{ "cells": [ { "cell_type": "markdown", "metadata": { "id": "2gzMGSGB5hHo" }, "source": [ "##### Copyright 2020 The TensorFlow Probability Authors.\n", "\n", "Licensed under the Apache License, Version 2.0 (the \"License\");" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "_RX4_K8Z5msT" }, "outputs": [], "source": [ "#@title Licensed under the Apache License, Version 2.0 (the \"License\"); { display-mode: \"form\" }\n", "# you may not use this file except in compliance with the License.\n", "# You may obtain a copy of the License at\n", "#\n", "# https://www.apache.org/licenses/LICENSE-2.0\n", "#\n", "# Unless required by applicable law or agreed to in writing, software\n", "# distributed under the License is distributed on an \"AS IS\" BASIS,\n", "# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n", "# See the License for the specific language governing permissions and\n", "# limitations under the License." ] }, { "cell_type": "markdown", "metadata": { "id": "kCQeUqKYe2dC" }, "source": [ "# Distributed Inference with JAX\n", "\n", "\n", " \n", " \n", " \n", " \n", "
\n", " View on TensorFlow.org\n", " \n", " Run in Google Colab\n", " \n", " View source on GitHub\n", " \n", " Download notebook\n", "
" ] }, { "cell_type": "markdown", "metadata": { "id": "stPyOUvle5ZG" }, "source": [ "TensorFlow Probability (TFP) on JAX now has tools for distributed numerical computing. To scale to large numbers of accelerators, the tools are built around writing code using the \"single-program multiple-data\" paradigm, or SPMD for short. \n", "\n", "In this notebook, we'll go over how to \"think in SPMD\" and introduce the new TFP abstractions for scaling to configurations such as TPU pods, or clusters of GPUs. If you're running this code yourself, make sure to select a TPU runtime." ] }, { "cell_type": "markdown", "metadata": { "id": "FOV8NaLkgBvW" }, "source": [ "We'll first install the latest versions TFP, JAX and TF." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "O0AI1GXeNgiw", "outputId": "f5d707d6-6ace-4611-b75e-7d03f064f08f" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[31mERROR: tensorflow 2.4.1 has requirement gast==0.3.3, but you'll have gast 0.4.0 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tensorflow 2.4.1 has requirement grpcio~=1.32.0, but you'll have grpcio 1.34.1 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tensorflow 2.4.1 has requirement h5py~=2.10.0, but you'll have h5py 3.1.0 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: google-colab 1.0.0 has requirement requests~=2.23.0, but you'll have requests 2.25.1 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: datascience 0.10.6 has requirement folium==0.2.1, but you'll have folium 0.8.3 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: albumentations 0.1.12 has requirement imgaug<0.2.7,>=0.2.5, but you'll have imgaug 0.2.9 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tf-nightly-cpu 2.6.0.dev20210401 has requirement numpy~=1.19.2, but you'll have numpy 1.20.2 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tensorflow 2.4.1 has requirement gast==0.3.3, but you'll have gast 0.4.0 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tensorflow 2.4.1 has requirement grpcio~=1.32.0, but you'll have grpcio 1.34.1 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tensorflow 2.4.1 has requirement h5py~=2.10.0, but you'll have h5py 3.1.0 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: tensorflow 2.4.1 has requirement numpy~=1.19.2, but you'll have numpy 1.20.2 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: google-colab 1.0.0 has requirement requests~=2.23.0, but you'll have requests 2.25.1 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: datascience 0.10.6 has requirement folium==0.2.1, but you'll have folium 0.8.3 which is incompatible.\u001b[0m\n", "\u001b[31mERROR: albumentations 0.1.12 has requirement imgaug<0.2.7,>=0.2.5, but you'll have imgaug 0.2.9 which is incompatible.\u001b[0m\n" ] } ], "source": [ "#@title Installs\n", "!pip install jaxlib --upgrade -q 2>&1 1> /dev/null\n", "!pip install tfp-nightly[jax] --upgrade -q 2>&1 1> /dev/null\n", "!pip install tf-nightly-cpu -q -I 2>&1 1> /dev/null\n", "!pip install jax -I -q --upgrade 2>&1 1>/dev/null" ] }, { "cell_type": "markdown", "metadata": { "id": "iEZhgnkYgG-N" }, "source": [ "We'll import some general libraries, along with some JAX utilities." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "Z0wJAO4FNaMx", "outputId": "e2651b04-a302-4ac1-d390-c9d54a865f15" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "INFO:tensorflow:Enabling eager execution\n", "INFO:tensorflow:Enabling v2 tensorshape\n", "INFO:tensorflow:Enabling resource variables\n", "INFO:tensorflow:Enabling tensor equality\n", "INFO:tensorflow:Enabling control flow v2\n" ] } ], "source": [ "#@title Setup and Imports\n", "import functools\n", "import collections\n", "import contextlib\n", "\n", "import jax\n", "import jax.numpy as jnp\n", "from jax import lax\n", "from jax import random\n", "import jax.numpy as jnp\n", "\n", "import numpy as np\n", "import matplotlib.pyplot as plt\n", "import seaborn as sns\n", "import pandas as pd\n", "import tensorflow_datasets as tfds\n", "\n", "from tensorflow_probability.substrates import jax as tfp\n", "\n", "sns.set(style='white')" ] }, { "cell_type": "markdown", "metadata": { "id": "5DHdAK2rgM5F" }, "source": [ "We'll also set up some handy TFP aliases. The new abstractions are currently provided in `tfp.experimental.distribute` and `tfp.experimental.mcmc`." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "X4Pe3mZKgO6i" }, "outputs": [], "source": [ "tfd = tfp.distributions\n", "tfb = tfp.bijectors\n", "tfm = tfp.mcmc\n", "tfed = tfp.experimental.distribute\n", "tfde = tfp.experimental.distributions\n", "tfem = tfp.experimental.mcmc\n", "\n", "Root = tfed.JointDistributionCoroutine.Root" ] }, { "cell_type": "markdown", "metadata": { "id": "b_Xkt7MtgThA" }, "source": [ "To connect the notebook to a TPU, we use the following helper from JAX. To confirm that we're connected, we print out the number of devices, which should be eight." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "HfgK8wpJgRdw", "outputId": "da2863e9-77c1-4da6-8822-675c59ae6854" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Found 8 devices\n" ] } ], "source": [ "from jax.tools import colab_tpu\n", "colab_tpu.setup_tpu()\n", "print(f'Found {jax.device_count()} devices')" ] }, { "cell_type": "markdown", "metadata": { "id": "2a__PBkDZBjm" }, "source": [ "# A quick introduction to `jax.pmap`" ] }, { "cell_type": "markdown", "metadata": { "id": "tTNW_FyuZEiM" }, "source": [ "After connecting to a TPU, we have access to *eight* devices. However, when we run JAX code eagerly, JAX defaults to running computations on just one.\n", "\n", "The simplest way of executing a computation across many devices is to map a function, having each device execute one index of the map. JAX provides the `jax.pmap` (\"parallel map\") transformation which turns a function into one that maps the function across several devices.\n", "\n", "In the following example, we create an array of size 8 (to match the number of available devices) and map a function that adds 5 across it." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "EV8gboOzZ9pt", "outputId": "f3140e2b-c507-4dcb-ed06-d71fc64f49ed" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ " [ 5. 6. 7. 8. 9. 10. 11. 12.]\n" ] } ], "source": [ "xs = jnp.arange(8.)\n", "out = jax.pmap(lambda x: x + 5.)(xs)\n", "print(type(out), out)" ] }, { "cell_type": "markdown", "metadata": { "id": "1C5WNNsmaTiE" }, "source": [ "Note that we receive a `ShardedDeviceArray` type back, indicating that the output array is physically split across devices." ] }, { "cell_type": "markdown", "metadata": { "id": "xWy1APt11voc" }, "source": [ "`jax.pmap` acts semantically like a map, but has a few important options that modify its behavior. By default, `pmap` assumes all inputs to the function are being mapped over, but we can modify this behavior with the `in_axes` argument." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "4eQ1g-xe42kY", "outputId": "7a482508-baa7-4cfb-9019-87878d4fb728" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[ 5. 6. 7. 8. 9. 10. 11. 12.]\n" ] } ], "source": [ "xs = jnp.arange(8.)\n", "y = 5.\n", "# Map over the 0-axis of `xs` and don't map over `y`\n", "out = jax.pmap(lambda x, y: x + y, in_axes=(0, None))(xs, y)\n", "print(out)" ] }, { "cell_type": "markdown", "metadata": { "id": "5GEq6tVq5Ux4" }, "source": [ "Analogously, the `out_axes` argument to `pmap` determines whether or not to return the values on every device. Setting `out_axes` to `None` automatically returns the value on the 1st device and should only be used if we are confident the values are the same on every device." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "ex1jYXJ95jlG", "outputId": "c1c938c0-7b45-4727-be54-ad93269b9c1b" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "2.0\n" ] } ], "source": [ "xs = jnp.ones(8) # Value is the same on each device\n", "out = jax.pmap(lambda x: x + 1, out_axes=None)(xs)\n", "print(out)" ] }, { "cell_type": "markdown", "metadata": { "id": "pEEXyAODauyi" }, "source": [ "What happens when what we'd like to do isn't easily expressible as a mapped pure function? For example, what if we'd like to do a sum across the axis we're mapping over? JAX offers \"collectives\", functions that communicate across devices, to enable writing more interesting and complex distributed programs. To understand how exactly they work, we'll introduce SPMD." ] }, { "cell_type": "markdown", "metadata": { "id": "e4MmQQamg7Wn" }, "source": [ "# What is SPMD?" ] }, { "cell_type": "markdown", "metadata": { "id": "DJKyMsQNnFUI" }, "source": [ "Single-program multiple-data (SPMD) is a concurrent programming model in which a single program (i.e. the same code) is executed simultaneously across devices, but the inputs to each of the running programs can differ.\n", "\n", "If our program is a simple function of its inputs (i.e. something like `x + 5`), running a program in SPMD is just mapping it over different data, like we did with `jax.pmap` earlier. However, we can do more than just \"map\" a function. JAX offers \"collectives\", which are functions that communicate across devices.\n", "\n", "For example, maybe we'd like to take the sum of a quantity across all our devices. Before we do that, we need to assign a name to the axis we're mapping over in the `pmap`. We then use the `lax.psum` (\"parallel sum\") function to perform a sum across devices, ensuring we identify the named axis we're summing over." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "fsZfnmr3eMYj", "outputId": "ddc7f720-8158-4293-a78b-b3c38ffdbf47" }, "outputs": [ { "data": { "text/plain": [ "ShardedDeviceArray([28., 28., 28., 28., 28., 28., 28., 28.], dtype=float32)" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "def f(x):\n", " out = lax.psum(x, axis_name='i')\n", " return out\n", "xs = jnp.arange(8.) # Length of array matches number of devices\n", "jax.pmap(f, axis_name='i')(xs)" ] }, { "cell_type": "markdown", "metadata": { "id": "eF2Nv7KXeTrb" }, "source": [ "The `psum` collective aggregates the value of `x` on each device and synchronizes its value across the map i.e. `out` is `28.` on each device.\n", "We're no longer performing a simple \"map\", but we're executing an SPMD program where each device's computation can now interact with the same computation on other devices, albeit in a limited way using collectives. In this scenario, we can use `out_axes = None`, because `psum` will synchronize the value." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "n2dPYYQO5-vz", "outputId": "fd72f196-8d8a-423c-ec79-6e6067b16f59" }, "outputs": [ { "data": { "text/plain": [ "ShardedDeviceArray(28., dtype=float32)" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "def f(x):\n", " out = lax.psum(x, axis_name='i')\n", " return out\n", "jax.pmap(f, axis_name='i', out_axes=None)(jnp.arange(8.))" ] }, { "cell_type": "markdown", "metadata": { "id": "JsoseMKEf-nA" }, "source": [ "SPMD enables us to write one program that is run on every device in any TPU configuration simultaneously. The same code that is used to do machine learning on 8 TPU cores can be used on a TPU pod that may have hundreds to thousands of cores! For a more detailed tutorial about `jax.pmap` and SPMD, you can refer to the the [JAX 101 tutorial](https://jax.readthedocs.io/en/latest/jax-101/06-parallelism.html)." ] }, { "cell_type": "markdown", "metadata": { "id": "TpRG6flRfoed" }, "source": [ "# MCMC at scale" ] }, { "cell_type": "markdown", "metadata": { "id": "ulERQF93geL1" }, "source": [ "In this notebook, we focus on using Markov Chain Monte Carlo (MCMC) methods for Bayesian inference. There are may ways we utilize many devices for MCMC, but in this notebook, we'll focus on two:\n", "1. Running independent Markov chains on different devices. This case is fairly simple and is possible to do with vanilla TFP.\n", "2. Sharding a dataset across devices. This case is a bit more complex and requires recently added TFP machinery." ] }, { "cell_type": "markdown", "metadata": { "id": "QfSa12Hqhy09" }, "source": [ "## Independent Chains" ] }, { "cell_type": "markdown", "metadata": { "id": "hqos2uVgh2Zw" }, "source": [ "Say we'd like to do Bayesian inference on a problem using MCMC and would like to run several chains in parallel across several devices (say 2 on each device). This turns out to be a program we can just \"map\" across devices, i.e. one that needs no collectives. To make sure each program executes a different Markov chain (as opposed to running the same one), we pass in a different value for the random seed to each device.\n", "\n", "Let's try it on a toy problem of sampling from a 2-D Gaussian distribution. We can use TFP's existing MCMC functionality out of the box.\n", "In general, we try to put most of the logic inside of our mapped function to more explicitly distinguish between what's running on all the devices versus just the first." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "LQAqJ4O3h1oM" }, "outputs": [], "source": [ "def run(seed):\n", " target_log_prob = tfd.Sample(tfd.Normal(0., 1.), 2).log_prob\n", "\n", " initial_state = jnp.zeros([2, 2]) # 2 chains\n", " kernel = tfm.HamiltonianMonteCarlo(target_log_prob, 1e-1, 10)\n", " def trace_fn(state, pkr):\n", " return target_log_prob(state)\n", "\n", " states, log_prob = tfm.sample_chain(\n", " num_results=1000,\n", " num_burnin_steps=1000,\n", " kernel=kernel,\n", " current_state=initial_state,\n", " trace_fn=trace_fn,\n", " seed=seed\n", " )\n", " return states, log_prob" ] }, { "cell_type": "markdown", "metadata": { "id": "sFPWI5GijOj4" }, "source": [ "By itself, the `run` function takes in a stateless random seed (to see how stateless randomness work, you can read the [TFP on JAX](https://www.tensorflow.org/probability/examples/TensorFlow_Probability_on_JAX) notebook or see the [JAX 101 tutorial](https://jax.readthedocs.io/en/latest/jax-101/05-random-numbers.html)). Mapping `run` over different seeds will result in running several independent Markov chains." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "N3mzWLKPjpsB", "outputId": "3e508a4b-d8b4-4047-afaf-65960d6a7eeb" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(8, 1000, 2, 2) (8, 1000, 2)\n" ] } ], "source": [ "states, log_probs = jax.pmap(run)(random.split(random.PRNGKey(0), 8))\n", "print(states.shape, log_probs.shape)\n", "# states is (8 devices, 1000 samples, 2 chains, 2 dimensions)\n", "# log_prob is (8 devices, 1000 samples, 2 chains)" ] }, { "cell_type": "markdown", "metadata": { "id": "pt5aCj0Ej3pA" }, "source": [ "Note how we now have an extra axis corresponding to each device. We can rearrange the dimensions and flatten them to get an axis for the 16 chains." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "uz1etedpjw_f" }, "outputs": [], "source": [ "states = states.transpose([0, 2, 1, 3]).reshape([-1, 1000, 2])\n", "log_probs = log_probs.transpose([0, 2, 1]).reshape([-1, 1000])" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 322 }, "id": "rrw6e5rhkLYy", "outputId": "3f4ad6ed-0356-47da-fa1f-e09adb90242e" }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light", "tags": [] }, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(1, 2, figsize=(10, 5))\n", "ax[0].plot(log_probs.T, alpha=0.4)\n", "ax[1].scatter(*states.reshape([-1, 2]).T, alpha=0.1)\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": { "id": "q7hL_vVAkuaR" }, "source": [ "When running independent chains on many devices, it's as easy as `pmap`-ing over a function that uses `tfp.mcmc`, ensuring we pass different values for the random seed to each device." ] }, { "cell_type": "markdown", "metadata": { "id": "EfbcQmFzlYau" }, "source": [ "## Sharding data" ] }, { "cell_type": "markdown", "metadata": { "id": "iKWgLWSPlfOv" }, "source": [ "When we do MCMC, the target distribution is often a posterior distribution obtained by conditioning on a dataset, and computing an unnormalized log-density involves summing likelihoods for each observed data. \n", "\n", "With very large datasets, it can be prohibitively expensive to even run one chain on a single device. However, when we have access to multiple devices, we can split up the dataset across the devices to better leverage the compute we have available.\n", "\n", "If we'd like to do MCMC with a sharded dataset, we need to ensure the unnormalized log-density we compute on each device represents the *total*, i.e. the density over all data, otherwise each device will be doing MCMC with their own incorrect target distribution. To this end, TFP now has new tools (i.e. `tfp.experimental.distribute` and `tfp.experimental.mcmc`) that enable computing \"sharded\" log probabilities and doing MCMC with them." ] }, { "cell_type": "markdown", "metadata": { "id": "FTl2U7kCmhC_" }, "source": [ "### Sharded distributions" ] }, { "cell_type": "markdown", "metadata": { "id": "GgqqVZjlmj3w" }, "source": [ "The core abstraction TFP now provides for computing sharded log probabiliities is the `Sharded` meta-distribution, which takes a distribution as input and returns a new distribution that has specific properties when executed in an SPMD context. `Sharded` lives in `tfp.experimental.distribute`.\n", "\n", "Intuitively, a `Sharded` distribution corresponds to a set of random variables that have been \"split\" across devices. On each device, they will produce different samples, and can individually have different log-densities. Alternatively, a `Sharded` distribution corresponds to a \"plate\" in graphical model parlance, where the plate size is the number of devices.\n" ] }, { "cell_type": "markdown", "metadata": { "id": "wNlR4OaWm1w8" }, "source": [ "#### Sampling a `Sharded` distribution" ] }, { "cell_type": "markdown", "metadata": { "id": "MJ0oarZsnBHP" }, "source": [ "If we sample from a `Normal` distribution in a program being `pmap`-ed using the same seed on each device, we will get the same sample on each device. We can think of the following function as sampling a single random variable that is synchronized across devices." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "fyiT7iPQm3cK", "outputId": "325ca0cd-864f-4248-89fe-dea76645c167" }, "outputs": [ { "data": { "text/plain": [ "ShardedDeviceArray([-0.20584236, -0.20584236, -0.20584236, -0.20584236,\n", " -0.20584236, -0.20584236, -0.20584236, -0.20584236], dtype=float32)" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "# `pmap` expects at least one value to be mapped over, so we provide a dummy one\n", "def f(seed, _):\n", " return tfd.Normal(0., 1.).sample(seed=seed)\n", "jax.pmap(f, in_axes=(None, 0))(random.PRNGKey(0), jnp.arange(8.))" ] }, { "cell_type": "markdown", "metadata": { "id": "TVJ48nEdnYmB" }, "source": [ "If we wrap `tfd.Normal(0., 1.)` with a `tfed.Sharded`, we logically now have eight different random variables (one on each device) and will therefore produce a different sample for each one, despite passing in the same seed." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "f-W8UhTxnmmy", "outputId": "c39ef11d-ec42-4dff-f876-64320240288c" }, "outputs": [ { "data": { "text/plain": [ "ShardedDeviceArray([ 1.2152631 , 0.7818249 , 0.32549605, 0.6828047 ,\n", " 1.3973192 , -0.57830244, 0.37862757, 2.7706041 ], dtype=float32)" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "def f(seed, _):\n", " return tfed.Sharded(tfd.Normal(0., 1.), shard_axis_name='i').sample(seed=seed)\n", "jax.pmap(f, in_axes=(None, 0), axis_name='i')(random.PRNGKey(0), jnp.arange(8.))" ] }, { "cell_type": "markdown", "metadata": { "id": "BqTKixQfARlQ" }, "source": [ "An equivalent representation of this distribution on a single device is just a 8 independent normal samples. Even though the value of the sample will be different (`tfed.Sharded` does pseudo-random number generation slightly differently), they both represent the same distribution." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "YwYeBAJHAYqm", "outputId": "132b416d-adb1-42af-9e9c-4dea25da155d" }, "outputs": [ { "data": { "text/plain": [ "DeviceArray([ 0.08086783, -0.38624594, -0.3756545 , 1.668957 ,\n", " -1.2758069 , 2.1192007 , -0.85821325, 1.1305912 ], dtype=float32)" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "dist = tfd.Sample(tfd.Normal(0., 1.), jax.device_count())\n", "dist.sample(seed=random.PRNGKey(0))" ] }, { "cell_type": "markdown", "metadata": { "id": "7vV2kIhxom_9" }, "source": [ "#### Taking the log-density of a `Sharded` distribution" ] }, { "cell_type": "markdown", "metadata": { "id": "9NLOMW6Po3E6" }, "source": [ "Let's see what happens when we compute the log-density of a sample from a regular distribution in an SPMD context." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "Wo_1xYRForc_", "outputId": "239f8b9c-7a3d-4b4a-9534-6be1f1ae1bb2" }, "outputs": [ { "data": { "text/plain": [ "(ShardedDeviceArray([-0.20584236, -0.20584236, -0.20584236, -0.20584236,\n", " -0.20584236, -0.20584236, -0.20584236, -0.20584236], dtype=float32),\n", " ShardedDeviceArray([-0.94012403, -0.94012403, -0.94012403, -0.94012403,\n", " -0.94012403, -0.94012403, -0.94012403, -0.94012403], dtype=float32))" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "def f(seed, _):\n", " dist = tfd.Normal(0., 1.)\n", " x = dist.sample(seed=seed)\n", " return x, dist.log_prob(x)\n", "jax.pmap(f, in_axes=(None, 0))(random.PRNGKey(0), jnp.arange(8.))" ] }, { "cell_type": "markdown", "metadata": { "id": "N1liUvISo8Lm" }, "source": [ "Each sample is the same on each device, so we compute the same density on each device too. Intuitively, here we only have a distribution over a single normally distributed variable.\n", "\n", "With a `Sharded` distribution, we have a distribution over 8 random variables, so when we compute the `log_prob` of a sample, we sum, across devices, over each of the individual log densities. (You might notice that this total log_prob value is larger than the singleton log_prob computed above.)" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "i4__zjePpVul", "outputId": "228b3ed0-b1a3-4ff2-ec57-58351c5ee1ec" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Sample: [ 1.2152631 0.7818249 0.32549605 0.6828047 1.3973192 -0.57830244\n", " 0.37862757 2.7706041 ]\n", "Log Prob: [-13.7349205 -13.7349205 -13.7349205 -13.7349205 -13.7349205 -13.7349205\n", " -13.7349205 -13.7349205]\n" ] } ], "source": [ "def f(seed, _):\n", " dist = tfed.Sharded(tfd.Normal(0., 1.), shard_axis_name='i')\n", " x = dist.sample(seed=seed)\n", " return x, dist.log_prob(x)\n", "sample, log_prob = jax.pmap(f, in_axes=(None, 0), axis_name='i')(\n", " random.PRNGKey(0), jnp.arange(8.))\n", "print('Sample:', sample)\n", "print('Log Prob:', log_prob)" ] }, { "cell_type": "markdown", "metadata": { "id": "u8iNYSVCBSEX" }, "source": [ "The equivalent, \"unsharded\" distribution produces the same log density." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "UM0QvFGJA7BA", "outputId": "f8af42f8-bbbe-42c8-bfb4-20cfdbaaf660" }, "outputs": [ { "data": { "text/plain": [ "DeviceArray(-13.7349205, dtype=float32)" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "dist = tfd.Sample(tfd.Normal(0., 1.), jax.device_count())\n", "dist.log_prob(sample)" ] }, { "cell_type": "markdown", "metadata": { "id": "3d94dKwmpcFh" }, "source": [ "A `Sharded` distribution produces different values from `sample` on each device, but get the same value for `log_prob` on each device. What's happening here? A `Sharded` distribution does a `psum` internally to ensure the `log_prob` values are in sync across devices. Why would we want this behavior? If we're running the *same* MCMC chain on each device, we'd like the `target_log_prob` to be the same across each device, even if some random variables in the computation are sharded across devices." ] }, { "cell_type": "markdown", "metadata": { "id": "CJBg-y09jI4L" }, "source": [ "Additionally, a `Sharded` distribution ensures that gradients across devices are the correct, to ensure that algorithms like HMC, which take gradients of the log-density function as part of the transition function, produce proper samples." ] }, { "cell_type": "markdown", "metadata": { "id": "ospRyIUlqld2" }, "source": [ "### Sharded `JointDistribution`s" ] }, { "cell_type": "markdown", "metadata": { "id": "_dLU_CC9rime" }, "source": [ "We can create models with multiple `Sharded` random variables by using `JointDistribution`s (JDs). Unfortunately, `Sharded` distributions cannot be safely used with vanilla `tfd.JointDistribution`s, but `tfp.experimental.distribute` exports \"patched\" JDs that will behave like `Sharded` distributions." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "CZHp912esU1l", "outputId": "36ef8d60-fab8-4d15-9f65-936fc584c893" }, "outputs": [ { "data": { "text/plain": [ "([ShardedDeviceArray([1.6121525, 1.6121525, 1.6121525, 1.6121525, 1.6121525,\n", " 1.6121525, 1.6121525, 1.6121525], dtype=float32),\n", " ShardedDeviceArray([ 0.8690128 , -0.83167845, 1.2209264 , 0.88412696,\n", " 0.76478404, -0.66208494, -0.0129658 , 0.7391483 ], dtype=float32)],\n", " ShardedDeviceArray([-12.214451, -12.214451, -12.214451, -12.214451,\n", " -12.214451, -12.214451, -12.214451, -12.214451], dtype=float32))" ] }, "execution_count": 0, "metadata": { "tags": [] }, "output_type": "execute_result" } ], "source": [ "def f(seed, _):\n", " dist = tfed.JointDistributionSequential([\n", " tfd.Normal(0., 1.),\n", " tfed.Sharded(tfd.Normal(0., 1.), shard_axis_name='i'),\n", " ])\n", " x = dist.sample(seed=seed)\n", " return x, dist.log_prob(x)\n", "jax.pmap(f, in_axes=(None, 0), axis_name='i')(random.PRNGKey(0), jnp.arange(8.))" ] }, { "cell_type": "markdown", "metadata": { "id": "2Jem6vNLsjgm" }, "source": [ "These sharded JDs can have both `Sharded` and vanilla TFP distributions as components. For the unsharded distributions, we obtain the same sample on each device, and for the sharded distributions, we get different samples. The `log_prob` on each device is synchronized as well." ] }, { "cell_type": "markdown", "metadata": { "id": "yGa7snFos2de" }, "source": [ "### MCMC with `Sharded` distributions" ] }, { "cell_type": "markdown", "metadata": { "id": "voZoIikGs69-" }, "source": [ "How do we think about `Sharded` distributions in the context of MCMC? If we have a generative model that can be expressed as a `JointDistribution`, we can pick some axis of that model to \"shard\" across. Typically, one random variable in the model will correspond to observed data, and if we have a large dataset that we'd like to shard across devices, we want the variables that are associated to data points to be sharded as well. We also may have \"local\" random variables that are one-to-one with the observations we are sharding, so we will have to additionally shard those random variables.\n", "\n", "We'll go over examples of the usage of `Sharded` distributions with TFP MCMC in this section. We'll start with a simpler Bayesian logistic regression example, and conclude with a matrix factorization example, with the goal of demonstrating some use-cases for the `distribute` library.\n" ] }, { "cell_type": "markdown", "metadata": { "id": "tTDEJ-20t-RU" }, "source": [ "##### Example: Bayesian logistic regression for MNIST" ] }, { "cell_type": "markdown", "metadata": { "id": "Btckhsht7TOd" }, "source": [ "We'd like to do Bayesian logistic regression on a large dataset; the model has a prior $p(\\theta)$ over the regression weights, and a likelihood $p(y_i | \\theta, x_i)$ that is summed over all data $\\{x_i, y_i\\}_{i = 1}^N$ to obtain the total joint log density. If we shard our data, we'd shard the observed random variables $x_i$ and $y_i$ in our model. " ] }, { "cell_type": "markdown", "metadata": { "id": "wUeySkfVuqI-" }, "source": [ "We use the following Bayesian logistic regression model for MNIST classification:\n", "$$\n", "\\begin{align*}\n", "w &\\sim \\mathcal{N}(0, 1) \\\\\n", "b &\\sim \\mathcal{N}(0, 1) \\\\\n", "y_i | w, b, x_i &\\sim \\textrm{Categorical}(w^T x_i + b)\n", "\\end{align*}\n", "$$" ] }, { "cell_type": "markdown", "metadata": { "id": "Q16-CWWi7ju5" }, "source": [ "Let's load MNIST using TensorFlow Datasets." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 202, "referenced_widgets": [ "a302adcb8d824d00be0125331fe7df1a", "4da2d87727e942c9956a2c507109c107", "eb1aba62376345729c10817275841ad1", "9a546ada258846cbad21774eea3af7cc", "a019862f72874e7cbc77c57cc1e31f9c", "f48b175a5b8e469da891b76be737e8cd", "bbfbc2dbad584046835b3fcac9917d57", "b14fcb50f7644de7b40c89486ab999ba" ] }, "id": "Tzsu879puFmo", "outputId": "87f3bc0b-55f4-4d98-f24f-3993cef281a3" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[1mDownloading and preparing dataset mnist/3.0.1 (download: 11.06 MiB, generated: 21.00 MiB, total: 32.06 MiB) to /root/tensorflow_datasets/mnist/3.0.1...\u001b[0m\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "WARNING:absl:Dataset mnist is hosted on GCS. It will automatically be downloaded to your\n", "local data directory. If you'd instead prefer to read directly from our public\n", "GCS bucket (recommended if you're running on GCP), you can instead pass\n", "`try_gcs=True` to `tfds.load` or set `data_dir=gs://tfds-data/datasets`.\n", "\n" ] }, { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "a302adcb8d824d00be0125331fe7df1a", "version_major": 2, "version_minor": 0 }, "text/plain": [ "HBox(children=(FloatProgress(value=0.0, description='Dl Completed...', max=4.0, style=ProgressStyle(descriptio…" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "\n", "\n", "\u001b[1mDataset mnist downloaded and prepared to /root/tensorflow_datasets/mnist/3.0.1. Subsequent calls will reuse this data.\u001b[0m\n" ] } ], "source": [ "mnist = tfds.as_numpy(tfds.load('mnist', batch_size=-1))\n", "raw_train_images, train_labels = mnist['train']['image'], mnist['train']['label']\n", "train_images = raw_train_images.reshape([raw_train_images.shape[0], -1]) / 255.\n", "\n", "raw_test_images, test_labels = mnist['test']['image'], mnist['test']['label']\n", "test_images = raw_test_images.reshape([raw_test_images.shape[0], -1]) / 255." ] }, { "cell_type": "markdown", "metadata": { "id": "dTXbO3FyuoqE" }, "source": [ "We have 60000 training images but let's take advantage of our 8 available cores and split it 8 ways. We'll use this handy `shard` utility function." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "cRvMbzl8vO3h" }, "outputs": [], "source": [ "def shard_value(x):\n", " x = x.reshape((jax.device_count(), -1, *x.shape[1:]))\n", " return jax.pmap(lambda x: x)(x) # pmap will physically place values on devices\n", "\n", "shard = functools.partial(jax.tree.map, shard_value)" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "w7MneZdGuTvk", "outputId": "44abc612-258a-463b-ee14-bef65c1c23b6" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(8, 7500, 784) (8, 7500)\n" ] } ], "source": [ "sharded_train_images, sharded_train_labels = shard((train_images, train_labels))\n", "print(sharded_train_images.shape, sharded_train_labels.shape)" ] }, { "cell_type": "markdown", "metadata": { "id": "h4cSdTI21z_W" }, "source": [ "Before we continue, let's quickly discuss precision on TPUs and its impact on HMC. TPUs execute matrix multiplications using low `bfloat16` precision for speed. `bfloat16` matrix multiplications are often sufficient for many deep learning applications, but when used with HMC, we have empirically found the lower precision can lead to diverging trajectories, causing rejections. We can use higher precision matrix multiplications, at the cost of some additional compute.\n", "\n", "To increase our matmul precision, we can use the `jax.default_matmul_precision` decorator with `\"tensorfloat32\"` precision (for even higher precision we could use `\"float32\"` precision)." ] }, { "cell_type": "markdown", "metadata": { "id": "67WV7IcdvUtg" }, "source": [ "Let's now define our `run` function, which will take in a random seed (which will be the same on each device) and a shard of MNIST. The function will implement the aforementioned model and we will then use TFP's vanilla MCMC functionality to run a single chain. We'll make sure to decorate `run` with the `jax.default_matmul_precision` decorator to make sure the matrix multiplication is run with higher precision, though in the particular example below, we could just as well use `jnp.dot(images, w, precision=lax.Precision.HIGH)`." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "VuJ9Um1xvSPt" }, "outputs": [], "source": [ "# We can use `out_axes=None` in the `pmap` because the results will be the same\n", "# on every device. \n", "@functools.partial(jax.pmap, axis_name='data', in_axes=(None, 0), out_axes=None)\n", "@jax.default_matmul_precision('tensorfloat32')\n", "def run(seed, data):\n", " images, labels = data # a sharded dataset\n", " num_examples, dim = images.shape\n", " num_classes = 10\n", "\n", " def model_fn():\n", " w = yield Root(tfd.Sample(tfd.Normal(0., 1.), [dim, num_classes]))\n", " b = yield Root(tfd.Sample(tfd.Normal(0., 1.), [num_classes]))\n", " logits = jnp.dot(images, w) + b\n", " yield tfed.Sharded(tfd.Independent(tfd.Categorical(logits=logits), 1),\n", " shard_axis_name='data')\n", " model = tfed.JointDistributionCoroutine(model_fn)\n", "\n", " init_seed, sample_seed = random.split(seed)\n", "\n", " initial_state = model.sample(seed=init_seed)[:-1] # throw away `y`\n", "\n", " def target_log_prob(*state):\n", " return model.log_prob((*state, labels))\n", "\n", " def accuracy(w, b):\n", " logits = images.dot(w) + b\n", " preds = logits.argmax(axis=-1)\n", " # We take the average accuracy across devices by using `lax.pmean`\n", " return lax.pmean((preds == labels).mean(), 'data')\n", "\n", " kernel = tfm.HamiltonianMonteCarlo(target_log_prob, 1e-2, 100)\n", " kernel = tfm.DualAveragingStepSizeAdaptation(kernel, 500)\n", " def trace_fn(state, pkr):\n", " return (\n", " target_log_prob(*state),\n", " accuracy(*state),\n", " pkr.new_step_size)\n", " states, trace = tfm.sample_chain(\n", " num_results=1000,\n", " num_burnin_steps=1000,\n", " current_state=initial_state,\n", " kernel=kernel,\n", " trace_fn=trace_fn,\n", " seed=sample_seed\n", " )\n", " return states, trace" ] }, { "cell_type": "markdown", "metadata": { "id": "pMq6alT_OmCC" }, "source": [ "`jax.pmap` includes a JIT compile but the compiled function is cached after the first call. We'll call `run` and ignore the output to cache the compilation." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "LYsRQkTUxCPD", "outputId": "06062688-9df6-4a98-d526-a1a639ba692b" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 24.5 s, sys: 48.2 s, total: 1min 12s\n", "Wall time: 1min 54s\n" ] } ], "source": [ "%%time\n", "output = run(random.PRNGKey(0), (sharded_train_images, sharded_train_labels))\n", "jax.tree.map(lambda x: x.block_until_ready(), output)" ] }, { "cell_type": "markdown", "metadata": { "id": "AZhCrhKqOxkc" }, "source": [ "We'll now call `run` again to see how long the actual execution takes." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "5SuYWRhUDGWB", "outputId": "88447cdc-6988-47d1-bd8c-1233bf09af51" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 13.1 s, sys: 45.2 s, total: 58.3 s\n", "Wall time: 1min 43s\n" ] } ], "source": [ "%%time\n", "states, trace = run(random.PRNGKey(0), (sharded_train_images, sharded_train_labels))\n", "jax.tree.map(lambda x: x.block_until_ready(), trace)" ] }, { "cell_type": "markdown", "metadata": { "id": "6njumqgPO1zG" }, "source": [ "We're executing 200,000 leapfrog steps, each of which computes a gradient over the entire dataset. Splitting the computation over 8 cores enables us to compute the equivalent of 200,000 epochs of training in about 95 seconds, about 2,100 epochs per second!" ] }, { "cell_type": "markdown", "metadata": { "id": "vWciJ-eLPRkr" }, "source": [ "Let's plot the log-density of each sample and each sample's accuracy:" ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 338 }, "id": "2PNzYZOXxd1R", "outputId": "6219df98-19ea-4c00-e82e-711cbc19e540" }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light", "tags": [] }, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(1, 3, figsize=(15, 5))\n", "ax[0].plot(trace[0])\n", "ax[0].set_title('Log Prob')\n", "ax[1].plot(trace[1])\n", "ax[1].set_title('Accuracy')\n", "ax[2].plot(trace[2])\n", "ax[2].set_title('Step Size')\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": { "id": "dTz4X8_8PXZP" }, "source": [ "If we ensemble the samples, we can compute a Bayesian model average to improve our performance." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "mP5Vd7w4PbJ2", "outputId": "4b04c2e8-9436-41bc-9884-6b27f95986e4" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Average Accuracy: 0.9188529253005981\n", "BMA Accuracy: 0.9264000058174133\n", "Accuracy Improvement: 0.0075470805168151855\n" ] } ], "source": [ "@functools.partial(jax.pmap, axis_name='data', in_axes=(0, None), out_axes=None)\n", "def bayesian_model_average(data, states):\n", " images, labels = data\n", " logits = jax.vmap(lambda w, b: images.dot(w) + b)(*states)\n", " probs = jax.nn.softmax(logits, axis=-1)\n", " bma_accuracy = (probs.mean(axis=0).argmax(axis=-1) == labels).mean()\n", " avg_accuracy = (probs.argmax(axis=-1) == labels).mean()\n", " return lax.pmean(bma_accuracy, axis_name='data'), lax.pmean(avg_accuracy, axis_name='data')\n", "\n", "sharded_test_images, sharded_test_labels = shard((test_images, test_labels))\n", "bma_acc, avg_acc = bayesian_model_average((sharded_test_images, sharded_test_labels), states)\n", "print(f'Average Accuracy: {avg_acc}')\n", "print(f'BMA Accuracy: {bma_acc}')\n", "print(f'Accuracy Improvement: {bma_acc - avg_acc}')" ] }, { "cell_type": "markdown", "metadata": { "id": "c2CjiAMgQ9Vd" }, "source": [ "A Bayesian model average increases our accuracy by almost 1%!" ] }, { "cell_type": "markdown", "metadata": { "id": "d9CYg3YC6XbK" }, "source": [ "##### Example: MovieLens recommendation system" ] }, { "cell_type": "markdown", "metadata": { "id": "kMd9c6SI7rm7" }, "source": [ "Let's now try doing inference with the MovieLens recommendations dataset, which is a collection of users and their ratings of various movies. Specifically, we can represent MovieLens as an $N \\times M$ watch matrix $W$ where $N$ is the number of users and $M$ is the number of movies; we expect $N > M$. The entries of $W_{ij}$ are a boolean indicating whether or not user $i$ watched movie $j$. Note that MovieLens provides user ratings, but we're ignoring them to simplify the problem.\n", "\n", "\n", "First, we'll load the dataset. We'll use the version with 1 million ratings." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 298, "referenced_widgets": [ "0509d5f63e8447c5bf3047dec3c9b436", "62ff554eba004da3809a1dd2e21311fb", "74764971a48a4f79a495f1132d805db0", "30b99f5c7029409ebca66ffedb065879", "76f980ecbd2b4b49a7c44b683f74e8c8", "cf5ca2562e75411c9e14dcb5d58736ed", "a03a5d0680884fa3a77c4b95ebfa0b98", "0033084241f445c9a6815a48a868d64b", "24c536ef29c547ef965bb3270637bcc2", "33822c1d39ca42f5b5a44f330ba19df3", "8074ff27fdd24ba6844f3a3f4eda7aed", "a1966d13e72d4a9f9fcda67545e8fee3", "f7503cf63443454a83edfeaa1bc3ca59", "6bd23dc4d2f04752a269bf7f3a25e10b", "ee5454904d4649e1a23c7b29eee96b27", "10eb91732d5b430481b75769cfdf6100", "95e6e8cbf69f495eb7fb6c84f2781d12", "cd2ca5c8fc214489a02e62e7750df78d", "56b9e4a682864227a14f95274afea6c8", "912901341ef44cf385d366dd62d64016", "e177f0e205af4369ad4e16af5c5f5f6a", "57ffaf62ae774f0cbcc3526d12902a76", "c400310f418546909ba8ffe2e10e3e12", "b4dbdb532ffd4b46ba6ebac9f1366b14", "445cf583fac44e3895e51e2cde2a350c", "cd2ab853d5d143daae3516a910d2cd87", "dd3fe337112740738f3b9fd788ba9405", "3ee6fe19d17d4df99ffeb289775b1c2e", "698a2d7c08864520af367c353735814f", "7154891fb6ac45c4a24e151af3fb6903", "3c0c2c21823b4206b1e217601b8b2533", "c2e9d94dd4cd459da8cd6c5d253b7ed5", "48ec20644a394848927cf72e8bea86b5", "4fc929851e25400eaa789a41aead047b", "55840316f537476b81ffe69cec8b7092", "a250abd4702646e8a30b3ff946418bc7", "ef4b4c0fa6be4907a9e86c77eb4249f6", "5f993f93d9764230b0fd70f0945a3739", "bc2e9dd37de44b5192881851fc42099d", "e3f937fc3dfb4486bfd7211b52877a80" ] }, "id": "8Ra9LNDC-bF9", "outputId": "16054a5b-648f-49e5-b51d-dfaaf303b025" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[1mDownloading and preparing dataset movielens/1m-ratings/0.1.0 (download: Unknown size, generated: Unknown size, total: Unknown size) to /root/tensorflow_datasets/movielens/1m-ratings/0.1.0...\u001b[0m\n" ] }, { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "0509d5f63e8447c5bf3047dec3c9b436", "version_major": 2, "version_minor": 0 }, "text/plain": [ "HBox(children=(FloatProgress(value=1.0, bar_style='info', description='Dl Completed...', max=1.0, style=Progre…" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "24c536ef29c547ef965bb3270637bcc2", "version_major": 2, "version_minor": 0 }, "text/plain": [ "HBox(children=(FloatProgress(value=1.0, bar_style='info', description='Dl Size...', max=1.0, style=ProgressSty…" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "95e6e8cbf69f495eb7fb6c84f2781d12", "version_major": 2, "version_minor": 0 }, "text/plain": [ "HBox(children=(FloatProgress(value=1.0, bar_style='info', description='Extraction completed...', max=1.0, styl…" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "\n", "\n", "\n", "\n", "\n", "\n" ] }, { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "445cf583fac44e3895e51e2cde2a350c", "version_major": 2, "version_minor": 0 }, "text/plain": [ "HBox(children=(FloatProgress(value=1.0, bar_style='info', max=1.0), HTML(value='')))" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "\r", "Shuffling and writing examples to /root/tensorflow_datasets/movielens/1m-ratings/0.1.0.incompleteYKA3TG/movielens-train.tfrecord\n" ] }, { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "48ec20644a394848927cf72e8bea86b5", "version_major": 2, "version_minor": 0 }, "text/plain": [ "HBox(children=(FloatProgress(value=0.0, max=1000209.0), HTML(value='')))" ] }, "metadata": { "tags": [] }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "\u001b[1mDataset movielens downloaded and prepared to /root/tensorflow_datasets/movielens/1m-ratings/0.1.0. Subsequent calls will reuse this data.\u001b[0m\n", "\r" ] } ], "source": [ "movielens = tfds.as_numpy(tfds.load('movielens/1m-ratings', batch_size=-1))\n", "GENRES = ['Action', 'Adventure', 'Animation', 'Children', 'Comedy',\n", " 'Crime', 'Documentary', 'Drama', 'Fantasy', 'Film-Noir',\n", " 'Horror', 'IMAX', 'Musical', 'Mystery', 'Romance', 'Sci-Fi',\n", " 'Thriller', 'Unknown', 'War', 'Western', '(no genres listed)'] " ] }, { "cell_type": "markdown", "metadata": { "id": "U5opyYKv-ioM" }, "source": [ "We'll do some preprocessing of the dataset to obtain the watch matrix $W$." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "OGK5nn3u-ltG", "outputId": "9ebed1be-b197-40af-85eb-117e5b3bb89e" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(6040, 3706)\n" ] } ], "source": [ "raw_movie_ids = movielens['train']['movie_id']\n", "raw_user_ids = movielens['train']['user_id']\n", "genres = movielens['train']['movie_genres']\n", "\n", "movie_ids, movie_labels = pd.factorize(movielens['train']['movie_id'])\n", "user_ids, user_labels = pd.factorize(movielens['train']['user_id'])\n", "\n", "num_movies = movie_ids.max() + 1\n", "num_users = user_ids.max() + 1\n", "\n", "movie_titles = dict(zip(movielens['train']['movie_id'],\n", " movielens['train']['movie_title']))\n", "movie_genres = dict(zip(movielens['train']['movie_id'],\n", " genres))\n", "movie_id_to_title = [movie_titles[movie_labels[id]].decode('utf-8')\n", " for id in range(num_movies)]\n", "movie_id_to_genre = [GENRES[movie_genres[movie_labels[id]][0]] for id in range(num_movies)]\n", "\n", "watch_matrix = np.zeros((num_users, num_movies), bool)\n", "watch_matrix[user_ids, movie_ids] = True\n", "print(watch_matrix.shape)" ] }, { "cell_type": "markdown", "metadata": { "id": "H8hKi0Aw-KVT" }, "source": [ "We can define a generative model for $W$, using a simple probabilistic matrix factorization model. We assume a latent $N \\times D$ user matrix $U$ and a latent $M \\times D$ movie matrix $V$, which when multiplied produce the logits of a Bernoulli for the watch matrix $W$. We'll also include a bias vectors for users and movies, $u$ and $v$.\n", "$$\n", "\\begin{align*}\n", " U &\\sim \\mathcal{N}(0, 1) \\quad \n", " u \\sim \\mathcal{N}(0, 1)\\\\\n", " V &\\sim \\mathcal{N}(0, 1) \\quad\n", " v \\sim \\mathcal{N}(0, 1)\\\\\n", " W_{ij} &\\sim \\textrm{Bernoulli}\\left(\\sigma\\left(\\left(UV^T\\right)_{ij} + u_i + v_j\\right)\\right)\n", "\\end{align*}\n", "$$" ] }, { "cell_type": "markdown", "metadata": { "id": "la6o8iam9nTN" }, "source": [ "This is a pretty big matrix; 6040 user and 3706 movies leads to a matrix with over 22 million entries in it.\n", "How do we approach sharding this model? Well, if we assume that $N > M$ (i.e. there are more users than movies), then it would make sense to shard the watch matrix across the user axis, so each device would have a chunk of watch matrix corresponding to a subset of users. Unlike the previous example, however, we'll also have to shard up the $U$ matrix, since it has an embedding for each user, so each device will be responsible for a shard of $U$ and a shard of $W$. On the other hand, $V$ will be unsharded and be synchronized across devices." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "4SUlkEg__ElT" }, "outputs": [], "source": [ "sharded_watch_matrix = shard(watch_matrix)" ] }, { "cell_type": "markdown", "metadata": { "id": "zdZgFkzH_BYI" }, "source": [ "Before we write our `run`, let's quickly discuss the additional challenges with sharding the local random variable $U$. When running HMC, the vanilla `tfp.mcmc.HamiltonianMonteCarlo` kernel will sample momenta for each element of the chain's state. Previously, only unsharded random variables were part of that state, and the momenta were the same on each device. When we now have a sharded $U$, we need to sample different momenta on each device for $U$, while sampling the same momenta for $V$. To accomplish this, we can use `tfp.experimental.mcmc.PreconditionedHamiltonianMonteCarlo` with a `Sharded` momentum distribution. As we continue to make parallel computation first-class, we may simplify this, e.g. by taking a shardedness indicator to the HMC kernel." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "XEPHnzdo-_G2" }, "outputs": [], "source": [ "def make_run(*,\n", " axis_name,\n", " dim=20,\n", " num_chains=2,\n", " prior_variance=1.,\n", " step_size=1e-2,\n", " num_leapfrog_steps=100,\n", " num_burnin_steps=1000,\n", " num_results=500,\n", " ):\n", " @functools.partial(jax.pmap, in_axes=(None, 0), axis_name=axis_name)\n", " @jax.default_matmul_precision('tensorfloat32')\n", " def run(key, watch_matrix):\n", " num_users, num_movies = watch_matrix.shape\n", "\n", " Sharded = functools.partial(tfed.Sharded, shard_axis_name=axis_name)\n", "\n", " def prior_fn():\n", " user_embeddings = yield Root(Sharded(tfd.Sample(tfd.Normal(0., 1.), [num_users, dim]), name='user_embeddings'))\n", " user_bias = yield Root(Sharded(tfd.Sample(tfd.Normal(0., 1.), [num_users]), name='user_bias'))\n", " movie_embeddings = yield Root(tfd.Sample(tfd.Normal(0., 1.), [num_movies, dim], name='movie_embeddings'))\n", " movie_bias = yield Root(tfd.Sample(tfd.Normal(0., 1.), [num_movies], name='movie_bias'))\n", " return (user_embeddings, user_bias, movie_embeddings, movie_bias)\n", " prior = tfed.JointDistributionCoroutine(prior_fn)\n", "\n", " def model_fn():\n", " user_embeddings, user_bias, movie_embeddings, movie_bias = yield from prior_fn()\n", " logits = (jnp.einsum('...nd,...md->...nm', user_embeddings, movie_embeddings)\n", " + user_bias[..., :, None] + movie_bias[..., None, :])\n", " yield Sharded(tfd.Independent(tfd.Bernoulli(logits=logits), 2), name='watch')\n", " model = tfed.JointDistributionCoroutine(model_fn)\n", "\n", " init_key, sample_key = random.split(key)\n", " initial_state = prior.sample(seed=init_key, sample_shape=num_chains)\n", "\n", " def target_log_prob(*state):\n", " return model.log_prob((*state, watch_matrix))\n", "\n", " momentum_distribution = tfed.JointDistributionSequential([\n", " Sharded(tfd.Independent(tfd.Normal(jnp.zeros([num_chains, num_users, dim]), 1.), 2)),\n", " Sharded(tfd.Independent(tfd.Normal(jnp.zeros([num_chains, num_users]), 1.), 1)),\n", " tfd.Independent(tfd.Normal(jnp.zeros([num_chains, num_movies, dim]), 1.), 2),\n", " tfd.Independent(tfd.Normal(jnp.zeros([num_chains, num_movies]), 1.), 1),\n", " ])\n", " \n", " # We pass in momentum_distribution here to ensure that the momenta for \n", " # user_embeddings and user_bias are also sharded\n", " kernel = tfem.PreconditionedHamiltonianMonteCarlo(target_log_prob, step_size,\n", " num_leapfrog_steps,\n", " momentum_distribution=momentum_distribution)\n", "\n", " num_adaptation_steps = int(0.8 * num_burnin_steps)\n", " kernel = tfm.DualAveragingStepSizeAdaptation(kernel, num_adaptation_steps)\n", "\n", " def trace_fn(state, pkr):\n", " return {\n", " 'log_prob': target_log_prob(*state),\n", " 'log_accept_ratio': pkr.inner_results.log_accept_ratio,\n", " }\n", " return tfm.sample_chain(\n", " num_results, initial_state,\n", " kernel=kernel,\n", " num_burnin_steps=num_burnin_steps,\n", " trace_fn=trace_fn,\n", " seed=sample_key)\n", " return run" ] }, { "cell_type": "markdown", "metadata": { "id": "3cUqCRXSWtOg" }, "source": [ "We'll again run it once to cache the compiled `run`." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "gNQFjD0lBA7N", "outputId": "ed5b6434-1129-4709-830c-f06807488018" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 56 s, sys: 1min 24s, total: 2min 20s\n", "Wall time: 3min 35s\n" ] } ], "source": [ "%%time\n", "run = make_run(axis_name='data')\n", "output = run(random.PRNGKey(0), sharded_watch_matrix)\n", "jax.tree.map(lambda x: x.block_until_ready(), output)" ] }, { "cell_type": "markdown", "metadata": { "id": "hBLG3HwyWx3c" }, "source": [ "Now we'll run it again without the compilation overhead." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "tfwsOUhmBMNB", "outputId": "8a4f5fb3-5cf4-44e0-94bc-e7429f003fe2" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 28.8 s, sys: 1min 16s, total: 1min 44s\n", "Wall time: 3min 1s\n" ] } ], "source": [ "%%time\n", "states, trace = run(random.PRNGKey(0), sharded_watch_matrix)\n", "jax.tree.map(lambda x: x.block_until_ready(), trace)" ] }, { "cell_type": "markdown", "metadata": { "id": "fl1uGAVWW1sR" }, "source": [ "Looks like we completed about 150,000 leapfrog steps in about 3 minutes, so about 83 leapfrog steps per second! Let's plot the accept ratio and log density of our samples." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 338 }, "id": "1wJrfbgYBS0t", "outputId": "232e5062-cddf-4779-a2ae-e4ac6ca22a67" }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light", "tags": [] }, "output_type": "display_data" } ], "source": [ "fig, axs = plt.subplots(1, len(trace), figsize=(5 * len(trace), 5))\n", "for ax, (key, val) in zip(axs, trace.items()):\n", " ax.plot(val[0]) # Indexing into a sharded array, each element is the same\n", " ax.set_title(key);" ] }, { "cell_type": "markdown", "metadata": { "id": "zGX10FOXYfvl" }, "source": [ "Now that we have some samples from our Markov chain, let's use them to make some predictions. First, let's extract each of the components. Remember that the `user_embeddings` and `user_bias` are split across device, so we need to concatenate our `ShardedArray` to obtain them all. On the other hand, `movie_embeddings` and `movie_bias` are the same on every device, so we can just pick the value from the first shard. We'll use regular `numpy` to copy the values from the TPUs back to CPU." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "OtN1xACUYw5O", "outputId": "71ea5fdf-390b-4ae1-e332-d7ab46e7536c" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "User embeddings: (500, 2, 6040, 20)\n", "User bias: (500, 2, 6040)\n", "Movie embeddings: (500, 2, 3706, 20)\n", "Movie bias: (500, 2, 3706)\n" ] } ], "source": [ "user_embeddings = np.concatenate(np.array(states.user_embeddings, np.float32), axis=2)\n", "user_bias = np.concatenate(np.array(states.user_bias, np.float32), axis=2)\n", "movie_embeddings = np.array(states.movie_embeddings[0], dtype=np.float32)\n", "movie_bias = np.array(states.movie_bias[0], dtype=np.float32)\n", "samples = (user_embeddings, user_bias, movie_embeddings, movie_bias)\n", "print(f'User embeddings: {user_embeddings.shape}')\n", "print(f'User bias: {user_bias.shape}')\n", "print(f'Movie embeddings: {movie_embeddings.shape}')\n", "print(f'Movie bias: {movie_bias.shape}')" ] }, { "cell_type": "markdown", "metadata": { "id": "Zn0qF_1uc17L" }, "source": [ "Let's try to build a simple recommender system that utilizes the uncertainty captured in these samples. Let's first write a function that ranks movies according to the watch probability." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "q-cc2vAfZYyz" }, "outputs": [], "source": [ "@jax.jit\n", "def recommend(sample, user_id):\n", " user_embeddings, user_bias, movie_embeddings, movie_bias = sample\n", " movie_logits = (\n", " jnp.einsum('d,md->m', user_embeddings[user_id], movie_embeddings)\n", " + user_bias[user_id] + movie_bias)\n", " return movie_logits.argsort()[::-1]" ] }, { "cell_type": "markdown", "metadata": { "id": "2sRtm0NaeTFG" }, "source": [ "We can now write a function that loops over all the samples and for each one, picks the top ranked movie that the user hasn't watched already. We can then see the counts of all recommended movies across the samples." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "id": "hfAjsSmueSZH" }, "outputs": [], "source": [ "def get_recommendations(user_id): \n", " movie_ids = []\n", " already_watched = set(jnp.arange(num_movies)[watch_matrix[user_id] == 1])\n", " for i in range(500):\n", " for j in range(2):\n", " sample = jax.tree.map(lambda x: x[i, j], samples)\n", " ranking = recommend(sample, user_id)\n", " for movie_id in ranking:\n", " if int(movie_id) not in already_watched:\n", " movie_ids.append(movie_id)\n", " break\n", " return movie_ids\n", "\n", "def plot_recommendations(movie_ids, ax=None):\n", " titles = collections.Counter([movie_id_to_title[i] for i in movie_ids])\n", " ax = ax or plt.gca()\n", " names, counts = zip(*sorted(titles.items(), key=lambda x: -x[1]))\n", " ax.bar(names, counts)\n", " ax.set_xticklabels(names, rotation=90)" ] }, { "cell_type": "markdown", "metadata": { "id": "XLmgJwXYecl9" }, "source": [ "Let's take the user who has seen the most movies versus the one who has seen the least." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "ncwX6MGSfVqf", "outputId": "877fa292-0165-45c2-8075-be602baf75ec" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "2314 20\n" ] } ], "source": [ "user_watch_counts = watch_matrix.sum(axis=1)\n", "user_most = user_watch_counts.argmax()\n", "user_least = user_watch_counts.argmin()\n", "print(user_watch_counts[user_most], user_watch_counts[user_least])" ] }, { "cell_type": "markdown", "metadata": { "id": "e2UfkAagel78" }, "source": [ "We hope our system has more certainty about `user_most` than `user_least`, given that we have more information about what sorts of movies `user_most` is more likely to watch." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 791 }, "id": "CS0clOvEelD6", "outputId": "2109267c-9a9f-451a-cb97-8ccefba2ae5d" }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAABIgAAAPaCAYAAAD8zWOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzde7hmVX0n+C9UgTCS4ABegtwSY/1yq1ZBW5KgMZ1O25lOPZiYRpkAGjsxqA3T02Ta6GhMOq3NGJjECwamM0kcoHFCJ4J0LjhMxxiC2oktJhD9QYxA4Y2bJiERi1Qxf7y7xmNRdeoUdW5vrc/nec7znr3W2u/+nQt1Ft+99t4HPfLIIwEAAABgXAevdQEAAAAArC0BEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABS1JVL6iqu/dj/0ur6o3LWdP0vgdV1a9V1Rer6r8u9/sDAKwlc7DVU1UfqKofX+s6YK1sXOsCYD2pqjuSPDnJ9iQPJvm9JP+yux9cw7LmTlW9PMmPd/dpO9u6+9wVOtxpSb4/yXHd/bcrdIwDzu5+RgCwVszBloc52PpWVScl+XSSQ7r779e4HHgUK4jg0bZ09xFJnpnkWUlet8b1sLgTk9zxWCYmVbUmIfl0xs2/vwDwtczB5os5GBxgrCCCPejuz1fV9ZlNUpIkVXVqkv89ybcluTPJ/9TdH5j6jkpycZIXJjk8yR9094umvp9I8tokRyW5Mcm53f3Zqe+RJK9J8j8neUqSX0ry60kuT/IdmZ1BO6u7t1XVC5JckeTtSX4qs7Nsr0qybdrvmCQXdfdbpvc+OMm/SfITSZ6Q5P+djv3AgjMYL0/y80n+uyS/2N1vnvY9PMkvJzk9yeeS/NrC709V/fT0vk9KsjXJ/9rd762qb01yaZJDqurBJH/f3U+oql9Pcnd3v2GJ35NXJbkgyROTXJnZWcRHdqnhXyS5ZMGxLu7uNy3hvf9lkn+V2b+B37jLe74gyRXdfdyCtjsyOxt3Q1X9wyTvSrIpyZeTXNnd/3oat9jvxweS/FGSFyQ5OcnmJH+R3ZjG3pjkHyX5B0l+P7Of09uTbEnSSf55d98xjf+uJG+barptOu5NU9/Lk/zM9H28L8kbkvy33f2MdlcLAKw2czBzsAVtd2QV52C7qqpXJPlfMvv9+K9JXtndd059b0vyw0mOTHJ7kn/V3X849e2p1g9Ob/2lqkqS7+/uDy2lFlgN0lPYg6o6LskPZPoDUlVPTfLbSf5dZn/0firJb1bVE6ddLs/sD/y3Z/YH+xen/f5Rkn+f5Iwk35DZH6337HK4FyY5JcmpmU0m/o8kZyU5PrMJypkLxj4lyWFJnprZ//j/h2nsKUmel+SNVbXzD+55SV6U5HuSHJvki5n9MV/otCSV5PuS/Mw0uUiSNyV52vTxwiQv22W/T03HOzLJzyW5oqq+obs/keTcJB/q7iN2Fzws8Xvyg0mek1lAcsZUw9fo7v9zl2O9aYnv/aIkz81sErGv3pbkbd399Zl9b35j+pr29vuRJGcneWWSr5vqWsxLp/FPnY7zocwmiEcl+URmP5+dk+LfzmzCenRmk6Pfrqqjq+rxU/sPdPfXJfmuJDcv5WcEAGvFHMwcbA9Waw6W6X1PT/L6zEKgJyb5wyRXLRjyx5mFmEcl+Y9Jrq6qwxarNcnzp9cnTN834RDrihVE8GjXTGc4jkjyXzL9j3hmE4Df6e7fmbb/n6r6kyT/Q1W9P7OJzNHd/cWp/w+m1x9N8qvd/d+SpKpel+SLVXXSzhUgSd7a3X+d5NaquiXJ+7v7L6fxv5vZMut3T2MfTvLm7t5eVe/JbCLztu7+m2n/P0/yjMzOTJ2b2Vmfu6f3+tkkd1XV2Qu+3p/r7i8n+XhVfXza9xOZ/XF/dXc/kOSBqnp7ZpOhJEl3X73gPf7v6ev6h0muXcL3eCnfkwu7+0uZnWH5/cz+AP/eMr33v5++rsfi4STfXFXHdPd9ST48te/x9yNf/dn9enffusTj/Fp3f2r6Gn43ybd19w3T9tWZnXFMkn+W5Pbuvnzavqqqzs9spdHVSXYk+Y6ququ7P5fZmUgAWI/MwczBFrNac7Cdzp3q/cT09bwlyeur6sTuvrO7r1gw9uKqekNmgd/HF6kV1jUriODRXjSttnhBkm/JbMlwMrvO+p9X1Zd2fmR25ucbMjvL9MCCiclCx2bBmYqe3Wzx/szOPu30hQWff3k320cs2L6/u7cv6Nvd/jvHn5jkvQvq/URmS6KfvGD85xd8/ncL9j02s2XLO33N2ZaqOqeqbl7w3t+Rr36v9mYp35M91bUc77111532wb/IbLnwJ6vqj6vqB6f2xX4/Hstxl/o78TVf7+TOJE+d7gnwkswmOJ+rqt+uqm/ZhxoAYDWZg321bnOwR1utOdhOJyZ524L3fCDJQZm+nqr6qar6RFX91dR/ZL76c9hTrbCuWUEEe9Ddf1Cza7Yvymw57NYkl3f3T+w6tqq+IclRVfWE6YzLQp/N7A/MzrGPz+xSoM+sVO0LbE3yiu7+o107anb9+2I+l9mka+fZlhMW7HtiZsuqvy+zpcXbq+rmzP5oJsnXXKe+Gyv5PVnKey9W399mtkx95/4bMltWnCTp7tuTnFmzewv8cJL/VFVHZ5HfjyUe97H6mq93ckKmM33dfX2S62t2P4N/l9nP7XkrVAsA7DdzMHOwaf+1noNtzWzF2JW7dlTV8zK7JPH7ktza3Tuq6ouZfg6L1Gr+xbomIILF/VKSO6rqGZndmPCPq+qFSW5Ickhm16v/RXffPS1DfldVvSazx7N+Z3d/MLNrla+qqv+Y2dmjtyT5yIKltivp0iRvrqqXdfedNbsW+7u6eylLkH8jyeuq6iNJHp/ZtfQ7PT6zP3D3JklV/VhmZ692+kKS46rq0O7etpv3Xsnvyf6+921JDquqf5bk/Zlde/64nZ1VdVaS67v73ulsUTK7jGvR34/9/7L26HeSvKOq/sfMfmYvzuy6/v9cVU+earghs7OaD061Jnv/GQHAWjIHMwdb6znYpUl+vqpu7u5bq+rIJP9kusTv65L8fWY/h401u3H41y+h1nun12+avl5YV1xiBovo7nuT/F9Jfqa7t2b2NInXZ/aP+9bMnmqw87+jszO73viTSe7J7AkNme4b88Ykv5nZGaGnZXYD4tXwtiTvS/L+qvqbzK5/fu4S9/25zJYJfzqzP9I773GT7v7zzJ4W8qHMJiKbM3s6xE7/JbOzXp+vqvt2feOV/J7s73t3918leXWSX8nsjNffJlk4ufinmd1n4MHMvr8v7e4vL+H3Y0V09/2Z3UzygsyWcf+bJD84Xe9+cJJ/ndkZvQcyu1Hmq6ZdF/0ZAcBaMgczB8saz8G6+71J/rck76mqv05yS2b3u0qS6zNbrX1bZj+rh/K1l7Htqda/S/LmJH80Xbp26v7UCMvtoEcescoNAAAAYGRWEAEAAAAMzj2IANbAtOR4d36gu/9wVYsBABiEORjs2bq8xKyqHpfkOZldu7p9L8MBgPmzIbNHEP9xd39lrYvB/AsABrHHOdh6XUH0nCTSWwA48D0vyY1rXQRJzL8AYCSPmoOt14Doc0ly5ZVX5ilPecpa1wIALLPPf/7z+dEf/dFk+pvPumD+BQAHuMXmYOs1INqeJE95ylNy3HHHrXUtAMDKcSnT+mH+BQDjeNQczFPMAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcBvXugAAAB6tqi5K8uIkJyXZ3N23TO2HJfnFJP84yUNJPtTdr5z6NiV5d5Kjk9yf5Jzuvn31qwcA5o0VRAAA69M1SZ6f5M5d2t+aWTC0qbs3J3njgr5Lk1zS3ZuSXJLkstUoFACYfwIiAIB1qLtv7O6tC9uq6ogk5yR5Y3c/Mo37wtT3pCQnJ7lqGn5VkpOr6omrVzUAMK8ERAAA8+NpmV069qaq+pOq+kBVnTb1HZ/kM929PUmm189O7QAAixIQAQDMjw1JvinJx7r72Ulem+S3qurr17YsAGDeCYgAAObHXUn+PtNlZN39kST3JdmUZGuSp1bVhiSZXo+d2gEAFiUgAgCYE919X5LfT/L9yf//1LInJfmL7r4nyc1JzpyGn5nZSqN716JWAGC+CIgAANahqnp7Vd2d5LgkN1TVrVPXuUleX1V/luQ9Sc7u7i8t6Duvqm5Lct60DQCwVxvXugAAAB6tu89Pcv5u2v8yyQv2sM8nkzx3ZSsDAA5EVhABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDghgyItj28/YA6DgDAercv8yJzKABYfRvXuoC1cOghG7LlgmtX/DjXXXz6ih8DAGAe7Mv8yxwKAFbfkCuIAAAAAPgqAREAAOuWS9MAYHUMeYkZAADzwaVpALA6rCACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgDggLPt4e0rMhYADlQb17oAAABYbocesiFbLrh2SWOvu/j0Fa4GANY/K4gAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGt6SnmFXVNUm+McmOJA8mOa+7b66qO5I8NH0kyWu7+/ppn1OTXJbk8CR3JDmru+9ZzuIBAAAA2H9Lfcz9y7r7r5Kkqk5P8qtJTp76fqS7b1k4uKoOTnJFkpd3941V9YYkFyZ5xfKUDQAAAMByWdIlZjvDocmRma0kWswpSR7q7hun7UuTnLHv5QEAAACw0pa6gihV9StJ/kmSg5L80wVdV1bVQUluTPL67v5SkhOS3LlzQHffV1UHV9VR3f3A8pQOAAAAwHJY8k2qu/vHu/uEJK9P8gtT8/O6+xlJnpNZcPTO5S8RAAAAgJW0z08x6+7Lk3xvVR3d3Vuntq8keVeS756G3ZXkxJ37VNUxSXZYPQQAAACw/uw1IKqqI6rq+AXbW5I8kOShqjpyajsoyUuT3DwN+2iSw6vqtGn73CRXL2fhAAAAACyPpdyD6PFJrq6qxyfZnlk4tCXJk5P8ZlVtSLIhyZ8neXWSdPeOqjo7yWVVdVimx9wvf/kAAAAA7K+9BkTd/YUkp+6h+1mL7HdTks2PsS4AAAAAVsk+34MIAAAAgAOLgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAY3Ma1LgAAgEerqouSvDjJSUk2d/ctu/S/KcnPLuyrqlOTXJbk8CR3JDmru+9ZvaoBgHllBREAwPp0TZLnJ7lz146qOjnJqQv7qurgJFckeU13b0rywSQXrk6pAMC8ExABAKxD3X1jd2/dtb2qHpfkkiSv2qXrlCQPdfeN0/alSc5Y2SoBgAOFgAgAYL782yRXdPcdu7SfkAUrirr7viQHV9VRq1gbADCnBEQAAHOiqr4zybOTvGutawEADixuUg0AMD++J8m3Jvl0VSXJcUmur6ofS3JXkhN3DqyqY5Ls6O4H1qJQAGC+WEEEADAnuvvC7j62u0/q7pOS3J3khd39/iQfTXJ4VZ02DT83ydVrVCoAMGcERAAA61BVvb2q7s5sldANVXXrYuO7e0eSs5P8clXdntlqo59e+UoBgAOBS8wAANah7j4/yfl7GXPSLts3Jdm8gmUBAAcoK4gAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwG5cyqKquSfKNSXYkeTDJed19c1VtSvLuJEcnuT/JOd19+7TPHvsAAAAAWD+WuoLoZd39jO5+VpKLkvzq1H5pkku6e1OSS5JctmCfxfoAAAAAWCeWFBB1918t2DwyyY6qelKSk5NcNbVfleTkqnriYn3LUzYAAAAAy2XJ9yCqql+pqruSvDnJy5Icn+Qz3b09SabXz07ti/UBAAAAsI4sOSDq7h/v7hOSvD7JL6xcSQAAAACspn1+ill3X57ke5PcneSpVbUhSabXY5NsnT721AcAAADAOrLXgKiqjqiq4xdsb0nyQJJ7ktyc5Myp68wkH+vue7t7j33LWTwAAAAA+28pj7l/fJKrq+rxSbZnFg5t6e5HqurcJO+uqp9J8sUk5yzYb7E+AAAAANaJvQZE3f2FJKfuoe+TSZ67r30AAAAArB/7fA8iAAAAAA4sAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBbVzrAgAAeLSquijJi5OclGRzd99SVUcnuTzJ05JsS3J7kp/s7nunfU5NclmSw5PckeSs7r5n9asHAOaNFUQAAOvTNUmen+TOBW2PJHlrd1d3b07yqSQXJklVHZzkiiSv6e5NST64sw8AYG8ERGtk28PbD8hjAQDLo7tv7O6tu7Q90N0fWND04SQnTp+fkuSh7r5x2r40yRkrXigAcEBwidkaOfSQDdlywbWrcqzrLj59VY4DAKyeacXQq5K8b2o6IQtWG3X3fVV1cFUd1d0PrEWNAMD8sIIIAGA+vSPJg0neudaFAADzzwoiAIA5M93A+ulJtnT3jqn5rnz1crNU1TFJdlg9BAAshRVEAABzpKrektn9hl7U3V9Z0PXRJIdX1WnT9rlJrl7t+gCA+WQFEQDAOlRVb0/yw0mekuSGqro/s5tOvy7JbUluqqok+XR3/1B376iqs5NcVlWHZXrM/ZoUDwDMHQERAMA61N3nJzl/N10HLbLPTUk2r1hRAMAByyVmAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADC4jXsbUFVHJ7k8ydOSbEtye5Kf7O57q+qRJH+WZMc0/Ozu/rNpvy1JfmE6xkeT/Fh3/93yfwkAAAAA7I+lrCB6JMlbu7u6e3OSTyW5cEH/d3X3M6ePneHQEUn+Q5It3f3NSf4myU8tc+0AAAAALIO9BkTd/UB3f2BB04eTnLiX3X4gyZ909+3T9qVJXvKYKgQAgFWw7eHtKzIWAObBXi8xW6iqDk7yqiTvW9D8garamOR3k/xsd38lyQlJ7lww5q4kx+9nrQAAsGIOPWRDtlxw7ZLGXnfx6StcDQCsrn29SfU7kjyY5J3T9gnd/ewkz0/ybUneuIy1AQAAALAKlhwQVdVFSZ6e5CXdvSNJunvr9PrXSX4lyXdPw+/K116GdkKSrctRMAAAAADLa0kBUVW9JckpSV40XUKWqvrvq+rw6fONSX4kyc3TLr+X5DlV9fRp+9wkv7GchQMAAACwPPYaEFXVtyd5XZJjk9xUVTdX1XuTfEuSj1TVx5P8aZKHM11i1t1/k+SVSf5zVf1FkiOTXLQyXwIAAAAA+2OvN6nu7luTHLSH7n+wyH7XJlnaXf4AAAAAWDP7epNqAAAAAA4wAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwG9e6AAAAHq2qLkry4iQnJdnc3bdM7ZuSvDvJ0UnuT3JOd9++tz4AgMVYQQQAsD5dk+T5Se7cpf3SJJd096YklyS5bIl9AAB7JCACAFiHuvvG7t66sK2qnpTk5CRXTU1XJTm5qp64WN9q1QwAzC8BEQDA/Dg+yWe6e3uSTK+fndoX6wMAWJSACAAAAGBwAiIAgPmxNclTq2pDkkyvx07ti/UBACxKQAQAMCe6+54kNyc5c2o6M8nHuvvexfpWv1IAYN4IiAAA1qGqentV3Z3kuCQ3VNWtU9e5Sc6rqtuSnDdtZwl9AAB7tHGtCwAA4NG6+/wk5++m/ZNJnruHffbYBwCwGCuIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBbdzbgKo6OsnlSZ6WZFuS25P8ZHffW1WnJrksyeFJ7khyVnffM+23xz4AAAAA1o+lrCB6JMlbu7u6e3OSTyW5sKoOTnJFktd096YkH0xyYZIs1gcAAADA+rLXgKi7H+juDyxo+nCSE5OckuSh7r5xar80yRnT54v1AQAAALCO7NM9iKaVQa9K8r4kJyS5c2dfd9+X5OCqOmovfQAAAACsI/t6k+p3JHkwyTtXoBYAAAAA1sCSA6KquijJ05O8pLt3JLkrs0vNdvYfk2RHdz+wlz4AAAAA1pElBURV9ZbM7iv0ou7+ytT80SSHV9Vp0/a5Sa5eQh8AAAAA68hSHnP/7Ulel+S2JDdVVZJ8urt/qKrOTnJZVR2W6VH2SdLdO/bUBwAAAMD6steAqLtvTXLQHvpuSrJ5X/sAAAAAWD/29SbVAAAAABxgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADC4jWtdAAAA+66qfjDJzyc5aPr4ue7+raralOTdSY5Ocn+Sc7r79rWrFACYB1YQAQDMmao6KMnlSc7u7mcmOTvJu6vq4CSXJrmkuzcluSTJZWtXKQAwLwREAADzaUeSI6fPn5Dkc0mOSXJykqum9quSnFxVT1z98gCAeSIgAgCYM939SJIzklxbVXcmuSbJOUmOT/KZ7t4+jdue5LNTOwDAHgmIAADmTFVtTPK6JKd394lJtiT5jSRHrGlhAMDcEhABAMyfZyY5trv/KEmm179N8lCSp1bVhiSZXo9NsnWtCgUA5oOACABg/tyd5LiqqiSpqm9N8uQktye5OcmZ07gzk3ysu+9dkyoBgLkhIAIAmDPd/fkkr0ryn6rq40nek+QV3f1AknOTnFdVtyU5b9oGAFjUxrUuAACAfdfdVya5cjftn0zy3NWvCACYZ1YQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAOyHbQ9vX5GxALCaNq51AQAAMM8OPWRDtlxw7ZLGXnfx6StcDQA8NlYQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAACsgW0Pb1+RsQDwWGxc6wIAAGBEhx6yIVsuuHZJY6+7+PQVrgaA0VlBBAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwuI1LGVRVFyV5cZKTkmzu7lum9juSPDR9JMlru/v6qe/UJJclOTzJHUnO6u57lq90AAAAAJbDUlcQXZPk+Unu3E3fj3T3M6ePneHQwUmuSPKa7t6U5INJLlyOggEAAABYXksKiLr7xu7eug/ve0qSh7r7xmn70iRn7GtxAAAAAKy85bgH0ZVV9adV9a6qesLUdkIWrDbq7vuSHFxVRy3D8QAAAABYRvsbED2vu5+R5DlJDkryzv0vCQAAAIDVtF8B0c7Lzrr7K0neleS7p667kpy4c1xVHZNkR3c/sD/HAwAAAGD5PeaAqKoeX1VHTp8flOSlSW6euj+a5PCqOm3aPjfJ1ftTKAAAAAArY6mPuX97kh9O8pQkN1TV/Um2JPnNqtqQZEOSP0/y6iTp7h1VdXaSy6rqsEyPuV/+8gEAAADYX0sKiLr7/CTn76brWYvsc1OSzY+xLgAAAABWyXI8xQwAAACAOSYgAgAAABicgAgAAABgcAIiAAAAgMEt6SbVAACsL9OTYn8xyT9O8lCSD3X3K6tqU5J3Jzk6yWNA7AwAACAASURBVP1Jzunu29euUgBgHlhBBAAwn96aWTC0qbs3J3nj1H5pkku6e1OSS5Jctkb1AQBzREAEADBnquqIJOckeWN3P5Ik3f2FqnpSkpOTXDUNvSrJyVX1xLWpFACYFy4xAwCYP0/L7PKxN1XV9yZ5MMkbknw5yWe6e3uSdPf2qvpskuOT3LtWxQIA658VRAAA82dDkm9K8rHufnaS1yb5rSRHrGlVAMDcEhABAMyfu5L8faZLybr7I0nuy2wF0VOrakOSTK/HJtm6RnUCAHNCQAQAMGe6+74kv5/k+5NkenLZk5LcluTmJGdOQ8/MbJWRy8sAgEUJiAAA5tO5SV5fVX+W5D1Jzu7uL03t51XVbUnOm7YBABblJtUAAHOou/8yyQt20/7JJM9d9YIAgLlmBREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEA1u28PbD6jjAAAAAPtu41oXwNo69JAN2XLBtSt+nOsuPn3FjwEAAAA8NlYQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIPbuLcBVXVRkhcnOSnJ5u6+ZWrflOTdSY5Ocn+Sc7r79r31AQAAALC+LGUF0TVJnp/kzl3aL01ySXdvSnJJksuW2AcAAADAOrLXgKi7b+zurQvbqupJSU5OctXUdFWSk6vqiYv1LV/ZAAAAACyXx3oPouOTfKa7tyfJ9PrZqX2xPgAAAADWGTepBgAAABjcYw2ItiZ5alVtSJLp9dipfbE+AAAAANaZxxQQdfc9SW5OcubUdGaSj3X3vYv17W+xAAAAACy/vQZEVfX2qro7yXFJbqiqW6euc5OcV1W3JTlv2s4S+gAAAABYRzbubUB3n5/k/N20fzLJc/ewzx77AAAAAFhf9hoQAQCwflXVm5L8bJLN3X1LVZ2a5LIkhye5I8lZ0y0AAAD2yFPMAADmVFWdnOTUJHdO2wcnuSLJa7p7U5IPJrlw7SoEAOaFgAgAYA5V1eOSXJLkVQuaT0nyUHffOG1fmuSM1a4NAJg/AiIAgPn0b5Nc0d13LGg7IdNqoiTp7vuSHFxVR61ybayQbQ9vX5GxAOAeRAAAc6aqvjPJs5P89FrXwuo69JAN2XLBtUsae93Fp69wNQAcSKwgAgCYP9+T5FuTfLqq7khyXJLrk3xzkhN3DqqqY5Ls6O4H1qBGAGCOCIhYc6u5/NlSawAOBN19YXcf290ndfdJSe5O8sIkv5Dk8Ko6bRp6bpKr16hMAGCOuMSMNbcvS6X3l6XWABzIuntHVZ2d5LKqOizTY+7XtioAYB4IiAAA5ty0imjn5zcl2bx21QAA88glZgAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAwAFu28PbV2QsAAeOjWtdAAAAsLIOPWRDtlxw7ZLGXnfx6StcDQDrkRVEAAAAAIMTEAEAAAAMTkAEAADslnsXAYzDPYgAAIDdcu8igHFYQQQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAALBtPPgOYT55iBgAALBtPPgOYT1YQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEEGSbQ9vPyCPBQAAAEuxca0LgPXg0EM2ZMsF167Ksa67+PRVOQ4AwDzZ9vD2HHrIhmUfC8DSCIgAAIA1ty8n7JxwA1h+LjEDAAAAGNx+ryCqqjuSPDR9JMlru/v6qjo1yWVJDk9yR5Kzuvue/T0eHMhWc7m0pdkAAADstFyXmP1Id9+yc6OqDk5yRZKXd/eNVfWGJBcmecUyHQ8OSO6FBAAAwFpYqUvMTknyUHffOG1fmuSMFToWAAAAAPthuQKiK6vqT6vqXVX1hCQnJLlzZ2d335fk4Ko6apmOBwAAAMAyWY5LzJ7X3Vur6nFJfinJO5O8dxneFwCA3aiqo5NcnuRpSbYluT3JT3b3ve4DCQA8Fvu9gqi7t06vX0nyriTfneSuJCfuHFNVxyTZ0d0P7O/xAADII0ne2t3V3ZuTfCrJhQvuA/ma7t6U5IOZ3QcSAGBR+xUQVdXjq+rI6fODkrw0yc1JPprk8Ko6bRp6bpKr9+dYAADMdPcD3f2BBU0fzuzknPtAAgCPyf6uIHpykg9U1Z8muSXJpiSv7u4dSc5O8stVdXuS70ny0/t5LAAAdjGtGnpVkvfFfSAZ0LaHty/rOIBR7dc9iLr7L5M8aw99NyXZvD/vDwDAXr0jyYOZ3Qfyh9a4Flh1hx6yIVsuuHav4667+PRVqAZgfq3UY+4BAFhhVXVRkqcnecm0gtt9IAGAx0RABAAwh6rqLZndc+hF08NCEveBBAAeo+V4zD0AAKuoqr49yeuS3JbkpqpKkk939w9V1dlJLquqwzI95n7NCoV1atvD23PoIRuWfSzAPBMQAQDMme6+NclBe+hzH0jYi6Xetyhx7yJgHC4xAwAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAABYgm0Pb1+RsQDrgaeYAQAALIGnnwEHMiuIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAPj/2DvzuFunso9/Dw4nDVIaUFGGX4kUpV6RqdL4qpCUMkbpLUW9KhUNZprMKpESMiREL4lkSKZMdYlISJLKkDJ03j+utc9zP/vsew9r77327Zzr+/mcz3mee+/1rOte97rXcK1rCIJgPicUREEQBEEQBEEQBEEQBPM5oSAKgiAIgiAIgiAIgiCYzwkFURAEQRAEQRAEQRAEwXxOKIiCIAiCIAiCIAjGxMOPPDaW7wZBEIyahSYtQBAEQRAEQRAEwbzKwjMX5K27nNbXd08/cKMxSxMEQVBPWBAFQRAEQRAEQRAEQRDM54SCKAiCIAiCIAiCoGGEa1oQBKUJF7MgCIIgCIIgCIKGEa5pQRCUJiyIgiAIgiAIgiAIgiAI5nNCQRQEQRAEQRAEQRAEQTCfEwqiIAiCIAiCIAiCIAiC+ZxQEAVBEARBEARBEARBEMznhIIoCIIgCIIgCIIgCIJgPicUREEQBEEQBEEQBEEQBPM5oSAKgiAIgiAIgiAIgiCYzwkFURAEQRAEQRAEQRAEwXxOKIiCIAiCIAiCIAjmER5+5LGxfDcIgnmfhSYtQBAEQRAEQRAEQTAaFp65IG/d5bS+vnv6gRuNWZogCB5PhAVREARBEARBEARBEATBfE4oiIIgCIIgCIIgCOZjct3Swp0tCOYtwsUsCIIgCIIgCIJgPibXLS3c2YJg3iIsiIIgCIIgCIIgCIIgCOZzQkEUBEEQBEEQBEEQBEEwnxMKoiAIgiAIgiAIgiAIgvmcUBAFQRAEQRAEQRAEQRDM54SCKAiCIAiCIAiCIAiCYD4nFERBEARBEARBEARBEATzOaEgCoIgCIIgCIIgCIIgmM8JBVEQBEEQBEEQBEEQBMF8TiiIgiCYxsOPPDZP1hUEQRAEQRA0g0HWgK3v5pQJgmAwFpq0AEEQNIuFZy7IW3c5rUhdpx+4UZF6giAIgiAIguYwyHqztV7MKRMEwWCEBVEQBEEQBEEQBEEQBMF8TiiIgiAIgiAIgiAIgiAI5nNCQRQEQSMp5TsePupBEARBEARBEAQRgygIgoZSKhZS+KgHQRAEQRAEQRCEBVEQBEEQBEEQBEEQBMF8TyiIgiAIgiAIgiAIgiAI5nNCQRQEQRAEQRAEQRAEQTCfEwqiIAiCIAiCIAiCIAiC+ZxQEAVBEARBEARBEATzJINkrK1+N7dcEDyeiSxmQRAENTz8yGMsPHPBea6uIAiCIAiC+YVBMuNWs9vmlguCxzOhIAqCIKhhkIXBsMTCIgiCIAiCIAiCSRIuZkEQBA2nlNlymEcHQRAEQRDkU9qdrd9yJevKLTcpGZvaHoN+d1SM1YJI0orAMcDTgb8C7zOz342zziAIgnmNUpZM3ayYmuBu1wQZSsoRbofBMMQaLAiCoDyl3dn6LVeyrsebjE1tj/ZypRi3i9nhwCFm9l1JWwBHAOuPuc4gCIJgxDTB3a4JMpSUo+kKu6DxxBosCIIgCIKBGJuCSNIzgdWA16VL3wcOlvQMM/tLj+ILAtx1113jEo9H/nnv2P52i9tvv33iMjRFjibI0EuOJsjQFDmaIEMpOZogQ1PkaIIMveRoggyl5Oglw3Z7njN2GQC+udvren8pg8ocH9qnETPEGqwx66/2/l+yXMg478oY7dG9XMg478o4yfbot1w8s97lRkW3NdiM2bNnj6VSSasD3zGzF1eu3QBsYWZX9ii7FnDhWAQLgiAIgqBJrG1mv5i0EPMSuWuwWH8FQRAEwXzFXGuwpmYx+xWwNvAnIKKmBkEQBMG8x4LAkvicHzSDWH8FQRAEwbxP7RpsnBZEzwRuBJ5uZo9JWhAPkrhCHy5mQRAEQRAEQQaxBguCIAiCIIexpbk3s7uBq4HN06XNgatiYRIEQRAEQTA+Yg0WBEEQBEEOY7MgApD0QjzF6uLA3/AUqza2CoMgCIIgCIJYgwVBEARBMDBjVRAFQRAEQRAEQRAEQRAEzWdsLmZBEARBEARBEARBEATB44NQEAVBEARBEARBEARBEMznhIIoCIIgCIIgCIIgCIJgPicUREEQBEEQBEEQBEEQBPM5oSAKgiAIgiAIgiAIgiCYz1lo0gIEQfD4RtKCZvbY/C5D4Eh6ArAk8JCZ/SnkaEbfbIocQRAE/ZA7hpce+3PG1sfLveXweJAxl1L39nhow0n04VjHlCPS3HdB0nLABsBzgIeAXwPnmdm/CtX/fGBbYP02GU4GTjazRwvJMdF2aIockp4EbNZBhpPN7JclZKjI8kTg1VU5zOyGgvWvA7wfb4tnAo8A1+B980gz+9v8IENFlg2Yu1+cZmZ3lJIhybFiRYbrzOz+QvUuALwP2A54KfAPYBb+TE4FvmJmN85HcjSibzZFjiBoZ5gxc9BxLnfuzllz5Kzbhlnr5a6LcueKzDbp+1nnjuHDjv2DrqlyxtYJ3ttYn9mIZMztj4OOBQOPO8PcW4m+P8z9DTo2TqIPD7OOKdH3U5msMXzS+9puhIKoA5JeBewDPAO4FLgL78gvAlYCjgH2MbOHxijDEcBqwA+AS9pk2DB99kEzu3SMMky8HZoih6RP4S//TzrIsCHurvlBM/vtuGRIciwDfB54I3BdmxyzgQPM7NtjluHHwL+A4+ncN98OfNrMzp6XZUhybA7sDtzM3P1ifeCXwK5mdtcYZXgysAveP/8N/DnJ8IIk035m9rNx1Z9kuBS4GH8eV7ROeCQ9E38e2wKHm9nx87ocDeqbjZAjCKrkjpm541zO3J275shZt+Wu9XJkHGauyKxv4GedO4YPUW7gNVXu2DqBeyvyzHJlHOKdHrjcMGu1zHsr1veHqC9nbCzdh3PftZJ9P2fcn/i+thfhYtaZnYCdzOzX7R9IWhTYAtgcOGqMMpxqZjt0uH4tcKKkp+GD4ThpQjs0RY6/AS/soAU+HdhPkoDnAmNVEOGDxpeB7dplSQudHSTtaGaHjlGGj3c4WXsEuAy4TNI+wDJjrL8pMgCsCqxlZvd0+lDS6/FTyZPHKMN5wLHAy83sz5W6FwDWwvvECmZ25BhleEunNjCzu5Nsx0paYoz1N0mOpvTNpsgRBFVyx8zccS5n7s5dc+Ss23LXejkyDjNX5NSX86xzx/DccjlrqtyxtfS9lXpmuTLm9seccsOs1XLurWTfz60vZ2ws3Ydz37WSfT9nDG/CvrYrYUEUBEHwOEbSwmb28LDfCYIgaCoxzg3P460NJT0VeNTMHsgs/zQzu3fEYjWCeeHecvvj460fz6vkvp/DvtdBGUJB1AFJK5vZdROW4dnAZ4DbgQOArwDr4iawO9sEgpZJWhxYGbCk9S1Z92LAAmb2tyTHusANZmaF6l8FuMXMHpA0E9gVWAP3F91rkmaAkj5jZl+aVP1VJB1pZtsXqGdr4HQzu0fSc/BTwFfgz2MrM7t53DJUZFkaeCd+ivEocANwnE3Qh1jSVmZ2dMH6noePUY8BHwE+C7wX9xPfwsxuLSTH68zsnPTzYsDBwJrA1cCO1dPGMcpwEvA9vH8WiRNXI8ciwMfxfnmamZ1Z+ewgM/vwpGQLgiqSzjCzt/T53acBz8PH2psHnXslrY+fAl9tZqd3+d6iuNvRc9OlPwJnm9mDXcrMBHYA/gMcCbwDeA9+kvxFM/v3KMpUys5KMs6Ze8zsvLrvD0PuuljSysBsM7te0grAW4Brzezcmu8/Fdgbb4Mnpst3APub2UFd6vkMHo/kbkkrAT/CA+H+FXi7mV3RQ86XMb0df9/lu1ljvKTLU7nvmtlfBiiXfW9pHni6md3Zdv3FZnZ9h+8vgPfBO8zsEknvANbD9x7fMLP/1NSTvQYYwTu9KO6ac7OZ/b3L97LWapKeDuybZDzNzA6pfHaymW1cU+55wKZMH0NOMrM/dPl+9joq5/4krYG/a9Uyh1pNWIQh3s/ccjOBnYF3t8m4f7fxu8PfGXj/miyo1gCu6WTp0+ff+JyZfaHms2Ljdw6R5r4z10i6RtJH08AwCY7CfWtfAJyL+0JvCxi+iBg7kg6q/Pwq3MzwK8D1kjYsIUOq+53AbcDvJW0KnA9sD5wvaZNCYhyHmzWCD3Kr4maRSwLjdOeahqQd2/8BO1V+njRvKFTPLhUz0AOBM4DlgG8ARxSSAUnvwf2q18UnvqWBjQFLSsUSMryp/R+wd+XnEhwOXIAvZP4PV2yvAJwAfLWQDOCLuBZ7AvcDG+Fj19cLybAOvrC7Q9KX0wZpEhwKrILf+76Sqs/h1ZMRKZjfkXRi+z9gncrPdeWWkXQWcA9wBT7e/FXSgZIW7lLuksrP78PXT4sBe0rauabMm/E4FFvi49gKwFbATZK6KbIOwueCN+Prg83wtcML8IO+UZVB0muB3wF74LEsNgIOknRF2mh2KnOPpK9LemmXe6hj4HWxpA/jc/NPJO2ExxFZETi4y3rlGHz+WAdfb+6Gz6lvlPT5LtVtVtn47Qd8wsyeiLtq1I79kl4m6Xp8XXkysBdwuaRT0qa2E7lj/JLAa4Bb099/c1LI9CL33l6Pxza5XtLlkpavfHxsTbGv4wcLB0vaC4/180f8GXSbywdeAwzxTr9d0n2SfpsUHL9J93OTpLfWlBlmrXYEcG+6x7elZ9cK0dIx1IekbYGLgGVxRcgd6ecL02edyF5H5dyfpF2Y2k++EPgL8HTgCnlg6E7kvp+55Y7E1/Z7AGcBh+Bj5h6SPlRXKGf/Kunkys9vwseEjYAzJG3RRcZubFdT18Djd2kiBlFnrscH/22AL0k6G/gWfnpUyuRqaTN7U5o87jSzddP1yyRlaTIzqG4i9gDeY2bnpsXFEXhgsxJ8Eh+8noxPIq9Kp1HPB04ETiogw4zKSd56wCvN7FH5SdLVBepv8XXgx/jJUYtZuPVMkb4pqU77PgOoW1CNmurYtYKZbZZ+/o6kjxWSAeDTwOrJkun5wFfN7M3yU+pD8MXguDkDD4xXNad+KvAJvE/8uIAMS5nZ1wEkfcjM9k7XD5LUcYIcEzMqP68FvMLMHgF2k3RtIRluN7PVJK0GbA1cIOlmXOl/nJndV0iOV5jZSwAkHQZ8X9K38AXLjK4lg2B8vAY4E/h5+n0GPqeeWVvCORr4Jn6SvAWwBD7G7oUv/Os2C7MqP38QeK2Z3Sq3Wjgfjz/Tzv7A2mZ2U/Wi3ALmR/iY24lXm9kqcquNu4Fnm9lDkk4FLh9hGXDl0XpmdpOklwMfNbPXSdoKOAxXOLVzP26d8H+SbsfHpO9Zf9kMc9bF2wEvBp4E3AKsaGa3y2ONnEPnw7UXmNlG6eerJF1iZvvILVmuwYPIdqKqUFjSzE4FMLOfyy1M6jgc2N7MLkrKhQ2Bj+Fz+6F4f2snd4y/28zeLg/M+178QONISccCR1l9hqrce9sTWMfMrpG0JXCupI2SNUTdHLAe8BLgCcCd+F7k/nTAcGWXunLWAEeT905/Dt+fLI6vb/7bzC6W9CJcudrJsmSYtdoKZrZJurdTccvkMyS9rUuZXYGXWVs8G0lfwBU53+pQZph1VM79bYvHcfpneiePM7PXy4MuHwW8vEOZ3Pczt9wrzWwlAEmnA+eb2e6SzsXXvYfUlMvZv1aVff8LvNHMrpbHIvsh8N1OFUm6rEaGGXjWtU7kjN9FCQuizjxiZqelzrwcHrn8QOCPkvYsJMNCkmbgg/QT5akISVrrmYVkqPLslkmwmV0NLFKycjP7U5o8b2+ZxZrZLZRTcj5UOX35O1OLzoUKygDwWnzAOd3MtjazrYF70s/bFJJhRpLjFR3+lXI9vLlyUnRTWri3XDNL8mhrAZD64/PSz+fhi50SbJ3+/18zW8/M1gPuSj+vX0iG6gahfRFZ0o95EUkvkpvhz07KoRaPFZJhNoCZXWnuxrUkvgl9O36KWIo541Iy2d8YN+0+lpj7g8mxCn7YszrwA3NX2AfM7BgzO6ZLuaeZ2ffM7G/JJeGNyapie+D1XcpVx5+ZLTcN8/gtde5BM9uVQ6nM7+g+3z+Svvdv4PctVxnzWCh1409OGfBDq5vSdy/HM9+Q2nO5mjJ/M7OP4dYFe+HuDbdJOl7S67rUBXnr4v+Y2YPmrr2/N7Pbk4z3UD8v/Ccp71pBohdIZf7FlBV3J66QWyyBb0DXTH9jJaYfnrQzy8wuSnWcjh9APmJmn6fzBhnyx/hWubvN7EAzWxkfl5+Gt+eo722mmV2T6jwGt4j7kaRuB4qPmmeYehD4p6W08X30x5w1QO47jZlda2Y/x8eOi9O133QpMsxabY6Czsxmm9mHcBfQM5mugK6yQLtyKHEP9cq5YdZROff3qJn9M/38d5IyI/WZuvvKfT9zyz0md8UCeAq+JyYptbuVq9Lv/rXaxk9J38VqXAIrrIgruD7R4d8/asrkjN9FCQuiHqSJbX9g/zQob1Wo6nNw88RFcLeZEyWdB2yAa01LsLSk/fDB7GmSFkwTB5TdYCxY+bndzLJUH94VOEfSMXicm3MlnYGnPuy2qB0pZnZ+Mk38stz17n8ouwEHP9VcorXwqCKp1AZ4R+DUZCJ7L25ZdwXu610ytsotknYDzgbehfvpI2lBCvVNMzsmnaYcmaxkdqd8n3hI0pPN7H4zm3PyIXdFKBmHZ1F80TYj1b+0md0h6Sl4jI8STFv8pYX18cDx8nhZpbhL0qrptBgze0zSu/HxalJub8F8jnnslXfKXSIukPS/9DdePSppOTO7WdLquAs+ZvYfSd02CkonvDOA5VvjVPqszo3linSKfgTQ2hwsg8cKuqpLXbMlLWQel2aOe6/cXWbBEZYBuCe9zz/BLTBuq3zWdX2WFOcnASdJWgpf1x6EW2r3ZIB1cVWOT7V9Vtf2X8Xd2a7CD50+AiDpWUw9i058CDg6WRDfgYcguA1XdGzdpdyDkl5tUxZEVWuqujkjd4yfSzFgnvr6Ukkf6VIu994WkjQrbcIxswskvQt/9nUKgD+mdf+TgRskfQWPm/QGPJ18HTlrgNx3enayFnoqfoD+KjO7VNKK1L8zw6zVfi/pNUkhRZLxE3IXvF1ryvxE7j73DaaPIe/H3cc6Mcw6Kuf+rpJ0JD6GbAL8IpWZRb0hQu77mVvuOPz9uBB4HcliKJXrNm/k7F+fL3dznpHKL2JTniPdDDOuBO5rKZqrSKpT4GaP36UIBVFn/trpYtJSX1xIho8B/42/AKcDq+GT8DnUm9SNmqr577dw39S704Kim6npqDmhMmjOiS+TJoj29IdjwczOk7QWbqK+LD5RPgvYzcxKKexasjwAbC/pjXh8qm4mxuNgI2pOksxsjRICJI3+aklZthJwIT7AnlU5ESnBDvjJ4ea4+2PLvW0x0gRYAjO7A3izpA/giuW6xd+4eLV1djOYQTmlOma2bM1Hj+IntSWoDRjfOkEvxA60nS6nhff7gO8XlCMI5sLMvifpZ0zFBOrF5/CNwl3As/FYPa2NwlwL8wrtcdhmp3LPxk35O7E1HoflO6STeHwTcxJTY3wntsE3qI/a9EQiSzM9PtqwZcDXIt/G2+/yJDOSntGlXCcFxZ24NdFeXeqCvHXxVyQ9ycwesEpAWUkvpGaTbGbflnQp7pq2c7LaaimlamPqJYuCjZKl90p4m95mPYJT4wFwT5InNbgHtwJq9avjasrkjvG1cSKtSyDhIe7teGBtfN/Q+luXyGN31u0jtsbdlVpWp+/E+9nvgQ90qStnDZD7Tn8uff5YKvNFRKtGKAAAIABJREFUSUsCz8Hfi05U12qXM9ha7b10UEaY2acldXQ7wg8qt8Df79YYchv+TOriPw2zjsq5vx3xZ701vn7dJ12fiT/3uRji/cwtt3c6/H0J8EEz+1ml3EvqypG3f/1o5eczcIvrf6cyp3WpawugLiPbMjXX28fvraDn+F2UyGL2OKI10U5ajqA5pEXNyp0018H8iaRlgf8ys4koAdSAFKYaMivKvEYTnkkQDEvqx8sDN1q5WF7zDJKW6cNdojEMO25JerqZdVRsdfjuEjUuQY1kkHubFP3IOIp3OlnJvBQPQVEiU2lfWdOGrKPxc3bJ9zMoTyiIalAD089J+r2ZdYyYP8Y6G5HCWx7E624zu03uP/0a4DozKxIoW9ITcY3vH83sBHn2k1bqzz1LDuINaIuJp/FO1mMH4Nk1PoWf9K6LP4+tzey3BWVpyjvSd5reMdSdlcJ0DHIsgwcd3RA/8fs77rN+GPCp5Aowbhm2xt+Ne5K7wTG4SfWvga3M7OZxy5DkaMQzCYIqlbn0djM7ftC5dFDlb7IU+gyeQecAPPjtuqm+ndusdqrlOs2z15pZnXvIMHW9k7lTTR9oPVIrD7tOHXSjm1PfoPNjZdx6Nx7cGvpLh70W7hL4R9wy4od4LI9/ABv3svQeZP7MXX8oP8199r1ltH9dmvtrgW9afZr7YWTMOtAZpFzlvm5P7mit+7oO+EbdfaWyb8fn8TuB9wE/wN37nok/77mCYufUN8ycnXt/yk9zX+T9lKe5/1gHGbumue/Sj0faHqnMAXiw/27ux53KvRz4s5n9cRJ7uV40ws+taagB6eck3d3+D3iepL+oPovUqGWYeArvJMfuuGn3+ZI+iJtnLgccIPe5LcE3gVcCW0o6Do+QfyzuZnZ4IRma0hZNSON9BO5rfTueDecCvB2OpN5tYOQ04R1RXpreUZObwnTUHI1nmng6bi58MO4SuliSqwS7VE6iD8RNlZfDYxEcUVtq9DTlmQRBldZc+r5B5lJlpsTGs/H8G89Qcy6uON4WMKZSPLfXVTfPHthjns2paz/cveB8fLN0Hr4Z/6HcFagjOetUZaQHH7K+nPmxNW6ty2Dj1pfxjLfH4W35RfNU8O+hy9ifOX/mrj9y09zn3ltO+9elud+E7qnWB5Yx953OLNe6r0Pa7mvjHvcFU1nTtsezpm1unllrbeALI6xvmDl74Po0XJr7dQeUMasPJ/mWZ8A099T3442Br3UqkNke4HGlzpJ0laSPJMVlVypzzAUT3Mt1JRREnWmln1sVT5X8ZzN7Md4pS20+r8F9Hv8LP31eA/gTnlXhFYVkaKVN3CjVv6h54LStKRcHCfz0YyXgVXhgxHXNbEdgTdzftgQrm9mmwNtw64TNzexEPI3rqoVkgGa0xe1mthp+kjgTH+Auk/QBeTDgEjzFzA4ysy8BTzWzA8wzg3wbD1xYiia8I600vYvh/fPn+Gbr13ROIzwOXmBme5rZVWb2cWAjM/sVfoJTql/CEFlRRkg1tt8KZvYVM/uLmX0HX2yUoinPJAiq5M6lR5On/F3azHbB46e80Mw+YmaXmdkeTMUGaSd3ns2p663AW8zsK3gsjjXMbH98k1iX+hny1qk5G91h6suZH3PHrZlmdnoaZ/9jZj8AMI9Z0i3rbs78mbv+uNvM3g48H4+hsy8eFHofeYDlUd9bTvuvh/eRdfAkKG8ws/2At+BJWUYp49HkvdM55XLvi3Qfg2ZNy6lvmDk7p75tgTXNbCf8nV7OPEPbG/HDrVHKmNuHX2lm25vZqXgsqA1S2dfSPSFNt/ZYr6ZMTnuAx+d6DvBFPJD2bZJOlLShPBt5J5qwl+tKKIg6M/H0c2b2WnyCOhqfjG7F04z+wcr5kDchhTfAw2b2z7TR+7slUz/zTCT9pjkcFbPxgHGtdKX/oT5l5ThoQls0IY33TEmzJC0BLC7pmTDHZL5kgOYmvCM5aXpHTW4K01HzqKTlkhzTsqIUlOPmymn8TZJWSPI8u1D9LZryTIKgE4POpbnK34XSIv0JeMajJwFIWoj6zDS582xOXVU37dmkgN1mdlvnr88ha52asdEdpr6c+TF33FpQ0hJp/F9cHtC5FfS12wY0Z/7MXX/kprnPvbfc1Oc5ae5zZMx9p3PK5d4XpKxpkv6LlDUt3Vu3rGk59Q0zZ+fUVzLNfW4fzk1zX6o9AGab2aNmdoqZvRW3eLoCt1SqG8ebsJfrSmQx60wj0s+Z2cGSfgJ8Q9L5JetOTDyFd+LvknbEF05/TWaAx+Ba3VIZq34t6QQ8Y9hPgGMknYKfgFohGaAZbdGENN7fA36L98Pd8Swk1+Ba/27ZBkZNE96RnDS9oyY3hemoyc2KMkp2BE5N7+a9wGXyLBzPpfuJ16hpyjMJgiq5c2luSuxz8Hd/EdzN80RJ5wEbAHVxUXLn2Zy6zgXOlHQunrn2dIBkjTvqNPc56cGHqS9nfhwmjXYrvtsH8H71NzwD8H5dyuXMn7nrj9w097n3ltP+uWnuc2TMfadzyuXeF+RlTcupb5g5O6e+0mnuc/pwbpr7Uu0Bc++J7sKtA/eVtGZNmSbs5boSCqLO5KQPHQtm9jtJ6wGfwP3SS9KIFN74qcB++GT9VnxwuQU369uqkAzvT3LMxv3PX4e3zy3peima0BYTT+NtZntJOhPX3F8jD5y9CXCemZ1SQoZEE96RnDS9I8UyU5iOQY4z5RY707KiJDneX0iGPwCryWN2rARciG+mzqqcTpWQoxHPJAjayJ1Lc5W/H8MVL7Nx5cvqwJa4MqfOzSZ3ns2paxfctWFV3GL86HT933icjzpy1qk5G91h6ht4fswdt8zsKEk/xC2d/irpDNya5HNmVpfWGjLmzyHWH7lp7nPvLWd9kpXmPlPGYdLcD1ou677SvZ2BW3mR6rmA3lnTBq5vyDk75/5KprnP6sOWn+a+SHskjuoi/8U1HzVhL9eVyGIWBEEwD6CGpOlVA9KzKtLcT6MJzyQIhkWR5n6kqHB68EHJHbdyypWeP0veW2kGkTH3nZ7EWFByXTGvPedRlAvKEgqiGjThNPfy1H4746a8faf2G4McTUnh3f48rm9pkgvK0Cn94SGlF1aTbosm9M0myFCRpSnvSBPS3A+U+nQMcjQhzX0j+mZTnkkQtDPMXJqzSVNGGvnceTazrqJp7nM3ujn1DTo/5o5bGiJFeCo/SJr7rDE+V8Zh7i1nfdKhzPXA93uUGUbGsae5z72vVC5rXTFofcPO2ZnPrVSa+9y+n5XmPpUde3tUyhWbY0oRQao7oAakucfNeJdj8NR+I0MNSOGd5Oj0PA4u+TxUn/7wSnVPfzhqOSbeFjSgbzZEhka8I2pWmvt1iTT3jeibNOeZBMEccudS5afEHjiNfO48m1lXyTT3WW04RH0l09xnpQjPnD9zx/jcNOa59zZw+9eU2aRbmVwZh3inBy43xH1Bxrois77sOTunvi5j8TjS3Of2/aw09wXbo+gcU5JQEHWmCWnuc1P7jZImpPCGZjyP3PSHo6YJbdGEvtkEGaAZ70ikuZ+iCWnum9I3m/JMgqBK7lx6NHnK35w08rnzbE5dJdPcH02+Ar3pae5zy+XMn7ljfOl7y2n/3DVNjoxHUy7N/TBrtZx1Rcm+n1tfyTT3ueVy37VS7QFl55hihIKoMxNPc09+ar9R0oQU3tCM55Gb/nDUNKEtmtA3myADNOMdiTT3UzQhzX1T+mZTnkkQVMmdS3OVvzlp5HPn2Zy6Sqa5H0aB3vQ097nlcubP3DG+9L3lprnPWdPkyFg6zX3uWi1nXVGy7+fWVzLNfW65YdLcl2gPKDvHFCOymHWmCWnuc1P7jZImpPCGZjyP3PSHo6YJbdGEvtkEGaAZ70ikuZ+iCWnum9I3m/JMgqBK7lyamxI7J4187jybU1fJNPe5bZhbX+k09znlcubP3DG+9L3ltH/umiZHxpJp7odZq+WsK0r2/dz6Sqe5zymX+66Vag8oO8cUIxREnWlP57kVgAqmuTdP7Xc5nvp0kNR+o6SaIvNyJpfmfuLPg/z0h6Nm4m3RhL7ZBBkSTXhHmpLm/hJgZSLNfSP6ZlOeSRC0kTuX5ip/c9LI586zOXWNKs19PzIOo0DPqW/g+TF33BpivMtJc581xk/g3nLWJ1lrmkwZS6a5z16rZa4rivX9Ie4vN819sT48xHqqSHskSs4xxYgsZkEQBPMAakia+yAIgnGgSHM/NNGGnYn5czLk9sfox0EwXhphxvR4QtILGyDDkb2/NXYZXjdpGaAxz2PrScsAjWmLJvTNicsA5d+RTotbSU/q9N2SyLONTBxJ1zZAhqb0zUY8kyCo0msuNbO/m9nlo9oQSvpcRpmseTazroHLpHK1Mo66DXvV16XMwPNj7rjVT7lRzZ+5Y/w4762mXE77Z61pusmY2x9H2Y+HWavlrCtK9v0h6ht4XzOBPpz7rhVpj1Su2BwzasLFbHD+jxTsaoK8YcL1A3yLybcDNON5fB43FZw0TWiLJvTNJsgAzXhHbighg6RFu3y88rjrr8ixUpePn15Kji4U65tNeSZBMABZc6mka82sV6rqTmwHfGHAMrnzbE5dOWUgQ8Yh2jCrPmrmx9xxa0zjXc78WTvGN+zectYntWVGLWNuf8ws17UtxrCuGGnfz62vBx3H4ob14dz11Mjaow9KzjEjJRREHZC0Y81HM4AnFpLh7i4yPLWQDPt1kWGxEjIkOZrwPE7sIsPTSsiQ5GhCWzShb05chiTHxN8RSd1800tl2HsADxg4o3Kt9XtJP+brgFvb5GhRJKtcU/omzXkmQTCH3Lk0d5Mm6bIu9T2zpkzWPJtZ18BlcmUcZqObWV/O/Jg7bmWVy5k/hxjjS9/bwO0/xJpmYBmHeKcHLjfkWm3gdUXhvp/7rHPG4tJ9OOtdK9geReeYkoSCqDNfA75H5067cCEZZgAbAP/ocL1UNp6PAPsBj3X4rOQGownP483AR4GH267PANYrJAM0oy2a0DebIAM04x05HbiAzouXJxeS4U/AqjVm+n8sJAP4Im5tM7tjgnI0pW825ZkEQZXcuTRX+bsiHqj0n23XZwAn1JTJnWdz6sopkyvjMAr0nPpy5sfccSu3XM78mTvGl763nPbPXdPkyJjbH3PKDbNWu5XB1xUl+35ufTljcek+nPuulWoPKDvHFCMURJ25AdjHzH7b/oGk1xaS4XJgCTO7poMMcw1SY+Ja4KQaGbYrJAM043lcBVxtZr/qIMMXC8kAzWiLJvTNJsgAzXhHfgdsa2a3dJChlCLgZ7iZ8PkdPqs7XRkHJwPLAJ36wCmFZGhK32zKMwmCKrlz6a3kKX+vBO4zs7k2E5LaNwItcufZnLpyyuTKeCv5CvSc+nLmx9xxK7dczvyZO8aXvrec9s9d0+TIeCt5/TGn3DBrtZx1Rcm+n1tfzlhcug/nvmul2gPKzjHFCAVRZ/ajPoD3JwrJsBGdNZ+Y2RqFZPg0bhbYiXcVkgGa8Tw+BPy55rO1CskAzWiLJvTNJsgAzXhHvoGbv861wMVPKcaOmW3R5bONS8iQ6qp9B8xsp0JiNKJvNuWZBEEbuXNprvJ3C+rH6GVqrufOszl15ZSBPBmHUaDn1Dfw/Jg7bg0x3uXMn1lj/ATuLWd9krWmyZQxtz/mlMteq2WuK4r1/dz6yBiLJ9CHc9dTRdojUXKOKUakuQ+CIAiCIAiCIAiCIJjPiTT3HZC0VB/fefaYZfi6pCW7fL6RpLFaKEhavcfni5RIx9eQ5/GOHp8/S9KrxilDqqcJbdGEvjlxGVI9E39HJD1hFN8ZUoYfSup4miPpyZJ2krT9OGVIdV0kaTNJc/lwS1pB0le6BAcclQxN6ZuNeCZBUKX0XCppZ0mLdPl8VUlvaLuWNc9m1jVwmWFkzCWnvpz5MXfcGqLcwPNn7hg/gXvLaf+sNU3T55th1mo564qSfX+I+gYeiyfQh3PftSLtka4Xm2NKEi5mnTlB0nXAccBlZvZvAEnPw9PqbQUciJs5jotzgJ9I+gvwS9zsbRYg4DXp88+MsX6AT0l6It4O7TK8AXgTsAswlw/liGnC83i1pN2Bk+jcFisCO4yx/hZNaIsm9M0myADNeEd+Lukk4Dgzm+N/L2kmsC7wQeAM4KgxyrA7sLekFXF/8lY7vBBYFjgcOGyM9bfYBPgscLCk3zH9efwd2NfMugV+HQVN6ZtNeSZBUCVrLpV0EfB14FQze7jtsxWAHYHfmdmhbUUfAa6XdEFNff8A/qetTO48m1NXTpksGYdow9w2yZkfc8et3HI582fuGF/63nLaP3dNM7CMuf0xs9wwa7WcdUXJvp9bX85YXLoP575rpdoDys4xxQgXsw5ImgFsCmwPrAn8G3/YfwZOBb5mZrcWkmUtfJJ6DvAQHnjrDDOrS/036vrXwNuhJcM/kwynAt8ys/sLyNCI5yFpaWBr5n4epwA/MrP/FJChEW2RZJlo32yQDBN9R+Snmx9OMizK1OS0JB4YcD8zu2ScMlRkeS6wDtOfxy/aF3EF5FgEeEVVDjP7fWEZJt43kxyNeCZB0CJnLk2nyJ/F57+BlL9pjNy4U31mdnWH72fPs4PWNUSZgWUcsg2z2iR3fswdtwYtN8z8mTvGl7q3VGbg9h9mTTOIjLn9cYhyQ63VBl1XTKDv5zzrrH1NyT6cyg38rhVuj2JzTClCQdQDSQvhaRMfMrP2NHtBYeJ5TBFtEbQj6TlMTU5mZv+asEhBEAQjo7Ty9/Ewzw4q47Bt+Hhokxxi/pwMuf2xCQdBQTAsTR1PQ0EUBEEQBEEQBEEQBEEwnxNBqoMgCIIgCIIgCIIgCOZzQkEUBEEQBEEQBMHjkhScOAiCIBgBoSBqOJL+t59rQVCaJvRNSWt3uPa+kjKkOp/c4drzSssRTCHpCZJeoC4pUgvIsJCkF6d/E8saKmlFSRuln58k6WmTkiUIJomkZ0xahm48DuSb2eHaU3uUmZVRz1f7uVbh+PSdnw5a1yTIHZMfT2O5PJX4Fo+XZ9IPkhZNz2Cl1r8x1THt36jrGBWSFh5EzmHuTdJTJK02nMR91fOczHKvGrUskyTS3NeQNnebAs9Nl/4InGRmfygsyruA/fq4NjYkPRP4MvA8M3uNpJcAa5rZ4aVkSHJsBpxlZvdJ+gKwBrCbmV1RUo4mkCLgbw+8Nl06B/iGmZUMKjbxvomnHX2nmRmApHcCHwO+U1AGgFMkvdHMHk1yLAmcBbx43BVLelO3z83sx+OWoSLLS4AjgFWBRSoyLFio/gWA9wHbAS/F04vOkvQInhniK2Z2YyFZXo6nKP03MANYSNLGZnZliforcmwJfApYGDgNWBo4hKmxIwiKI+lA4AvAg3i2qNWAHczsuz3KHQ4cbGbXDVjfK4ET8YPR56b3c3sz2z5H/po69jWzXSX9AJhrLjazd45DPkkbAMtRWdPXpKqvllkUD+5bLXNDr7qAo4H3VP7Ok4GzgW6bo1slfQ841Mxu7qMO8BTW7azT5ftPkLQxsEynObHbPJiCHT/dzO5su/5iM7u+pszzgW2B9ZkKkvxrfMw/ubUWqCmbNSYPUW4W/sza+0inQ74XA3vh89ZncuZLSa/A22ZTPKX50T2+f2L7u9HpWo+/0co+taWZ1VqTDbOfkfQhYB/gXqCVzWo28IIR1/cAbeNHWsNcBry/td4dVX3J+u7bwNJm9vykhPlvM9ujS5m3Awfh2f7A1zizgV5rvax7S+/0EcBjwLJpfNzdzN7arbLM9r9M0iXAIWZ2Xo/7qXK4pMfwd/K43CD3klYrvU7sRCiIOiBpW2AP4IfAbenyssCFkj5vZt8qIMPrgNcDS0mqbrgXw1/EknwD3+zumH7/LfBdoKiCCJ+sTpCnLtwQ+Bo+QK057ool/YWpQW0GMNvMnpk++6mZbTBuGdrYD3gZPqgDbAksD4zdgqdhfXML4ERJr8cVhp8DSj8L8AXyt4H3phPgs/E0rCX4RPp/Fp7R49r0+yr4pFtMQQQcBnwGn5DfAHwI6Jo6dsRcnP7tDFxhZo/BnEXChsCRkg43s+MLyPI1YBsz+2mSYX18vHp1gbqrfBR4OXAheGoeSc8uLEMQtPNaM9tF0puBO4DN8LGqq4IIuBFXyN+FL8S7bsYrfBl4I/A9ADO7XNIxnb7YScHT2qy2lEA1dfwi/X9GH/Jky9cm69H4+30lvnmCDsqptjIDb3Qr/LmiCHsCfq/f61HmJfiB1nmSbsA3Xh3bSNKmwDvxTeCJlY8Ww9NU1/EpYAfgWUzNiS1mUzMPprXDCennm4F3mdlN6eNjccVle5kj0vUfALsCd+Hz74vweeaTkj5oZpfWyJo7JueW+wGuVPolrvjpxjHAUfj9nCdpMzO7qFcFkpYA3gtsA8zED+keNLM39iHf8h2uvbCPOmcCb8OVUevia7BtehQbZj+zC7DygIYCOfXthiscj8LX1FviWa5+jytJ1h1xfYcBX8LHBICr8b6/R5cy++Pv6aV1qd9ryL23z+Pr27Ngzvi4XB/15bTHsvh8tKekxfB55jtm1nUta2YvlXs17JjKfhc4LCOz3heBNw9YZuSEgqgzuwIvM7N7qheT1crFwNgVRMDDTGlaH6xc/xOwd4H6qyxtZodL2gHAzB6WNMiAMCoeSf+/DvimmR0n6eOF6n55l8+2KCRDlQ2B1SoWKycCV1BAQUSD+qaZXStpZ9yCakHg9Wb255IyJDkOlPQ1uQn8WsC+ZnZKobrXA5D0feCjZvbL9PsawE4lZKgwy8x+KmkBM/sT8BlJvwL2LVT/W9rHbQAzuxtf8BybFrIleGJLOZRkOE/SlwvVXeVhM3tAUvVaPxvqICjBa4BTzOxOST0tYM3sy8CXJb0BX4gfKOlbwOFpzKljYTO7oe09eLjmu90UPL+o+8DMTk//91TsdGAQ+aqsCbzYzB7p+c0pcja61bInpnn3DcCZZnZQtwJp/P2SpL2BjYBDJR0EHIwri6on7TcCZ+IHPmdWrt8HdHNVusHM3iTpy2a28wD3syewjpldkyx0zpW0kZn9mvoDr1PNbIcO16/F2+ZpdFe25Y7JueWWN7MX9fE9gJkt6zNJVwInJUXMtsBHuhyG3oErrnYws4tT+e26VSTp/bjicEVJl1U+WgzoZiWzKq4I2hy4CldqvcjMPtjH/Q2zn7kr453JqW8TM1u98vvXJV1hZqtL2mUM9S1mZmen9xMz+4+kXmPPva3nPCDZ92Zmd7X1/V7KTshoDzN7mKn14pq4++reSWH/xTSe1ZW9EDcmeSnwI+Bjks4CPmFmv+1DXsxs4sohCAVRHQt02mQA91DIQsLMLgAukHSyDWhKPQamTUByf/PSliIAs+VuZu8C/jtdW7hExd0mhR6L0nHRMudsMZv5qG+2WS6B3/8NwE6SOppOj0mOqv/5N/FTkfOBqyWtZP2Z7I+KlVvKIQAzu0zSKgXrh6mx4t60iLsdPx0qQnXclvQUfGF8Zd13xsw/Ja1rZucnedah+wn4uPirpBVJ44WkLfDnEgST5G5Jh+FWM/vIY3QN4op6KbAS7kr6X8C2kg4ws7o4Nf+W9CSm3oOVgI4uAN0UPC0lUCc6WR61le3mMtO3fG38sY/vtDPwRlfTY4V8AD+VPx93817UzLqOban8e3Gl3k34fLkebtmzfut7STHza0k/MrN7BxDxeGB13L15EGaa2TWp7mMk3Qr8SNIm1DxLMzu72x9McneTPXdMzi33e0lP7mUBkXhI0rPN7C4zO1/SUsBSwN9wi+Q6vooflu4t6SjgpD7q+j/gd7iisGr1dR9wTZdyV+HKwpeb2W0Akr7UR30w3H7mnLT2PJ7Ku9ljnZdT36KSXtCyPJG7Mz6x098bUX2PJSVgq18tzZRlYR2nSvogbn1XbYtea5zce7tf0rMqMq4L/L1HXXP9zX6ft6Rl8HHu3cC5+Hi1PvAT3HujrtzauOX8K1OZVrkfASv2IW9jCAVRZ36SNH7fAFqT6DLA+/EBrSQ3yV3eevoOj5FTkkntkyVthU/wRxWsv8WHcQuZb5rZLXK/2Z+VFEDSa4Ar0ynOtrjJ475mdktJOfBB6iy5eTm4mWbXRcsYeF+bNh8o1jcfbPu9iLVOB87scG3j9K9fk/1R8aCkLSzF8JD0HsorJE6Q9HTckuwX+Ibvc4VlyPZXHzE74aevrVOuhfF+UZqPAscBSpuffwIl2yEIOvFuPC7KMWb2N0nL4m5WXZG0Or4Afz3er19jZrcmhfB1+Ea1E3vi67el0rz5Bvqw/k0b8heZ2WlJgbNwF8VFy/JojfSv5S73brpvrgeWT1LLZeJG4KeSfsj0jVq3GEQ5G92W1fCMyv8vBz5Oj9gjkg4G3oFvkt5TOVg6TlLdqfoRnSzKuijZcmMQLSRpVsuKycwukPQuXMHRNbi23HX5w8y9Pu8VOyd3TM4t9w/gckk/Yfrz7rRW2xmPbXRX+s5s3DoIuqwnzF0OPwW8Cbfu+QretuuZWcd1elJS/gFYuXVN0sLA0yy5h9fwITze0AXpXRkk5uQw+5lWApRNK9d6rfNy6vsMHgfnCvw9exnwgTT+/KBH2Zz6DsVjNC4haQ/8PnfrUWbP9P8hTB8Xein5c+/tk7hS+vmSzgdWYMpQoBsDt4ekM/D4oUfgnhp/TR9dnMaGunLX4iEVDgK2sCnX5+8mZW5fSDrDzN7S7/fHxYzZs0vGtH18IA90ugXuX9nKRHQb3nmPtcH8LYeV5XSmfIfnDJhm9vlSMiQ53oO/jDOAH1mPQJLzKpKuwU+oVsL7w3fxWArrdy04ejkWwP3tW+a+5wJHFu6bu1d+nQW8BbjczLYuJUNTkPRUM+vnNGOcMrwIN4ttBca+Fg/Y+JsJyTMTdzkrGYOoVfev8IXzWWb2snTtBjMbecaRHnLMBFpaVBvQDWSUciyHLBx5AAAgAElEQVSIn17NSHJ0W3wHwVhJ/fG0nEVwWoQfBHy3/bRa0g5mdkRNucWAp+Pu2TOA/7OpWDN1dc0JCmxmL5CfiBxiZr2CAl+ErwseSr8/ATjXzLrGH5P0gn7lk/Ttus/wGIm1sVgkdTrQmm1mYznQkIcC+JaZ/a3DZ0t2ssJObd9iFrAJ7kbW0W1antlrB2Bt4PK2j2fXrdEkfQb4pZmd03b9FfizXqPLfV0I/Aa3Zquuz/uJHZU1JueUa1urzaHbPqJ62NTtWpfyz8AVDVsBi5tZbWYoScfjz+5hPND3EsBeZnZAjzpWxpVR7waegiuOTjaz+3qUK7qfyakvWcu0+t4vu7k1jai+tfA10wzg9OQqNRZy7y2N4WviMl7c75p70PaQWw+eOsg6Ke3JVjOz9rFnYCStbg1IvhQKooYj6TfWv+/wPI0ys56MWIYrzWy1tOD5t5kdJOmq1iZ0fkaezeQkM9uwYJ19Z+cYowwzgOtLKx/aZFgQ2NbMjkzPgUkoZpIsyzH38ygZKBtJvzKzV1TfzUm8p8rPFDRPyhEELeRZYta0ATNvStrAKrG90rX1rUu2mdwxWtJVuMLhwso4cp2Zrdyj3I3AC1sHNml8/o2ZPa5cDDqRNuW3mNmD6fcnAstaTbavEda7MK40W7fH9waNQTSMTD37QpeyKzMVkPe8buOxeqQC78OtZ2Baa91e1/r8W2uYWa0FXWtuThvz1+JWTJea2Uv6/PsL4QqAbfB4Uk8eVMZBSJZjc6zLWm5uY6hnYabP2ZNwUa9Fg1lXtpcd6N4mMe4M+pwlXdNvn308EC5mNag5ae4H8R0eC0OY0Y6a9qwn78JdfEpaMy0kT0f7DtzlEAaLmzAUknYys69J2p/OaXRLuh628wBTFnelGCQ7x1gws9mS/ihp8U4npIVkeEzS9rgV2STHir3xFPO/YXpGnaIKIvL91UeGhssUNM/JEQRtXIrHsTgOnzuAvpTJ+zN3ZqkDOlybwxBjdG5Q4J8BP9ZUFrL30sMdXh4MdT/8vVyItmypXcrtCnyjtTGTu/huY2b79yi3Eh4DCOCn1mcAVTwgcDWl/SPpWm0iD3lq6SNw6+tFWtfNbJC102zc9akXX9GUi9m1ZpYTo6lfrpO0lJndOUihNCZ/mikX9U9K2tPMDqspUk0N3h4/pdatR9KmZvYDTbkjTsM6uCHK3bFfibsbVcstRmbMz27KocTM9P86wI/N7J8aIBGOuSvPKbg70ZKdvqPh4oO1/sb6eF9/Fr6+WRj4K1D7jubsnyS9A/g6g6eQH6g+TWUj7Ng2PWTcCnf7Whg4DX83D8EVfN3ky723gcedVF9O+6+Huy32/ZwTN0la1sxu7fG9rkja3syOHOZvjIJQEHVAzUhz3wrCO4jv8Lg4Gd/wnUvFjHaCtLKe3KE+sp6MmM/iC53zzOz6pEHvaqY+Ylp94IGu3yqApgeKXgAPEFnanWmQ7Bzj5B/AVZJ+zPTNTsn39GeSNjGzfoJDjotNgeV6mXgXINdffZQMkyloXpQjCKq8NP1fzTzULR358rhrzVM0PcbMYkBXC4tEzhidGxT4f/AAp5uk388Eei34v4WnN57mrtQHm5vZnCyRZvZXSe/GFWkdkfReXGncautPSdrVzHqlqwdY0CqusuZZgXrtJQ7DY498GY+t9CE8VkctbZvWBYCX4NlK676/ML5B3QxPmQ2wXHJf+h8zG8cB0uLAtXKXwur6vJeyYSc8U/LdMMcd6yK8nTrxc7yPHwMcN4CSc2X8EO0VHT6rWzsvjW+6n9hW7j7cXWwc3CCP+/oiXFn2hNw/ZPVJY4aJD9Zifzyswwm4QnpbfG/YjZz9037kpZAftL5WRsZuWRvr2AnvJxeC+zpKenYf5XLvLWfcgbz2P4DBnzPAk4FrJP2C6XPMoMYU/SjCx04oiDrThDT3rSC8Rpd0j4VY3My2n7AMMHzWk6Exs9NwbXnr9xtxa6JS9R+R/p/Ld1zSInOXGCvVQNGPAodTPlj0xC3sEtenf5NkK2AXSQ/hz6avE+gR86cGKIdaGdzWI8NffYTkpMQdB02RIwjmYGbr9f7WNF6Nj3HPYu6MR71SP0PeGJ0VFDhtZA5K//rlITM7bkD5oHNGnl5r+48Dq5vZXQBpY/cToB8F0SOanoVoOXpvvGaZ2U8lLZA28J+Rx4nbt0uZ6qb1UWB/q2Tp7MCe+AZtmZYCRZ5u/tD02cd7yJjDcenfoNxnlbgrZvYXSbXzppmtK8/2tCVwiTwO17eBs7ttss1s9/R/33EhW2tcSa83s1JJebbEY2/92swelGfR+uQoK7AUFypZWr/GpuKDHYkrD/r9OzdKmplcY78p6XJc+VlHzv4pN4X8QPWZ2emS3oLHfLrKurjpdiDXujL33nLGHcjcv2Y8Z3Cl49AeLa33dtKEgqgzTUhzXzQIdQ+yzGjHQFbWk1GSfME/y5QZ5TnAl0r6BsuDoS1uKbJ+Ojn7IL5org0EOGoa0kebYGHXlLboampbiEskfR8/taw+j9IuZpjZP3AroqIk1w3IyxQ0z8kRBJ2QxwXaBljBzD6Z5vOl6jYPaYN3jKStzOzoQevLGaPTJuGVDB4UeEU8U87SZvZ8SasB/21me3Qp9mNJbzSzQces30naGc8aNQP4GH1YNbeUQ62f1SEjaQ2fBy6S1HKPehNT7vZ1tDaO90paFbfCWqKHfD0DPbfxNuDFZvZw5W/cm1xhrqOHgkgpc5lNZR7qhxNbioYBOUfSN5k6bN4Kz568EnQem80z5e4B7CHpbcDRuDVGbSDnOteyyt/s5GL2ajO7CA+nMEg2uGzM7CFJN+BWYrfgSt9+rXoG5RlMD0fwcLrWDy0LljskvRW4FXhajzI5+6fcFPID1ScPCbAZ8CvgY5L2MbOD+5Qx17oy995yxh3Ia/+c55wzZtHpHcP3NdelNezECAVRZxqT5r7NjafFP4BLBtT25tTdMvFdjDwz2pGSTlmOApZPv9+KT5IlOQh/bz6aft8OOBhf6I6dNFh9D3iipPNwjfYJeErSzUvIUJFlCfzeW5nUzgF2MrO/FBSjCRZ2AEh6Pe42UQ1q94VS9TfESqRlkv7hyrViMYgk/YXOpvMlranObPt9kJS486IcQdCJL+PWQKvh1gL34ynqa7NGJe7O2bjWrKX6OUxYBD+pXsj/jPpRrh4KfAl34wK4Gs8wuUcHuVpj1gzg05Luxzew/Y5ZH8FPrfdKf+diPOZRN26W9HncXR58bfv7Lt+fg5mdIWkdpg7J9rEe2eCAE+SxkfbG3VoWBD7XrUBy03hrmzXQD83sNTVFHqkqhyry/kvSXNc7sDlwQFpjHmFmd/QqANyWFBvnAT/F3Wb6UTC1UmVv0Hb93dSMzUmhuiGuTHoJ3sd6WS+15uMl8Pg+reDuG+AxseZSEKW/fxHTrfRaDDyXSzoXt7zbp075q0q2QNxCfyn6iGeTycDxwSp8TdLi+Lr7+/je6GOdvjjk/mngFPKZ9b0NeKmZ3SdpKTysSr8KoizrSjLuDQYfd4Zs/9Zz/ix+j7XPua3OnD3RZ/HD3WvT76sA1wBLS9rOzHLc/0ZCKIg682HcUmUbpqe5Px6f5EvyLDyLxg/T7xvh2t53SjrRzPasLTk81Y6ZY0Y7UtKi8Ah8sbasPJje7mbWz6A0Kl5hlSj1ki7G03KW4ou4/+75uGvbecDnzOzAgjK0OAI32d8FH+Tfn66VdLlrguUOkvbBF2Mvxhc4GzGA2fKQdR9rZu9NJvudAg322nCNjAy3kVEzcSsqM3v+pGVIrG1m/ZzoBcEkWA94GXAlzImdM6t7EWD6xnUWrpS/kt4b16pL9CzgLcydDn0ayg/wvpiZnZ1O6DGz/3RRUgw1ZqWT8fXlWX2wlOWnBx/AA8Vek34/B08z3m+dNwI3DvD9lqX32UnRM8t6u4U/ySqxdpI1ULfsVA9JWqldeSfPfvSvmjJVGTdKVmw74FbJvwAONbNuyoNn4/P+a/G12SqSLjOzTlYB1boGmiMk7Ytv5i8GDjOzC/op13ItS1YXqyYrJJK72tdryrw//T+qufxT+AH7O3H5O/FR8uLZTEPSb3Drj32s3l0zJz4YSa7vpx9/RTqo7kL2/snMFhjk+0PU95ClkABmdqf6i+lD+n6WdWXmvc2pk/7HnWHav/WcL6P3c66Ssye6CY+RdgVAsjbdGdgCV0KGgqhJmPv0Hkt5ZVAnlsJ9xVunKF/Ag26thWduGqeCaHkz++wY//6gfB6fjM8CMLPLkx9qSWZIemJlEbYohdwOW5jZ2enH4yR9aULKIfBgxBtXft9d0tUlKlZGdo4x82Z8s3OFme2Q3tNvFKr7q+n/ccRYGIhB3UZGTUOsqACQ9FrgVy0zYUlPBVYbt+VnhR/RJbNTEEyYf5lnFwPmuE73nEvbN67JLaeTtUN7uWmHCZL2AnoF9M8N8P6YpJlMuV8szZSCqZ3DzeyNA/59JD3fzG6puJK2rgPdXUjN49+8q+7zHvU+F3dtWpXp1rIdlWZJqbMFfngCflLez4ZtAUmLtlxP5Gm0Z3b5/hdxBdQX8bUxeNajz+BWVj1JVumfknQ6fiD8Bkm3AB8ysws7fP+x9Hnr31LUP+dpaLAscp/ALaVXBvZvdwfs4xBomZZyKH3/lqQk6iXjm4D1KzL2dH+UNMvMqpYav8IVKt3etdx4Nu1sgAf4XavuC5YXH2wOkjZg7oxYndabpfdPOfU9o20NPe33GhfE9oQArbFxkWRdOZZwG4OOO2S0R91+olJXr31Fzp5o1ZZyKNVxpaRVzOw3aT09MUJB1AF5RoF/tMxVk2/lGngAtRIBqqss3XaK8ndJS5rZ/ZLGndb7zbj5W2OwuX3lS6c2/y4eZ+X49PtmeDrEYsgzPLQGjvuqv49rcK5hAUnPtKlMHM/Es42UICc7xzj5l5k9Kmm2PLDdHZKKxINqTS6tU8W0kMbMJpHpLtdtZKTUuZr14a4xStrTcd9Hj3TcI2aii4sg6MG1kt6DH7osi1sazLUJ74WZ3ZBOXQflAaYsxOvIDfB+KHAqnip8D+B9wG413x3YUiJxEG4F1e5KCj2snJKlwPZUNv/AN/p0jzoKV568FLe0/yBwc009S+NWI3fgp/Ez8IDEn5a0Zg83ru/jsXpamb0+SJcAsGb2Q0n3Arvj7nYzcNe+9/ZjcSOP5bgZsCPu7vKZdJ+tjFfLdihzHe5acwYeT2j7ftpQg2eRG9aa5y5JnwW+mX7fBg9L0E3GPXGXodZad+/0zHrtCW6V9D3c+qpjv+hAbjybaSRrujtxhVRHUt/fhrnDAfQME5Hc0lbHLRZb1jJ1682B90+SfmpmG3RYv/TjbpqzXzuX6Wvo6u919/VA5bP2NUatq9iQ9wYDjDuJnPbIccmskrMn+qekzVtWS5I2B1pxzSaxl5lDKIg6cy6ugX5Y0m541qzTgE0lvdDMep5WjZAb5FH2v51+3xL4jTxjVRNSzpfkfknPYmoSWRcompnIzPaVdA1TPqa7Vix6SvASfICuDsyt33v68Y6YA/C0wdWgcSPNPFGHme2eTNUPBn5nk8+cdX86WbkYD6T6J6YG+SJIeiFu9bgKMFue6eR9PU4mR02u28ioqbptzMIXFI/UfHdczDDPgAHMcTMp+X4+RdIbqVEU2QQChwdBhZ1xhfKSuMXHj+gjG5mmxx9aAF/U93y3NT0G0QL4Ru83PYplBXg3s+9I+j2+wV4U2LKTBUpiRtuhT/vf6njoY2ZvSf/nuLQegis8qnFYVsVdb3qxhJl9S9JHzewSSb8ELsEtvNv5HHC0tWXlkbQ7rsipzS5kZntLuhP473TpCDPrehhnZj9n7rg+/XIr7rq/s5ldUrn+C3kcnU78HM+uty5u8fKIpF9Vx/0aBsoi14+Cqwfvw13Krku//zRd68ameDbnB5OMXwOuovem+yX4cz1PHp/pEOsdRyUrno2kFfC90SDB4I/A977rAYfhcZ9+3quuxH/hgdDHtZbYIv1fxFXeBshuV+Hn+Jh2DHBc1YChB8Pe2yDjThaW4ZLZRs6eaGvgWEnfxvdw1wNbyl2GS+oa5iIURJ1ZwKb8o98OrJvMH7+Kb3xKPrRt8Em2FTjsfGBXXDk0sFnygEhSbSaBPsxaR80ncfey50s6H1iBqcVDMZKZbfHsSKnuUhY6PUmL4CvxxRHA18ysSKp3SZvhC4P7cdPWdxR03enEu/AF4sfxjc9TmR4UuARH46fKLdfY96RrryooQ5bbyKjpcOr/OUmX4m4Ipbhf0istpWaW++v3Ex9kVDwTn6s6tX+xwOFBUMMM83gnczLRSHpKH+Wq669H8RgO/Yy11XfvUeBw4JQeZVqb6IEDvJvZL/CAzL3odOjTd/BWSd8CjjLPPNUv6wArWUqRLulEfGPSD61YSg9Ieh7wZ+qzQK2N3187ezEV/6gWS5nr+pQrmzRP7WpmHcNKmNl2Ndd3TOWfg1ssnIDP/Yv3qtPys8gNTLKs2aTnF6fzN1xR0+Jf6Vqvuu4GviSPv7URcKikg/A9zCFV97NKmax4NriCp69g8BXWMLNVJF2TlJCH4gYA/fDHPr8HGfsnM/tT+j/HarHIfs3M1k0Kky1xb4pr8bX42a3xpKbcMPcGg407MFx7ZLlkpj3RFUxZ/PXcE5nZb4CXK8VXs+mx2c7pVec4CQVRZ2ZLeoZ55PEHSSdHZvZI4RNgkmVEXWyRcWeLupMJazCrmNllktYD1sQnkYvNrKgFkToHAv4Hrsneb0JuPZPEan4eN7sBa5rZ1alP7I4H7C5Kmqi+gi9srsJPH79UWo7Ek9pOWL8radfCMozEbWTUSHoBrjApyf8CP5TUWiCsRMEA7sBNZrZ+768FwUQ4n7ndLTtdm4blB8+9sP0QQdL6dJk3BrXO0VTmnE78C/gdHmT4nsr1X5vZywapp40r8aw7i+EHAsdY7+D0f8Wzs7WsXGfS/3ry58l691DgCtzN/wc13320k8tVWkt3dcVK7hkfZu54LyPPnpusOz/GgHFHJa2OWyy9Du+3VzLlltKN7CxyA8qXk+a+ZaF3MXCWprJ9bUF/Cs9WnJr34u56N+GubevhhxJzzUlqi6NF/9kCBwkG36LV5x+Tx7j6R+pr3e6n1Y43Aj+V9EOmWxR2cj3K3j9JWhOPt/MCvO/344ZVbL+WlCd7AHtIehs+7uyHW9B0JfPeoPO40y2u1TDtMbBLJoCk/zWz/ago2yvXupVbjjTOaSqG3MQP8EJB1JkvAT+TdABuTneSpJOA11PIckTNCMJ7/whMW0eKecDXiVjvJH6KWy5VTbPvxAPjHUbv9LLzDJLWwuME/BMf5GdJepeVCUj8HzO7GsDMfiZpUoG6v4GbbX8T2Bg4ED9ZmQRXSFornVwj6dX0yNIzBrLcRkaNpvu5L4BvgHYqKUMyg14JN0sHuGQAc+wgmCeRxwBZGI/XUHWtWgx3Xejnb4i5g5X2igXYKf5XXzHB0gayWtdtNV/t5kqzEO5+ezzT03cPFWfCzA4BDpG0Cj73XCrpejPbsEux6/HT/xPS75sCv2qtN7utL20qxMKxki4AnmJm19V8vZsSqJebzsm4C+C5lAmncLWkNcys1uqgAwfha8I98QPLXsqJFtUscrPxe+w7i9wA5MRUad9UV90AeyoyJR2MH4T8CHhPpW8cJ6nO3b0aR2sWHsfwD0AvBe0gweBb3CtPYX42rgC7B4+R1Y1qnJ6bcTf+FnXv7zD7p2/h1s6X0n/fL7ZfkwdP3hDYCrcQPJT+M4Xl3Nug4w4M1x5Vl8zZ+CFCL5dMcE+CdmVQp2tzSMrN7fCxrhrXKhRETcTMTpT7j38UP/VdEA9edxxTAdvGTROC8DYiyKlqAs7Sv+Z5lKxjZq0NH5LOwE9a/gvoddoxr3EIsIVNBUdeG1eSrVqg7oUlvYipPjqr+nsfJ0+jYqnWQlzST/CTjaJUrNoWxk9Zfpc+Wp4+zPhHSTKPneY2MiGqfu6P4sFmi8dsSwqhSU30daf6QTBJdsMtPmcz3e3rPlzB3hVJH8E300viwWjXBi6gJlmEpOVxC8+naHr8op4KqWRhdAy+YX0MH2P/So01YnKJ6iX/r9su3dqrTJ9cj1tgLc+U23cdM3GL1xXT77/G7+0V1KwvNXf2IoB7gHtUyTbWxiqS7u5wfQbe/t1Y3MxqYxTVIeldwEmdLJd6sDpwUZo/51iCd3NFMbM1B5UvlcvKIifpcODgHhvjaj05ae6HDYh9Kx6np9NhSMe/3W6pJ88U1k8IjUGCwbd4s3n2ud3w+ENPpXeimQ92co1Lsr6wpsww+6eHzGyg1OxD1tc3kvYF3obvew7LUMLk3Fur7kWB5+C6i/9IWqnLWj+7PQZ1yZT0OtyAZClNj3W3WB9ybIpnP5t0HNW5CAVRDWZ2OVNBtSZR/+7p/5wgYqMiN9jfqCkSsK1PltD0NJ6LAE9LcVeKBiVuAtXJwcwu1Bj96NtYlLk33q3f+4oRMSLmnIIm8+ZJZB1oQnr7N3X7vLS5rJn9QZ4V5UVmdpqkJ0la2MzuHXfdGj5bx0gws71K1BMEg2Cebv7zkg42s//J+BPb45mlLjKzDSWtjMdprOPV+En3s5huHXEfva0b98fXQSfglkbb0iGb1YBMW8+Y2VBup8lyaCtgc/zE++j0cy2Z68pW9qK6mGadwi8sn1FPi+skLZU2a4OwOXCApKPwwNa9rENafGTAegbOiCXpNd3+nnmQ7W7cCJwi6S78gO7k/2fvzONuG+v+/z4nw6F06NEg1YPok5JQ0Zw0UD2lkqKUqTTIQ8mT0I/mkiFCyJzQpDQaMxUyK6UPkqKS50mOoZA6vz++1zr3uve917j3Xvvu2J/X67zOvddea6/r3vfea13X9/sZahbCOom5TxYcb7XdV2qU+dBUwfY5kj5fY78mZvDZMf9M//9L0pnAajUW56dKem1vkykVh84miha9GGT99ENJr6p6v4d1vsTI3IuZcs5+xdHdCDuJtYDP9875ywqqCW1+NyTtSHhN3cEUS6xsrt/4/ZD0Ats/LZrPlsxjH2Dq+phvevwJ+EzFaf80G4tDMCkQNYakV9geuXHUbFhwdbGYqgO3NzUbBb5OULO/nh6/iZAgPoLhdQJrIXVZei/oXUgPM5wl6W1O0ayS3kokcYwctlfp4jw1IE03wpv2uMbNcmDkGFzPtX3JqM9XgDKtd+d0WUlbE/5HSxEGlCsTE+qXlx03JHSaRDLBBP+m2EPS3LRQW4tYcJxaQ6Zzn+17Jc2VNMf2takY3BeJ1XO8pG1sH9d0kA4D3SUdyVRHSbqcWEy1goefgPQtoii0ge3aJrpNZXpuEZAx4NxtBcLT7qdM93sp9SCyvanC/+7dwOWSfkJErp9bcVx2H314elwnUKBpIlaeIbcm01P0FhKFz7IxHgAcIGkTwt9nf4VJ+eEVxZdOYu4TM+eeniZqJTTdgyhLJly6ZP/zbb9E0udsf5ia3kjp2AuB/yIKnVcBd0r6ocvTqW8nJHJbpOsAibF+NgUpVW3WT7mm0hzi+ng34bVT2VwacL12CsE4PpZq2Vcrhtkgv1vCrsBada8pLd+PbYCf0n8+WziPTdeO8yV9qy67L4eLJZ1MvP/569xEYvZviKOBJ3VwnqIP6GrAE+k2znxWQO3NzYYG23sq0pA2TJv2sf299HNnBrSSjiMWoFcyXbfaxbnzF/oPSsomHUsTlPP/6WIcswSlhdyOcbikfxKFkJOaTNAGxRBo6cPGLsT340KISBRFlPDIYftPqZN6qFMU9QQTTDAD5wIvVqS3nEGwXzYhJull+JvCd+Qa4HOSbqHefOj2fo23iol4Vsz5g6TXEk2gR9U4V2ewXVgcK0JTmV7Ba8wDng78dkTNxJOo72syDbZvBj4i6XvE4ncTSb8FdiximCiCDE4i2EALJV1FSOjLzKMbJWLZXmQZIemq/OOGuISwv1iHsDfYXtJ+tr9QsH+XMfcmpO7fZLpUr6x5mfcgepAwcy/zcnyMpP8ANk7SsmmsNveXO2Z4hMOYeivgq0SB5xrKm1zbEYbIRwA75IpDH7Z9YslxTTGuptLcuozjFpKyDIP+breNmixg+12KRMNdbPdKgetgB0n7ZNfD9Bn9qO1dSo7JrgE75bZNPIhmK3o0hHnU0U0PBb0LLkWM5scIzXhjKuxiglbmZsNGKgh9r3LH0eL5hM572N3IOpgwIxIGuFkOHbbXUfhAvQ/4lKQTCY340NNRiqAwL9wOWMP27qmT+3h3Y1yexwO27+mhPzf1pGiN1En9j4wh0dV5J5jg3whzEhNoS+DLtvdRRCZX4X0EM3BXIi59NeqFQ+QXgPOIxfWVlE/ED1IY2u5FBDLMBz5Q41yoR+IKjETiKmlFIkL85cTC4ixgZ0cKbxGayvRQpIUeSkg8/oeQ3S0JPELS1ra/NfAvk4Nr+Dn1g6SlCM/Q9xGFw72IItH6wIkUSwSPAI4kWBQQhcojiISyIjROxMqhcUNPkZq2I+F3chLwYts3S3okUfzpWyByhzH3xLrylwRDKkPp7+qGaYEEa+4WoimZMb2ypmWR3DFDxkx6KXBKYjCWzg3SPlsA35N0LPGZ2M0tvXRKzvM76J98Jel/KDE7HhAXS1rb9ig9Kw+3XcdXqghnpbX5KUxn2gzVbzT9rb9CmG83xYvy13jbf5H0korzzbbm6iJMCkT98d/EF7FfEaJTjxFFrN9ehK76i8BTKqrjXYzp2cCttitj/4aM1uZmgyKjsqogxraK9jwC1KaSDxuzTPI3QQ6pO3qhpHWIFJEPSPoRMZkpShAZJg4gvD7WIzpzdxOT1pFL7Xrwl7RAy+jgWwFV0c/DxiWEd8FJTO+kjrUzpIgtXgDs5+I0pgkmGDXmSVqaWGwdkrZVNn5yFP57ifSXWujTdHsaFTHItk9OP15GA6D8GZ0AACAASURBVD+dYUhcFYlit9k+tWLXI4gFeeantEPaVsZobiTTS/h8OsfyBOPj9bbPV3ggfYVYtA+MtADPz7EWEhKfM6tkYgk3E2bdH7R9cW77TySdXXLco20fk3t8rKSq5Ms2iViD4DhiHfD+/DrA9l1JElaIupJCDRhz74b+VokN+DYiGWwhIfE5rayxYnsvYC9JF9gu9XXqg/Mk/YpY/75H0vJUXHdy78nRxPf4R4Q07dVpPLXv6TXXT43TsAY83wbAtpLM9OLLMOdtgzK4M8bb5rltjf1Ga74fN0paJTERm6BfYXLJgnGs6vACe1q/54dd+GqDSYGoP35BpCDMqKZKqj0hGQQKt/YPAe8laL9PHxGNtw0+C6wt6TtukTIxAFqZmw0J2Y2xLMa2S1wPnCPpO0y/oHfmQSTpicQNq3fS0ZVB9AQ9SAyiHYkb/lHp30ZEsaixFKEFXkpE4V4Jizoo88oPGQl2ITqsknQz0Ql9bcdjWCf9/97cttlAHf42sdjdn+mTrQkm6BJfI3xQbiDSox5H7l5WBElrECyPlW2vKmk94HW292lyctu/Ssf2O8egHpDDkLg+AXi1pN2cS07tgyfb3iz3eG9JV1e8dhuZ3sOyuZekj2fsWdu/UEU4RcH7uQC41vaCnu2X99l3ReBQSQfZPqJinM9ygR+P7bL5+78kybbTmJ9CdcGyUSJWz2JwWvJqGl/potD2M0qeK3xfGkoKB425n5OOyYqhZwJHOXn39Oy7Snr+DqKhAsFO20nhs/R220cXnatFcQhifvRM4Cbb/1AYjVelrubfk2sJm49sW9N7euH6SYOlYTU+Xw5lEqhhYY6kZSj4PaqIDy1YZkWo834sB/xc4V+Wb+5VEQEuS1LMfYnfczfi+9YPXyS8sH7Q57kug3YKMSkQ9cce5D4UPWgcTdkSvyU6ZP+P6Eg8N38THmcX2vbLFR4bbSh4jaHBzc0GRpKVtaY9jwDzgN8QXZcMXSdoHUPQPdchOkDvTWPqDKn48DZmmnU/lHyQAEjyjLuJG89Wnko3OTExaLrAfY5Ev2xMc+kofjUPh7HsBkRRbE5s6jbmfrZSh21/Z9xjmGAC2x+TdDCwINH67wE2qzqOMAP+JDHRB7iaYLDsU3ZQT5EiM8ItkmgParo/sMTV9h6waMFdhrmSHuOITidJnKoMpdvI9PLzi7t7nquS0X6UKJhlEsJnAD8HVpb0TtuLGm+2D+33ApK+SHjnlBaIHB5wjQy4E/Yg2LdZce2ZVL8ny2rqj/wd20Xrhgy9i8H856hyUShpJcJLaKO0/48JOWFVOlhtSeEQ7lv7EoWkTKq3NbAG/b0p9ycMxKdJ4yR9gCgcPY5g7dDz/CCM/jWJ1Kkn5L6f95f9QsO8l6f101zi89WLQdKwys5Xul5zC6sESYcDh7i+KfPaxO+Wv57VlQVm53wZIds9JF3nlrd9fZNxV7z/GU5M/5riAwRj/irid/o+BcU3J3/KIRa+ho5JgagPXJJSZvunHQ3jOuID9tY+z429C50WW1d1dLpZ43mTLkrvJzrw+YJEpxKzpjTeEWFF20dL2sX2xZJ+BlxMeGV1hW8QE92fUXGTHwWK6KEZOqaJbmu7X/cV25t0NIZfSHob0S1ahZBZlMbOjgqps/sAwRxanWBRdQbNEj+mxEbdg4jzfasimvepk0LRBONEvmDTU0ipkujMt326pM/AIs+IquQzmF70eRC4kQIG3RAWhI0lrmn/39u+T9LGxCL7CNtVvi/7AVdJyooPr6YgWSlDS5neqppKb83/PIdiX58MNxKyqCsAEnPrg4Rs6WRqMLNt3yGpsgnWkC2Tf/3TJT2dYN8CXGL7/wrOMRc4kGiK/Y14D+YpTKo/2I8tk84x6GLwK0RKWiZ92y5tq5IutpEUImk+IKYX2spS2gA2BtbLGlTpc3IF/QtE6/Sw37JzHChpD+AFBecYhNH/A6YKE/MISfzvgJEv1NP6YTVH2uyM9ZMHS8MqRNV6Lf2dP0w0e/N/641KXvZ6QkJ/GyG7+1auKdkP19iuZKCVjHF34tq2EiFJXopoUr+wxrFLEp9jgF+ncc54PyTtb3tX28erRWK5I65+uybHpPMuSzBG82vKicRsgv6wveG4x5BBY04P8/iM2/rhNEI+czZjNMpuQuMdIbJJ+T2SngT8GXh0h+cHWN32mtW7jQz5ycaTgLvS4/nA7+lm0vHqfj9n6Jht+EHCh2glomj3Xaa8MUYOhcfErravURj7X04ULVeVtJZrJnUMCbPFj+lLRAcy65jdSizKJgWiCcaJNqbREIbASzJVfFmZEgaLpJfYPr9t0SddU7OF0jmuJ3FvI3H9OvAcSasSLJkzgeOB15UdZPsESVcylax6ECHbK4TayfTynfBeJkzVQv2ZWXEojflKSc+wfV0NhlQ25tK/cw6NDbjT63/BkTb0/T7bevF+onH5jJwkTYSk+/0Ei3cUWMn2x3OPP6kwea9CY0mhpLcQxccViKLt6un4vrLMHDJGSIZsftQPZX/PBUXskDJGv6TSBmVvkS6xUgYxTy6FpAsJKdEcoihxp6Qf2i5jKd6UCuD5+f2nymRYOaVFX1Ss144BfkUwrj9KFDmuKNkf2wcAByikgO8D9pd0NGFG3Y/RNujaZEviO3dpOv+tCnP2Ukh6ITHfyRdytyho1OXvEZ8jDP8boSl7UdKOBBv2r0xPpJ5IzCb4t8CsSA9jiMZtA2BZ2zt2eL4iNKHxjgoXKEzUDyNuJvcTUaBd4iZJy9nupbx3gmyykejvF9j+Rnr8JqCNPr4NsonGPEI2kafxX0qHbMP0d3gX1Zr+UeHxnoonfRuxoHuTwojyQkJO0RVmix/T2ra3TqwEkvSlSoIywQQjRW/BRjVMoxMOI3y0VlREXL8D2LNk/wNJi1pJF7vcz2caFMa/ryWk1ACfkfR826VR3y0lrv9yeKK8hpDd7KtqLyEkfT4tNq9Nj5cg3p+yglRjmV6/xXgD/E3Slk6m36mokSWATVs4qn+K8H8QzJQ6CXKt2DL0v18XJRC9DXiLcya2ti3p7YS31qgKRDdKWt32jQCSnkwwOarQRlK4B/As4Azb6yr8ceokoZ1BmFsflx5vnbb1w42S3ugeI3ZJm9HermBbYO+6O9s+R9LnW56rDh7hSLfbCvgq0Sy6hvJr3ReJ9XlWnHwnwZopY6dkSovtie/LkcS1Z3vC46kMq9veTNKmtk+WdCpQxxAeYl34NKLA/zxge0n7uUc2SJjHD4K/p+tjfludotOhhOXC+bDIp/NL9JeYzSn4uRZashd3BdbyLAz/mRSIJqiDsaWHwciM29riZ6nzVSeOd5RoQuMdFfaw/Q/gK5LOBx5J90lRC4DLJZ3BdLPurj2IXmx7p9z5vylpry5OnC20JJ0M7GL7Z+nx+kxR0UcKSaXFsBq09GEhb3L7AhJLxvadqoiyHcVYPAv8mOiRXqYi1aRANMGsgktMo3v2O0HSTUQBZFlga0d6YxHy37mmBdrNgXVt3wugMCC9iuiyV2FpoqG2BKmxXCEbmCfpscTvlRW86lwvnihpR9uHpmvMSUBVYmVbmV5bbEvME7KEsl8C75D0cGYulO/tebyQ8OT8vOslcTZiy0jaHHgzsIqmZHMQ88sixsby7pNw5Iicn19jjI2gKa+dZYBrFOa5EPe4StuLlpLCB23fngqO2D5L0udqHPc/BIsrS9H7NsW+UbsR8eVvYsqk+nkEG+4VNcfZi9LvjKbbAmReZEtXHHOy7S0l7Wz7oIbjyV77pcAp6btWNRd5ju1FnkGSLiI+z4XIKS1ebTtvy7GTpMsoL5plc4QHUtP3r1SoASQ9izD8fiVxzXlx+vw/kihWTysQ2S5LVayDWxIbaGG6zu1BXEcq4ZzHku0LVWyqv7SmjOPzP2fHVsm+2rAXb5uNxSGYFIhKoQJJU++2hwDGmR4GIzBuGwCHE8yZWxhdHGQdNKHxjgrHEZ00bP9e0nIEJfO5HY7B6d+4MUfSi7KFiqQX0P0ifK2sOARg+1JFBHEX2D/385qEh1qGhXQnq3pA4SNxO9H9/e/cc12zd2aLH9MFCj+HpSVtSMgATxvDOCaYYBHUzDR6Gmz/hBqR2wm998km+CvTiwT3pW2lyMkG7mBKRlMlG/gCcS87x/blklYjGiBV2Bo4XeEF8jrg9gr5CjSU6Q0K29cBz05zhIxpmuGsnn0H9TBsypa5npDMrc906dxdhCl2P5SZUfcWuIaBvITvq7mfSxu3BWysRahopN2vkP/dIGknggHyiIpxAmxo+3BirpyNYyPCULv3/NemRfhOBNtiDlEIeZ+rvbeKUPUdz/+NHyTkmFtXHLNW+n9rQsLZBOdJ+hWx3n5PYjNXsQnnSHp4VpgmCuF15/fzJa3o5J8laUWi2FmG61Nh6CSiUHcnFRIzYv7/RcJbbNE10vZdiXk5bOxEMHHWIq7JF5LWHxU4S9LbbH8VQNJbKWa0Lct0xn0jE3kasBdzhcqz0vf0FKavKSceRLMcY5c0SXqC7a5ZGb14N2NKD4PRGbe1xInApwjZyDjldv1ovKd3PIY/aypNYhliEvPVqoOGiSFMJoeFHYGTJWU39GUIzXSXuFfSVrZPBEjFidLo0GHB9nOynyVdlX/cMT5CmHg+HPhy1uVNLMSuC4lj9WPKYU+iq3s3ce/6LlPSkgkmGBdqm0bDoiLGHkSB5gBC+v5yYoG/TQmrV5Iu7fMzUNncuYi4z2byqq2An2TFLRf7uzWSDaSO+K22l89t/h3V5sPYvl/SG4kF0yW231/jlE1legNDIXF9efr5TDc0gK2LpmyZJEm+RtJ3bVdJcTI8oaD4MgdYuerg1GD8cfp3TtUcfwB53yDFqr0IVviHCUnOfKL4VoX9mOlT1G8bEOxewsKiNlSQXka8/48qOW4u8PqcDL0uLpe0AFhG0u0956taC+0J/CdwU5JILUG1BP9E4GJJmbT1LVSYrOfwBeBqTTetL5XW285Sbg9I18flKVlPKFLRvm37yILXK00abAPbtwGvVBg6z3VFYqCmp19/UNJR6amlgf+jj9rC9ioDDrMJe7HXxy1/75sVHkRzFi7sOhl79kNTkqa3M/1LOR9Y3wM4sbcYyx8Jk9VDbc+owHc0hv/st30ctDhJr2Sm0/7Hi48Y+vmvtF1Jg+9gHHOZblJ9NnCk7ZF1AfuMYQ5hrHkxsAlwdlfsOkmb2/6GpL4TFtuHdTGOnjEtxVRSgm2PkrLf7/xrEj4ST0+bfkHIL64rPmok4xjrdyRNXpZLE89s28OBOVWTigkmmGB2QtKZhLxrOUKu8VXC824jYAvbfWWukop8ZIDyiGdJZT4cC12Q8iPpItvPLztvn2MaXTc105R2GYJt/U+oNKXNzFtfSyygvlch0+t3/FJMT90pM9DdjWhinZw2bQEcb3u/JuesMabXE0XErDt/LfBZlyQ2StoCuMz2b9Ljw4j04N8S3iUzZCySSj1uqhpXkh4NvIyYv21EMOfOsV2nADMU5BnPDY55ZlFxRdLqhOfWwUxn7s4H9rb91NaDnXmuUsZPWUFN0s/z8q0G53wsUdDrFwTSdy2U5si/tF2adltw7KuIzwjE3Lp2A1jS2kz5Z51XUjzP9l8L+G1OSvsI4D/7ffZzx/zM9gZFz5cct1HvWrbfttxzrZKCi9atueOGvn7N3keiQflpotD2SduVXnKzERMGUX/MJknTKkT1+FMKbfOhwAnu0JR3tugjJX2WoKE/nZBIbEoURrrE6ZI2aXKxHgVSIWgajbcrpAp+hvcAPwLOAw6RtGzZRHGIWIuIuO/HVOms6t3zXsCUseISkpbo6L0AKmn8Dxk4zGDv7Nk2Ctp/X6hPklwO9wE32v59B+MoXWyMo4g6wQQZUid9B3IJYQTrr8if43GJrToX+IPtT6btv5b07qLzlBWAKsY3F9jZ9s8bHDOIbOBqSevbvrRknzyeXb3LjPH9njCfPZdYOH64xWu8gZCWrJQ2ZXL3slSstwPPy+5Jkg4mvHOGViBK190vEUbbP0vjWh84TNL9JRYJe5Bk8ZI2BV5DNIifRRicv7L3gEGZy7b/N7FgbiHSTrehg1ALSSulc21LvD9rFOz3HCKV9TxHuMLTCVPzFwBFhccXpNd+LNPZgXcxZObsAIwqCGPsVdzHQ6rinH+WtEGTJpPDg/AWSSs0lcylz2srW490zap93SLSEvPWEA+kbWXXmB9K+hBBolj0ntSY8zZimBFF3lsJlmmvzK6QadPlulXS+bZfkiRlH7H9GWp6fUn6uu03V20bByYFoj6YTZKmxEL4CmHw93xiwvGZRHn+hO3bS19gAEj6iu23KwzOZiy6K6jZo8BriFSgK2y/W9LHgS93PIZ3AbtrTHK7WYKseDon9/+zgQ9RPVEcCmzvnf7fdtTnqkD+vciQf29G/l7kkYrIIgxPgW4Monu6PPPU3NxvcUKZ/8cSwOqSDrPdiFbfAmUyvwl1eIJx4zBCepEt9t5OJMu8p2D/B2GRoXJvjPLQ5d7pPCcCTdgGg8gGngX8VNINTF9w9Z1ntVwAvYEwAN4MOFDSnURzJysY1Slcf54wdb6kAWN5Tr5hYftuFcTb92m6TEPJAnRngkmWLwhepfB/+SjFi+2FudfcBDgmFekuLSs8DgJJ3yeav5cShdEXuH80+DDOtQTRTN0e2IC4B21s+5KC/fck5nIm1hqHEI3xQ4gCUF+kos3xkraxfdwwf4chYzng5wqj7/z3rM6CfF6SKr2c+D6fRRSR/7fkmAXE5/CHPeebIXEqkc5BfE7fUnSSAddrD3MEzmT7PpA+N2XIWHT7UmPOm2OYPbKniTaf8P4pwgnACwlJ7HFlrKaC8/YyLYFqhmVDPDL38+Y0I5Ks3mfbmoMNZziYFIhKkKqBY5U0wSKq3HsI2uvZwFFE1+0MomAyKmQu9B8a4Tma4D7bD0paKGlJ23+Q9ISOx9C4a7e4wfYkASlhNr0Xkt5CdGJWAP5A3HiuobgzM0z0LoyamvstNnBPfHcv0gLoChr6LrQYx7iLpxNMUIYXA0/LigyKBKmyyf+jc6y4/M8AK45ojI3YBrZXHeBc/129yxSKFoK5scxYENq+grj27J8YUusSBaM9gWOp19C4w/ZFTcYKXKZIMMsaetsDlxfsmzVdilA0xlX7scUcqUVlcpP86z2f6WyXUTV45jIVYrGQCoPwAWQ2BxJeiD8nTIU3A35VVBxK2ApY0/ZtCoPdXwIvqfqbJ9ncvVlxKEkY3wTcRFhkFBZxJR0OHNJRQ/7E9K8NjiDej12Jgsi70rayhK5fUjNti+lm5BnmE3H3Vde4QdZr/5C0mu2bACQ9mYqie4u5byuGme1tFBYBm5NUCkRT4STnbARKkF+zzSOMrWuFITRA44abpHcRDNqnaLov3nxmR/DOpEBUhtkgaUqdhqcTF6H1bP8lPXVR0k6PDGky0ZqiPQLcnS4OFxGdij8Bf+9yAL1dO0Va1XbEhGeChzDSZGpN26clDfdSrm96OQzsQXShz7C9rsJL7U1dnHjAhdFDCrb/JmmvUZ9nIjGbYJbjL4RhaHYPXxIo68SfzRQrLv8zFKdNDYpWbIM2soFsniXp0RWMhAytG3eJ4bkh4eW0HiGNrivb+bak9wJfY7p8rkxashPB4jk4PT6bggJ5tvBM18j7gSOJxfg7iXSyIrRNFjtf0snAbcQi/IJ0/kcz/IUkALZfnRgazyWavZ+QdK+LvXF+wBRL40nEonohsZj8PVB0/3034RH5GdvnAkiqWsz+3WEIjO3rJblmQfDbhM/UbxT05R8R6oc3pPF9oOTY64FTFUl8hwLfKpGatoLCm3DpXnlaWlPc3/+oGXiy7c1yj/eWVOov00SOmB+bpKWJovEHCa+10obSgOu1jxHsxbyxdZWRdqM57yAMM4dFwHHAcQpvphOIRL1Kz9M+TMv/J+kShtugK2teFM21ziQS9A5hZsGsiTxwZJgUiMoxGyRNxxFu8TOqubbXmrn78JFkKx9mJpOqr0HjCLElUdX+EHHRXJ6S1JNRQdLjiBvhtsRNutPkrjSGselWiyibPDTldgBI2gbYnZjAnkYkmRxKjRSaIeJB27dn1GDbZ0n6XIfnnzWQtAbREV/Z9qqS1gNeZ3uf8Y4sYPtbHZwmW0CvSBhWZovolxGSkkmBaILOkZs8/5JI6vlaerw5cFnRccNgxEl6GbGgOURhOjvf9vUlh7RlGzSWDUjagAh9mAs8UdKzgR1s79Bv/zYLwfReP4NYmFxAyMWuLGN39EEWYX0oNeXUaYG3e8PhvtHTTbv3k3QFxYlM89PisZ907ZF9tmXYhZCnPYGQXmXhEiLS8gqRmrTfbFrMUESPbwS8grge/5NofPZF1oCR9EXgAtvfSI/fRLl30eMJ5cHnFTHmJ1C97pvfIwGal39fXZzct4KT0TcxV/+m7fdJmkcxW4z0mgcQCVqbEElp+0s6Gji8THon6ZHAPUkOuhbhTXmq+weEfJZgZhzVs31L4m89Q/LVB3MlPcbJ2kPSY5highWN8THE5+hJtl+sMJB+vu2+/qGJ2fdOIkXuPMK76+YaY8uOb7xes/19hZl/Nl/9rO0bK86zDS3mvLaPSwylJzPd5L7oc5X9Tm8l0hb/ThRUvlF2npLXWo1iH622KGte9C3IpsLV7yS9s4LRNzZMCkTlGLukyfY3Ja0lacO06dymGswh4BjgV4R+9KMEY+aKjseA7T/Dog7bJ6v2HyZS9+F1BFPoucCpRFJSZaTpiNBvAjq0lIgKzBqZXdsJ2giwM/G+XAgRYZYKiV3ifoWvww2SdgJuJrosD0V8iTDVzOLcrya6mfuMa0BdI1tQp67gM23/Nj1elalO/gQTdI385PkqYl4BIYddclQnlbQ70RlfiejaLknMbV5YdEwv26DGOQaRDRwAvIrUcLJ9ucJrsuqcTRaETyYWLH8kjF9/37A41EpWnYoEb2PmorBsUb6MpNWzhWpaVJZ5ldxC8SL/lqKDHN4rM8yybf8E+EnJ+SCKC/tJOgY4wvYfKvbPcDVTMfefcP3Qghfb3ik3xm+WsVGTBOcwwqh7bWLuPk/SBcBX3T+O/PdMZzTk39eFTJeP55Fn4TyXlABt+z5JdZlYlxAJdOsAzwO2l7Sf7S8U7H8u8GJFMMcZhKHxJvT3StqI+J704jji2lOnQLQf4SeUZ9pUFT6/TLCpssL4r4mi84wCkaTNCWaLgVe1XOu1Wq+lQnlZsbwXrea8kj5NsJOuY0rGVvi5yhW1vwZs2aRYlo7PN7TnEtf9nZu8RhUGbF4cLumfRHHtJNv3VR3QFSYFonKMXdIkaUdCOpJdkHaX9CnbX+pwGKvb3kzSprZPlnQqcWHuFE07bEPGn4iu26HA5rb/LummDs47DQNOQIeCPpTNcaLtBG3YeMD2PZLy27ouWu1FdEo/TBRI5gPv7XgMswXzbZ8u6TOwyHC2X1fxoYD/zIpDALZ/m4pEE0zQOXKFy0faviv/XGIEjApbEguaS9M4bi06n6SdbR+kSCKbgZLCxiCygaVs/6rnHlLnmlV7QWg7S7l8ISEx+4DC3+OnhEn1KTXO1wbfIJgGP6O+nGdP4JLEGppDFA0K53q2NxxwjI1he1NJqxBSrssVUsTDMjlXyXFtG81zlIunV1gc1CrYOZKtdpG0G/B6ggE/o0DkCg+9EixITKM/MuU3kzFi5pUch6RnATsSiXEnEYWwm9P381qm/HV6Mcf2vZK2JBIQ95FUFOn+MPcxVbf9T0m1zNZtn5A+j9l7dFCNIs7Ktg9XMjx3GEAXne9rwO+I7/3ePdeCukbatddrGszYuu2c982EVO+uyj0DmwN3AO8Hdsydr65iId/QfhC4rWlRfJSwvY6kFxEFxE8pghG+5OQHNU5MCkTl2JL4QI1T0rQzsG6O0vho4mbeZYEou6E/kGiqfwUe3eH5M7TqsA0J5xLUyY0JA+Bx+TLNGt2qpCcSGuBnMr1z2ZkhcdsJ2gjwF4UeeyGApK2IDm1nsP3j9OMCEs03FRRH5c8xm/FPSUsy9fdYmQoj0GEjUct3YmbXvOv40tskfZQpav12hN/GBBOME+cx00C/37Zh4e+2/9GzoCnyY8m6uGXeNTOQyQYIqUtT3K/w8ciuWU/LjaMMjRp4jjSxHwE/Steo/yKaCu8iUnL7QtI5tl+mmRLzOgu11W03SuaxfaqknxJR9RCpaXW8mTpFYjR8RNL3iPdvE0m/BXbMCjm9aCo7ymFH4GRJ2edyGWKd0mS8/yAKdq0kOiXYmSjurAx83MnHiPh8lUrMCBbPF4H3O+dlZfsuSZ8qPCrYUEsTUr1D0raixf8ykpZ1j1dW+s4tXTG+RUgFoSbMnmlFE0nL018GCVG0GxRN1muDGFu3nfP+qUFxCIr9tWphljW0+yJdJy6UtA7wXaJw/yNgN9u/Hte4JgWiEmSSJqKa26mkKYe7nIuyt/2/kpp8uYaB69OF5iSCAnonY5CY0b7DNjBsv0XSCgRN+sD083xJq+a78x2MY5AJ6LBxDDEhWod4X95LmF12ijYTtBFgF+L7IUk3A38DXtvRucvwUTr0TSvouC8ALs4VsLrAYYRp5oqS9iG063t2eH6AbxE06rMZQQx3A7yDkJRlCTHnpG0TTNA5FB5pSxF+HsswtViqijseFLcokpUWJlbDHhQs9DLpjfsYzKZu7yjwKaIB9HhJxxFSma1qHFd7QZgajBvm/q1KsHpOppoVno2ljcT8JknLORd1Xwe2/yzpDNJapd8Cf5yQtBTwFqL7/zCCxXsKUdQ6kYiy74fasqM8HIlsqxGeOWlTX7+dzpEYSjPmpba/Syx6y7BOL6tD0sNt31sgg8vwNaLZcQNhsvw4iouqXyOUINtnBQqFPPMIhl8sy+NUSUcAyyl8e95HzJ1nwA0lrQWofk8+/QAAIABJREFUvV6zfYXCPmMr25Wm1D34AA3mvJrytbpYYQr/Daab3PeVmLUt8PQpZGeYdV6p6Z6yI7AB0cjLksq/y5QEu3NMCkSzFJqKtTxL0lHA0enxNoTWtjPYziYGByRZ0/LEza1rtO2wDQW2/0p0KQ6RtC7Rib9M0g22n9fFGCR9g/Jo2y7ZCSvaPlrSLrYvlvQzIjGjdmrDoBhggjZUONI+NiAu5nNi06ygsRZ1qkaFxwIvAr6THm9KGM++WWGiXtYNHBoSFfwmYsKyLLB1h8XCDCt0JH8the0/0lGa3QQT1MCewN7EfSzP0LkL2L/qYEWHaC9mMvPK5BAQbL4TiEXs3wjvjLfVGbCklYi517bENXWNOsc1ge0fSTLBUp4DfNIVRrEJTRp4txAFofMJycbFdT0vnMyCmyzYcg2DBQTD9wymLwoLfV8kvZEobK+UNlWaYQ8CtYtav5lgvX3Q9sW57T+RVJZ43ER21IsliKLgEsDqkgpj7v+NcK6kLdK9CknPJIqWTys7yPbHJB0MLEgy8nuAzQp2/zjBVPqDpBvStjWIRfg+g/8KhWPcV9LbiHXTq4GDbbcxvq97vn7rtdNL9v9nYrA1PY8bznl363m8U+7nMm+rthibV6rqJ1GikETeTTDotvKUn+qJiZU1NkwKRLMXP+h5/LLczwvpsBsu6Qu2d4FFxn1I+gLBmugSbTtsQ4ftq4CdJO1K6Lm7wvc7PFcVss7VPZKeBPyZ7qWHN9NugjYUSFra9v0KrzIIdhfA0mniNu5uZ1Wk7bDxeOBZqZiKIvnxW4Tnxc+YSsAZOVzPZHSUuFbS47NJ7ziRFtW9UtATxjeiCR6qSKycj0k6xPb7W7zEKUT3+VgaMPOS5OWV6Vo913ZZLHrGdNqUCKbYgJgvb+ySxBlJWyaZVytmcfKd+FKSIK0GVBaIehaElxFMrKIF4Qq2u/TRzAqAprlH4r6EX8kl7uMdMwK0iVp/lgtStmy/s+S4JrKj/H47EsELdzAlmV5IfFaGhsQq+X+29x7m65bgaIIBtCPwRML4uZDRIilLbnsg/31M3+m+3+v0t9xK0upEOjXAVTWLsAPB9lfpOO04NU+vTA+Xotwf6MeSDiEK6Ivev7LCo6ZSk6/rs20G3N7bqi3+CTzO9jR5o8K7tjAZbxConU/utr1jzGB7kxEMszYmBaICJAryWok22TmcYi1nCfrFaL6k81GEWfhWNO+wjQyJ3vv1Ds/XledSHVyQOpeHER3L+xktVbcfZkzQlMxPKyZow8LFhGfGPUyP/B1ptzOPAlkXaQzzR33+HqycFYcgUlQkrWT7bkl1zUlbY5Yx7FYAfqHw0ch3zTv1IJL034RH10oEm+tFBINgUiCaYJzYQ9Jc14uozmOu7aKo81IoF6+cSdX7SRskHUh4u/ycYB1sBvyqrDiU8CGC+fAtGnopSbqQ8GuZQ6S73Snph7Z7O++FqGJJdlwc6ivRa4A7bBdGvxdB0v4EW+ReQja3HvDuKtaGG0atS3oOwdZC0usIH5zrCYPZqsJSbdlRD3Yl1iUj9VVJrJJXEUy/kcP28ZJ+TTR0bgfW85TFRz9kXj13Eqy5Jue6kRqF1yJIehkz2YuHlezfqRehpDcBBxLNOqg3F90i/f+a3LaqwmOrJOWc1CyPBcC1thdUHd8A+9JfsrlMeq4We7QhGvvkpn0W3Zdy24fNqGqMSYGoAGnSciLQmHq3uEARufhmYBVJ+SLIfIKi3eVY5hB06KfRrUH3BD2QtC2ReJJNXL8i6XzgkQ3p2cPAD+jW6LQXW0C76N8hosxI9YDORhH4laQjie4+wNbAdQojyS4kd7OJYXdS+jdu7EBILn9qe+O0GP9/Yx7TBBM0iajO42JJazdt3ikSDd9JvXjldxPF/884BR5IqsPGnCPpiwTLeUbhvkxSBTzC9oIkK/gqwaK4hpnSjOz3ub3fdmaRx0ZJ8wKofD++Lem9hHdMvsBeNfd8ue1dJb2GCBR5C/E3rivrqYxal/QJoln5sDQ3filwFhFisyZT3kJ9MYDs6LZRF4dy+IGkDzGTVVJ77p9YLI/ylGF10X7rAccTXq/rAkdLeoftO/rt78FixVsjKReeTTBz8teQMjT2IsyrNsq2FWBf4I3AFXWZd00ICRo8SfmjwHOYCtV5Rvp5ZUnvtF04h2tYSFnD9gW9Gx0+XqNaQzb2yW14X+oUkwJROW6UtIrDBPehiOuJBfj6TJe83UXHyUi2F0q6RdIKeYbCBGPBG4lO2wKiGHMuUTDqrDik8Rmd9uJk4FlKCS8dnncRBuzSDhvbEcWHLFHkPCIh559EZ2WkmE0Mu1k0lvscUcBzJc2xfa0ifWSCCcaJJhHVeWwAbKvw68kXDqo8iDanfrzy44G3Ap9PLNkTqDdffjNxf+z1V6qDLEnppcApqUlZxkS5h5AaHUt4QnbmedeARZG9B08mWOenpsdvoDoJNpMjH0o7Vu6LCUbaH+sU99Qsan0zohD0CEJWvpIjcWsp4Oo6g2spOzorFd1OYfpnfxQeRBl7aF8avP+STiEKrA8QBc4VJX3a9n4lh30L2Mb2+ek1PkCwXZ9cca5+hbgFwKW2b+jz3KB4PvB0RxJcXbTxIhxEtfEn25c1PF/2nV7T9iGJ9bS87ev77DpokvKNRFrdFem86xEp4VsR8+m+BaL0ud+aKELVKaQ8vGQMo2rotvHJbXJf6hSTAlE5lgN+rojOzlfQu44pHgtsXwNcI+m7RZX8jrEAuErSD5n+9yjrQg0dacKwuu0rK3deDGH7tUmCuR5x03oTkex2J3Cu7e07GEZmdAotjE6HiGUkbUaw7GZQZ2cDTbRLpJtcUWTqyCOKB+xYD2sMO9s+qGgsXV+vgL9JWpKYrH9O0i10IH2cYIIKNImozqOt92HteGXbdxLS6cMUBq7bEeO9APiqC5KVknxlX0m32m7KHjxP0q+Iefl7FL40he+H7dUkbUgwrnYhggGOc8RwjwxNWBRZ80LSjwnZ0F/S409SIUcfgJV7e2IIvAr4bGom1bneHUf9qPX7Hebe90m6MftcOcymC6XUFfenhbY/XDHGLH1y8/xxDNmDCAZ6/5WYcG8Cfkws/i8BygpE6ztn6mv7wLTuqsImxBw0a1hvRHgdfkbSPrZnyPYkPdU90eH9thXglhr79KK2F+GQVBtfTAy3b1OziChpd4LJthJxLV6KkDy+sHdf90lSzjHF6ly/n5kVh9LrXSnpGbavS0qRIrwBWLUBg+0+Sas5fN0WQZECOKpwozY+ubXvS11jUiAqx4nUp6UuzlgiXXA60dCW4JcUxNJ2hVQEOIKYGK2iMCHb23bnkeaJBbCm7dNS1Xqprgp5ibp6OZFM8iPCRP39xM1t5AUiD250Oix8hOiWPYaZUoBZQRPtEpLmEdru3mtFV0WRpl37USCbfMyGsUDIHZYi/Cs+TSwm3j7WEU0wQbOI6kXIWAZ1oZbxyrnnfw7sImk3IpBiW2IOUHbMSZI2Bl6eNp1p+6yKoe5IGMnfZPsfqbBRGj1t+zyisPRwQu58bloY9/VE0XCin9uwKB6XFYfSuP+S/t6jwFuJe9Dxtv8qaRVqSK1tP6Pkud6/95zEXJ4LLOxhMZctcvvdE5ZI430iwbYtG2On3qQt55hLpv9fAvzQ9t9UkdCWFYcScyULUijzIMqwEHiG7d+n459IMM7WJyRd/XydTmKmBUG/bYuQYypdD5wj6TtMv4bM+L5pyg9xPvW9CIeh2liZKMptzfQCblkRcUui6HtpGtutqRFeiJZMMYiG1Za2T06vsyWQeaOVMf1+T4Vcqwf7A9+VtAvp9yLe1wMIVtzQ4XZJlK3uS11gUiAqwTglAppdZquNNbSjwCyR0nyM0M/+CKYZjHUKSVsTxYmlgNOIm8KhTE1IR3nuNYENCSr8OsRN7QKik9mY2jogZlDwVV+rPTBsnwacJukA2x/s4pyzHN8gPpM/I0zLO8VsuEZki4nZMBZFGs3mjjSaewmt+wQTjB1uFlG9CJLmEwvpdZieyrdRwSFDiVdOBZFvUCOIIRWTtiYkExCS7OMrFk9rEgugJ2jKw6LyGirpqcS99/XE73Juye7DiH5uw6L4paSjiLQqiCJbqSxKEXV+OFE0y+R32C5lA6Viwxdyj28m2EGlkLQiwZ54OfG5OAvY2f3jqtcmWOxZMehepsuwisY27Z6QGCOfIBqfm1aNMXdcvpBCViAZJhTm2bvTfI75q9Q0XBPYPRXPqs61EeFB9FhijbEU8Bei8VaG1fK/u+1bFAmCt6lHnpn+vo8hmIBrMt2WoEyOBDHnz/AbwjcnQ9HfOy+VqsUmHJJq478JhUOTpK6/p6J0fluVLLMNUwziu/8VSVnx7pfA1qnIXWbIvyvwPUlnUVGcS9u/mj57xxCfXQhPsk+6nt9XY0j6H9v7kvPJzW0rQvbZanVfGiUmBaISSFqD0HevbHtVhVbydbb36eD02cVl/fQv+0C/lalqaFdoo6EdGtIi543AX22frUjleQWhRf2Eh+t8X4l088lv6nwhTFDKnw1cmMbkEXbkevFL4kbwCeB0211HqecxWxL29lG7RJ7FDavbXnNcJ5e0ue1vqL83QWnayOIId5xGM8EEDbE+8PJ0Pz3L9pk1jjmGKC48hTA83Y5I0ewLdx+vDMHQe57tuwFSIeynlC+efsBUkWEesVj+HdCXNaIwcH4H0X0/Hvh4lfzCwzE4rs2iyGF7pnvT/ZhiKXKGw4C9iI7/JgTD6u6inSX9lnj//tf2BhWv3Q9HEHObXdPjHdK2N/buOID8CgBJrwA+Q8iGtrf905rHtS2ktMHOtJtjbk0wKK5xeIytTBSayvB5goX+NYLJsz2wSo1z/VnSHkyFYmxDSAwfxswCx9uIefPjmb74XkAFo8TJFFspITf/XBHTJiMYSNrI9o97jikqZmc4Qn18s2oSA37XsDgEcIukFxJsuLnAHlSrNRozxQBsXwc8WxFOQHaNTChjWe5OSODWoaZBuO2jgKMkPTo9HrXNwRbM/Cz127YI/e5PdYqqXWBSICrHlwhX/c+mx1cDXwH2GfWJcxeXHQizvL+nx0cSTJ4uUVtDOyIcRiy450m6iYgp/D5xYTqCqYjGLnC3pMcyZUK2ISnqtGM8YPuenkJVVbTqsPA6ojCzD+G38BPCcPI8V6RVDAuaRQl7CW0TeQZGiWwAgJqygWHhJknL9dz0u8TTiQ7/c/o8N85C5jgxcBrNBBMMG31YNvvXYNlAFKE3k7Sp7ZMlnUo5ayY7309sv7Bq25AwJ38NtH23yv01ZsiHFKaxZcb+hwJXAbcS8dSvyc8HyhaTSYqzL8HOyTNR6njZzKM+iyJ73TJvusLz2D4nNV7+BOwl6TLgcwXnGFR+9WTbeQbb3pJqGU7XhaTnEIWh/wD2sP2jhi/RtpDSBq3mmGmt8p3c4z8QzI2q466XtGRqOB4l6XKiQFiGdwAHM1XUO5e4pizJlF9T9voHAQdJ2sP2p6vGU4DzaJ6au1+f5/ttyyPPPppH+HzWNSK/tIVkaSdifrAWMX++kOoY+EZMscTs+q3CuDm/PRtf1e/3LOApbRrSoy4MpYLvK5mZXjmfcslp7+s8j/hObwasMNRBtsCkQFSO+bZPV8TQkdgBXTMCHs10hsoDaVuXWIH6GtpR4EXEwm9Z4I/Aox1mgEdSzzV/mNidkJetKuk8YA2iYNI1/qLQh2eFqq2IieLI4Yih/H467yMII7uXEJ5AC20/tYNhzJqEvYS2iTzDQCYb2J6YeB5J3JS2J1JuusQCwpfqDKZfK7ryIHI637aaJB5maJVGM8EEI0Yblg1MzYceUCSM/ZV6c6Jp6ZaJZfCoRiOuj8skHQt8OT3envDsq41UHPl8yS6DRH0fQyRhrUMsBN9LFH3qjKvxedXOmy4rRtyR5Ga3AitWnOdhwGW2yxbfRZgr6TG2b0+v9RiGn3b0M+BmIs3tpZKmsQfq3CdbFlLaoNUcs6U0MPOz+oOk1xLvUeV3MzWt31TwdNH865uS5tm+T+ETti5wRNlcQS1ScyWtTrAcH6npASaVSbvusTdJ15I67EqIQgo0kCylxu4rJS0LzLV9T9G+OTRlin0R+C+mz9fz46sqTl9PSAHrjK1rPECMqze98k9EQbgQ6TqzNcGEXZWYrz1zNMNshkmBqBz/VKS/ZBfIlYFKCt2QcS7wQ0nZBePt1OiWDRknUVNDOyLcn26G90q6KZPtjKNgZ/vSdFN/PnGTuMiReNI1diH+JpJ0M1H179QoO9E2N2TKj+iJxARo5BiSVnuYaJvIMzAy2YCkV9vOe0zslDquXcqLnP6NC7sxxUg4h/Iu3UihPl5Y/baNGoPKISaYYERozLJJuD4Vhk4ipM53UiIxS0yl/wHmS7o999SyNI8Zr4udCPnbwenx2YQkuxA9nfW5BAty6YLdB/XIXNH20ZJ2sX2xpJ8BFxMei6VIi8iPEP4vb1N4ID3V9ndKDmvjTfc1Sf9BLLB+QhS0/1/ZAUlSe09WAKhzEklHO5JX9yNScrMF7KuplkY1xccZjMnaqpDSEm3nmI2kgQkHSVohHXcyUUT5QNWJ1E5K/nXgOZJWJRQIZxKyvbJGb5aa21sAKEvNfQHBIH8sM+Pgd+13QAkWMuWjU4hUID3Z9pENXx+Fl+qTiVAioJJ1tDNwlO3/S/uWMsVs/1f6vy3T7y7gijE2HwvhCE44X9K3bF9b5xhJmxJFoRcQPr/vBE603ZchOQ5MCkTlOIyIClxR0j4EZXHPjsfwfuA9TFXJf0AwBDrDgBORYWBpTZnK5X+GHD26QzyWSHM4TdIjJD2q6yJF6iJtQHQo5sSmWhGTA0PSYQRjaFXClPo8IinpYttd+zH9h6TvMh6fsDxaJfIMGfMlrZjdsBWmjPO7HIDHb8w8p+DncWCs/liSng6sZPvsnu0vB/5Yg9I9wQSjRCuWje0sNvgASZcCywOnlxxyJFGgOIRYrGa4a1QMQ9v30ry4kO+sP0jcS7Ye2qCmI2us3SPpSURiVF1m+peIzvg66fGtxKK+rEDU2JvOdpY8dnoqCM6rKV02cIGkbzJdUltUNFg3PX+CpCuJhhfAQbaHmpo7hHlJq0JKGwwwx2wkDUznypo6lwGrNxhmXko+j2hUXkKs3YrwL4ch82uAw2zvWyUldIvU3LRuOl7SNraPq3NMBk0PKZpLmKJXpSBmBdIdaLhGTLKorYnvTt7fp6xAtBJhP3ImcKjtVs1hhWXHh10d8vLr9K/p62/lHlPqftuGAdvXSnolMwMUPt5n928TTcw1PZXiN6tsECYFohKkG8ZNRNV8WWBr2xd2PIZ/ENS8L3Z53jxUkKjWocRsWaZfqPI/d/qF0hjTw/pgaeJivkQMTXV0vMPAX4gO6UV1u3QjxNh8wvJwy0SeIeMLwNU9HdC2WvtG0Owxh15Y8HNn0Ozxx/oM/aUHfya+L+OQxk4wQYbGLBsARQjAb23fa/snSeYsCkxVHSEWCwh5w6zFAJ31NrggFV0OI9hX9wPfrHns2ra3TvIcHD41VSzFgbzp0jz4H5U7BpYgPgv5glSte0Hq/lcyAIruc7nXGdX97gcOP6dFhRRVxJG3RQ+jLW3SgsQUKUNtaWCP9GoGKhgsM+SOklYi5uNlmJeKEq9lquFfq5lUtzjUc8xx6buSrRHOtF1V7Ml7ED0IfL5BAeZcSW+yXff7DPAGYFU38CW0vbPCIPztwJeTmuNQgsE0Y12QpFR7A08i5K3fIK737yMarFXna9t8/CBTIU9l2waGpM8SRcunE+vDTSn2DH4NIRO+TtL3CBbbrMKkQFQB2z8h6K1jgUIDfAzjZUgMYpg2MGyv0tW5amCc6WGLoEhy+xThMZPJHuvoeAeG7Y+O+hwNMBt8wjI8iqAu56+rleaMQ8SJwAVMdUAPsd2VD9JaFJtDd4ln5GQkK+R+ngMsdDeG3bPFH+txtmd4tNn+RaLXTzDB2NCSZQMxkX5u7vEDadswItzHihYLyVawnUleviLpfOCRdaUR9EjEFP5CVQWizrzpeosGNZC/Z+RRds/I7nMrEqzQ7Lr+MsICYlQFovNobpLcFj8kbAOylOD5RELYfcCWti8pOK6JNLAs1rxx1LftP6U1Uxm+QDBlzrF9uaTVmPodSyFpbUKWVttfSTPN+A9QhRm/p0KKHp4e31u0bx9sA+wq6e+EHK7O3Of3TLEKa8PhPXQkcDvxvn4E+ISkD9r+es/uRxOMvh8Ab2GKzfn8OvPUJG39KLnrI/CpoqKWpGcDGxAKoHxBdz7R4B8FXkMwEq+w/W5JH2eKITsNDoP6H6VC/VbAgcAT0jFftT1OqwZgUiDqi0SHLEsFWr/D4RzGmBkSvRIzNTNMW9wwzvSwPHYB5PEly80WzAafMFKB6p3AddSn6A7z/HMIid/TKDZnHBls753+H8Q4dRhoQk8fCTx7/LHK5IVLljw3wQQjh9oZFwM8LDFKsv0f6CnK/1uizUKy5/ilmP4+lrIB0oLrCemYf0l6Wk0G8gWJObC0IsX1g0S3vAydedOpuUfS9QTbtjY8FXv+A+CZtn+bHq/KFCNuaFALk+Qh4DtEMu130hg2BTYipDEHEYvvGWgiDXSfiO8m6Fn4Z75d/Yp9+XMeyXQJ1s3UVwB8ieb+So3N+FPR6iSSWbGkq4CtbN9UY4xtCuW7At+TdBbTC7iFhc7Ewno3UZC6PI3vgvQdOJ/wespjNdtPT8ceQ/ydVm5Q/Poica3KPBzfSciGtyvYf2XivXg40xuXdzG6dOH7bD8oaaHCSP4Pkp5QdkCaIx4MHJwIINsRn49SQ/4u8G9/Ux0RsjjO1wBPJSqfEHSwrqt6s4khkaGWYdpiirGlh/Xg1klxCJgdPmEAmxMxuXeN4dzYXijpFo05uUvSp4E9naJIUwfsSNtVkalDgZNh9yzBEpI+wcwFcFfS3P+VtK7tq/IbJa1L9+l2E0zQizbGxQD/kLRatlhSmKt24r/XFA2LNq1S3SS9gVg8rZQ2VaYUStqRaDq2YSDvSZh+300kI36XqQZmXzSVhyjMdg+3/a4mxyU09Ui6f4D7xn9mxSEAR4x3Y3ampBcA2znMsvuhjUnyoNjQuUAFh+fmx5K0qCrSfJHhcXpcKReTNJ+Qiua9Wy6oGGN+4f8goWyoY24toviS9zA9oeo4Wvgr0c6M/wiiiHVserxN2vaKqgHa/l2SHa5u+8qq/RN2J64f6zC9wVmGq4DjgBfbXrQGSt+BY/vsf39unwcUgUNNmFHPsb129kDSRcA1RTvbPg04TdIrbXdFaLg7FagvIvyn/gT8ve7B6e91paSmJuYjwaRA1AcOR/LMuOu5ucXO94k/fJcYO0NCLQ3TFlOMPT0sYW9JRxEMlXzFvxPGymyBZ4FPWMKfxlUcymEBkcLyQ6abc3aZ8rAS0WHeguiAnMLMTtJDBd8iGGVnM54F7CeICdLHgUvTtvUJmva7xzCeCSbIo7FxccLHiCCAvNdaZSEhLcp2YIotcBbw5Wx+N0y0KdrQPtXt84Tn2SW2684NdwXWalMYSeytT6V/tZDm0v1eq++9yWG2u3a/52qgqUfSIA3X2yR9FDgqPd6OCKuohMKaYGui8byQkkQ9tzBJHgLmSnq+7YsAJD2PKSlh4edMLQyPJb2FKISuQMjyVycW/6XSuTaMZYU9w7uJ7+ZlwIsIxkudAlH2+1T6K+XQxoz/0baPyT0+VtLONcaX+Todkca6SpJa7W27bJ3yLOApDa+Fq7oglCZjlPfur+mejNMe12iczZH08FxRaVnqeUetIOmRtu9Kc6H1iSZmYfLlANiSeN8/RDArlyeax41Q9L52jUmBqByPIirMWQVwaUYXKVmE2cCQGMQwbbFBmmSsTlBrO08P68Fr07+nMAZJ02xA6jJeZns9xugTlnCxpJOJrvi4Cna/pMCotSvY3lbS24EriY7R1rbPHeeYxogVbO8wrpPbPkPSO4mC0EFp8xXADrbPGNe4JpggoZVxse3vS3oJU4Wez9q+scah+xL+EFl3e2vifj6KAnqbok2rVDfgjmwR3wC3NS0OaTBj5jxTYB5hGF71u/1Y0iHEwj3f8KiSwTXySLL93KLnauAdhDwk8286J23rizRneR3xt30ucCqwnO1ajPysOCRpecJr8Cb38ZkbEnYk/IT+Rsx1lwHeqjCFP7DkuMaGx8AeRJHiDNvrSnoFU8nNM5DexzcSn/1zUtHnFURR6hMOY/oi7EAUCX5qe2OF6X2RR1IvTlF9f6UMbcz4/yVJdvjQJNVC3XXGxwhm1Y8AHD5LT6445npCinVPxX7TrgOabrVBOl/RdWCXnsc/6LtXMU4k5tmnpMdvoV5Rby/bX5O0PrAxMRf6IvD8huevhO0/5x5+ctiv3zUmBaJyfI34QGYO628mOuKdYTYwJDyYYdpigyTv+2Ra9F835uG8AVjFdm364uKG1GW8R9I8jz9NLaM675Tb1mnBrimNfxSQtBwxAb6WSKt4NmHa+VDEtZIeP04paKJWP1T94iaYhcixSVobF9u+nljUNMHGwHq2H0zj+DpRMB1FgahN0aZVqhvwbUnvJear+fdxxgJdU8lUZ6W/wyk9x5QVXw4h3q9fMLNzX8o86L03JSlyVcrSFun/1/Scp0oG18YjqRXStb2wkNEHfwJuINKeNrf99zS/L4WkE4nG7DUKX59rCHnZipL2tH1U+Ss0h+0LU2FBU5ucsa3KEpfaGB4/aPt2JS8x22dJKpNtHUYEY8xL798yRCP7JQR7ZouSY+9zmCvPlTTHEU1eZWydNYjPtv0XavgrpWMeBnyhhVRyD+BCSVcT37W1CQlqLdi+rad4U8VIuQu4oua1uNV1wD1etk1h+3OSrmGqMfBh26fXODTzq3sFcJTtkyR9qOyAplBB0neGGuyoWYlJgag2fy3NAAAgAElEQVQEtveSdAmQGantZbtp1XMgSNrZ9kHkGBKSjrDdmURAgxmmLW64WtL6ti+t3nWkuIn6sa+LM0xMCL/J9C5jV7Hq2fkGMlscBJo9EfMQE4fjgU8T9NpjJP3QdiMT0GFA7TwNhnHebLIwH/iFpJ8yfdL1bzlZmGCCISFrMDUyLpb0FdtvV0GIiKvDQzKZV4aF1Iy3boHaRRsYaCEJU1KvQ5n6nYrkbL3z17z8oar4sh3BulqLuMaf5PZ+d/cQDYRC2G6btNjYI2kQSDP9bGwXMRvOJRa4GxNSqvNrnmY9R/gBRKHgOtuvVBjgfp8piduwsTrw4vTzP4Bf1zimseExcH+SU94gaSfCOPoRJfu/iIgSXxb4IyHJekCRqFXFqPqbwrbjGuBzkm6hXPqZjf9fqVC3dnr8Dyrm4G2lkg7f2aczZQR+ie3/q3n43QoD6cyWZEPgzopjfk29vy0M9zpQG5K2sn0icHqfbWVYmCSMWxDNSxh+itn3q3fpD0mnE0W3HzSU+I0ckwJRAXrkK63/+EPA6yT93va307gOBB7Z8RhaG6YthngW4X9wA9MLEl0m20F0oX4s6TvUvwkvjliCkFXlvSzGcpEdV0GC8oj5rt+Ld+Z+578Cb6irnR8m2noaDAn5+8VJHZxvggn+bTAA0/EL6f+23d8ziFjh49LjrdO2UaBJ0WYgzx3bVRHz+X3bFl2wfRxwnMKEeWvgIkm/AD5ZJXPSdA+iucQ8qpKFLellwJq2D5H0GGD5xB4rG2djj6S2UEM/G9tvkbQCkd53YPp5vqRVnTO77oM8Q/qFhO0Etm+VNJJ7fJKKf5YpFvRHJH3YdqFXUkIbw+O9iHXNhwmT8flAmaTx/rSYvldhdvwA1A7xeR9RINiVaGStRsSM18GNklaxfXPN/aG9VLLt33V3Ql62qqTzgDWYKoz0Rb9rsqQXFex7HC2vAwPig4TMrGpbL95PfK6Ochhor8GQWe0DsqOOJOR3B0s6ghjnX4YzssEwKRAVYBbJVzYDzpb0Z4JquwrNKK3DQGvDtMUQ/z3uASQsDfwGeEZu26yqPo8aieJ7CHCDx2wQPc6ChGdPxDyOmNONmaIBn5kYkF2jkafBMDEolXqCCRZnKGLH5UiZyZpe89PTB9u+ut9xtq9IjbutWjJt/ofwH3ljevxtotE1VCQpyro5xkddtF1IZuetnZgm6eu9TMZ+2/ohLbIOJMyYP05IWKsWhnlbggeBwwn/nUJI2p0wIF+JuM8vBRxDFEj67T+IR1JbNPazSWyLQ4BDFImS2xH+UzfYfl7RcZIeTzRdNiQSzTLM63vA4PgQ8Czbt6XzP44oqFYViBobHtv+cfpxAfUi55eWtCZReM3/DBXvh+3ML+peIiodSR+jJBErh+WAn0v6CdO/o2Xfm8ZSSUlvJAoHVxC/1zGSdrBdlMS3CLYvlfRSwmNnDnCR7SoGUXbelQgCwLbp2DVKztPmOtAYCpPtDQg5Zf47Pp8aTCDbFwOvzz2+gelWEEODpBWJ7/bL0qazgJ1t/2/J+E4FTk33xfcBv5R0JnCQR2OkXRuTAlE5xi5fsX1nulj8OI1nU3dvijyIYdpiBaeEu3FjNhQDxolUkDmWoJEvLemNuUnGODC2gkQGhX5/B2CjtOkcIqXnwQ7HsBvRVTo5bTpA0vG2S6OaR4CmngZDR4EufQFwMXCc65vXDmMsTWNvJ5hgVPg4U2xkiCLAwYRJ6u6U+IcMwrQhYrsPJ4oTAEjaiJhbDQ2JxfAVkhSlCpKOdsSbt/LcSfPDg4HHU4OtlLB6n22liXJJArQxsXhci0in3KCC+QK0Zo1tSXjYXZpe49Z0HStCPwZthlE1z1r52WSwfRWwkyLW+vUlu34GuJrw9vlJVjSU9FzC82ckyIpD2c/qY0rcB00Mj1sViwlpWd7fMf9zm7/1tkwvuhXhRKoZK9PQkrX3KeD5GVsusV6+C1QWiNI5F5BMqquQ5kibEsbpGxB1gY1tX1Kwf+vrQO41nkIwA09TmJ4vZfuOgt1XJq4DD2f6d/wuophVda4liCLsOkxn929Xd7wNcAShaNiVuA6/K217Y9lBCdnn9gGCMXiCpNNtjy3yflIgKsfY5CuaqbOfR2i2L5LUtaQpb5gGobeubZi2OEFxh9wLeDLTu3VdS8yaat8XN+xJ3ECvTt2SvRnyRL8hxl6QIIwb/5MpA8m3E5+P93Q4hrcDz3MybpR0MPBTgl3VJZp6GowCtxETm6xY9haiQPRmYrLSCRtR7WJvJ5hgVFjDdn7x8jfbhwJIqiPJbcu02Y+ZjM5+24aBJlKUdWEg+de+1ExMk/QuoonwFEl5H8X5VPtB3Ur4vRxHFPkWAssoGV+Xvf9tOuvA323/o6coUWYEO46mWSs/m14kidTXS57/hqQLgccxnenye2IROgr8JjFrMpbduwjvyyo0MTxuVSy2vUqNcTRBqReZpOOB84BzG8rL8q+xEsHqud72Lyp2vy8vpbR9g6TSQBpJtxc8NQdYaPsxfY45kCjE/pz4Xm8G/KqoOJTQ+jqQzrk18BGC/XMaUQA6lALmWCoenibplY7QjaY4glirvZSQL74VGJX1w5Ntb5Z7vHdu3dwXkjYjEgMfR1wjn2b7nrSWuJEoNo0FkwJRCcbM0hiqy/ogcBimrUVQaaGZYdrihlMIv5djGSOLSg2174sh/pV1l2yfK2n/MY9nNhQkXkzcXP4Fi1J6uo69n+Ncqoftu9P70jX6eRq8t+MxPJNgLdwPoDDQPJtgeJVOGoaMNrG3E0wwKvTOO9+a+3mFGsc3YtpIWh14CvDIVCzNMJ9gIowCbaQobdEkMe1Mwr/wEGC33Pa7qJaH/ANYkZibZh3yDFVMpzad9VskvZAwmZ1LNCoL72c9f9sZcKTPDhv9/GxG0jxNbJ7beraNMiHzPUSx5ufE3/dsorhYhSaGx4MWi4eFqsb/z4nP6gGS7iSKRecRBaO+DC5JbwaOJgyidyYaeDcDa0jay/aX+hyTXY9Ok7RnOn4OwdapYg/dA9xBrE1+RL31ybsJRvNnbJ+bxlD1XgxyHYDw23k2cCFENF6SL5bC9pkKT7Le5nyVomd928+Q9HPbn5F0GCNKNQTmSnqM7dsBFL5pVR5x2wKfsz3ND8/2g2ktMTZMCkQl0BhTgWaLlClD+sCP06x7tmCu7U+PexC00L4vZliqV3eef1yjmzxsNDVZHAX+QnhTZZ2mJYGyDu0ocJmkY4Evp8fbA5d3PIa+ngape35Oh8N4LNPjfv8BrOhIW6mKnR0q+sgDOj3/BBPksKSk5bJCsu3rYJEMso6nRFOmzQsIKcJjmVkUGVV3tokU5RkF3f/Czj9MW0zWTkyz/Tvgd4Q0BEmPrmDx5I9dpc5+BWjcWSd8Qk4gxvo3YkH5tpL9dyt5biHTZUhDgfv42SwuSHP+aQweSa8jpE5l+JrtaQWiJCXrh0GLxbVRIPmG+J49quxY2/sD+6dC5bqED9SbCKPxv9ru13DZk5ijL08U155l+9cKL6kziXliL+5herriJ3LPLSQYO0VjXE2RWLYNUYT5DiFlL2sSPp54zz+v8PQ8gYq6wBDYWw8khkx+W6UNgiJc4NnAldQ3P4ep+fA/JS1re0Eq3IwC+wFXScrSIl9NMOEKYfu/Sp773hDH1hiTAlE58nrHeQRF7RKiEjxSlFzMgG6ikttQFh8CuFjS2h6tW38dDKR9XwzQq0En97hOF2Oo6FeQ6Aq5QvYvic/n19LjzQl2WZfYCfgo0XmEmBh9onj3TvFRpgpXXeB84AcKPxKIxc2FSXPfZYGmTeztBBOMCqcQQRfbOYULpOLQl4kiRyXUIN3KYRp/vKRtHAk8I4ebGdVfTywkmqJ3MXlo7rlSDyJJ6xNM6LnAE5PsdAfbdRgibdC4s54YM69MhbC5tks9bWy/dGijrQmFN8yxwMq2V5W0HvA62/s0eI2XzLaGsKTNCUuL79m+XtImBENqGaoLRCcxU7bZbxsMWCxuiLIGd63mt8Nf7G9E0eHvxH30NwW7/yv3+/wuK5rZ/qOkvgURN0gkLDj+POA8SQ8ninvnStqniNTgMK8+DDhM4e22HdFsvQD4qu2hm/gDf0lrlWw+shUhW6vC84GnO1IKm+AORVrg6USK5f8RQTJDh+0TJF1B1AogjKZLWfyK9OMPM9MjaaPCgzrCpEBUgl6JWdKQHlqw+7AxG9g6bSiLiyU05Qm1JLCtJDO9W9e1B9FQtO//rhiBBr0ViliGGbpgGzK9kH0VIaeA+Gws2cH5F8H2vVR0TMaIrqVuOxI0/cys/Azg8DTBeW6H42gcezvBBCPEJwn/ij9IuiFtW4Og/Rd2yDOoYbpVhq6KQ1Dc4Cto7N2fmD2NMOBi8kDgVaREqiQ7HWX6YuPOeq9kLDEOFgDXOkx4i46bQyx017C9u6RVgMc3kOE1wf9n77zDZKmq9f0O6RxQPAaCwNUfCPKJBAURI1lFUa5eFBQBQVDwqsARVBQxowQFBBEzUUDAhIIoQZLkHBQ/UEQyegHJUeb3x9rNqenTebqruvvs93nm6a7q3l1rZrqra6+91vd9l3g/75O2rwKOBr7UxWscSAlup52i0A18G+GgtZ1CS+gDRIV604SBQmdqCeqquYlq6mc1GTbtZHGndJm0nUKq0FuPaBu/kdCw2R+43M1NQIqf/3on7I4NKiTNIK4htrO9YQfPfxlRRfQuYtG0I0v3tOg9W2E08i6i9WkQCaLZRMJQkm4mqgM70UO8tZuDSHpWuh59u8Pc4HNEtdRzGZAUh6TFCVflP6XthTqo0jwM+DNx3f554txVqXtZjZwg6gLbd5ZVpTGdk1kfY+ilZHFcGRpNqERpve+ZlhxCnMyvZe4ERCmC9hVrpQGgsJ/eFLjP9hkKjaw3E8KnX211QV8ipfw/aqRE0LfTT2V4Gra3mUy/SROqrRTaQKun3Vfa/muHL9Gtu1UVFBf4ZhITvGZtz0802d8xkl7AnKTzhW7uCFRjIdt/rmvzmHYcLTiJ+J7seGWdmCy9mjnaSKum+8tI+pDtZouoBxDthGsQSagHgW8xR0Ozn8xyaHTuDc9UmHT7d6xCo68VbwFWT21ASxBC2Ks1q9ArsCUxT1iaqdXd9xNC6o2YVrK4RA4mqrG/DPzB9t0djJHmCMEX708wZxGv1eBXEy36mxHnuiPaPP9/iUTeo4RRyVcatZm2I123nJh++k6qSHsN8TeYiF0dOXPfAJwp6VdMXZxvtgh7LuEsfASwtUObsysHuh44mTjH1c4BCwK/ofWC4Aq23y3pnbaPk/QLOkzqDZqcIGpBXXXAfMSXVbO2q0HFUKZF31x0W7I4xnzAYUU7FIxz7/uIsR1h674K8aV8rO37qghE0l+IhNWRLghFl8ShxN9gpqSbiFL0k4F1iVWoptbV/URSswvRCebY5w46hl1sHyTpGzSuImjk5jJQUt/9maTv/NSL3/XFYybTL1JCqNOkUJGu3K2qoH6BT6HL1tCBx/a0qgklbURMfGqaPodJ2sr26S2GPZ5aXWttHi9n7iqHvpAqei60/XK6M034K/Bx25en11kD2BXYinCGbJYgWp9IPF4BYPseSTObPHe6/CdVctf+jsvQRXVIYiAT8WnwSK2dz+HMekMHySFsHwQcJGkPd6jT2YdkcVm8gKhQXBfYJc2HzieEqs+xfWeDMV23jaYqrK2J68oFiUqXh22/rYPh3yEqyG8jBPzfXjxHliFL0gUziI6UBUiGzG6vGzqTaOdbtbCv1Xl/EUmvAl5VV9EGDEyndEbxuirJgLQ799TkBp5Q6EDdByw+gNi6JieIWlNs3XiKWAH6RMkxlGnR15BeSxbHjNXbP2XwtJgAA9VMPudlUtvCEZKWIxJFF0i6FtirAp2qLYi2pi9I+jnwnUIicdCsDaxMaEPdASzuEGP+Ae3dcfrJwy0eO6CkGGoTrZaaGWUhaVNiBXSptGuCNholmcwQ05W7VQ1JL6+fFEh6c5tESr+YJOycB8HXgHUKeicrEW1OrX6vrxEJq6UV4q9vJRIvfcf2pKRbJT2vy8WTV9SSQ+l1rpC0qu3r1doZ87F0TADSe2RQVTqHAr8EFpP0JaKC43PdvECnyZQSWbxucfy5xe0OFod/Jmmm7cdS8nJ14Put/vfTSBaXQmp/+236ISVX301UFK1Ag+/SHnWlbifE2HestURK6nQBuPJK8k6Q9DGiJfNe5iRT2+qG9lApfzBxHlyeufVKB6ZTWmwpU2cuZjekxNCxhMbxv8ktZsPPMLRuUK5F3xT6VbKY6SutJsCZirD9d0kHEja0XyEuvktNENm+EviQpOcSK1CnSvo78C3bvxjw4R+3PQk8LOkm20+kmHopue8Z218u61gtYvj+sMSS2A/YHLgolVlnMqNMt+5WNY6V9FaH+DGS1iEqLtV6WPfUaRDNB6xGCPYPggVrySEIod9U1dIU26cqdBQ3IpInezFYx8v7CQ2i31JInLdZ0HpE0ha2jwOQtAVzHIlaVQ5cK2lLYCLpD32WZKndbxyitDcRGiqLANvYHsixSuQMpi6On1nY7qRS7wTg1WnR7PvEtdCRDJHuXZIKWcn2SSnZs1C7tsykL7Ne4Wc5ovXr2D6G9i0iUbu3pMOAn3U6cBhkSTpkN2AV96C7psj6voKp3TQN9YRsfxf4rqSf2i6lgp1ISp0vqRbTB4C9Ww2wXUvMH5DaEJ9LCGpXTk4QNUDDITxbo0yLvnpGqWRx0PRkRdtvhmjSmeGZ8vmNiNWbVYiLo9fY/nuFYb2GuIB5hPii+Yik99p+7wCPOaNQxlu8D4Uv80Ej6VXFVecGj88AlnOdDW+fY2j1/TGZLlzK5F4PRqA1kykdd+luVeATwEmSNiCqHX8MNLUYnibF9qengG/YvnhAx/qXCg5tkrahg2SP7ZsoWG1LuoVwrhoEf6K79jKI79SjU3veZBq/TWrvaWVpvytRLboUcDHhurVb1xF3zl+Av9j+vwEeozT6sDD+dGoBfTtwqO39JF3VdlRJpM/HZwkNz5OIyr7v0MJ9VtKfgWWBy4gOio8TWn59dSO1vbukzxLtadsRAuYLSFrf9rh0btzVY3JoZ2BH4nN9KVG1fg7tBafnMpaS9IFmiaXpYPuwlDCutRd+uNNKsrSwuxjwdzcXPi+VnCBqTOXCswVKs+hrwDBUUA0LvVrR9hXNEQO+1/aZGk4x4HmJ24iWqiOIyqFJYOGk6TCoPueGSPok8QX6N0IY+bepqufrkgZdvr0IU8t4i/fLPGd+Nk0gjiUmB3cTCSoRbRQbE5OFgSWImLr6WmOSSNr9PwqTskGSJtAAv0zVoMczVdwxV4NmRhJJyxOtAwvUFq1s17cRTMH2WZIOIhz9Xgj8j20PKMT/2J4iiJp0gZqKpDZpH7+f0O/5Q4tj7QgcI+l7xHnmKnprFxuYWHIvC1upKmpNSYum7aKuXtP2ufS8D6efgSFpNiGCvXja/iewT9LiaTbmxbZvGWRcQ8BMSUsSVVW1drthEuKeTYjcnwehkCzphW3G7AScb3sgOl1FUpXvycDJqWrpA8DBqUXzvwZ9/EFRuyYGTk/nup8y9Xqk3bXyDoTQ/Pm2N5K0CuGs145DJG1eO9dL2pxYLBiIk5ntsxWu1zVX34ZI+gmxcHB1ajG7GniAaFf9nO0fDSK+bsgJosZULjwrablUhVC06NuSEFodyBu7nhEqWSyDx3vJeg+AoRADzjzDk0TW/5NE4qF4ITSwPucmLAts0qQ6ZpDVQ9hedpCv3ym23yNpLeJi4ovAfxGVVNcSWhFre8AC3vUrsCmerxNuOmWW2T9EvAdr78nvFLazBlFmJEmTi22IRZGa+80kc+tM1J5fX9G3CKHjuI6kdQZUEb4rczvmNNpXZEliVfxXafudxEr55pJOsP21RoNs/w14bWqVoYuKqnr6nshX6FfK9klp+0DmmAUcbLtlZUkvicBmFZz9/D9L2gr4CPE+vJg4p64FHCjp/2wf02TotamN5DDgF/2uQBkSvkV8Ns+0fZmklxCJzmHhCYdDW3Ffy4oN22f2erDUznZLN5pMheP+C9gf2D9dR4wyp9Rtb1a438m18mMO0ef5JE3Yvk6duYpvBZwg6S3EZ/QLwIYdR90F6Xx1DNEGh6Qrga1StWY9a9i+Ot3fGrje9lsk/Rcxp8sJomHEwyE8e4Wk+wkHsbOBs2wfXdKxM3NTmo5KG4ZFDDjD8CRGAGx/vMVjQyF6Vwa2LyFZYFdJmhx9nXDd+CJwXKroKgXb7cQRM5lR5H+INtFOK+DqK/quJZKjr6bPSRFJaxItvovVJSpmES0trVgaeFVt4ijpK8DPCfekiwlh6eKxXj7XK8R+iJX52+sTEM3GJAYxH/gKcHhhe2NCp+NZRPVN0wUthXX8h4Dr6SARWKD4/55JGLxcRCyu9YsdgPcWJngAv5f0PqIDoVmCaGliYvwRorLhp8Bh4/T9bPsHwA8Ku/5Bi/atCrgnJRZqznNbEZXgg6IvmkzpuqYjetFYGjS2l5vmSzyi0Fa7GthX0q10sMhl+1pJuxJVh/MDb7F99zRjacb3iPd+7Zy3LfE/f3OD5xar0d5ILGBi+zZJQ+HKmRNELXC1wrMvIDLN6xFq+QdKuo+wVczJopLxNK1o+8jjHgIx4MzwIOlftJjolKWPlQnSCtCXiYuCvYHNq+wpl/Qt27Pb7ctkRoRb6GLBpg+aKt2wDNG+8iymJioeICYLLccWqwps/1vSUrYflNSo0qR+Rb7IAsCzJe1m+7AOxwyifealtk8tbD9i+zsAktq58W4GLO9wkOqYBhWcS9FAh2SavLAuOVQ79jWpvapZbA8T7ehHKKzdtwV+IenfwI9tH9znOEtD0htsny+pmRRDu8ReWcwmWtAl6WaiwniTAR6vVE2mXjSWyiRVQ27ebl8DPkr8TrsRC28vISpvmh2nvmV3knAi30XSoByfF6873x4uaZcWMS5N2NqvRywi1ihNt7MVOUHUAA2B8GzqQ708/eyftGe2JHp6tyHs+zLzHjM0BGLAmaFizXS7PZFY/gHxntiesBLNlMuNwJ2EDtQEsIOmivuXaXIAsE6DfeuWHEMm0y92A34j6XSmali0/FxJ2h34YW0lXdILgO1sf6NfgaVWqpMkvcX2aV0O/3OqBK6tPm8DXK8Q1v9P/ZPbrcinxMiZRDtTR2MGQP0c4/2F+89rM/bObpNDjbB9Z4etKN3Qqk25I6dZh7X7npKOJwTTDySqqyol6ae0WnBq1uq0LXA+jQXEO6n8KgXbN0h6DbAi8f1s23N9vlqhgpV5B5StydSLxlKZrNBg30rtBtm+Lt19mKgsbEf953DQTr4AT0tSQe9oRRqcuxN7E3pxTwB/rGkwSXotsQhSOTlB1JihEJ5Nk//1iBLZVwJ/TTF1pIqeGUuGRQw4MyTUtLEkbWx7zcJDO6WLvS82HpkZEMcTn8VG7RylfUYlbUbY2y8r6YTCQ7OIVdNMZhT5DOFk80qmth61Ywvb+9Y2bN8j6f1A3xJEBf4o6avAS2xvmdpNX2b7Vy3GbEfoYxySts8Gdid+x7d1G0BKjJTtmFjPgpIWrem+OYSnkfQc2rfcXSjpOOBEpiYCu9Egmo+o5GrkQDsdlmimdURoErZE0ixgC2IR+kWENtUH+hfetPhkun078DIieQURa1NRd9sfljQfMLtRddWQUUu6LkByT+9kXpcSSycQ76sXpZbSHWzv0GJYT5pMKYFd61y4yPY97cYkutZYKgNJHyZaM1dMOlw1ZtHifVUYvwQhFr48hdxFs8ojV+P4vAdwXqoQmwBWo0mVk+0TJZ1HGCYUPy+3MGCB/U7JCaLGVC48K+lu4CbiZLQvcGWqKiodSasRfZSvIE6sANjOIqclM0yaN5mhY5akxZzsdiUtxhxB0HkeSWvbPm/Qx7G97aCP0SE3EC0lazG1teQBorIgkxlFXgWs2IOeV6NV+0FdA3+XqCJ8Zdq+DTiOOQLUc5GqZT7Z5OFOqxXqX/PbvYzrIz8l2iy2q1UDpeTQD4lEeitqLXo7FfZ1q0H0FNFW8omOI+6MM2jsVgktzq1JKPeDhBbTHwhdqVO6rWAZJE623KlF57W1z5mkk4EL2ox9WtLRxMR4KJH0MWAforq6NqfqdF53AJGsPQYgJXxamvn0osmkELP+CXAlKdGgcEFs6t5XoGyNpU45jaiuPoSpVWYP0Jl0y88JPbIzaF6V05D0uXslhS4L21/p5jU6wfbvFO5qtSq7i2rX402efxchYVPcd0e/4+qVnCBqwJBMwo8mWgO2I0ohz5Z0TnpDlc13gT2Jk+NbgY/RusQ2k8mUz7eAqyTVkgEbE73a8yypzWJb4qJ8AnhppQGVSFrFvVrSr6sWqMxk+sgNhMZPt25dNyax0gOJc8EniKrsQbCa7W3SRI+0ot9SNF7STEJGoH6FfBBaGWWxF1H1frukG9O+lxLaKC0naLbX7+WAZWhOTeMYBxIthLvY7ndVU795PjGhfjRtz0j72vFXScvavnlQgU2T3YBV3Jsr8UK2/1xXndNSD03SIoQm0Etsb0m8/19Gi2QxkThcp1BxtxIxJ+wkQVS2xlJHpL/3PwjZll54XptKrYZI2odI5q5MnHfeSSSZBoLtf0o6jXQOl7SIOzdUGCpygmhIsf1JAIUC/dqEZsQukhYFzrf9kRLDmWn7TEnz2b6T6Ju+lKhsymQyQ4Dt76SS1Zq+zCG2r60ypiqQtABxEbA94Si0ALCR7YsqDawicnIoM2Y8AFwu6fdMbT1ql0jZmViV/zqxun4BLUROp0m9e9hMoi2lFScSbVcX148fVZI4/1ZJkHn1tPvKpL/TlpRgq1VbnNaqgiLpdG4K3JuuV3cmjAIMfNV236zWJS3dbqVf0gvrF3Rtr9yvGErgeKLNr1bptTlREQ0NI5wAACAASURBVNaORYFrJP2RQhK3AxHisrirx+QQwONpTlarznk57cXdu64mBBasJYcgWjMVDl5t6YfG0pByXSefuwa8nTj3XG57R4U75A/7Hx5I2pTQEVsq7Zog3isj2W2TE0RDTlp5uow46T6HqOD5AGGTWRa1/tV7Jb2COMG17bPOZDLlYvsaynNaHDoUrpNbEH+DIwgHyD/Pq8mhTGYM+Uv66Yo0sdhA0rPSdkdiwj1yrqQ9CCOJ9YBdidXrVqxgu61Yaz2a260HQuPkQtt/6Pb1BkFKCHVVrSXpU4RQ93Fp1wGSjrT9zSZDDiWqE2ZKuglYGDiZWDD5PvC+XmJvwvGSriMqNS6x/XiK+cXENfq2wP5EW0zxd+pVALp0bO8p6WJCBxVgT9utXPBq/CT9DBU1DVng9PSZ+SlTE8ydaMt+jWiVWlrSEcT/eqs2Y7quJgT+JWlb20ek2LehTZtpqlQqUkuCzUgaSyNZxSLpROIzMwu4VtL5TP2/tUs8Pmb7KUmTkha0fbvCaXYQ7EckUi+qShKmn+QE0ZCSBEbXIwSqlwUuIUQLtwMuLDmc45Ng2t7AH4ls6BdKjiGTybRA0ouIL6hXMLXXeuCaaUPEjsT5cW/bZwFIyuLtmcyY0Eh8VNLanYxN4sAikgi112tnt94LnwM+TbTi7wf8mrh+asVNRUHnLliSqDKvVSS8E7gU2FxhH/21Ll9vWNgaeF3t7yHpYMIlq1mCaG2ijWQRwmRmcdtPKJzh+r1osg6wGWEA8XpJjxPfuXcDvwTe36TFqpnG1FBi+zeSLnLnjl3YbqnJUyH1ya3NCvc70iCyfaokEy7XE8BeHVTD9VJNuCNwjKTvpdiuon0i6qH03HrN3JGuYiGSvDWO7WH8gyl5dgFwpKQ7mdM22W/utd1Sp2uUmJiczNfOw0hqFTmLSApdYLtdGWMppDLHmT1cxGQymQGisH3+KXERuh3wv8DfKnJzqARJzyWslLcj9BKOIqysX1xBLCK02+o1RUpdJU6ClbfYfiytYq4OfN/2fWXGkcn0k3p9Mdst9cUkvZdILjwPuJ2wW77a9hoDDhWFVf2OtpvamEs6lhDg7qp1Lp33N699ntM58OfAu4CLbTdyUxx6JF1re9V2+wqPXWl79fr7afuKQf2fU0vzYsCj/WxjqxoVHLtsd+rY1bXb1LiTqpX+TSR4PkpUE15je88Oxj4boupooEGWTF3r6Om2T2vz/OcDywE31sTuuzjWksTff37ib/9c4GDbfbOSL1Rv7ZKOdTxTz+EjWb2VK4iGFNsdrYiVje0nCZe3TCYzXCxm+8eSZtu+MJWHXwjMMwki2/8mWg0OVbgvbkdUC5wLHGP7+yWG81NCV+RwunTd6DMnAK+WtBzRanEacCTw3xXGlMl0zTT1xfYgJWBsry7pzcB7+hzfwsDHgRcDP7N9jqSPAJ8nHHiaJogIrZy2ds8NWKaY7LX9b0lL2X4wVbaMKpdKOpw5eiHbA5e1eP6MJOY7UXcfChW1/SbpLFVhHjNounbsSvTsNlUGqapu83b7mox9PVER+BLi3DMBTNpeosWwRtWE+zR5/Y2b7AfAdjsHPyR9y/bsdvuqokHr6P6tWkdTYv9w4u83Q9Km3bTO2r47vc7itveaXvRNqa/e+g5jUL2VE0SZTCYzHtTcNB5KWgh3A4tXGE+lJD2m2emC5F1EpUGZCaL5bA+Di9zTtp+U9HbgUNv7Sbqq6qAymW7og77YU8lhZgEA26dL6rfRxo+AFxHtDPtIuoOw/P5wu8ndNCo9/5zaqA5P29sA16eqpaGboHfBTkRirZZUO4PWzmeLAMW/cfF+bpXonq4duxI9uU2VyAoN9nWq/fVj4KvARXT42UqL6l9LP+2o2b/PJJy3aiYjqxIyI20TRET7Yz3rNthXFd22jn4OeL3tqyStT7R1dpwgKlbCAR1XwnWD7XYtgyNJThBlMpnMeHBuKsU9FLic6H3/WbUhVU+6QDsx/ZTJhZJWS4mqKpmZyqw3IS62YKpOQSYzCkxXX+xxSROE3f1OwM3As/sc4xrAqkkUdVGismQ5d2BpntoUPk/BtQv4WgftCdsRmpCHpO2zgd2JCezbuv4NhoQkIv6Z4j5JhxG/b6PnL1tCWPMSvTh2Qe9uUwNF0oeBHYAVJV1SeGgWnVfuPWq7Kx0cNRaRb9g6anv9NOY4YLbti9P2WkT7UqvjbEYIJC8r6YTCQ7MIq/thYaIoUZIqHVtdjzxt+6r03LMk7d/l8XqthOsYSSsDS9k+o27/m4A7OhRAHzpygiiTyWTGANu11aejJZ1DuB7eWmFI8ySa41SzIPDBJGpZ7Ecv26nmW8QF8Jnp4uglhMtRJjNKLE3oi30jJcKPortr2D2Jc+LuhPX0LEITpJ88mlqOahOfGzpJDiW+Tfw+tVaQDxFJn4YJkRpJk6OZ+HHH4sIjwpvaP2U0kLQhsJLtQ1ICf5btG6qOq0BXjl2avtvUoDkNuJH4TH2qsP8BOhcx/62kt9k+tYvjFt0SZwLvoHWrJMAqteQQgO1LJDXU3ipwAyHEvRZTBbkfAM7sPNyB023r6EL17aLF7Q6SL71WwnXD3sT3Sz13E+2EI9nOnxNEQ07KrG4HrGh7d0nLAkuXqZQ+LGKrmUymM2oCfJJuIfQwMuUxbE41h9v+QWH7H0R7TiYzMkxXX6ygW3E/g0s0LFNXMbB0cbuN4PSrba9W25B0AXB1uwMqXJG2ZO7rs5bi1iPKUFQ+NtOKqdGunVDSZ4CNgaWIhMWCwGHAG/sV43TpwbFrum5TA8X2P4jvvlW6HSvpX8zRlNlD0oNEhXZbDaL61lFJX6d9ZffDkray/ZM0ZkvaVAHZvhq4WtIjtosVREjagC7asgbMTkTFY7F19Kstnl/fOkphuxP3uV4r4brhhY0qxW1fm7QfR5KcIBp+DiBsTNcgVr4eJFaEy0zODIvYaiaT6Y6huKCel7B9TtUx1HEEMYGssQjwG+C1lUSTyUyTbvTFJM0PbArcZ/sMSTsDbyaq6r7aZ+epQ9tst2JC0rNSaxXE57ST8/eJwELAxdRZao8hw6Il9KkWj03SXitmC2BNQlcG27dJek6fYusbtm8iqu064SfAjPqWyNQ6OervyzX7+FoP0X7R7oNEJXityuZa4AMdvv5nCM2dIt8k5pCVk85vu3fx/GWneciuKuF6ZFaLxxbs87FKIyeIhp/1CVviKwBs35NWjMpkWMRWM5lMdwzLBXWlSHoncKvtK0o8Zq3VrMj9hI7KfiVa194tad9UgbowsdJ7TEnHzmQGRof6YocSVQMzJd0E1D4D6xJJpff1MZ7pOEb+hNAt+2nafi/RRteOFWx3KrI79DQ5b0Iky1q5RZVGTStmGjyajAOK+4bqu1rSG4nWmVplWrtqmX2IpOuP6vZvAYhw8hpVvme7Jz2vuorC+QgnxetbjbF9PbBm0jGrtauuDfylxXFWAFYEnlNX4TaLSDZXjqRXExXWK6dd1wHftN2u5W46XEAkhDqthOuFf0la3faVxZ2SVgfu7fOxSiMniIafx2xP1r5IJM1H+VUBwyK2mslk6kgls83I5/hgU+BVkm63vVFJxzwTeClhKQ/h3nEHsAyxKrt1SXHsBpwgaVdi9ewU298u6diZTNWsTUxIFiE+f4vbfiI5fw3NNY3tfSVdzZz2t91t/66DoTdJWrQo/DriTKtFN53nfmz7fklHE25QO9s+rS/R8UxVTFM6EBa/NSVgJtM1/R7An/oVX584jDA1uJzOOgc2oHFlyBFEq+QoJ4heOI2xRQ2ip4DvAb/ocOyzgW0lbUskl17a4rlvALYlOk7qNZZ26/B4A0PS64jKuu8RFvcTRCfMaUnX6eJW43s85gRwoe2X03klXC98FThJ0ldIVYHE7/Z5wlxhJMmTh+Hn2tR/OpH0hz4LnFdyDK9hOMRWM5nM3JzS4rF+91qPJLa3AUjitmWxru3X1TYknUysZr0OGLirRd0k5iPAqYTD0SGSFulgEpPJjAOP254kdD1usv0EgO2nJfVbrLQnUhvcpbbXADpJChW5H7hM0u+Zen02khPyPrTobmv7AIUl9hKEVtXBRJtJv3iIOZo0NWrbk8D8bcbvRFSHrUJoy5zH1DbgYeA+2904f85v++n6nbb/I2mu/WUzTd2oiVR923BxvtV3abdVhZIWAN5JiDe/hpinb2T7olbjbB8JHClpW9tHdHPMkvg0sJ3tXxb2/VLSRcS89l39PmAqrrhV0vNs39fv1y8c5/eSPkQkhA5Kuy8HdrD9+0Edd9DkBNHwsyuhQ7QU0WP+a8rPBs9u/5RMJlMFtkdWBK9sbJdZ7ruYpJm2a5O2GcDz00XLoyUcvziJqd2uSazQdzKJyWTGgRkF15vifQhXocpJk+iH6s4XHQ+nc5vueYFatcv6hHj5BalKp2/Yntbr2b4LeEtK4s9XYrtxNxwr6SOEnk0x8dgsGbJwo4WHJBA8Y3BhdkytqmYmUVV2bdpelaj6aJUgWo34Pu04Iagm9vY1GiVwJR1ItORdQ1RevRv4c7vkUBq7nO2/A5c0qiofAqv1leuSQwDYPknSNwZ43PuBKyX9lvgf1o7b1wR6qlDsZxK6cnKCaMhJZcMfTj9VcfsAejYzmUymrxTcRupp6zYyAE4g2nNrgpHvAX6WLphvHvTBpzuJyWTGhHoXnOL9geu+SFq3w6oYA+dK+hlTJzItha6nqXs0jjwqaXdior12ajNZqOKYgOZVLDUJiXbuZyXzT8KK/Dtpu1111PFEBcv2th8AkDSL0PnqphJpINR0oyQdB8yutTRJWgvYpc3wq22v3uUhH27/lLnYkdAo3Nv2WSm+Ts9R3wbeQeOK8k7cvgZNq4rlttXMkpYgiiVebHud5GT5etvfazP0Twxf++ZIkBNEQ4qkdVo9bvvcsmIheiufBZxF6GqcafvOEo+fyWQyndBPt5FpYftzqXx6vbTrS7Z/k+5vWlYcklYB/l5zR0rn8mVt54umzNjTBxec6XIgnTkILUBMZIqC000nh5I2s32ipI82erxdYmmM2Rb4KKHhdJek5RkeUf7pVLGUzd7Ed9cVjVrHGvAVourldkk3pn0vJboevjSA+HpllaLeje1LJK3aZkzXieQeE7dLA+8HvpHa4Y+iw3m67Xek27kqyiU9t4dY+s1CddWbUx7rYPwPiTb52vnuL4Swf8sEUU6g905OEA0v+xfur8RU1ftJSrS5t72ypBcS4okbAvtIeiAJf2UymcxQYPsfSc/jpNoFU8Xx/IawlK+SI5lqaf9k2jc0ybRMZoxpaSoi6YPA2bY/2OXrrkJUZry6wWND5YhVJrZvoCCLYPtvRLKjcqZZxVI2d3TjLmX7KWCr5KZVq7a5cgi7Dx6WtJXtnwAkjdd2FSw3d3uQdB2yKXCv7TMl7Qy8magU/Krt++vH2P434bp4aKqQ2Y5wXzyXaJf8frdxJK4BXtzj2H5RX8lZpJPz1TK2vydpR4BkNNA0cVn4+99n+4xO/v6ZqeQE0ZBi+5kvfUlXFrfLJpXovhj4f+nnHuCPVcWTyWQyzUh6Hi+QNF+HK58DoZldcwXi/vM77MBrx38iCWFmMpnB0669ZlPgAEn3EyLyZxEJo3+0GmT7i+m228TSSCLpeiK5vY/tYxs8vq/t3SWdSOPz7uYlhNkpvVSxlM2ZkvYlWseKGkQttWxSQmjYkkJFPggcLelHxPvkWmCbVgNs91LxeyiRxJ0p6SZgYeBkYF2i7e59bY55DTBb0qcIAecPpnG9ULbz9Vz0oZLzqeJGqopq9XtN6+/fK5KeA6xg+4pBvH6Z5IvE0aDq1aD7gOuAHwBb27694ngymUymFRcBv5B0LFP1PMos4S/aNc8kNDHuKPH4NZ6U9BLbNwGklotObIszmcw0sf31No9vkkSU1yAmL+8BDpT0b+As29u3Gp+Ejj8LvMT2lpJeBrzM9q/68xsMDRsSLThrN3m8tmh58qADadbWV6OD9r5eqljKZqt0W0ysDYOWzbSwfT2wpqRF0/aDAzrU2sDKROXMHcDiaXHmB0RFT0ekxZ0TmZ6OU9VzyH7wC0nfBxaVtC3RanZYi+f35e/fDUlj7PvE9dWyktYEvmh7k0Ecb9DkBFGmE74ObEDYFL5R0unEhcv/VRtWJpPJNOSV6fZ/C/smKVHjoV6YVtJpVFN5+WXgfEmnECtub6Na04NMJlMgVTpeRtjVn0okQz5OTM5bJoiA7wJ3MuecdxtwHDCyCaLUHvKFWpUUgO07iIlew7Yn279J415SHDcgahX9ixFJvTPT9oZEBVi7BFHXVSxlM87uqGmRZHlggQEKhD9ue5JIBt5k+4l0nKclPdHnY9HIuazAyM/1be+XEqnPBTYGDq4lWJtQ6t8/8WXi3HBqOtZl6b02koz8m2Zcqfuwz6wX9yrTstD2fsB+khYiyvL2A15Efv9kMpkhpKb1MGQ8B3hh2Qe1fbKkdQkNOQiHlGFuAchk5hnStd16hC37K4EbgHMJseVLO3iJ1WxvI2kjANsP9dvWvWxSm/DbgK4SPb2O65ZaW19Kur8i2YsjaTng4A7Gl1XFMm2Se9TM2rbtWyoMZ9pI2hv4EKHrWqukHcTi0YzCvG1G3RxuZvNhPdPIvazGYy0eGxlsH0PngvNl//0BSML4xV2PD+pYgyZP8IeX+g97vTVraWWekt5NrIy8ibC4PIM5KyaZTCYzFEhazvbfm62mlZlYr9Mgmo84Z+/ffMRAuQmoVTTdXFEMmUxmbv5EtMR+FfhdWvXuhikTEEkzifPNqHOKpE8STk7FNuF2rVi9juuF/1dLDqVj/D0lidqSEnpvSvdPs336AOLrGUkbEGYGSxKJlIUI/dElqoyrD2wGLG/7gV4GS3qH7ZObbReoF2Wun8P1lXGu+IJnEpU7kSq/avtbaIuV+vdPPChpydrrS1oP+PeAjjVwcoJoSBmyD/u7iYTQfrZvrjiWTCaTaca3gXfQeDWtbP2EogbRU8BNtu8s8fgApD74nxMTyQmirP7d4yCimMkME0mDoilN2lj+G1iHsALfT9IfiWTu2bbv6uCw50rag1glXw/YFTipm7iHlFoV0H7EuXsi3c4/oHG9cJekzwM/StvbAW3/Z0l4eBuiFRBCpPxI298cQIy98g1iYfh4Qh9re2DZKgPqE3f2mhxKvJOpOlf128D0RJlTkndL5k6GfLrX1xwDfk5UfZ1BBxqKfRDF7oXPEO1ly0k6G3gpcX4fSSYmJ8dBuypTBpKeBWD74apjyWQymWEnuYXV6o2drIDLjuF8Qs/jzLS9AWHz+oayY8lkxhlJZ7V4eNL2Bm3GPxt4I6Frs2ka87I2YxYk9CH/m0iG/Jpw+ir9XFM1qbVuVdtXl3S8pYmWslpL85mEfX1LMwJJ1wBvqLWWpVaz822vNsh4u0HS5bZfJek626ukfZfZXrPq2KaDpJpExolMdWcr08CiJZJ+Q1RsXUwhGWL7y5UFVTHF9+EwI2kW8HriXHyB7VxBlBlfJL0EOJboj5+UdCWwVc0VJ5PJZIYBSRPADNuPpe3XEhdaAFfYfqjp4P7HMiyVO8+qJYcAbP9B0gElx5DJjD3T0T6TtDihRbQekXB4ETFBbHfMJ4GvpZ+xQtKKwEq2T0rJs4Vs39vs+UmA9miglERLSgS9p4ehE0XdIdsPpu+uYeLJdHu7pE2I1uTnVxdO36gJjO9U2NeRBpGk99j+Wd2+z9nu92dvBdsr9fk1R53rJC3dLvlaNbbvl3QmKb8iaZEBtbcOnJwgynTC9wmL+8PT9rZp35urCiiTyWQasBfwNPD5tH0CcDcwA/gF0cZRFgcB29VV7nwbKLty5xFJ69k+O8WxLsNnqZzJjDySbiSuk47odCIj6VCiYmg5QpT6bMLC+ULbTQVOk2PXpsB9ts+QtDNxTWaiQvD+6fwuVZOsrD9DJPhPApYBvsMcsf1m/FXSsmXJISgUaV/BVCHno9oMu1TS4cAP0/b2NHFnq5CDJD0P2JNohZsFzK42pOkzTQOL3SXdYfsCAEmfIN6P/U4Q3SRp0WEWLy8LSScSCbxZwLWpIrpY+dVMg6h0JG1KVBQulXYNsr114OQEUaYTFrd9WGH7cEm7VBZNJpPJNGZjokWjxj22X50mU+dQboJoWCp3dgF+Jqk22VyI0JXLZDL95cPEAtr1ki4ADgN+lap8mnEPUc1wQa3ysUMOBVYhXG5vAhYmtFDWJRbw3td9+EPFLsCawHkQ/bmSOnGBXBS4Jmk5FUWq+z6RTEm5HYkJ4aXA2sT3TLsE0U7EIkbN8ewMQqR8KEitevfbvo/4vVaoOKS+ktqAxNSk3rkdDN2UEEHfjKjyew/wlj7GtV+6ez9wmaTfMzUZMi9qEBX1nY6tLIrO2A/YHLjI9tNVBzNdcoIo0wlPS5JtwzNlv21FwjKZTKZknq7TSPspPGN/vFCTMYNiKCp3bF8qaQUKWkiMR6tAJjNUpM/62ZI+TkwUdgYOlXQscFgjbRzbn6/f1yFrAysTbj13EAt5T0j6AXBNj685TDxh+6E6y+hOdJV+kn7KYAdgLUI/aCNJqwBfaDcofUd9ZtDB9Upq1duL/lu/V46k9wLfBJ4H3E4kv64mhLhbYvtWSVsRSYt7gTf1WZO19lpOP5lot+v1HFk299aqy8aBnCDKdMIewHmSriJK5lYDtq42pEwmk5mL5xY3bO9b2Fys5Fgqr9xJK+7LAFfbvi7pnOxHVDk8r8xYMpl5haR1dhhwmKSXERUlH6e/rQaP254EHpZ0k+0n0rGflvREH49TFfekxciaZfRWwG3tBtk+ctCBFXjM9sOS5pM0kc6xK7YbJGlX4MdJr+QoIsm0s+3TBh5x51wlaS3bl1QdSJ/ZA3gV8Hvbq0t6M210pAptTjWeJqrTfiypb9Vp87IIdQvezhzJgKFE0iLp7i8l/S/h/Fes/BrJlv6cIMq0xfbvJK0MvCbtusj2/1UZUyaTyTTgCklb2z66uFPSlsCVZQZSdeWOpO2JNpT7gH8lO+YjgN8TrRuZTGZApMTQB4GtiEqFj/X5EDMkrUQs2hXvQ6F1ZoSZTbSUSNLNRPXlJu0GSXopoQO1jO3lJK0B/LftLw0gxkeSi9zVwL6SbqWzJOC2tg+QtD6wJLAd0W42TAmiVwHnJ12tYqveWtWF1Beesv3P5DCK7dMl7dtmTL2N/SmDCS2QdDDwpZogu6QXAJ+3PfIaUGPKQ0QCsXb+/U5hO2sQZcaXdOGxctq8LieHMpnMkLIHcK6ktzLHAWgtYENgnbKCGJLKnV2BNWz/SdIbCPHbLepdWDKZTH+Q9BxC+2c7YFngGGAj29cN4HCLMLUFqHh/khHH9g2SXgOsSEy0bLsTaYPvEmYF+6Ttq4CjGYz+3EeJytDdgK8DL6Gz6vra77E+cIztC5LuzzCxc9UBDIjHk2PcjZJ2ItzZnt1qQMlVaQBrF936bN+TWtTnRSSpaRXbMCQsbQ/bZ7cv5ARRpimSZhKlcm8CbiS+pFeQdBrwvlYOG5lMJlM2tm+UtDqxWr9R2n0lkSi5s4wYhqhy50nbfwKwfb6kv+XkUCYzUG4nErH7Ab+23YlmTk/YXnZQrz0MSHr53Lt0v+3b2wydlare94bBttwVEn8PAx/qYuijknYHtgDWTgmLsjXyGiLpONtb2D5H0jYVJEcGzZ7Ac4DdiWTiLCLR15YSq9MaVZws2OdjjAp3AJ+qOohOkPSt+iqvRvtGhZwgyrSippi/jO1/AyTbyx8TJ9evVBVYJpPJNML2XVTbsz4slTsL1bWdPF3ctv3nkuPJZMadFctKRM8D/BZ4EeHoBDGR/6ekx4jz6UVNxv0ntX3VtIuWITRj+s40EgbbEkmJ3W3fJWl5otpsGHhZ4f4uwFgliGz/Id29n1j87oayqtMulXQQkWieIBIkl/b5GKPCg7bPqTqIDmlUpT6ylV85QZRpxf8Ab64lhwBs3ydpR6JXOieIMplMZirDUrlT34JCYXuSaIfIZDJ9IieH+sqvgLNt/wpA0juBDYBfAgcxRxOznkPTcxaT9CXgA8DnBhRjTwkD2zcQGku17b8Bew8mxK4Z+fbEAVJWddoniPf4lcT/42QK75d5jIn2T6kWSZsRrpXLSjqh8NAsKnCu7Rc5QZRpxYxGekO2/5XazzKZTCYzlaGo3Bn3FpRMJjPWrFdszbB9kqQv295F0sLNBtk+StJNhKD1IsA2ts8bUIxdJQwk7Wt79wauWLXY++KGNU1mSXob8X31HEkbFx+0Xb/oMC9RSnWa7QcIgftM6EcOOzcQwuVrMVXA/AHgzEoi6gM5QZRpxaMtHhvZrGgmk8kMkFy5k8lkMtNjPkmvt30BgKTXATUx2Lkm5ZL2t71b2lzY9u4lxNhtwuCP6bbeFWuYuIU58hK3MlX/ZZK5v9vmJUqpTpPUUBPJ9qH9PtawUxTrHlZsXw1cLenXoxBvp0xMTuZqwkxjJN0HnN7goQlgQ9ulWTZnMplMN0jaEFjJ9iGSliRWe2+oOq5MJjP+5PPP9JC0NmFzX1uMXAR4P9F28+568WRJV9heo/7+gGP8ANFashpwGClhYPu4QR870xvJafChVO21CrAK8AvbHbWKSXojUZ02AfxmENVpkg4vbM4k3O4usv2ufh8rk2lGriDKtKJVz+swr4BkMpl5GEmfATYGlgIOIRxADgPeWGVcmUxm/Mnnn+lj+7wk3qw5u56ZxDcSTp5ocn9g9NrOJmkp4GBi4g/wB2CXrGFVCmcB60halHAWvQ54KyEc3hRJ8wOXpsTjH1s9d7rYntJelt4v3xnkMTOZenKCKNOUMbS3zGQy8wZbEJbylwDYvi2tHM5zSHqZ7b+025fJZPpGPv9Mk4LN/X/S7QptbO5nFLTeiveBgWq//QX4SyO9zhYcDZxLuIQBbJf2deuqnK6+1QAAIABJREFUlemeCdsPS9oC+KHtL0m6tt0g2/+R9JCkmbYfKyHO4rHvlLRimcccZiQdZ3uLquMYd3KCKJPJZDLjxqO2n5RU3Dev9lMfC9S3WzTal8lk+kM+/0yfbm3u67Xfivf7rv0maTbwGWDxtP1PYB/bB3UwfCnbRRfgvVLCIjN4ZkqaAbyZqO6DOUnIhkiasD0JGDhX0s+Ah2qP91sbqE6DaD7g1cA/+3mMEUftn1IehWR2Q8oyJuk3OUGUyWQymXHj1qQVMClpPmAP4E8Vx1QqkhYDliAuiIur6bOAZ1UWWCYz/szz558+0JXNfZmujZK2Aj4CbANcTJxb1wIOlPR/to9p8xJ/lbSC7b+m11uecELKDJ7jgbuAG4HzJb0QaFcRdDmxoLIA8TleqfDYIBK/ry7cfwr4M/CJARxnVCmlhbQLTiHeBxPAiwn3skniWusWYLnqQuudnCDKZDKZzLixE3AUIUD5CHAesGWlEZXPloSO3NJMXU2/H9ivkogymXmDfP6ZPj3Z3JfEDsB7k3tRjd9Leh9RldIuQbQw4XpU07J5A5GsOAGGw+4+LTAcQrS9TRKGNbvY/lelgU0T21+WdDBwfxKqfgh4d5thE2lsKdbzZR1nhNmw6gCK2F4OQNK3gXNtn5i23wOsU2Vs0yEniDIdkfpfV0pf0s8GFhonO79MJjM+2L4LeIukRYD5bD/Ubsy4kVodDpK0h+2vVx1PJjOvkM8/faErm/uSeWFdcggA29ckx7p2HMPUJNKxfYusf3yfqJbZLW3vkPZtWllEfUDSBPE7rAjsDixGLKI007aCBppWRfrVQtTM3r5wnHnO5r4RQzz3XMf2TrUN2z+TtGeVAU2HnCDKtEXSNsBngYWAk4BlCEX9LKiXyWSGhma94DUtkFHtBZ8OteSQpCUIy9za/lsqCyqTGUMkbdxkPwC2f9vo8UxDPgYcL2mKzX1aoDywurAAeLDFYw+3GzwiBjDL2y5W1nxR0lWVRdM/DgCWJFrGdif+l98iWgSbsTzRRtQoQdRPfatDiHa2axscK2uYDT8TktauORlKegNzktojR04QZTphNuHIcR6E12jq281kMplh4pQWj/VdqHQUkLQ+0e6yJCHGuRBwD6FPlMlk+sen0u1MQkek5o60KuFolhNEHdKDzX2ZLNGi2mOxZoMkHWF723R/T9t7FR472fY7+hvmtJhP0hK2/wnPLDCM7GS3wPrA6sAVALbvkTSz9RD+bHv1gUcWbnbbEK2pRwLH2r6vhONm+sPHgOMk1ZLECxOOliNJThBlOuEJ2w/VOXI8VVUwmUwm04haL3hmCt8kevaPJ1ZNtweWrTKgTGYcsb0+hA0zMNv2xWl7LeZYmmdaIGmG7cdTex7A39LtApIWsP1Is7ElcgZThYSLnNli3GqF+5sCexW2l5luUH3mm8CVkmqLLhsTrm2jzmO2J2vzmSQiPxSix7aPAI6QtByRKLpA0rXAXravqTS4TFtSUvslNE5qjxw5QZTphHuSBtEkPOPgcFu1IWUymUxjmrSa3W+7lc7A2GL7BkkLJqveH0m6DBjZ3vhMZshZpZYcArB9iaRVqwxohLiQSGQ/xBxnoOLt/NWFFkxDRHiiyX0YshYi20dJugJYL+06yPY4OPFdK2lLoh1oWUI+47w2Y84feFQFbP9d0oGE29pXgNOAeTJBJOlSWnw2bLdqDayCBYDH0+0KkkZW2iAniDKdMJsQ0ZOkmwlXjk2qDCiTyWRa8FvgRYRjF4Td6D8lPQZsYfuiyiIrnyfT7e2SNgFuBp5fXTiZzNjzsKStbP8EIE1Ih6HyZeixvUa6HYd2pnomm9wfSmxfB1xXdRx9ZldCh2gp4GLg18wR4m6I7Y+XEFdNQHsj4INEm9kJwGts/72M4w8pn6w6gE6R9DFgH+Be5gjpj6y0wcTk5NCfozJDgKT5CdX/CaJs7j8Vh5TJZDINkfQt4Gzbv0rb7wQ2AH4J7Gv7NVXGVyaStgB+B6wAHEckyz5Rm7xmMpn+khyPjiYmeZOEFtE2tq+vNLARJOnDrAz8fYjdizoiaZPUqnBWLtyfIFyCn11JYA2Q9HpgP2JyuwCpgst21q4bEJJuB+4AjgDOpi6JOKqVKPMKkm4C1rf9j6pj6Qc5QZTpiNQP/l8Uqs7yySqTyQwjkq6y/cpG+yRdY3u1ZmMzmUymH0haFMB2K9erTIEkqv8dYhX+04R22oLAs4kk288rDG9aSFq31eO2zykrlnZIuh74KnARYW4AwDhMfiVtSDiTFeczlVvIpw6N2qS81lZZY9L2SFaizCtIusD266uOo1/kFrNMWyTtDHyNMSmby2QyY898kl5v+wIASa9jjgPL082HjSfDekGcyYwrkjYC3pTun2b79IpDGhW+QbT8PJdwpXyX7XOShtPRwMgmiIYpAdQBj9o+tuog+o2kIwhX5iuYk/hqWymRuii+Z/vDg4rN9rKDeu1MKZwuaT/gp8BjtZ2jWkyRE0SZTpgNyPYdVQeSyWQyHfAx4HhJNd2PRYD3S3o2cGB1YZWPpCOBV9HlBXEmk+kNSZ8iXIiOS7sOkHSk7W9WGNaoML/tUwEkfaWWVLF9bZ2TbmVIOpHWwrmblxjOoPitpLfV/hdjxOuBlW0/2faZBWz/R1KuPM604gPpdrPCvpEtpsgJokwn3JaTQ5lMZlRIdqPL09hu9MiKwqqK19HDBXEmk+mZrYHX1VrLJB1MOCHlBFF7iomX+ta8Yan+PLnqAAaFpH8xp71pD0kPEq5M46JBdOs0xv5B0iHAUYTLHjC6FSKjRqqEXsn2IZKWBGbZvqHquGrYXq7qGPpJThBlOuGLkn5EOAMVy+Z+W11ImUwm05KxsRudJtO5IM5kMt0zUdQdsv1gcijKtGc5SSc0uD8BLFtNSFOxPc6LDGtWHcCAuQE4U9KvmDqf6aTl+n3p9u2FfSNbITJKSPoMsDHhPncIoUt2GPDGKuNqhKQlgJm1bdu3VBhOz+QEUaYTNkk/KzK1RSEniDKZzNAxbnajvSDpo+nudC6IM5lM91wq6XDgh2l7e+CyCuMZJWYX7p9S99hQVO5I2sX2QUlvZC5sf7rsmPrIBsBZtm+uOpABMRP4G7BqYV9HLdfjViEyYmxBJC8vAbB9m6TnVBvSVCRtQFSoL0nMlRcC7gFGsuouJ4gynfA/wLK2H606kEwmk+mA3YBVxsFxZRq8unC/pwviTCbTEzsBnwcOTttnEI5QmTaMSHVOLdH+cK8vIOlk2+9otl0hmxKaWfcTVutnAWePy3ep7Q9O9zXGpUJkxHjU9pN1OmTDdh3zDWBDwnlxDWJhYNkqA5oOOUGU6YSbgKxfkclkRoW7xuWCdhr8yvZJVQeRycxr2H4Y+EzVcWQGg+3vp7v7TWPh9ItttivB9iaS5iMmuOsC7wEOlPRvorJo+0oDnCap1XMHksMgcBrwI9udOJmNVYXIiHGrpDcCk+n9uQfwp4pjmgvbN0haML2ffiTpMmDPquPqhZwgynTCjYQ4W25RyGQyo8BY2Y32yBeBnCDKZEpG0q7Aj23fL+koYC1gZ9unVRxapr/cIukk4HDb53cz0PblrbarxPbTREvkZZJOJaoiPg5sTlRFjDL7AasDh6ftbYCXAp20BY5VhciIsRMhDr4K8AhwHrBlpRHNTa2Q4nZJmwA3A8+vLpzpkRNEmU6YQW5RyGQyo8NY2Y32SBbFzWSqYVvbB0han6g22I5oN8sJovFCwPuBg5IeyhHAUbZvazloagLxaKIdeCgSiJJWAtYD1gdeSWjYnQtsC1xaWWD9YyNgDdtPASQR9MvpLEE0VhUio4Ttu4C3SFoEmM/2Q+3GVMBBkp5HvB+OA2YxVVNtpMgJokxb+tGzm8lkMmWRxSQBWLqZiCqMvJBqJjPM1Mw81geOsX1BaovIjBG27yUclQ6RtAqhffd3wmGpFcUE4hIMVwLxT8BFhGbW7zppvRoxJpi6wD1J54spY1UhMgpI2rjJfmC43LRtH5fuXgqsUGUs/SAniDIdofg0voKpwmxHVRdRJpPJTEXSDNuPp1WmubD9SNkxVcjTTENENZPJ9MyjknYnnHfWTronC1Uc08gh6Qe2d2i2PQykxN/GRIXNOkQVUTuGOYH438Tv8SVgP0l/BM4hhKrvqjKwPvF74FRJR6TtbYDfdTi2UYXIJ/oeYabIp9LtTKLS7tq0vSrhaDY0CaJxIyeIMm2RtDOwI7AUkRldm/jCyAmiTCYzTFxIaAM8xNwrg5PA/FUEVRF32v5y1UFkMvMg2wIfBXa3fZek5YFjqg1pJPlNm+1KkXQA8F6i6uZIYOsORauHNoFo+2TgZABJzwbeSIhVf1nSpO2XVRlfH/g0MZ/ZNG3/EvhBh2NPsf0AhQqRYbNaHzdsrw8g6Thgtu2L0/ZawC5Vxjbu5ARRphN2IEQWz7e9USql/ULFMWUymcwUbK+RbodlNbZKsgZRJlMBtm+goD1h+2/A3tVFNHoknZf6hNB5lQTTnHuA19q+tctx2zLkCURJixNaROsRlU4vAi6uMKS+kAS4v5t+uuVsYgGq3b5M/1mllhwCsH2JpFVbDchMj5wgynTCY7YfljSfpAnb10laseqgMplMphWSnktc4N5k+5qKwymb3asOIJOZl5C0r+3dJZ1IAyMP25tXENaocgQFlyJJixKtQK+tKqBCLC+2fYvtrzV4bA3bV7QYOz/wKdsfru0bpgSipEOJiqHliEqZs4lk1oW2H68wtGnRSo+PcDq9ETje9hMNxi5AVHjNJ2lh5iy+zAIatrNn+s7Dkray/RMASVsSbmaZAZETRJlOeETSgsDVwL6SbmXeatXIZDIjgKSfAN+wfbWk5xPnrAeAxSR9zvaPqo2wPIbBESeTmcf4Y7o9udIoxoO7Cwm3hYm/6bBU2fyKVDUi6RLbaxUe+xEtKkps/0fSagOObzrcQ1iKX2D7saqD6SOt9PgWADYH3gm8p8HjnwO+SCR9i6/zALB/vwLMtOSDwNGSfkT8H64l9KOGBkkvAvZjbr3ekXTPzQmiTCd8lMie7wZ8nbCK3rrSiDKZTGZu1rB9dbq/NXC97bdI+i9igjHPJIgymUy51FqibB9ZdSxjwG7ACckS/q2E/su3K46pRrF9t96xrJPW3j9IOoTQ8XzGrtv2n/sQ27Sw/fmqYxgEnejxSbquxdgvSzrE9sf7HlymLbavB9ZMlYTYfrDikBpxGPBTIkG0JfC/wN8qjWga5ARRpimpFHaG7dpJ82HgQ5KeRZRkZjKZzDBRPC+9kRCgxPZtksbNrjeTyQwhkpYibMvXT7v+AOxi+87qohoN6hwoPwKcSrQ5HSJpkSFxoqy3SW/2WDPel27fXjduJCsNxoi3t3qwlhyax1vXK0PSRsCb0v3TbJ9ecUj1LGb7x5Jm275Q0sWEccpImoXkBFGmFfsAZu5V9/cBItwAMplMZmiQtDRwH3EB98XCQzMbDhhTJL281ePDsFqdyYwpRwPnMsdlZ7u0702VRTQ6FB0oa7drAp9keJwoZ0paiYiteB86+J6xvdwgg8v0hu1/NNqfW9erR9KniJay49KuAyQdafubFYZVT02/6iFJLwbuBhavMJ5pkRNEmVZsQGOh0yOIE2ROEGUymWFib+Aq4ov6j7UkiKTXArdUGVgFnMKcCdaLiQvaSUJY8xZChDSTyfSfpWx/pbC9l6QtKotmhBgRB8pFgN8Wtov3O65UlbQEU7VK5rXvqFEht65Xz9bA62qtZZIOBs4HhilBdG5KIB4KXA48Dvys2pB6JyeIMq2YP1lCTiGJ7M21P5PJZKrE9omSzgNeSCSxa9wCfLjxqPGktkot6dvAubZPTNvvAdapMrZMZsz5q6QVbP8VINmY31BxTCNDkje41PZQ2ofbXnY64yVtABwJLAn8h9D4vAdYYtrBZQZBbl2vnomi7pDtByV1ovdVJnvYfpIQ0z4HeA5wW8Ux9UxOEGVasXCjnm9JzwZmVBRTJpPJNMX2XcBddfvuqCicYWAd2zvVNmz/TNKeVQaUyYw5CwNXS6q5mr0BOF/SCZDt7tuRFiEfkjRzzJy0anwD2BA4nnA82x5YtsqA5hV6dZrKreuVc6mkw4Efpu3tgcsqjKcRRxDi1Ni+JQlqnw68tsqgemUUSjkz1XE8cKSk59R2SJpFlFOeWFlUmUwmk+mUCUlr1zYkvYH83Z/JDJJjCPfXY9PPx9LtKekn0x4TLRuflvTR2k/VQfUL2zcAC9qeTBo2b606pnmEw4Az0v0tgT8S1VytqLWu30huXa+KnQhNn4PTz7+AYXOUu1vSvgCSFibaD4+pNqTeyRVEmVZ8hciI3i7pxrTvpcCvgS9VFFMmk8lkOudjwHGSHk7bCwNZDyWTGRDZ5r4vLAD8CVipsG9c2nmeTLe3S9oEuBl4fnXhzFN07TSVW9erx/bDwGeqjqMNuwEnSNqVSPieYvvbFcfUMxOTk+Nyvs0MCkkrAKunzStrffWZTCaTGX4kLUQ4TwLY9hOtnp/JZLpH0hG2t03397S9V+Gxk22/o7LgMkNDEiz/HbHgeixhHDDb9shWG4wKki62/RpJFwLvJapSrm/XYpaplpR0+bHt+yUdBawF7Gz7tIpDQ9Iihc2FgVOBs0mFFPUyLaNCriDKtCUlhHJSKJPJZEaTBQhHjQWAFSRlm/tMpv+sVri/KbBXYXuZkmMZeSSJubVijqououkh6Uhi4niW7fuAS4AVKg1q3mOsnKbmIba1fYCk9Qlx9+2IVrPKE0TAQ8xxjK3drgl8Mm3PX11ovZMTRJlMJpPJjCmSPgbsA9wL1NwnJ4G8YprJ9JeJJvdhfNqjSkHSzsCOwFLApcDawDnAyCaIgGuIxOEBkv5NJIvOJhJGWcumBGx/Kt19xmnK9nVVxpTpiP+k2/WBY2xfIGkotBRtD0Uc/SYniDKZTCaTGV92A1ax/Y+qA8lkxpzJJvcz3bMD0UZyvu2NJK0CfKHimKaF7f2B/dPEdnXCEes9wIGS7rO9fJXxzSuklqD/IubAT0t6ea6oHXoelbQ7oZ+4drK4X6jimBqSWvqfya+MaovZWGa9Mv1F0ss62ZfJZDKZoeOunBzKZEpBki6RdEnxvqRLgRWrDm7EeCwJ084naSJVeYzF39D208AjwKPp59/A3yoNah4hVdTeTdiP11wFT640qEwnbEtUE+5u+y6iAnqoNLskbSrpNuAx4EGi9ezBaqPqnVxBlOmEY4E1OtiXyWQymeHidEn7AT8lLlwAsgZRJtN/Nq46gDHiEUkLEq5R+0q6lRHV8qgh6X+JqqFXEJbp5wL7A5fbfqrC0OYlckXtCGL7BmB2YftvwN7VRdSQ/YDNgYtSEnikyQmiTFMkLQYsAcyUtBJzeupnAc+qLLBMJpPJdMoH0u1mhX1ZgyiT6TO2z6k6hjHio0QLyW7A14nz1daVRjR9Dib0lL4M/MH23RXHMy+SK2pHCEn72t5d0ok0aNu1vXkFYTXjXtsXVB1Ev8gJokwrtiQytksDvy3sv5/IlGYymUxmiLG9XNUxZDKZTCdIei6wJyDgSmBv2x+qNqq+8QLgjcC6wC76/+3de7ClZXmm8asbwRYVFOUYk6GByS2IqBwcEhEaSMgoEmfwUDpqiUETY0FQE2PGQzEeyUgwgaBjEmLEE0WYjDEDxERDNwgtxwgSYR4kBCkJGMJBQGkQ2PPHtzYs231Y3b12v+tw/apWre97v+6qu6zNtteznvd5kycDl9ANqr6wqm5rmG1a2FE7Xi7uvY/DNsAv9boEz+Ynf7bGcgbRspkZ5+hpYUneU1UfbZ1DkrRxkuzATx4X7ak5kkZKr1MA4ALgV4Ebq+r4hpGWTJKnAK8A3gPsUVVjvYVuHCT5lzmWZ6rKjlptkiRzbSubGdf/ru0g0iD+Osle6y9acZek0ZbkMOBMYEe6o2K3Au6k2z4sSaNkr6p6DkCSTwPfaJxnqJJsTzeHaPa1Ericbq6nlpgdteMpyc50WzQP7S1dAJwwSl13k3bcvQUiDeK8vusVdB80vkv3f2ySpNF1MnA4XdvzvsCxwK4tA0nSPB6cvaiqB5O0zDJUSa6j+917JbAaOA5YW1UPLvT3NHx21I6dz9ENdT+hd/9rvbVfapZoDkmeARzYu/1GVd3VMs+msECkRa1fcU9yOPCSRnEkSRugqm5IsmVVzQBnJLmSbs6HpCWQ5Nyqetl895rXyiR/Od/9iA2l3VDHA5dU1bpF/6SWhB21Y2vnqvpg3/2Hk7y2WZo5JPkV4PPA1b2lTyd5fVV9tWGsjWaBSBusqv4hycmtc0iSFvXj3vutSY4Cbga2axdHmgonLnKvub19vfvz5vxTY6iq/qF1BtlRO6ZuTLJHVd0IkGR34IbGmdb3EeDgqroeoHf69+cAC0SaTOvNH1oOHAA8sVEcSdLgTk3ydLqOobOAbfnpD2GShqiqrlroXnOrqjNbZ9Bks6N2LD0JuCbJ7KlmLwIume0uHJHOwi1ni0MAVXV9ki1bBtoUFog0iP5vcB4GvgO8sVEWSdKAquqs3uUVwB4ts0jTIMk7gT+vqh8k+Rzdl2q/VVV/3ziaNO3sqB1PX+i9Zo3iUPc7khxTVZ8BSPJG4I62kTaeBSItyqn/kjRekhy80POqumhzZZGmzDFV9fEkh9LNNvk1uhN4LBBNuSRbAJ+qqre0zjKl5uqofUfbSFrMmHQWvhX4fJJPATN0s4he1zbSxrNApAX1Jv0/UFX3JdmX7ojBqqpzG0eTJM3vlL7rPYHr++5ngBdu3jjS1Hik934o8IWqWptkoo5A1sapqkeS7NM6x7Syo3a8JPlMVR3Tu35fVX2479lIDP5P8iZgTW8+0oFJngJQVfe3TbZpLBBpXkmOBz4KPJjkg3RV9suAtyY5oKocuihJI6iqDpi9TvLN/ntJS+qBJO8GXgu8OMkyutOStIGSbEXfZ5Wq+lHDOMNyQZLTgc8Cj32IrKrr2kWabEleVFWXJHnpXM+r6vzNnUkD6S+mHg18uO/+ZzZzlvkcDXw8yQ+ANcDq3rsFIk2stwIrgW2AbwMrq+r2JE+lKxRZIJKk0TfTOoA0RY4B3ga8u/dvpt35yfkZWkSS/wr8MbBzb2kZ3e+xLZqFGp7X9N6P7FubAXZrkGVaHANcArxrjmczgAWi0bRsnmsYkX/XVNVRvQ7RfYFDgFcCf5jkHmB1VR3bNOBGskCkhTxUVf8O/HuSG6vqdoDedrMHG2eTJEkaGb0ZM+/qnzFTVf8MnNQu1Vg6GXg1cGlVPdo6zDA513Pzm/3vsaoObZ1FG2RmnuuR0vsddSVwZZK/BQ4HjqP7HWaBSBOn/z/GHy/wTJI0QpLs1Xe7Isme9H0D53YGaficMTM0d1XV2tYhllJvxueK2fuquqVhnIk239ayWW4xG1lJcvkc18uAn2+U6Sf0/m21im7m3POBG4CL6LrWrmgWbBNZINJCnpvk33rXT++7XkY3+V+SNJrOW+++/x/AbmeQlo4zZjZSkq17l19K8pvA2cC62eeTMIMoyWHAmcCOdAPNtwLupDvxTktjdmvZCuAA4Nre/XOBy3GL2ahasLA3Ir4NXAp8CPhKVU1EA4UFIi3ECf+SNIbcxiA144yZjXc/3f9Ws92On+i7n5QZRCfTbUE5m25uybHAri0DTbrZrWVJzgLeXlWX9e5fCJzQMpvmV1UXts4wgF8FDgb+B/CxJBcDF9KdbHZ7y2CbYtnMzEQUuiRJkiRpZCW5qqr2S/JPVbV3b+3Kqtq/dbZJl+Taqnruemvfqiq3hWqT9Y64P4huWPXRwExVPbttqo2zvHUASZIkaZIk2SHJz82+WucZJ0n+aJC1MTU70/PWJEcleS6wXctAU+SHSV4/e5PkdcDYb1tUe0m2B14CHAW8HPhZ4LamoTaBW8wkSZKkIXDGzFAcPMfaIZs9xdI4NcnTgfcBZ9HN9HxH20hT403A55KcQbdl8VrgjW0jaZwl+STd76aVdEOp1wBvA75RVWN74rcFIkmSJGk4nDGzkZK8iu5o6F2T/GXfo22ZnE6P86rqXroPk3sAJNmmbaTpUFXXA/sneWrv/r7GkTT+7gSOB9ZW1brF/vC4cIuZNliSnZM8sXUOSZKkUVNVNwBbVtVMVZ0B/OfWmcbEDXQnMN7Xe599fYpu+8YkWDPgmpZAkt2BXwBenOSlScbhpKypl+Tche5bqar3V9UFk1QcAjuItHEuALZLcmpVfbR1GEmSpBHxEzNmgJtxxsxAquoa4Jokf1NVd7XOM0xJnkC33XB5kifx+Elt2wJbNws2RZKcBLwZuJ5u+yd0W8085n70nbjIvYbIApE2WFXt2RvGdVDrLJIkSSNkdsbM+4Ev0hUA3t420tj5kyQ/dcxyVb26RZgheS/dh9oZ4Id96/cCpzRJNH1eBeze2+KnMVJVVy10r+HymHtJkiRpEyQ5k26r0OqqurltmvGWpH9w8ArglcB1VXVCo0hDk+T0qjqudY5plOTrVfXi1jm0YZK8E/jzqvpBks8BBwC/VVV/3zjaY5KsmKRtZhaINK8k59B90zGnMf8mR5IkaSiS/Dawiq67+h66YtEauoLRLc2CTYAkWwF/X1WrWmcZliRPo/t5uamqvtU4zlRI8jG648fPAR77MF9VbjEbYUm+VVX7JDkU+D3gA8BpVbV/42iPSXI78AXgk1X1z63zbCq3mGkhIzEATJIkaZRV1SnAKUmWAy+g+/D/SuAPk9xdVbu3zDfmZoCfaR1iUyT5PHByVV2TZDvgGrrtZc9M8t7eMHMtrQN678f3rTmDaPTNzos6FPhCVa3t/Z4dJfsAvw5ckOQ64BNVNbafo+0gkiRJkoYkyZ50H2ZWAfsDN1bVEU1DjZH1OtiX0334+lpVva1dqk2T5Lqq2qt3fQJwZFUdkeRZwLlV9fy2CaXRlGSlKfpOAAAVZklEQVQt8GW6AeMvBr4PXFtVezcNNockWwAvB/6IrrB1Ol2xaKy2n9lBpHn1WjHnVVW/u7mySJIkjaokv0lXEHoe8B3gIrrhw1dV1cMNo42j/m/eH6brvLmsVZgh6f+AeBDwJYCq+t5cA7k1PEn2Wuh5VV23ubJooxwDvA14d1XdnmR3uu1cIyXJ1sAb6LLeCJxB90XB+cBhDaNtMAtEWsgPF/8jkiRJU+804Aq6+RgXVNX3G+cZW1V1JkCSJ/fuJ+Lfo0l2Ae6mKyT2H9O9okmg6XHeAs9mgN02VxBtmF5Hzruq6i2za70ZPye1S/XTkpwOHA38DfC6qvqn3qMvJvl/7ZJtHAtEmldVfaB1BkmSpDHwDLrOkEOAE3rFjUvoBlVfWFW3Ncw2VpLsBnyRrhuLJN8EXl9VNzUNtmlOAq4GHgIunu1aSXIg4BDzJVRVK1tn0MapqkeS7NM6xwBuBp5TVXfP8ezQzZxlkzmDSPNK8qqqOifJnHu+q+qTmzuTJEnSqEvyFOAVwHuAPapqi8aRxkaSrwJnAX/RWzoG+G9V9cvNQg1Bkp2AnYBrqmqmt7YL8ARPupPmluQk4KnAZ4H7Z9fdGrh07CDSQvamOwrygDmeWVmUJEnqSbI93fah2ddK4HK6bhgNbvuq+nTf/V/0BjuPtaq6Hbh9vbV/bRRHGhev6b0f2bc2ElsDk9zB3J+JlwEzVbXDZo40FHYQSZIkSZugd7TxrsCVwGq6rWVrq+rBhrHGUpJ/BF5bVdW7/3ngrKrar20ySXpckv+w0POq+u7myjJMdhBpXkkOXuh5VV20ubJIkiSNsOOBS8btOOMR9R7g60mupvsmfh+604EkTakkO9A30H0UtmWOawFoMXYQaV5Jrui73RO4vu9+pqpeuJkjSZIkacL1tuv9p97tpVX17y3zSGojyWHAmcCOwCPAVsCdo7B9q/dZed5iyrh+VraDSPOqqsdmDyX5Zv+9JEmStBSq6o4kF9PNctoFsEAkTaeTgcOBs4F9gWPptvOOgt9pHWApWCDSoGw1kyRJ0pJI8nng5Kq6Jsl2wDXAvcAzk7y3qs5om1BSC1V1Q5Ite6f/nZHkSuB9I5DrwtYZloIFIkmSJGkTJdkC+FRVvaV1ljG1b1Vd07t+A3B9VR2R5FnAuYAFImn6/Lj3fmuSo4Cbge3axXlckv9ZVe9Ocg5zNFNU1asbxNpkFog0ryR79d2uSLIn3bBAAKrqus2fSpIkafRU1SNJ9mmdY4z1D/g+CPgSQFV9L4md7NJ0OjXJ04H3A18EtgXe3jbSYy7uvZ/bNMWQWSDSQs5b7/78vusZYLfNmEWSJGnUXZDkdOCzwP2zi36pNpgkuwB3080eOrHv0Yo5/4KkiZTkTGANsLqq7gYuB/ZoGuqnPTPJyqo6s3WQYbJApHlV1crWGSRJksbIa3rvR/at+aXaYE4CrgYeAi6eLaolORBofqS1pM3qW8DRwMeT3ENXLFpDVzAald8Hs/l+QC8bsKaqvts01SbymHtJkiRJzSXZCdgJuKY3kHa2q+gJI/ShUNJmkmQ58AK6rsJVwIuAu6tq94axHtPLty9wCI/nu4eukHVsw2gbzQKRJEmSNERJdqBvW5TFDUnaOL05uIfSFWD2B26sqiOahppDb37v4cBxwC5V9dTGkTaKW8wkSZKkIUhyGHAmsCPwCLAVcCewQ8tckjROkvwmXUHoecB3gIuAU4CrqurhhtEe0ytcraIrXj0fuIEu5zHAFc2CbSILRJIkSdJwnEz3DfLZdNsOjgV2bRlIksbQaXRFlg8AF1TV9xvnmcu3gUuBDwFfmd0WO+7cYqaBJDmCrjLa3y79wXaJJEmSRkuSq6pqvyT/VFV799aurKr9W2eTpHGRZBvgILrZPocATwYuoRsGfWFV3dYuXSfJy4CD6fJtTXfs/YV0g6pvb5ltU9hBpEUl+X3gAOA5wJeBlwNfaxpKkiRp9Py4935rkqOAm4Ht2sWRpPFTVfcC5/deJHkK8Aq6jqI9gC3apetU1bnAufBYvtmC1geSzFTVs1vm21gWiDSII+mmx19VVb+R5IPAnzXOJEmSNGpOTfJ04H3AWcC2wDvaRpKk8ZNkex4/vWwVsBK4HPhis1BzWC/nocDPApc1jLRJLBBpEOuq6uEkM0m2rKpbkzyrdShJkqQRc17vm+8r6L7lnt0qIUkaUJLr6Oa3XQmspjsZbG1VPdgyV78kn6TrGFpJ9zt/DfA24BujlHNDWSDSIO5LsjWwFjgzyW3AA40zSZIkjZo1dMOpF1uTJM3veOCSqlrXOsgC7qTLuXbEc24Qh1RrUUl2BO6mKyi+E3gacFpV3dI0mCRJ0ghI8gS6I+3XAr8ALOs92hZYPa6zKCRJ08UOIi2q71jBh4APt8wiSZI0gt4LnAjMAD/sW78XOKVJIkmSNpAdRJIkSdIQJDm9qo5rnUOSpI1hgUiSJEkaoiRPozvR5qaq+lbjOJI0dpJsAXyqqt7SOss0Wd46gCRJkjTOknw+yfN619sB1wIfAb6a5M1Nw0nSGKqqR4B9WueYNs4g0kCSvBQ4rHd7QVWd3zKPJEnSCNm3qq7pXb8BuL6qjkjyLOBc4Ix20SRpbF2Q5HTgs8D9s4tVdV27SJPNDiItKslHgN8H7uq9PprkQ21TSZIkjYz+I44PAr4EUFXfoxtcLUnacK8BjgTOBs7rvc5tmmjC2UGkQbwKeEFV/RAgyanAN4H3N00lSZI0IpLsAtxNN3voxL5HK5oEkqQxV1UrW2eYNhaINIi7gR/13a/rrUmSJAlOAq4GHgIunt3+kORA4JaWwSRp3CXZgb5ie1X5e3WJWCDSINYCf5vkzN7964GLe3OJcB6RJEmaZlV1TpKvAzsB1/Q9ugXwBB5J2ghJDgPOBHYEHgG2Au4EdmiZa5JZINIgnt97//W+tX17rxnAApEkSZpqVXU7cPt6a//aKI4kTYKTgcPpZhDtCxwL7Noy0KSzQKRFVdWhrTNIkiRJkqZLVd2QZMuqmgHOSHIl8L7WuSaVBSLNK8nB6y3NAP8G3FhVjzSIJEmSJEmaDj/uvd+a5CjgZmC7dnEmnwUiLeSUOdaeCTyc5OiqunZzB5IkSZIkTYVTkzydrmPoLGBb4B1tI002C0SaV1UdMNd6bzj1HwK/tHkTSZIkSZKmxHlVdS9wBbAHQJJt2kaabMtbB9D46Z1aZmufJEmSJGmprBlwTUNiB5E2WJLlwBatc0iSJEmSJkuSJ9Adab88yZOAZb1H2wJbNws2BSwQaV5J9ppj+RnAm4Gvb+Y4kiRJkqTJ917gRLpDkn7Yt34vc8/J1ZAsm5mZaZ1BIyrJv6y3NAPcAXwV+EhVPbD5U0mSJEmSJl2S06vquNY5pokFIkmSJEmSNJKSPA1YBdxUVd9qHGeiOaRakiRJkiSNhCSfT/K83vV2wLXAR4CvJnlz03ATzgKRJEmSJEkaFftW1TW96zcA11fVc4D9ALecLSELRJIkSZIkaVSs67s+CPgSQFV9j24urpaIBSItKslhg6xJkiRJkrSpkuzSO+J+FXBh36MVbRJNBwtEGsQfDLgmSZIkSdKmOAm4GvgOcHFVXQeQ5EDglpbBJt0TWgfQ6EqyB/DzwDZJXtr3aFtg6zapJEmSJEmTqqrOSfJ1YCfgmr5HtwBvaZNqOnjMveaV5I3AMcD+wJV9j+4F/rSqzmuRS5IkSZIkDZcFIi0qyTFV9ZnWOSRJkiRJ0tKwQKSBJNkd2J2+bYlVdX67RJIkSZIkaVicQaRFJfko3V7P64FHesszgAUiSZIkSZImgAUiDeLVwO5VdW/rIJIkSZIkafg85l6DuM3ikCRJkiRJk8sOIg3iG0nOAs4B1s0uOoNIkiRJkqTJYIFIgzig935835oziCRJkiRJmhCeYiZJkiRJkjTl7CDSopIsA34N+I9V9XtJdgV2qaq1bZNJkiRJkqRhcEi1BvFx4HDgv/Tu7wP+qF0cSZIkSZI0TBaINIhDgdcBDwBU1Z3AiqaJJEmSJEnS0Fgg0iDWVdVjw6qSLAeWNcwjSZIkSZKGyAKRBnFtktcBy3rzh/4X8PW2kSRJkiRJ0rBYINIg3gmsAnYGLqP7uXlXy0CSJEmSJGl4POZe80ryAWANsLaqHmwcR5IkSZIkLRGPuddCtgNOA3ZLcgWwmq5g9I2qeqhlMEmSJEmSNDx2EGlRSbYHDqbbZrYKWAlcVlWHN4wlSZIkSZKGxBlEWlRV3QH8HfC3vde/As9qGkqSJEmSJA2NHUSaV5KX0HUMHQI8EbiE7vSyi6rqtobRJEmSJEnSEDmDSAs5D7gU+CDwlap6tHEeSZIkSZK0BOwg0rySPIeue+gQYF/gRuBCukHVV1TVI+3SSZIkSZKkYbFApIEleTZwOHA8sEtVbdM4kiRJkiRJGgK3mGlBSZYD+/H4CWYHAXcBf9UulSRJkiRJGiY7iDSvJOcDvwjcw+Nby1ZX1c0NY0mSJEmSpCGzg0gL+SvguKq6qXUQSZIkSZK0dOwgkiRJkiRJmnLLWweQJEmSJElSWxaIJEmSJEmSppwFIkmSJEmSpCnnkGotKskzgdOBXwJmgK8CJ1TVHU2DSZIkSZKkobCDSIP4E+AG4HnA84Hv9NYkSZIkSdIEsINIg9i9ql7Rd39ikqubpZEkSZIkSUNlB5EGsTzJDrM3vWt/diRJkiRJmhB2EGkQfwB8M8l5vfuXAr/XMI8kSZIkSRqiZTMzM60zaAwk2RtY1btdXVXfbhhHkiRJkiQNkQUiSZIkSZKkKecWMy0qyS8CHwN2o/uZWQbMVNUOC/5FSZIkSZI0FiwQaRB/DnwIuBR4pHEWSZIkSZI0ZBaINIgHquqLrUNIkiRJkqSl4VHlGsT5SV7SOoQkSZIkSVoaDqnWvJLcAczQzRx6BnAf8CDOIJIkSZIkaaK4xUwL2b91AEmSJEmStPQsEGkhhwGrq+rm1kEkSZIkSdLScYuZ5pXk/wIHAT8A1gCrgTVV9d2WuSRJkiRJ0nBZINKCkiwH9gUOAVYBLwLuoessOrZhNEmSJEmSNCQWiDSwJHsBhwPHAbtU1VMbR5IkSZIkSUPgDCLNK8medF1DhwLPB24ALgKOAa5oFkySJEmSJA2VBSIt5NvApcCHgK9Ule1mkiRJkiRNILeYaV5JXgYcTDd/aGvgYuBCukHVt7fMJkmSJEmShscCkQaS5Cl0J5odAhwNzFTVs9umkiRJkiRJw+AWMy0qyfZ0s4hW0c0j+lngsoaRJEmSJEnSENlBpHkl+SRdx9BKuqHUa4DVwDeq6sGG0SRJkiRJ0hDZQaSF3AkcD6ytqnWtw0iSJEmSpKVhB5EkSZIkSdKUW946gCRJkiRJktqyQCRJkiRJkjTlLBBJkiRJkiRNOYdUa1FJVgCvA3an72emqn63WShJkiRJkjQ0Fog0iHOArYDLAI+3lyRJkiRpwlgg0iD2qKo9W4eQJEmSJElLwxlEGsRNSZ7aOoQkSZIkSVoadhBpED8Arkzyd8C62UVnEEmSJEmSNBksEGkQ1XtJkiRJkqQJtGxmZqZ1BkmSJEmSJDVkB5EGkuQI4PnAitm1qvpgu0SSJEmSJGlYLBBpUUl+HzgAeA7wZeDlwNeahpIkSZIkSUPjKWYaxJHArwDfr6rfAPYDtmsbSZIkSZIkDYsFIg1iXVU9DMwk2bKqbgWe1TqUJEmSJEkaDreYaRD3JdkaWAucmeQ24IHGmSRJkiRJ0pDYQaRBvBZ4BPgd4DpgBnhV00SSJEmSJGloPOZekiRJkiRpyrnFTItKEuC9wB70/cxU1QubhZIkSZIkSUNjgUiDOAf4HPAZuq1mkiRJkiRpglgg0iAerqqTW4eQJEmSJElLwyHVGsRXkrykdQhJkiRJkrQ07CDSIL4GfDnJo8CDwDJgpqp2aBtLkiRJkiQNgwUiDeJPgTcB/4gziCRJkiRJmjgWiDSIu6rqf7cOIUmSJEmSlsaymZmZ1hk04pK8B7gL+Etg3ex6Vf2oWShJkiRJkjQ0dhBpEB/uvX8SmKE3gwjYolkiSZIkSZI0NHYQSZIkSZIkTTk7iDSwJFvR9zPjFjNJkiRJkiaDBSItKsnRwGnALrjFTJIkSZKkiWOBSIP4GPBq4NKqerR1GEmSJEmSNFwWiDSIu6pqbesQkiRJkiRpaTikWvNKsnXv8gTgHuBsPOZekiRJkqSJYweRFnI/j88cAvgEziCSJEmSJGni2EEkSZIkSZI05ewg0qKSbAPcX1WPJtkb2Bv4P1X1UONokiRJkiRpCJa3DqCxsBp4UpKdgL8D3gT8adtIkiRJkiRpWCwQaRDLquqHwMuAP6uqXwH2a5xJkiRJkiQNiQUiDWJFkicCvwz8Q2/tkYZ5JEmSJEnSEFkg0iDOBm4HVgKX9LaarVv4r0iSJEmSpHFhgUiLqqoPALsBB1bVo8D9wCvappIkSZIkScNigUjzSrKy974XsDPw7N71zwHbtswmSZIkSZKGx2PutZA/phtMfd4cz2bouookSZIkSdKYWzYzM9M6gyRJkiRJkhqyg0jzSrL1Qs+r6kebK4skSZIkSVo6Foi0kPvptpItm+PZDLDF5o0jSZIkSZKWglvMJEmSJEmSppwdRFrUXFvN3F4mSZIkSdLksINIi0ryKN2Wsn4/Bi4H3lJVtflTSZIkSZKkYbGDSIN4L/AA8Gm6eURvBJ4J3AT8CbCqWTJJkiRJkrTJLBBpEK+sqv367k9LclVV7Zfkt5ulkiRJkiRJQ7G8dQCNha2T7DZ7k2Ql8OTe7cNtIkmSJEmSpGGxg0iDeB9weZKr6LaYvQB4a5KnAOc0TSZJkiRJkjaZQ6o1kCQ7Ai/s3V5WVf/WMo8kSZIkSRoeC0QaWJKt6Os686h7SZIkSZImg1vMtKgkRwOnATv3lpbRHXu/RbNQkiRJkiRpaCwQaRAfA14NXFpVj7YOI0mSJEmShssCkQZxV1WtbR1CkiRJkiQtDWcQaVFJ/jtwD3A2sG523RlEkiRJkiRNBjuINIiP9N4/QTd7yBlEkiRJkiRNEDuIJEmSJEmSptzy1gE0upL83ALP9t2cWSRJkiRJ0tKxQKSF/PXsRZLL13t2xmbOIkmSJEmSlogFIi1kWd/1lgs8kyRJkiRJY8wCkRYyM8/1XPeSJEmSJGlMeYqZFrIiyZ503UL91wAr2sWSJEmSJEnDZIFIC9kaOL/vvv/aDiJJkiRJkiaEx9xLkiRJkiRNOWcQSZIkSZIkTTkLRJIkSZIkSVPOApEkSZIkSdKUs0AkSZIkSZI05SwQSZIkSZIkTbn/D8Kdjm7xnHEUAAAAAElFTkSuQmCC\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light", "tags": [] }, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(1, 2, figsize=(20, 10))\n", "most_recommendations = get_recommendations(user_most)\n", "plot_recommendations(most_recommendations, ax=ax[0])\n", "ax[0].set_title('Recommendation for user_most')\n", "least_recommendations = get_recommendations(user_least)\n", "plot_recommendations(least_recommendations, ax=ax[1])\n", "ax[1].set_title('Recommendation for user_least');" ] }, { "cell_type": "markdown", "metadata": { "id": "fIReD8VthTUL" }, "source": [ "We see that there is more variance in our recommendations for `user_least` reflecting our additional uncertainty in their watch preferences.\n", "\n", "We can also see look at the genres of the recommended movies." ] }, { "cell_type": "code", "execution_count": 0, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 482 }, "id": "MEmslkFVgaOP", "outputId": "21bd9181-22be-44b4-b95e-cf5a076fe0e2" }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAABIgAAAJRCAYAAAAuzYweAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdfbhlZ10f/O9khpeYtEGSYJqQBAXmV6URCCKpvKiXbe1bRAUDecgLWisBDLbSGqCKiMKTB8LjAyQxc0mpeUhMJVIIUNsobSnNE0BNGRB4+BGFJEMi5A2QCElwJv1jr1MOcfY5ZzJn5uyZ+/O5rnPts9dv3Wvfe+2T7Hu+615rbbrvvvsCAAAAwLgO2egOAAAAALCxBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABCy0qvqtqvq1fdG2qn6tqm6vqs8/8B4CABx8jMHWbm/2FSySLRvdAdhXquq5Sf5lkr+T5C+TfDbJpUl+o7vv28i+sfGq6oQkL01yYnffutH9OVBU1Q8kuay7H7nRfQFgMRmDsRJjsAemqu5L8tju/tON7gsHLzOIOChV1UuTvDHJ65Mck+TbkpyT5KlJHrwPXm/zem9z2q4Qd985IckdD2RgspGfy776WwOA9WAMxhoYg8GC8j8+DjpVdUSSVyc5q7vfsaz0kSTPW7beQ5K8JslpSR6S5J1J/mV3f21plkSSX09yXpKdSV7R3f9uavtbSb6W5MQk35/kmVX1ySRvTvKMJHcl+fXuftO0/vcmuTjJ1qnd5d3987vp+9LrvjmzI29/UFVnJ/mFJP88ycOS/Jck53T3nVObpyV5XZLvSvKVJL/U3b817Yc3J/lHSb6a5DeTvLa7d1XV86ft/WGSn0xyZ5Izpv796rQ//nV3X7rs/X41ybcneXqSjyZ5VpKXJTk7yReSnN7dH5nWP3aFffGqqa93J/mxJDclObu7/3iqPzHJv03y2CS/l+SbjjRW1T9N8mtJHpXkk9O++Nha2i7bxt9L8p4kD6mqu5L8bnc/v6p+JMn/meS4JNuTvLC7//+pzQ1JfiOzv6GqqsO6+6+WbfNRmR0hfdDS8qp6f2azbd5SVY+Z+vaEJF9P8l+6+znTen972l9PSnLb9Bm+fdm+/6a/tSTvm/O+fit79jl95/SenpDk5iQv7+53T7V/nOSCJMcn+YvM/lv4jST/adl+S5Kt3X3L7voDwFiMwYzB5rVdto2Dcgy2m/e50r56WWZ/A49IsiPJv+nud0613fa1qj4wbfqj00yif9bdv7OWvsCeMIOIg9HfzezL9apV1js/sy/jJyR5TGZfSK9cVj8myRHT8n+W5KKq+tZl9f8js8HN30hybWZfdh+d1v+hJP+iqn54WveNSd7Y3X8zyaOTvH2Ffh2T5OGZfRn9TJJzk/xoZl9Mxyb5YpKLkqSqTszsH+xvTnL09F62T9t589T/75janpXZQGTJU5J8LMmRSX47yb9P8uRpX5yR5MKqOnzZ+qcl+cUkRyW5J8kHk/zP6fnvJvm/pz4dssq+SJIfmV7vYUneneTCqe2Dk7wrydumfXBlZoOgTPUnJnlrkhdM/d6W5N1V9ZDV2i7X3e/LbNB2S3cfPg1Mtia5Ism/mPbl7yV5z7TdJacn+SdJHrZ8YLJGv5rk95N8a5JHZvb5pKoOS/IHmX0Gj0jy3CQXV9V3LWu7/G/tmlVeZ62f04My+5x+f3rdc5NcXlU1beffJnlBd/+NzE4R+K/d/Zf55v12uHAIgGWMwWaMwcYcg2Xa7tx9Na3yZ5mFfUck+ZUkl1XV31qpr939jKn++Gm/CYfYJ8wg4mB0VJLb73dk4drMjpg8JMkPJ/kfmX3xf/eyo0CvzewL4uVTs68nefW0nd+bjnJUkg9N9au6+/+b2p6U5OjufvVU+0xV/WZmXzRXT9t6TFUd1d23L9vG7uxK8svdfc+07XOS/Gx3f256/qokN1XVmZl9ab2vu6+Y2t6R5I5pCuxzkzyhu7+S5CtV9YYkZ2b2D/8k+eyyo3G/k+TfTO/3niS/X1X3ZjZQWRrsvLO7r5vWf2eSF3X3/7us/c9O6z15lX2RJNd09+9Nbd+W2YAgSU5J8qAk/890jYLfrarlR/l+Jsm27v7w9PzSqnrF1O6+Vdqu5jlJ/mN3/8HUrwuS/FyS70vy/mmdN3X3jj3Y5nJfz2zAeez0WS4NMv5pkhuWPoskH6mqdyT5icwGDcmyv7XMjvqtZK2f0ylJDk9yfnfvSvJfq+q9mQ3AXjX197uq6qPd/cXMBsUAsBJjMGOwkcdgS1baV/+9u69ctu7vVNXLk3xvZsHqvL7CfiEg4mB0R5KjqmrL0gClu78vSarqc5nNnDs6ybckue4bEyayKcnyc4vvuN8Riq9m9g/qJcu/pE5McmxVfWnZss2ZDYKS2dGvVyf5VFV9NsmvdPd75/T/tu5e/gV0YpJ3VtWuZct2ZnZO//GZHYW4v6My+6K+cdmyGzM7mrTkC8t+/1qSdPf9lx2+wvrz1l1tXyTJ8jtWfDXJQ2t2TvmxSW7ub76A5fL3cGKSs6vq3GXLHjy1u2+Vtqs5dvn60zTwHfnmffZABybJbIr6ryb5w6r6YpI3dPdbM3tPT7nf/tqS2VG4B/K6a/2cjk2yYwqHliz/G3lWZkcrz6+qjyV5WXd/cA/6AcB4jMGMwXbXdjUHyxhsyUr7KlV1VpKfz+z0s2T2+R21Sl9hvxAQcTD6YGbTb5+Z5B1z1rk9sy/Ux3X3zQ/wdZZ/Ce7I7GjQY3e3Yndfn+T0aervj2d2ZOXI6ZSdlba7tO2fWnb04n+bvjy/dzfbuD3fOALxyWnZCZldZ2ZfW3FfrOLPkxxXVZuWDTJOyDcGYDuSvKa7X3P/hlX1/au0Xc0tSU5atr1NmQ3+lu+zle68svRZfktm1+xJZlPVkyTd/fnMzjdfumbB+6bzyXdkdjTp76+w7X1xx5dbkhxfVYcsC4lOSPLpqb9/lNl1HR6U2ZHJt2e2P9x9BoB5jMGMwYzBVt5XJ2Z2TaofSvLB7t5ZVdszC0nn9rXduYz9REDEQae7v1RVv5LZOcSbMptS+5dJvjvJYdM6u6Ypt79eVT/b3bdW1XFJ/k53Xz134/P9YWZTiM9L8qYk9yb5ziSHdvcfVdUZSa7u7tuWHaXYNWdb93dJktdU1dndfWNVHZ3k+7r7qiSXJ3lFVZ2W5D9kdi7z8d29varePrU7K7PzwX8+s4sO72sr7otV2n4wyV8leUlVXZzk1MwGX/9tqv9mZkfy3je9zrck+YEkH1hD29W8PcnLquqHpu39XGaD3GvX0nj6bG9OckZVbcvswpGPXqpX1U9kNhD4XGana92X2d/AezObpXNmZtcESGbXMbirp4sz7iMfzuzI4S9MU9+fmtk+e3LNzvn/iSTv7e4vV9Vf5Bt/r19IcmRVHdHdX96H/QPgAGMMZgxmDJZk5X112PT6t019+8nMrvW4Wl+T2RjsO5IIi9hnXKSag1J3vy6zL+NfyOx/pl/I7AJx5+UbXzbnZfY/2A9N/wB+X2bntz+Q19uZ2XnMT8jsLgq3J3lLZoOFJPmHST4xnUP/xiTP7e6vrXHzb8zsIoK/X1Vfyezc+adMr3tTkn+c5KWZ3QVje5LHT+3OzWxQ9pnMzl/+7cwumLdPrWFfrNT23syO7j0/s/fznMwGXUv1P87sqMqFmX1p/um07qpt1/DandmFId889fnUJKdO212rf57kX2c2xf5x+eaBzZOTfHj6G3h3kp/r7s9M1yf4B5ldH+CWzKZ+/1+ZXathn5ne16mZXSjy9szu8HJWd39qWuXMJDdM/22ck+nuM1P9isyua/Clmt0tBQCSGINN7YzBBh6DrbKvPpnkDZmFal/IbObU8hlqu+3rVHtVZtcz+tIUTMK623Tffc4WAAAAABiZGUQAAAAAg3MNIoADSFV9IrMLX97fC7r78v3dHwCAERiDMQKnmAEAAAAMbiFnEFXVQzK7QNefJ9m5wd0BANbf5iR/K8kfdfc9G90ZjL8AYBBzx2ALGRBlNjj5HxvdCQBgn3t6Znf5YeMZfwHAOP7aGGxRA6I/T5LLL788xxxzzEb3BQBYZ5///OfzvOc9L5m+81kIxl8AcJBbaQy2qAHRziQ55phj8shHPnKj+wIA7DtOZZqjqh6V5F3LFj0syd/s7odX1dYklyY5MskdSc7q7uundnNrqzD+AoBx/LUxmNvcAwAsoO6+obufsPSTWVj021P5kiQXdffWJBcl2bas6Uo1AIDdEhABACy4qnpwkucleWtVPSLJyUmumMpXJDm5qo5eqba/+wwAHFgERAAAi+9Hktzc3f8zyfHT7zuTZHq8ZVq+Ug0AYC4BEQDA4vupJG/d6E4AAAcvAREAwAKrquOSfH+Sy6dFO5IcV1Wbp/rmJMdOy1eqAQDMJSACAFhsZyf5j919R5J0961Jtic5faqfnuQj3X3bSrX93GcA4ACzqLe5BwBg5vlJXnK/ZeckubSqXpnki0nOWmMNAGC3BEQAAAtsul39/Zd9KslT5qw/twYAMI9TzAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwQwZE935950Z3YZ84WN8XAHDgM05ZDD4HAObZstEd2AgPftDmnPrSqza6G+vuPW945kZ3AQBgtw7W8deBxngRgHmGnEEEAAAAwDcIiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAY3Ja1rFRV70ry7Ul2Jbkrybndvb2qbkhy9/STJOd199VTm1OSbEtyaJIbkpzR3beuZ+cBAAAA2HtrCoiSnN3dX06SqnpmkrcmOXmqPbu7P7585ao6JMllSZ7f3ddU1S8mOT/JT61PtwEAAABYL2s6xWwpHJockdlMopU8Kcnd3X3N9PySJKftefcAAAAA2NfWOoMoVfWWJP8gyaYk/3BZ6fKq2pTkmiSv6O4vJTkhyY1LK3T37VV1SFU9vLvvXJ+uAwAAALAe1nyR6u7+6e4+Ickrkrx+Wvz07n58kidnFhxduP5dBAAAAGBf2uO7mHX325L8YFUd2d07pmX3JLk4yVOn1W5KcuJSm6o6Kskus4cAAAAAFs+qAVFVHV5Vxy97fmqSO5PcXVVHTMs2JXluku3TatclObSqnjY9PyfJlevZcQAAAADWx1quQXRYkiur6rAkOzMLh05N8m1J3lFVm5NsTvLJJC9Kku7eVVVnJtlWVQ/NdJv79e8+AAAAAHtr1YCou7+Q5JQ55Seu0O7aJCc9wH4BAAAAsJ/s8TWIAAAAADi4CIgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAY3JaN7gAAALtXVQ9N8utJ/l6Su5N8sLt/pqq2Jrk0yZFJ7khyVndfP7WZWwMAmMcMIgCAxfW6zIKhrd19UpJfmpZfkuSi7t6a5KIk25a1WakGALBbZhABACygqjo8yVlJHtnd9yVJd3+hqh6R5OQkf39a9YokF1bV0Uk2zat192379Q0AAAcUAREAwGJ6dGaniP1yVf1gkruS/GKSryW5ubt3Jkl376yqW5Icn1lANK8mIAIA5nKKGQDAYtqc5DuSfKS7vyfJeUn+Q5LDN7RXAMBBSUAEALCYbkryV5mdJpbu/nCS2zObQXRcVW1Okunx2CQ7pp95NQCAuQREAAALqLtvT/LfMl1PaLo72SOSfDrJ9iSnT6uentkso9u6+9Z5tf3ZdwDgwCMgAgBYXOckeUVV/UmSf5/kzO7+0rT83Kr6dJJzp+fL28yrAQDslotUAwAsqO7+TJIf2M3yTyV5ypw2c2sAAPOYQQQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4LasZaWqeleSb0+yK8ldSc7t7u1VtTXJpUmOTHJHkrO6+/qpzdwaAAAAAItjrTOIzu7ux3f3E5NckOSt0/JLklzU3VuTXJRk27I2K9UAAAAAWBBrCoi6+8vLnh6RZFdVPSLJyUmumJZfkeTkqjp6pdr6dBsAAACA9bLmaxBV1Vuq6qYkr0lydpLjk9zc3TuTZHq8ZVq+Ug0AAACABbLmgKi7f7q7T0jyiiSv33ddAgAAAGB/2uO7mHX325L8YJLPJTmuqjYnyfR4bJId08+8GgAAAAALZNWAqKoOr6rjlz0/NcmdSW5Nsj3J6VPp9CQf6e7buntubT07DwAAAMDeW8tt7g9LcmVVHZZkZ2bh0KndfV9VnZPk0qp6ZZIvJjlrWbuVagAAAAAsiFUDou7+QpJT5tQ+leQpe1oDAAAAYHHs8TWIAAAAADi4CIgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiJy79d3bnQX9omD9X0BAADAetuy0R1g4z34QZtz6kuv2uhurLv3vOGZG90FAAAAOCCYQQQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIPbstEdAABg96rqhiR3Tz9Jcl53X11VpyTZluTQJDckOaO7b53azK0BAMxjBhEAwGJ7dnc/Yfq5uqoOSXJZkhd399YkH0hyfpKsVAMAWImACADgwPKkJHd39zXT80uSnLaGGgDAXAIiAIDFdnlVfayqLq6qhyU5IcmNS8Xuvj3JIVX18FVqAABzCYgAABbX07v78UmenGRTkgs3uD8AwEFKQAQAsKC6e8f0eE+Si5M8NclNSU5cWqeqjkqyq7vvXKUGADCXgAgAYAFV1WFVdcT0+6Ykz02yPcl1SQ6tqqdNq56T5Mrp95VqAABzuc09AMBi+rYk76iqzUk2J/lkkhd1966qOjPJtqp6aKZb2SfJSjUAgJUIiAAAFlB3fybJE+fUrk1y0p7WAADmcYoZAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxuy2orVNWRSd6W5NFJ7k1yfZIXdPdtVXVfkj9Jsmta/czu/pOp3alJXj+9xnVJfrK7v7r+bwEAAACAvbGWGUT3JXldd1d3n5Tkz5Kcv6z+fd39hOlnKRw6PMlvJjm1ux+T5CtJ/tU69x0AAACAdbBqQNTdd3b3+5ct+lCSE1dp9o+S/HF3Xz89vyTJcx5QDwEAAADYp1Y9xWy5qjokyQuTvHvZ4vdX1ZYk/ynJq7r7niQnJLlx2To3JTl+L/sKAAAAwD6wpxepfnOSu5JcOD0/obu/J8kzknxXkl9ax74BAAAAsB+sOSCqqguSPDbJc7p7V5J0947p8S+SvCXJU6fVb8o3n4Z2QpId69FhAAAAANbXmgKiqnptkicl+dHpFLJU1bdW1aHT71uSPDvJ9qnJf07y5Kp67PT8nCRvX8+OAwAAALA+Vg2IqupxSV6e5Ngk11bV9qp6Z5K/neTDVfXRJB9L8vVMp5h191eS/EyS91bVnyY5IskF++YtAAAAALA3Vr1IdXd/IsmmOeXvXqHdVUmueoD9AgAAAGA/2dOLVAMAAABwkBEQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExDBMvd+fedGd2GfOZjfGwAAAHtny0Z3ABbJgx+0Oae+9KqN7sY+8Z43PHOjuwAAAMCCMoMIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcFs2ugMAAKysqn45yauSnNTdH6+qU5JsS3JokhuSnNHdt07rzq0BAMxjBhEAwAKrqpOTnJLkxun5IUkuS/Li7t6a5ANJzl+tBgCwEgERAMCCqqqHJLkoyQuXLX5Skru7+5rp+SVJTltDDQBgLgERAMDienWSy7r7hmXLTsg0myhJuvv2JIdU1cNXqQEAzCUgAgBYQFX1d5N8T5KLN7ovAMDBT0AEALCYvj/Jdyb5bFXdkOSRSa5O8pgkJy6tVFVHJdnV3XcmuWmFGgDAXAIiAIAF1N3nd/ex3f2o7n5Uks8l+eEkr09yaFU9bVr1nCRXTr9ft0INAGAuAREAwAGku3clOTPJb1TV9ZnNNHrZajUAgJVs2egOAACwumkW0dLv1yY5ac56c2sAAPOYQQQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADG7LaitU1ZFJ3pbk0UnuTXJ9khd0921VdUqSbUkOTXJDkjO6+9ap3dwaAAAAAItjLTOI7kvyuu6u7j4pyZ8lOb+qDklyWZIXd/fWJB9Icn6SrFQDAAAAYLGsGhB1953d/f5liz6U5MQkT0pyd3dfMy2/JMlp0+8r1QAAAABYIHt0DaJpZtALk7w7yQlJblyqdfftSQ6pqoevUgMAAABggezpRarfnOSuJBfug74AAAAAsAHWHBBV1QVJHpvkOd29K8lNmZ1qtlQ/Ksmu7r5zlRoAAAAAC2RNAVFVvTaz6wr9aHffMy2+LsmhVfW06fk5Sa5cQw0AAACABbKW29w/LsnLk3w6ybVVlSSf7e4fq6ozk2yrqodmupV9knT3rnk1AAAAABbLqgFRd38iyaY5tWuTnLSnNQAAAAAWx55epBoAAACAg4yACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAYnIAIAAAAYnIAIAAAAYHBbNroDAADsXlW9K8m3J9mV5K4k53b39qramuTSJEcmuSPJWd19/dRmbg0AYB4ziAAAFtfZ3f347n5ikguSvHVafkmSi7p7a5KLkmxb1malGgDAbgmIAAAWVHd/ednTI5LsqqpHJDk5yRXT8iuSnFxVR69U2199BgAOTAIiAIAFVlVvqaqbkrwmydlJjk9yc3fvTJLp8ZZp+Uo1AIC5BEQAAAusu3+6u09I8ookr9/o/gAABycBEQDAAaC735bkB5N8LslxVbU5SabHY5PsmH7m1QAA5hIQAQAsoKo6vKqOX/b81CR3Jrk1yfYkp0+l05N8pLtv6+65tf3XcwDgQOQ29wAAi+mwJFdW1WFJdmYWDp3a3fdV1TlJLq2qVyb5YpKzlrVbqQYAsFsCIgCABdTdX0hyypzap5I8ZU9rAADzOMUMAAAAYHACIgAAAIDBCYgAAAAABicgAgAAABicgAgAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGJyACAAAAGBwAiIAAACAwQmIAAAAAAa3ZS0rVdUFSZ6V5FFJTuruj0/Lb0hy9/STJOd199VT7ZQk25IcmuSGJGd0963r13UAAAAA1sNaZxC9K8kzkty4m9qzu/sJ089SOHRIksuSvLi7tyb5QJLz16PDAAAAAKyvNQVE3X1Nd+/Yg+0+Kcnd3X3N9PySJKftaecAAAAA2PfW4xpEl1fVx6rq4qp62LTshCybbdTdtyc5pKoevg6vBwAAAMA62tuA6Ond/fgkT06yKcmFe98lAAAAAPanvQqIlk476+57klyc5KlT6aYkJy6tV1VHJdnV3XfuzesBAAAAsP4ecEBUVYdV1VXvD4IAABnTSURBVBHT75uSPDfJ9ql8XZJDq+pp0/Nzkly5Nx0FAAAAYN9Y623u35Tkx5Mck+R9VXVHklOTvKOqNifZnOSTSV6UJN29q6rOTLKtqh6a6Tb36999AAAAAPbWmgKi7n5JkpfspvTEFdpcm+SkB9gvAAAAAPaT9biLGQAAAAAHMAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxuy0Z3AACAv66qjkzytiSPTnJvkuuTvKC7b6uqU5JsS3JokhuSnNHdt07t5tYAAOYxgwgAYDHdl+R13V3dfVKSP0tyflUdkuSyJC/u7q1JPpDk/CRZqQYAsBIBEQDAAuruO7v7/csWfSjJiUmelOTu7r5mWn5JktOm31eqAQDMJSACAFhw08ygFyZ5d5ITkty4VOvu25McUlUPX6UGADCXgAgAYPG9OcldSS7c6I4AAAcnAREAwAKrqguSPDbJc7p7V5KbMjvVbKl+VJJd3X3nKjUAgLkERAAAC6qqXpvZdYV+tLvvmRZfl+TQqnra9PycJFeuoQYAMJfb3AMALKCqelySlyf5dJJrqypJPtvdP1ZVZybZVlUPzXQr+yTp7l3zagAAKxEQAQAsoO7+RJJNc2rXJjlpT2sAAPM4xQwAAABgcAIiAAAAgMEJiAAAAAAGJyACAAAAGNyqF6muqguSPCvJo5Kc1N0fn5ZvTXJpkiOT3JHkrO6+frUaAAAAAItlLTOI3pXkGUluvN/yS5Jc1N1bk1yUZNsaawAAAAAskFUDou6+prt3LF9WVY9IcnKSK6ZFVyQ5uaqOXqm2ft0GAAAAYL080GsQHZ/k5u7emSTT4y3T8pVqAAAAACwYF6kGAAAAGNwDDYh2JDmuqjYnyfR47LR8pRoAAAAAC+YBBUTdfWuS7UlOnxadnuQj3X3bSrW97SwAAAAA62/VgKiq3lRVn0vyyCTvq6pPTKVzkpxbVZ9Ocu70PGuoAQAAALBAtqy2Qne/JMlLdrP8U0meMqfN3BoAAAAAi8VFqgEAAAAGJyACAAAA/rd7v75zo7swvI34DFY9xQwAAAAYx4MftDmnvvSqje7G0N7zhmfu99c0gwgAAABgcAIiAAAAgMEJiAAAAAAGJyAC5jpYL053sL4vAACAB8pFqoG5DtaL023EBd8AAAAWmRlEAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAa3Lv13dudBeG5zNgX9my0R0AAADgwPDgB23OqS+9aqO7MbT3vOGZG90FDlJmEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMLgtG90BAAD+uqq6IMmzkjwqyUnd/fFp+dYklyY5MskdSc7q7utXqwEArMQMIgCAxfSuJM9IcuP9ll+S5KLu3prkoiTb1lgDAJhLQAQAsIC6+5ru3rF8WVU9IsnJSa6YFl2R5OSqOnql2v7qMwBw4BIQAQAcOI5PcnN370yS6fGWaflKNQCAFQmIAAAAAAYnIAIAOHDsSHJcVW1Okunx2Gn5SjUAgBXt9V3MquqGJHdPP0lyXndfXVWnZHZhxEOT3JDkjO6+dW9fDwBgVN19a1VtT3J6ksumx490921JslINAGAl63Wb+2cv3Xo1SarqkMwGJs/v7muq6heTnJ/kp9bp9QAADmpV9aYkP57kmCTvq6o7uvtxSc5JcmlVvTLJF5OctazZSjUAgLnWKyC6vyclubu7r5meX5LZLCIBEQDAGnT3S5K8ZDfLP5XkKXPazK0BAKxkva5BdHlVfayqLq6qhyU5IcmNS8Xuvj3JIVX18HV6PQAAAADWyXoERE/v7scneXKSTUkuXIdtAgAAALCf7HVA1N07psd7klyc5KlJbkpy4tI6VXVUkl3dfefevh4AAAAA62uvAqKqOqyqjph+35TkuUm2J7kuyaFV9bRp1XOSXLk3rwUAAADAvrG3F6n+tiTvqKrNSTYn+WSSF3X3rqo6M8m2qnpoptvc7+VrAQAAALAP7FVA1N2fSfLEObVrk5y0N9sHAAAAYN9br7uYAQAAAHCAEhABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEAAAAMTkAEAAAAMDgBEQAAAMDgBEQAAAAAgxMQAQAAAAxOQAQAAAAwOAERAAAAwOAERAAAAACDExABAAAADE5ABAAAADA4AREAAADA4AREAAAAAIMTEAEA/6u9e4+yq6oPOP6FPCqKCOJbaBGFH5DwJqA8KipqtWK0oAtiFERE1FofwZbWxwKrFipUGowLIghBC6LUglgqPioCQsIbAuiv+AYhUlGsCjEhpn/sPcnNZGbuncm9cydzvp+1sjL3nnvO2fuce/be57f3PleSJEkNZ4BIkiRJkiSp4QwQSZIkSZIkNZwBIkmSJEmSpIYzQCRJkiRJktRwBogkSZIkSZIazgCRJEmSJElSwxkgkiSNyYqVq/qdhJ6YrPmSJEmSRjK13wmQJG2cpk+bwqHzLut3Mrru8tNn9zsJkiRJ0rhzBJEkSZIkSVLDGSCSJEmSJElqOANEkiRJkiRJDWeASJIkSZIkqeEMEEmSJEmSJDWcASJJkiRJkqSGM0AkSZIkSZLUcAaIJEmSJHXFipWr+p2ExvMcSBqrqf1OgCRJkqTJYfq0KRw677J+J6PRLj99dr+TIGkj5QgiSZIkSZKkhjNAJEmSJEmS1HAGiCRJkiRJkhrOAJEkSZIkSVLDGSCSJEmSJElqOANEkiRJkiRJDWeASJIkSZIkqeEMEEmSJEmSJDWcASJJkiRJkqSGM0AkSZIkSZLUcAaIJEmSJEmSGs4AkSRJkiRJUsMZIJIkSZIkSWo4A0SSJEmSJEkNZ4BIkiRJkiSp4ab2cuMRsSOwCNgaeAh4U2be08t9SpLUDytWrmL6tCn9TkbXTdZ8TXa2wSRJ0mj1NEAEnAUsyMzPR8Rc4GzgxT3epyRJ4276tCkcOu+yfiej6y4/fXa/k6CxsQ0mSZJGpWcBooh4GrAX8NL61kXApyLiqZn5v21WnwKwbNmyXiWPlY/8qmfb7pf77rtvzOt6PNaajMcCPB6txnwsHvsj06ZOvpm5G5Ivvx/r8nh0rqWOd3hSl21AG8z2VwP06ppu5XnuL8/x5Dce5xg8z/3WjzbYJqtXr+7JTiNib+CCzJzR8t7dwNzMvKXNugcC1/QkYZIkaSI5KDOv7XciJpOxtsFsf0mS1CjrtcF6PcVsrG4EDgIeAFb1OS2SJKn7pgDPpNT5mhhsf0mSNPkN2wbr5QiipwH/A2ydmasiYgrlIYk7dDDFTJIkSWNgG0ySJI1Fzx6mkZkPArcBR9a3jgRutWEiSZLUO7bBJEnSWPRsBBFAROxE+YnVrYBfU35iNXu2Q0mSJNkGkyRJo9bTAJEkSZIkSZImvsn3e82SJEmSJEkaFQNEkiRJkiRJDWeASJIkSZIkqeEMEEmSJEmSJDXc1H4nYCKJiJ8AfwJsk5mr6ntHA+cB78rMT41ye9sBL8vMhV1NaJ9FxDTgQ8ARwHJgFfDfwImZuXLQZ88BFmXmNeOe0C4aTZ7HIS2rgSdm5u/Gc7+diIitgPuBhZn57jaffQ1wf2beUF/vA7w3M9/Q+5T2RkQsoZQh04EdgTvroicBD2XmPh1u5yPAXZl5cUScBGyemSfU8uhVmXl41xPfI7VcXQ78AXgCcBdwamZe18dkjYuI+CiwdWa+vb5+FXA5MDMz76rvfRX4j8w8dxTb3RI4LjP/uQfJliYN6+7JpZM2RkRcBZyWmV/twf73AHbMzC92e9taazRtyfr544HNMvOTXUzDe4ALM/PBXu1DnWn6vUU/OIJoffcDL295fTRwyxi3tR1w3FhWjIiJHLw7D5gB7J2ZuwGzgKTcGK8REVMy89iNPThUdZRnMQdYDBwZEdPbfPY1wL4DLzLzpo29AM/M/TJzD+CVwMOZuUd9/eZOt1Gvmw9n5sXdSldETOnWtsbo8MzcPTOfR/nZ7SsiYr/WD0TEphGxSX+S1zPfBg5uef1CYMnAe/W8HAhcNcrtbgn87VgSNMHrFqnbrLsnl9G0MXphD+D1Y1nRsndURnWeM/OsHgRu3gM8rcf7UGcafW/RDxZW6zufEhS6IiK2p/R4LwWIiAeAvTLzgfp6PrAMOINy0zMDWAlkZr4eWAA8JyJuA36QmYdHRNTPP4UyyuCMzDyvbm81cDLwl8DXIuK1wJsz88a6/H3ATpk5pqBTN0TEDsBrKaOsfguQmY8BCyPi6IiYC/wW2AGYGxFnUHtyIuJ8yiiCHYDnAl+m9KafDGwLfDIz/7XuZ9jjNN7a5HlKRJwG/EX9+NeAv8vMVd3Kb0T8FfBxSu/nv7ek6/3Adpn5zvr66cAdwHMy85EeHY52jqHcuP49MBv4UkQ8G5hPOQ4AF1GCrq8GDomIY4F/AX5G+a7sAxARbwLeD6wGfgi8LTMfrKNo5gC/BmYCDwOHZeayccnh2E2NiLOBF1DydERmfi8iDqYcn5uBPYEPRsThwE3tRi1GxFHAOyhl+W+At2dm1mO0zrUI3NaTXI1SZn45IvYFToiIuyjl5pOAPwVeEBEfoARSpgO/BI7JzJ/WEZk3AZ+hXG+bAW8Ajgf2Ax4FZmfmsojYFfg0pfx+HKXX6YxxzOaA6yh1wNMz8xeUfJ1MqWMWUM73/wGPRMQllGOwGXBRZn48IjYFPgW8mFKW/C4zD6jrblnrlkcyc/+IeCZw5uBtwJpRXF+o21kaEfcCQTnu21Our9f1sdyQus66e1Iaqo2xCyUQuDmlvf44gIg4EDgzM/ccWDkibgLmZeZ32tSf67UxKO37jwBb1LL3akrb5abMfErd/nYDr1vqrPMpZe/CiLiMYcpprWOo83wSw9RbQ4y0nkM5b7sBPwfeBZwGPA+4EZibmasjYg7wbsq1C3BCZn6rtkOeBVwSEcvr9l7fso8pwKkMX34sp4wi3xa4HjgqM1d3/zA1hvcW48wRROu7Cti1Dmc7CrigZdki6oigiNicMmT5HMqIoy0yc5fM3B14W/38O4G76yiCw2vvwYWUoW6zKD3HJ0bETi37eDQzZ2Xmhyg3Bu+o+9sEeDvlxqCf9gTuycxfD7P8+ZQCdmZmDnVDOgN4BbAz5eZuLuWm6QDgYxGxeYfHaTyNlOfjKD1Ke9V/e7LuqLENym9tOH6GcuO7B6XROuBc4LD6XRxIy4X9amBGxG7A1pTh++dRCnSAzwOLM3O32oP7mcy8EvgKcEq9Pi4YtK2ZwCmUKZq7UaZqndnykVmU79kM4G5K5T/RzQDOqvn5IvDBQcsW1mPR0bD4iDiI0mD588zcG/gE8NmWj7S7FvtpCSXPUII7czJzp3qNnVLLwN0pFf6pLettDVxbG/znAt8CFtRjejPw1/VzPwEOycy9KD1Jx0XEzr3O1GCZ+ShwA3BwRDyRErD6GqXMgDKS6CpKPTM/M/cF9gZeEREvBXYHXgQM1C2vquu9k7Uj1Pav7w23jQFbZOa+mfmW+nofSmNoZ2AapXySJhPr7klkhDbG54BP1/bAGZT2AZl5LbB5XY/acbAVcHUH9ed6bYzMfAj4MPDNWvb+TQfJ3hq4MTP3ysyzaF9ON94I5xk6r7dmAe/LzJ0onUcX1vV2AXYFXlI/dyXw/NqmOIJyn0dmfowyo+Tweq7vHrT9duXHTMpI8hmU83xI50dArby36A8DROtbTbl5O6L+u7Bl2QLgzbVRMBf4epa5qbcDO0fEgoh4Hes2BFrtSCnUvlB7H66hDHNuvXFZ1PL354CXR8STKUGoX2Tm7RuawR67NjN/OMLySzPzD7UhlMAVmfnHzPw5JWq7DZ0dp4niEOD8zFyRmSsohVdrRbCh+d0PuCUzs25vzfOsMvNXlILwjfU7+VbKqIl+eQtwQe0l+TKwX0T8GbA/sGZYbmb+soNtvYhyrB6or89m3eP63cy8t/69mNLLO9FlZt5a/x6c5nsy8/pRbu9QSgBhSf3enELprRrQ7lrsp9apZFcM+k68IiIWR8SdwAmsDaZAGUHzn/XvW4D7WoJfN1N6BwEeD5wbEUuB71J6AnfvdiY6dBUlEHQg5ZysAu6JiBn1/YEpZ/PrebyBkt6dgR9RGsHnRsQbh9tBRDxhhG0MuGDQaldm5sP1el3CxnENSd1i3b3xGa6NMZPSXiYzF1NH/VeLKCM2qf8vquu3qz+71cZYTrmn6LSc1tDn+dl1Waf11ncz8776962UuvfhLCMIb2dtW+G5wJV1NPPFwDMi4hkdpLGT8mN5XXbLCOlUe95b9IFTzIa2iFLwXJ2ZD5URxJCZ99bhqbMpPbjH1fd/VBv7L6H0OH289lQMtgnwy9qbNJw1Dy/MzN9HxIWU55ccTP9HD0EpaHeIiK2G6ZVr9/DF5S1/rxri9VRKkK7dcRpP7fI8kg3Kb0S8us32zwT+DXgQ+F5m3jPK9HVFlDnBc4A/1OGbUG5sj+7RLoc6jhPdSGkey0NLNwE+m5kfHmb5RH4Q6izWPsB7TTprpf9JYFZm/jgi9mfdIH1r8H246wnKtI5lwNGZ+VhEfJ067aAPvk25+fsN8J363tWU+uJAypDp1ZQ8r/fQ3JZA0iHAqRGx1xD72HSkbVSDvw+Dj91mnWRG2ohYd08SG9DGuABYHBH/ABxJmeIN7evPTtsYj7FuZ/vgeub3uXZqUSfldKN1cJ47rbc6uX6hjFKel5mX1indj9CdtsLG2EadcLy36B9HEA0hM38EfAD4xyEWn0kZwrpyoMc/IrYBVmXmpcB7gacCT6Y8W+JJrZumPGtiTU9wHYq8xQjJWUB5UNretMxh75faiPkKcHadMkGdy38sZf53V3bD6I9Tz7TJ81XAURExLcqvpRwFfGO0u2D4/C4G9ozyLAWAYwelbSnwEOU72c8A4uySnNwmM7fLzO2Al1FG2l1HuS4AiIin1D8HXx+tvg28sqUn562M/rhOdpcDb6rlz8B3cu8+p6mtiJhNmS57+hCLtwBWAMtqY+34Me5mS+DeGhyaCRw0xu10w/WUHyw4jLUPo76aMh3u4XoNXwOcOLBCRGwbEc+IiKcCj6/Dpk+kBJm2p1w7j6+jD8jyfJUht9HbrEkTl3X3pDJSG2Mp5SaSKM+3W9NBm5k/o0wVmU955MNP66Kx1p+D2y3LgGkRMTAiZc5wK1pOd2S483x0j/a3JfDj+vcxrPvw+pHaqN9kw8sPtee9RZ8YIBpGZi7MIZ7bkZnfoUQYW4cD7wpcHxG3U4aM/lNm3k956GBGxJ0RcUkd2ngocERE3FGHNH6atQ9HGyodPwa+T+npWNGt/G2go4B7gJvrNJClwE4MP7VuVMZynMbBcHleSDnPt9Z/d1CeO9CxkfJbpzAeB1weEbcydM/GOcAfga7/pOsoHEPpDV2jBlA3BU4CDqjXwe2U4aJQhoTPiYjbWnoGBta9k9KI+kZE3EEZCt72p06bJDOvpgSyv1KP652UynQiuiQibo+IH1DO/yszc8ngD9Wbpi9RGvRLWNtwG62PAm+t352TKAGZvsjM5ZS8UOsFKA/JfDZrA0ZvAHaJiKV1WtzFlIbrtsA36/m9A/gvypz7X1Gut6URcV2bbUhNZt09OYzUxpgHvKue3/dSytdW51NuBM9vWXes9ee3gCfU+mx+/Q68m9JWuYEy6mAkltMjG+k8v7AH+3sPcGlE3ELpfHmoZdl84LzaRt1l0HobXH6oI95b9Mkmq1f7UPXRiIjnUJ5p8bwchwcK1p6o71OGpP681/vTxicizqFE2D/R77RIkqT2rLslSRNRY+fWjUVEfIQSzZw3TsGh4ym/dHS6wSENFhHPogyXXAZ08msakiSpj6y7JUkTmSOIJEmSJEmSGs5nEEmSJEmSJDWcASJJkiRJkqSGM0AkSZIkSZLUcAaIJEmSJEmSGs4AkSRJkiRJUsMZIJIkSZIkSWq4/webLACIkugXFgAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light", "tags": [] }, "output_type": "display_data" } ], "source": [ "most_genres = collections.Counter([movie_id_to_genre[i] for i in most_recommendations])\n", "least_genres = collections.Counter([movie_id_to_genre[i] for i in least_recommendations])\n", "fig, ax = plt.subplots(1, 2, figsize=(20, 10))\n", "ax[0].bar(most_genres.keys(), most_genres.values())\n", "ax[0].set_title('Genres recommended for user_most')\n", "ax[1].bar(least_genres.keys(), least_genres.values())\n", "ax[1].set_title('Genres recommended for user_least');" ] }, { "cell_type": "markdown", "metadata": { "id": "ZwCsrFFgnALF" }, "source": [ "`user_most` has seen a lot of movies and has been recommended more niche genres like mystery and crime whereas `user_least` has not watched many movies and was recommended more mainstream movies, which skew comedy and action." ] } ], "metadata": { "accelerator": "TPU", "colab": { "collapsed_sections": [], "name": "Distributed Inference with JAX", "toc_visible": true }, "kernelspec": { "display_name": "Python 3", "name": "python3" }, "widgets": { "application/vnd.jupyter.widget-state+json": { "0033084241f445c9a6815a48a868d64b": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "0509d5f63e8447c5bf3047dec3c9b436": { "model_module": "@jupyter-widgets/controls", "model_name": "HBoxModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HBoxModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HBoxView", "box_style": "", "children": [ "IPY_MODEL_74764971a48a4f79a495f1132d805db0", "IPY_MODEL_30b99f5c7029409ebca66ffedb065879" ], "layout": "IPY_MODEL_62ff554eba004da3809a1dd2e21311fb" } }, "10eb91732d5b430481b75769cfdf6100": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "24c536ef29c547ef965bb3270637bcc2": { "model_module": "@jupyter-widgets/controls", "model_name": "HBoxModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HBoxModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HBoxView", "box_style": "", "children": [ "IPY_MODEL_8074ff27fdd24ba6844f3a3f4eda7aed", "IPY_MODEL_a1966d13e72d4a9f9fcda67545e8fee3" ], "layout": "IPY_MODEL_33822c1d39ca42f5b5a44f330ba19df3" } }, "30b99f5c7029409ebca66ffedb065879": { "model_module": "@jupyter-widgets/controls", "model_name": "HTMLModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HTMLModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HTMLView", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_0033084241f445c9a6815a48a868d64b", "placeholder": "​", "style": "IPY_MODEL_a03a5d0680884fa3a77c4b95ebfa0b98", "value": " 1/1 [00:01<00:00, 1.17s/ url]" } }, "33822c1d39ca42f5b5a44f330ba19df3": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "3c0c2c21823b4206b1e217601b8b2533": { "model_module": "@jupyter-widgets/controls", "model_name": "DescriptionStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "DescriptionStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "description_width": "" } }, "3ee6fe19d17d4df99ffeb289775b1c2e": { "model_module": "@jupyter-widgets/controls", "model_name": "HTMLModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HTMLModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HTMLView", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_c2e9d94dd4cd459da8cd6c5d253b7ed5", "placeholder": "​", "style": "IPY_MODEL_3c0c2c21823b4206b1e217601b8b2533", "value": " 1000209/0 [20:19<00:00, 751.68 examples/s]" } }, "445cf583fac44e3895e51e2cde2a350c": { "model_module": "@jupyter-widgets/controls", "model_name": "HBoxModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HBoxModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HBoxView", "box_style": "", "children": [ "IPY_MODEL_dd3fe337112740738f3b9fd788ba9405", "IPY_MODEL_3ee6fe19d17d4df99ffeb289775b1c2e" ], "layout": "IPY_MODEL_cd2ab853d5d143daae3516a910d2cd87" } }, "48ec20644a394848927cf72e8bea86b5": { "model_module": "@jupyter-widgets/controls", "model_name": "HBoxModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HBoxModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HBoxView", "box_style": "", "children": [ "IPY_MODEL_55840316f537476b81ffe69cec8b7092", "IPY_MODEL_a250abd4702646e8a30b3ff946418bc7" ], "layout": "IPY_MODEL_4fc929851e25400eaa789a41aead047b" } }, "4da2d87727e942c9956a2c507109c107": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "4fc929851e25400eaa789a41aead047b": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "55840316f537476b81ffe69cec8b7092": { "model_module": "@jupyter-widgets/controls", "model_name": "FloatProgressModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "FloatProgressModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "ProgressView", "bar_style": "danger", "description": "100%", "description_tooltip": null, "layout": "IPY_MODEL_5f993f93d9764230b0fd70f0945a3739", "max": 1000209, "min": 0, "orientation": "horizontal", "style": "IPY_MODEL_ef4b4c0fa6be4907a9e86c77eb4249f6", "value": 999347 } }, "56b9e4a682864227a14f95274afea6c8": { "model_module": "@jupyter-widgets/controls", "model_name": "FloatProgressModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "FloatProgressModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "ProgressView", "bar_style": "success", "description": "Extraction completed...: 100%", "description_tooltip": null, "layout": "IPY_MODEL_57ffaf62ae774f0cbcc3526d12902a76", "max": 1, "min": 0, "orientation": "horizontal", "style": "IPY_MODEL_e177f0e205af4369ad4e16af5c5f5f6a", "value": 1 } }, "57ffaf62ae774f0cbcc3526d12902a76": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "5f993f93d9764230b0fd70f0945a3739": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "62ff554eba004da3809a1dd2e21311fb": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "698a2d7c08864520af367c353735814f": { "model_module": "@jupyter-widgets/controls", "model_name": "ProgressStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "ProgressStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "bar_color": null, "description_width": "initial" } }, "6bd23dc4d2f04752a269bf7f3a25e10b": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "7154891fb6ac45c4a24e151af3fb6903": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "74764971a48a4f79a495f1132d805db0": { "model_module": "@jupyter-widgets/controls", "model_name": "FloatProgressModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "FloatProgressModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "ProgressView", "bar_style": "success", "description": "Dl Completed...: 100%", "description_tooltip": null, "layout": "IPY_MODEL_cf5ca2562e75411c9e14dcb5d58736ed", "max": 1, "min": 0, "orientation": "horizontal", "style": "IPY_MODEL_76f980ecbd2b4b49a7c44b683f74e8c8", "value": 1 } }, "76f980ecbd2b4b49a7c44b683f74e8c8": { "model_module": "@jupyter-widgets/controls", "model_name": "ProgressStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "ProgressStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "bar_color": null, "description_width": "initial" } }, "8074ff27fdd24ba6844f3a3f4eda7aed": { "model_module": "@jupyter-widgets/controls", "model_name": "FloatProgressModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "FloatProgressModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "ProgressView", "bar_style": "success", "description": "Dl Size...: 100%", "description_tooltip": null, "layout": "IPY_MODEL_6bd23dc4d2f04752a269bf7f3a25e10b", "max": 1, "min": 0, "orientation": "horizontal", "style": "IPY_MODEL_f7503cf63443454a83edfeaa1bc3ca59", "value": 1 } }, "912901341ef44cf385d366dd62d64016": { "model_module": "@jupyter-widgets/controls", "model_name": "HTMLModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HTMLModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HTMLView", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_b4dbdb532ffd4b46ba6ebac9f1366b14", "placeholder": "​", "style": "IPY_MODEL_c400310f418546909ba8ffe2e10e3e12", "value": " 1/1 [00:01<00:00, 1.04s/ file]" } }, "95e6e8cbf69f495eb7fb6c84f2781d12": { "model_module": "@jupyter-widgets/controls", "model_name": "HBoxModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HBoxModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HBoxView", "box_style": "", "children": [ "IPY_MODEL_56b9e4a682864227a14f95274afea6c8", "IPY_MODEL_912901341ef44cf385d366dd62d64016" ], "layout": "IPY_MODEL_cd2ca5c8fc214489a02e62e7750df78d" } }, "9a546ada258846cbad21774eea3af7cc": { "model_module": "@jupyter-widgets/controls", "model_name": "HTMLModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HTMLModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HTMLView", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_b14fcb50f7644de7b40c89486ab999ba", "placeholder": "​", "style": "IPY_MODEL_bbfbc2dbad584046835b3fcac9917d57", "value": " 4/4 [00:01<00:00, 2.08 file/s]" } }, "a019862f72874e7cbc77c57cc1e31f9c": { "model_module": "@jupyter-widgets/controls", "model_name": "ProgressStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "ProgressStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "bar_color": null, "description_width": "initial" } }, "a03a5d0680884fa3a77c4b95ebfa0b98": { "model_module": "@jupyter-widgets/controls", "model_name": "DescriptionStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "DescriptionStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "description_width": "" } }, "a1966d13e72d4a9f9fcda67545e8fee3": { "model_module": "@jupyter-widgets/controls", "model_name": "HTMLModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HTMLModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HTMLView", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_10eb91732d5b430481b75769cfdf6100", "placeholder": "​", "style": "IPY_MODEL_ee5454904d4649e1a23c7b29eee96b27", "value": " 5/5 [00:01<00:00, 4.53 MiB/s]" } }, "a250abd4702646e8a30b3ff946418bc7": { "model_module": "@jupyter-widgets/controls", "model_name": "HTMLModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HTMLModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HTMLView", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_e3f937fc3dfb4486bfd7211b52877a80", "placeholder": "​", "style": "IPY_MODEL_bc2e9dd37de44b5192881851fc42099d", "value": " 999347/1000209 [00:09<00:00, 188965.53 examples/s]" } }, "a302adcb8d824d00be0125331fe7df1a": { "model_module": "@jupyter-widgets/controls", "model_name": "HBoxModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "HBoxModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "HBoxView", "box_style": "", "children": [ "IPY_MODEL_eb1aba62376345729c10817275841ad1", "IPY_MODEL_9a546ada258846cbad21774eea3af7cc" ], "layout": "IPY_MODEL_4da2d87727e942c9956a2c507109c107" } }, "b14fcb50f7644de7b40c89486ab999ba": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "b4dbdb532ffd4b46ba6ebac9f1366b14": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "bbfbc2dbad584046835b3fcac9917d57": { "model_module": "@jupyter-widgets/controls", "model_name": "DescriptionStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "DescriptionStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "description_width": "" } }, "bc2e9dd37de44b5192881851fc42099d": { "model_module": "@jupyter-widgets/controls", "model_name": "DescriptionStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "DescriptionStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "description_width": "" } }, "c2e9d94dd4cd459da8cd6c5d253b7ed5": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "c400310f418546909ba8ffe2e10e3e12": { "model_module": "@jupyter-widgets/controls", "model_name": "DescriptionStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "DescriptionStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "description_width": "" } }, "cd2ab853d5d143daae3516a910d2cd87": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "cd2ca5c8fc214489a02e62e7750df78d": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "cf5ca2562e75411c9e14dcb5d58736ed": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "dd3fe337112740738f3b9fd788ba9405": { "model_module": "@jupyter-widgets/controls", "model_name": "FloatProgressModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "FloatProgressModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "ProgressView", "bar_style": "info", "description": "", "description_tooltip": null, "layout": "IPY_MODEL_7154891fb6ac45c4a24e151af3fb6903", "max": 1, "min": 0, "orientation": "horizontal", "style": "IPY_MODEL_698a2d7c08864520af367c353735814f", "value": 1 } }, "e177f0e205af4369ad4e16af5c5f5f6a": { "model_module": "@jupyter-widgets/controls", "model_name": "ProgressStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "ProgressStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "bar_color": null, "description_width": "initial" } }, "e3f937fc3dfb4486bfd7211b52877a80": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "eb1aba62376345729c10817275841ad1": { "model_module": "@jupyter-widgets/controls", "model_name": "FloatProgressModel", "state": { "_dom_classes": [], "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "FloatProgressModel", "_view_count": null, "_view_module": "@jupyter-widgets/controls", "_view_module_version": "1.5.0", "_view_name": "ProgressView", "bar_style": "success", "description": "Dl Completed...: 100%", "description_tooltip": null, "layout": "IPY_MODEL_f48b175a5b8e469da891b76be737e8cd", "max": 4, "min": 0, "orientation": "horizontal", "style": "IPY_MODEL_a019862f72874e7cbc77c57cc1e31f9c", "value": 4 } }, "ee5454904d4649e1a23c7b29eee96b27": { "model_module": "@jupyter-widgets/controls", "model_name": "DescriptionStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "DescriptionStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "description_width": "" } }, "ef4b4c0fa6be4907a9e86c77eb4249f6": { "model_module": "@jupyter-widgets/controls", "model_name": "ProgressStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "ProgressStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "bar_color": null, "description_width": "initial" } }, "f48b175a5b8e469da891b76be737e8cd": { "model_module": "@jupyter-widgets/base", "model_name": "LayoutModel", "state": { "_model_module": "@jupyter-widgets/base", "_model_module_version": "1.2.0", "_model_name": "LayoutModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "LayoutView", "align_content": null, "align_items": null, "align_self": null, "border": null, "bottom": null, "display": null, "flex": null, "flex_flow": null, "grid_area": null, "grid_auto_columns": null, "grid_auto_flow": null, "grid_auto_rows": null, "grid_column": null, "grid_gap": null, "grid_row": null, "grid_template_areas": null, "grid_template_columns": null, "grid_template_rows": null, "height": null, "justify_content": null, "justify_items": null, "left": null, "margin": null, "max_height": null, "max_width": null, "min_height": null, "min_width": null, "object_fit": null, "object_position": null, "order": null, "overflow": null, "overflow_x": null, "overflow_y": null, "padding": null, "right": null, "top": null, "visibility": null, "width": null } }, "f7503cf63443454a83edfeaa1bc3ca59": { "model_module": "@jupyter-widgets/controls", "model_name": "ProgressStyleModel", "state": { "_model_module": "@jupyter-widgets/controls", "_model_module_version": "1.5.0", "_model_name": "ProgressStyleModel", "_view_count": null, "_view_module": "@jupyter-widgets/base", "_view_module_version": "1.2.0", "_view_name": "StyleView", "bar_color": null, "description_width": "initial" } } } } }, "nbformat": 4, "nbformat_minor": 0 }