T`3T!2Lj5ZhQ#97aweO}yyeo1zKekNMa zg`d*8cWZ2rj+Sf0P_H1WB}E_&#YfQf?#kQf70q1g70KJ^7u78HOXqG3{Hue~aU$;^ zWgElNwadfO3f>kpY=7;VH$3K=0Qj7uY!`qsriiZTVW7{W8DvDjsi8Khy(zC5Vq^jM za*T>na0S;smu=2U!H2Nv1~Kk3Z5^|3f<*w8ny`rRmt1n8m)nT3Vx~q%zABFLAvesV|AKf3g z#p7JR3JoGGN&}~=8y%1e-*I%S~Qg)S1miX3iDV?;PkwB)1SRsBzc1EFST6d-wCdMsAVMyEroUFVL_Sv(~iu zzA<{Uv)2x`A!q+&4PJyNVWo5D>BGpfuge_I??N1^7koa{kUkV?&K8OH+Xk1dXoJ~C zy4HMs*@~^)6>K;dX**1E5M?U$w*JEOoe!Z-{mb42< z1n>S+QII0N;fI#VM%<6LLFpVTq2X{vX%wob$@w}tzM0SGQYHe}6k)&Ruqi@~3_1`e z`kr_Wgls40@QgxI)^%R{XwOx$8ex~JIrY9h5V`KRC}OP;Zg%q|TjR)2&Arhr>iD9l z_|%m2cVJStAAh@+Y0b|7>Ecc?_N6-^2770;eGgB^f4zR_F@?XpFo~F~M!EAIe8_v| z86L@gr_6ZN=KKM?@E8kPev)?Rc(YA{X4t;$X3#x0Wpwi@JV}KdqFr2WNCaWGdzQ{< z6WotJ8*sP#yiCsSpA%yCIyF5_5(|&}<#nq5I;C!0IUjcz&6=Da*RMCd3t;CsbfJu5 ziITCGW|zHj2})eI2t?`J&bO{{$E(3#<4HFR!{@>u-dzT+)G1pL*=bjz*lCv`lR1*l zb|u+pR9%8-8kRFAM;HpGTO;uAS|HYN=YID#1*<95M8a52k#oM-^+A;?^Zw}iHR$(T zRO27~|6v=?D%IcGU#&L%WAHxOr|A4yE5bO7 zt73tfJOnXg&b{?P{|A77=x+&k;@#UAq|~*&di(s{HETe{r8i@>=*gz&DucGj)#h<@ zv5J1F0v3uOxX%TX6TYxvMxUZE2vLZL#+5E`@R>DWoR&2Ml~n}`6xWJ^`Y5MgO>Igb zr1qK}ieSj!@_(VppqC{utg{zqeD~scW6}wnPuB#l4jArpK;~H*Sv%`>rj2VRzt0*M z*?9!NXpOMP{M`cYq{1F+!oGuXX)485;T}K!cjqy1&9POKTKiF}3hf`)M-H`WQLQ}a z!xNU&QN3lW$NX?Zb$BLD89Ik!TeQiw^Wj=cVxyUn5sK@OYmmv~a}MzOj^-LsMCFvm z51z7#KwmIVF|0mOI;^)=3_Xf!iZ3WQa)D3{iSG{rf8t)S6J+=kfBoxxIG^4q+T}=3 zS=XQg%k8V1{aZk<9YXuK$jsFr7PDASR{%2Hu|;;KcNgD@2wLABF%MRfqR2T@T;B%- zgL?>(&|SKnn@owKMBfM20)3mF4?=k^%ub#hs2qVzOv9MXQK_F@N_v)&J!7nrzCx4T za=>#f(n>%z!GkH6eAgH+bY$slxP?DF0zUe#Vh6EusEk@MKzJFHd!O^w4cZAzJ2D(^WCtiDF)|pHYq&hA^zMkRui$wB zcU0}(0nckoYplLt*2EG~8AnlI!h#099-eHIKPHY?Z^v6zYk?+ZK*(ZVrOj zQ1V2g!$!uw%UAtPVSrdt$-0xsSI{e~iPzp;E)_UO3(@k>w+&(Rl@#2IEuu3jl!>hB zwhvu<5MA36$4gL!tz$JSU3n0#FP*j38b!3z-64~bL4WAMy6hwgat)w=Lh8(oJYTf7 zLp-%%jHFFt{ezk0`C_nBUSrKCl44D?4mo}*W4L;8sOi{;AWD}%@S0=PbWeJW3#(jK zJj=mmwt9U1W$44on~c&2hTeSG4n~9saQxR*vvz`RcsV`t;MNDa`}%mD8ESlke4q*y z`sHWt2JgC9xw)!Ze!*kz)~ChPOb`b3``lOGYMFQe5eY`SC;HVsC;UnluKT$~_#Qk!9VEu^n#x+8gXy7Ng(DbJu|7*Zs z5Rzf;KL(2&(uXSCh%BIO%P%iCs{rdPBP=T~t4Qu2aDoZ0%Z%a`JM+6lB>Bg}&z%l` zI!cg2KloTdv17Sh3jq#=V&5!5<6wtAtRQoXXM;aEHuQju%x^z5+*8y`ixMem5g|E! zMBLvD!gdmhvI{>Ykh-v?=jCZe%(uh+vvSA&jKJL6%(M$!K{lqa{OLk=Gg-W+;prm( z{a!Kic`+@S!h}>TK*H_RBZZ(593Vn&#%pjlc-?|^n*KB|fY2<~0(v;V%K6r!7wnVb z8uY~XfQ1BdlWgA)hMo*J0Y(ZivoZc(u@dbBP0Kxch$|5>T~+#&D41(m(`k1 z4n$T-rbQ$VBG_(^^FxX%owk=oJcRJ^;^EV+`HQ>pHD-KznV6jLJ zWem76jsbu}Ar?_0ObwMAMk1Kl0shzTT|Wmhm0Lb5FYKFu9IC_si8zs{@F}eDsj%jN zPk=bm0GeB=K!Fd^Pv@Ub|>hjt+D_Vdl4)@C;f?(Lg7~&oUS~R2GP$lWJNc z)LcG4bIJSSKjH-QN(HjQ9w|lsS%OGq1GLh93k``2hGe?IdrkJNXw9lTU8jYw z=wJ5kVBu_HuIC#)X(-d*NS6Xs<=>A<&$^poQ}*-Pv)&kNsF%OEyzw8evfH)GHBV%o zlmNb}pj1L6Je<%lT035DF4VDG*75A8EE`R4^Ff0h=p97;$xb#o26dH-)XkxG+uyO~ zPBdi0%tz_(v(Bn*v6n$dBrxL=N~%p58|`@{zmL;0g4i;Isag9a@aO*WwO1p%bR~2L z;cOUlws`80pcS$FZZT0LjMwltsj1|5lm{R+1O}n3`9b38QvHz&^y<;+U*vCMdz)Fy zaRTJk=?!rQQskxdcydvVe%vGf%#QvP+z;nLRzhd)hh~P##WYr)sTAx&elai;J*E@B zLb)AGsyKuMPsF}L&7H#D5Pdw@!O_IHCnTOaDm$!+%K#d8@;yp)IXuA;$?A$1^OL z*paEh(TdoaTbHX8id2eJ;^m2|gM-9+>lz`Ea1foR|EQz%M^zUlP&m_RY;ta`+@w`0 z-&j~&wd~Ffd3*nYp|!U4w9N{UNa5fZfAc>2*nAQeRM${M4IP;Bytyp_{`ijkBDBd; z(Da{Z#{46@i)}O+icflrs$fjSDK)5&v^=QcvAF6$kMZ3lj?S(wA|^7ILLx8o<#k*f zMLn7{Tg~&#{t(Z#Qn$biPEt@^fi=NjQE}_{kK8PHMmh>g`sa(B?UCo2=Y-OD9`2%U z9E?ma^1xh6~a=;DIIa$mnmLZ{&`Gg#C_wneWJtzj>o&gYn8zd2u5f^nagA z#Fh2z%zBb!%1UI)5_`MK0$aA1*_FkulmwmD14qjebM@`Z3QCskFx;aNHc|1BvG5Bd z|G>jWDG!D~mSa|E)GD&@Q8MZr5HvBcGt#OZ92O}KLO_)Z?(OC40L-UEkLfy6DsD1Sa#5<*VPVP;5W`S`)}iU|rqbCTVzF3K z913*uXr&Sfcr2+7#YKwqC8{Mj3*`%z3(^*dEXiqN?uvJ%5DS79C=ZASGBia=^5Z23 z3$PZXjbXRBUZu_n!1n_=2t~ZoACxJug?%NHO6(W{BEK1ws4?V76-Sj|6D6$+8A)X0 zbPGw~_gUi8%n|>PVu%Q$5WS2VM2Zv+f|=Lc{~m8l@@9lu(OJoP$6F28ky4#utBk97 zt+<4z!mq;iF$ui;LqP1)m}RbQ$ka==br?MPwF0>sSp(OY46K!3&_yXluX8Cnm6#6E z&rSp>^uj^~%_u_?KIWe8r5@7R#wm|O&lZ^I=MfC~eCL@g)RjQ{_ zBt}w>$c*mQ>R*egw~3~6g~%w4>!&CzdZU^@-GD|G&pA(2Yu#n=tGe5x8`=|@Za;|M$^C_ ze|-nY?k{ckQ(84=UUgggZwZu3Qisp{ROmg)T1^LD0h@E~Z|KsM90vPVP}q?Y_{RRm zt;jgqt8)!}<7s&5Wgyv9MctsWKj5IlE}mLw>e79_=)tQ7`|r+WY6FsWpl^ zumwH}0ob86e$4qPiYuOvFcp~W=P2=nB)Tt~gu-!oir&4IyFZjp$X!y%;rej`vF*>j zm1Bm~@_ZA$(!|+T%*ScVDnj~HayeyU2fsKtwRLEp9^31P!?A31dv*oSgUPqQ;eXpF zfYtvs(t~KWsX{$}|Hp*~QJWsNcLW!k145K>3ka>HmsY69FfKlJ#(BwbooF-C#k_;o zLdq20h7OzJtZYg%RVgYsL3K=!6LLV)Or+jRf$Y#DEULxu`Fu;(W9+EItB`cpZ1F!L z^+$3O=M7m=q;I0)h2?6r`bd)hisTB2b~!I$!fn-}R^N|LF(@2evS1lIA3LwJ{%V8a z229`5@nLqY1wgB|NqC{E=KSXE#6V`T`ibn+k#?z5-)6*gVsp-X4n&9nZ>SN#mz|>&S9g-hq5nYKN@^TRivwb~ zHOjIUXGB!9%iY^rV3^CJlAN0nu|TK#JlLk^wsU*|m?+EFw?#hjL45z0N27%7vT?b2 zyED|}5ITd}dcrKev3?kLLx|Ots@AqcXxr8OdT-wFV!7`jVg?6tXgkI!t4}3S005H! zn?Xf)&U$XUZ1{J6_4nVQz8b-ItrEWoH_~#X^>z{C=KbqQBpd%XnbKu&DGdHe^00L$PY!&6=P~eM*gWE2G zx`w;bhR#7DS0N8u6gcfRE~FR*VA4`ZlU?(E{BU;l;C#8QH-W$6x9lBy20fP>sGJ6F zh7Y@^W{0I0@q-|Rs4IUN6nABN`ViYk!a)3Z7lgCtF#y~g=^E4|_XOIzc^x36>(OcS z(y96qwn1R_mqesv>+}n;CLVo&J9*3fHcJMxgM7-mEHT(T2r~E?6HreM=uv(fXHRZD zmAmeLvB1TTsH=h#x7-anxkfevS>w*>-JQQp+BGPkI+&|>h>uVF%p9$3=fd@9B|6@O zBF8pu1A8QpKrH6R0UukVPvW3n?D2q@sqgCFBld3Cu|O1^e`%+5_Efif!8D>NY#WoV zB*ZYfff@H^{yfLee!ay8HdFH1%vLHXs3M1e1Tr1^XyE~gcXCefe_ut=b`fH;&c%PixP zJOzgCTHecQ+biYiWM_@;WJ7HwmF8ke)fA?&x_BFs?P&@b;e+}Buo3&d+uTHZRBoLx zI&E)@8!hb&WdHu^T-gTD=gz(v>k`N0Rf{Ri!awkH6LC_dDRM0pmpA76;Lqb~C!)^K zM-T3)Og+M`qb#FLiv?M*l2guyFFpo3;%H0@+ZJAhw&p-qpPABLO!0Kv(G7Rtby^N8 zGKWU0N5j8#=$fklOD@JDvrTb~QMFM~Tuw?$eSIxxOLQ=`o^zHj8;uA-iI7yD{?TW()*7;9tF`16na(;$;Bkfv7&OlDx0RvNdNQ?y7DZTo!3-G}RL*cqfkj z4UONxILDd-IXd}y=ZS}UWY=;c;j$72@3tAQ)NlSv3sP>7$IIiVH5IMbkiKDIUU|On zL15HAXeC01?_&KK`47V%)sP%J3>=|~SKdy4c$~hQdrq!P?7sU>JY;Y0Ttm3&Y38J0 z{+QLFZkw9`ew~|VZKV>qaGGWXRw@ide5!fn9K&7QKOrGJJYB(GoSmJY6LT1DQf_5k zqv2;dc6?W`eEprxj`5wItcN z&ZC>{F8;!HJbU27J_?fi6YpT){)_d_zKHi8xF9&H^_YCczHBFPHKgbAeu2EQBcw=s z!Jq!Da(Yl}wJvaw?2X`j!0euCK@PHsFFPhQlovjsEt zJ6!<%_iUc&2?3DUE4Kb(8*!HOab~3qhZ?UzyF%c~?0C@xUz{oim&2`kOTQ9yPIpnw zrZZgU`&)99z7XzK5@PxYP`ncDbteUYgUx4c0Rel4UbrY)s9&?nzis$ALWbWHv1n|PKW6ePR=Kx2s$ zj0U0zhsRXrpjOXCmi#1>3b}P?Ik1fo3bl&!2>B->p8E;UrxxJerwB!u>(lfV+bX|? zcuta&{2=&cm6F4%^=c_&e2KS@FLKk-Payy!d)-;hG_IuM}Kofo=3ptS*g>Y92sT1T~`*Hs{`{d zUp_*lO`s}=wC=l!K98n~ ze#N`tQJSaXGh_rTb}J6`g02I9_r;AqV|se!FG+}rIHSc%Dr#A2SXJnV0YXHlJQZa^=C)U$7-&O^@-R;EO zjd18wi2dzkN8M8V$0S6_2Ka>uEeJWdEFW4g@8L*Eb;=c_&%3}ZAnqlW$IB%&U3y21 z_H((ABl+p*u5D~R`e9I$uF~qMx)HQ|LyWJb#LJh)X4|z@z*7*vn^%fv-n$`t0|3Og zV;3y`$$U(Lpf66j2Id*b9juvM6w=e~^tAfxJT9GbHF9GPz=UBOE5GB0REpp22&-UD z=?MBGza$8Y#|rBK>5xqmZuX&OrEa#$YZld!mK}Q>F|!lO@2|QP`>sJRWZtC^R=}>j zgGB*j4RL8@WqxkRw@#IL6dTkP#E`1-823~O%wj*u+2jSBws1HLIJ#7A>&8+ig~A-} zCxWuC0xAb%hGZN1eYTp(l9tdftC0hYV|fQ;RRT1Py4XmdCmzMeW-scPpwvswEctL% zg)06qKEgH8aYMo1apG^6bQKxjcWdLl7_zKkR8E`&D>@qCHC%SlH4Sn&J}3AKy)}JC z&}6<87VHmdZQ=0l!*G->MK5Gelb7FkOG=tH)Q1kmZ`G@)`ibeq^E@v0EUs6+@#EZumOUYt4;aX-^`sY*$o|wXVN&944u>Ts$qz!AT)rG^h)D)i@H>J&d*JRPco(j*=f1R@cqehR zc~`S`g(8j0g+|E-vxqpmvNgM8!m(dPWL{g+?d_@`FnJz1ItI$^(bo2CeSju#yHF~% zm-})gAvyQ(#hCTgz_27TqQeu28hz4c7{ZkK1Ew|{Q%5pe*Nh2tx;sf+5q1=X4 z>+^oQv^-i{7OG|PBsg?KRZm*Ax!<&>9pnmtGDMEhZ)pXu3kpMvWxEN#SRY1EmD^C# zNq3A|-a@J0Nx!hx=l*b`DdEWiM@G=g^YW_b+D-@Zx;cC z`9r7zuR~UbD5WjMJ*`qdcq6Dlrl9wfdKjDs!Uj=$OJv5}O)aw$$t-Cl4QtD_j(Hbd z?V>U6@J4L4kbUI;X&&&66j-lg%DF$;c{qy6?Zp~`G@giPIclmFhIYD}^{&74!gXx! z-gw~u7#|UCBQb`Z+z-9mj|h6H+2#S<>;!L>l*A{~c1M4Ki0rxg6#HS^TXykg*_vra zIbMHcXIT|0XOq7#dq8OO4NQ_%Z6~)0--rmzQ4%JGE&ZuSHPf~#;7xj0z|>TZEL&-VH57tCst%wi|MiQqvu73HRBuut zN38KLSQ%y1IW;oJMzwrBmfwJ6bz@)*jse|tz&38Yal71xJtwt^*&xN`T1I%O7pRY6 zNKoA>@SHsPtp2)1zWt<$Elfl=Zmyvb!DuAqb_?U(<~Y7PN;ltqB6a>3z1zDB;=+}u zi8WOX#=MvTJ#izP&;&`()Mo#&E3Du=*6Mpuh~o~{E8f8U6MzA5r~RGj;qi@G z{*o5zq5}I>1g3rX7}^4ZQPUGskI|6DPngh(`b&1hjDv5sn|)&X3l-~zStQ1n zG4UleWymA$yUbVYcPS_!b8l&Mku;FbEqtSl3ET1IHm@j6_8_a}u%={W|A#Ko^?Bp` zk?h6(4=pXa@?R`zOk&hRJb9!Fsd>dp@PP3U?qfbAqqrb{smo%{9Z;Z^De#c5j!&K&pgPI`M>25%!b&9RL6%fn z3hVPiw2OL?gh$fYwms%KejS41AED4ol!|dYw>yrXl=qJvJEf9Hl)$6(Z$P~1TP~A0=)oZ(9o<3-kCqI~XzL}?btd8z&_OS4?Vzpp- zxa?MeG>y!9)+omw&C>}UR(qGZ?E<=zLo~st(9ZXsg>({=`A!&$(w7+fF80tNTl0Ge z#&Zu{pS}Dl+cn%DBJN1QD=N1hP4 zU)blxK@pUmB?}=ZW~31kLN%?JR#%trIXqFw4L0SaLy8s{rFI!G1ZDW(Qy}80M#Z8U zIo0?r$4EQ`0i85+czD4?>ZM* zTs4LlQK(J$8uDoipGf1o!;W6w1 z$A@PJLHky53r(0DO9hUeOeqS^1+X$QGX3B9Riy@u7n<>EHZmD{iLyq%6Qn69rN09V zEoBufsDo~kygYShb%BgTb%Xh8K`WViJhNDQk?VZBIb3HB#6GH@Lq$*7ahReJkDlws zhl=XfXxGMu?Oa|CpYzH`*Zq3tYorW}1;a$M%HU27Dj%AJXTxLUo~g{K1OwMjY(6=N{h$Y}SFB zYwe~54n&*-3+L}acbL>}7Ol)ox=CpnaWon&dcAoZetY$K$%mnfLqp+(?&`Zr3Ra^2 zW*6ME!R~pHbKI9!WoAjtLg>jUs8kZD67j;Zp=%->>i;6D3U$(H@u@7HSrm+Cva_V^`46IZP0txZ4F%)RV#Ux=W){%eEJEPXp{pYJ^5u3Ay&+-)xaZ@kO5;e?Cdt3ufbA5zO<^m5DFAF?~> zJJ(>bNC~_xH@)+}mloC z09CUR_V$yZSu8r0S@3CXk(B<)n-+KoKESi#0d-fH7^TSChExZ~(52ny4bBUve3{XG;82P4M-r$P# zu(x`vM9H|v&Kf!TX7}$d%*Go``}5(lJvgRm+wrQ~?62W)y7} zzLVB0s>zD^S-w2q5k1`P)=9BFBd3eJXd2AdaRWvR7Udn2X_(PtG>rquqZ{8VUy!|h zu6B3=3Or}9RVA)w{IK}}iuJJIBvIRGJcf<48E^4FCON6~vXw799bc!ql&x=HXkK82 zTY-Z3w*@g+E(l{&u$^HZrR+33)~+%Fj>*;anXk~FvdcLo@rwHCzex=wT$FVzldJoO zAx^W+%xTwO5u!DxkNVF8`ecpk*>#fQh-~DSmXuqitd(})xE3#f2qenG8k-;3<%$eKMb{Go!+fvl~(c0pXoh!Xob2UFm}57CX$xhrIP=)3Wl2OIr7EQPC>>u z1iSpw!g26tz=`0bEQpXwpYM|Mx;FoQ74ap~1#o~1x-#goACA!GA~%v6(`X3i#c<_# zqcrU&-`N2eC+C7&|8WxhJZ2E^Z{bS*NxzO~joEAY^S@nThKe7GE zI_6L49Z#4MoP5^){Y*{fl$l?oblBV+lq5{&le=a<|ldLb6`K+z& z;W{RQ;B{@|E~k zSf?G1c)dsR>7xFTfPDl$&iEjwsRNwCtZ|!k z`#u3w%AC|2z+Am9_rEg^N`-(75)i;^Bu&PJ#gT_ov@#1OxsY~P; z?4@Iem;@ZXg!^R~OZI_>0Mq333rFl~P84}9Ch z-8C7X0V~y@#ER@+0(WDM4*0jJ==I#lz%IfmyS5GBR#rqonW{8inMm2Tp)#u`=60tI zn@UYb>H*k1Eq&0NR|gPt1p?G1E`GRG!5A~saGcu)9c%wyWG6c<;$aYSYFAH;5j$8c zD9gw=dzQp>A3W#G4fA@0*No+e9WyUWQb^C`7)6Ohr=GsLIdAKB>Njfl1GQTD$;cvOt%v_+YEB9v(FAp6$C+t5`pPK3x2VzJWPl%!{M{6g z`Qa?WSCxYoiy}tjKQI6Eo%+>&1!23pcx!cSY5G16QP$e4x$xUF)p0Wi3Wz4SrThjR zZQhmMjwQK8u+5zv3Oy!niL}h&IgL?OX>uG2(HMoeC8cOa zii>^xtVTvlJA^A{5k9vB6-rArAib)n(n6kMtzkjLM{X7=`>V;0dFxvHwstd>fV(F2oC}~n9A~(%9M3T_AC}9u$ zE!gN6XLO>FV8OifRcI2IJ+30&8nh9ZaILq9;c(`)a|`B zU03UKU8+B{0^HV_JtM%V&~pMti?$wsaU(LuF3LFUh{5>P(JQ;l|64;@0AG6RJA);j zhcuT3CB5oOX}yfMX;hp{Qic}<@sZhEC<;*(lf(=UUF<99_E_;#%cN(o8QGCBdV%nF zh09Klo7)48DKGKYqYohA!>`R{hjc@0J}%|ynWmROanvIryYwPwI0!-c-(s^rd5rG$&5cFH9NknNn$lKYln-OL zK|c$(tIsn^zm!EAxHQJ87f@Q@t`@V$Hw4Z6{1g;}L35PR#*u)rHmHuwMw_v<-wZ~i zf(FAqUco$rPthS{MC61i257`fig1Gr6wox=`~B7kpvAQx&smJ^XT)3WMGgQIU3FOUtWid(IvJCzhy6^;riUzM^C;~*Rus(c+@bsxS}zvE#NRwf<$2q5+! zkVo1?WQ6L;?ds&C7{yY&7vrYCiH#+qgbvOQ*+q^e=o*;fJ9#GmU8)VaiF|ZC&~Rk< z2pe41l2G&OYzXt;v7)lUgpc$J!cy~Xio0LNx8a(Uy}tum=FMJ5R=&sEx>h$97iXVz ziFDaLs{KAjg0rjsI^D3lx=w1#q;fr9O_bGM5dJF~pROb0$&=_Kl)g$R^pDbFDl>Pl zSyBFo*pL;=eA>SYT|30-0>Kv4`AK<(EefV{Oo3zSYN|RHCC>8R*QJ`m=k8!?V`l+~ z%@=&`vE0D7CM**RrU=q|jr|un%ny9eZdbTwKTePS?&~B*3B!S1rSiZ9M}DQm)Uo4y zQBxgS1%m}#7Ia*;Iz(%O@*B#`(G01)Ut7ad_V{Ms^sreXi~flg6UT??biXabtT<2* zwuWM@P7J}$X^%+=;)X5fmN}AFSRdM1wzj4rqC^6NAq0^~o)0#bd-c`Uaaamp(F`1* z6MwA*A~6Dzx4s55A1S3zjLjQ@AfoB`)LCcG|SuY zj+zZL3T0m-rPJ{~X#aj0SwT?U@s@`+wYzv2)6tinBHoozG8%e3Rvir_+2RUGC+gl~ z=UMPB=}h*@cV05_Xh=OZ4>tAY{sQYhdShBf+_^5Bt?jRE$L4y+`(Of}WY3`U&5e1d z$`GfsKsm&il3f=eEjW*{{i#s04QkcOlJ*P;+ntQqRE>+RdD%^=wXk(gni4dNB32KP z6r0GChv$zQf3YTU=FCW%lVFD~iZe_5|KgU{Xk_gSuraT1a&dy_ImS`O8oTJZtm8_P zDnIVDBh%S73*(X$B+P&}?OnMixC^B@wUnh2%ancnvF`86G41tyh-88`5^(GIrIP{- z2RI_$=@ZYvf>8WzE2;c$S*87RyW#gacu5e0l1R{s6)8Ssxoi*_wE$Dr?k#L(Y>qfe zZL5{{B%+1${3i`YL1)8)&yh?u!C`&wIpR)fT>VQ(X|pi{d1Aqo-VBJIUCGJbG$&Ry z$xOo}FVqJzj4MqF)*5t@( zJG2S6v_;D6b@MY#F!wYYOkQm)nrOlF;^e~|Ix|F>eE|F|iZY2R71bELa@8D=9}BJd zG#1*e@4wh$FXga`lQ&IicBV>Zt4An!)8qh2`_8 zPa0COq7_hLbcI2veJ1P4;Vu^C&sii{BJ)Dftc8rce zycj(ekE<&}hw*dl<^`+acH0ub+NP~aUj>s2v|jz(AV`={waT2QY?c_iPs&*s^N;e` z-9zH-bg6g#Tem@0Nfg5P>t;=6Qvn+gi!WwNRU%b#0{?XBgLNbv1xXtAmiu#S!<8zr zhErWNcOC3-Ix9&2JM;U#N}FR`D8w{hlikT&r!L#FMup9o)$j%>zp51gFK>=6Iner; zp%0xosppzPLrZ=@YkO(u?%(8+O|{QRjDAZ5A_O+0Qj(bL@WqzVU-twzfAik7L>Cu1a^zilrYS&{aC)8d}xYJH&r7RzE zA>gz!ko~Sz%%N#HilPJ{W~Kl3XZ5F?O;f`UkG;|Tt-fJTy%7*ufhc1k0?&ZysAs0Y z%Q(hH(P8u?d1A0Httdwb0jwE2QrhAT4Rz64FhDoTQQMaQnFz`+T11PgYn?Cks@h;a z=)w-;NKEgaAbCU!xO;}+>}>ihWkxvwc`16(qoDy?p)zK`kgg3(3fR5&E_zUqw1Xq` z7Cx$=h-{bcc(eE54JP!BljwJLb|9oHsh_or>#i`+cHJ(aQ zwag8Jr2B)nllv+bbFA-Lf&ATHLKs89S14th7gg2+69M_AQNf}zbvwkpe-|Jjw~D zzL#0Qr^uq;@}cLpyDnZff-6Dv?yi8c7;dK;vD0jzE_-p zUQ{$IFQA6ZdokQ8&QNb$kj)%fy$I#N7`d31G9|*?UnaIEi=fD5PO$q_ly`Qfl zs=fuBn2dM=MOt45Q{kbX!_Pj8UkS+A8P+GUSk8%P>@rS?r3&dH=%u&t z8n735qia)(dl$nKRbjOzez|eGsRIk>ma8Ra;T(fL%ZfkiXNtu{?OF#bs(+4KAL6nh ze2siIU0v{8+E}*CPgIh2r61mH`{#zC+3gm4sZ2rwa8KxOjCIB$6@tAOsI3msM66SEPV>xmzp<%(?Ff+>;e3ijF|%}1J)yxcrj(59(JiWBULw&uwRmLy z?G;#)ZPq;wx{I>+(Zs?Ipg~whc|Ah&4NzasUKWYCmP+cIbQzECaYCM+8M0k{2Lfh6 zbn}OWa!t{aFIEM*iq306>Ie0Uyqj*5DDwcFovGqJt3jByKP8_FY~7#l<$P?jL@3Q- zIp5vH#B$`|`-ekzCiJ_u(6IwAE&hZ?%E-n2Zna-E!twY4{R<+~PDDMip0cEGcGf4Q z!pacV(i|gRfdVJfpo!Uy3sLm1NRIjL3f-^L4zX(%Klf{FP4J=O_%IbL5uAh_bxvbI z2^PjBAP@I6WN$tb?3QNwFgEou%?^R(MqB5Z)K|6GJ*6S>GW+?}!r+_?9h3a2;D1?@Eu~%_R zrjhEr`Ui(rE2pi<$wtps3$D``S1}3TZimV`pSIQh^|*0uqhHQ(;O)ASUyZ=m&Qbs* zh}l04@A>-~x2Zn#ibqT=%WS(OKliiGxgCxSC?k3UHHge~!R^tmr7hDbPVe|gb}(VT zt7Ui-=~qPnFP<;cPJZQr`7t=7CUzcZsNQowO-9BRap^w@TLLx%_o@_DzA0jiX z_J~#K&$N68a!S>Rx-J4+CLd$TIO^;q0ZZtjZj9O@Ed%p6L6fY>B<`5w#qgMv;R8%c zB@d985o$4VQmmS0P{sN(Q|xk--+Vav;e|bAmOItsYM;VnA+cOtlg)aluuLcH?tfdK zs7mz~5Ve3`xh%#B^*;g76))<6yonB*s(<1fmLnexW?i+%WuwdjPPVd7jeW-DfjFOQ z?I#QC_4{JJ`;e=;rp)D{RGSi4tl1h=^!+$d^$x-J zu+IDaPYVARd}tHFmq1el6qoZ-1U46tFtANF{=Mg3CAQv^@9 zWkv*!43~~&1SXdmQv^G=>tzIWAObTnw_=6_SOI^jcU?TDg;qR2;d+k zATSawzLx=C0HV?el~Nez5D%p=qq72mhs~N1H!#D%{mQ_n1};%l$ihnHz$b@@CsDzci(n50z0}cVhAuw7d3n5a}64R2I;1(K(U~3C29ipfi zeusZ4bmTYSs*WTJVq^q62Z0M%?BIW>3RGBWj0UmbTSIm@*lr>w^Nf=M&*(mz$R)@C zT2MKeX2K*Io~6DZgy-D&2jK=TFe{H}S70My98$^~xEDa2Rw}R*e)FKHkCySPf>P8# zRKW-a9ehP#krRQ0CfNfGk`csZP*oKf0~de6Tnj)Yt4IYoYZO!+j1dB6(_@0VYJOvo zu3k8j8s1=Di+~9vW|E*j6~`TTEl6ctY2>QM1QJ9<;HaE*j^BZ2(leVll@gK$8ICe3 zLvW?Rf<9v{s@J_=K752bDyL!%rN_6ImxJ%_2!*d<_GTe+u{I@{wl!k8S!jZ-(G-7# zCf0~iL8yE!OLKy1HG*CcT3f?c1))|ovURf%=}?=JDaz|QmSC0kbtsvFtgclF7M!kS z$rQx9ZcS<0GVfv?Vo9hmr7a1Cs&z@|9DHC(LM>`r5@KDfoh}J&sC`kWU9EjdC`4^a zLIb}Sg&KYrCHHCiy(om=RY9hLs7rr_C}V0}iir}TL%ohrdR5y{62d>OD1?7}%LV>E zlK6F!n34=0Xmg|^#F|)ZSrWn$B|bqfV2)Jp>ee@3Z<;&LiocEy~BTIb~D4n zKk@9zlkfU?TH*>kWRtqQZf29)=~*+Y=1`|!uAAPg;RXMhHshPL*`eTIGQPq6tdAIQ zbZ>t=o+S7htvbqW9Iv;XU$^I9cAufcExMc4xUA6e=p19R8xs54Yr#@nWLvOgh~1=^tu}irAmgVt|3O15bV|nW-~&J+akoxl(fQZiNJbk_vFZw zPjyXhQ}k0+vss-YL-o2kn;B}Q*cu@g5cIYPQWX%295?F-wtx_~MhJgfW{F;fWxn00 z=Fg1XvV`?#j$xRCEE473l#?M3pD((Lv|SKwq2Qo*(wvWmPbZ)Ilv_Nmd5U5r_B?aw zgK0ConM@zO7@ju2RDX_cK2@JE+jQFe{Fu!^_D9AG)a9EAJ`e-|A}f-`$?r7_P*f zoBKxpfB#_Pay#;sIDgqEqz%X4`9t|lEn6P6VuJ_K$H(pFSoMFqJ3rZs<5e<9 zdK#4Opf8V8@^L4P;fMe~@F0@kgw!v(zUGl7_PIJhFJXC~>|rn3;e{V4jEp`mbhSD9 z-iLvPo2*;h=A$YYXaqWjiXX3d%wekKy)(HBT!yJBT7W3gPM*7o44A$SJmP_|Do=~K zxT(@kCNQqd?9RD)d`zdNN$`nCserzhS&alSfAbVnsO*!)7mL=(l{0M6i6cpE0=PiN zr#Pf#j5kIkW6-*Oc4+Jy)^kwU3i2SNfCWA_6QkrjM=VLh1~f@cmVxW3$pd2U67wC@ zPVGD&9goRl%10dsMj<;SpKJpI0wKi$&mtk=bSfQNGq@i6GM!G>ZT<=@L&NmsnSJ2- ze>}V^BiYQOcb<->TjnROqeWO5H+1JnCTYe3QC#}k;8B7Pt>DNg36%2J0p1%LWYQ^b zTbsb1V@gvMq%ip*Q%)0-Ih|T1(wP+pIxll>LlF*EbSVuLDrqy%AET46 zG?R<(A_P6?k*{1|wBT8^&jFJuO+Ynve_E!8Ps@PMLzx*Fu;PI}(l|lz)e3vUlX=zh zJRqkh1Ip)sZBQEJRLTjgeVmL7M9L~q5!*d{k{`1JeL8Sg>AeOenG!NHUotNhu0z&F z`^3}gORFH- z!E%TF40S;7&FybDX`^{`$a@Mz@9A*XBswZ6*^J0C)WNh$+#?Q}YMJ078Te}i($-lxN6syzL8 z^zP(iOBJx<_HsCV0Ja3R&KC1EY;m?k7OdQDvS7UjS+FDLo(Ipu0$E&9^E6=CGuf)HX!2hoY-o;0;s(8Da^>P3qT*e&iw z#~1r7M6n5O$H~nP&|w1t^{96Zf=|vlQ8R7MdsnxYH>2yze=oi3%iG!2XnZ^CUGOjp z!vEY3FMGr5>*?fkzP@Z`v#Z;)Prc!EI{9--3wn@VPo_MMLWkdm)5j|ob)2+Y0-yf- z^!e)-AjrvNJbBR7K8m&cjcTj>MLD?2U!~Nh)t-Hp`DGy8HuO1nC|3 zE?U(>$FS}F7(0ISV$F|uIxhX_lc$uQz#7eQPg@)q1?hj#fsAh5x|i zb^m$U_1}M+>7@7j=#QqCYOJ2Q72$8nxNH;&opOrewMKpIEq(?xWD0ZpXw#&NtIN9@tRI1VdGtO)ZG znky&GDw`DIa=|sGMHx4WppKdQ=G6*wPih4hKe)% zORdyrKaL}_kGo6+Xtq^?8Dv?nnAEX?>8E^7{#N5aevaK$A0C0uzLzV!pl;=NY) zf@@mQo$&2;bQfIv0Is_WH?W_L>2Om+`d7I52Hd&_X1Am#;X7Z!oDbmkSh(YFFgJl# z!JS)SUI^Ts2=m!8|L?GXR|{iw1X_qwv|O`)KlP=4;uLq+dMj~C9%~K|r!-jS)q0V> zxUckm2>Wf(Hi5JaZ7b4$v$lVWu427A?Lgbn_9BsZUWtm(eu6}zcWLfIJJL=fBdXDI zR0JHI0mtlzV}A?B`NHuSE17`zC=>q!Cl$geVQ^|&S_G#B(-JuSJ2->=%*43KtTk|d zwm*!S0AqW=IbIZVM&@FiW!`f*e>;q8L-A&1!BDvHa~R(NCiv0AFtKj@JBuXcYVJmZ zX^2R2Z>@I~Sv*X050NEF=plu7PpuEr>cMo6w&_VjX)losi>n*0UhT~{@(T0IG>pDL z`-tQ$K>Ka%lk-h>OSZBhkS~U3iK}Sdmm0CMLw?ES0~buG>S%xJY0d_KH~d&oUScLh&(x{ zc`O}8M~giD4O;#(7Cy(GX`C{DDA?bO#fmeY|H69M>x`GBCt$-!`Uz|_0(xWTG(M|f zV?4EKG5|JBq1f3>Gu+X%h^Ie6-}UrQ(C--ZkAkhYzyS6Z=uQ8uPccSJn}?bw&{yd= zG3~aYW&1R}I=(jlpm`FVOsCMPVme@@o5YW`%`~pH;bKywwKAQRXy!P7I*ZPruZdZ| z8a-?Xf*UcP%_cs@X3R{pWtR3cvo`-r^Bg*x#)!$rSHNWBTVrzY9WmRmBFy$IxT75A z;wxr$awKS16;rlK z^XqgjohRmef3&<1NdKX2=8L)bO!ERUm-^u9o!?yV4y(Gs z8(rybSbYoLfpcJln133rxTiROcl9l&i>Z5xdqd36 zd}|M=o-3K?8s4i zHH(!oVrM?p$|_dIicJpC%4$|#6`QqRD{epI#Af%?%G<1ej2Bz@lUCN!bF`4YPw%p& zd-@Vq)-k(he)l}@p5fi|xqBw}V#^}Bo|Oi)A#FsxsSj<;mK&Ix(5AE*ZO*I3%q^&U z>fT53+JM9;AoqHTo_+MNiW+bO+6)XK5LIkDjL&=ykf2-k{a=Cas~j=xut3?xGjzF?yVy zpeJcL&7+s-WqO5H(5tkPR?*!wpB9KG`;5-&0dx|79YSBCQ8bc{rW5G|v6VQ<*s95} z8mA6hb47cbD)x43RPJ%FyU(5Kqf_u#+&MqVhres7?Wc+T{R~t-t*cBI``H*&J`dN* z46zSZqVmPJT6scS3F5=G_R=5S(@k7j{iu_ qZ%}t