From 6fb6e7ab318ef55e2b49e6cbe3f0034017416f73 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?D=C3=81NIEL=20UNYI?= <daniel.unyi.42@gmail.com>
Date: Wed, 8 Jan 2025 01:18:11 +0100
Subject: [PATCH 1/2] Update tutorial notebook

---
 docs/notebooks/segger_tutorial.ipynb | 1848 +++++++++++++-------------
 1 file changed, 914 insertions(+), 934 deletions(-)

diff --git a/docs/notebooks/segger_tutorial.ipynb b/docs/notebooks/segger_tutorial.ipynb
index 6a3c11b..a15e7fa 100644
--- a/docs/notebooks/segger_tutorial.ipynb
+++ b/docs/notebooks/segger_tutorial.ipynb
@@ -1,940 +1,920 @@
 {
-  "cells": [
-    {
-      "cell_type": "markdown",
-      "id": "21ed4db6-5234-46b1-9f38-b5883ac88946",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-11T22:22:55.404267Z",
-          "iopub.status.busy": "2024-09-11T22:22:55.403876Z",
-          "iopub.status.idle": "2024-09-11T22:22:58.089917Z",
-          "shell.execute_reply": "2024-09-11T22:22:58.089303Z",
-          "shell.execute_reply.started": "2024-09-11T22:22:55.404248Z"
-        },
-        "id": "21ed4db6-5234-46b1-9f38-b5883ac88946"
-      },
-      "source": [
-        "# **Introduction to Segger**\n",
-        "\n",
-        "\n",
-        "**Important note (Dec 2024):** As segger is currently undergoing constant development we highly recommend installing directly via github.\n",
-        "\n",
-        "\n",
-        "Segger is a cutting-edge cell segmentation model specifically designed for **single-molecule resolved spatial omics** datasets. It addresses the challenge of accurately segmenting individual cells in complex imaging datasets, leveraging a unique approach based on graph neural networks (GNNs).\n",
-        "\n",
-        "The core idea behind Segger is to model both **nuclei** and **transcripts** as graph nodes, with edges connecting them based on their spatial proximity. This allows the model to learn from the co-occurrence of nucleic and cytoplasmic molecules, resulting in more refined and accurate cell boundaries. By using spatial information and GNNs, Segger achieves state-of-the-art performance in segmenting single cells in datasets such as 10X Xenium and MERSCOPE, outperforming traditional methods like Baysor and Cellpose.\n",
-        "\n",
-        "Segger's workflow consists of:\n",
-        "1. **Dataset creation**: Converting raw transcriptomic data into a graph-based dataset.\n",
-        "2. **Training**: Training the Segger model on the graph to learn cell boundaries.\n",
-        "3. **Prediction**: Using the trained model to make predictions on new datasets.\n",
-        "\n",
-        "This tutorial will guide you through each step of the process, ensuring you can train and apply Segger for your own data."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "XEY6CTzK0648",
-      "metadata": {
-        "id": "XEY6CTzK0648"
-      },
-      "source": [
-        "Installing segger from the GitHub repository:"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "TIQnPzfx08Zr",
-      "metadata": {
-        "id": "TIQnPzfx08Zr"
-      },
-      "outputs": [],
-      "source": [
-        "!git clone https://github.com/EliHei2/segger_dev.git\n",
-        "%cd segger_dev\n",
-        "!pip install \".[rapids12]\" -q"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "q3SNnImS09_N",
-      "metadata": {
-        "id": "q3SNnImS09_N"
-      },
-      "source": [
-        "Downloading the [Xenium Human Pancreatic Dataset](https://www.10xgenomics.com/products/xenium-human-pancreatic-dataset-explorer):"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "Qjdt3f-U0_i9",
-      "metadata": {
-        "id": "Qjdt3f-U0_i9"
-      },
-      "outputs": [],
-      "source": [
-        "!mkdir data_xenium\n",
-        "%cd data_xenium\n",
-        "!wget https://cf.10xgenomics.com/samples/xenium/1.6.0/Xenium_V1_hPancreas_Cancer_Add_on_FFPE/Xenium_V1_hPancreas_Cancer_Add_on_FFPE_outs.zip\n",
-        "!unzip Xenium_V1_hPancreas_Cancer_Add_on_FFPE_outs.zip\n",
-        "%cd .."
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "trM8h-Ek16sJ",
-      "metadata": {
-        "id": "trM8h-Ek16sJ"
-      },
-      "outputs": [],
-      "source": [
-        "from segger.data.parquet.sample import STSampleParquet\n",
-        "from segger.training.segger_data_module import SeggerDataModule\n",
-        "from segger.training.train import LitSegger\n",
-        "from segger.prediction.predict_parquet import segment, load_model\n",
-        "from lightning.pytorch.loggers import CSVLogger\n",
-        "from pytorch_lightning import Trainer\n",
-        "from pathlib import Path\n",
-        "import pandas as pd\n",
-        "from matplotlib import pyplot as plt\n",
-        "import seaborn as sns\n",
-        "import scanpy as sc"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "db009015-c379-4f50-97ed-81dca9df28ac",
-      "metadata": {
-        "id": "db009015-c379-4f50-97ed-81dca9df28ac"
-      },
-      "source": [
-        "# **1. Create your Segger Dataset**\n",
-        "\n",
-        "In this step, we generate the dataset required for Segger's cell segmentation tasks.\n",
-        "\n",
-        "Segger relies on spatial transcriptomics data, combining staining **boundaries** (e.g., nuclei or membrane stainings) and **transcripts** from single-cell resolved imaging datasets. These nuclei and transcript nodes are represented in a graph, and the spatial proximity of transcripts to nuclei is used to establish edges between them.\n",
-        "\n",
-        "To use Segger with a Xenium dataset, you need the **`transcripts.parquet`** and **`nucleus_boundaries.parquet`** (or **`cell_boundaries.parquet`**, in case the Xenium samples comes with the segmentation kit) files. The **transcripts** file contains spatial coordinates and information for each transcript, while the **boundaries** file defines the polygon boundaries of the nuclei or cells. These files enable segger to map transcripts to their respective nuclei and perform cell segmentation based on spatial relationships. Segger can also be extended to other platforms by modifying the column names or formats in the input files to match its expected structure, making it adaptable for various spatial transcriptomics technologies. See (this)[https://github.com/EliHei2/segger_dev/tree/main/src/segger/data/parquet/_settings] for Xenium settings."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "f488a0b7",
-      "metadata": {},
-      "source": []
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "598b4b16",
-      "metadata": {},
-      "outputs": [],
-      "source": []
-    },
-    {
-      "cell_type": "markdown",
-      "id": "9d2b090b",
-      "metadata": {
-        "id": "9d2b090b"
-      },
-      "source": [
-        "### **1.1. Fast Dataset Creation with segger**\n",
-        "\n",
-        "Segger introduces a fast and efficient pipeline for processing spatial transcriptomics data. This method accelerates dataset creation, particularly for large datasets, by using **ND-tree-based spatial partitioning** and **parallel processing**. This results in a much faster preparation of the dataset, which is saved in PyTorch Geometric (PyG) format, similar to the previous method.\n",
-        "\n",
-        "**Note**: The previous dataset creation method will soon be deprecated in favor of this optimized pipeline.\n",
-        "\n",
-        "The pipeline requires the following inputs:\n",
-        "\n",
-        "- **base_dir**: The directory containing the raw dataset.\n",
-        "- **data_dir**: The directory where the processed dataset (tiles in PyG format) will be saved.\n",
-        "\n",
-        "The core improvements in this method come from the use of **ND-tree partitioning**, which splits the data efficiently into spatial regions, and **parallel processing**, which speeds up the handling of these regions across multiple CPU cores. For example, using this pipeline, the Xenium Human Pancreatic Dataset can be processed in just a few minutes when running with 16 workers.\n",
-        "\n",
-        "Below is an example of how to create a dataset using the faster Segger pipeline:"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": 10,
-      "id": "vlDtoWZb24FJ",
-      "metadata": {
-        "id": "vlDtoWZb24FJ"
-      },
-      "outputs": [],
-      "source": []
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "e933ebf3",
-      "metadata": {
-        "id": "e933ebf3"
-      },
-      "outputs": [],
-      "source": [
-        "xenium_data_dir = Path('data_xenium')\n",
-        "segger_data_dir = Path('data_segger')\n",
-        "\n",
-        "sample = STSampleParquet(\n",
-        "    base_dir=xenium_data_dir,\n",
-        "    n_workers=4,\n",
-        "    sample_type='xenium', # this could be 'xenium_v2' in case one uses the cell boundaries from the segmentation kit.\n",
-        "    # weights=gene_celltype_abundance_embedding, # uncomment if gene-celltype embeddings are available\n",
-        ")\n",
-        "\n",
-        "sample.save(\n",
-        "      data_dir=segger_data_dir,\n",
-        "      k_bd=3,\n",
-        "      dist_bd=15.0,\n",
-        "      k_tx=3,\n",
-        "      dist_tx=5.0,\n",
-        "      tile_width=120,\n",
-        "      tile_height=120,\n",
-        "      neg_sampling_ratio=5.0,\n",
-        "      frac=1.0,\n",
-        "      val_prob=0.1,\n",
-        "      test_prob=0.2,\n",
-        ")"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "6ab27f9a",
-      "metadata": {
-        "id": "6ab27f9a"
-      },
-      "source": [
-        "#### **Parameters**\n",
-        "Here is a complete list of parameters you can use to control the dataset creation process:\n",
-        "\n",
-        "- **--base_dir**: Directory containing the raw spatial transcriptomics dataset.\n",
-        "- **--data_dir**: Directory where the processed Segger dataset (in PyG format) will be saved.\n",
-        "- **--sample_type**: (Optional) Specifies the type of dataset (e.g., \"xenium\" or \"merscope\"). Defaults to None.\n",
-        "- **--scrnaseq_file**: Path to the scRNAseq file (default: None).\n",
-        "- **--celltype_column**: Column name for cell type annotations in the scRNAseq file (default: None).\n",
-        "- **--k_bd**: Number of nearest neighbors for boundary nodes (default: 3).\n",
-        "- **--dist_bd**: Maximum distance for boundary neighbors (default: 15.0).\n",
-        "- **--k_tx**: Number of nearest neighbors for transcript nodes (default: 3).\n",
-        "- **--dist_tx**: Maximum distance for transcript neighbors (default: 5.0).\n",
-        "- **--tile_size**: Specifies the size of the tile. If provided, it overrides both tile_width and tile_height.\n",
-        "- **--tile_width**: Width of the tiles in pixels (ignored if tile_size is provided).\n",
-        "- **--tile_height**: Height of the tiles in pixels (ignored if tile_size is provided).\n",
-        "- **--neg_sampling_ratio**: Ratio of negative samples (default: 5.0).\n",
-        "- **--frac**: Fraction of the dataset to process (default: 1.0).\n",
-        "- **--val_prob**: Proportion of data used for validation split (default: 0.1).\n",
-        "- **--test_prob**: Proportion of data used for testing split (default: 0.2).\n",
-        "- **--n_workers**: Number of workers for parallel processing (default: 1)."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "70755046",
-      "metadata": {},
-      "source": [
-        "### **1.2. Using custom gene embeddings**\n",
-        "\n",
-        "In the default mode, segger initially tokenizes transcripts based on their gene type simply in a one-hot manner. However, one can use other genes embeddings (e.g., pre-trained embeddings). The following example shows how one can employ a cell-type-annotated scRNAseq reference of the same tissue type (not necessary same sample or experiment) to embed genes based on their abaundance in different cell types:"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "3bf18259",
-      "metadata": {},
-      "outputs": [],
-      "source": [
-        "from segger.data.utils import calculate_gene_celltype_abundance_embedding\n",
-        "scrnaseq_file = Path('my_scRNAseq_file.h5ad')\n",
-        "celltype_column = 'celltype_column'\n",
-        "gene_celltype_abundance_embedding = calculate_gene_celltype_abundance_embedding(\n",
-        "    sc.read(scrnaseq_file),\n",
-        "    celltype_column\n",
-        ")\n",
-        "\n",
-        "sample = STSampleParquet(\n",
-        "    base_dir=xenium_data_dir,\n",
-        "    n_workers=4,\n",
-        "    sample_type='xenium', # this could be 'xenium_v2' in case one uses the cell boundaries from the segmentation kit.\n",
-        "    weights=gene_celltype_abundance_embedding, \n",
-        ")"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "9962e4b8-4028-4683-9b75-d674fa6fb01d",
-      "metadata": {
-        "id": "9962e4b8-4028-4683-9b75-d674fa6fb01d"
-      },
-      "source": [
-        "# **2. Train your Segger Model**\n",
-        "\n",
-        "The Segger model training process begins after the dataset has been created. This model is a **heterogeneous graph neural network (GNN)** designed to segment single cells by leveraging both nuclei and transcript data.\n",
-        "\n",
-        "Segger uses graph attention layers to propagate information across nodes (nuclei and transcripts) and refine cell boundaries. The model architecture includes initial embedding layers, attention-based graph convolutions, and residual connections for stable learning.\n",
-        "\n",
-        "Segger leverages the **PyTorch Lightning** framework to streamline the training and evaluation of its graph neural network (GNN). PyTorch Lightning simplifies the training process by abstracting away much of the boilerplate code, allowing users to focus on model development and experimentation. It also supports multi-GPU training, mixed-precision, and efficient scaling, making it an ideal framework for training complex models like Segger."
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "4db89cb4-d0eb-426a-a71f-d127926fa412",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-12T00:49:07.236043Z",
-          "iopub.status.busy": "2024-09-12T00:49:07.235854Z",
-          "iopub.status.idle": "2024-09-12T00:49:08.351946Z",
-          "shell.execute_reply": "2024-09-12T00:49:08.351565Z",
-          "shell.execute_reply.started": "2024-09-12T00:49:07.236028Z"
-        },
-        "id": "4db89cb4-d0eb-426a-a71f-d127926fa412"
-      },
-      "outputs": [],
-      "source": [
-        "# Base directory to store Pytorch Lightning models\n",
-        "models_dir = Path('models')\n",
-        "\n",
-        "# Initialize the Lightning model\n",
-        "metadata = ([\"tx\", \"bd\"], [(\"tx\", \"belongs\", \"bd\"), (\"tx\", \"neighbors\", \"tx\")])\n",
-        "ls = LitSegger(\n",
-        "    num_tx_tokens=500,\n",
-        "    init_emb=8,    \n",
-        "    hidden_channels=64,\n",
-        "    out_channels=16,\n",
-        "    heads=4,\n",
-        "    num_mid_layers=1,\n",
-        "    aggr='sum',\n",
-        "    metadata=metadata,\n",
-        ")\n",
-        "\n",
-        "# Initialize the Lightning data module\n",
-        "dm = SeggerDataModule(\n",
-        "    data_dir=segger_data_dir,\n",
-        "    batch_size=2,\n",
-        "    num_workers=2,\n",
-        ")\n",
-        "\n",
-        "dm.setup()\n",
-        "\n",
-        "\n",
-        "# if you wish to use more than 1 device for training you should run this:\n",
-        "batch = dm.train[0]\n",
-        "ls.forward(batch)\n",
-        "\n",
-        "# Initialize the Lightning trainer\n",
-        "trainer = Trainer(\n",
-        "    accelerator='cuda',\n",
-        "    strategy='auto',\n",
-        "    precision='16-mixed',\n",
-        "    devices=1, # set higher number if more gpus are available\n",
-        "    max_epochs=100,\n",
-        "    default_root_dir=models_dir,\n",
-        "    logger=CSVLogger(models_dir),\n",
-        ")"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "207864b8-7e52-4add-a4a2-e95a4debdc06",
-      "metadata": {
-        "id": "207864b8-7e52-4add-a4a2-e95a4debdc06",
-        "scrolled": true
-      },
-      "outputs": [],
-      "source": [
-        "# Fit model\n",
-        "trainer.fit(\n",
-        "    model=ls,\n",
-        "    datamodule=dm\n",
-        ")"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "e6214a79",
-      "metadata": {},
-      "source": [
-        "Key parameters for training:\n",
-        "- **`--data_dir`**: Directory containing the training data.\n",
-        "- **`--model_dir`**: Directory in which to store models.\n",
-        "- **`--epochs`**: Specifies the number of training epochs.\n",
-        "- **`--batch_size`**: Batch sizes for training and validation data.\n",
-        "- **`--learning_rate`**: The initial learning rate for the optimizer.\n",
-        "- **`--hidden_channels`**: Number of hidden channels in the GNN layers.\n",
-        "- **`--heads`**: Number of attention heads used in each graph convolutional layer.\n",
-        "- **`--init_emb`**: Sets the dimensionality of the initial embeddings applied to the input node features (e.g., transcripts). A higher embedding dimension may capture more feature complexity but also requires more computation.\n",
-        "- **`--out_channels`**: Specifies the number of output channels after the final graph attention layer, e.g. the final learned representations of the graph nodes.\n",
-        "\n",
-        "Additional Options for Training the Segger Model:\n",
-        "\n",
-        "- **`--aggr`**: This option controls the aggregation method used in the graph convolution layers.\n",
-        "- **`--accelerator`**: Controls the hardware used for training, such as `cuda` for GPU training. This enables Segger to leverage GPU resources for faster training, especially useful for large datasets.\n",
-        "- **`--strategy`**: Defines the distributed training strategy, with `auto` allowing PyTorch Lightning to automatically configure the best strategy based on the hardware setup.\n",
-        "- **`--precision`**: Enables mixed precision training (e.g., `16-mixed`), which can speed up training and reduce memory usage while maintaining accuracy."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "9a7d20c6-ca16-4beb-b627-afb41e3fb491",
-      "metadata": {
-        "id": "9a7d20c6-ca16-4beb-b627-afb41e3fb491"
-      },
-      "source": [
-        "### *Troubleshooting #1*\n",
-        "\n",
-        "In the cell below, we are visualizing key metrics from the model training and validation process. The plot displays **training loss**, **validation loss**, **F1 validation score**, and **AUROC validation score** over training steps. We expect to see the loss curves decreasing over time, signaling the model's improvement, and the F1 and AUROC scores increasing, reflecting improved segmentation performance as the model learns.\n",
-        "\n",
-        "If training is not working effectively, you might observe the following in the plot displaying **training loss**, **validation loss**, **F1 score**, and **AUROC**:\n",
-        "\n",
-        "- **Training loss not decreasing**: If the training loss remains high or fluctuates without a consistent downward trend, this indicates that the model is not learning effectively from the training data.\n",
-        "- **Validation loss decreases, then increases**: If validation loss decreases initially but starts to increase while training loss continues to drop, this could be a sign of **overfitting**, where the model is performing well on the training data but not generalizing to the validation data.\n",
-        "- **F1 score and AUROC not improving**: If these metrics remain flat or show inconsistent improvement, the model may be struggling to correctly segment cells or classify transcripts, indicating an issue with learning performance.\n"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "43a9c1a4-3898-407d-ac0f-f98b13694593",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-11T22:06:58.182616Z",
-          "iopub.status.busy": "2024-09-11T22:06:58.182357Z",
-          "iopub.status.idle": "2024-09-11T22:07:01.063645Z",
-          "shell.execute_reply": "2024-09-11T22:07:01.063184Z",
-          "shell.execute_reply.started": "2024-09-11T22:06:58.182599Z"
-        },
-        "id": "43a9c1a4-3898-407d-ac0f-f98b13694593",
-        "outputId": "70ba8e1b-7814-497a-c8b6-8aa7295cd4d9"
-      },
-      "outputs": [
-        {
-          "data": {
-            "text/plain": [
-              "Text(0.5, 0, 'Step')"
-            ]
-          },
-          "execution_count": 88,
-          "metadata": {},
-          "output_type": "execute_result"
-        },
-        {
-          "data": {
-            "image/png": "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",
-            "text/plain": [
-              "<Figure size 600x600 with 1 Axes>"
-            ]
-          },
-          "metadata": {},
-          "output_type": "display_data"
-        }
-      ],
-      "source": [
-        "# Evaluate results\n",
-        "model_version = 0  # 'v_num' from training output above\n",
-        "model_path = models_dir / 'lightning_logs' / f'version_{model_version}'\n",
-        "metrics = pd.read_csv(model_path / 'metrics.csv', index_col=1)\n",
-        "\n",
-        "fig, ax = plt.subplots(1,1, figsize=(2,2))\n",
-        "\n",
-        "for col in metrics.columns.difference(['epoch']):\n",
-        "    metric = metrics[col].dropna()\n",
-        "    ax.plot(metric.index, metric.values, label=col)\n",
-        "\n",
-        "ax.legend(loc=(1, 0.33))\n",
-        "ax.set_ylim(0, 1)\n",
-        "ax.set_xlabel('Step')"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "e73687e1-ee8f-46e9-8bd2-1ddc571ef94b",
-      "metadata": {
-        "id": "e73687e1-ee8f-46e9-8bd2-1ddc571ef94b"
-      },
-      "source": [
-        "# **3. Make Predictions**\n",
-        "\n",
-        "Once the Segger model is trained, it can be used to make predictions on seen (partially trained) data or be transfered to unseen data. This step involves using a trained checkpoint to predict cell boundaries and refine transcript-nuclei associations."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "9807abf3",
-      "metadata": {
-        "id": "9807abf3"
-      },
-      "source": [
-        "#### **Requirements for the Faster Prediction Pipeline**\n",
-        "The pipeline requires the following inputs:\n",
-        "\n",
-        "- **segger_data_dir**: The directory containing the processed Segger dataset (in PyG format).\n",
-        "- **models_dir**: The directory containing the trained Segger model checkpoints.\n",
-        "- **benchmarks_dir**: The directory where the segmentation results will be saved.\n",
-        "- **transcripts_file**: Path to the file containing the transcript data for prediction.\n",
-        "\n",
-        "#### **Running the Faster Prediction Pipeline**\n",
-        "Below is an example of how to run the faster Segger prediction pipeline using the command line:"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "PEOtAs-t9CiY",
-      "metadata": {
-        "colab": {
-          "base_uri": "https://localhost:8080/"
-        },
-        "id": "PEOtAs-t9CiY",
-        "outputId": "8b7a5375-9ebc-4bb4-9421-254410319120"
-      },
-      "outputs": [
-        {
-          "name": "stdout",
-          "output_type": "stream",
-          "text": [
-            "Starting segmentation for segger_embedding_1001...\n"
-          ]
-        }
-      ],
-      "source": [
-        "dm = SeggerDataModule(\n",
-        "    data_dir='data_segger',\n",
-        "    batch_size=1,\n",
-        "    num_workers=4,\n",
-        ")\n",
-        "\n",
-        "dm.setup()\n",
-        "\n",
-        "model_version = 0\n",
-        "model_path = Path('models') / \"lightning_logs\" / f\"version_{model_version}\"\n",
-        "model = load_model(model_path / \"checkpoints\")\n",
-        "\n",
-        "receptive_field = {'k_bd': 4, 'dist_bd': 12, 'k_tx': 15, 'dist_tx': 3}\n",
-        "\n",
-        "segment(\n",
-        "    model,\n",
-        "    dm,\n",
-        "    save_dir='benchmarks',\n",
-        "    seg_tag='segger_output',\n",
-        "    transcript_file='data_xenium/transcripts.parquet',\n",
-        "    receptive_field=receptive_field,\n",
-        "    min_transcripts=5,\n",
-        "    cell_id_col='segger_cell_id',\n",
-        "    use_cc=False,\n",
-        "    knn_method='cuda',\n",
-        "    verbose=True,\n",
-        ")\n"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "3aa5002c",
-      "metadata": {},
-      "source": []
-    },
-    {
-      "cell_type": "markdown",
-      "id": "0a823035",
-      "metadata": {
-        "id": "0a823035"
-      },
-      "source": [
-        "#### **Parameters**\n",
-        "Here is a detailed explanation of each parameter used in the faster prediction pipeline:\n",
-        "\n",
-        "- **--segger_data_dir**: The directory containing the processed Segger dataset, saved as PyTorch Geometric data objects, that will be used for prediction.\n",
-        "- **--models_dir**: The directory containing the trained Segger model checkpoints. These checkpoints store the learned weights required for making predictions.\n",
-        "- **--benchmarks_dir**: The directory where the segmentation results will be saved.\n",
-        "- **--transcripts_file**: Path to the *transcripts.parquet* file.\n",
-        "- **--batch_size**: Specifies the batch size for processing during prediction. Larger batch sizes speed up inference but use more memory (default: 1).\n",
-        "- **--num_workers**: Number of workers to use for parallel data loading (default: 1).\n",
-        "- **--model_version**: Version of the trained model to load for predictions, based on the version number from the training logs (default: 0).\n",
-        "- **--save_tag**: A tag used to name and organize the segmentation results (default: segger_embedding).\n",
-        "- **--min_transcripts**: The minimum number of transcripts required for segmentation (default: 5).\n",
-        "- **--cell_id_col**: The name of the column that stores the cell IDs (default: segger_cell_id).\n",
-        "- **--use_cc**: Enables the use of connected components (CC) for grouping transcripts that are not associated with any nucleus (default: False).\n",
-        "- **--knn_method**: Method for KNN (K-Nearest Neighbors) computation. Only option is \"cuda\" for this pipeline (default: cuda).\n",
-        "- **--file_format**: The format for saving the output segmentation data. Only option is \"anndata\" for this pipeline (default: anndata).\n",
-        "- **--k_bd**: Number of nearest neighbors for boundary nodes during segmentation (default: 4).\n",
-        "- **--dist_bd**: Maximum distance for boundary nodes during segmentation (default: 12.0).\n",
-        "- **--k_tx**: Number of nearest neighbors for transcript nodes during segmentation (default: 5).\n",
-        "- **--dist_tx**: Maximum distance for transcript nodes during segmentation (default: 5.0)."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "b0917be9-4e82-4ba5-869d-5a9203721699",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-11T23:06:23.977884Z",
-          "iopub.status.busy": "2024-09-11T23:06:23.977517Z"
-        },
-        "id": "b0917be9-4e82-4ba5-869d-5a9203721699"
-      },
-      "source": [
-        "### *Troubleshooting #2*\n",
-        "\n",
-        "In the cell below, we are visualizing the distribution of **Segger similarity scores** using a histogram. The **Segger similarity score** reflects how closely transcripts are associated with their respective nuclei in the segmentation process. **Higher scores** indicate stronger associations between transcripts and their nuclei, suggesting more accurate cell boundaries. **Lower scores** might indicate weaker associations, which could highlight potential segmentation errors or challenging regions in the data. We expect to see a large number of the scores clustering toward higher values, which would indicate strong overall performance of the model in associating transcripts with nuclei.\n",
-        "\n",
-        "The following would indicate potential issues with the model's predictions:\n",
-        "\n",
-        "- **A very large portion of scores near zero**: If many scores are concentrated at the lower end of the scale, this suggests that the model is frequently failing to associate transcripts with their corresponding nuclei, indicating poor segmentation quality.\n",
-        "- **No clear peak in the distribution**: If the histogram is flat or shows a wide, spread-out distribution, this could indicate that the model is struggling to consistently assign similarity scores, which may be a sign that the training process did not optimize the model correctly.\n",
-        "\n",
-        "Both cases would suggest that the model requires further tuning, such as adjusting hyperparameters, data preprocessing, or the training procedure (see below)\n"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "a450d3ca-2876-4f48-be89-761147b17387",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-11T22:07:04.216273Z",
-          "iopub.status.busy": "2024-09-11T22:07:04.215965Z",
-          "iopub.status.idle": "2024-09-11T22:07:08.177601Z",
-          "shell.execute_reply": "2024-09-11T22:07:08.177158Z",
-          "shell.execute_reply.started": "2024-09-11T22:07:04.216257Z"
-        },
-        "id": "a450d3ca-2876-4f48-be89-761147b17387",
-        "outputId": "0576e0b8-4823-4701-b661-dc6b513841f2"
-      },
-      "outputs": [
-        {
-          "data": {
-            "image/png": "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",
-            "text/plain": [
-              "<Figure size 600x600 with 1 Axes>"
-            ]
-          },
-          "metadata": {},
-          "output_type": "display_data"
-        }
-      ],
-      "source": [
-        "fig, ax = plt.subplots(1,1, figsize=(2,2))\n",
-        "sns.histplot(\n",
-        "    segmentation['score'],\n",
-        "    bins=50,\n",
-        "    ax=ax,\n",
-        ")\n",
-        "ax.set_ylabel('Count')\n",
-        "ax.set_xlabel('Segger Similarity Score')\n",
-        "ax.set_yscale('log')"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "5492fb96-bf8e-49d5-b40e-7e6b3f871bbe",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-11T22:34:15.990223Z",
-          "iopub.status.busy": "2024-09-11T22:34:15.988880Z"
-        },
-        "id": "5492fb96-bf8e-49d5-b40e-7e6b3f871bbe"
-      },
-      "source": [
-        "#### The Importance of the Receptive Field in Segger\n",
-        "\n",
-        "The **receptive field** is a critical parameter in Segger, as it directly influences how the model interprets the spatial relationships between **transcripts** and **nuclei**. In the context of spatial transcriptomics, the receptive field determines the size of the neighborhood that each node (representing transcripts or nuclei) can \"see\" during graph construction and model training. Segger is particularly sensitive to the size of the receptive field because it affects the model's ability to propagate information across the graph. If the receptive field is too small, the model may fail to capture sufficient context for correct cell boundary delineation. Conversely, a very large receptive field may introduce noise by linking unrelated or distant nodes, reducing segmentation accuracy.\n",
-        "\n",
-        "#### Parameters affecting the receptive field in Segger:\n",
-        "- **`--r`**: This parameter defines the radius used when connecting transcripts to nuclei. A larger `r` expands the receptive field, linking more distant nodes. Fine-tuning this parameter helps ensure that Segger captures the right level of spatial interaction in the dataset.\n",
-        "- **`--k_bd` and `--k_tx`**: These control the number of nearest neighbors (nuclei and transcripts, respectively) considered in the graph. By increasing these values, the receptive field is effectively broadened, allowing more nodes to contribute to the information propagation.\n",
-        "- **`--dist_bd` and `--dist_tx`**: These parameters specify the maximum distances used to connect nuclei (`dist_bd`) and transcripts (`dist_tx`) to their neighbors during graph construction. They directly affect the receptive field by defining the cut-off distance for forming edges in the graph. Larger distance values expand the receptive field, connecting nodes that are further apart spatially. Careful tuning of these values is necessary to ensure that Segger captures relevant spatial relationships without introducing noise."
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "7ece1ac0-0708-45e2-87fc-1b25782831f8",
-      "metadata": {
-        "id": "7ece1ac0-0708-45e2-87fc-1b25782831f8"
-      },
-      "source": [
-        "# **4. Tune Parameters**"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "896b8288-5287-4d10-a206-e68c0e4731c6",
-      "metadata": {
-        "id": "896b8288-5287-4d10-a206-e68c0e4731c6"
-      },
-      "source": [
-        "### Evaluating Receptive Field Parameters with Grid Search\n",
-        "\n",
-        "To evaluate the impact of different receptive field parameters in Segger, we use a **grid search** approach. The parameters `k_bd`, `k_tx`, `dist_bd`, and `dist_tx` (which control the number of neighbors and distances for nuclei and transcripts) are explored through various configurations defined in `param_space`. Each combination of these parameters is passed to the `trainable` function, which creates the dataset, trains the model, and makes predictions based on the specified receptive field.\n",
-        "\n",
-        "For each parameter combination:\n",
-        "1. A dataset is created with the specified receptive field.\n",
-        "2. The Segger model is trained on this dataset.\n",
-        "3. Predictions are made, and segmentation results are evaluated using the custom `evaluate` function. This function computes metrics like the fraction of assigned transcripts and average cell sizes.\n",
-        "\n",
-        "The results from each configuration are saved, allowing us to compare how different receptive field settings impact the model’s performance. This process enables a thorough search of the parameter space, optimizing the model for accurate segmentation."
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "b0c1a7a8-acb2-4aae-8ae4-8aa9a4196717",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-12T01:10:47.781418Z",
-          "iopub.status.busy": "2024-09-12T01:10:47.781067Z",
-          "iopub.status.idle": "2024-09-12T01:10:48.706615Z",
-          "shell.execute_reply": "2024-09-12T01:10:48.706194Z",
-          "shell.execute_reply.started": "2024-09-12T01:10:47.781401Z"
-        },
-        "id": "b0c1a7a8-acb2-4aae-8ae4-8aa9a4196717"
-      },
-      "outputs": [],
-      "source": [
-        "import itertools\n",
-        "import pandas as pd"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "0bd0803c-e58d-4f43-9627-d2c1ab187d5e",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-12T01:16:31.976312Z",
-          "iopub.status.busy": "2024-09-12T01:16:31.975947Z",
-          "iopub.status.idle": "2024-09-12T01:16:33.168389Z",
-          "shell.execute_reply": "2024-09-12T01:16:33.167956Z",
-          "shell.execute_reply.started": "2024-09-12T01:16:31.976295Z"
-        },
-        "id": "0bd0803c-e58d-4f43-9627-d2c1ab187d5e"
-      },
-      "outputs": [],
-      "source": [
-        "tuning_dir = Path('path/to/tutorial/tuning/')\n",
-        "sampling_rate = 0.125"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "b879a0b5-150c-4240-99ec-81075855aa52",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-12T01:16:33.169525Z",
-          "iopub.status.busy": "2024-09-12T01:16:33.169189Z",
-          "iopub.status.idle": "2024-09-12T01:16:34.147222Z",
-          "shell.execute_reply": "2024-09-12T01:16:34.146804Z",
-          "shell.execute_reply.started": "2024-09-12T01:16:33.169508Z"
-        },
-        "id": "b879a0b5-150c-4240-99ec-81075855aa52",
-        "jupyter": {
-          "source_hidden": true
-        }
-      },
-      "outputs": [],
-      "source": [
-        "# Fixed function arguments used for each trial\n",
-        "transcripts_path = xenium_data_dir / 'transcripts.parquet'\n",
-        "\n",
-        "boundaries_path = xenium_data_dir / 'nucleus_boundaries.parquet'\n",
-        "\n",
-        "dataset_kwargs = dict(\n",
-        "    x_size=80, y_size=80, d_x=80, d_y=80, margin_x=10, margin_y=10,\n",
-        "    num_workers=4, sampling_rate=sampling_rate,\n",
-        ")\n",
-        "\n",
-        "model_kwargs = dict(\n",
-        "    metadata=(['tx', 'bd'], [('tx', 'belongs', 'bd'), ('tx', 'neighbors', 'tx')]),\n",
-        "    num_tx_tokens=500, init_emb=8, hidden_channels=32, out_channels=8,\n",
-        "    heads=2, num_mid_layers=2, aggr='sum',\n",
-        ")\n",
-        "\n",
-        "trainer_kwargs = dict(\n",
-        "    accelerator='cuda', strategy='auto', precision='16-mixed', devices=1,\n",
-        "    max_epochs=100,\n",
-        ")\n",
-        "\n",
-        "predict_kwargs = dict(score_cut=0.2, use_cc=True)"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "fbd831c9-3a50-4e3b-97d3-3c152ae01188",
-      "metadata": {
-        "id": "fbd831c9-3a50-4e3b-97d3-3c152ae01188",
-        "jupyter": {
-          "source_hidden": true
-        }
-      },
-      "outputs": [],
-      "source": [
-        "def trainable(config):\n",
-        "\n",
-        "    receptive_field = {k: config[k] for k in ['k_bd', 'k_tx', 'dist_bd', 'dist_tx']}\n",
-        "\n",
-        "    # Dataset creation\n",
-        "    xs = XeniumSample(verbose=False)\n",
-        "    xs.set_file_paths(transcripts_path, boundaries_path)\n",
-        "    xs.set_metadata()\n",
-        "    try:\n",
-        "        xs.save_dataset_for_segger(\n",
-        "            processed_dir=config['data_dir'],\n",
-        "            receptive_field=receptive_field,\n",
-        "            **dataset_kwargs,\n",
-        "        )\n",
-        "    except:\n",
-        "        pass\n",
-        "\n",
-        "    # Model training\n",
-        "    ls = LitSegger(**model_kwargs)\n",
-        "    dm = SeggerDataModule(\n",
-        "        data_dir=config['data_dir'],\n",
-        "        batch_size=2,\n",
-        "        num_workers=dataset_kwargs['num_workers'],\n",
-        "    )\n",
-        "    trainer = Trainer(\n",
-        "        default_root_dir=config['model_dir'],\n",
-        "        logger=CSVLogger(config['model_dir']),\n",
-        "        **trainer_kwargs,\n",
-        "    )\n",
-        "    trainer.fit(model=ls, datamodule=dm)\n",
-        "\n",
-        "    segmentation = predict(\n",
-        "        load_model(config['model_dir']/'lightning_logs/version_0/checkpoints'),\n",
-        "        dm.train_dataloader(),\n",
-        "        receptive_field=receptive_field,\n",
-        "        **predict_kwargs,\n",
-        "    )\n",
-        "\n",
-        "    metrics = evaluate(segmentation)\n",
-        "\n",
-        "\n",
-        "def evaluate(segmentation: pd.DataFrame, score_cut: float) -> pd.Series:\n",
-        "\n",
-        "    assigned = segmentation['score'] > score_cut\n",
-        "    metrics = pd.Series(dtype=float)\n",
-        "    metrics['frac_assigned'] = assigned.mean()\n",
-        "    cell_sizes = segmentation.groupby(assigned)['segger_cell_id'].value_counts()\n",
-        "    assigned_avg = 0 if True not in cell_sizes.index else cell_sizes[True].mean()\n",
-        "    cc_avg = 0 if False not in cell_sizes.index else cell_sizes[False].mean()\n",
-        "    metrics['cell_size_assigned'] = assigned_avg\n",
-        "    metrics['cell_size_cc'] = cc_avg\n",
-        "    return metrics"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "ba2dcc9a-3a06-4b84-a487-59a768eed5d5",
-      "metadata": {
-        "execution": {
-          "iopub.execute_input": "2024-09-12T01:16:35.184598Z",
-          "iopub.status.busy": "2024-09-12T01:16:35.184180Z",
-          "iopub.status.idle": "2024-09-12T01:19:55.171470Z",
-          "shell.execute_reply": "2024-09-12T01:19:55.170810Z",
-          "shell.execute_reply.started": "2024-09-12T01:16:35.184582Z"
-        },
-        "id": "ba2dcc9a-3a06-4b84-a487-59a768eed5d5",
-        "scrolled": true
-      },
-      "outputs": [],
-      "source": [
-        "param_space = {\n",
-        "    \"k_bd\": [3, 5, 10],\n",
-        "    \"dist_bd\": [5, 10, 15, 20],\n",
-        "    \"k_tx\": [3, 5, 10],\n",
-        "    \"dist_tx\": [3, 5, 10],\n",
-        "}\n",
-        "\n",
-        "metrics = []\n",
-        "\n",
-        "for params in itertools.product(*param_space.values()):\n",
-        "\n",
-        "    config = dict(zip(param_space.keys(), params))\n",
-        "\n",
-        "    # Setup directories\n",
-        "    trial_dir = tuning_dir / '_'.join([f'{k}={v}' for k, v in config.items()])\n",
-        "\n",
-        "    data_dir = trial_dir / 'segger_data'\n",
-        "    data_dir.mkdir(exist_ok=True, parents=True)\n",
-        "    config['data_dir'] = data_dir\n",
-        "\n",
-        "    model_dir = trial_dir / 'models'\n",
-        "    model_dir.mkdir(exist_ok=True, parents=True)\n",
-        "    config['model_dir'] = model_dir\n",
-        "\n",
-        "    segmentation = trainable(config)\n",
-        "    trial = evaluate(segmentation, predict_kwargs['score_cut'])\n",
-        "    trial = pd.concat([pd.Series(config), trial])\n",
-        "    metrics.append(trial)\n",
-        "\n",
-        "metrics = pd.DataFrame(metrics)"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "id": "dcfa5570-ada2-4102-aae0-a3830d304c5f",
-      "metadata": {
-        "id": "dcfa5570-ada2-4102-aae0-a3830d304c5f"
-      },
-      "source": [
-        "### Interpreting Output Metrics\n",
-        "\n",
-        "The key output metrics include:\n",
-        "- **`frac_assigned`**: The fraction of transcripts that were successfully assigned to a cell. A higher value indicates that the model is doing a good job associating transcripts with nuclei, which is a strong indicator of successful segmentation.\n",
-        "- **`cell_size_assigned`**: The average size of cells that have assigned transcripts. This helps assess how well the model is predicting cell boundaries, with unusually large or small values indicating potential issues with segmentation accuracy.\n",
-        "- **`cell_size_cc`**: The average size of connected components that were not assigned to a cell (i.e., nucleus-less regions). Large values here may suggest that transcripts are being incorrectly grouped together in the absence of a nucleus, which could indicate problems with the receptive field parameters or the segmentation process.\n",
-        "\n",
-        "These metrics illuminate the effectiveness of the model by highlighting both the success in associating transcripts with cells and potential areas where the model may need further tuning.\n"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "id": "1a89aed4-c53b-460f-8a6f-f690920b6829",
-      "metadata": {
-        "execution": {
-          "iopub.status.busy": "2024-09-12T01:19:55.171961Z",
-          "iopub.status.idle": "2024-09-12T01:19:55.172161Z",
-          "shell.execute_reply": "2024-09-12T01:19:55.172071Z",
-          "shell.execute_reply.started": "2024-09-12T01:19:55.172062Z"
-        },
-        "id": "1a89aed4-c53b-460f-8a6f-f690920b6829"
-      },
-      "outputs": [],
-      "source": [
-        "metrics"
-      ]
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "21ed4db6-5234-46b1-9f38-b5883ac88946",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-11T22:22:55.404267Z",
+     "iopub.status.busy": "2024-09-11T22:22:55.403876Z",
+     "iopub.status.idle": "2024-09-11T22:22:58.089917Z",
+     "shell.execute_reply": "2024-09-11T22:22:58.089303Z",
+     "shell.execute_reply.started": "2024-09-11T22:22:55.404248Z"
+    },
+    "id": "21ed4db6-5234-46b1-9f38-b5883ac88946"
+   },
+   "source": [
+    "# **Introduction to Segger**\n",
+    "\n",
+    "\n",
+    "**Important note (Dec 2024):** As segger is currently undergoing constant development we highly recommend installing directly via github.\n",
+    "\n",
+    "\n",
+    "Segger is a cutting-edge cell segmentation model specifically designed for **single-molecule resolved spatial omics** datasets. It addresses the challenge of accurately segmenting individual cells in complex imaging datasets, leveraging a unique approach based on graph neural networks (GNNs).\n",
+    "\n",
+    "The core idea behind Segger is to model both **nuclei** and **transcripts** as graph nodes, with edges connecting them based on their spatial proximity. This allows the model to learn from the co-occurrence of nucleic and cytoplasmic molecules, resulting in more refined and accurate cell boundaries. By using spatial information and GNNs, Segger achieves state-of-the-art performance in segmenting single cells in datasets such as 10X Xenium and MERSCOPE, outperforming traditional methods like Baysor and Cellpose.\n",
+    "\n",
+    "Segger's workflow consists of:\n",
+    "1. **Dataset creation**: Converting raw transcriptomic data into a graph-based dataset.\n",
+    "2. **Training**: Training the Segger model on the graph to learn cell boundaries.\n",
+    "3. **Prediction**: Using the trained model to make predictions on new datasets.\n",
+    "\n",
+    "This tutorial will guide you through each step of the process, ensuring you can train and apply Segger for your own data."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "XEY6CTzK0648",
+   "metadata": {
+    "id": "XEY6CTzK0648"
+   },
+   "source": [
+    "Installing segger from the GitHub repository:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "TIQnPzfx08Zr",
+   "metadata": {
+    "id": "TIQnPzfx08Zr"
+   },
+   "outputs": [],
+   "source": [
+    "!git clone https://github.com/EliHei2/segger_dev.git\n",
+    "%cd segger_dev\n",
+    "!pip install \".[rapids12]\" -q"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "q3SNnImS09_N",
+   "metadata": {
+    "id": "q3SNnImS09_N"
+   },
+   "source": [
+    "Downloading the [Xenium Human Pancreatic Dataset](https://www.10xgenomics.com/products/xenium-human-pancreatic-dataset-explorer):"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "Qjdt3f-U0_i9",
+   "metadata": {
+    "id": "Qjdt3f-U0_i9"
+   },
+   "outputs": [],
+   "source": [
+    "!mkdir data_xenium\n",
+    "%cd data_xenium\n",
+    "!wget https://cf.10xgenomics.com/samples/xenium/1.6.0/Xenium_V1_hPancreas_Cancer_Add_on_FFPE/Xenium_V1_hPancreas_Cancer_Add_on_FFPE_outs.zip\n",
+    "!unzip Xenium_V1_hPancreas_Cancer_Add_on_FFPE_outs.zip\n",
+    "%cd .."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "trM8h-Ek16sJ",
+   "metadata": {
+    "id": "trM8h-Ek16sJ"
+   },
+   "outputs": [],
+   "source": [
+    "from segger.data.parquet.sample import STSampleParquet\n",
+    "from segger.training.segger_data_module import SeggerDataModule\n",
+    "from segger.training.train import LitSegger\n",
+    "from segger.prediction.predict_parquet import segment, load_model\n",
+    "from lightning.pytorch.loggers import CSVLogger\n",
+    "from pytorch_lightning import Trainer\n",
+    "from pathlib import Path\n",
+    "import pandas as pd\n",
+    "from matplotlib import pyplot as plt\n",
+    "import seaborn as sns\n",
+    "import scanpy as sc"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "db009015-c379-4f50-97ed-81dca9df28ac",
+   "metadata": {
+    "id": "db009015-c379-4f50-97ed-81dca9df28ac"
+   },
+   "source": [
+    "# **1. Create your Segger Dataset**\n",
+    "\n",
+    "In this step, we generate the dataset required for Segger's cell segmentation tasks.\n",
+    "\n",
+    "Segger relies on spatial transcriptomics data, combining staining **boundaries** (e.g., nuclei or membrane stainings) and **transcripts** from single-cell resolved imaging datasets. These nuclei and transcript nodes are represented in a graph, and the spatial proximity of transcripts to nuclei is used to establish edges between them.\n",
+    "\n",
+    "To use Segger with a Xenium dataset, you need the **`transcripts.parquet`** and **`nucleus_boundaries.parquet`** (or **`cell_boundaries.parquet`**, in case the Xenium samples comes with the segmentation kit) files. The **transcripts** file contains spatial coordinates and information for each transcript, while the **boundaries** file defines the polygon boundaries of the nuclei or cells. These files enable segger to map transcripts to their respective nuclei and perform cell segmentation based on spatial relationships. Segger can also be extended to other platforms by modifying the column names or formats in the input files to match its expected structure, making it adaptable for various spatial transcriptomics technologies. See [this](https://github.com/EliHei2/segger_dev/tree/main/src/segger/data/parquet/_settings) for Xenium settings."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "9d2b090b",
+   "metadata": {
+    "id": "9d2b090b"
+   },
+   "source": [
+    "### **1.1. Fast Dataset Creation with segger**\n",
+    "\n",
+    "Segger introduces a fast and efficient pipeline for processing spatial transcriptomics data. This method accelerates dataset creation, particularly for large datasets, by using **ND-tree-based spatial partitioning** and **parallel processing**. This results in a much faster preparation of the dataset, which is saved in PyTorch Geometric (PyG) format, similar to the previous method.\n",
+    "\n",
+    "**Note**: The previous dataset creation method will soon be deprecated in favor of this optimized pipeline.\n",
+    "\n",
+    "The pipeline requires the following inputs:\n",
+    "\n",
+    "- **base_dir**: The directory containing the raw dataset.\n",
+    "- **data_dir**: The directory where the processed dataset (tiles in PyG format) will be saved.\n",
+    "\n",
+    "The core improvements in this method come from the use of **ND-tree partitioning**, which splits the data efficiently into spatial regions, and **parallel processing**, which speeds up the handling of these regions across multiple CPU cores. For example, using this pipeline, the Xenium Human Pancreatic Dataset can be processed in just a few minutes when running with 16 workers.\n",
+    "\n",
+    "Below is an example of how to create a dataset using the faster Segger pipeline:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "e933ebf3",
+   "metadata": {
+    "id": "e933ebf3"
+   },
+   "outputs": [],
+   "source": [
+    "xenium_data_dir = Path('data_xenium')\n",
+    "segger_data_dir = Path('data_segger')\n",
+    "\n",
+    "sample = STSampleParquet(\n",
+    "    base_dir=xenium_data_dir,\n",
+    "    n_workers=4,\n",
+    "    sample_type='xenium', # this could be 'xenium_v2' in case one uses the cell boundaries from the segmentation kit.\n",
+    "    # weights=gene_celltype_abundance_embedding, # uncomment if gene-celltype embeddings are available\n",
+    ")\n",
+    "\n",
+    "sample.save(\n",
+    "      data_dir=segger_data_dir,\n",
+    "      k_bd=3,\n",
+    "      dist_bd=15.0,\n",
+    "      k_tx=3,\n",
+    "      dist_tx=5.0,\n",
+    "      tile_width=120,\n",
+    "      tile_height=120,\n",
+    "      neg_sampling_ratio=5.0,\n",
+    "      frac=1.0,\n",
+    "      val_prob=0.1,\n",
+    "      test_prob=0.2,\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "6ab27f9a",
+   "metadata": {
+    "id": "6ab27f9a"
+   },
+   "source": [
+    "#### **Parameters**\n",
+    "Here is a complete list of parameters you can use to control the dataset creation process:\n",
+    "\n",
+    "- **--base_dir**: Directory containing the raw spatial transcriptomics dataset.\n",
+    "- **--data_dir**: Directory where the processed Segger dataset (in PyG format) will be saved.\n",
+    "- **--sample_type**: (Optional) Specifies the type of dataset (e.g., \"xenium\" or \"merscope\"). Defaults to None.\n",
+    "- **--scrnaseq_file**: Path to the scRNAseq file (default: None).\n",
+    "- **--celltype_column**: Column name for cell type annotations in the scRNAseq file (default: None).\n",
+    "- **--k_bd**: Number of nearest neighbors for boundary nodes (default: 3).\n",
+    "- **--dist_bd**: Maximum distance for boundary neighbors (default: 15.0).\n",
+    "- **--k_tx**: Number of nearest neighbors for transcript nodes (default: 3).\n",
+    "- **--dist_tx**: Maximum distance for transcript neighbors (default: 5.0).\n",
+    "- **--tile_size**: Specifies the size of the tile. If provided, it overrides both tile_width and tile_height.\n",
+    "- **--tile_width**: Width of the tiles in pixels (ignored if tile_size is provided).\n",
+    "- **--tile_height**: Height of the tiles in pixels (ignored if tile_size is provided).\n",
+    "- **--neg_sampling_ratio**: Ratio of negative samples (default: 5.0).\n",
+    "- **--frac**: Fraction of the dataset to process (default: 1.0).\n",
+    "- **--val_prob**: Proportion of data used for validation split (default: 0.1).\n",
+    "- **--test_prob**: Proportion of data used for testing split (default: 0.2).\n",
+    "- **--n_workers**: Number of workers for parallel processing (default: 1)."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "70755046",
+   "metadata": {},
+   "source": [
+    "### **1.2. Using custom gene embeddings**\n",
+    "\n",
+    "In the default mode, segger initially tokenizes transcripts based on their gene type simply in a one-hot manner. However, one can use other genes embeddings (e.g., pre-trained embeddings). The following example shows how one can employ a cell-type-annotated scRNAseq reference of the same tissue type (not necessary same sample or experiment) to embed genes based on their abaundance in different cell types:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3bf18259",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from segger.data.utils import calculate_gene_celltype_abundance_embedding\n",
+    "scrnaseq_file = Path('my_scRNAseq_file.h5ad')\n",
+    "celltype_column = 'celltype_column'\n",
+    "gene_celltype_abundance_embedding = calculate_gene_celltype_abundance_embedding(\n",
+    "    sc.read(scrnaseq_file),\n",
+    "    celltype_column\n",
+    ")\n",
+    "\n",
+    "sample = STSampleParquet(\n",
+    "    base_dir=xenium_data_dir,\n",
+    "    n_workers=4,\n",
+    "    sample_type='xenium', # this could be 'xenium_v2' in case one uses the cell boundaries from the segmentation kit.\n",
+    "    weights=gene_celltype_abundance_embedding, \n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "9962e4b8-4028-4683-9b75-d674fa6fb01d",
+   "metadata": {
+    "id": "9962e4b8-4028-4683-9b75-d674fa6fb01d"
+   },
+   "source": [
+    "# **2. Train your Segger Model**\n",
+    "\n",
+    "The Segger model training process begins after the dataset has been created. This model is a **heterogeneous graph neural network (GNN)** designed to segment single cells by leveraging both nuclei and transcript data.\n",
+    "\n",
+    "Segger uses graph attention layers to propagate information across nodes (nuclei and transcripts) and refine cell boundaries. The model architecture includes initial embedding layers, attention-based graph convolutions, and residual connections for stable learning.\n",
+    "\n",
+    "Segger leverages the **PyTorch Lightning** framework to streamline the training and evaluation of its graph neural network (GNN). PyTorch Lightning simplifies the training process by abstracting away much of the boilerplate code, allowing users to focus on model development and experimentation. It also supports multi-GPU training, mixed-precision, and efficient scaling, making it an ideal framework for training complex models like Segger."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4db89cb4-d0eb-426a-a71f-d127926fa412",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-12T00:49:07.236043Z",
+     "iopub.status.busy": "2024-09-12T00:49:07.235854Z",
+     "iopub.status.idle": "2024-09-12T00:49:08.351946Z",
+     "shell.execute_reply": "2024-09-12T00:49:08.351565Z",
+     "shell.execute_reply.started": "2024-09-12T00:49:07.236028Z"
+    },
+    "id": "4db89cb4-d0eb-426a-a71f-d127926fa412"
+   },
+   "outputs": [],
+   "source": [
+    "# Base directory to store Pytorch Lightning models\n",
+    "models_dir = Path('models')\n",
+    "\n",
+    "# Initialize the Lightning data module\n",
+    "dm = SeggerDataModule(\n",
+    "    data_dir=segger_data_dir,\n",
+    "    batch_size=2,\n",
+    "    num_workers=2,\n",
+    ")\n",
+    "\n",
+    "dm.setup()\n",
+    "\n",
+    "is_token_based = True\n",
+    "num_tx_tokens = 500\n",
+    "\n",
+    "# If you use custom gene embeddings, use the following two lines instead:\n",
+    "# is_token_based = False\n",
+    "# num_tx_tokens = dm.train[0].x_dict[\"tx\"].shape[1] # Set the number of tokens to the number of genes\n",
+    "\n",
+    "\n",
+    "num_bd_features = dm.train[0].x_dict[\"bd\"].shape[1]\n",
+    "\n",
+    "# Initialize the Lightning model\n",
+    "ls = LitSegger(\n",
+    "    is_token_based = is_token_based,\n",
+    "    num_node_features = {\"tx\": num_tx_tokens, \"bd\": num_bd_features},\n",
+    "    init_emb=8,    \n",
+    "    hidden_channels=64,\n",
+    "    out_channels=16,\n",
+    "    heads=4,\n",
+    "    num_mid_layers=1,\n",
+    "    aggr='sum',\n",
+    ")\n",
+    "\n",
+    "# Initialize the Lightning trainer\n",
+    "trainer = Trainer(\n",
+    "    accelerator='cuda',\n",
+    "    strategy='auto',\n",
+    "    precision='16-mixed',\n",
+    "    devices=1, # set higher number if more gpus are available\n",
+    "    max_epochs=100,\n",
+    "    default_root_dir=models_dir,\n",
+    "    logger=CSVLogger(models_dir),\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "207864b8-7e52-4add-a4a2-e95a4debdc06",
+   "metadata": {
+    "id": "207864b8-7e52-4add-a4a2-e95a4debdc06",
+    "scrolled": true
+   },
+   "outputs": [],
+   "source": [
+    "# Fit model\n",
+    "trainer.fit(\n",
+    "    model=ls,\n",
+    "    datamodule=dm\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e6214a79",
+   "metadata": {},
+   "source": [
+    "Key parameters for training:\n",
+    "- **`--data_dir`**: Directory containing the training data.\n",
+    "- **`--model_dir`**: Directory in which to store models.\n",
+    "- **`--epochs`**: Specifies the number of training epochs.\n",
+    "- **`--batch_size`**: Batch sizes for training and validation data.\n",
+    "- **`--learning_rate`**: The initial learning rate for the optimizer.\n",
+    "- **`--hidden_channels`**: Number of hidden channels in the GNN layers.\n",
+    "- **`--heads`**: Number of attention heads used in each graph convolutional layer.\n",
+    "- **`--init_emb`**: Sets the dimensionality of the initial embeddings applied to the input node features (e.g., transcripts). A higher embedding dimension may capture more feature complexity but also requires more computation.\n",
+    "- **`--out_channels`**: Specifies the number of output channels after the final graph attention layer, e.g. the final learned representations of the graph nodes.\n",
+    "\n",
+    "Additional Options for Training the Segger Model:\n",
+    "\n",
+    "- **`--aggr`**: This option controls the aggregation method used in the graph convolution layers.\n",
+    "- **`--accelerator`**: Controls the hardware used for training, such as `cuda` for GPU training. This enables Segger to leverage GPU resources for faster training, especially useful for large datasets.\n",
+    "- **`--strategy`**: Defines the distributed training strategy, with `auto` allowing PyTorch Lightning to automatically configure the best strategy based on the hardware setup.\n",
+    "- **`--precision`**: Enables mixed precision training (e.g., `16-mixed`), which can speed up training and reduce memory usage while maintaining accuracy."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "9a7d20c6-ca16-4beb-b627-afb41e3fb491",
+   "metadata": {
+    "id": "9a7d20c6-ca16-4beb-b627-afb41e3fb491"
+   },
+   "source": [
+    "### *Troubleshooting #1*\n",
+    "\n",
+    "In the cell below, we are visualizing key metrics from the model training and validation process. The plot displays **training loss**, **validation loss**, **F1 validation score**, and **AUROC validation score** over training steps. We expect to see the loss curves decreasing over time, signaling the model's improvement, and the F1 and AUROC scores increasing, reflecting improved segmentation performance as the model learns.\n",
+    "\n",
+    "If training is not working effectively, you might observe the following in the plot displaying **training loss**, **validation loss**, **F1 score**, and **AUROC**:\n",
+    "\n",
+    "- **Training loss not decreasing**: If the training loss remains high or fluctuates without a consistent downward trend, this indicates that the model is not learning effectively from the training data.\n",
+    "- **Validation loss decreases, then increases**: If validation loss decreases initially but starts to increase while training loss continues to drop, this could be a sign of **overfitting**, where the model is performing well on the training data but not generalizing to the validation data.\n",
+    "- **F1 score and AUROC not improving**: If these metrics remain flat or show inconsistent improvement, the model may be struggling to correctly segment cells or classify transcripts, indicating an issue with learning performance.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "43a9c1a4-3898-407d-ac0f-f98b13694593",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-11T22:06:58.182616Z",
+     "iopub.status.busy": "2024-09-11T22:06:58.182357Z",
+     "iopub.status.idle": "2024-09-11T22:07:01.063645Z",
+     "shell.execute_reply": "2024-09-11T22:07:01.063184Z",
+     "shell.execute_reply.started": "2024-09-11T22:06:58.182599Z"
+    },
+    "id": "43a9c1a4-3898-407d-ac0f-f98b13694593",
+    "outputId": "70ba8e1b-7814-497a-c8b6-8aa7295cd4d9"
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "Text(0.5, 0, 'Step')"
+      ]
+     },
+     "execution_count": 88,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "data": {
+      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2oAAAKACAYAAADtg4tbAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/TGe4hAAAACXBIWXMAAC4jAAAuIwF4pT92AAEAAElEQVR4nOzdeXhTZdoG8PtkbZrupaVAadn3TUBQ2TcBBRQRZRAH1HHDEcdlnM8ZRcXRcWQUV0YBBVFHUQRFUAREQDYRAdnBAqVAge57mvV8f4ScnNMkbdqmTdrev+tiJsl5z8mbFEuePM/7vIIoiiKIiIiIiIgoZKiCPQEiIiIiIiJSYqBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGIYaBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGIYaBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGIYaBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGIYaBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGIYaBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGIYaBGREREREQUYhioERERERERhRgGakRERERERCGGgRoREREREVGICdlA7cSJEwgLC0NycnJArvfZZ59h1KhRiIuLQ1hYGNq3b48HH3wQJ0+eDMj1iYiIiIiIAkUQRVEM9iQqMpvNGDlyJHbu3IlWrVrh/PnzNb6W1WrF9OnTsXLlSq/Hw8PD8eGHH+LWW2+t8XMQEREREREFUshl1Gw2G26//Xbs3LkzINf761//KgVpY8aMwddff42dO3fiP//5D2JiYlBWVoYZM2bg119/DcjzERERERER1VZIZdRycnIwbdo0/PDDD9JjtcmoHT58GL1794bD4cCUKVPwxRdfQBAE6fjx48dx7bXXoqCgAEOHDsXWrVtr/RqIiIiIiIhqK2QyauvWrUO/fv2kIE2lqv3U3n77bTgcDmi1WixYsEARpAFAly5d8PTTTwMAtm3bxqwaERERERGFhJAI1G6//XZMmDABGRkZEAQBTz/9NIYMGVLr63799dcAgKFDh6J169Zex/zxj3+Ubq9atarWz0lERERERFRbIRGo/fzzzwCANm3aYOPGjXjhhRdqfc3Tp0/j0qVLAJyBmi8JCQno3LkzALD0kYiIiIiIQkJIBGotW7bEK6+8guPHj2PUqFEBueaxY8ek2x06dKh0bNu2bT3OISIiIiIiChZNsCcAANu3bw/ImjS5zMxM6XZKSkqlY1u1agUAyMvLg9lshl6vr9VzZ2VlITs7u1rnmM1mnDt3DrGxsYiPj0f79u0RFhZWq3kQEREREVHDFBKBWqCDNADIz8+XbkdERFQ61mg0SrcLCwuRmJhYq+deuHAhnn/++Vpd4/Dhw+jevXutrkFERERERA1TSJQ+1gWz2SzdriozZTAYvJ5HREREREQUDI02UJNn6Sq25a9IvpVcXWT3iIiIiIiIqiMkSh/rgrzcsby8vNKx8uO1XZ8GALNnz8bUqVOrdU5aWhpuvvnmWj83ERERERE1fI02UIuMjJRul5aWVjrWdVwQBMTExNT6uRMTE2u9zo2IiIiIiJquRlvnl5qaKt2+cOFCpWNdxxMSEqDRNNrYlYiIiIiIGohGG6h17dpVun3q1KlKx54+fdrjHCIiIiIiomBptIFay5YtkZycDADYsWOHz3HZ2dk4efIkAOC6666rl7kRERERERFVptEGagAwefJkAMDGjRsVG2DLffjhh9JtNvMgIiIiIqJQ0KgDtfvuuw9qtRpmsxkPPPAA7Ha74viJEyfw4osvAgAGDhyIAQMGBGOaREREDZpos8N+uQDWQ2dh/uE3WH4+CdFq8zrWdvoSSt5eh5I319bzLImIGpYG3TkjPT0dbdu2BQAMGzYMW7ZsURzv0aMHZs+ejbfeegvffPMNhg4discffxwtWrTA7t278cILL6CgoABarRZvv/12EF4BERFR9TiKTbCdvAD76csQoo3QdEiCOiUBgkbtObaoDLaj52A9cg62I2fhyC6C4Y5hCBvXt9LnsKVdRPnq3Shfswf2ywXQ9kiFblAX6K7rCk2vVNh/vwjL3jRYf02D9bd0OC7lAw5RcQ0hOhz6Ub2hH9cXmjYJKN9wAOZ1e2E7dt55PCocxgfGQdA16I8iRER1ptH/dnz11VeRmZmJL7/8Ejt37sTOnTsVx/V6PT788EP0798/SDMkIqJAEe0OiAWlcOQUwZFbBEeJGaqIMAjR4VBFGyHEGCEY9RAEwfNchwP2jBzA7oC6bSIElbLoxFFUhvK1e2HZcRSCXgd1crz0R4gyAioBUAkQVAIcOcWwHjsP27FzsB07D/ulfKjiIqBOioWqRSzUzWMAQYBYboFYboVotgI2OyACV/7H+X8OB0SHCIiAWFoO28lMOC7ker5wrRrqNolQRYVDLCmHo8QEsaQcYmGZx1DLrhNQf/13aLuneLx3pk+3wfTZT7AdzlAcs+47Beu+Uyh9a53/P4vCMpSv2oXyVbu8Hy8qg2XHMehH9PT7mkRETUmjD9S0Wi1WrlyJFStWYMmSJdi/fz+KiorQvHlzjB49Gn/961/RrVu3YE+TiKjJEi02OArLIKgECLFGzwCpsAy23zNhS7sIx+UCOLIK4cguhCO7CI4SE1BudQY75RaIpeUemZ2KhEgD1KkJzj+tEyAWlsJ69DxsJy8AJotzTHwkdNd1gX5wV6gSY1C+Zg/Kv9sHlFtq/DrtucWw/36xxudXymqH/feLsFc9ErDZUfT4UsSt+YeUzRJFEcXPfALT/7bVahpCpAGqhCjYz+cCFu+ljwCg6ZIM/YT+0HRuVavnIyJqzARRFCv/F43qxZEjR9CjRw/p/uHDh9G9e/cgzoiIGiNRFL1mk1zH7GkXYb+QB1VSDNStm0FlDFOOMVshFpsgxEf6vI7rWo7MPNjSLsJ+6hJsZ7MgFpbBUWyCWGyCWGSCo6gMYmEpxFKz+0SdBurmMVC1iIWg1zqDs4v5AXntjYkQEQZ1+xYQC0pgP5dTaXAqRBqg6Z4CdWoCyldslx43zpmAiMduAgCULf0Bxc9/pjhP0yMFYTcPhLZPO1j3psGy4xgsv6Q5g1WdBtqeqdD27wBtvw7QdGwBdfMYCOF6AICjpByWLYdQvn4/LD8ehFhqhqZLK+hv7I+wG/pD0z6pDt4VIqLGhYFaiGCgRkSVcQU+oskCiCJEuwOwO6QqOWlcsQn2jCzY0rNhz8iC/XyeMzBylcKVmaFqEQttn7bQ9mkHbZ+2cBSWwbLlEMxbDsORmae4nhAbAXXzGIhlZjjyiiGWlAMAVEkxiHjsJoRNuQ6C2p0Bsx7OQOnb62D56YgyAKtvOg1UzaIgRIRdKQEs9W8+ruDTxz+NQmwEwiZdDcEYBseFXNjP58J+IRdimfPn4ipVFAw6aDq1grZba2i6JkPdJhGOvBI4LuXDfjEfjqwCQBAghOkghGkhhOkA1xozQTaPK6WUEFSAVg1N2+bQdG4FVXK8FCiL5VbYzlyG/dRFiBabM6sVYYAQEQZVbARUreKksUV//8idNdOoEffV3+HIKULB3W9KwZ62T1tEzZ8FTceWHq9fNFthz8yDukUchDCtXz8K0WKDWG6BKircr/FEROTEQC1EMFAjIsAZkIm5xbCfy4Ht9CXYjpyD9UgGbEfPQSw2BXt6HjRdkhHxj6lQxUei9PU1MG84UCfPI0QaoOnUEupW8VAlRjv/JERDFWUAwnQQ9FpnwGMMkwK0ihk/0WqDI78U9nM5sGdkw342C/ZzORAiwqDpkgxt19bQdGoJ0WqDZdcJZwZp+1E4sgqhHdAJhqmDoB/VC4LevwAlFDlKypE79jlpnZu6XRIc2YXS3y1VyzjEff0PqBOigjlNIiICA7WQwUCNqOkQy8yw7Pkd9rNZcGQXwp5V6Fx3dTEf9nM5EMuCmImC88O6I6eo0jVGfl+reQw0HVpA3T4JqvhIqCINEK78UUUbnU0+Ypz/D5sdjksFsF/Mh/1SvrNcrt2VDFKL2EpLLcl/5h3HUHDHax6PC+F6xK78G7TdWgdhVkREVFGjbyZCRBRIot0BR04RVLERfrcVF6022H6/CMtPR2HZehiWvWk1D4J0GkClgqAWAJXKXSJ3haDXQp3SDOrURGezjJQEZyAUEQYhwgBBr4Ht94uwHjgD64HTzu5+GjV0g7pCP7wHdEO7Q50YDdHhgCOrEPbzuXBkFTrL6OIioYqLgGixofT1NSj/6mevU9R0aw3jnAnQXdel2uVuqmgjG0zUMf2grjDcMQymT7a6HxQERL3xJwZpREQhhBm1EMGMGlHoEUUR1l3HUf7tPtjTLztL5jLzAKsdQrge4feMRvh9Y6GKNCjP2X8alp+OwnYyE7bfM2E/cxmw+tWPTyJEGqBOSYCmSytou6dA0yMFmq6tFc8VbNaD6Sj+5xew7jkJwFkGafzLROiv7+PRuZFCS8USyIj/mwLjA+OCPCsiIpJjoBYiGKgR1T3R4YDtcAYcBaXOzYG1aggaNYRwPYS4SKhijRA0ajgKy1D+5U6UfbwV9tOXKr2mEBsB459vhH5UL5jX/QLTl7tgP325yrkIkQZo+7S9st4qBurEaKiaOzstqlOaQRVtDNTLrlOiKMK6Nw2w2qG9phMDtAbElp6F0oXfQtuzDQwzhrG0lIgoxDBQCxEM1Ijqjv18Lkxf7kT5yp3OVua+CAKEGKOzs2It9svySq2CpkcK9MOc5YXaPm2dwSIRERGRF1yjRkQNkqOwDCX/WgnbmcvOJhXNopx/ogwQS8vhKHLu1WVLvwzr7pM+260riCLE/BLlY4IA3fAe0A/v4cx2tW4GVWwEyj7agrLFG3w2/tAO7ATdgI7QdGwJdaeW0LRt3qC7BRIREVH9YqBGRAEjiqJzn6gzWXDkFUPbuy3UrZv5d67dAdux87AdzYD1cAZsRzJgP5sN3fAeiHzmNkUpoP1iHgpmvQnbiQvVn6RKgDq5GUS7HbDaIdrszr3BKjT3EGIjYLhtEAx3DIMmJcHjMhGPToJhxnCUvr3WuS+V1Q51agLCplwHw+Rr/H7dRERERN6w9DFEsPSRGip7ViHK3t8Iy7YjsKVnASZZyaBKgH58PxjvHwttrzZezxdtdpSv3o2S17+RGhtUpGoZh+j/3AXddV1gS7uI/D++7rExc1XUbRJhmDoIYVOuhTopVjkHUYRYemVD59xiiDY7tD3b+L2hrz2rEI68Ymg6t+I6HyIiIgoIBmohgoEaNTT2rAKUvfs9yj7ZCpitVY7XDuwEwy3XONvGt24GVfMYmDccQMmrX8F+qvKGHQAAQYBh2mCUf/srxMIy6WHDjOFQt4iFI6cIjtxiOIpNzk2Po9x7dWmv7gBt/w4MooiIiKjBYOkjEflNdDhg3X8G5V/thunzHT4DNFWzKECjguNSgfSY9eeTsP58UjZIABwVvidSq6DpkgxNjxRou6fAkV+C0rfWATY7IIowffqTe6wgIPK5aQifOTKAr5CIiIgoNDBQI6IqWQ+mo3zNHpR/+6vXkkNVsyiE33s9dNd2hrpNIlRR4RDtDpg3HkDZe9/Duv+050XlQZogIGzyQBj/MsljPZh+VC8UPrJEmXXTaRD9+p8QdkO/QL1EIiIiopDC0scQwdJHChbR4QDKrRDC9V6Pl773PUr+tdLrMVVCNMIfGIfw6UMgGLyfL4oirL+egmnFdtjTLsJ+LgeOnCLpuH5cX0Q8NgmaTq18z9FkRvG/voTp4y1QxUUg+q37obu2czVeJREREVHDwkAtRDBQo2Cw/nYGhY++D/uZLET841YY/3S98vjx88i78QXA7nA/qBKgu6Yz9BOvhmHyNRDCdNV+XtFkhv1CHoSocKgTo/0+z1FYBiFMyzb3RERE1Oix9JGoCRJFEablP6L4n58DVjsAoOTFldB2aw3ddV2dYxwOFP/9YylIUyXFwPjQjdCP6wt1QlStnl8w6KHp0KLa56miw2v1vEREREQNhSrYEyCi+uUoNqHwz4tQ/OynUpAGABBFFD7yvlSWaPpsO6z7TkmHo16eifA7h9c6SCMiIiKiqjFQI2pCrIfPIm/SP2Fet1d6TJUQDeicyXVHdiEKH/sA9qxClLz8pTRGf2N/6If38LgeEREREdUNBmpETYDocKB0yUbkTf4X7GeypMd1g7oi7tu5iPz7rdJjlm1HkDf5XxCLnHuVCZEGRM69vd7nTERERNSUcY0aUSPnyClC4RNLYdly2P2gIMA4ZwKMcyZAUKtgmDkSlp3HYd5wwHnOhVxpaMQTN0PdPKZ+J01ERETUxDGjRtSImXccQ+645xVBmqp5DGI/eRQRj06CoHb+ChAEAVGvzIKqVZzifE2vVBhmDK/PKRMRERERGKgRNUqiw4GSt9aiYMYCxZ5lutG9Ef/dXKmzo5wqxojoN+8FrgRvUAmIeulOKZgjIiIiovrD0keiBsxRWAr7uVyoW8dDFW10PlZQisJH34flx0PugToNIv8xFYY/joAgCD6vp+vXATHvzYbps58Qdsu10PZIreuXQEREREReMFAjaoDsWQUoXbgepv9tBSw2AM59zjSdWsF26pJijZk6JQHRCx+AtkeKX9fWj+4N/ejedTJvIiIiIvIPAzWiBsSeVYiyd9ej7JOtgNmqOOa4VADLpQLFY/rRvRH16t3cKJqIiIiogWGgRtRAmL7+GUX/txwwWZQHVALgED0ei/jrZITfPxaCimvMiIiIiBoaBmpEIU60O1DyyiqUvfe94nFVQjSMs8fD8IchcOSXwHYiE7aTF+C4mA/9hKuh69c+SDMmIiIiotpioEZUB0RRhCO7EKqYCAi6mv9n5igqQ+EjSxSNQYTocBgfnoDwGcMghOkAAOoWcVC3iIN+eI9az52IiIiIgo+BGlGAOApLYdl+DOath2HZegSOywVQd2iB2E8eq9aG0Y6SctiOn4ft6DmULf8R9rSL0jFN51aIXvwQNCkJdfAKiIiIiChUMFAjqgHRbIX1cAZsh886///QWdhOXvBYK2ZPu4iCu99C7Od/hcoY5vt6dgfKFn0P02c/wX422+sY/Zg+iFpwD1QRvq9DRERERI0DAzUiAGKZGdBrq9zcWTRbUfbxVpS+sw5iXolf17YdyUDhQ+8hZsmfIWjUHsftlwtQ+MgSWHef8HkN48M3wvjoJDYGISIiImoiGKhRkyaKIgofXgzz2l8AQYAQa4QqLhKq+Eho2idB26cttL3bQt2uOcrX7EHJa2sUe5RVpGoWCd3Q7tAN6wHr/tMwLdsMALBsOYziZ/6HyJdmKDacNm89jMLHPoCYW6y4jhARBk2XZGi6JiPshv7QXdu5bt4AIiIiIgpJDNSoSbPuTXMGaQAgihDzSmDPK4E97SKsP5+E6X/bnMc0asBmV56s10LbIwWanqnQ9kiFtmcq1B1bSFmvsIlXw3G5AObv9gEATJ9ugxCmhTolAY6cItjOZruf23XJiVcj4omboW7djNkzIiIioiaMgRo1aWZZN8VKyYM0rRqGO4Yh4s83QtUsyucpgkqF6AX3IP9yIaz7TgEAypb+4H2wXovI56bBMG2IIuNGRERERE0TAzVq0uRt740P3wjdiJ7OrNqlfNiOZMB64AxsJzMBuwMAEDZpAIyP3wRNaqJf1xfCdIhZ/BDybvmXzyYh6vZJiH7nfmi7JNf+BRERERFRo8BAjZos+6V82I6dl+6H3XwNNO2TPMaJZWbYfs+EKjEa6hZx1X4eVXwkYj56FMVz/wdHbjFUzaKcf+IjoWmXhLAJ/SGE62v1WoiIiIiocWGgRk2WZeth6ba6dTOo2zX3Ok4I10Pbu22tnkuTkoDYZY/U6hpERERE1HSwWwE1WeYf3YGabkRPrg0jIiIiopDBjBo1aPbMPJi3HYEjpwiOnGKIecVwlJmh698BhjuH+9xkWrTYYNl+VLqvH9GzvqZMRERERFQlBmrUIIk2O8qWbETJa18DFpvHccum31C6eAOMs29A+B3DIIRpFcete9MglpQ77+i10F3TqT6mTURERETkF5Y+UoNjO3EBeVNeRsnLX3oN0lzE3GKUvLACOSP+AdMXOyCKonTMLFufprumMwQDm3kQERERUehgRo0alNJ316Pk1a8Aq3tfM1WzKGiv6QxVfARU8VEQi8pQ9slWwGQBADgu5qPor8vgyCuB8f6xAJT7p+lH9KjX10BEREREVBUGatRglK/f58yiyYTdci0i594OVYxR8Xj4/eNQtvBbZ8B2JetW8vKX0HRqCU3HlrCfzJTG6oZzfRoRERERhRaWPlKDYfpih3RblRSDmA8eRvRrd3sEaQCgTohC5LPT0GzzP6FKjnc+KIoonLMYZR9sco9r1xyaNv5tXk1EREREVF8YqFGD4Cg2wfKTu0tj1H/ugn5kryrPUyfHI2bJn6UNpcVikyJQ0zObRkREREQhiIEaNQjmzQelEkYhxgjdQP+7NGq7JCPqtbu9HtMN5/o0IiIiIgo9DNSoQTCv3yfd1o/pA0FbveWVYeP6wvjoJOWDBh10A9iWn4iIiIhCDwM1CnmiyQzzFnc7/bDxfWt0HePDN0IvO1c/rIfH/mpERERERKGAXR8p5Jm3HpFa7QuRBugGda3RdQSVCtGv3o3i5rFwXM5H5DO3B3KaREREREQBw0CNQp75u1+l2/qRPSHoa54FE8L1iHpuWiCmRURERERUZ1j6SPVGFEWIouj7uMMB25nLEG3uzaxFsxXmzbLNqcf3q9M5EhERERGFAmbUqF7YM/OQP2MBxKJShN81GuH3jIYQppOOW/acRPHcT2E7fh7qds0R8/7D0LRtDsuOYxCLTc5BBh30w7oH6RUQEREREdUfZtSoXpS8+hXspy/BkVOMkvmrkTPyGZhW7YL9cgEKH3sf+bfNh+34eQCA/fRl5E15Gdb9p1H+nazb44ieEAz6YL0EIiIiIqJ6w4wa1Tn75QKUr9mjeMyRmYeixz4AVALg8CyHFPNKkPeHVyGoBemxsHE16/ZIRERERNTQMKNGda5s2WbA6lx3JkQaAK3afVAWpGm6pyDiyVvcx8stEEvNzts6DXQje9XXlImIiIiIgooZNapTjpJymD7ZKt03zpkA/dirUPLvVTCv2wsAEKLCEfHXm2GYPgyCWgVtnzYouP+/7rVpAHRDu0MVEVbv8yciIiIiCgYGalSnyj/fDrGoDIAzm2aYNgSqSANi3rkf1gfHw3b0HPSjekEVHymdo7uuK2I/fxIFs96A43IBAMBw88BgTJ+IiIiIKCgYqFGdEW12lL6/SbrvCtJctD1SoO2R4vVcbddkxH39d5Qt2Qh1yzjob+xf5/MlIiIiIgoVDNSozpjX74PjQq7zjkaN8LtGVet8dVIsIp++rQ5mRkREREQU2thMhOqEKIooXbxBuh82oT/ULeOCOCMiIiIiooaDGTUKOHtWIczf/Qrbb+nSY+F/uj54EyIiIiIiamAYqFFA2M5cRtnSH2DZeRz2tIuKY7pBXX2uRSMiIiIiIk8M1KjWRKsN+XcugON8rtfj4bPH1/OMiIiIiIgaNgZqVGvm7/d7BGmqFrHQXdsZYROuhn5Q1yDNjIiIiIioYWKgRrVWtnyLdFs3sicin50GdUoCBEEI3qSIiIiIiBowBmpUK9bj52Hdc1K6b/zzjdCkJgZxRkREREREDR/b81OtmD7aIt3W9EiB9qp2QZsLEREREVFjwUCNasxRVIby1bul++F/HMFyRyIiIiKiAGCgRjVWvmoXxDIzAECIDkfYxKuDPCMiIiIiosYhZAI1k8mEl19+GX379kVERAQiIyPRq1cvzJs3DwUFBbW6dmlpKebPn4+BAwciOjoaer0ebdq0waxZs/Dbb78F5gU0MaIoouyjLdJ9w22DIRj0QZsPEREREVFjIoiiKAZ7EpmZmRgzZgyOHj3q9XibNm2wbt06dOvWrdrXTk9Px7hx43DixAmvx9VqNV5//XX8+c9/rva1A+nIkSPo0aOHdP/w4cPo3r17EGdUOfOOYyi44zXnHUFA/JZ/sokIEREREVGABD2jZrPZcNNNN+Ho0aMQBAEPPvggNm/ejC1btmDOnDlQqVRIT0/HpEmTUFRUVK1rW61WTJw4UQrSJkyYgJUrV2LLli3497//jejoaNjtdsyZMwdr1qypi5fXaJmW/yjd1g3rziCNiIiIiCiAgt6ef/Hixdi7dy8A4I033sDDDz8sHRs2bBiuvfZa/OEPf8CpU6fw6quv4vnnn/f72v/73/9w+PBhAMCDDz6IhQsXKq49ceJE9OvXDyaTCX/9618xadKkAL2qxs2eVQDzxgPS/fA/jgjeZIiIiIiIGqGgZ9TefPNNAEDXrl3x0EMPeRyfNm0aJkyYAAB46623YLPZ/L72+vXrAQCCIOCll17yON61a1fcddddAICTJ08iLS2t2vNvimxHzwMOZ8WsKiEaumE9qjiDiIiIiIiqI6iB2vHjx3H8+HEAwPTp06FSeZ/OzJkzAQD5+fn48ccfvY7xJisrCwAQHx+PmJgYr2O6du0q3b506ZLf127KHIWl0m11qzgI6qDH+0REREREjUpQP2Hv3LlTuj106FCf4wYPHizd3rp1q9/Xb9GiBQAgJycHeXl5XsecOnVKut2yZUu/r92UiQXuQE2IjQjiTIiIiIiIGqegBmrHjh2Tbnfo0MHnuKSkJBgMBo9zquIqmQSAZ5991uP42bNn8cEHHwAA+vTpg3bt2vl97abMkV8i3VbFGIM4EyIiIiKiximogVpmZiYAQKPRSNkvX1zZLtc5/pg6dSpuvPFGAMDbb7+NyZMnY/Xq1fjpp5/wxhtvoH///igqKkJUVBQWLVpUw1fR9DgKyqTbQjQDNSIiIiKiQAtq18f8/HwAgNFohCAIlY41Gp0BQWFhod/XV6vVWLVqFV555RXMnz8fX331Fb766ivFmLFjx+KNN95A586dqzf5SmRlZSE7O7ta5zSkRiaibI2aKiY8iDMhIiIiImqcghqomc1mAEBYWFiVY12lj65z/HXixAns378fpaWlXo///PPPWLFiBf7+979DownM27Fw4cJqbSPQ0DgK5IEaM2pERERERIEW1NJHV5fHqrJpACCKouIcf2zfvh2DBw/GqlWrkJSUhI8++gi5ubkwmUzYs2cPpkyZgoKCAjz77LO44447YLfba/ZCmhhFMxEGakREREREARfUQC0iwtkxsLy8vMqxrjF6vd6va5eVlWHq1KkoKipC8+bNsXv3bsyYMQNxcXEICwvD1VdfjZUrV+LRRx8FAHz++ed45513avhKmhZHgbyZCLs+EhEREREFWlBLHyMjIwE4g6qquEoXY2Nj/br2V199Je2LNm/ePCQnJ3sd9/LLL+Pzzz/HhQsX8M4772DOnDl+Xb8ys2fPxtSpU6t1TlpaGm6++eZaP3d9kDcTYekjEREREVHgBTVQS01NBQBYLBZkZ2cjISHB51hXt0d/9zr75ZdfpNuuzo/e6HQ6jB8/HkuWLMHJkydRUFDgc3NsfyUmJiIxMbFW1whVosOhaCbC0kciIiIiosALaulj165dpdvyjacrunjxIkwmk8c5lZE3D6kq8JIHVUVFRX5dv6kSi8sBhyjdZ0aNiIiIiCjwghqoDRgwQLq9Y8cOn+O2b98u3b7uuuv8unazZs2k22fOnKl07Pnz5wE4m5rIzyNPDlk2DYIAIdIQvMkQERERETVSQQ3UOnTogJ49ewIAli9fLnV2rOjDDz8E4FyfNmzYML+uPXjwYOn2xx9/7HNcSUkJvv32WwDAVVddhfBw7gtWGTHf3UhEiA6HoA7qXyEiIiIiokYp6J+yH3zwQQDAwYMHMX/+fI/jK1aswLp16wAA9913n99dH8eMGYP27dsDABYsWIBt27Z5jLHZbPjTn/6EnJwcAMBDDz1Uo9fQlHAPNSIiIiKiuhf0QO2+++7DVVddBQD429/+hpkzZ2LTpk3Ytm0bHn30Udxxxx0AgDZt2uCpp55SnJueng5BECAIAoYPH644ptVqsXjxYmg0GlgsFowePRqzZ8/Gd999hz179uCjjz7CddddhxUrVgAArr/+esycObPuX3ADJw/UhGgGakREREREdSGoXR8BQK1WY+3atRg1ahSOHz+O5cuXY/ny5YoxrVq1wtq1axEdHV2ta48YMQJffvkl7rzzThQVFeG///0v/vvf/3qMmzhxIj755BOo1epavZamQCxka34iIiIioroW9Iwa4Gy5v3//fvz73/9G3759ERkZCb1ej65du+Kpp57CwYMH0b179xpde9KkSfj999/x9NNPo2/fvoiKioJOp0OrVq0wZcoUfPPNN1izZo20pxtVTln6yPV8RERERER1QRB9dfCgenXkyBH06NFDun/48OEaB6d1qfj5z1C29AcAgGHmCEQ9Pz3IMyIiIiIianxCIqNGDYe8Pb8qJiKIMyEiIiIiarwYqFG1sOsjEREREVHdY6BG1SIWuJuJCAzUiIiIiIjqBAM1qhZm1IiIiIiI6h4DNaoWR0GJdFuIZtdHIiIiIqK6wECN/CY6HBDlGbVYNhMhIiIiIqoLDNTIb2JJOeBw7+bA0kciIiIiorrBQI38Jspa8wOAEMXSRyIiIiKiusBAjfzmkHd8jAqHoOZfHyIiIiKiusBP2uQ3dnwkIqLaOr//AnJO5UIUxaoHExE1YZpgT4AaDjFf1vExhmWPRERUfVve2IHcM3kwNgvH9X8fiZR+ycGeEhFRSGJGjfzmKJRn1NjxkYiIqqc4qwS5Z/IAAKU5ZYhqHhnkGRERhS4GauQ3lj4SEVFtnP3lnHQ7ulUUYpKjgzgbIqLQxkCN/CbfQ01goEZE1CTYrfaAXevsz+5ALXVA64Bdl4ioMWKgRn5jRo2IqOkwl5jx6X1f4r83fIDjG0/W+noOmwMZv56X7qcOSKn1NYmIGjMGauQ3eaAmRDNQIyJqSOxWO/IzCiA6/Ou2uH/lIWSdyIbdYsfe/x2o9fNfOnYZlhILAECtVSG5T8taX5OIqDFj10fym6jIqLHrIxFRQ+GwO7D6ibW4cOAiOgxrhxvnXV/peGu5Fb+tOizdLzhfCIfdAVUt9s88u8edTWvZqwV04doaX4uIqClgRo38pih9jGXXRyKihiJ9dwYuHLgIAEjbehrmEnOl449+ewLlheXSfbvFjpKskkrOqNrZPRnSba5PIyKqGjNq5DexkKWPRER1zWF34LdVh1FwoRAdh7dHq94tIAhCra554MtDivsl2aXQR+i9P7/NgX0rfvN4PP98IaJaRFX6PJmHL2HrmzsAAOPnjpa6OpYVmHD5RLY0juvTiIiqxkCN/CKKIhwFZdJ9NhMhIqobPy/biz3L9wEADq4+gsTOCeg3rTc6DG0HlcZ3IYwoihAdokd5Yu6ZPJz79YLisZLsUsS3jfN6nd+3nkbRpWKPx/MzCpB6tfdMmCiK2LfiN+xY9DNEu3MN3IZ/bcbUt26GoBKQ8ct54MrSuIgEI+Lbxvp8HURE5MTSR/KLWFIO2NwtmhmoEREF3ukd6VKQ5pJ1IhvfPb8JH97xKc4fyPR63uUT2Vh004dY9of/Ifv3HMWxA7K1Zi4l2aUejwHOgOvX/+33eqzgXKHXx8uLyvHN39dj+393S0EaAFw8fBlHvzsBADi7x92WP+Xq1rXOEBIRNQUM1Mgv8kYiACBEs5kIEYW23Ut/wYoHVyP954yqB4eAgguF+P6lzdL9is02ii4V45unvoO52HN92ba3dqC8sBzFl0uwbu4GaQ1aeVE5jn/v2Vq/JNv7erOMveeRnZYr3e92Qxfpdv75Ao/x+RkF+N+fVuLMzrPuB2Ux2Pb3dsNUYFJsdN2G69OIiPzCQI384pCvT4s0QNCogzgbIqLK5abn4edlv+LS0cv49tkNKLpYFOwpVcpabsW6uRuk9vVagwa3//cWTF9yK7qM6SiVM1rKrDi89pji3KyT2cg8dEm6X5hZhM2vboMoijiy7jhsZpvH8/nKqP366QHpdurA1ugwrK10P/9cgcf4rW/tQPFld9DX5poUTF9yK7QG58qK8sJyrH36e5jyTQAAQSWgdf/kyt4KIiK6gmvUyC/c7JqIGpLc03nSbavJhh/+sw03/+fGoJfcWcosyDx4CZePZ0FQqxAWpYchKgxp284gR5bJGv3kcMS1ca7jGvv0KEQ2j8AvHztLEg98eQh9pvaE+soXZge+9CxtPLn5FFr1aYnfVruPhccZUJbnDJiKvWTULp/IVqxl6/+HPohIcHf4Lb5cApvZBo3e+dHBYXPgwm8XpePX3NUfA/7YD4JKwMBZ/bH9v7sBQBFEJnVLRFik9yYmRESkxECN/CIvfRQYqBFRiKvYDCNj73kc/e4EustK+epL4cUiHP7mGM7vv4DLJ7IV67i86TO1JzqN7KB4rNfkHvj1s9/gsDlQkl2KtC2n0Xl0R5Tlm3Dyh9+lceFx4SjLczZ++nHBT1IDD5Vahf53XIVtb+0EAJTmlKGiExvd12neNRGt+rSEaBeh0qjgsDkAESi4UIRm7ZxNSPLO5kvZOpVahX5/6ANB5QyE+9zaE8fWn0DumXzFc7DbIxGR/1j6SH5x5DOjRkQNR+FFz66FP72zE6W53kv+6oq13IoV96/C3k/249LRrCqDtJY9kzD4gWs8Ho9oZkTnUe7gbd/nByGKIg6vPQq71QEAMMaH4/aFN8MQHeYcJHuqDsPaIqlLonTf2xo1eWljx+HtIAgCVBoVolu6W/IXyNapydvtx7eNlTJtAKDWqDHi0SEez8H904iI/MdAjfzi4B5qRNSAeFuTZi6x4MfXfoIo+g6W7FY7bBa7z+PVlXUyBybZxtEAoI/Qoe11qeg4oj1a92uFhI7NEJkUgdQBrXHD89dLJY0VXTW1l/u6J7Jxfn8mDn51RHqs503dEdUiCtf/faTHub2n9EREoruMsbzIDGu5VTGmMNP9nsW0ipZux7aOkW7nZxQo5uCS2DnB4zlb9W6JrmM7SfcN0WFo7mUcERF5x9JH8ovINWpE1IAUyTJqba9LlboSntqejrQtp9FxRHuPc4qzSvD57NUwFZZj/NzRaD+krceY6io4725pH5sSg3HPjEKz9vEee535I6FjMyT3bYXz+5zryNa/8INU5qjWqtBjYlcAzoYe/ab1xq+fOTetTuycgBbdm8Nhd0BQCRAdzkC1JKcUsckxAADRISrKRaNkWbTY1u6gLV/Wov9yFYEaAAx+8Fpkp+Ui51QuBs7qL5VGEhFR1RiokV8UzURiGagRUegSHSKKL7uDjuv+NADWMqu0B9mWN7ej7aA20OiUmauTm9OkbogbXv4RMzonIFKWhQKc+4zlpedDUAkwxBgQFqmvNPiQB2qJnROQ2Kl2GaW+t/WSAjVXkAYAnUZ2gDHOvW3KtfcOgFqnRt7ZAgy6fyAEQYBao0Z4nEFan1aa7Q7USnNLYZdlEqNbREq3Y2QZNVfpo91mR84pd/MTX5my8FgDbn/3FjhsdujCdTV70URETRQDNfKLopkISx+JKISV5JRK67YAIKpFFEY9OQwfz1wBu9WBsjwTck7lIKlrc8V58tI/S4kFP8zfipteuUHqFGm32vHdcxtxanu6NE5QCQiL0iP5qlYY83/DoQ1T7n0mD9RikqNRW20GpiA2JUZRggg4Sxvl1Bo1rr1ngMf5Ec2MUqAmb9FfmOkObA0xYYqgSp5Rc216nXcmXwrsVBoV4tvF+5yzRqcGdNzShYiourhGjfzC9vxE1FDIyx6dQYcWMa2iFZmh4kuezTTk+4EBwNk953Dk2+MAnK3o17/wgyJIA5zZO1NBOX7/8RSOfnfC45oFF2SBWqvaB2qCSlCsVQOAFj2a+732S9FuXx6oydb0yZuHAMqMmqmwHOVF5Yqyx2bt4jyyk0REVHsM1MgvDNSIKJQ4bA5cPHwJRZc9uzvKG4lEtXAHHVFJ7nK+iu37ASjKJV1+ensnCi8WYcO/NiNt62n3AS/VjvLmGoCzTLLwQmAzagDQ5fqOCHN1dgTQ+5aelYxWikhw//4uVWTUvL9ngLN8UWd0Z9jyzxVW2UiEiIhqj6WP5Bfuo0ZEdaXocjF+33wKANBrcneP8kG5snwTDq89hkNrjqIkqwRqrQp3LL1N0ZlQkR2SrbWKai4L1CoEZaIookiWZRPUAkS7CEuZFZ/+aSXMJRbpWJtrUnDjC2NhN9tweN0xaWPn3DN5imuW5pbBarJJ92MDFKhpw7QYP3c0flq4C616t0AnL41RfJEHavIW/UWZvjNqgiAgJjlaCs4KzhUoG4nUct1dXRNFEQ6Ho+qBRERXqFQqqew9mBioUZVEUVS052dGjYiqw2FzIOv3bAACwmMNMMSEQaVR4czODBxeexRn95yT9vyyW+0Y8Md+HtcovFiE3e//gt+3nFKsP7NbHUjbdgZX33GV9Ji89FGRUWvhO6NmLrHAanK3qx98/zX4aeEu6ZhL636tcOO866HRqaHRqdGqdwvpWG56PkSHKDUXka9PM0SHQR+pr/yNqoaU/sm444Op1T5PHqgV+8ioRVfIqAHOdWquQC3nTJ6ykUiX0ArURFGE2WxGUVERiouLYbFYqj6JiKgCg8GA2NhYREZGQqUKThEiAzWqklhmBqzubmAM1IioOtY+873UHt9FpVbBYffMcpw/kOkRqImiiK+f/NajgYZL9u85ivvyza7l5Y6RzWXrsyqsUSuWBW4avQZX3dYLF367iNM70qXHW/ZMwsQXxyk2do5LjZNu28ptKLpcLAU6gW4kEgjyNWolvtaotfIWqMVIt09tOwOHzfmzU2tViG8b5zE+WERRRHZ2NnJzc6seTERUCZPJBJPJBL1ej9TUVKjV9b8Wl2vUqEpifqnivhAd7mMkEZFScVaJR5AGwGuQBgC5Z/I9HivNLVMEacb4cLQb3Ea6n52mDNSUa9RkpY+yoK34crFi4+viLHfgFtk8AoIgYOTjQ6UMVFK3REz69w3QGpRlmbpwreK6uafd5Y+hGai5v2gryy+D3WaH1WRFWZ5Jerxi6SOgbCgiz77Ft4uHWhsajUQYpBFRXTCbzcjMzFT8m1FfmFGjKskbiQgRYRC0/GtD1FBYyizQGrRBq7W/ePiS+44AqcQRcGZj2g9ph/ZD2uC75zcBcO4NZiowwRBjkMbJg5/wuHDc9fkdKLpUjNNXOjAWnC+EpcwKXbgWNosdJTnu31nyMr5IWUBlKbPCXGxGWJSzKYe8FNKVeTPGh2P6kluReyYPLXu2gErj/bvN+LZx0vm5Z/LQblAbaV4uIROoNZNVRIjOINhS6i75VGtVMMZ7fhknb9Ev52+3yfpgNpsVQZpOp0NsbCyMRiM0Gv67RUT+EUURFosFeXl5KC52/m4vKSlBcXExoqI8v8iqS/zNRVWSr0/jHmpEoU8URZz6KR27P9iD3DP56DquM8b83/CgBGuZh9yBWpfRHTH6/4ajvLAcpkIzIhON0EfoIYoiflzwE8qLzACca72S+7gDNfl6qIQO8VBr1IhpGQ2tQetcVyYCOadz0bJHEoqzit3BoKAsdzREh0Gj18Bmdjb4KLpUIgVq8tb88k2uDTEGJF/VqtLXGN8uDmd2ObOG8oYioRioafQahEWHobywHICz/NF1G3AGsyq1Z0Aac2Vj7IoSQ2h9WlGRO9On0+mQmprKAI2IakSj0cBgMODcuXMoLXV+Di4tLa33QI2lj1Qlka35iRqMc/suYMWDq7Hume+lMsJj60/g9x9PBWU+F49clm636JkEtUYNY7wRzdrFQR/hbK4hCIJinVPF7onyQK1ZO+c4QSWgWQf3JsuudWryRiKRiRGKsjxBEHy26Jfflo/xR3zbWNncne+56BCVe6j5CHSCQdH5MaukykYigLPE09jMM9PWPIQ6Prq++QaA2NhYBmlEVCuCICA62v0lmytgq08M1KhKij3UYhmoEYUim8WOb/6+Hqse/QaXj2V5HP9p4S5FV0NvRFHErvd/wVdPrkPWyZxKx/rDUmZVrB9r0SPJ59i4NrJg53SFQE12v1l7d3CW2KGZdNtboCZfn+aiaCgia9GvXKNW3UDNHWTmZ+TDYXOgJKcUdou7CVNMcv1+C1sZRaCWU6oM1LysT3OJrRBsqnVqxMmC1GBylSq5GI38t4qIas9gcFd3WK3Wel+nxkCNqiQWyPYWYukjUUg6te20okMh4NzvS611/povyS7F3v/tr/Qa5/dnYs/yX3H253PY9s4On+NMheVS17/KXD6eBdHu/EdNZ9Qhvo3vD/WKjFq6u6GI3WZH3ln3/XhZoJbQURaopTmzbvLuhVFJnkGHr4yaPGiLSopAdcS2joGgdpaV2q0OFFwoVJQ9hseFQxeu83V6vavY+VHR8bGSQC0mJUZx31WGGgoq7pPGbBoRBULFTo/1vScjAzWqkmVPmnRb3TJ02jATkdvZX85Lt5u1j8dt79yMm/59A666rbf0+K+f/abInlQkL1PMOpHt9ZvDw2uPYfHNH2L5nZ/BUlZ5hk7eSCSpW6LXtU/SnNspSx9dz52fUSgFhSq1CnGyYCGhoztoyz2dC7vNrsioRXvLqMk7P14J1Gxmm6LroXyNmj80eg1iWrnLY3LP5IXk+jQXeUORkuxSFGXK3rNKM2rK1xHqG10TETV0DNSoUo7CMlh2HpPu60f2CuJsiJoG0SHip4W7sGTKchxYeajq8aKIc/suSPf7/aGPVGZ49Yy+0toiu8UubeLsjXwtmNVkU+yz5bL/84MQHSIKM4twZmd6pfPKlAVqLSspewSAuDbuQM1cbEZpbpnHnGJTYhRrzuLaxEnBn93qQP7Zggqt+b1k1GSlj0VXGogUZ8uqBlQCjAnVrxyouMYupAO1imvU5O9ZZRk1WYt+AEgMoY6P5HT48OGgPG96ejoEQYAgCHj66aeDMofKuOY2Y8aMYE+FqFoYqFGlzD/8Jm12LcRHQjugY5BnRNT47V62F/tW/IbSnDJsW7gT5cXmSscXnC9EiWyNVet+7i6FunAtBj9wrXT/1E9ncPaXc16vIw+KAChKDgHnOrj88wXSffnG0hWJDhGX5I1EqgjUDNFhirbwroYiOadljUTaKzP6Gp1asbYt+/cc5WbXXjJq3kof5ZtfG5sZa1TOV7GhSEMJ1LLTchVr6bxlIV1iU5SvI5Ra8zd1RUVFePDBBzFp0qRgT4WIAoiBGlXK/N0+6XbY9X0gVFK6RES1l7btNPZ8+Kt0X7SLOPfr+UrOADL2uo/Ht4uDMU7Zna/z6A5o0dMdKG17ayfsNrtijM1sUwQXAJB/tkBxv+BcgbTmDFCu66oo72w+zCXO5g6CSkBSt8RKXwNQoaGIK1CTd3yUrU9zkZc/Xjh4UdFqvqrSR3OxGeZSC4oue+6hVl0NKaMWKVuj5tqqAAAMMWGVrqWLbhGF2Culp7EpMYhLDY1GIgQ89thjePfdd+t9/QwR1S1+6iafHCXlMG91l1Hox/cL4myIGr+c03nY8NJmj8fP7vGeAXORlz3Ks2kugiBg+JxBzg2n4QyiMn5RBn+5Z/IgOpRr0vIy8j3GyMn3HqtIvn9as3ZxfjXTiFesU3M+t7wDZLN23gI1d0MReTMVtc65DUBFxrhwqcEK4FynJs+oRQUgUCs4X4jCzNAN1HyVdla2Pg1wrhG85bUJGPXXYZj86gSfG4BT/bPZbFUPqkNt2rSBKIoQRRH//Oc/gzoXosaEv2XJJ8uPhwCL85e/EB0O3bWdgzwjosarvKgca/+xHlaT5weus7+c99kS2GF34LwsUEvpl+x1XGKnBLS7ro10/8LBi4rj2RXKHgHPjJq8GyNQeaB28Yg7UKuq7NGlYlaqvKhcsU5OHsi5JMha9JsKZBs3N4+AoPLc4FtQCYhIVK5TK86SZ9Sq15rfJaZVtBQAig4RdqtDdix0WvMDgN6ogy5c6/G4tzV9FUUkRKDHhK7VbrhCRETVx0CNfCpf7y6/0o/uDUHLdsdEdcFhd+C75ze5OzIKwOgnh0vHS7JKPIIml6yTOVKJoUqtQqveLXw+j/zYRVnGCwByvQRqeRnK58yrkFErulzsM4CUZ9T8DdSayQK1vPQ8qeU+AOgj9Yq1VS4JHTyzbIDvjZsB5Tq14ssVMmrV3OzaRaVRIdZLKWBEghHaMM+gKNjkLfpdqsqoERFR/WKgRl6J5RZYfnSXPYax7JGozpzcfEqxzuy6Pw1A9xu7KIIQXw1A5GWPzbslVlpiKA+YLp/Ihk3WRMJbRq0srwxmWSOTihk1W7lNsSZMOi/fhMIL7k6CLXs09zknOfkaNavJhjM7z0r3m7WPgyB4Zsj0Efoqm4ZUdqzoUrFyjVotMkXyjKCLvG1/KPEW9DJQa3iee+45CIKADz/8EABw9uxZqcPhc889B8Dd8XDZsmX49ddfMWjQIBgMBsTHx2PEiBE4d879u8XhcODzzz/HH/7wB7Rr1w6RkZHQ6/VISkrCmDFjsHDhQpjNns2NKuv6OGvWLGf59fDhAID9+/dj5syZSElJka596623YuvWrXXzJvlp/fr1mDp1Klq3bg29Xo/4+HgMGjQI8+fPR2mpZwdcuR07duCPf/wj2rZtC71ej+joaHTr1g2zZ8/GwYMHfZ5nsVjw3nvvYdSoUYiPj4dOp0NiYiKGDh2KV155BUVFvrdToaaBgRp5Zd56BGKZ85exYNRDN7hbkGdE1Hgd/PqIdLvd4Dbof8dVAICUAa2lx32tU5M3Gknxsj5NLrFTM6h1zo6Gdosd2SezATjb++ecyvN6jiurZjPbvO7BVuSl/FG+f5qxWbiigUdldOE6RMo2mz75o3sPR2/r01zk69RcvAVvLvLyxsLMIkV5ZWQ1N7uW87ahd6itT3MxNmOg1tQcP34cw4cPx86dO1FeXo68vDycOnUKrVo5f29cunQJAwYMwO23347PPvsMZ86cQUlJCSwWCy5fvoxNmzbhoYcewjXXXIOCgoIazWHJkiUYOHAgli9fjnPnzknX/vLLLzF8+HC89NJLAXzF/ikuLsbNN9+M8ePHY+XKlTh//jwsFgvy8vKwc+dOPPnkk+jcuTN+/fVXr+c/99xzGDx4MD766COkp6fDYrGgqKgIx44dw3//+1/06dMHL774osd5eXl5uO666/DAAw9g8+bNyMvLg9VqRXZ2Nn766Sf87W9/Q8eOHXHgwIE6fgcolDFQI6/M693dHnUje0EIwdIdosYg51SuogxxwB/7SZmj1KvdgdqF3y4qOvQBzuBJXmLYuq/39Wkuaq0azbu4uy+69jkryS5VZM6iZWuqXC368zMKPJqNAN47P8oDtRbdk7xmwnyRZ6VKc8qk2946ProkegnU/C19vHwsS9pQGwAiE2tW+gj4yKiFaKAWmchArTF44IEHsH//fkycOBEA0KJFC+zfvx/79+/HAw88oBg7f/58WCwW/Otf/8L27duxZMkSzJ07FyqVCqIoYtKkSVIwcscdd2DVqlXYtWsXvv32W7z44ouIj3f+N3jgwIEaBVRHjhzBAw88gMjISLzwwgvYtm0bfvzxRzzyyCNQqZwfR5955hkcOXKkiisFjs1mw6233oqvv/4aAHDVVVdh6dKl2L17N7799lvcfffdUKlUuHDhAkaMGIETJ04ozt+8eTOef/55AED//v3x8ccfY/fu3diyZQteeeUVJCQkQBRFPP3009i5c6fi3L/85S/S+/3ggw9i/fr1+OWXX7B27VrMnDkTAJCVlYU//OEPsNuVXXqp6eCiI/IgWmwwb/pNus+yR6K6c+ibo9LtxM4Jir2pWvZMgtaggdVkg81sw4WDFxXBW+bhS9IeWFqDxq8W+C17JiHzSiORi4cuAdOULfAjEoxo2SNJKl3Mv5JRy033nnFz7UUmp9jouqd/69Nc4tvGIX1Xhsfjzbw0EpGOdaheRi1KljWTZ9PCovRem2z4y1uzk1AN1CquUVNrVYp97KhhSEpKQlJSEuLinH/3dDod+vTp43Wsw+HAq6++ij//+c8AgEGDBknH1q5di19++QUA8OSTT+Lf//634tzx48djxowZ6NatG0pLS7Fq1Sq88sor1ZprTk4OEhMT8fPPP6NNmzbS48OHD0dKSgoef/xxOBwOfPzxx/jXv/5VrWvX1IcffogNGzYAAKZMmYLPPvsMGo37o/H48eNx44034tZbb0VxcTHuuusuRcC1fPlyAEBCQgI2b96MyEj3751hw4Zh7Nix6Nu3L+x2O5YuXYrrrrsOAGA2m7FixQoAwP3334+FCxcq5nXjjTciMTER8+fPx/Hjx7F7927Fz4uaDgZq5MGy4xjEYpPzTpgO+uE9gjshokbKUmbF8e9PSvd73qQsMVZr1Uju0wpndjnXamX8cl4RqJ371b0+rVXvllBrq96oWb5OLfPwJYiiqGja0axDvKIphiuj5mqXX1FxlrL00WaxI+tEtuz5/FufJj2/l6wUBCCuje9ALbGjZ7atsg6Gvkoxa9rxUXrO5pFSYO0SuoGaMqMWmRQJVSPdJ1MURaDUd4fSemGMqFZmuS5otVrcc889Xo/9/vvvSE1NRXZ2Np588kmvY1JSUnDttddi06ZNyMjw/DLFH08++aQiSHO555578MQTT0AURRw+fNjzxDry2muvAQDi4+PxwQcfKII0l1tuuQX33nsvFi1ahF27dmHnzp1SwHX58mUAQLNmzRRBmkuvXr3w9NNPw+FwYMCAAdLjBQUFsFicTaDatWvndW4PP/wwioqK0L59ezRvXr3fo9R4MFAjD+Vrf5Fu64f3gBCuD+JsiBqvk5vTYCmzAgB0Rh06j+zgMSZ1QGspUDu75xyGzL5WOiZvQNK6b+Xr01zkjT1MBeUouFCI3NOyQK1dPOKubGoMyDJqso6P+gid1GlS3jERALJPZkut6TV6jdf1Y5XxVj4Y3TKq0kyXsZkRhugwmK40NtEZdQiL8v17KyLeCEEtKDbvBmq+h5qLoBIQ1yYOl49lXXkgdMsJIyqsUausVLTBKy1B0eQxQZ1C1OqNQETtvgiorR49esBgMHg99thjj+Gxxx6Dw+GQyhC9SUpyftFjtVqrHOvNmDHefw7R0dGIi4tDbm4uSkrqJ6i+ePEijh51VjTcdtttiIry/d/A/fffj0WLFgEANmzYIAVqnTt3xvr163Hs2DHcf//9ePrpp9G6dWvFua6mLnIJCQmIjY1Ffn4+XnrpJbRo0QK33nqr4ufTunVrvPvuu7V9mdTANc6vz6jGbGcuo/zrPdJ9/fi+QZwNUeN2aI277LHr9Z2gNXgGIykD3OvOcs/koSTb+SGmvNiMrJPuzFXr/pWvT3MJiwpDnCxjdvHQJUVGLaFDvOJ4YWYRbBY78mQdH1NkWb2Ka9QU1+oYD7Wm6iyfXGxKjMf+Z5WtTwOcXe3kAWFUi8hKsxcqjQqRXtrTR9YyUAOUgWZk8who9KH5fWjFjFqoBpQUOK6mIZVxBV52ux3p6enYtGkTFi5ciPvvvx89e/bExx9/LI11OBy+LuNTamqqz2NGo/PvZH1t3i1fC3f11VdXOrZ3797Q6XQe582ePVua96JFi5CSkoI+ffrgb3/7G7Zs2eLztahUKjz++OMAgMLCQvzxj39EfHw8brzxRrz55ptIS0vzeh41PQzUSKFk/mrA5lzzom7dDGHjGKgR1YXLx7MUJYI9JnnvrBrTKlqx3ursL84s2snNacCVhJAh1uC9ZNAHeTlixq8XUHC+ULrfrF08olq6y+BEh4icU7kovOju+NjmmhTpdsWujzny7FwVAZY3Gr1G0czEOaeqX5s8UIv2o8ukt+6OtS19BJzbCLjEpXh2gQwVYdFhUgdQgIFaU1BZxghwBl4ffvghhg4dCqPRiLZt22LMmDF46KGHsGjRIhw+fLjaGbSKXEGNN64vV3ztzRhoubnu31WJiZWv71Wr1VIzlbw8d3VBp06dsG7dOkU552+//YZXXnkFI0aMQEJCAv70pz/h2LFjHtf8+9//jqeffhparfMLOpPJhG+//RaPPPIIOnbsiB49euCVV15BWVmZx7nUdITmV30UFNb9p2H+1t1+1vjEzRD07PZIVBfk2bSWPZN8BiOCICD16tbS+CPfHsfxDSdxfn+mNKb1Va08slCVadkzCUfWHQcApG09LXVzVOvUiEmOhkqjQkxytLQ+7dRPZ6SgUK1TI7lPS+la5YXlsJqsUjYw57T7Q0xlLfUrE982DgXnZMGjHwFf9xu64MCXh+CwO9Dthi5Vjo9qHokLuKh4zN9tBCrTdWxnHPr6KMryytD39t61vl5dEQQBkYkRUpBeMThuVIwRztLDIM8h2CrLMptMJkyaNAmbNm2SHlOr1ejYsSO6d++O/v37Y+TIkXjnnXekBhoNnTwg9Gf9oKvzYsWxw4YNw8mTJ/Hdd99h1apVWL9+vbR2raCgAO+//z6WL1+ODz/8EH/4wx8Uz/nCCy/goYcewueff441a9Zg+/bt0j51R44cwd/+9jcsXrwY27ZtQ4sWLWr9mqnhYaBGAJy/sIr/9aV0X9MjBWETKy8FIKKaMRebceIHd2lLxSYiFaUOdAdq8lb+Ll3GdqzW87eQdWK0yza9jm8bB5XG+Y15bEqMFKilbT0tjYlLjUVEghEqtQoOu7P0qfhyCeLaxEIUReSekmfU/M/yycW3jcOpbWfc1/Ej4ItNicG9q/8Ia7nNY/2VN96CstquUQOAsEg97lx+Oxw2h1/NXYKpz5Se2PrWDsSlxiia1DQ2giAEfX1YqHv66aelIG38+PF48skncc011yAsLEwxrrqdHkOZq1Mm4GyDXxmr1Yr8/HyP81y0Wi0mTZqESZMmAQAOHz6M77//Hl988QV+/vlnWK1W3HfffRg3bhxiY5WZ9qSkJMyZMwdz5syByWTCTz/9hG+//RafffYZLl++jLS0NDzxxBP45JNPavuSqQFi6SMBACw/HoJ1j7v7XMT/TYFQyxIHIvLu+KbfYSt3rl0Iiw5Dh6Heu365JF/VymtHvla9W+DWN29C22t8r/vwJqZVNAwxYR6PyzNXcakx0m15aWR8m1io1CpEyPbhKrqyTq34UrHUHAXw3hjEH/IyTk2Yxu+yPH2E3q8gDXBm1CoKROkj4AwMQj1IA4Det/TAvV/NxB0f3Baya+mo7tntdixevBiAsznGN998g+HDh3sEaQBw7ty5+p5enenZs6d027U1gS8HDhyA1er83da5c2fp8dLSUvz66684ffq0YnyPHj3w+OOPY/fu3XjwwQcBACUlJdi+fbs05uLFi9i8ebNiHZvBYMD111+P119/HUeOHEFysnPt8fr162v4Kqmh4ydxgmh3oOTfq6T7uiHdoB9c+Tf8RFRz8rLFrtd3qvJDst6oQ9tB7mCsZc8k3LJgIqa8MQmtele/HEYQBLTs4bm/mTwDFutjfVXclSBKHugUX1mnJi97jEyKgD6iZh1jk/u2gj7CuXC/w9B21Srr9FfFfdY0eo3X4LWxM0SH1cn7S/WrNmvHsrOzUVzs/LKlZ8+eUKu9f8mQlpYmbdAM1F/Tj7rSokULdOvm/Kzz+eefS++BN65AFgBGjhwJACguLkZ0dDT69++PefPm+Tx37Nix0m2Tybn10X//+1+0bNkSo0aNUgRvcvHx8ejfvz8AoLy83M9XRY0NAzVC+Ze7YDtxZT8mQUDE/00J7oSIGjn5JtGJnf1rXz/mb8Mx6q/DcOubk3DrWzehdd9WtdqXqYWXjagTfGTU5OLbOAM4eYdE1+vJqdDmv6YM0WGY/v5UTPjnWIx6YmiNr1OZih0eI5sHf58roprS651fitSktX10dLS0f9jOnTu9BiwXL17E7bffLq3TAiDtA9aQzZkzB4BzM+57771X8fpcVq1ahSVLlgAA+vTpg+HDhwMAIiMjMWTIEADAihUrFEGsnGtja0EQ0Levs0HbuHHjpOD6H//4h5Stk8vMzMRPP/0EANJ51PSw1qGJE212lLy+RrofdvNAaLunVHIGEdVWkayDYmUbM8vpI/ToMaFrwObgPaPmDq5iZXupybnKGSMVGbUrgdopWSORGnR8lItKikRUAJp7+BKZGAEIkJqkRCYGv9kDUU25NkTOzc3F/PnzMWLECMTGxqJ9+/ZVnmswGDB+/Hh88803yMzMxLBhw/D444+jffv2KCgowLZt27B48WLk5OQozisqKqqyk2Sou/fee/HFF1/ghx9+wIoVK/D7779jzpw56NKlC/Lz8/HFF19g2bJlEEURBoMBn3zyieILnblz52Lr1q0oLy/HiBEj8PDDD2Pw4MGIjY3FuXPnsHTpUnz33XcAnHu1dejg3Cuzbdu2uPPOO/Hhhx9i586d6Nu3r/S8NpsNhw4dwquvvorc3FwIgoCnnnoqKO8PBR8DtSbOsvM4HJlXPlxp1Ih47KbgToiokTOXWlBeZJbu12UwUpmETglQ69RSM5GIBCPCotylf7pwHSISjCjJLpUe04RppPnK29u7Sh9zFR0fa7Y+rb6otWpENHO/vkDsoUYULDfeeCPmzZsHURTx5JNPAgDuuOMOxb5nlXnrrbewb98+XLhwAfv378eMGTM8xnTr1g233347nn32WQDAsWPHpDVUDZVKpcLq1avxhz/8AevWrcO+ffswa9Ysj3Ft2rTBF198IZVKuowYMQILFizA448/juLiYrz00kten2fUqFF4//33FY+9/fbbOHv2LLZs2YLDhw/jvvvu8zhPq9Xi1VdfxQ033FDzF0kNGksfm7jyr36WbuuH94C6tX9lWERUM/KyR7VODWNceFDmodGp0bxzgnS/WQfPDFjFrFpcaqy0nkm+Rq3ocglsZhvyzxW4r1eL0sf6Ig/OghUwEwXC1VdfjS+++AJ9+/aFwWBAVFRUtfbfSk1Nxf79+/HEE0+gS5cu0Ov10Ol0SEpKwqhRo/Dee+/h119/xQMPPCCVSX7++ed19XLqVWRkJNauXYs1a9bglltuQcuWLaXXPnToULz77rs4dOiQtF6sokceeQR79+7Fvffeiy5duiA8PBw6nQ7Jycm4+eab8cUXX2DTpk0ee8hFRETghx9+wEcffYTx48ejRYsW0Ol0iIiIQNeuXTFnzhwcPHgQDz/8cH28DRSimFFrwkSTGebv90n3wyZfE8TZEDUedpsdeen5aNYu3qNRQ7EsUItsHhHURg6pA1sj80q7/1a9PJuSxKXG4tyvF6T78W3dDUbkGbXSnFLknMr12I8t1LW5JgUXDzv3O0od0Hjb01PTMGXKFEyZ4rnG3N8NpBMSEjB//nzMnz/f55iwsDCv66natGnj83mWLVuGZcuWVfn86enpfs2zJvx5DyZOnIiJEyfW6Pp9+vTBokWLqn2eSqXCjBkzvGYwiQAGak2aeeNvEEudJVhCRBj0o3oFeUZEDZ8oivh89lfIOpGNzqM7YNwzoxXHC2Xr06L9XJ9WV66a2gvmEmdDgN639PA4XjGjFt/GXc4YmeAO1ESHiPQ952TjYqX92EJZv2l9ENs6BsZmRjTvkhjs6RARESkwUGvCTF/tlm7rx/eDEKYL4myIAu/ikcs4sPIQ2l6Xgi5jOtXLcxZcKETWiWwAwIlNaRj1xDBoDVrpeNFFd0Yt2OV22jAthjx4rc/jcanKFv1xsoyaRq9BeJwBZXnOdtNndp6VjsWH+Po0F7VWjY7Dq262QEREFAwM1JooR24xLFuPSPcNkwcGcTZEgSeKItbP24SiS8U4+WMamrVvVi8NLiylyrKg/HMFSOzkXgsmX6NWcS+vUFMxUJNn1ABn50dXoOYKToGGsT6NiEJbRkYG8vLyqh5YiZSUFMTFNYwvjoi8YaDWRJWv2wvYHQAAVfMYaAd2DvKMiAKr8EKROygSgUNrjmDEX4bU+fNaTcpALS89XxmohVBGrSrG+HB0HN4Ov285jXaD23jMN6p5BC4fy/I4r7at+YmI5s6diw8//LBW11i6dKnXLo5EDQUDtSaqfLW77DHspgEQ1KG/noSoOi4evqS4f/z7kxh03zXQhWt9nBEY1vIKgdrZfOm2KIoNKqMGAOOfG4PBWSVe9xmL9BFoNpTSRyIiolAWMp/OTSYTXn75ZfTt2xcRERGIjIxEr169MG/ePBQUFNT6+seOHcNDDz2Ejh07Ijw8HFFRUbj22mvx1ltvwWKx1P4FNCC2s1mw7j8t3Q+7iWWP1PhkHrmsuG8ps+LkD7/X+fNayyoGagXSbXOxGZZS9+8bfze7DiZBEBDVPFKxyauLvEW/iyHWELQtB4io8XBtNF2bP8ymUUMXEhm1zMxMjBkzBkePHlU8fujQIRw6dAhLly7FunXrPDYa9Nfbb7+Nxx9/3CMg2717N3bv3o2PP/4Y69evR2xsrI8rNC7yvdPUnVpC041tqanxuVQhUAOAg18fRfcJXb0GHS6H1x7Dqe1n0Pe23mjdt1W1n9dablPcl2fUCmVlj1qDBoboMDRk3jaJDvWNromIiBqKoGfUbDYbbrrpJhw9ehSCIODBBx/E5s2bsWXLFsyZMwcqlQrp6emYNGkSioqKqr5gBcuXL8fDDz8Mi8WCpKQkLFiwANu3b8eaNWtw/fXXAwD27NmDO++8M9AvLSSJoohyWbdHw83XVPqhlaghMpdakHM61+Px7N9zcPm455oql5LsEmx+bRvSd2Vgw79+9Hv/ITlbhUCt8EIR7FY7gAqNRJK8Z6kaEu+BGtenERERBULQA7XFixdj7969AIA33ngDCxcuxIgRIzBs2DC88cYb+OSTTwAAp06dwquvvlqta+fk5OAvf/kLAKB169bYvXs3/vKXv2DQoEGYOHEi1q9fj8mTJwMA1q1bhx07dgTuhYUo27HzsJ9xf1ANmzQgiLMhqhuXjl4GrsRY+ggdWvRMko4d+vqoj7OAvIwCiHbniSVZJSi+XFLt57ZUaCbisDtQcKEQAFAk20MtKin0yx6r4q30sVl7ZtSIiIgCIeiB2ptvvgkA6Nq1Kx566CGP49OmTcOECRMAAG+99RZsNpvHGF/eeust5Oc7y46WLVuG1NRUxXFBEPDyyy9L91etWlXt+Tc09gx3C211p5ZQJ/Pbb2p85GWPSd2bo9dN3aX7JzefQnmx2et5pTllivtZJ3Oq/dy2Cs1EAPc6tYbU8dEf+kg9dEbl/ovs+EhERBQYQQ3Ujh8/juPHjwMApk+fDpXK+3RmzpwJAMjPz8ePP/7o9/VXrFgBABg+fDhGjhzpdUynTp3w5z//GQ8//DD69etXnek3SGJpuXRbFWMM4kyI6s5FWaDWonsSOgxrJ60Hs5ltOPb9Ca/nleaWKu5n/57tdVxlKmbUACD/yjq1htbx0R/y8kdBJXjsvUZEREQ1E9RAbefOndLtoUOH+hw3ePBg6fbWrVv9uvbZs2dx4oTzw9itt95a6di33noLb775JqZPn+7XtRsyscQdqAnGht3IgMgb0SHi4lFZoNajOTQ6Nbrd0EV67NDXR72uPyvJVgZqNcqomTyz/q6GIoqMWgPo+OgPefljTHI0NPqQ6FFFRETU4AU1UDt27Jh0u0OHDj7HJSUlwWAweJxTmUOHDkm35Zkym82Gs2fP4tSpU02uLT8AiCUm6bYQwUCNGp+8s/mwlDj/2xZUApK6JgIAek5yd43NzyjAhd8uepxbmlv70seK+6gBzk2vPfZQawSljwAQmeTOqLHskYiIKHCC+tVnZmamcxIaDVq0aFHp2JYtW+LUqVPSOVWRB3StW7dGdnY2nn76aXz22WdS98iIiAjcdtttePHFF5GUlOTrUtWWlZWF7OzqlUylpaUF7Pkr4yhxr81RMVCjRki+0XV8uzjowp1rqKJbRqF1v1Y49+sFAEDmoUtI7tNScW5pjjKjVpZXhtLcUhjj/S8TtnrJqOWfK0RZXhlsZvex6EZS+thpRHscXH0EANB1bKcgz4aIiKjxCGqg5mr0YTQaq2xTbTQ6PygVFhb6de3cXHdr7gsXLmDixInIylK25S4pKcEHH3yA9evXY8OGDejevXvFy9TIwoUL8fzzzwfkWoGmzKgZgjgTorqReVi+Pq254lizdvFSoFaS5dnRsaTCGjUAyPo9B22rE6h5yajZzDacP+DO4OkjdNBH6v2+Zihr1bslZv7vDxAdImJbxwR7OkRERI1GUEsfzWZndicsrOrMjqv00XVOVUpK3B/Cbr31VmRnZ+ORRx7ByZMnYTabceLECcyePRuAM7N30003obi42NflGg15MxGWPlJjdOmospGInLzxRXGFQE0URY+ujwCQXc3yR6uXZiIAkL47Q7rdWMoeXWJaRTNIIyIiCrCgBmquLo/+bPrqWvjvqzNkRSaTO3N07tw5vPHGG3j99dfRsWNH6HQ6dOrUCe+88w6effZZAM592t56663qvoQGRyxmMxFqvEwFJuRnFEj3W/RQZtQUgdpl5Rcz5YXlcNgcHtes7jo1eemjoHb/bju7RxaoNZJGIkRERFR3glr6GBHh/NBUXl5exUj3GL3ev3IheZaud+/eePjhh72O+8c//oFFixbh4sWL+N///oe///3vfl2/MrNnz8bUqVOrdU5aWhpuvvnmWj93VRyy0kdVJAM1alwuHnWXNxtiDYhuqQyIIhPdgVrRpRKIoih9UVSS65lNA4Csarbol5c+xreJQ84pZxm2qcD9e66xZdSIiIgo8IIaqEVGOj+slJV5/4AkV1rqXDsSG+vfHj2uawPAxIkTfY7TarUYPXo0PvroIxw5cgRlZWUIDw/36zl8SUxMRGJiYq2uUVcUpY/MqFEjc+mIu5FIi+7NPbL1kbJW8laTFZYSi7RWrFTWml9r0EoljMWXSmAqLJf2YauKvPSxeZcEKVCTayx7qBEREVHdCWrpY2pqKgDAYrFU2SXR1e2xZcuWlY5zkXdxrOqc5ORk6ba8CUljpNhHjc1EqJG5WEkjEQAwxIRBrVNL94tk69TkjUQSOsQrArPs3/0vf7SWu0sfm3fx/oUNSx+JiIioKkEN1Lp27SrdPnXqlM9xFy9elNacyc+pTI8ePaTbru6Svsj3U4uJifHr+g2VYo0aSx8pQERRhKUsuPsSOmwOXDruLn1s0cNzyw1BECqsU3MHavJGIhEJRiR0bCbdzzrp/iLJYXfg6HfHcXpHusem2aJDhE0WqCV2TvA6V5Y+ElFDt2zZMgiCAEEQsGnTJsWxNm3aQBAEDB48uEbXnjVrFgRBQJs2bQIwU++OHDni9fHazp0okIIaqA0YMEC6vWPHDp/jtm/fLt2+7rrr/Lp23759odE4Kzv37NlT6VjXnmtxcXGKksnGiKWPFGgOmwOfz16N9yYuw8GvvP/DVx+y03KkIEmlVqG5jyBJvk5N3lBEvoeaMd6IhE7uQE2eUftp4S5sfHkLvvn7epzdc05xbXk2DXA2LzHEeP53xkCNiCg4zGYz5s6di6uuuirYUyGqUlADtQ4dOqBnz54AgOXLl3t8O+3y4YcfAnCuTxs2bJhf146Li8OoUaMAAOvXr8fZs2e9jrt8+TJ++OEHAMCkSZOqNf+GRrQ7IJbJN7xm6SPVXuahi7h0NAsOmwMHvjwUtHlc+M29T1li5wRo9N6X4MrXqclb9JfKmokYm4UjUZ5RuxKo5abn4bfVh6XHLx1T7s1YcQ81bZgGcW2U62oN0WHQhWurfD1ERBR48+fPxwsvvACr1ftWKkShJKiBGgA8+OCDAICDBw9i/vz5HsdXrFiBdevWAQDuu+8+v7s+AsDjjz8OwPntycyZMxV7qwGA1WrFPffcA7PZDEEQcN9999X0ZTQI8mwawNJHCozyYnfwbyowVTKybp3fnyndTr7K97pUX6WPJbKMWkQzIxI7uTNyBecKYS61YPt/d0O0u79QspQqyz0Ve6gJgEavQVyKMlBjIxEiauzS052l4fKKqFBhs9kqPR7Kc6emJ+iB2n333Seln//2t79h5syZ2LRpE7Zt24ZHH30Ud9xxBwBnzfBTTz2lODc9PV2qjx4+fLjHtceMGYN77rkHALB161b0798fixYtwq5du/DFF19gyJAhUhD4wAMP4Nprr63DVxp88kYiAEsfKTDk+4aZSy0+M+N1yWF34MJBd0at0kAt0fum16WyZiLGZkZEt4yCzqiTHtv32QHFptWAl0BNVvqoDdNCEATEpVYM1NhIhIiIiKoW1Pb8AKBWq7F27VqMGjUKx48fx/Lly7F8+XLFmFatWmHt2rWIjo6u9vXfe+89CIKAJUuW4MSJE7j//vs9xvzxj3/E66+/XtOX0GCIsj3UEKaDoFH7HkzkJ3m5n2gXYTXZ6r20LzstVwqaVGoVWnT3bCTi4i2j5rA5UJbn/u/D2CwcgkpAQsd4XDjgDAD3fLTP41qVZdS0Bud7EJsaoxjD9WlERETkj6Bn1ABn+/z9+/fj3//+N/r27YvIyEjo9Xp07doVTz31FA4ePIju3bvX6NpqtRqLFy/G5s2bcfvttyM5ORk6nQ7JycmYMGEC1q1bhw8//BA6na7qizVwDllGTRXhfwkpUWUU5X4ALKVmHyPrzoUD7rLH5l0SKg0U5YFaaU4p7DY7ygpMEB3uTKAxzggASOwoa0jiJVForjRQc34P5plRY6BGRIHx6quvSpVF27Ztq3TsI488AkEQoNVqkZXlXl9bUlKC119/HePGjUNycjIMBgOMRiNSU1MxdepUfPPNN9WeV1WdEx0OB7744guMGjUKLVq0gNFoxFVXXYU333wTdrvdr+dYv3497rrrLnTu3BkxMTHQ6XRITEzEkCFD8PLLL6OoqEgx3tWl8vnnn5cec713s2bN8nvuALBr1y7MnDkT7dq1Q1hYGGJiYtCvXz/MnTvX5zZP8iqwLVu2wGKxYMGCBRgwYABiYmIQERGB3r17Y968eSguLvZ6jUCp6c+8sk6fcoMHD/batbPie7BhwwZcddVVCAsLQ/PmzTF+/HiPvZVr8l7LnT9/Hs8++yz69u2L2NhYGAwGdOnSBQ8//DDOnDlT9ZsVZEHPqLmEhYXhySefxJNPPun3OW3atPG7zGrEiBEYMWJETafXKHAPNaoLFTsdmkstiPDecLHOyNentepT+b6JkQnuQE10iCjNKVOsrdMZdVKglyjr/OhiiAmDqcD535KlVBmk2iqUPgLOVv+6cC0sZc6xzKgRUaBMnz4dTz75JBwOB1asWIGhQ4d6HecKjADnspDEROcejzt27MDkyZO97mWbkZGBjIwMrFy5ErNmzcLSpUsDMufS0lLccsst2LBhg+LxAwcO4JFHHsGqVavQvLnnPpguxcXFmDp1Kr7//nuPY9nZ2cjOzsb27dvx3nvvYcuWLdKevYFgs9nw0EMPYdGiRYrHzWYz9u3bh3379uGtt97Cp59+inHjxvm8Tn5+Pq699lrs26es1Dh48CAOHjyIZcuW4aeffkKrVq0CNneXYPzMvdm6dSteeOEFKTDPyspCUVERwsPDAQTmvV6xYgXuuecelJaWKh4/ceIETpw4gQ8++ACffvppSDcTDImMGtUPZaDG9WkUGLYKGTVzSf3up1ad9WmAs8FHeJz7i4riyyXKjo/x4dLthAqBmkavwbX3uLcVqbz00fk9mCAI6Da+MwBnNi+5ikCSiMhfLVq0wMiRIwEAK1eu9JmN2rp1Ky5edP6enDFjBgDnHrXjx49HdnY2IiIi8Le//Q3r1q3D7t27sWrVKsyePRtarfMLp2XLluHbb78NyJxvv/12KUgbNmwYVq5cid27d2Pp0qXo1q0btm7dipUrV/o8/+6775aCtBtuuAGfffYZduzYgQ0bNmDBggVISUkB4Mze/PWvf5XOmzRpEvbv369YArN//37s378f8+bN82vu999/vxQ4tG/fHgsXLsTOnTuxceNG/OUvf4FOp0NBQQEmTpxYaYZz9uzZ2LdvH8aNGye9/uXLl6Nbt24AgDNnzuCxxx7za07VEayfuTfz5s1DZGQk3nnnHWzfvh1vvPEGnnjiCel4bd/rdevWYdq0aSgtLUVkZCSefvpp/Pjjj/jxxx/xzDPPIDw8HGVlZZg2bRpOnDhRZ6+ztkImo0Z1TxGosZEIBYilYuljPQdq1Vmf5hKZGCGtSSvOKpayXYAzA+YS2zoGmjCNlCnrO623opTRXKHMs2IzEZehDw9Ctxu6ICY52ue2AURENTFjxgxs2rQJWVlZ2LJli7Q1kdxnn30GAIiIiMDNN98MAPjPf/4jldh9/vnnGD9+vOKcyZMnY+TIkbj11lsBAKtWrcINN9xQq7l+++23UhO36dOn4+OPP4YgCACAgQMH4rbbbsPYsWN9dlw8ePCgFMTddtttWLFiheL4mDFjcPfdd6Nbt264cOECvvnmG1itVmi1WsTFxSEuLg5JSe5/I/r06eP33H/44Qd88MEHAIBBgwZh/fr1iIhwV2iMHj0aU6dOxejRo2EymXDnnXciLS1NCnzkLl26hCeffBL//ve/pccGDhyIm266Cd27d8f58+fx1Vdfobi4OKD7+wbjZ+6Lw+HARx99hAkTJgBwvqcutX2vy8vLMXv2bADO7bq2b9+Orl27SucPHz4cgwYNwg033ACTyYR//vOf+Oijj+rkddYWM2pNiLyZiBDJ0kcKDJuX0sf6VJ31aS6KvdQul6A0R55RcwdqKrUKvW/pAQBI6NgM/ab1gc7ovn7F0kdvzUQAZ1YtoUMzRfBGRPVDFEU4CsuC+qcuu+FOmTJFKherGLgAzhKyVatWAXB+EHeNzczMRPPmzdG/f3+PD+wukydPlrZFysjI8DqmOhYvXgzAGTAuXLhQCtJcwsPDsWzZMqhU3j+eHj16FB06dIBOp8MzzzzjdUxUVJT0esrLy72W+NXEa6+9BgDQaDT45JNPFIGDy3XXXSfNKyMjQyo3rSg+Pt5rFi8qKgq33XYbAMBiseDkyZMBmbtLMH7mvrh6RXhT2/d648aN0tz/+c9/KoI0l7Fjx2LMmDEAnIFhMDpW+4Nf7TYhDtk+aiojm4lQYHisUSup32Yi1Vmf5qJo0X+5BA67Q7pvbBauGDv4/mvQZ0pPGGLCoNaooZe17LearHDYHVCpVdJ9F20Yf70ShQKxyITs3o8EdQ4Jv70BITq86oE1EBERgUmTJuGzzz7DqlWrsHDhQmg07t8/GzduRE5ODgBIWx4BwKeffgrAmdnwRaVSISEhAefPn0d5ebnPcf6w2Wz44YcfAADjxo3z2cm7ffv2GDx4sNdytmnTpmHatGlwOBw+gzkAiqxZbecNOPfd3bp1KwBn1q6ydW/33nsv/vGPf0AURWzYsAHTp0/3GDN06FCf+wK3bdtWul1x/9/aqu+feWWuvvpqr48H4r12ZW0FQfD6/rssXrwYDocDKSkpHl8ahAp+kmhC2EyE6oJn18f6y6hVd32ai6JFf1aJouNjhCyjJj3WzP2YfG81ALCWWaGPdP6DKw9aNQZmz4iofsyYMQOfffYZcnNzsWnTJkVzBVfZY1JSEkaPHu1xrivgKS8vx5kzZ5CWlobjx49j//792LZtGy5cuACg8g/3/rhw4YJUdte7d+9Kx/bv37/SNV6uOYuiiAsXLiAtLQ2///47Dh06hF27dimadNR23gBw9uxZqSGFrwDDpVmzZmjXrh1OnTqFI0eOeB1TWfBhNLr/valqc+6aqq+feWV8NUoJxHvtykS2bdu20q29WrduXd1p1zsGak2IWCwrfWQzEQqQioFafTYTqcn6NMBzLzVB5f4mrWJGrSJduDJQM5da3IGa7L3QMVAjonoyduxYJCQkIDs7GytWrJACNbPZjK+++gqAMxulViv3Ty0qKsIbb7yBFStW4NixY14/mAuCEJCyMPmWAHFxcZWOrazrIwB8/fXXWLhwIXbs2OHR0Q9Apdm2mpC3gXd1zKxMYmIiTp06hby8PK/H5cFYRfLMTl2U49Xnz7wyUVFRXh8PxHvt+rsWHx9fy1kGHwO1JkQsZddHCrxglj7WZH0aUKH0MasEao37H3VjM9//gAKASqOC1qCB1eR83fIMonzzbw1LH4lCghBlQMJvbwR9DnVJo9Hg9ttvx9tvv42vvvoK7733HnQ6Hb799ltpPzFXt0eXtLQ0jB49GmfPnpUec+0x1atXLwwcOBBjx47FyJEjFWPqg7cGHIAzwzNz5kx8/PHH0mOCIKBt27bo3r07+vXrh+HDh2PTpk345z//GbD5yIMWf0rkXN03Q62crr5+5v5k4ny9N4F4r+sqExkM/CTRhCg3vGbpIwVGMEsfa7I+DVA2E6k434gqAjXAmVVzBWry5imuxwBm1IhChSAIdbY+LJTMmDEDb7/9NgoKCrBhwwZMmDBBai7SuXNn9OvXTzF+2rRp0ofxRx99FLNmzUL37t09sm6BWiclXzdWVYOP/Px8r4+/8847UpA2cOBAPPvssxgyZIhHs4mabNJdGXkGUJ4Z9MU1pqrMYX2r7c/c32xfxc3GqyMQ73VsrLM7s6+MZkPCro9NiFjMjBoFnjyLBNRf6WNN16cBzk2r1Tq112PhcVV/oJOvU1Nk1EzMqBFRcAwcOBAdO3YEAKxZswYmkwlr164F4JlN++WXX/Drr78CAO655x689tpr6NWrl9cP7AUFBQGZX8uWLRETEwMA2Lt3b6Vjf/vtN6+Pv/vuuwCcH8Q3btyI8ePHe+0IeO7cudpNtoJ27dpJ5Yq//PJLpWOzsrKQnp4OwBkgh4pA/MzlTWpMJpPXMaIoIjMz0+sxfwTivZbvR+daF+nNihUrkJKSghEjRgS8w2agMFBrQlj6SHXBo/SxnjJqNV2fBji/FZSXP7oYosOg8RHAyfkM1OT7qDGjRkT1zNXVce3atVi/fr20fkve7REATp06Jd2+6qqrfF7viy++kErLaltOplarpXbsGzZs8PlhPicnR9oQuyLXvNu3b+9zf7G8vDxs3LhRul9x3jVZv6bRaDBs2DAAytbv3ixZskS67dqMPBQE4mfuCrQB36379+7d6zMj6o9AvNeu2w6HA59//rnP89euXYtz585h586daNnS/y976xMDtSZEsY8aSx8pAERR9Cx9rKeMWsbe89Lt6qxPc5E3FHGpan2ai7xFv9lHRo2BGhHVN1fm7OLFi9IarUGDBilavgPKJgvr16/3eq2ff/4Zjz/+uHTfYqn97/aHHnoIgiDAYrFg5syZMJuVa5ptNhvuuecen23hXfM+cuQIzp8/73G8qKgI06dPVwQKFectb4tfnbLOOXPmAHC2j7/zzju9NjHZuXOn9L63bNlS2hMtFATiZ96zZ0/p9pIlSzx+fiUlJXjkkdpvhVHb9/rWW2+VGpH84x//wJkzZzzO37ZtG/73v/8BAKZOneo1MxsKGKg1IYqMGvdRowCwWx0Q7co69fpqJnJq22npdsrVydU+X75OzcUY7986Ft8ZNe6jRkTB0759e1xzzTUAILWor5hNA4DBgwdLH2TXrl2LKVOmYM2aNfj555+xevVq3HXXXRgyZIgi4KnNuiOXa665BrNnzwYAbNq0CQMGDMDHH3+MPXv2YMWKFRg8eDDWrFkjbcpd0ZQpUwA4y+6GDx+OxYsXY+fOndi4cSNefPFF9OjRA99//73inIrzlneUnDt3Lvbu3YujR49WOfexY8di1qxZAJwf8vv06YN3330Xu3fvxubNm/HYY49h5MiRMJlMUKlUWL58eaXdHetbIH7mKSkpuO666wA4y1NHjx6N1atXY9euXXj33XfRr18/7Nq1C+3atavVXGv7Xuv1eqlM9vLly7j66qvx8ssvY8eOHdi4cSOeeuopXH/99XA4HIiJicFLL71Uq/nWJX6SaEIcsjVqqkhm1Kj2bBXWpwH100yk6FIxLh93L0bvMKx9ta9Rm4yaP2vUmFEjomCYMWMGdu/eDcDZPdFbVsdgMGDp0qWYPHkyLBYLVq1ahVWrVnmMu/nmm2E0GvHJJ58gIyMDZWVlPoMof73xxhuwWCxYvHgxDh48iDvvvFNxvE+fPpg0aRLmzZvnce68efPw448/4vDhwzh16hTuu+8+jzHJycl4/PHH8eijjwIAjh07hsGDB0vHR40aBYPBAJPJhAULFmDBggUYNGgQtm/fXuXcFy1aBLVajffffx9paWl48MEHPcbEx8fj448/xqhRo6q8Xn0K1M988eLFGD58OLKzs7F9+3aP9+3RRx9FXFwcnnnmmVrNt7bv9eTJk/H+++/jwQcfRG5uLp566imPMc2bN8fq1auRkpJSq7nWJWbUmgjRagPM7g+RXKNGgWAxeQnUyqxw2Otuk0wASNvqzqbFpsQgvm1sta/hbY2aPx0fAUBndAdhllL3e6BYoxbGQI2I6t+0adOk9vbjx4/3uZfUDTfcgL1792LGjBlITk6GVquFwWBA27ZtMWXKFKxbtw6rV6+WslhWqxWrV6+u9fzUajUWLVqEjRs3YuLEiUhOTkZYWBg6deqEuXPnYufOnT7Xn8XExGD37t14/vnn0bt3bxgMBmi1WiQkJGDIkCGYP38+Dh8+jIceekh63RXXKLVu3RrfffcdBg0ahIiICBiNRo8SPl+0Wi2WLFmCbdu2YcaMGUhNTYVer0ezZs0wcOBAzJ8/HydOnFBsOB5KAvEz79atG44ePYonn3wSnTt3ll7/+PHjsW7dOrz22msBmWsg3uu7774bx48fx5w5c9C5c2eEh4fDYDCgZ8+e+Mc//oHDhw/j2muvDch864og1vWOduSXI0eOoEePHtL9w4cPo3v37gG7vqOgFNl9/iLdTzj4BlRRjb9dMdWtvPR8fDRzhcfj96+9C2GRdVde+/ns1bh45DIA4Oo7++K6Pw2o9jUyfj2P1Y+tVTw24rEh6HVT1f/d7V76C35e5uye1eX6Thj7j5EQRRFvj1osBal//HgaYlvHVHteRFR9drtd0bWtU6dOHt3siIiqK9i/W5hRayLEYmUbVcHIjBrVXsXW/C6WOlynVpxVIgVpANBxWM1q4b2VPvqdUQuXlT6WOUsf7VaHIpPI0kciIiKqDQZqTYR8s2shXA9BzR891V7F1vwudbmX2qlt7u5N0a2i0KyD97KeqkQmeFmj5m8zkQhZoHbltVZcr8dmIkRERFQb/CTRRHAPNaoLFVvzu9TlXmq/y9andRjWDoIg1Og6Gr0GhlgDTPnubHNt2vNXXK/HNWpERFRbFovFr66UldHpdNIm0NSwMFBrIuSlj9xDjQLFV6BWV50fS3NLkXnoonS/pmWPLpGJEVKgJqgEhMf499+GouvjldJHm8mdXVTr1FBpmLUmIqLayczMrHSDan+kpqYiPT09MBOiesVPEk2EMqPGPdQoMHyWPhbXzRq1tG1ngCvtjyKTIpDYOaFW14tKcncWC48z+B1cyTNqrtJH7qFGREREgcRPE02EfI2aihk1CpD6Ln1Mk61P6zisfY3LHl3kLfqN8f5vTOqtmQj3UCMiokBr06YN2KC96WJGrYkQS+Slj1yjRoHhK6NWF6WPZQUmXDiQKd3vUMuyRwCISY6Wbke38L5vjzfyZiJ2qwM2s417qBEREVFAMVBrIsQSNhOhwKvY6dClLro+nv7pDESH81vFiAQjkrom1vqaXa7viFa9WyC6VRT6Tuvj93nyjBrg3ORbmVFjsQIRERHVDj9NNBHKQI2ljxQYFTsdupjrYB+19D3npNsdhrWDoKpd2SPgDLhuffOmap+nNWggqAQpcDSXmmE1MaNGREREgcOMWhOhCNS42TUFiLzToaB2B051UfpYXuT+O9ysfc32TgsUQRCgC3cHY5ZSq6KZiIYZNSIiIqolBmpNhLKZCAM1Cgx5cGKMc28WXReljzbFGrDgB0KKTa9LLYrSRx2biRAREVEtMVBrIhTNRCJZ+kiBIS/3M8bLArU6yKjZzHbptkYf/EBNb3Rvc2EusSiaiWhY+khERES1xECtiVDso2bkPmoUGPKMWkQzd3v7ulijZrMoN5QONkXpY5kyoxYKGT8iIiJq2BioNRFiMZuJUOApMmoJ7kCtLtao2cwhXPpYYuE+akRERBRQDNSaCAf3UaM6IA9O5BtG1/UatVAofay46bViHzU2EyEiIqJaYqDWRIil7lI0NhOhQFGUPsoyanaLHTaL3dspNSa/nkYX/EBIb3QHauZSZtSIiIgosBioNQGiKCqbibD0kQJEnkWSr1EDAlv+KDpE2OWBmj4E1qgZK3R9LOc+akRERBQ4DNSaArMNsLo/5LL0kQJFnkUyxBoUe6mZSwPXUES+Pg0Ija6KFQM1WzmbiRAREVHgMFBrAuQdHwEGahQYokNUrBvThWsV67bMxYHLqHkEaiGQUWPpIxEREdUlBmpNgLyRCAQBQjjb81PtVQyetGFa6CtsAh2w56qw3i0U1qh5lD6a5M1EGKgRERFR7TBQawKUrfnDIAhCJaOJ/CNfkwU4y/0UWaYq9lLLPHQRZ/ecgyiKVT6XPChUaVRQaYL/q6tiUKrMqAU/kCQiqkvLli2DIAgQBAGbNm1SHGvTpg0EQcDgwYNrdO1Zs2ZBEAS0adMmADP17siRI14fr+3cG4Ldu3fjxhtvRGJiIsLCwpCamoq5c+dWes7SpUshCAJmzJhRT7MkAOCniSZAsdk1yx4pQOSBCQRny3x9hDtba64ko5bx63msfmwtAGD4Xwaj9+QelT5XqLXmB5QbXptLrWwmQkTUAJjNZrz44ot4+eWXYbEEfiuZUPfLL79g+PDhMJvdX6ZmZGTAYPDdaC49PR1//etf62N6VEFofOKhOqXo+GhkoEaBocgghWkgqASPTaB9Ob/vgnT78DfHqg7UzPJALfjr0wBAZ5QFpcXmkNuQm4iIPM2fPx8vvPBCsKcRNK+88ooUpD333HMYM2YMzGYzOnbs6HX8pUuXMGbMGOTm5tbnNOkKfppoAhwl7owa91CjQJFnkFxdGJWlj74DNUuZO8jLOZWLokvFiEqK9Dk+1PZQAwCd0Z01U2QXwTVqRNS0paenB3sKPtlstkqPh/LcA+H48eMAgOuuuw7PPvtspWP379+PW265pdG/J6Es+As9qM6JskBNiOQeahQYFTNqQMV1W77XqFnKlEHcmV1nK30uRUYtRLJV8qC0IgZqREQUikpLSwEAbdu29TnGarViwYIFGDRoENLT06FSMVwIFr7zTYAiUGPpIwWItdyzHb2iHLCSNWryjBoAnNlZRaAWimvUfARqgkqAWhca5ZlERERyDocDAKDReP+3ND8/H7169cJjjz0Gk8mEiIgIrF69uj6nSDIM1JoANhOhuqBoR+8qfYzws/SxQhB3fv8FjyybnCKjFiJBkFqr9hqQacM07KxKRPXi1VdflTovbtu2rdKxjzzyCARBgFarRVZWlvR4SUkJXn/9dYwbNw7JyckwGAwwGo1ITU3F1KlT8c0331R7XlV1TnQ4HPjiiy8watQotGjRAkajEVdddRXefPNN2O12r+dUtH79etx1113o3LkzYmJioNPpkJiYiCFDhuDll19GUVGRYryrS+Xzzz8vPeZ672bNmuX33AFg165dmDlzJtq1a4ewsDDExMSgX79+mDt3rs+1XOnp6dLzbdmyBRaLBQsWLMCAAQMQExODiIgI9O7dG/PmzUNxcbFf74G/5M999qzzi9EPP/zQ6+svLCyUyiOHDRuG3377DZMmTQrofMh/ofHVNNUpsVjWTISBGgWIt3b0/jYTqbimy2514Owv59FxWDuv4xVr1EIkowY4s2omi0nxmIYdH4monkyfPh1PPvkkHA4HVqxYgaFDh3od5wqMAGDMmDFITEwEAOzYsQOTJ09Gdna2xzkZGRnIyMjAypUrMWvWLCxdujQgcy4tLcUtt9yCDRs2KB4/cOAAHnnkEaxatQrNmzf3eX5xcTGmTp2K77//3uNYdnY2srOzsX37drz33nvYsmULUlNTAzJvwLm+7aGHHsKiRYsUj5vNZuzbtw/79u3DW2+9hU8//RTjxo3zeZ38/Hxce+212Ldvn+LxgwcP4uDBg1i2bBl++ukntGrVKmBzr47evXvjmWeewZQpU4Ly/OTGjFoTwGYiVBe8lT4qmolUVvpYavV47MzOdJ/jQ3GNGuB9nRr3UCOi+tKiRQuMHDkSALBy5Uqf2aitW7fi4sWLACDtg3Xx4kWMHz8e2dnZiIiIwN/+9jesW7cOu3fvxqpVqzB79mxotc7f7cuWLcO3334bkDnffvvtUpA2bNgwrFy5Ert378bSpUvRrVs3bN26FStXrvR5/t133y0FaTfccAM+++wz7NixAxs2bMCCBQuQkpICwLOl/KRJk7B//37cf//90mP79+/H/v37MW/ePL/mfv/990tBWvv27bFw4ULs3LkTGzduxF/+8hfodDoUFBRg4sSJlWY4Z8+ejX379mHcuHHS61++fDm6desGADhz5gwee+wxv+bkj5YtW0qvtUWLFgCAiRMnen39KSkpOHDgAIO0EMFPFE2AsvSRzUQoMGzlnu3oFfuoVbLhtbcyx/RdGXDYHVCpPb8/UnSYDJGuj4D3dWpsJEIUWkRRrDTDXx90Ebo6K4meMWMGNm3ahKysLGzZsgWjRo3yGPPZZ58BACIiInDzzTcDAP7zn/9IJXaff/45xo8frzhn8uTJGDlyJG699VYAwKpVq3DDDTfUaq7ffvst1q1bB8CZDfz444+l92XgwIG47bbbMHbsWGzfvt3r+QcPHpSCuNtuuw0rVqxQHB8zZgzuvvtudOvWDRcuXMA333wDq9UKrVaLuLg4xMXFISkpSRrfp08fv+f+ww8/4IMPPgAADBo0COvXr0dERIR0fPTo0Zg6dSpGjx4Nk8mEO++8E2lpaVKwK3fp0iU8+eST+Pe//y09NnDgQNx0003o3r07zp8/j6+++grFxcWIjPTdEdlfOp1Oeq06nfPfrbi4OK+vn41DQkvofOKhOqMofWQzEQoQi6Lro6uZiLzro//NRADAVFiOS8ey0LJHkscxuyX09lEDGKgRNQSWEgvenRCYsr2aemDtXdBH6qseWANTpkzB7NmzUVZWhhUrVngEajabDatWrQLgDL7Cw8MBAJmZmWjevDlat27tEaS5TJ48GXq9HmazGRkZGbWe6+LFiwE4A8aFCxd6BK/h4eFYtmwZOnXqJDW9kDt69Cg6dOiAjIwMPPPMM16fIyoqCuPHj8eSJUtQXl6O7OxstGzZstZzf+211wA4m3B88skniiDN5brrrsMzzzyDv//978jIyMAXX3yB6dOne4yLj4/3msWLiorCbbfdhtdeew0WiwUnT55Ev379aj13argYNjcBoqxNuhDJQI0CwyZvJmLw0kyk1AJRFD3OE0VRkVGLTHL/Y3d6R7r35zKH5ho1r6WPIVSaSUSNX0REhNTsYdWqVR77hG3cuBE5OTkAgDvuuEN6/NNPP8WlS5fw888/+7y2SqVCQkICAKC8vNznOH/YbDb88MMPAIBx48YhOjra67j27dv7bOQxbdo0/P777zCZTOjRo4fP55JnzWo7b8DZrn7r1q0AnFm7yta93XvvvVIAWnEdnsvQoUOh13sP3OVt80tKSmo6ZWokGKg1AY4Sd0ZNxdJHChDlGjXP0kfRLio6Q0rnmWyALH7rPKqjdNtXm/5QXaMm3/TaRctmIkRUz1zrznJzc7Fp0ybFMVfZY1JSEkaPHu1xrqvUrby8HMeOHcM333yD+fPnY/r06UhOTsb58+cBwGuGqzouXLgglVr27t270rH9+/ev9LhrzqIo4vz589iyZQsWL16MOXPm4Oqrr8ZLL70kja3tvAHg7Nmz0v5jV199daVjmzVrhnbtnI2xjhw54nVMZYGe0WiUble1OTc1fqHziYfqDPdRo7og79yo8VL6CDg3vdaFayucpyyJ7Dy6I/Z+sh8AkJeej8LMIkS3jFKeE6Jr1Lw3E2GgRhRKdBE6PLD2rqDPoS6NHTsWCQkJyM7OxooVK6SOg2azGV999RUAZzZKrVaWjhcVFeGNN97AihUrcOzYMa9BjSAIXqsjqku+JUBcXFylYyvr+ggAX3/9NRYuXIgdO3ZIAZRcoNdZyVvuuzpmViYxMRGnTp1CXl6e1+PyYKwieTloIN53athC5xMP1RlFoMbSRwoQefCkuxKcaHTOvcXsV9rpm0ssiEhQnifv+CioBcS3jUVcaizyzuYDAE7vTMdVt/ZSnNOg1qiFUMaPiJwffOtqfVio0Gg0uP322/H222/jq6++wnvvvQedTodvv/1W2k/MlXVzSUtLw+jRo6V9tQDAYDCgS5cu6NWrFwYOHIixY8di5MiRijH1wVsDDsCZHZs5cyY+/vhj6TFBENC2bVt0794d/fr1w/Dhw7Fp0yb885//DNh85AGTP01hXN03uacm1RY/UTRyoihCLGEzEQo8eemjvBxRH6FDWZ7z75y3Fv3y9Wm6cGcntLbXpUqB2pmdZz0CNcUatRAKhLwGauHMqBFR/ZsxYwbefvttFBQUYMOGDZgwYYLUFbFz584eTSmmTZsmBWCPPvooZs2ahe7du3tk3QK1Tkq+bszbvm1y+fn5Xh9/5513pCBt4MCBePbZZzFkyBCPxh412aS7MvIMoDwz6ItrTFWZQ6KqcI1aY2eyAA7ZN0Fco0YBIl9/Js8iKTo/emmJLe/46CqLTLk6WXos94znP9CKoDCESh914cyoEVFoGDhwIDp2dK75XbNmDUwmE9auXQvAM5v2yy+/4NdffwUA3HPPPXjttdfQq1cvr0FaQUFBQObXsmVLxMTEAAD27t1b6djffvvN6+PvvvsuACA2NhYbN27E+PHjvXZfPHfuXO0mW0G7du2kcsVffvml0rFZWVlIT08H4AyQiWqDgVojJ9/sGgBULH2kAJGvUZNnkaraS61iRg0ADNHuv5fy67rYLCHa9dHLuhM2EyGiYHF1dVy7di3Wr18vrd+Sd3sEgFOnTkm3r7rqKp/X++KLL6Qyvto2tlCr1ZgwYQIAZzfEzMxMr+NycnJ8dkt0zbt9+/Y+9xfLy8vDxo0bpfsV512T9WsajQbDhg0D4OyiWdlWBUuWLJFuuzYjJ6opBmqNnLzsEWoVoOeHSAoMq2LDa1mgZlS26K9IvkbNlVGTZ6as5VaIDuUCaru5Aa1RYzMRIgoSV+bs4sWL0hqtQYMGKVq+A859vFzWr1/v9Vo///wzHn/8cem+xVL7TcMfeughCIIAi8WCmTNnwmxWfplns9lwzz33+Gyp75r3kSNHpG6UckVFRZg+fbqidLLivOVt8atT1jlnzhwAzlb9d955p9cmJjt37pTe95YtW+K2227z+/pE3jBQa+QUe6hFhHFhKwWMTd6e31fpY1Vr1K6MVQQ3orIdP9DA1qiF0PyIqGlp3749rrnmGgDAvn37AHhm0wBg8ODBUvfCtWvXYsqUKVizZg1+/vlnrF69GnfddReGDBmiCHhcTUlq45prrsHs2bMBAJs2bcKAAQPw8ccfY8+ePVixYgUGDx6MNWvWSJtyVzRlyhQAgMlkwvDhw7F48WLs3LkTGzduxIsvvogePXrg+++/V5xTcd7yjpJz587F3r17cfTo0SrnPnbsWMyaNQsAsG3bNvTp0wfvvvsudu/ejc2bN+Oxxx7DyJEjYTKZoFKpsHz58kq7OxL5g58oGjmxWNZIJIJljxQ4itJHeUZNvul1FWvUXAGaax82aYzJqgjerObQXKPmtT0/m4kQURDNmDEDu3fvBuDsnugtq2MwGLB06VJMnjwZFosFq1atwqpVqzzG3XzzzTAajfjkk0+QkZGBsrIyn0GUv9544w1YLBYsXrwYBw8exJ133qk43qdPH0yaNAnz5s3zOHfevHn48ccfcfjwYZw6dQr33Xefx5jk5GQ8/vjjePTRRwEAx44dU2ygPWrUKBgMBphMJixYsAALFizAoEGDsH379irnvmjRIqjVarz//vtIS0vDgw8+6DEmPj4eH3/8MUaNGlXl9YiqwoxaIydfo8bNrilQHDYH7Fb3fjvVWaNm9ZJR0+g1gCzZW3GdmiKjFkJr1Lxn1BioEVHwTJs2TWpvP378eEWZo9wNN9yAvXv3YsaMGUhOToZWq4XBYEDbtm0xZcoUrFu3DqtXr5ayWFarFatXr671/NRqNRYtWoSNGzdi4sSJSE5ORlhYGDp16oS5c+di586dPtefxcTEYPfu3Xj++efRu3dvGAwGaLVaJCQkYMiQIZg/fz4OHz6Mhx56SHrdn3/+ueIarVu3xnfffYdBgwYhIiICRqPRowTTF61WiyVLlmDbtm2YMWMGUlNTodfr0axZMwwcOBDz58/HiRMnpH3siGordD7xUJ0QS2V7qDGj1uQ4bA5k/HoecamxiEry/g9fTci7MALKcr/qZNRca9QEQYDOoJWOVQzUGtIatVAqzSSipic+Pt7v9WQ9e/bERx99VOmYyZMne914edasWVIpYEWuroeVGT16NEaPHu312BNPPIEnnnjC6zGj0Yi5c+di7ty5lV4/JyfH57Fhw4b5zKD5M/chQ4ZgyJAhVY6Ta9OmjV8bWFf2vgaCP6/PG26+HRz8RNHIcQ+1pm3Hop+xb8Vv0IRpMOvT6TDG1a5kxaViIKWRZZGqtUZN1kREW0mgFrJr1AxaZyZQrPAYERERUS2x9LGRE4tlGTW25m9y0n92thC2ldtw7lfPDlk1Je/4qFKroNa6f5VU2fXRS0YNUDYUke/RZrfZ4bC7yyxDaY2aoBI89lLTGEJnfkRERNRw8RNFI6csfeQataZGntEqzS0L2HXlgZTGoFF0E9VHuteoWbzto1bquUYNUAZqFllGTZ5NA0JrjRrgDDblr4lr1IiIKFAsFotfXSkro9Pp0K1btwDNiOpTaH3ioYBzyEofVSx9bHLkAU9gAzXvrfkBZfDlbY2a4lyDj4yarDyyYqv+UFqjBjjX5JVku/fT4T5qREQUKJmZmZVuSu6P1NTUGq9No+BioNbI6QZ2BkTnWjVN7zbBng7VI1EUYS2ro0Ct3HuwBShLH72uUZNveK3IqLl/HSlKHysGaiFU+gh4NhThPmpEREQUCPxE0ciF3dAPYTf0C/Y0KAhsZhtEh7vLRVlAAzV38FSx1E/e9dFSZoXD7oBKrZI9Jm8mImtCIlvrJc8EWmWBmlqnhqAKrU3b5fNW69SK10pERFQb/naLpMaJnyiIGil50w4AKMkt9TGy+iotfZTto+ZtHspmIt7XqFl9rFELtfVpgDIwZdkjERERBQoDNaJGSp65AgKcUfOxzgy4kiWTJb3Ki9wNbURR9J1RU5Q+ygO10NxDzUURbLLskYiIiAKEgRpRI2X1ksmqmN2qKZu89LFCoKZSqxAm6/xYXugO1Kwmm3LPMR9dH30HaqEXCOmYUSMiIqI6wECNqJHyFpSV5gWm/NFSSekjABhi3FtBmBSBmjLL53sftQYUqDGjRkRERHWAgRpRI+UtUAtU+WNlGTUAMES7t4IwFbi3iJB3fFSpVVDr3KWM/uyjFoqBmrzLJfdQIyIiokAJmUDNZDLh5ZdfRt++fREREYHIyEj06tUL8+bNQ0FBQcCfb/fu3VCr1RAEAZs2bQr49YmCzVtr/EC16K+smQhQMVBzZ9Tk69O04VrFRtn+ZNS0IRioJfdt5fU2ERERUW2ExKeezMxMjBkzxmPn9UOHDuHQoUNYunQp1q1bF7Bd1a1WK+699144HI6AXI8oFMmDHZe6CNQ03jJqPkofFR0fjRWakCgCNXdwJg/U1CHYTKRZuzjc/t/JKLpUjPZD2gZ7OkRERNRIBD2jZrPZcNNNN+Ho0aMQBAEPPvggNm/ejC1btmDOnDlQqVRIT0/HpEmTUFRUFJDn/Ne//oXDhw8H5FpEoapi10cAKM0JzBo1+T5qOq+Bmo/SR3nHR0OFjaLlpY+ycaG+Rg0Akro1R6eRHaDWhl4gSURERA1T0AO1xYsXY+/evQCAN954AwsXLsSIESMwbNgwvPHGG/jkk08AAKdOncKrr75a6+c7fvw4XnrppVpfhyjUeW0mEqiMWrkso1ZVMxF56WOp74xaQ91HjYiIiKguBD1Qe/PNNwEAXbt2xUMPPeRxfNq0aZgwYQIA4K233oLNZvMY4y9RFHHvvffCbDajWbNmNb4OUUNQsT0/AJTmBSpQk60b89JAQ7FGrdD7GjV5t0SgYqDmvfQxFNeoEREREdWFoAZqx48fx/HjxwEA06dPh0rlfTozZ84EAOTn5+PHH3+s8fP997//xfbt25GQkID/+7//q/F1iBoC76WPddBMxFBVe3556aMsoxbuZaNs2fVF0bnhWqhveE1ERERUF4IaqO3cuVO6PXToUJ/jBg8eLN3eunVrjZ7rwoULeOqppwAAr732GuLj42t0HaKGwnvpY4DWqCkCtarWqMn2UVN0ffSdURMdIuwWZ8mjspkIM2pERETUNAQ1UDt27Jh0u0OHDj7HJSUlwWAweJxTHbNnz0ZRURFGjx6NGTNm1OgaRA2Jt4yaucSiCHxqSrGPmpc1amGy0kdLqQV2q/3Kbf/WqAHuvdS4Ro2IiIiaoqAGapmZmQAAjUaDFi1aVDq2ZcuWinOq4/PPP8eaNWtgMBjw7rvvVn+i1ZSVlYUjR45U609aWlqdz4uaFm9r1IDANBRR7qPmmVELl5U+Au51ahZTJV0fKwR8VilQ4xo1IqJQs2zZMgiC4HU/2jZt2kAQBEVFVHXMmjULgiCgTZs2AZipd0eOHPH6eG3nHsqee+456WfGz50NQ1A/9eTn5wMAjEajYuNbb4xGIwCgsLCw2s8xZ84cAMDcuXPRvn37Gsy0ehYuXIjnn3++zp+HqDLeSh8BZ/ljdMuoGl9XFEVlMxEvpY8avQZag0ZqCmIqMCGimVGZUauwRk1QCYpzXIFmQ2jPT0REDYPZbMaLL76Il19+GRaLZ+UJUSgJakbNbDYDAMLCwqoYCan00XWOvx5//HFcvnwZPXv2xOOPP179SRI1UN5KH4HaZ9TsFjtEhyjd99ZMBFCWP7rWqSm6Php1Hud4a9Ef6hteExFRwzF//ny88MILsFq9f5lJFEqC+vW0q8tjVdk0AFIHOF+dIb358ccfsXTpUqhUKixatAharec3/0SNlbz0MSxKj/Ii55cctQ3U5GWPAKDxUvoIAIZoA4ovlQBwlz5W1YTE+ZizS6TXNWo6ZtSIiEJdenp6sKfgU1XbPIXy3KnpCeqnnoiICABAeXl5FSPdY/R6vV/XNplMuO+++wAADz74IK655poazrL6Zs+ejalTp1brnLS0NNx88811MyFqkiyyoCi2dQwuHrkMIACBWrnyHzlvzUSAip0frwRfimYifmbULJU3LiEiIiJqjIL6qScyMhIAUFZW9QfH0lJnW/HY2Fi/rv3cc88hLS0NLVu2xEsvvVTzSdZAYmIiEhMT6/U5ieTsVrvU3h4AYlpHywK12rXot5a7gy21Tg2V2nuWW95QxGvpY7ivjNqV53GtUSvnGjUiIiJqeoK6Ri01NRUAYLFYkJ2dXelYV7dHV/fHyhw6dAivvfYaAOC+++7D6dOnceDAAcWfjIwMafypU6ekx7mwlBqDio1EYpJjpNuBLH30Vr7oolijdmXTa+WG154ZNZ3sehYva9S44TURkdurr74qdfHbtm1bpWMfeeQRCIIArVaLrKws6fGSkhK8/vrrGDduHJKTk2EwGGA0GpGamoqpU6fim2++qfa8quqc6HA48MUXX2DUqFFo0aIFjEYjrrrqKrz55puw2+1ez6lo/fr1uOuuu9C5c2fExMRAp9MhMTERQ4YMwcsvv4yioiLFeFeXSnmzN9d7N2vWLL/nDgC7du3CzJkz0a5dO4SFhSEmJgb9+vXD3LlzkZub6/Wc9PR06fm2bNkCi8WCBQsWYMCAAYiJiUFERAR69+6NefPmobi42K/3oC5kZ2fjueeeQ///Z+++w6OqtgYO/8709IQkEJIAofcq0otSFBFEBBuo2BW91nuvvXeveu3YP5BrAaWIFJUO0nvvoRMggfQ67Xx/DDkzk5lUQkLIep/HxzPn7H1mT4bArFl7r921K2FhYQQEBNC4cWPGjRvH6tWrS+ybnZ3N+++/T+/evbX3pH79+lx99dV88cUXJdaZ2L59Ow8++CAtW7YkICCA4OBgmjVrxl133VXqn+1LWbV+Pd26dWvtODExkejoaL/tTp48SV5enk+f4mzcuFGbg/zKK6/wyiuvlNj+wQcf1I4PHTp0QcvBClEVbEUKiUTEh2nHOWfON1DzrPhY/F8hAUUyaqqqli+jpk19lH3UhBDCnzFjxvDUU0/hdDqZOnUq/fr189uuMDACGDx4sDbrZ+XKlYwcOdLvl+VHjx7l6NGjTJs2jTvvvJOJEydWyphzcnK44YYbmD9/vtf5LVu28NhjjzFjxgzq1atXbP+srCxuvPFG/vrrL59rKSkppKSksGLFCr766iuWLl2qJQUqg91u5+GHH+brr7/2Ol9QUMCmTZvYtGkTn376KT///DNDhgwp9j5paWn07NmTTZs2eZ3ftm0b27ZtY9KkSfz999/ExcVV2tjLYtasWdx5552kp6d7nT98+DCHDx9m8uTJPPzww3z88cfo9d5fnB46dIiBAwdy6NAhr/OnTp3i1KlTzJ8/nw8//JCFCxfSsGFDrzbfffcdDzzwgE+QnpiYSGJiIpMmTeKBBx7giy++KFNdi0tJtWbUunXrph2vXLmy2HYrVqzQjnv16nVBxyTEpcAzc2UMMBAcHaQ9zkmtvDVq/vZQK+S1Ri0jzxXguYtFlmGNmut5ZOqjEEL4V79+fQYMGADAtGnTis1GLVu2jJMnTwJw2223Aa4vwa+55hpSUlIIDg7m6aefZu7cuaxZs4YZM2bw0EMPaUXYJk2axLx58yplzDfffLMWpPXv359p06axZs0aJk6cSJs2bVi2bBnTpk0rtv/dd9+tBWlDhw5lypQprFy5UgsECoOAw4cP8+9//1vrd91117F582YeeOAB7dzmzZvZvHkzr732WpnG/sADD2hBWtOmTZkwYQKrVq1iwYIFPP7445hMJtLT0xk+fHiJWaCHHnqITZs2MWTIEO31T548mTZt2gCuoOfJJ58s05gqy6JFi7jxxhtJT0/HZDLx+OOPs2DBAlatWsWECRO07a0+//xzxo8f79N/3LhxHDp0CKPRyDPPPMPChQtZt24dM2bMYPjw4QDs37+fe++916vf3r17GT9+PA6Hg2bNmvHNN9+watUq/v77byZMmKAlT7766it+/vnnC/tDuBip1ax9+/YqoHbo0EF1Op1+21x77bUqoEZERKj5+fmV8rwTJ05UcX1sVBcsWFAp9zwfO3bs0MYDqDt27KjuIYka7MT2k+pH/b5QP+r3hfr19d+rGUkZ2uOP+n2h2q32Ct97z8L92n1+vn9ase0O/H1Qa/e/cVPU7DPZXmOw5dt8+iz58G/t+rJPV6pOp9OrT/qJjAqPWwhx6bLb7equXbu0/+z2iv8dV9NMmjRJ++ywcOFCv23uv/9+FVCDg4PVnJwcVVVV9cknn9T6zZs3z2+/adOmaW3uuecer2slfY5q1KiRCqi9e/f2Oj937lytz5gxY3w+9+Xk5Kh9+vTR2jRq1Mjr+tatW7VrN910k98xZ2RkqHFxcSqgWiwW1Wq1el1/+eWXtXv4U9zYFy5cqPXr3bu3mpWV5dN35cqVakBAgAqoDRs29HruQ4cOeX3Oe+qpp/yOPT4+XgVUk8mkZmZm+h1jRXm+9v3792vnbTab2rhxYxVQAwIC1JUrV/r0zcrKUnv37q31/+uvv/y+trffftvvc994441am+PHj/uMSa/Xq8eOHfPpd/z4cTUiIkIF1EGDBp3Py6+Q6v67pVozaoAWlW/bto333nvP5/rUqVOZO3cu4FpvVtaqj0LUZrZc742lAyODvK6fT1bNs5hISWvUPKc+5qbne1V81Ol16E2+682Mgd5THz0LooCsURNClJ+qquRmF1Trf6qqlj7QCho1ahSBgYGA6zNTUXa7nRkzZgAwcuRIrW1SUhL16tWja9euXHPNNX7vPXLkSO1zl+fa/or65ptvAFfV7wkTJvhMYwsMDGTSpEnFbsW0a9cumjVrhslk4sUXX/TbJjQ0VHs9+fn5pdZAKKvC2gcGg4Eff/xRq1zuqVevXtq4jh49qk03LSoyMtJvFi80NJSbbroJcNVv2LdvX6WMvTQzZ87Upiy+8MILfmevBQcH89NPP2lZ1vfff1+7dvr0ae24SZMmfp/jmWee4eGHH+a///2v17TJwr7BwcF+p7zGxcXx6quv8swzz3itJawtqn0e0f33388333zD5s2befrpp9m5cye33347JpOJmTNn8umnnwKuxZ3PPvusV9/Dhw/TuHFjwJU+X7p0aVUPX4iLkjXHey2YwaT33kvtTC6h9UIqdG97XvmnPuZn5lOQ415EbAw0+p1n7rnmzZpn81qfBjL1UQhRfnk5Vh6/6adqHcNHv4whMPjCfNEcHBzMddddx5QpU5gxYwYTJkzAYHD/XblgwQLOnDkDwNixY7XzhdPInE5nsffW6XRER0dz/PjxMm2lVBK73c6iRYsAGDJkCGFhYX7bNW3alD59+vidOnjLLbdwyy234HQ6S9xXNyYmRjs+33ED2Gw2li1bBrjW+JW07u2+++7j+eefR1VV5s+fz5gxY3za9OvXr9jEQ+HnWnAV56gKCxcu1I7vueeeYts1bNiQIUOGMHv2bJYvX05+fj4Wi4WmTZtiMBiw2+3885//xGKxcM0113jtX9ylSxe6dOnic8+WLVsCkJGRwc0338xbb71Fq1atvNo88sgj5/sSa6xqz6jp9XrmzJmjvSmTJ09m8ODB9O/fn48++giHw0FcXBxz5swp9pdaCOHNa43auSxVUGSgdu58Kj9a87zXvxUnMMydUVMdKlmn3f/gmIL8B3hFi4nY84tsri2BmhBC+Chcd3b27FmvD90AU6ZMAVzBy6BBg3z6FgY8+fn57N69m9mzZ/Pee+8xZswY4uPjOX78OFByQFcWJ06c0KoZduzYscS2Xbt2LfF64ZhVVeX48eMsXbqUb775hkcffZTLL7/ca1um8x03wJEjR7Rtoi6//PIS20ZFRWlZpZ07d/ptU1KgFxTkngFT2ubclaVwnA0bNiyxkAu460sUFBSQmJgIuF5zYbbr+PHjjBgxgujoaEaPHs0333yj/Rny5/bbb9cC65kzZ9K6dWtatmzJY489xrx58yol0K7JLopPPbGxsWzevJlPPvmEqVOnsn//fqxWK02aNOH666/nX//6F3Xq1KnuYQpRY1jzPDJqAa6iHUGRQZw9lAac315qnpUbDSVk1EzBJnR6HU6H6x/JjCR3qeTCMRVVdB81e4E7o6boFHSGav9uSQghLjpXX3010dHRpKSkMHXqVK3iYEFBAb/99hvgykYVrdSXmZnJxx9/zNSpU9m9e7ffoEZRlEqZuum5JUBpn+lKCxZmzZrFhAkTWLlypRZAeSop21YRniX3y7JPbt26dUlMTCQ1NdXvdc9grCjP2SYXcsqsp8LXV9bXVsjz9X322WfodDq++eYbVFUlIyOD6dOnM336dMAV4N5zzz3cc889XhnfyMhI5s+fz2233ca2bdsA2LdvH/v27eOTTz4hMDCQ4cOH88QTT9C9e/dKeb01yUURqAFYLBaeeuopnnrqqTL3SUhIqPAf4jvvvLNWznUVtYPXGrWgys2opR9N145D6vnO0S+kKAqWMAu559bDZZx07wtTXEat6D5qRfdQq21leYUQ5y8gyMRHv/hOP6vqMVxIBoOBm2++mc8++4zffvuNr776CpPJxLx587T9xAqzboUOHDjAoEGDOHLkiHucAQG0atWKDh060L17d66++moGDBjg1aYqeE6Z8+R0Ohk3bhw//PCDdk5RFBo3bkzbtm257LLLuOKKK1i4cCFvvPFGpY3H87NmWf4dKqy+WVP+zSp8feV5bUXbm81mvvrqK5555hmmTp3K7NmzWbt2rdZ+/fr1rF+/nkmTJjF//nxCQtzLL9q3b8+WLVtYvHgx06dPZ968edqfudzcXKZOncovv/zCO++8U6444VJw0QRqQojK47lGzXhuY+nASgrUzhxyf4MW1SSyxLYB4R6B2okM7by/za7Bz9RHq5TmF0KcH0VRLtj6sIvJbbfdxmeffUZ6ejrz589n2LBhWnGRli1bctlll3m1v+WWW7QPw0888QR33nknbdu29cm6VdY6Kc91Y6UV+EhLS/N7/vPPP9eCtO7du/Pyyy/Tt29fn8IeFdmkuySeGUDPzGBxCtvUlNlgheMsz2vz7OepcePGPPPMMzzzzDNkZmayZMkS5s6dyy+//EJGRgZr1qzhnXfe4c033/TqpygKAwcOZODAgYDri4T58+czc+ZMFi5ciKqqPPvss4wYMUJb11YbyDyiWkAtyEctZm8VcWmyFqn6CEUzahWb+mjNtXlNYYxqUvI/QgFh7oIimZ4ZNT+bXYO/NWoSqAkhRFl0796d5s2bA/D777+Tl5fHnDlzAN9s2vr169m4cSPgKh7x3//+lw4dOvgN0opuflxRsbGxhIeHA7Bhw4YS227dutXv+S+//BKAiIgIFixYwDXXXOO3+uKxY8fOb7BFNGnSRJuuuH79+hLbJicnc/jwYYAaE1C0b98ecK3FKy1YW7duHeDKehYWPlFVlSNHjrBkyRKvtqGhoYwYMYKvv/6aTZs2aT/DP//8U2uTn5/Ptm3bfNbzNWvWjIceeogFCxbwzjvvAK6MatGN0i91EqhdolRVJfedl8m6+2YyrxuA8+jh6h6SqEKeBT8KpxMGe5Toz61gRi31cKq2abXBYiAsNrTE9p4l+j2LiRjLmlHzWKMmgZoQQpSssKrjnDlz+PPPP7X1W57VHgGtCARA586di73fr7/+qk1dO9/CFnq9nmHDhgEwf/58kpKS/LY7c+ZMsR/GC8fdtGlTr6lznlJTU1mwYIH2uOi4K7J+zWAw0L9/f8BVRbOkrQq+/fZb7bhwM/KLnWeRmf/7v/8rtt2RI0e096Znz54EBLj+jX/mmWdISEhgwIABWpBaVJMmTbQvEvLy8rTzjRo1omPHjjz88MPFPm/hmsuifWsDCdQuUYqi4Dh4AOexI+B04jh0oLqHJKqQzU/VR8+pj9kVDNTOHHRPe4xMiEDRlTyf3bNEf2FREShhjZrXPmp276mPfvZdE0II4VaYOTt58qS2Rqt3795eJd/BVcChkGd2w9PatWv55z//qT22Wq1+25XHww8/jKIoWK1Wxo0bR0FBgdd1u93OPffcU2ylv8Jx79y5028lwczMTMaMGeM1dbLouD3L4pdnWuejjz4KuEr133777X6LmKxatUr7ucfGxmp7ol3sRo4cScOGDQF4/fXXWbt2rU+b7Oxsxo4dqwW+jz/+uHZt6NCh2nFxa8h27NihZc08p+EW9l2+fLmWAS6qsHJp0b61gQRqlzB9k2basfOgBGq1iWdlRlNQYdVHd6CWl56H017+ksVnD7orX0WWsj4NvKc+eipL1Uenw0lBlmeFScmoCSFESZo2bUqPHj0A2LRpE+CbTQPo06ePVr1vzpw5jBo1it9//521a9cyc+ZM7rrrLvr27esV8BQWJTkfPXr04KGHHgJce3d169aNH374gXXr1jF16lT69OnD77//rm3KXdSoUaMAV1bliiuu4JtvvmHVqlUsWLCAN998k3bt2vHXX3959Sk6bs+Kki+99BIbNmxg165dpY796quv1orQLV++nE6dOvHll1+yZs0aFi9ezJNPPsmAAQPIy8tDp9MxefLkEqs7XkwMBgMTJ05Ep9ORm5tL//79efLJJ1m0aBFr1qzhyy+/pHPnzqxcuRKAcePGMXLkSK1///79ueKKKwBXFrZPnz7873//096bt956iyuvvBKbzYbJZOJf//qX1vfpp5/GbDajqiqjRo3iH//4h/Zn8bfffmPs2LG8++67gGt6b+EattpCPvlcwvSNm2HD9ReWQwK1WsVrjVqA7xo1VMhNyyU42nduf0FWASe2nSS2fQyWUO9AyzOjVtr6NPCe+uipLPuoAeSmu6c4yNRHIYQo3W233caaNWsA1zoif1mdgIAAJk6cyMiRI7FarcyYMYMZM2b4tLv++usJCgrixx9/5OjRo+Tm5hYbRJXVxx9/jNVq5ZtvvmHbtm3cfvvtXtc7derEddddx2uvvebT97XXXmPJkiXs2LGDxMRE7r//fp828fHx/POf/+SJJ54AYPfu3fTp00e7PnDgQAICAsjLy+PDDz/kww8/pHfv3qxYsaLUsX/99dfo9Xq+++47Dhw4wPjx433aREZG8sMPP9S4gGLAgAHMmDGD22+/naysLO1nU9Rjjz3Ge++953P+559/ZvDgwezYsYOVK1dqQZ2n4OBgJk6c6LWPXqtWrfjhhx+4/fbbyc/P5/PPP+fzzz/36duxY0dmzZp1nq+y5pGM2iVM55FRk0CtdvHKqJ1bD2a0GDEFuzNZ2Wd8pz+qTpUZ/5zD7Of+ZNqjs7yybqqqctZz6mNZArXiMmrFrVErkjXL8wzUTBKoCSFEaW655RatvP0111zjNc3R09ChQ9mwYQO33XYb8fHxGI1GAgICaNy4MaNGjWLu3LnMnDlTy2LZbDZmzpx53uPT6/V8/fXXLFiwgOHDhxMfH4/FYqFFixa89NJLrFq1qtj1Z+Hh4axZs4ZXX32Vjh07EhAQgNFoJDo6mr59+/Lee++xY8cOHn74Ye11//LLL173aNCgAX/88Qe9e/cmODiYoKAgnymYxTEajXz77bcsX76c2267jUaNGmE2m4mKiqJ79+6899577N2712tNVU0yYsQIDhw4wPPPP0/nzp0JDQ0lMDCQVq1acd9997Fx40Y++ugjv9snxMTEsGHDBj777DMGDBhAdHQ0BoOBsLAwOnbsyLPPPsvu3bsZPXq0T9/Ro0ezY8cOHn/8cdq3b09ISAhGo5GYmBiuvvpqvv32WzZs2FDq/nqXIkWtqt30RIl27txJu3bttMc7duygbdu253VP55kUNo95gENKDKeVOvR99jY69G91vkMVNcC3N0zWSvCP/nQEcR3qAzD59imkndsHbfhbQ2jSO8GrX9L2k/z6D/c3Vjd+fj2x7VwllXPO5vLtDZO1a/f9No7ACP8Zs0LHNp1gxhO+ZZKHvDiQloOa++3z+VXfavuntbqqBXvm7wOg+ZVNGfrK4BKfTwhROzkcDvbt26c9btGihU8FQyGEKK/q/rtFMmqXMCUyirXmDszXX85WXVP2r00svZO4JPgrzw8QFOWeL+9vL7X9yw56PT663l3i+IzH+rTAOgGlBmlQfEatuKqPAMYAd+bMK6Nmlg9dQgghhKg9JFC7hCmKQv1I94fepENnS2gtLhWqU8XmWZ4/yB0Uea5Tyz6T49PvwNIigdoGd1Utr2mPjcu2iWdxa9TMxaxRA+91arlpMvVRCCGEELWTfPK5xNVvEA7n9i48mVy2OdiiZvPcQw3cxUQAwuq75/0fWnmYHnd1RVFcJfZP70kmO8U7eDu1O5mC7ALMwWavjFpUGSo+AljCzH7PG4up+ui65h5vnhQTEUIIUYtZrdYyVaUsiclkok2bNpU0IlGV5JPPJS6uTTxsdJXXPZunx1pgxyQfeC9ptlzvPWM8pxm2GNCMdZNdJZtTDpwladsp4jq61q8VnfYIoDpUjm9OomnfxuUuJAKgN+gxB5soyPYeU3FVH8E7sMzLcO+lI+X5hRBC1DZJSUklbkpeFo0aNSp2I2pxcZOpj5e4ep1aYFBdhRlUFE4eTi2lh6jpPNen6Qw6r42iIxvXocFlcdrjLdO3A66Kjp7THj2zV0c3HMfpcHL2sHs/nbKU5i/kb/pjcVUfwTuj5rA6PMYka9SEEEIIUXtIoHaJMzZuQl3StccntvlmTcSlpbhCIoU6jWqvHSf+fYjM01kk7ztD5qks10kFuo51f3t3dMNxMk5kuoMmBeokRJR5PAHhvgVF/I2rkLGYa7JGTQghRG2TkJCAqqrn9Z9k02ouCdQucYrFQkyAe9rZiV0nqnE0oir420PNU0KPhoTFhgKuAiLbftvJAY9pj7HtYmg9pIX2OP14BodWH9Eeh8eFYbQUH2gVFRDmnVHT6XXoTcVnx0wBxQRqMmVXCCGEELWIBGq1QP267oIOSef20BKXLptHRs1fdkqn19HxBo89+2bvZt9i94boza5oQmi9EMIbhGnnNk/brh2XdX1aoaIZNVOQUStg4o+xuEBN1qgJIYqh03l/nLHb7dU0EiHEpcThcHg9Lvp3zYUmgVotEOtRSv3kWfnH61JX2tRHgDbXtNT2KyvIKiDzZJZ2rVm/JgA06tpAO5ednK0dRzUtW8XHQkXXqBU3tVG7XlygVkIWTghRuymKgsnknkGQk5NTQmshhCibvDx39WmjseQvmi8ECdRqgbi2jbTjVKuJ/CLl28WlpbSpjwDmYDOth7T0OR/Tph4hdYMBaHh5vN++5SkkAr6bXpdUSARKCNRk6qMQogQhIe7tR9LS0iSrJoQ4L6qqkpGRoT0OCgqq8jHIJ59aoG6XVhjVRGyK6wPwyX0nadyxYTWPSlwonptdFxf0AHS6oT3bZu70Otf8iibacVynWBS9gupQvdqUe+qjT6BWWkbN/19LEqgJIUoSGhrK2bOu/R6tVitHjhwhIiKCoKAgDAb5+0MIUTaqqmK1WklNTfXKzkugJi4Iff1Y6ukyOa66pqwd37xfArVLmDXHI6MWVHz2KqJhOI26NeDIumPauWb9GmvH5iAT9dvUI2n7Ke2cwWwgrH5oucZTdOpjaRm1YouJyBo1IUQJzGYzkZGRXsHa6dOnq3lUQohLQXBwsFfWvqrI1MdaQFEUYkLciyGTdh2vxtGIC60sa9QKdb6pg3Zcv109QosEYQ27ek9/rJMQgU5fvr82fIqJVHiNmgRqQojiKYpCdHQ0kZHlW0crhBAlMZvNxMbGVvn6NJCMWq0R26gOnCvcd+KQbHp9KStPoNbo8gYM+Gc/Tu1OpuvYTj7XG3aNZ83EDdrj8q5PA9+pj8YKr1GTYiJCiJIVBmuhoaFkZmaSlZWF1WotvaMQQhQRGBhIeHg4ISEhVV7tsZAEarVEfPc2sH0fAKeyDTgz0tGFhVfvoMQFUdY1aoXaX9eG9te18XutXqu6mIJNWLNdH3TKuz4N/Ex9DKpgRk2mPgohykBRFCwWCxaLhbp166KqKk6ns7qHJYSoQXQ6XbVk0IqSTz61RHyv9vCtK1BLV4LJ+nsFYcOGVfOoxIVQlqqPZaUz6Gg1qDnbftuJoldo3KNR6Z2KMAYYMZgN2AtcFdhMAVL1UQhRdRRFQa+XjLwQouaRTz61RGS9EMx6JwUOV+r2+LKNEqhdoqw5ZZ/6WBZ9xvegfrt6RDQIJ6JheIXuERBuIeu0ay+20jJqxY1Z1qgJIYQQojaRYiK1hKIoxMQEao+Tdh1HtRZU44jEhWLLK1vVx7IyWoy0GtyCeq3qVvgenuvUKr6PmnwjLoQQQojaQwK1WiSuVax2fMoejH3LxmocjbhQPIuJlGWNWlWI7VBfO45pU3LA5688v96oK3e1SSGEEEKImkzmEtUicY0jgYMAnFIisK/+G2O3XtU7KFHpylP1sar0uPtywmJDCY8PI7pZVIltDX42vNbL+jQhhBBC1DLyFXUtEtsoQjs+rdTBtnoFqlTCuqSoqlqpxUQqiznIRKdR7UnoXvpG63qDHr3Je5qjrE8TQgghRG0jn35qkdhG4dpxphJE7tkMAvfvwdDSf2l2UbVUVSX1cBoF2QXYrQ4cVgeqUyW2fQyWUEvpNwBXH4eqPTZeJBm18jIGGHFY3Zu0y/o0IYQQQtQ2EqjVIuGRgVgCjeSfmxp3SokgZMpkDC+/U80jEwBznv+LgysP+5y3hFkY98MtZQrWPKc9wsWTUSsvY4CB/AyPx7KHmhBCCCFqGZn6WIsoikKcx/THU0od7CuWYt+8oRpHJQCyz+T4DdIA8jPyObz2WJnu4zntEcUV8NRERQuK6GXqoxBCCCFqGQnUahnP6Y+JSiwniOT4J1+Sm5lbfYMS5Gfkux8oEBYbijnYnQ07c/Bsme5j8ywkEmBEUZRKG2NVKlqtUja7FkIIIURtI59+aplYjw2Lt+qasVXXDE4Dt0yl3eXxjH2oJ5H1gqttfLVVQbZ7T7vgqCDu/HkM63/czKqv1wJw9mBqme7jmVGrqevTwF+gJmvUhBBCCFG7SKBWyzQuYdPiHeuP88p9vzCyh5nel0egCw1H37gpurr1qnCEtVNBtjvAMgebAYhqWkc7d6bMgZpnaf6auT4NfKc+GiWjJoQQQohaRj791DJNWkVz/bgubPj7MLkZeeSezSAf9wf6ArvClBVWNi7fwI2O5UQoOQS++SHGy3tU46gvfQVZ7oxa4ZTHqCaR2rns5GzyswqwhJhLvM/FuIdaRRTNqMk+akIIIYSobeTTTy009OaODL25IwAFv/1C7ucfsllpxix9L/IUV2XB/boG/FcZzVjHItpO+gpD1+41dr1TTeAvoxYcHYQp2IT13LWzB1OJ61i/xPvYvKY+1tyMmqxRE0IIIURtJ8VEajnTiBsJ+fBLej8ykueGK7SNzNOu5Stm/k8/hMUHjNh3ba/GUV76PNeoFWbUFEUhqrHn9MfSC4pcMhm1QJn6KIQQQojaTQK1Wk5RFAztOmIefgP1HnqQRyeP565/9sV8bt8qVdExR9+TSf9Zis1qr+bRXrr8ZdQAopq6pz+WZZ2aZzGRouu8ahIpJiKEEEKI2k4CNeFFURR6DmzG0x9cS50w94fj9SkhvP/P2eR4rKUSlSffc41aiHvKYmQTd0btbDkzajW56qPPPmqSURNCCCFELSOBmvArvnEdnvv8BpoYzmjnDiWm89v3G6txVJcuq9fUR4+MmlegloqqqiXex2sftSBZoyaEEEIIUVNJoCaKFVonmEduqUc3527t3JrFieR5TK8rpNrt5L75Itn/uBvH/j1VOcxLQnFTHyM91qhZc21kncoq8T6XzBq1AO/ATNaoCSGEEKK2kUBNlChg2Aiu128gQM0HoCDfzppFiT7tbIv/wrZ0AY69u8if/E1VD7PG8yom4jH10RxsJiTGvQF5aevUvNeoXUoZNVmjJoQQQojaRQI1USJdaBhBg6/icude7dyy2Tt9puDZd27Tjh0HfQO50lgXzCPntee87lObFJdRA4hqXPaCIpfKGjWZ+iiEEEKI2k4CNVEq8y130MN4UHucdDyL/TtOe7Vx7HVPj1RTTqMW5Jf5/s4zyeR98Cb2vxeT9/4b5z/gGshfef5CUU3LXlDEcx+1mjz1UYqJCCGEEKK2k0BNlEoXE0uDB+6kufO4dm7J/y3RjtWCfJyHPLJoqooz6USZ7+84cggcDgCcx4+i5uSc/6BrENWpUpBTfEYtskk5Mmp5l0gxEdlHTQghhBC1nARqokxMw0bSu5E767Nlbw7pB13BmOPAPnA6vNo7jx8t873VlGTvviePF9Py0mTNsYLHTFKfjJpH5ce0Y+nYrd4/a697eRYTkX3UhBBCCCFqLAnURJkoisLlL48nlFwAHOhZ/vEM17HHtMdCjhPHynxvZ9FA7UTtCtQ8pz2CbyYsvEEYeqPrV1V1qKQdSSv2XlaPzJwxsOZm1IoGmbJGTQghhBC1jQRqosyMsXH07hKmPV55QMFhteHYt8unbXkyas4z3oGaI6m2BWoe68qCTOj03r+WeoOeiEYR2uPipj867U4cHtm2mrxGrWhGTdaoCSGEEKK2kUBNlEv/+69CpzoBSFOD2DZzpd+MmrMcGTWfqY+1LlArvpBIoSiP/dTOFFNQJC/Tu4BLTQ7U9EY9AWEWABSdQmB4QDWPSAghhBCiakmgJsqlTsO6tAvP1B7P/2039uO+QVm5Mmq1fupj8YVECkU28az86D+jdnq3++cYEG7BHOL/XjVF34d7EdEwnO53diUoMrC6hyOEEEIIUaVkPpEot/5XxLNtVjYAiRkWFuq6cJVzIxiNYHMVs1DT01Czs1CCQ0q9X9Gpj5JR8xXV1LPyo/+MWtL2U9pxbPv6KIpSSSOsHq2vbkHrq1tU9zCEEEIIIaqFZNREubW5aSDt1EPa44W6y9ijNMDQvhNY3FPUylJQRM3Lhews73NnU1Dz8iptvBe7smTUPCs/5pzJJS/Dd5+6pB2egVpMJY5QCCGEEEJUNQnURLnpI+pwa6s0otR0AFRF4Wf9ADLi26CLb4AKHFLqMfOnHaxasL/EeznPpPg/f6rs+7DVdPlZpWfUgqKCvKYyFt342l5gJ3mPOzMZ20ECNSGEEEKImkwCNVEhof37cYd9AUbVNdUxV7Hwf1vC2GBsy8eGG5hguJ4F63OY9OEK1i07WOx9ihYSKVSV69QcNgfJ+86gOtXSG18A1jJMfVQUhaimHgVFEr3XqZ3em4LD5iryYjAbiG4edQFGKoQQQgghqooEaqJCjH36U59URjn+1s4dPVnAzwfqckKJ9mo79+etOIsJgpwpp/2fr6J1aqpT5ef7pvPzfdNY8M6SKnnOorymPpZQACS6mTv4OrrB++dz0mPaY0zruugNskG0EEIIIURNJoGaqBBdVF30rdtxmbqfHo6dJbY9eTSdLauP+L1WtJCIdj7pOHa7k2Vz97BtXdlL/ZfXmUOpnD3kyk7tXXQAVa36rFpZ1qgBNO7VSDs+sv4YBR5TJpO2eaxPk2mPQgghhBA1ngRqosKM/QcCMMK5iqYhruIfRqOO7s7dPGn7lQ6KOzibN3Wb3yBITfFYo+ZViOQ43/1nGT9+vprPXlnItrUXJljzDHacdie2PPsFeZ4Sx1CGqY8A8R1jCQh37S3mtDtJXHkYcGUFvQuJ1L8wAxVCCCGEEFVGAjVRYabhozAOHIK5UxeeePcanvtoGO98MZTRjuXUJ5UB1vVa26MHzrJjve/eap4ZNUO7jtrxlsMONq44rD3+49dtF+Q1WHOsXo89g6aqUtaMms6go1m/Jtrj/UsSAUg9muYOOBWIaVO3wmNxpqWiOp0V7i+EEEIIISqHBGqiwhSTicBnXiH4vc8wJTQmoUU0IbHRKGHhAMRxljYxDq393E/m+9zDc42avmMXAHIxMTO7nVe7xF3JHNl/ptJfg2dGzd/jqlDWjBpA8yubasdH1x8nP6vAa9pjVNPIEoO9Esfxyw9k3Xwt2Q/ejmqv+syiEEIIIYRwk0BNVDpdXAPt+Mrjs7Xjg6kG9m7yzqp5Vn00tOsAej1z9D3JUgJ97rto1q5KH6tnNsv1uBoCNa/y/CUHWXEd6xMQ4Zoi6nQ4Sfz7UJGNriu+Pq1g9nRQVZyHErFv2Vjh+wghhBBCiPMngZqodJ6BWoJ6mqZO955o8/5vtXas5uejZmW6+9WL5UBEO9brWmnnWngEHuuXHyIjNbdSx1o0MCsauF1oTrsTa65Ne1xaRk2n19G8v/f0R8+Kj7HtKhaoqaqKmuou+a8WU41TCCGEEEJUDQnURKXTxTf0ejzAuVk73n0wl8W/7yInq8C74qNOhzUwhF/zOmunGkQqPPraYMIjXdk1h93Jsnl7XcfHj5L/40QciSVvqF2aooFZfhVPfbTmej9/SeX5C3lNf9x4nIwkd7Ab26GChUTy88DqUVjlbOVPMxVCCCGEEGVnqO4BiEuPvmVr94PAINpe3ZcGv5/mmK4eAFO+XMuv366nbbMgmurakEUgKZZ6nHxsLqlWV6CiUx3c2ioLk9nAlcNaM/N711S8ZfP2MGR0WwqeewLnyRNYp/9MyOTpKMEhFRqrb0atagM1r0BRAVNgyRk1cE1vDKwTQG5qHqrDXUkzpF4wIXWDKzQONT3d+/GZFP8NhRBCCCFElZCMmqh0hi7dMN9xL4a+Awj+4AsCxt3LEGWTVxuH3cm2PVnM1Pdlof4yttriSU7K0q5f6dxK/WxXef++17TAaHJt4JyVns+6qStxnnRNp1SzMrEtW1ThsfqsUavijJpXIZEgE4pOKbWPTq+jmcf0x0Lnsz7NmZHm/fisBGpCCCGEENVJAjVR6RRFwXL7vQS99Bb6Zi1QgoJp07E+j9mm09uxg2BTyeXfGzpPM9C5CWfScQCCQy1095jut/ivQ3juyGZdMK/CYy0oWp6/ygO1spXmL6qFx8+jUP0Krk8DUDPSvR5LoCaEEEIIUb1k6qOoEobe/YnfuJZ45xmuCz5E0hOfsPqL2Zw6nkEE2cS0iqPByKupa8wl7OV70QHOkydwpiSji67LgOtas+KvfQAcT1c4pMTQRHUV0XDs3IbjxDH0HkVMPNntTlBVDEa9zzVrNRcTKU9pfk/128UQFBlIzll3cZUKr08D1HTvjJoqa9SEEEIIIaqVZNRElTD26qcd604n0So8j1vrH+ARx2/c5ljIsN6hdOvfhEY926KPiXU1VFWsi/4EIL5xHVp1cgci0/X9yMWdgbIt+MPr+VRV5cDO03zxxmIeHvE9j47+kS/fWsLmVUew2dx7u1V3MZH8cpTm91R0+qMp2ERkQkSFx1E0o6ampcpeakIIIYQQ1UgCNVEldJFR6Fu7N7G2rVyKM8U9vU6Jruv6v6Jgumqou938uaiqa6LjkNHttfPJSgST9Fdjw5Ulsy6ch+p0Yj9xgjXvfs/bD/3Kf/49j82rjqCqYLc52LTiMF+8sZh/j53CTxNWU5BvoyAr32ucVZ1Rs3pOfSxDxUdPba5phaJ3rWlr2icBnb7iv85F16ihqqhpqf4bCyGEEEKIC06mPooqY+zdH8fuHQAU/PIDngvNdNH1tGPToKEUTP4WAOexIzj27MTQuh1tusRxdYM0/jrmyhwd0tVnCgMY61iAcvoUG1/5jFkb7JwmAsgpdhy52VaWztlD+tlc7FkFgLuAR7UWEynH1EeAui2iuOG/wzl7KJVWg5uf1zjUjAyfc84zrmmnQgghhBCi6l00GbW8vDzeeecdunTpQnBwMCEhIXTo0IHXXnuN9CKlw8srLS2Nt956i549e1KnTh1MJhMxMTEMGzaMX3/9VcvYiAvLeOVVYLG4HhQUeO3b5RkQ6OrHou/g3k/Ndq5YiOpwMCj5Ty537tGubdM14Vf9FXylH8bXG0LOBWluzdvV46EXB/CPVwbRrX8TTGb3dxNbVh8lW+f9K5BfJMMG4DyVRM7Tj5L7+vOoeXkVeOXFK8iqWDGRQvGdYuk4sl2F+npSi2bUkHVqQgghhBDV6aLIqCUlJTF48GB27drldX779u1s376diRMnMnfuXNq0aVPue69evZpRo0Zx8uRJr/OnT59m7ty5zJ07lyFDhvDLL78QElKxvbhE2ejq1iPo1ffIef05yHaX4kdRUOpEebU1Db6WvG2ujbKtSxZgefAxHAf2Q3Ymo/ibDCWIfYqreMgGXUuvvoqq0kFNpL9xL21fnoQS5NpbrEO3BuTn2fjohb84uNs17fKUyUCTfJuWUytIz6Wo/P99h33TOgD0LVphvvn28/5ZFDqfjFplKrqPGoBT9lITQgghhKg21Z5Rs9vtjBgxgl27dqEoCuPHj2fx4sUsXbqURx99FJ1Ox+HDh7nuuuvIzMws172PHDnCtddey8mTJzEYDDz44IP88ccfrF27lh9//JEePXoA8OeffzJ27NgL8fJEEYYulxP86Xfo4htq53QxsSgG7+8MjP2udGffsrOwr16Bff1qAPQ4ubPlaRo0reNz/5ZBGTxm+oPbHItokH8c67xZXtctAUZuuq+b9jhfryPD4P41sNvBduqUVx/7NvcecLaNa8v5it1Uq5Wc154l66Fx2PfsBCpenr+yFS0mApJRE0IIIYSoTtUeqH3zzTds2LABgI8//pgJEyZw5ZVX0r9/fz7++GN+/PFHABITE/nggw/Kde8XXniBtLQ0FEXh119/5YsvvmDIkCF069aNMWPGsGLFCkaPHg3A7NmzmTt3buW+OOGXPr4hwZ98i3HQNejqx2G592GfNkpgEMY+V2qPc99/nYKp/9Meh/ToziOvDqZ+w3AA4gLyeHiohSenPEKT4Vdo7QpmTvWpXtikVV269musPT5tNOC5s1vmV19px86zZ1BPubOxjp3bUa0VKzhinTcL+99LcO7fS+7LT+PMzLhoMmo+xUSQvdSEEEIIIapTtQdqn3zyCQCtW7fm4Yd9P7DfcsstDBs2DIBPP/0UexlLhufk5DB9+nQARo0axfXXX+/TRq/X88UXX2AyuT4g//DDDxV5CaIClJBQAp9+mZDJ0zH2G+C3jemqa90P8vPB5g6QDN16El4nkBc/G8HbE2/kxV/H0/Eft6Lo9ZiuvxH0rmqQakoytuWLfO498pa26FRXeGbXKZz12GMt5++V2PfuBsCxa7t3R2sBjnPZsPI6tXw9p4jAgYKaeob8zz/wDtT8VH20b1pP/uRvcKYkV+g5y0K1WiHXd8qnKoGaEEIIIUS1qdZAbc+ePezZ4yoMMWbMGHQ6/8MZN24c4CoKsmTJkjLde/PmzeSdK/wwfPjwYttFRUXRvr2r7Pv27duLbSeqnr5jF/SduvqcN3S5HF3jZq5jg47IesHodO7Kjbqouq7CJecU/PKjTxGQ8JRE2juOao/PGPXYzt3CqpqwLfkLAPu5KpX5GLWtAOxbN5b7tcybvJ43d7fiA+NNvGS4iy/0w5m5NIOkM1kU7upWdOqj81QSOc8/QcH/viP39efK/Zxl5W/aI7iyiUIIIYQQonpUazGRVatWacf9+vUrtl2fPn2042XLljF48OBS712vXj1efvlljh8/TseOHUts63S6MisFBVVbml2UTNHpCHr3E5xHDqHm54HDAUYT+mYtUBSlxL7m0WOwLXRtgu1M3EfWPTcT8ODjGPpeiaIo2Devp539ODv0jXAoCk5F4ZjZSJTNQR5m7KtX4LjvUXavP8JK/WB2KgkYcHC5cw/91+0goZR6Iqrd7iqSotfz95/7+G3KDjg3Zqti5KASy0FiwQlKoI5gh5Pdu5OJaBZJYJArw2tbt8q1aA5w7N6B48A+9M1anOdP1c9Y/Ux7BCkmIoQQQghRnao1UNu9e7d23KxZs2LbxcTEEBAQQF5enlefkjRv3pxXXnml1Hapqans2OHKmjRo0KBM9xZVR9Hp0DduWu5++qbNMfYbqE17VFOSyX39OQzdexP43GvYN61HwUK01c4psxGAPL2OY3od36hX0f70QQ7fPZWUlLZa3tmKjpX69qxKVOn46gIGj2pPs7b1fIJG++4d5L7+HGp2NntufJ4fphylJKqikGXQ88vkTcz4aQtd+iQweGRbos9VvSxkXTiPgAsQqDk9Kz5aLK5ppgA52ah5eSgBAZX+nEIIIYQQomTVGqglJSW5BmEwUL9+/RLbxsbGkpiYqPWpLB9++CE2mw2Aq666qpTWZZOcnExKSvmyEQcOHKiU5xZuAU+9iC4unoJpP8G599i+diXZ/xyPM3EfVrUrdRxOsvUOsg3uNWr5ion1SitI8d1TDUBFYcva42xZe5y6sSF0v7Ip3a9sSt3YUBwH95Pz3BOQncVupQGTfj6Mei7SMzhVGuXbsIToaHF5IPtW7mW3LgGnR6BntztZt/Qg65YepIkulP5KI1qpR9ABtkV/Ybn3Hz4VMs+XZ0ZN3yABR+I+OJdldp5NQe9RoVMIIYQQQlSNag3U0tJcHxCDgoJKncoWFBQEQEZGRqU9/6ZNm3jvvfe0+99zzz2Vct8JEybw6quvVsq9RMUpZguWu8djvOpa8j99X9sLzXlgHwBWTChAwwI7OTYH6UY9mXodqsefRUVVaaUepVd8Hhm6YJYdCyZFCdeuJydlMfvHLcz+cQsNG4UQcHwfqq0v6OGgUh9nYZCmOmiU78CiqpDp4IpHR9Ip/12mrLSSo1fI0uuwBhvI8Zh9e9BZl4OGIcQ7Uxjn+Ivw9DTsG9di7N67Un9OnmvUlDqRKKmRWiER9ewZkEBNCCGEEKLKVWsxkcI1YZbC/bJKEHBu+lVlrSM7duwY119/vXa/V199lejo6Eq5t7i46OMbEvj2R5iuG+113qq61oIpQLBTJb7ATstcK12th7nMuZdBjo08a/+Jux1/0qFbPP0HJvAv+1Tusv9B88B0n+c5eiSLvY767NM1YJ+uAXbF9T2IWbXSIj/LFaSdc+ZgGsqt96MDQhwqja05vJj9DXdfH01CC+/Nv4/rovnMcD0nqYNt/jy/r1E9lzGsCDXdnVFTwsLRRbqfX0r0CyGEEEJUj2oN1AqrPJaWTQNQz33ILa4yZHmcOHGCQYMGcezYMQCGDh3Kk08+ed73FRcvRafD8o9/Yr7lDu2cFfe+ZYV7mOmBBgYbtziWcrVzAxFkA2Bo0x59p8vQAW3Uo9yfPY3X7qrDsIRU6uG/GAeAQbVzl+NPTE691/mU/WcoyHfv3mbCih4nbZZ+zTPvDeHR9ido6zykXc9QgplguI7dq/ehZrk3fledTnLffIHMoX3J/e9bqA6H1/OcPp7Bolm72LvdexNvT2qmO0uthIejRLm/sJBNr4UQQgghqke1Tn0MDg4GID/f/1ogT4VtzGbfvabK4+DBgwwaNIhDh1wfgrt168bUqVPLFCyW1UMPPcSNN95Yrj4HDhzwu9ebqDyKomC55yGUsHDy//ctNms4nNs+LLR+KCn7XUGJLaI+FIlP9G3ao4SFQWAQ5OaAw0HA12/TH+gHnKQOR3T1UBUdpisGYejQGcffi2i44Xci1Uz+xnsz6+R9KYTUDdYem3SujJiakox96QIaHl7NnY6zrFZbM9PQD1WFfMXMt+rV3PLVPHo9OhqjyYB95TJsSxe6xv3H7yhGE5Z//JODe1KYP307W1YfRVVBp1N48u0htGgf4/NzcXpk1HRhEeCxlYFUfhRCCCGEqB7VGqiFhIQAkOtns92icnJyAIiIiKjw823cuJGhQ4eSnOzaPLhr16789ddfWsBYWerWrUvdunUr9Z6i8phHj8E08mZst0+FXFd2KjQmxB2oBUV4BWpKTCy6OpEAGDp0wr5mpdf9FCAuxE7CoA6Yho5An9AEAGevemTdMZ38XO8gDSBl/1niO8W6x1QnBM4lyvK/nYCaehaAns7d1Hvy33z74RpsTgWHoufHxXn8vOwHYhuGU//0TqJ0nXCgw4YB25wkkjZM4mCR+MrpVJkxcQNPf3Ctz5cSXmvUwsJRPKZRyqbXQgghhBDVo1oDtUaNGgFgtVpJSUkpcY1YYbXH2NjYYtuUZP78+YwaNYrsbNdUtj59+jBnzhzCwsIqdD9Rsyl6PdYcq/Y4NCZEO7Zi9ipTb2jbXrtmGjYK+7o14HSgxNTH0PlyjF17YOjRB8XkHZDpwiMw3ziGjO9/8Xn+tGPpZJ/J0R4HNG4AW13Haqo7StQlNKHzwDY87sxjwn/XkqO41mo6HSrHD6VxnFjQF/mdKCa2Orgnha1rjtKpZyOv816BWngEOvdSOsmoCSGEEEJUk2pdo9a6dWvtODExsdh2J0+eJO/cdCzPPmU1a9Yshg8frgVpQ4cOZf78+RKk1WKqqlKQ7T9QK8i1Ybi8l/bY0PEy7djYvRchP/5GyI+/Efq/mQQ++RzGfgN8grRC5lvGoV5xre/zO1VObD2pPbbUjcDQw7eao6FDFwCaD76Mx7un09J5FINqL9NrbBKaz0NP9aRDN/f+gDO/34TT4fRq51VMJDTMa42aU9aoCSGEEEJUi2rNqHXr1k07XrlyJT169PDbbsWKFdpxr169/LYpzoIFC7jpppuwWl0fyseNG8e3336LoZL3ohI1i73AjtPuDlhC63sEalkFBDz4KHlOB7o6kRgHD/Xqq4sqe3VQxWhEHTQSFvhWazy587R2bA42Yx5ym8+0SkOHztpxw38/xn0P3ob91J8kE8EJJYrjShSZShAB3XtgDjTBqsWY87NorR6l4dlklK//IOKOp9i+HlQVTh5NZ83iRHoNbg6A6rB7FSfRhUegWtwbXKtnz6CqaqWu4RRCCCGEEKWr1oxas2bNaN/eNa1s8uTJWmXHor7//nvAtT6tf//+Zb7/0aNHufHGG7Ug7R//+AcTJ06UIE14ZdOgSEYt24oSVY+gV94l4NGnznuD6fws/1tKOKzuCo3mYBP69p3Qt2rr1UbfoZN2rAQFEfj0K+h1CvVJpau6j+udq7int5N7X72W258ezNivHmFoGxsNVdc6TDUlmfAPn6JrY3eg9fsPm7FZXVk5z4qP4FqjdjrfzHzdZSRRB2xWr0BOCCGEEEJUjWoN1ADGjx8PwLZt27TNpz1NnTqVuXPnAnD//feXq+rjXXfdpW2Qfeutt/Lpp59KZkAAUJDtDp70Jj1BkYHaY9WpYs2r+L5kPs9VTKDmyRxsRlEUzDfdpp3TNWmOLiLSq52hXUfMt45zn1AUzLe7N2rXxcQS9P4ELPc8BIUBptPJoL0/oldcX4SkpuSw8JVJ2PfuQk1Pd9/LaOLE6QLeeXEpC/Rd+cxwPYeUeqhnksv/ooUQQgghxHmp9tTS/fffzzfffMPmzZt5+umn2blzJ7fffjsmk4mZM2fy6aefApCQkMCzzz7r1ffw4cM0btwYgP79+7N06VLt2uLFi1m8eDHgysSNHz+eLVu2lDqeTp06VcrrEhc3z4yaOdiEOdj7C4CCrALMQf7XnZWXZ0YtomE4aUfTfdqYQ1zPZehzBeY77sWxczvmcff5vZ/5tntwnjiGbeVyzGPv1KpMFlL0esy33IHhsm7kvPYs6qmT1CGLHvYdrNS7Mth/bbbSefsT1Hn0ca1fZmgMn72ykPxcV5BqU4z8n/4aHt+YSJMmzc/nRyCEEEIIIcqp2gM1vV7PnDlzGDhwIHv27GHy5MlMnjzZq01cXFy5KzR+99132nFaWhr9+vUrU7/ipl+KS4tnRs0cbEZn0GEMMGI7l0lzXQ8ppnc5n8sjUIttH0NGUqbX+rjCMcC5vd5uv7fE+ykGAwHPvU4AJW8Wr2/eiuDPJ5H3zsvY169hkHMT63WtsCpGcpUAPrNfza0/zCIeyMfId3l9SMvM8bpHvmLmsx+O8XSvDOrFSfEdIYQQQoiqUu1TH8FVcn/z5s28++67dOnShZCQEMxmM61bt+bZZ59l27ZttG3btvQbedi6desFGq24FFiLZNQAzCHurFpZpiuWVX6We0P3oMhAIhN89wIsHENZKYpSpmm8utAwAl//APOd9xMSHcaVoUe1a8lKBJ8mX86fusv5UT+IJJs7MO3eMQSd6goms20GPnz2T9LO5Pjcv5DqsGP7ezH2LRvL9TqEEEIIIYR/1Z5RK2SxWHjqqad46qmnytwnISGh2AzYjh07Kmto4hLkOfXRdG6KoznYRPa55Vj5WVZ/3Sr2XFmeQaGZ6OZRpBw469Wm6NTLyqTo9VjG3o1l7N2MdDhxTPibRfMSURUFp6Jjkb6LV/vBI9sy+q4uNB71MFNsrkqsqWdy+c+/5zH05g50u6IpZov3Xx3WaT+T/+3nAAS+8i7G3mUv+iOEEEIIIXxdFBk1Iapa0amPABbPjFp2ZWbUPJ4rxBWoFVXejFpF6fQ6bnqkP/9od4I6aobP9S69GzHqnstRDAZ6DmzGCId7u4Czp7P53yerePqOqUz7bj1nTmVp16yL/9KO8yd9jer0ntoJkJqcRfqcuVj/muP3uhBCCCGEcJNATdRKRYuJuP5/YaY+FnhMfbSEmqnbwl+gduEyav60vvVqnrRPo6djp3aucTTc/a9+6HSuKZXGQUPo49zBNY61KB6Z69xsK/On7+CFe6czd8pW7GfP4jx4QLvuPJyIfY1770OAOT9t4dk7p/HM5yf57MMNrHxnMgX5lVdZsziqqnLyWDrWgrJtEi6EEEIIcbG4aKY+ClGV/GXUCisvuq5X3tRHz4yaJcRMVNMoUIBzsY9Or8NgqdpfRX2nrgTExnBD0gp6OHeRrERw+ejRmMzucehbt0NXP44BJ7fQznmYNe1vZd0RHQV5rqDH6VSZNXkTe5bt5CYCCCVP61vw0yQMPfuiKArL5u7h9x82A6AqevYojdizQmXK+p+5rG9jRt55GeF1Aqls1gI7X765mB0bThCXEMHTH1yLJcBY6c8DYN+9g/wvPkLfpDmWR/+NopPvwIQQQghxfiRQE7WS34zahSomkuk59dGCKdBIeHwY6ccyzp0zVfn+fopOh2noCPK//ZxYUolVUzHWqePdRlEwDriagh//j7qkM+LUdK5p1Ih1ibCUdqTbXD+3vUcK+NAwmlsdi2mhngDAsXcX9k3r2Gmrz09frPE7BmuBg9ULD7B7cxLjXxhA45bR2jWnU2XN4gOsXXKQ3KwCbDYH1gIHdpuDRs2juOvJPgSWkIW0We1MeH0RuzYlAXDicBqLftvJtbd2Op8fW7HyP/sAx77dOHbvwHDZ5Rj7DrggzyOEEEKI2kO+9hW1kjXHu8CH5/+h8tao2QvsOKwO7bHlXNaursc6taqe9ljIeNW17k2xAaVOpG+bgVdrx2pKMsbt6+mdu54nc3+kYwP3dMhsJZBv9dfyvWkIu5WGOFDY++2vfPPuUlSnq12Yms0d9vlc7tyDRXX/fNPP5vLeU3+wepFr+uThfSm8+8+5TPrvCnZvTuLIgbMkHUnnzKks0s/msnXNUb79zzKcDv/r3Gw2B1+8uUQL0gr9NW0HWRn5fvucD+fZMzj27XY//7rVlf4cQgghhKh9JFATtZL31EdX8GTxCJjyKylQKxrwFWbtPAuKmCppY+3y0kXUwXzbPQDo23dC37KNTxt9g0boW7fzOR+AldvSZnDLLS3Qq65AVFUUdqiN+D/DNbxlGMu3h5tiK3Bds5gU7rH/QXv1EDc5lvGS/X9c51hF4QxBu83BxA/+5r1/z+PtJ+ZwaG9KiWPfseEEs/632ee83ebg67eXsGP9cZ9r+Xk25k2p/G077Ou9M4b2jetkP0YhhBBCnDeZ+ihqJe+pj4UZNY81apU09dFz2qPepMdwbg1Y076NWf3depx2J40uj6+U56oIy9i7MN9wM1gCip1+GfDUS1rpfX2L1hT88B3YbJB2hm7bfybGfoifDANJVtz7w2UqQdqx3qDj3s4Z1F+R6jqhKBhVB32d24kLM/A/R1+yz/2c9u887fXcl/VJ4LK+CZhMBgwmPUtm72brGtdecH/8so0GTerQtV9jAJJPZvLzhDXs3HhC69+ldyMaNY9i5iTX/m5L5+5h4Ig2RMVUzmbmAPYN3hk0NeU0zmNH0DdMqLTnEEIIIUTtI4GaqJU8M10mf2vUKqmYiFdp/mCzFgxFNAhnzLejyUjKJKF7w0p5ropSAkou5KGPb0jQK+9qj53Hj2JbMA8Ax5YNxAFP2n/lYM9b2GBuz9bVR/CclXjnY71oMvNtCieAmkbejHXmVFBVmiRv5qnHRvLNn3kcO5iq9anfMJxbH+xOq06xXmNp0jKat5+cw8mj6QBM+nAFgcEmNq8+yt9/7sXpcGeyOvZoyL1PX4HqdLJ07h7SUnJw2J3M+t8m7vl35ezzpjrs5G9YzzJdJ9II5krnFuqQjX3jWgnULiFOhxOdXiagCCGEqFryL4+olaqqPL/nfSyh3mvRIhvXoUnvBHSGmvVraL7+Jp9zelQ6Xt2JB58fwH++GcF1rKGj8wB32OfTJfAUjgN7tbamwdd4FdsImvEdT717NQOua02jZpHceN/lvPjZCJ8gDcASaOThlwYSeO49sxbY+eiF+Sybu6dIkNaA+5+9AoNBh9Fk4Lrrm2nX1i45yNHEsz73BsjPtbF51REWzNhB2pmcUn8Wjt07+SuvFX/ou7NG35ZPDSM5pkRj37iu1L7i4qeqKl+/s5RHRv/Awt92lt5BCCGEqESSURO1TtECH+7y/Bc4oxZSPUVDKpu+RSv0bdrj2LXdfVKnw9CxCwAh9etw5WUh2NcsAiB/4pfgOPfztgSga9wU89g7sf29GFQV57EjsGgOtzw4GuepJHJff57cKckEPv8Ghg6dfZ6/bmwo9/y7H5+9spCiS8HqRFq4/q5udLuiibYfnJqXS9vpbxCj9uKU4iqY8tNTPzDwulaYW7VBb9Bx8lg629cdZ9+OUzjsrnTgn79u5x+vDPKqRlnUpt/WskTvHmO2EsgX+uHctuVvethsOJNPkf/dBHR1Y7Dc/4iU7a9hDuw8zYblhwCY+f1G+l/bCqNRX82jqjyqqoLTiaK/dF6TEEJcSuRTg6h1igZhWjERj33U7AV27B7BXIWfyzOjVk3VHS8E00jvrJq+RSuUYPe6L2PfK7Vj56FEd7uWrVH0BvRNmmO8eph2vmDytziSjpPz3BM49u1GTT1L3kfvoDp9KzuqdjtN5n3ONXZ3EY8gNY/rHCt5rvkOegxoqgVpALbli1FOJzHU4c5yHcwL5ZupSXz26kI+fnE+v3y9jt1bkrQgDSArI5/3n/mDLefWxBV1+kQGP6z2XddnU4xMclzB4i//JPfZx7H/vQTr9J+xLV3g9z7i4rV7y0nt2Fbg4PC+M9U4msrlzMwg+/7byLp1OPad26p7OEIIIfyQQE3UOp6l+XUGnVbgo2iZ/MqY/uhZTKTo1MeazNjnSpRId6bJ0Oly7+s9+4Kfb+kNrdtrx5Y7HwBLAABqRjrZD9zuyq6d4zx2BPuaFT73KPh5EvYVS7nSuZV77PO4mb95xv4zfZ07UP9ehH3HFq/21r/mANBKPUoTZ5LP/YoK0DvQn5uOaitw8MUbi1k6Z7dXm4J8O1++toB89dyfHdXKmDHNMesKK2DqmPrHGX5Kbk4qrgDWsd17XCWx2RxM/Xotz909jUWzdpW5n6hcuzd7/3nZt+1kMS1rHtuCeTgPJ6KmpVLw64/VPRwhhBB+SKAmap2ipfkLC3wYLAav9WLl3UvN375eBZfg1EcAxWDActcDrgcBgRivGe59PSQUQ5fLffrpW7fVjnWRUZhvud19MT/Pp33BtJ+8HjsO7KPgx4na47ad6nHl5HcIatncfZsJH2mZOMfxo1qApAB3P38VHeoVEO9Mob56lnrGHOrGhtIwPogrHZt5yD6Ll/O/Y/xABUugEQDVqfLThDV88vICZkzcwPI/9jLpv39z4liW9pw3BW2h/9jePHGdhTA1Wzu/UdeSdw0386u+H8m7D/v5SbqoNhuq3Q5AZloe/33mTxb9toszp7L45eu1HDlw6WRyaoq8XKvPNhH7tp8upnXN4zx2lBTC+EXfn1V7z3/2gBBCiMongZqodfyV5gdQFKXCJfoXvb+ML675P9b9b5PXec/92CyXUKAGYLp6GCE//kboT7PQx/puMWDsc6XPuaJ7splHjUGJ8l4DZujSTTt2bN+CffcOwBXM5L73urbeTakbQ9DL76KLqotl/OPuPvv3YJs/F0D7P4AuoSnRfboy/pWrecwxgyft0/hX7g+8+kw7nohZz1DnOhqrp9Cj0mj+d/zr35cREeWuiLlj/XH+/HU7P3y6io0rDmvn+zm20rVPIxRFodHgnvzD/huxqjuwcip61ula8/bRrkz6YDlJR9K8Xq99724yb7yGrJuv5fDfW3nzsd9J3J2sXVdVmPLlWtmbrYrt234Kp9P7Z35g92nstksjqHGePM5U/RWs17Xi14y2l1S2UAghLhUSqIlax6s0f5HNpitSUCTzZCY7Zu/GXmBn3fcbcHh8kMvPzPd770uFrm6M19o0T4be/UHnnv6oi41HF1HHq41isWC5/1F3n179CHzrQ/RtO2jnCn79CVVVKfhxIs6D+7Xzgf98DiXItV+boW0HjFdepV3L+/y/2Hduwzp/nnbOdPW1KIqCvlFj9G3cUzDzvvoE++q/vQdvLaDOtM/5172NqG/ILPb1N3aeZKhzHYZuvVyvMaEJEXUCeNQ+kxvtS4lQ3Nk1JzpWLUrklfG/8clL89mzJQmHw8mpb/6Pg7nBLM9uwHvvbiDtTK7P8yTuSmbd0oPFjqM8nBnp2LdsRLVWTsGcvFwrv36zjilfriGnkqqlXgyKTnuES2udWtrxVI7oYrTHq+btqMbRCCGE8EeqPopax19p/kKeWa/8Mn7ozDzt/jDusDnJOp1NeHyY67my3M91qWXUSqMLC0ffsTOOzRsA32xaIdOVg1FCQlEz0zH2H4ii12O+cSy55woc2FcuJfuum3CeOObuc+1Ir8wbgOXeh7GtWQF5uZCfR86/H3ZtzA2g12McOMTdf8hw8s5VrXRs2aidV0LDUDMzzp3fgGnLBh5Fxy6lEclKOGeVMFIJIVUJIUrN4FbHYsy9+2Lo0cfVX1EwDRuJOvkbuhkO0e+Vh/j7o+ksSI0jTQnVnmfHhhPs2HACRQFVbQOGNq4L52bP6nQKN93fjd1bTmobfE//vw106tkQs8VY9jehCDUrk+wHbkc9m4Kh7wCCXnqrwvcC11q6z19bxL5tpwA4mpjKE29dfUlURvQsJKLoFNRz2bV9O07RrG296hpWpVDtdnadNXt9VbtpbRJjrXaMJvlYIIQQFwvJqIlax3uNmnfw5LWXWhnXqOUU2W8r46Q7A+P1XJdQMZGyMo8eox2bhgwvtp2xa3dMA65G0bs+JBp69kUX18B10en0CtKUejFY7v+Hzz10desR+NLbYDj3QbMwSAMMPfp4ZfOM/QdqhUw8BTz+NIaefb3OGXDSQT3EIOdmbnYsZbxjNs/bf+IB3XzqPng/gS+97VV23zzmToLe/5zgb37EfHl3enUO5Wn7VG62LyY2xO51b3+zGQPMCo++fhUDrmvDjfddjuHcusn0s7n8MdW7Op9qt5P39afkvPwUjqQT7N1+inVLD3I2Odv3xoB16QLUs651V/a/F2MvR4GTopxOlYkf/K0FaeAqZz/5o5U1fppm+tlcbVN1gB5XNtGO93q83ppKTTnNLhp4ncsvcLJt3fFqGpEQQgh/5KszUeuUlFGrUKB21nuqWkaSO1DznPp4KZXnLytjt16ETJ4BCuhifDewLo6i02G+cSx5H73jdV7friMBjz+DEhjk//m6difwudfIfeMF8Cjtb/LYCgBACQzCeMUgbH/O1s7pEppg6H0F+tbtyNq6CXJdAbiuSXMCHnsaNTMd6+/TsW9ci65pcwL/9QL6Js0pStHrMXS8zD3mZi3R/zWHrup+usUEcOzpl5k/Ywe7Nrmn1gWruUSoWdQnlUHN9CR0vhPVWkDoHz9yZSOVBYmuoHL+jJ30vqoF0fVd000Lpkwm59cpbFRasGLLb5y2udfUJbSIokvvRnTpnUDdWFc2z7bwT6+x5k38ihPjXkSn09G8XT2tsE5ZTP9uvbbHmKe1SxKpFxfKsDGdynyvi83uLe73JiommF5XtWD1Itc2E4m7krHbHBhqcNYw78hR9itxPufXLU3ksj4JVT8gIYQQfkmgJmodryxXSNGMmmcxkbKt4ckuklHLPBeoqarqdY9LcY1aWejqlz1A82S85jqcKck4Dh3A0KEzxr4D0NUtfcqZse8AAp54jrwP3gBcRUcM3Xr6tDMNGe4VqJlvvRNFp0OJqkvQ6+9T8NuvGDp0wjTsBpRzWTpjjz6oNhsYDGUOavTNW2rHzoP7ad0xhjZd4jjz91pSX3uZCLIx4ZFp267HmZlB/tefYvtrDv0xsC7wDjJsRuw2B1+8uZiEZpEo1nwKlu5np2EsOUoA2Lyf9/C+Mxzed4YZEzfSon0MVw+MpcGu7RSO+oASy9xdCRx/9i8A2nSJZezDvbQgsFBGai4H96RgDjASFhFAaEQAqxcdYMHMnVqblg2M2JNOkOioC8DvP2wmun4I3a9sWqaf0cVmj0eg1rpTLE1aRmEw6rHbHFgL7BzZf4ambWru9Me9G49hU3yn0G5fd5ycrAKCaunfVUIIcbGRQE3UOiVm1EJ8M2o5qbks/3QV5mATfR/uibHIGqGcIsUfMk66yrbb8uxeJfstoZbKeQG1hKLTYbnz/gr1NQ0ZhhIZiX3dKkxDR2hTKj3p27TH0OcK7CuWYrisu2s65DmGDp0xdOjsf1zG8q0R0zdpDjqdK8OXn4fzxDH0DRMIXP0XRtJdbTp3xXnkMGrqGXA4yHv3VezrVgFgxs61ecv5yeAa3/GDqRw/mHru7i2gSLwYERXoU5Bk3/ZT7Nt+ivqGUfR07mKXksAeXUOvNrs2JfHKQzMZPqYzA0e0ZtfmJFb8tY/t6477VD/01KBBCLcd+Qynw8lnhutJUSIA+P7DFWxedQRLoAlLgIGAIBMNmtShRfsYgi/i3wVVVb3Wp7XuFIvRZKBJq2j2bXdNe9y3/XSNDtR27EoDXH/3tXEe5ohSjxwlALvdyaaVR+g7pEX1DlAIIQQggZqohUpeo+YO3AqLiSz8z1IOr3YVdIhuHkX769p49ck5W2SN2rmMWn5Wvtd5c4h3UCguLOPlPTFe7ptJK6QoCoHPv4F6JhklMhrFzwbdlUEJCEDXoBHOI65pgo4De9FFRWP7e7HWxnTNCBx7d2KdPgVAC9IKdVIPsEZpx0HVf3BgVG10de6jLztp/J/vODzpFzYt2c82XRNOK+61eSeVKGbo+xU7VluBgxkTN/D7/zZht/vuC1hUVEww90ZuwnLQ9Wf9bvuffGYeTY7TqH3o9/l5KBDXuA4tO8QQ1yiCoBAzgSFmgkPM5OVYOXE4zfXfkTRsVgedejbkimGtCQy6ML8/TocTRadoGdJTxzJI95jO3LJjfQBatI/RArW9209yzc0dfG9WA6iqys4T7ui+rfMwEUo2K/WuYj9rlyZKoCaEEBcJCdRErWMtY9XHgqwCkraf1II0gDMHz/rcr+jUx4ykzHPTHt0BoTHAiN5Qc9e0XKoUgwGlHGvnKkrfrKU7UNu/FzUnG/LPBfLBIRh790MXG68FapqAQFAUlNwc7rTOY3PDweQ3bIFt104c6ek40BEZFUAX5RCW04cByHvjBaL27uIq4CrnRg4pMSwJ6M5uWwxFJThPcq1zLekEM8vUj2yn6/ehaJAWGhGA0aQnMy0Pm9W1/UREVCAPj4kn8N0PtHZRZDLOOo+vLddjt/vPwqlq0axgyQ7vO8Ofv27nyuGtGXR9W0LCKicbl5Gay5/TtrPir32ERQRw2z960apTrNf6tAZN62jP17JDDHPO7b+euCsZu92pFXqpSU4cTiPN6s4Kt1KPUo80VuIK1PZvP0XamRwiovyvAxVCCFF1JFATtY7n1EdTUJGMWpFAbdW367yuZ53yrqanqqpPMRFrjpWCrALyM4tfCydqF33zltgWuQp52NeswHkmRbtmuvIqFJMZfYtW6OIaeFW4DHjkX66pkB+8SQBWeh2dC0fnet076NkvcezbS/4XHwLg2LvL63pj9RSNc2dxigiW6DuxTd+Meo0iGTYgiqZffXVu5uRpWhQcY56+B2t1rV1j1it07NmIvle3oHWn+uj0OhwpyWQtWUJueCyRPbtge/phim7/3Fg9xb9a7uPYFXeQl2sjP9dKfp6N9NQ89u84RVZ6PuWVn2vjj6nbWPjbTrr0akTrzrG06RxHeKSreEramRwSdydzaE8KTlWlbZc4WnWs77fgR1ZGPn9N286S2bu1oDM5L4sPn/+Lq0a14/gh94bkrTvFoubl4kxJJqF5LAaDDrvdSUG+naMHztCkVV2ve6uqyp5PJ7Ny4QEK6jXi2n+PIKGF94buF5LqcJSaGd62zv3nq4EzmVDyCFHzqKNmkKqEoaqwftlBrhrVvoS7CCGEqAoSqIlax3vqY/FVH88eSvNaYwaQVaTsuTXbij3fu+Q6uLJqns9T2/ZQE970zTwKihx3Z2gJCsZ801jg3B5sQ4aT/90EAIxXDMY46BoAbMsXY1+/2ue+xoFXY2jXCX2T5uR//7VWqbLw3oYu3bCfm2IZQxq3OpYw7oEuBIy8HgBr4LMU/PIDzhPHCMTKaMdyejh2cVqJoKXtGHWCr8XSqR+KXofz5AlyHrsPJS2VICDfZAarx+/S2Lso+HEiAJFbFtPoH/ehb9QK59kzOJNPoaYW4OxoIVmpz4GCCBL3pZOelktuVgE5WQVkZxVgMOiJbRhGXKNw4hpHkpGay5I5e8jLcX25YitwsHbJQdYucW3+Xb9BGAX5dlJTvLPai37bRWCwiQ7dGtCqU30y0/JIPpHJ6aRMjuw/i7XA93dWVeGvad6bPrduXYfsxx/AeXA/xiHDadyqM/t3nAZcZfoLA7Xc7ALWLE5k2cwtnDztBJrACdj2xBwG39CO4WM7Y7YYPJ5LJTUlB6NJT3CoBZ3OPRWxIN/O6RMZnDqWQXZWPkajAZNZj9GkJzDIRONmESgrFqFE1MHYvZfWz751E7lvPA9GE0Hvf44+Nt7nNQJsX3NYO26lHkWpEwWpZ+jsPMAivata6dolEqgJIcTFQAI1UesU5HhOfSw+o1Y0SAPIPJ3l9Ti7yPq0QhlJmVjz3GX4JFCr3fTN/K/5CXzmFa9tC0w3uoI21eHAfOMYbd1UwL9eIPeN53EeO4KuXn109ePQNW2OecSNgGu7AdM112Gd/rP73k+9hOGybmSfOIrz4AHXSZ0e85WD3M83dISruuaRQ9hWLMX6x+/EJ58iXj0DgHX2dNSsDCwPPk7Os4+jpnlMV/QI0oz9BmIedz+2lctxHnaVsc994wXU/DzUU+6phAChQBeDgctbtcU0ZDjGq0Zor9O2dhW5b70A2/NQwsJR6kTRv2dH1sQMYOHsvWRleGfjTh7LKPZnnpttZc3iRNYsTiy2TbCaSy/nTrbompF8rghKIYNBR8Mjq3Ee3O8a25+zaT68O/vPxXKbVx3BYXeyZ+tJDu5O9rumT1Vh/vQdbFl9hJvu705ejpVdm06wa3MSGal5gGtz85BwC6HhAeTmWElNzva7v16hEIONKwvW08O5i5D7H8J841hS9hzmrxd/YZNtGE4UWj87k673Dadd1zivDdKzMvI5uM/9HrYxJ6Nv1Qb7quV08QjUjh1M5cThNOISInye//C+M2xde5SW7WNo2bF+qdVPnU6V5fP2cPTAWQaMaEN84zolthdCCOEmgZqoVRw2h1cGzDejVnLBAmu2lYLsAi3AK1rxsVDGySyvb8ll6mPtpgQFo4uNx5nk3lDYfNs9GHv08W6n12O+5Q6f/ro6kQT/98sSn8N841hsK5agJp/GfOcDGHu5ioYEvvAmOU88gJqRjunaEejCvT98K4qCPqEJ+oQmmG+5A9vSBRRMmaytqbMtXYhtzUrIz3N38symWSxY7n0IRVEwj76VvPdd2yIUBmx+2e04dmwlb8dWnGmpWG65A+fJE+S+9SLkun6n1PQ01PQ0dAf3c8VNZgZMHM+ODcfZvTmJXZtPcKbINOTIesE0bV0Xa4GdneuPYStmjRxAoJrHlc6t9HLuxISd/s5tzNb1YI2+rdamSaso1N8mePVLOLEBcBV0Kdz+wJ+mzhNkKEGcUcIBSE7K4rNXFvpt63SqZKTmaYFbabLsRn7X92KJriN9v13OqVV6tuzNxUlLrQLoxhTY+NYSjCY9bbrE0rlXIzp2b8jOjSe0IDBUzSG+fgC6+q791OqSTnyIleNZrr8DP35xPvc/cwXN2rper8PhZN6Urcz5eSuqU2Xuz1tp3rYew8d2KjZgs9kcTPzgb22/vXXLDvHAc1fQ/vIGPm2FEEL4kkBN1Cqe2TTwzaj5y3w17tmIoxuO4bC5vjHPOp3tEaj5z6hlJmVi8Sh6YAmVQK2207frqAVqhm69MN9+T6XeXxcZRch3U1Czs9FFRrmft0Ejgr+b4toWoEXrEu+hGAyYBl2Dsc+V5L7yNPaNa10XPII0810PYr7hFuyb1+M4lIixaw/tw75xwNXkT/oK1WMNHoASEopSJxIlJBTHoUTIcQdZBf/3Bbr6sVh/+dF76qYH68xfCRlxI5f1SdA2ZE5OymDPxJkoyxfQyHmSCHM8pqbXYN+5jZy8v9mjNGS7rjFnjFFEdmlLvfhwwveuJWLbUhqppzFhxzTqFgztOsEbzzPKsYJW6jFmGvtRYAnlqsY5qJu9iwfFb52HwXSP3+xZEPl0deyhu3M30WRiQ898XVeWGTqVmCHzx2DUExMfSnhkIHabE2uBnYKkU5zOULErrn+2s5Qg5ul7wN48fPZoOMdmdbB1zTG2rjmGTqcQ4PFFVCv1KIa4OHSx7o2v+4Yc5+esJgCkn83l/af/4Ia7u3JZ7wS+e385B3ae9rr//p2n+e9zf9G8XT2uuakDbTrHotO7Cqzk59r44o3FXsVZrAV2Pn91Ebc90os+V0tlSSGEKI0EaqJW8Vw3pugVjAHevwImPyXAe97XjdSjaWSccJXdzzydTVTTSACyzxaXUctEkYya8GC58wHUnGyU0DAC7n8ERVf5FQMVswXF7FsVURcWji4svOz3sVgIfO09ct94Hvvqv7XzpuGjMN86DkVRMPbsi7FnX+9+RiNBr71H/k+TUIKCMXS6DEOny9BFuYtuqA47jr17yHvnZZwnT4CqkvfGC173CXjiGXQNE8h940XUsylgs5I/8SsCn37ZdY+cbIK+epOOq5ZrfZyJ+8hP3AeAGeioHqSj4yA4wNL6IQy9W5I9bQqorgIi5rF3a/v0Bb74FrmvP0dbxxFaW/+H0qQDulUpFI2vjA4rfRLyWXrAhMGop1mburRsE0XC1tnU3zofPSoEh6CPbQ37dnOtcy2X9WrMLycbc+xgKhHRQbTpHEvrzrG0ahyIMTiQzDyVzLQ8MtLyMJkNxDQII6pusBbwAORPmUzBd9+SSSBLdJ1YY2iLXfX+81NHzaCfcT8h+ans1CWwS9+YfNU97dHpVMnxKHDU2nkEXf0+6Oq717J1ydmG/b5bmP5/63E6VJxOlWnfrmfGxA04He6fhiXQSH6ue2r3/h2n2b9jARFRgfQY2IyO3Rvw84Q1HDngWyXX6VSZ/PFK0s7kMGxMpzJvHC+EELWRBGqiVinI8l6fVvRDgk6vwxRkwnou89ZiQDOim0YSGhOiBWpZHuvUPDNqYXGhWpuMpEyvDa6LZu5E7aOLrkvQK+9W9zDKTDGZCHzpbfI+fQ/b/LkYBw/F8vCTpX6w1jdvRdDL7xR/X70BQ5t2BL7+PtmP3uuTRTNdOxLT0OsBsNx5P3kfvAmAbdGfOEbdgjMtjfzP3veaRur3ecLCUTPSAcj/+XsMm9eD0xWk6erHYR57l9bW2Ls/loeeIP/T99EB7NnmDtJ0ekwjRmOdORWAa4/8wpAvpxBcLwJl23ryPn4D9fQp7V6WOx9AMRrJ27cbgPpb/uL5n38nJ8dBULARx6rlFPz8Go79e7Dq9QS1bENYxy40rBPpWtSWpMPZui26lq79Gq1/zqbgXIGZUHIZ2fQsw58bwR/fr2H9yqPUdabT17md9rEOgl9+m+wHbqOD4xB2xzJO3PsG21MsbF59lMw0d1bUpNporp5AVz/eK6PG2WQGXducJq2i+frtJdrG6Z5BWvcrmzDm4Z4cP5jK7B+3sGere3PwtDO5/DF1G39M3eb1XnTr34SB17fh89cWaeOY/eMW0s7kMvYfPdHra942B0IIURUkUBO1SkkVHwvVb1uPI+uOYbAY6HF3VwBC6gZr17NOu6dteZbmj21f3x3MJWcTWi9EuyZTH0VNpBgMBD7xLDzxbKXfW9+oMYEvvEHuC/8Ep2sqoa5JMyzjH9PaGAcPpWD6zzgPHwRVJftfD3tNmwQwXN4T8x33Yl+5DOvi+WAtwHzHfRj7DSDrzhshOwtyc7BvWq/1Md8zHsVo9LqPafgoHPv3Yvtzttd548Crsdx5P9b5c13PnZuD8c0nKSjI99pKAVzTW03DrofcXPI+fQ/sdtT0NBwb12POSCNnymScxzw2AXc4cOzajmPXdp+fj+HyHhi69iT/q4+1c7rYeILe/BBdRB3GPD+MG/9eQu57r6MEBhH00n/RN26K4fKe2NetwoCTZvsW0eHFN7n1oZ4c2pvC5lVHODprIT3yt2DGjq5+LLp69UGndwWxqorzVBJNWyfwwqcjmPjBcnZsOOH6mQUYGPNQT3oObAZA83YxPPn2EPZtP8VfP65n5/YUnKpvED/gmqbc9HBfdDqFZ/57LZ+8tIBT54rArPhrH5F1g7j21k4+/YQQQkigJmoZq2fFRz/THAGGvDiQvQsPUL9dPSIahAMQEuMOuorLqMW2q8fuv/aCCqpDJSXRPe1Hqj4K4ct4eU8CHnuKvE/fR4mIJPCFN72mbip6PZZ7/0HuC0+6TngGaYqC+fZ7MI+9G0Wnw9CqLZZ7HkJVVS3rZxlzJ/lff+r1nPpWbTH2G+gzFkVRCHjkXzgPJ+LYs8v9HDffjhIYhHnYSAqm/g9AqwTpvqke8023YR57F4reACGhGLr2wL5mBQC5L/1by+aVlX39Guzr17jHVyeSoHc+QRfhrppo7Hslod17g17nel7ANGwk9nWrALCtXIrjwF50TVvQtHVdmjQNI3PqoxQumtPHxrs2fa9bT6vO6Tx5An3DBELCLPzjlcGsWXyAk0fT6TukJXVjQ73GqDocNNqziDu2fUWWTc9mXTPW61pySolEUVWudq5n4KqZOAfVQde6HVH1Qnj6/Wv57MU/SdyXSqPmUQy8vi1CCCH8k0BN1Cpem10XMx3REmqh4w3tvM6F1nNn1DI9qs15lucPrR9KcHQw2ef2WivIkg2vhSiNaej1GPsPBqMRxeT75YmhW0/0nS7DsWWjdk7f6TIC7nsEfYtWPu09p2aaRoymYNav3lMTH3ik2OmbislM4EvvkPOvh3AmHcc04kb0jRq77nX9jRT8Ph3yvNel6lu3I+CJZ9A3buZ13njlVVqg5hWkKQrGvgMw3XQbAI5tm7Dv3gFWKyjgPJOC88A+74EFBRP09kfo6sdSVNGfmaFbT5TouqgpyeBwkD1+HLrYeAx9+mNo1U4L0tDpUaJdFR31sfHYCwM1jymlOp1Cr0HNtcdqTg6OfbtR7Taw2SiY9jOO7ZsBCAH6ObfTr85pzoY1Qk3cSxSZcMYVqIZMno4SEEhQiJkH4nby+75UhjRugMmaAwHhft8PIYSo7SRQE9VGVVVseTZMgSWXxK9MeenudRrlyXKF1PPNqKmq6lWePygqkLDYEC1Q8yQZNSGKpwQFFX9NUQh44lny3n7JtQ/cmDsxdOtVpiIUismM5b5/aMVKDH0HuKo8lkAXXZfgr37AeSYZncem0bqougT951Psq5ajhIahi41HF98QXYNGfsdi7NmXvMBAbbsBDAaMg67BfNNt6Bs00toZWrbG828HVVWxr11J/sQvXfvfBQQS9Np76Js0pywUvQHTdaO1dW3gCr6sv/yIZ81bpW49FIPrI4AuNg42uc7bVi1H36gJ+rbtteymqqrY5s0i7+tP3K+n6PPWicJy70MYr7yKUEXB+uuP5E/6ChwO1PQ0rPNmYR51K46jh1EWzmGE0wlzwdasIeZhI8v02oQQoraRQE1UC9WpMuOJ2ZzYfpI+D/Sgy80dq+R5c1M9AqvIwDL3C/HIqOWczcVudWDLteL0KNMdHBlEWP1QTmw56dNfMmpCVJw+Np7gT/+vQn1N/Qeh6A04T5/ENPyGMvVRLBb08Q19zhtatcXQqmxT9ZSAAAKfepmC6T+jb9EK8w23oqtbr/R+ioKxRx8M3XrhPHgAJTLKa7pjWZhH3Yqi02FdMt83O3eO3qOIiGdA6tiykZwtG8FoRN+6HYZOl2HfvgXH5g3FPp/xikFYHvk3utAw9xhuuQNnRjrWaT8BUDDtJ0zDR5E/8Uv3msS4BpiGDC/XaxNCiNpEAjVRLU7tOs3xc/vrbJyytcoCtRyPTWUD65QjUKsb7Nqq6NysoeyUbGx57vLUBrMBU7CJsCJrOAp5VoAUQlQtY58rqud5e/fH2Lt/hfoqOh36ZhXba0wxGl1r5m66DWfyKWwrl2Fb9BeOvbu0NnqPgNPQvTdM/ApsHjk3mw3Hts04tm32vrklAF10XdDpUELDMV13A6YrBvsdh3nUrVhn/Qo2G+qZFPInfIh9xVL39bse0LJ6QgghfMnfkKJaZHlMD8xNy8Vhd6A36C/483pl1MoRqOmNeoIiA7Wpjlmns3FY3etOgqICURSF0Pp+AjWl+MIlQghxIenqxmAeeTPmkTfjOHII29KFoKqYbxyrtdE3TCDkuynYVizBvnkD9h1bfdbiARh69SPg0ae8NlQv8bmjojEOHopt3iwArHNnup+zeSuMfQec56sTQohLmwRqolrkemS2UCEvLY/g6ODiO1SSHI9ALTAyoFx9Q+qGeARqWTid7r2FgiJda2z8ZdTMQSavza+FEKI66Bs1Rj/uPr/XdPVjMd84FvONY1Htdhz7dmPfshH7lg2QnY1p9BiMVw4u9wbV5ptuc2154HR6nbfc+9AF2fRdCCEuJRKoiWrhGTC5HldNoOYZIJYnowYQGhPMqV2nAcg8ne31gSU4ynUvf4GaTHsUQtQkisGAoU17DG3aw5g7z+te+rgGGPsNcGXyzjF0uRxDl27nOUohhLj0yddZolp4bhTt7/GFYMu3ee2jVp5iIlCk8uOpLK891IKiXBm1gHALxgDv7z+kkIgQojYz33y712PLPQ9X00iEEKJmkUBNVIvcIhm1oo8vzHN6TLdUICCsnFMfPSo/ZiVnewWXwecCNX/r1KQ0vxCiNtM3a4nlgUfRNWqM5eEn/e5/J4QQwpdMfRTVoujUx6oI1DyfMyA8AJ2hfN9TeGbUMk9lYfbYMNszOxcWG8rZg6naY8moCSFqO/PoMZhHj6nuYQghRI0igZqoFkUDs6qY+ljRio+FQj0yatnJ2djy7O77Rbk37A0rklGTQE0IIYQQQpSXTH0UVc5pd5Kblud1rmiG7ULwDAYD65Rv2iNASIw7o+awOb0DvyjvjJonmfoohBBCCCHKSwI1UeXyMvK0jaMLVfUatYpk1MxBJkzB/vdDC6rjzqiFxoZ4XZNATQghhBBClJcEaqLK+ZvmWNVTHwMrEKgBhNYL8TlnCjJhCjRqj32mPoZKoCaEEEIIIcpHAjVR5fxNc8xNzUNVVT+tL8zzVmTqI3hXfixUtMx/aEwIeOwJKxk1IYQQQghRXhKoiSrnL3tmL7BjzbVd0Oc932Ii4D+jFhwd5PXYYDZ4tavocwkhhBBCiNpLAjVR5Ypbj5Z7gac/5nisUavo1Ef/GbUgn3Pd7+qKKchEk94JxLStV6HnEkIIIYQQtZeU5xdVrrj1aDmpuUQ0DL8gz6mqaqVk1PwGalG+92ozpCWtBjdHp5fvQoQQQgghRPnJp0hR5TyrL3qfv3AZtYKsApx2p/Y4MLKCUx9j/Ex99JNRAyRIE0IIIYQQFSafJEWVyzmbU8z5CxeoeRYS0Rt1mIsps1+aED9r1Pxl1IQQQgghhDgfEqiJKueZUQuu655KeCE3vc4tsj5NUZQSWhcvMCIAvdH71yYoyn9GTQghhBBCiIqSQE1UKVVVvTJndZtHaccXspiIZxavooVEABSd4hVcAgRXcBqlEEIIIYQQxZFATVQpa64Ne4FdexzVLFI7royMWubpLNZOhoGUrgAAJ49JREFU2kDSjlNe5z0zaudbLr/oOrXAYtaoCSGEEEIIUVESqIkq5Zk1U/QKUY3ruK8VU2SkrJx2J7/9ay5rJm5g5j/nkJvuvl9lbHZdKMQjo2YJs2Aw6c/rfkIIIYQQQhQlgZqoUl4BU0QAQR6bRRdXZKSs9i1JJO1oOgD2fDsnt7uzap4BYtB5TlUM8cioBcv6NCGEEEIIcQFIoCaqlNdeZpFBXtMQ8zLycdgdFbqv6lTZ8OMmr3PJ+89ox94ZtfML1KKbuqdr1kmIOK97CSGEEEII4Y9seC2qlGchkcA6Ad7TEFXIS8+vUJbq4MrDnD2U5nUuxSNQq8w1ak16J9D5pg5kJmXS4+6u53UvIYQQQggh/JFATVQpz8xWUJ1AjBYjpiAT1hyr6/rZ3HIHaqqqsv6HTT7nvQO1ysuoKTqFfg/3Oq97CCGEEEIIURKZ+iiqlL+1Yp5ZtdwKVH48uuE4p/ek+JzPTskhLz0Ph91BXka++3nPs5iIEEIIIYQQF5oEaqJK+Vsr5lncoyIl+j2zaQ0vjycg3KI9Tt5/hrw072qS55tRE0IIIYQQ4kKTQE1UqRw/GTXPNWPl3fQ6aftJTmw5qT3udnsXoj020U7Zf4Ycj/VppmATBrPM+BVCCCGEEBc3CdRElcotskYNvDNc5c2orf9hs3Yc2z6GuI6xRQK1s36fUwghhBBCiIuZBGqiyjjsDvLS3WvFAiN9pz6WZ9PrtOPpHF5zVHt8+e1dAKjrGagdOFOpm10LIYQQQghRFSRQE1Wm6Foxvxm1cmx6fXi1O0irkxBBo24NAIhu5g7U0o6lk3Ei0+c5hRBCCCGEuJhJoCaqjOf6NM+1Yl5r1MqRUTu81h2oNe7VCEVRAAiPD8MYcG4dmuqqCllICokIIYQQQoiaQAI1UWVyitl0OtCz6uPZXFRVLfVetnwbJ7a6i4gkdG+oHSs6haimkdrj5L3u0v0SqAkhhBBCiJpAAjVRZXI9pjV6Bmqex/YCO9ZcW6n3Or4lCYfVAYApyET9dvW8rnsWFPEke6gJIYQQQoiaQAI1UWU8M2qema2AMAuKXtEel2XT6yNrjmnHDS6LQ2/Qe10vLlCTjJoQQgghhKgJLppALS8vj3feeYcuXboQHBxMSEgIHTp04LXXXiM9Pf287z9lyhQGDhxInTp1sFgsNG3alPHjx7Nv377zH7woE89CIZ6VHhWdQmBEgEe70gO1w+vc69M8pz0WqltsRk0CNSGEEEIIcfG7KHb+TUpKYvDgwezatcvr/Pbt29m+fTsTJ05k7ty5tGnTptz3ttlsjBkzhmnTpnmdP3jwIF9++SWTJ0/m+++/Z/To0ef1GkTpcr0yat5TEIPqBJJzJvdcu5IDtbTj3pUcC6s9eqqTUAedXofT4fQ6Lxk1IYQQQghRE1R7Rs1utzNixAh27dqFoiiMHz+exYsXs3TpUh599FF0Oh2HDx/muuuuIzMzs/QbFvHvf/9bC9IGDx7MrFmzWLVqFe+//z7h4eHk5uZy2223sXHjxsp+abWKvcDO4TVHycvIL7aN18bTkUFe17wKipQSqHlOe4xsXIeQusE+bQwmPXUSIrzOKTqFgHBLifcWQgghhBDiYlDtGbVvvvmGDRs2APDxxx/zyCOPaNf69+9Pz549ufXWW0lMTOSDDz7g1VdfLfO9d+zYwaeffgrAqFGj+PXXX7US7j179uTaa6+lZ8+epKen8+STT7Js2bJKfGW1hzXXxrRHfiPlwFlC6gVzy9ejCAz3LdrhOaXRX0bNXzt/PMvyJ3T3zaYVim4eyZnEs9rjgPAAdPpq/25CCCGEEEKIUlX7p9ZPPvkEgNatW/Pwww/7XL/lllsYNmwYAJ9++il2u73M9/7ss89wOp0YjUY+/PBDLUgr1KpVK1544QUAli9fLlm1ClBVlYX/WUrKAVdAlHU6m4XvLPUpsa+qqlemzCej5rWXWvGBmr3AzvEtSdrjRn7WpxUqWlBEKj4KIYQQQoiaoloDtT179rBnzx4AxowZg07nfzjjxo0DIC0tjSVLlpT5/rNmzQKgX79+NGjgP/Nyxx13aMczZswo872Fy+ZftrF/SaLXuUOrj7B9lvd6Q2u2VSunD75Bk2dxkZI2vfYsy28MMBLbPqbYtkULisj6NCGEEEIIUVNUa6C2atUq7bhfv37FtuvTp492XNbpiQcPHuTUqVOl3js6OpqWLVuW697C5dimE6z4co32WG90/3Fa/vkqzh5O1R57ZtN0Bh2WUO+1YoFlnPp4xGPaY8PL4tAb9cW2jWomgZoQQgghhKiZqjVQ2717t3bcrFmzYtvFxMQQEBDg06cy7g3QuHHjct1bQObpLP54ZQGq0zXFMaReMLd+M1or1uGwOvjz9UXYz2W/vNanRQSg6LynoXpm2EoqJnJ4rbuQSEnTHgHMQSbC4kLdzxEpgZoQQgghhKgZqrWYSFKSa62RwWCgfv36JbaNjY0lMTFR61PWewM0bFjyB/q4uDgAUlNTKSgowGw2l+k5ipOcnExKSkq5+hTdmuDAgQPnNYZCZxLPkHYs47zvo6pQkF1AdnI22WdyST+WTn5WAQB6o55Od7TlVO5JGt0Ux5IP/wbg5NYkMh9LI6JhBKlH0jiZ43pPIuvWYefOnV73zzqTrV0nF6Z//BuoKqpTxZ5vpyDHSn5mAYf2HtH65IXn+NynqILIPE7uc933jCOu1PZCCCFqvqZNm2KxSJVfIUTNVq2BWlpaGgBBQUE+hT6KCgpyFZ/IyChb0FF4b4DgYN/y7f7uXXj/unXrluk5ijNhwoRyVaf05/rrrz+v/lXtreGv+L/grz7LRnhqmp/znjaU/pzv9n+99EaljUUIIcQlZ8eOHbRt27a6hyGEEOelWqc+FhS4MjJl+darcOpjYZ+y3rss9y+8d3nuL4QQQgghhBAXSrUGaoVVHkvLpgFauffiKkMWd++y3N+zlHxZ7y+EEEIIIYQQF0q1Tn0snJKYn59fatvCNmVdP+Y53bG0+3teP9/1aQAPPfQQN954Y7n6ZGZmsmHDBkJDQwkPD6dBgwYVGsuBAwe8pk3+9ttvpRZTETWLvMe1g7zPtYO8zxdG06ZNq3sIQghx3qo1UAsJCQEgN7f4Kn+FcnJyAIiIiCjXvT37lnZvRVEIDw8v0/1LUrdu3Qqtc+vZs+d5P3dRzZo1k3n6lzh5j2sHeZ9rB3mfhRBCFKrWeX6NGjUCwGq1llolsbCKY2xsbLnuDXDixIkS2xZej46OxmCo1thVCCGEEEIIIao3UGvdurV2nJiYWGy7kydPkpeX59OnMu4Nrs2xy3NvIYQQQgghhLiQqjVQ69atm3a8cuXKYtutWLFCO+7Vq1eZ7h0bG0t8fHyp905JSWHfvn3lurcQQgghhBBCXEjVGqg1a9aM9u3bAzB58mSv6ouevv/+e8C1Pq1///5lvv/IkSMBWLBgQbEbZRfeG2re3mVCCCGEEEKIS1O116IfP348ANu2beO9997zuT516lTmzp0LwP3331+uSoj3338/er2egoICHnzwQRwOh9f1vXv38uabbwLQvXt3rwyfEEIIIYQQQlSXag/U7r//fjp37gzA008/zbhx41i4cCHLly/niSeeYOzYsQAkJCTw7LPPevU9fPgwiqKgKApXXHGFz73btWvHQw89BMDs2bPp168fM2bMYPXq1Xz44Yf07NmT9PR0jEYjn3322YV9oUIIIYQQQghRRtVe4lCv1zNnzhwGDhzInj17mDx5MpMnT/ZqExcXx5w5cwgLCyv3/T/44AOSkpKYPn06q1atYtWqVV7XzWYz33//PV27dj2v1yGEEEIIIYQQlaXaM2rgKvyxefNm3n33Xbp06UJISAhms5nWrVvz7LPPsm3btgrvK2M0Gpk2bRpTpkxh0KBBREZGYjQaiY+P584772TTpk3cfPPNlfyKhBBCCCGEEKLiqj2jVshisfDUU0/x1FNPlblPQkJCsQVIirr55pslIBNCCCGEEELUCBdFRk0IIYQQQgghhJsEakIIIYQQQghxkblopj6KyhEdHc3LL7/s9VhcWuQ9rh3kfa4d5H0WQghRHEUt6yIvIYQQQgghhBBVQqY+CiGEEEIIIcRFRgI1IYQQQgghhLjISKAmhBBCCCGEEBcZCdSEEEIIIYQQ4iIjgZoQQgghhBBCXGQkUBNCCCGEEEKIi4wEakIIIYQQQghxkZFATQghhBBCCCEuMhKoCSGEEEIIIcRFRgI1IYQQQgghhLjISKAmhBBCCCGEEBcZCdSEEEIIIYQQ4iIjgZoQQgghhBBCXGQkULtE5OXl8c4779ClSxeCg4MJCQmhQ4cOvPbaa6Snp1f38GqNAwcO8Nhjj9GuXTtCQkKwWCw0btyYO+64g/Xr15faf8qUKQwcOJA6depgsVho2rQp48ePZ9++fWV6/rVr13LTTTcRGxuLyWQiLi6OG264gUWLFpWp/7Fjx3jkkUdo3rw5FouFqKgo+vbty3fffYfT6SzTPWqj//znPyiKgqIo2O32EtvKe1wzzJ8/n5tuuokGDRpgNpuJjo5m6NChzJ49u9S+8h4LIYSoFKqo8U6cOKG2adNGBfz+l5CQoO7cubO6h3nJ++qrr1STyVTs+wCozz33nN++VqtVHT16dLH9AgMD1V9//bXE53///fdVnU5X7D3+/e9/l9h/6dKlalhYWLH9BwwYoGZlZVX453OpOnDggBoQEKD9nGw2m9928h7XDDabTb3jjjtK/D2+//77VafT6dNX3mMhhBCVSQK1Gs5ms6ldu3ZVAVVRFHX8+PHq4sWL1aVLl6qPPvqo9g9+06ZN1YyMjOoe7iXrt99+UxVFUQG1Tp066uuvv64uW7ZMXbVqlfrxxx+r8fHx2gel999/36f/Y489pl0fPHiwOmvWLHXVqlXq+++/r4aHh6uAajab1Q0bNvh9/pkzZ2r9mzdvrk6aNEldvXq1OnnyZK8g/rPPPvPb//Dhw9qHu5CQEPXdd99VV65cqc6ePVsdOnSo1n/06NGV+nO7FAwcONDrg3BxgZq8xzXDXXfdpf0sOnfurP7www/q6tWr1e+//15t3ry5du29997z6SvvsRBCiMokgVoNN2HCBO0f308++cTn+s8//6xdf+mll6phhJc+u92uJiQkqIAaFRWlHjhwwKdNSkqK9iEvMDBQPX36tHZt+/btWkA9atQon2/qd+/erX3I69evn8+98/Pz1QYNGmgBeWpqqtf17Oxs9bLLLlMBNSwszOe6qqrqTTfdpAKqyWRS169f73P9wQcf1P4cLV68uMw/m0vdd99955Ox8BeoyXtcMyxcuNAr0MrPz/e6npaWpv2uh4aGqrm5udo1eY+FEEJUNgnUarhWrVqpgNq6dWvV4XD4bTNs2DAVUCMiIor9tl9U3KJFi0r9pltVVXXWrFlauy+//FI7/8ADD6iAajQa1aNHj/rt+/7772t9i34b7xmMT5kyxW//DRs2FJvRO378uKrX61VAffDBB/32z83NVevWrasC6rBhw4p9jbXJqVOn1IiICC1ALylQk/e4ZujXr58WCJ09e9Zvm6+++kr7Of/+++/aeXmPhRBCVDYpJlKD7dmzhz179gAwZswYdDr/b+e4ceMASEtLY8mSJVU2vtpi1apV2vHw4cOLbTdgwADtePv27drxrFmzAOjXrx8NGjTw2/eOO+7QjmfMmOF17bfffgMgMDCQkSNH+u1/2WWX0bZtW7/9Z8+ejcPhAGDs2LF++wcEBHDjjTcCsGDBArKysvy2q00eeeQR0tLSGDBgANdee22JbeU9vvglJSXx999/A/DEE09Qp04dv+2uu+467rjjDp588kmioqK08/IeCyGEqGwSqNVgngFCv379im3Xp08f7XjZsmUXdEy1Uc+ePXnmmWe47bbbiIuLK7adZ7W1goICAA4ePMipU6eAkt/D6OhoWrZsCfi+h4V/Drp3747JZCr2HoV/DtatW0deXp5Pf7PZTPfu3UvtX1BQwNq1a4ttVxvMnj2bX3/9FYvFwpdfflliW3mPa4YFCxagqioAo0ePLrZdTEwM33//PR988AE9e/YE5D0WQghxYUigVoPt3r1bO27WrFmx7WJiYggICPDpIyrHwIEDefvtt/nf//6HXq8vtp3nB7PCb9zL+h4CNG7c2KdPTk4Ox44dK1d/u93OgQMHtPOF92vYsCFGo7HU/kXHUNtkZWXx0EMPAfDCCy/QvHnzEtvLe1wzFGa5AwICaN26tXY+Ly+P/fv3c+zYMS2QK0reYyGEEBeCBGo1WFJSEgAGg4H69euX2DY2Ntarj6h67733nnZ81VVXAd7vR8OGDUvsX5itS01N1TJyFelftF/hcUX71zZPP/00x48fp23btjz11FOltpf3uGYoDFri4uLQ6XRs2bKF4cOHExoaSosWLWjYsCFxcXG88cYb2ntTSN5jIYQQF4IEajVYWloaAEFBQSiKUmLboKAgADIyMi74uISvzz//XFv/0q1bN3r06AG430OA4ODgEu9R+B6C+3083/6e96ho/9pk5cqVfPnllyiKwtdff11i5qKQvMc1w9mzZwEIDw/np59+olu3bsyZM8drA/OTJ0/y4osvMmjQIDIzM7Xz8h4LIYS4ECRQq8EKv421WCylti2c+lj0m2Bx4c2fP58nnngCcGU/P/74Y+2a5/tR2vtY+B569jvf/p7HFe1fW1itVu677z5UVeWBBx6gV69eZeon73HNkJ2dDcDRo0e56667CAgI4JNPPuHkyZPk5eWxatUqrrzySgBWrFjBvffeq/WV91gIIcSFIIFaDVZY5bG0bBqgra0orjKkuDCWLl3KDTfcgM1mA+DNN9/Usmng/X6U9j56ro8p7He+/QFtXV1F+9cWb775Jrt376Z+/fq88847Ze4n73HNUFiYIzk5GUVRWLhwIY888ggxMTFYLBZ69uzJ/PnzGTRoEAC//vqrVsBD3mMhhBAXgvwtXYMVTnHJz88vtW1hG7PZfEHHJNzmzZvH0KFDycnJAWD8+PE+a5o8pymV9j56Xi98H8+3P7inQlW0f22wa9cuLTj75JNPCAsLK3NfeY9rBs9M1AMPPMDll1/u08ZgMPDhhx9qj3/66SdA3mMhhBAXhgRqNVhISAgAubm5pbYtDBYiIiIu6JiEy6RJkxgxYoT2Lf19993H559/7tOu8D0E93tUnMLriqIQHh5e4f7g/eeg8B4V7X+pczqd3HvvvVitVoYNG1Zi6XZ/5D2uGTx/ztddd12x7dq1a6dVbV2/fr1PX3mPhRBCVBZDdQ9AVFyjRo0A19qZlJQUoqOji21bWN2rsPqjuHDeeustnn/+ee3xo48+ykcffeR3SlLhewhw4sSJEu9beD06OhqDwfWrGx8fj16vx+FwlLk/eP85aNSoEceOHatw/0vdV199xerVq9Hr9dx3331s2bLFp01qaqp2vHXrVvR6PSaTiTZt2sh7XEPExMRox6W99vj4eI4dO8aZM2cA+T0WQghxYUhGrQbz3OsnMTGx2HaFi+GL9hGV7+mnn/YK0l555RU+/vjjYteNlPU9BNemukX7mEwmmjRpUq7+RqPRa6+mwvsdPnwYh8NRav+iY7jUFW4K7HA4GDFiBJ07d/b5b/bs2Vr7rl270rlzZ4YOHQrIe1xTtGvXTjv2rMLoj9VqBdAyYvIeCyGEuBAkUKvBunXrph2vXLmy2HYrVqzQjstaqU6U36uvvsp//vMfwLVI/8svv+Tll18usU9sbCzx8fFAye9hSkoK+/btA3zfw8I/B2vWrMHpdBZ7j8I/B127dvUqK1/YPzc312+2qGh/o9Hod/2O8E/e45rB8+/TdevWFdvO6XRq71NCQgIg77EQQogLRBU1Wvv27VVA7dChg+p0Ov22ufbaa1VAjYiIUPPz86t4hLXDzJkzVUAFVIPBoP78889l7vvII4+ogGo2m9UTJ074bfPee+9p91+7dq3XtenTp2vXpk+f7rf/+vXrtTbvvvuu17Xk5GRVr9ergPrII4/47Z+bm6vWrVtXBdRrrrmmzK+tthg3bpz287XZbD7X5T2++OXn56thYWEqoLZr10612+1+2/3222/az3nSpEnaeXmPhRBCVDYJ1Gq4CRMmFPsPt6qq6pQpU7TrTz/9dDWM8NJ35swZNTo6Wvs5f/XVV+Xqv337du0D1vDhw30+IO7Zs0cNDw9XAbV79+4+/fPy8tSGDRuqgJqQkKCeOnXK63p2drbatWtXFVCDg4PV5ORkn3vcdNNNKqCaTCZ1+fLlPtfHjx+vvb4//vijXK+vNigtUJP3uGZ47rnntJ/BP//5T5/rSUlJaqNGjVRAjY6OVjMzM7Vr8h4LIYSobBKo1XB2u13t3Lmz9o/vHXfcoS5YsEBdtmyZ+vjjj2sfHBISEtT09PTqHu4l6cUXX9R+/t27d1c3b95c6n/79+/3ukfht/GA2qtXL3X69OnqqlWr1P/+979qRESECqhGo1Fdv3693zFMmzZN69+gQQP1q6++UlevXq3+8MMPatu2bbVrH3/8sd/+hw8fVoODg1VAtVgs6iuvvKKuWLFCnTt3rpaRBdSRI0dW+s/vUlBaoKaq8h7XBLm5uWq7du20n8WgQYPUadOmqWvWrFEnTJigxsXFademTp3q01/eYyGEEJVJArVLwIkTJ9RWrVpp/wgX/S8uLk7dsWNHdQ/zkuX54a2s//Xv39/rHlarVR01alSx7c1mszplypQSx/Hee++pOp2u2Hs8/vjjJfZfvHixGhoaWmz/vn37emUQhFtZAjV5j2uG5ORktXv37sX+jPR6vfrZZ5/57SvvsRBCiMokgdolIi8vT3333XfVLl26qCEhIarZbFZbt26tPvvss+rZs2ere3iXrJSUlHIHaf4CtUJTpkxRBw0apEZGRqpGo1GNj49X77zzTnXnzp1lGs/atWvVW2+9VY2Li1ONRqMaGRmpDh06tMzTnE6cOKE+9thjavPmzVWLxaIGBwerPXr0UCdMmFDsmh1RtkCtkLzHFz+Hw6FOnDhRHThwoBodHa1aLBa1WbNm6j333KNu27at1P7yHgshhKgMiqqqKkIIIYQQQgghLhpSnl8IIYQQQgghLjISqAkhhBBCCCHERUYCNSGEEEIIIYS4yEigJoQQQgghhBAXGQnUhBBCCCGEEOIiI4GaEEIIIYQQQlxkJFATQgghhBBCiIuMBGpCCCGEEEIIcZGRQE0IIYQQQgghLjISqAkhhBBCCCHERUYCNSGEEEIIIYS4yEigJoQQQgghhBAXGQnUhBBCCCGEEOIiI4GaEEIIIYQQQlxkJFATQgghhBBCiIuMBGpCCCGEEEIIcZGRQE0IIYQQQgghLjISqAkhhBBCCCHERUYCNSGEEEIIIYS4yEigJoQQQgghhBAXGQnUhBBCCCGEEOIiI4GaEEIIIYQQQlxkDNU9ACGEKI99+/Yxd+5c/vrrLw4dOkRycjJ5eXlERkYSHR1Nu3btuPrqqxkyZAjR0dFlvu/vv/9OQEAAgwcPvoCjF0IIIYQoG0VVVbW6ByGEEKU5ePAgjz76KHPnzi1Te5PJxN13381bb71FREREse2OHj3Ko48+yqxZs5g4cSJ33nlnJY1YCCGEEKLiJKMmhLjorVixgmuvvZbM/2/v/mOqrB44jn8uVySHaBo/A0kQpcLVpEaEV4Vcu1BQblquTU1ksYphyaK2lFWImyv/CLHUrVRInVqmVI4kU8ryR5S1aZhAgS0pQCVQVBDw+0frGVcul4v6ree292tje87hnPOchz/YPjvPc05bm1FntVoVGRmpoKAg+fj46Ny5c6qpqVFLS4skqbOzU2vWrFFZWZl2796t6Ohop2NPnz5dtbW1/8hzAAAAuItv1ACYWkNDg2bMmGGEtMDAQK1du1ZnzpxRdXW19u/frz179ujw4cM6e/asKisrNXPmTKP/yZMnlZKSotbWVqfjX758+R95DgAAgMEgqAEwtVdffVVnzpyRJIWEhKiyslKZmZkaOXKk0/b33nuvPvjgAy1dutSoq6ur07Jly/6R+QIAANwIBDUAptXZ2anNmzcb5aVLlyo8PNytvkuWLJHNZjPK77zzjrq7u2/4HAEAAP4fCGoATOv48eNqb283yvHx8YPq/8wzzxjXLS0tqqqqumFzAwAA+H8iqAEwrUuXLjmUGxsbB9XfZrNp/Pjxuv/++5Wamiovr7/+5VVUVMhischisejkyZNG+/T0dKM+MTGx33E7OztVXFysWbNmKTIyUr6+vvLz81NUVJTmzJmjDz/8UANtqFtfX2/c6/bbbzfqS0tLlZqaqjFjxmjo0KHy9/dXYmKi3nzzTZ0/f35Qzw8AADwX2/MDMK3GxkYFBwcb5QcffFBlZWWyWq3XNW5FRYWSkpJctpk2bZoqKir61JeXl+vpp59WXV2dy/6xsbEqKSlRTEyM09/X19crIiJCkhQdHa0ffvhB6enp2rJlS79jBgUFacOGDUpOTnZ5bwAA4PlYUQNgWkFBQbr77ruN8meffabExETt3bt3wBUrV0aPHi273S673a6bbrrJqP/7sGy73a64uLg+/datW6eHHnrIIaQFBATIZrPJZrPJ39/fqD9y5IgSEhK0f/9+t+bUO6RZrVbFxsZqypQpuuWWW4w2jY2NSk1N1Y4dOwb9zAAAwLOwogbA1LZt26bZs2f3qQ8NDVVaWpqSkpI0depUh5W3wRg7dqzx+qOrA6+/+uorJSUlqaurS5IUFRWloqIi2e12WSwWSVJPT48++eQTLVy40BgzMDBQ33//vW699VaH8XqvqPU2c+ZMFRYWKjQ0VJLU1dWl4uJiPf/888arj35+fjp27JjbG6sAAADPw4oaAFN7/PHH9dJLL/WpP3XqlNasWaPZs2crJCRE48eP14IFC7Rx48ZBf8s2kJ6eHi1YsMAIaRMnTlRlZaWSk5ONkCZJXl5eeuSRR3Tw4EEjhDU1NTmdvzMZGRl6//33jZAmSUOGDFFGRobKysrk7e0tSTp37pxee+21G/V4AADAhAhqAExv+fLlWr9+vUaNGtVvm9raWq1fv15z585VSEiIJk+erHXr1t2QA61LS0tVU1MjSbJYLNq4caNuvvnmftuHhISoqKjIKG/ZskWnTp1yeY+oqCitWrXKIfj1ZrPZtGjRIqO8devWPputAACA/w6CGgCPMH/+fFVXVys/P18TJkxw2fbKlSs6cOCAMjIyFBMTo6+//vq67r19+3bj2mazOXw315+UlBTj+7Kuri6Vl5e7bJ+dne3wvZwzWVlZxnV7e7s+//zzAecBAAA8E0ENgMfw9/dXXl6eTpw4oaqqKhUWFmrGjBkOG25craamRg888MB1bcDRO+jdd999bvXx8vLSpEmTnI7hzMMPPzzgmOHh4Q7ftX355ZduzQUAAHgeghoAj3THHXdo4cKF2rFjh5qbm3X06FGtXLlSaWlpGjZsmEPbzs5OzZkzRz///POg79Pd3e1w1tqKFSuM888G+tmzZ4/Rz9Wrj8OHD9e4cePcmk/vM9caGhoG/TwAAMAzENQAeDyLxaKJEycqOztbH330kf744w8VFBTI19fXaHPhwgUtW7Zs0GO3tLRc11EAfzt79my/vxs9erTb4/T+Nu5Gb5oCAADMY8i/PQEAuNFGjBihxYsXKzk5WdOmTVN7e7skaefOnXr33Xf73bDDmas3I4mNjVVAQMCg5+Rqxezv3Rzd0d3dbVxzugoAAP9dBDUAplRdXa2cnBw1NjaqqalJO3fudPjmyx333HOPsrKy9Prrr0v6a3Xs9OnTgwpaV692ZWdn93vW2rVqa2tzu+2ff/5pXLv6Ng8AAHg2Xn0EYErd3d3atWuXvv32W/366686ePDgNY0THx/vUO7p6RlUfx8fH4ew9uOPP17TPFxpbm52CGCuVFVVGdcD7X4JAAA8F0ENgClNmDBBfn5+Rrm4uPiaxmlpaTGu/fz8rum1xYSEBON6165dbvfLzc3Viy++qLfffluVlZUu27pzhEBNTY1+++03ozxlyhS35wIAADwLQQ2AKVmtVs2aNcsof/PNN9q0adOgx+ndx263y8vL8d/e1WVnUlNTjevjx4+rtLR0wD4HDhzQihUr9MYbbygrK2vArfTXrl074Ji9D9EOCAhQYmLigH0AAIBnIqgBMK3FixfLx8fHKKenp+u9995zu39BQYH27t0r6a+dIXNzc/u06X3I9KVLl5yOM2/ePAUGBhrlzMxMl1v9t7W16amnnjLKw4cP19y5c13O9eOPP3b5bOXl5XrrrbeMclZW1qA2IQEAAJ6FoAbAtMaNG6fCwkKjfPnyZc2bN09Tp07Vtm3bnG7CcfHiRX366adKSkpSXl6eUZ+Tk6O4uLg+7XtvyFFWVuZ0J8Vhw4Zp1apVRrmpqUnx8fHasGGDOjs7HdqWl5dr8uTJDt+S5eXlOQS9/qSnp6ugoEAXLlww6jo6OrRy5Uo9+uijxvd1UVFReuGFFwYcDwAAeC7LFfZ3BmByRUVFysnJUVdXl0O9t7e3wsPDFRwcrCFDhuj06dOqra1VR0eHQ7vMzEytXr3a6WuOzz77rFavXm2Uo6OjFRYWppEjR2r79u0ObfPz8/XKK6841Pn5+SkmJkZWq1W1tbV9zjZ74okntGnTpj5HAtTX1ysiIsIoh4aGGodi+/r66q677pLVatXRo0fV2tpqtAsICNC+ffsUExPT798LAAB4PoIaAI/w3XffKTc3V/v27XO7T2RkpJYvX67HHnus3za//PKLJk2a1Gd1ztvbW+fPn9fQoUMd6rdu3apFixbp999/d3lvHx8fvfzyy1qyZInTgHh1UDt8+LCee+45HTp0qN8xbTabSkpKHPoBAID/JoIaAI/y008/qbS0VIcOHVJ1dbUaGhrU3t6unp4ejRgxQhEREYqLi1NaWprsdrusVuuAY544cUL5+fn64osv1NzcLG9vb4WFhWn37t267bbb+rS/ePGiNm/erLKyMh05ckTNzc3q6OjQqFGjdOedd2r69OmaP3++wsLC+r3n1UGtrq5OY8aMUUlJiUpKSnTs2DG1trYqODhYCQkJevLJJ5WSknJtfzQAAOBxCGoA8C9wFtTGjh37700IAACYCpuJAAAAAIDJENQAAAAAwGQIagAAAABgMgQ1AAAAADAZghoAAAAAmAxBDQAAAABMhu35AQAAAMBkWFEDAAAAAJMhqAEAAACAyRDUAAAAAMBkCGoAAAAAYDIENQAAAAAwGYIaAAAAAJgMQQ0AAAAATIagBgAAAAAmQ1ADAAAAAJMhqAEAAACAyRDUAAAAAMBkCGoAAAAAYDIENQAAAAAwGYIaAAAAAJgMQQ0AAAAATIagBgAAAAAmQ1ADAAAAAJMhqAEAAACAyRDUAAAAAMBkCGoAAAAAYDIENQAAAAAwGYIaAAAAAJgMQQ0AAAAATIagBgAAAAAmQ1ADAAAAAJMhqAEAAACAyRDUAAAAAMBkCGoAAAAAYDL/A0RNUrDduAHDAAAAAElFTkSuQmCC",
+      "text/plain": [
+       "<Figure size 600x600 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
     }
-  ],
-  "metadata": {
-    "accelerator": "GPU",
+   ],
+   "source": [
+    "# Evaluate results\n",
+    "model_version = 0  # 'v_num' from training output above\n",
+    "model_path = models_dir / 'lightning_logs' / f'version_{model_version}'\n",
+    "metrics = pd.read_csv(model_path / 'metrics.csv', index_col=1)\n",
+    "\n",
+    "fig, ax = plt.subplots(1,1, figsize=(2,2))\n",
+    "\n",
+    "for col in metrics.columns.difference(['epoch']):\n",
+    "    metric = metrics[col].dropna()\n",
+    "    ax.plot(metric.index, metric.values, label=col)\n",
+    "\n",
+    "ax.legend(loc=(1, 0.33))\n",
+    "ax.set_ylim(0, 1)\n",
+    "ax.set_xlabel('Step')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e73687e1-ee8f-46e9-8bd2-1ddc571ef94b",
+   "metadata": {
+    "id": "e73687e1-ee8f-46e9-8bd2-1ddc571ef94b"
+   },
+   "source": [
+    "# **3. Make Predictions**\n",
+    "\n",
+    "Once the Segger model is trained, it can be used to make predictions on seen (partially trained) data or be transfered to unseen data. This step involves using a trained checkpoint to predict cell boundaries and refine transcript-nuclei associations."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "9807abf3",
+   "metadata": {
+    "id": "9807abf3"
+   },
+   "source": [
+    "#### **Requirements for the Faster Prediction Pipeline**\n",
+    "The pipeline requires the following inputs:\n",
+    "\n",
+    "- **segger_data_dir**: The directory containing the processed Segger dataset (in PyG format).\n",
+    "- **models_dir**: The directory containing the trained Segger model checkpoints.\n",
+    "- **benchmarks_dir**: The directory where the segmentation results will be saved.\n",
+    "- **transcripts_file**: Path to the file containing the transcript data for prediction.\n",
+    "\n",
+    "#### **Running the Faster Prediction Pipeline**\n",
+    "Below is an example of how to run the faster Segger prediction pipeline using the command line:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "PEOtAs-t9CiY",
+   "metadata": {
     "colab": {
-      "gpuType": "T4",
-      "provenance": []
-    },
-    "kernelspec": {
-      "display_name": "Python 3",
-      "name": "python3"
-    },
-    "language_info": {
-      "codemirror_mode": {
-        "name": "ipython",
-        "version": 3
-      },
-      "file_extension": ".py",
-      "mimetype": "text/x-python",
-      "name": "python",
-      "nbconvert_exporter": "python",
-      "pygments_lexer": "ipython3",
-      "version": "3.11.9"
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "PEOtAs-t9CiY",
+    "outputId": "8b7a5375-9ebc-4bb4-9421-254410319120"
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Starting segmentation for segger_embedding_1001...\n"
+     ]
     }
+   ],
+   "source": [
+    "dm = SeggerDataModule(\n",
+    "    data_dir='data_segger',\n",
+    "    batch_size=1,\n",
+    "    num_workers=4,\n",
+    ")\n",
+    "\n",
+    "dm.setup()\n",
+    "\n",
+    "model_version = 0\n",
+    "model_path = Path('models') / \"lightning_logs\" / f\"version_{model_version}\"\n",
+    "model = load_model(model_path / \"checkpoints\")\n",
+    "\n",
+    "receptive_field = {'k_bd': 4, 'dist_bd': 12, 'k_tx': 15, 'dist_tx': 3}\n",
+    "\n",
+    "segment(\n",
+    "    model,\n",
+    "    dm,\n",
+    "    save_dir='benchmarks',\n",
+    "    seg_tag='segger_output',\n",
+    "    transcript_file='data_xenium/transcripts.parquet',\n",
+    "    receptive_field=receptive_field,\n",
+    "    min_transcripts=5,\n",
+    "    cell_id_col='segger_cell_id',\n",
+    "    use_cc=False,\n",
+    "    knn_method='cuda',\n",
+    "    verbose=True,\n",
+    ")\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "3aa5002c",
+   "metadata": {},
+   "source": []
+  },
+  {
+   "cell_type": "markdown",
+   "id": "0a823035",
+   "metadata": {
+    "id": "0a823035"
+   },
+   "source": [
+    "#### **Parameters**\n",
+    "Here is a detailed explanation of each parameter used in the faster prediction pipeline:\n",
+    "\n",
+    "- **--segger_data_dir**: The directory containing the processed Segger dataset, saved as PyTorch Geometric data objects, that will be used for prediction.\n",
+    "- **--models_dir**: The directory containing the trained Segger model checkpoints. These checkpoints store the learned weights required for making predictions.\n",
+    "- **--benchmarks_dir**: The directory where the segmentation results will be saved.\n",
+    "- **--transcripts_file**: Path to the *transcripts.parquet* file.\n",
+    "- **--batch_size**: Specifies the batch size for processing during prediction. Larger batch sizes speed up inference but use more memory (default: 1).\n",
+    "- **--num_workers**: Number of workers to use for parallel data loading (default: 1).\n",
+    "- **--model_version**: Version of the trained model to load for predictions, based on the version number from the training logs (default: 0).\n",
+    "- **--save_tag**: A tag used to name and organize the segmentation results (default: segger_embedding).\n",
+    "- **--min_transcripts**: The minimum number of transcripts required for segmentation (default: 5).\n",
+    "- **--cell_id_col**: The name of the column that stores the cell IDs (default: segger_cell_id).\n",
+    "- **--use_cc**: Enables the use of connected components (CC) for grouping transcripts that are not associated with any nucleus (default: False).\n",
+    "- **--knn_method**: Method for KNN (K-Nearest Neighbors) computation. Only option is \"cuda\" for this pipeline (default: cuda).\n",
+    "- **--file_format**: The format for saving the output segmentation data. Only option is \"anndata\" for this pipeline (default: anndata).\n",
+    "- **--k_bd**: Number of nearest neighbors for boundary nodes during segmentation (default: 4).\n",
+    "- **--dist_bd**: Maximum distance for boundary nodes during segmentation (default: 12.0).\n",
+    "- **--k_tx**: Number of nearest neighbors for transcript nodes during segmentation (default: 5).\n",
+    "- **--dist_tx**: Maximum distance for transcript nodes during segmentation (default: 5.0)."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "b0917be9-4e82-4ba5-869d-5a9203721699",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-11T23:06:23.977884Z",
+     "iopub.status.busy": "2024-09-11T23:06:23.977517Z"
+    },
+    "id": "b0917be9-4e82-4ba5-869d-5a9203721699"
+   },
+   "source": [
+    "### *Troubleshooting #2*\n",
+    "\n",
+    "In the cell below, we are visualizing the distribution of **Segger similarity scores** using a histogram. The **Segger similarity score** reflects how closely transcripts are associated with their respective nuclei in the segmentation process. **Higher scores** indicate stronger associations between transcripts and their nuclei, suggesting more accurate cell boundaries. **Lower scores** might indicate weaker associations, which could highlight potential segmentation errors or challenging regions in the data. We expect to see a large number of the scores clustering toward higher values, which would indicate strong overall performance of the model in associating transcripts with nuclei.\n",
+    "\n",
+    "The following would indicate potential issues with the model's predictions:\n",
+    "\n",
+    "- **A very large portion of scores near zero**: If many scores are concentrated at the lower end of the scale, this suggests that the model is frequently failing to associate transcripts with their corresponding nuclei, indicating poor segmentation quality.\n",
+    "- **No clear peak in the distribution**: If the histogram is flat or shows a wide, spread-out distribution, this could indicate that the model is struggling to consistently assign similarity scores, which may be a sign that the training process did not optimize the model correctly.\n",
+    "\n",
+    "Both cases would suggest that the model requires further tuning, such as adjusting hyperparameters, data preprocessing, or the training procedure (see below)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a450d3ca-2876-4f48-be89-761147b17387",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-11T22:07:04.216273Z",
+     "iopub.status.busy": "2024-09-11T22:07:04.215965Z",
+     "iopub.status.idle": "2024-09-11T22:07:08.177601Z",
+     "shell.execute_reply": "2024-09-11T22:07:08.177158Z",
+     "shell.execute_reply.started": "2024-09-11T22:07:04.216257Z"
+    },
+    "id": "a450d3ca-2876-4f48-be89-761147b17387",
+    "outputId": "0576e0b8-4823-4701-b661-dc6b513841f2"
+   },
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 600x600 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "fig, ax = plt.subplots(1,1, figsize=(2,2))\n",
+    "sns.histplot(\n",
+    "    segmentation['score'],\n",
+    "    bins=50,\n",
+    "    ax=ax,\n",
+    ")\n",
+    "ax.set_ylabel('Count')\n",
+    "ax.set_xlabel('Segger Similarity Score')\n",
+    "ax.set_yscale('log')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "5492fb96-bf8e-49d5-b40e-7e6b3f871bbe",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-11T22:34:15.990223Z",
+     "iopub.status.busy": "2024-09-11T22:34:15.988880Z"
+    },
+    "id": "5492fb96-bf8e-49d5-b40e-7e6b3f871bbe"
+   },
+   "source": [
+    "#### The Importance of the Receptive Field in Segger\n",
+    "\n",
+    "The **receptive field** is a critical parameter in Segger, as it directly influences how the model interprets the spatial relationships between **transcripts** and **nuclei**. In the context of spatial transcriptomics, the receptive field determines the size of the neighborhood that each node (representing transcripts or nuclei) can \"see\" during graph construction and model training. Segger is particularly sensitive to the size of the receptive field because it affects the model's ability to propagate information across the graph. If the receptive field is too small, the model may fail to capture sufficient context for correct cell boundary delineation. Conversely, a very large receptive field may introduce noise by linking unrelated or distant nodes, reducing segmentation accuracy.\n",
+    "\n",
+    "#### Parameters affecting the receptive field in Segger:\n",
+    "- **`--r`**: This parameter defines the radius used when connecting transcripts to nuclei. A larger `r` expands the receptive field, linking more distant nodes. Fine-tuning this parameter helps ensure that Segger captures the right level of spatial interaction in the dataset.\n",
+    "- **`--k_bd` and `--k_tx`**: These control the number of nearest neighbors (nuclei and transcripts, respectively) considered in the graph. By increasing these values, the receptive field is effectively broadened, allowing more nodes to contribute to the information propagation.\n",
+    "- **`--dist_bd` and `--dist_tx`**: These parameters specify the maximum distances used to connect nuclei (`dist_bd`) and transcripts (`dist_tx`) to their neighbors during graph construction. They directly affect the receptive field by defining the cut-off distance for forming edges in the graph. Larger distance values expand the receptive field, connecting nodes that are further apart spatially. Careful tuning of these values is necessary to ensure that Segger captures relevant spatial relationships without introducing noise."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7ece1ac0-0708-45e2-87fc-1b25782831f8",
+   "metadata": {
+    "id": "7ece1ac0-0708-45e2-87fc-1b25782831f8"
+   },
+   "source": [
+    "# **4. Tune Parameters**"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "896b8288-5287-4d10-a206-e68c0e4731c6",
+   "metadata": {
+    "id": "896b8288-5287-4d10-a206-e68c0e4731c6"
+   },
+   "source": [
+    "### Evaluating Receptive Field Parameters with Grid Search\n",
+    "\n",
+    "To evaluate the impact of different receptive field parameters in Segger, we use a **grid search** approach. The parameters `k_bd`, `k_tx`, `dist_bd`, and `dist_tx` (which control the number of neighbors and distances for nuclei and transcripts) are explored through various configurations defined in `param_space`. Each combination of these parameters is passed to the `trainable` function, which creates the dataset, trains the model, and makes predictions based on the specified receptive field.\n",
+    "\n",
+    "For each parameter combination:\n",
+    "1. A dataset is created with the specified receptive field.\n",
+    "2. The Segger model is trained on this dataset.\n",
+    "3. Predictions are made, and segmentation results are evaluated using the custom `evaluate` function. This function computes metrics like the fraction of assigned transcripts and average cell sizes.\n",
+    "\n",
+    "The results from each configuration are saved, allowing us to compare how different receptive field settings impact the model’s performance. This process enables a thorough search of the parameter space, optimizing the model for accurate segmentation."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b0c1a7a8-acb2-4aae-8ae4-8aa9a4196717",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-12T01:10:47.781418Z",
+     "iopub.status.busy": "2024-09-12T01:10:47.781067Z",
+     "iopub.status.idle": "2024-09-12T01:10:48.706615Z",
+     "shell.execute_reply": "2024-09-12T01:10:48.706194Z",
+     "shell.execute_reply.started": "2024-09-12T01:10:47.781401Z"
+    },
+    "id": "b0c1a7a8-acb2-4aae-8ae4-8aa9a4196717"
+   },
+   "outputs": [],
+   "source": [
+    "import itertools\n",
+    "import pandas as pd"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0bd0803c-e58d-4f43-9627-d2c1ab187d5e",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-12T01:16:31.976312Z",
+     "iopub.status.busy": "2024-09-12T01:16:31.975947Z",
+     "iopub.status.idle": "2024-09-12T01:16:33.168389Z",
+     "shell.execute_reply": "2024-09-12T01:16:33.167956Z",
+     "shell.execute_reply.started": "2024-09-12T01:16:31.976295Z"
+    },
+    "id": "0bd0803c-e58d-4f43-9627-d2c1ab187d5e"
+   },
+   "outputs": [],
+   "source": [
+    "tuning_dir = Path('path/to/tutorial/tuning/')\n",
+    "sampling_rate = 0.125"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b879a0b5-150c-4240-99ec-81075855aa52",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-12T01:16:33.169525Z",
+     "iopub.status.busy": "2024-09-12T01:16:33.169189Z",
+     "iopub.status.idle": "2024-09-12T01:16:34.147222Z",
+     "shell.execute_reply": "2024-09-12T01:16:34.146804Z",
+     "shell.execute_reply.started": "2024-09-12T01:16:33.169508Z"
+    },
+    "id": "b879a0b5-150c-4240-99ec-81075855aa52",
+    "jupyter": {
+     "source_hidden": true
+    }
+   },
+   "outputs": [],
+   "source": [
+    "# Fixed function arguments used for each trial\n",
+    "transcripts_path = xenium_data_dir / 'transcripts.parquet'\n",
+    "\n",
+    "boundaries_path = xenium_data_dir / 'nucleus_boundaries.parquet'\n",
+    "\n",
+    "dataset_kwargs = dict(\n",
+    "    x_size=80, y_size=80, d_x=80, d_y=80, margin_x=10, margin_y=10,\n",
+    "    num_workers=4, sampling_rate=sampling_rate,\n",
+    ")\n",
+    "\n",
+    "model_kwargs = dict(\n",
+    "    metadata=(['tx', 'bd'], [('tx', 'belongs', 'bd'), ('tx', 'neighbors', 'tx')]),\n",
+    "    num_tx_tokens=500, init_emb=8, hidden_channels=32, out_channels=8,\n",
+    "    heads=2, num_mid_layers=2, aggr='sum',\n",
+    ")\n",
+    "\n",
+    "trainer_kwargs = dict(\n",
+    "    accelerator='cuda', strategy='auto', precision='16-mixed', devices=1,\n",
+    "    max_epochs=100,\n",
+    ")\n",
+    "\n",
+    "predict_kwargs = dict(score_cut=0.2, use_cc=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "fbd831c9-3a50-4e3b-97d3-3c152ae01188",
+   "metadata": {
+    "id": "fbd831c9-3a50-4e3b-97d3-3c152ae01188",
+    "jupyter": {
+     "source_hidden": true
+    }
+   },
+   "outputs": [],
+   "source": [
+    "def trainable(config):\n",
+    "\n",
+    "    receptive_field = {k: config[k] for k in ['k_bd', 'k_tx', 'dist_bd', 'dist_tx']}\n",
+    "\n",
+    "    # Dataset creation\n",
+    "    xs = XeniumSample(verbose=False)\n",
+    "    xs.set_file_paths(transcripts_path, boundaries_path)\n",
+    "    xs.set_metadata()\n",
+    "    try:\n",
+    "        xs.save_dataset_for_segger(\n",
+    "            processed_dir=config['data_dir'],\n",
+    "            receptive_field=receptive_field,\n",
+    "            **dataset_kwargs,\n",
+    "        )\n",
+    "    except:\n",
+    "        pass\n",
+    "\n",
+    "    # Model training\n",
+    "    ls = LitSegger(**model_kwargs)\n",
+    "    dm = SeggerDataModule(\n",
+    "        data_dir=config['data_dir'],\n",
+    "        batch_size=2,\n",
+    "        num_workers=dataset_kwargs['num_workers'],\n",
+    "    )\n",
+    "    trainer = Trainer(\n",
+    "        default_root_dir=config['model_dir'],\n",
+    "        logger=CSVLogger(config['model_dir']),\n",
+    "        **trainer_kwargs,\n",
+    "    )\n",
+    "    trainer.fit(model=ls, datamodule=dm)\n",
+    "\n",
+    "    segmentation = predict(\n",
+    "        load_model(config['model_dir']/'lightning_logs/version_0/checkpoints'),\n",
+    "        dm.train_dataloader(),\n",
+    "        receptive_field=receptive_field,\n",
+    "        **predict_kwargs,\n",
+    "    )\n",
+    "\n",
+    "    metrics = evaluate(segmentation)\n",
+    "\n",
+    "\n",
+    "def evaluate(segmentation: pd.DataFrame, score_cut: float) -> pd.Series:\n",
+    "\n",
+    "    assigned = segmentation['score'] > score_cut\n",
+    "    metrics = pd.Series(dtype=float)\n",
+    "    metrics['frac_assigned'] = assigned.mean()\n",
+    "    cell_sizes = segmentation.groupby(assigned)['segger_cell_id'].value_counts()\n",
+    "    assigned_avg = 0 if True not in cell_sizes.index else cell_sizes[True].mean()\n",
+    "    cc_avg = 0 if False not in cell_sizes.index else cell_sizes[False].mean()\n",
+    "    metrics['cell_size_assigned'] = assigned_avg\n",
+    "    metrics['cell_size_cc'] = cc_avg\n",
+    "    return metrics"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ba2dcc9a-3a06-4b84-a487-59a768eed5d5",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2024-09-12T01:16:35.184598Z",
+     "iopub.status.busy": "2024-09-12T01:16:35.184180Z",
+     "iopub.status.idle": "2024-09-12T01:19:55.171470Z",
+     "shell.execute_reply": "2024-09-12T01:19:55.170810Z",
+     "shell.execute_reply.started": "2024-09-12T01:16:35.184582Z"
+    },
+    "id": "ba2dcc9a-3a06-4b84-a487-59a768eed5d5",
+    "scrolled": true
+   },
+   "outputs": [],
+   "source": [
+    "param_space = {\n",
+    "    \"k_bd\": [3, 5, 10],\n",
+    "    \"dist_bd\": [5, 10, 15, 20],\n",
+    "    \"k_tx\": [3, 5, 10],\n",
+    "    \"dist_tx\": [3, 5, 10],\n",
+    "}\n",
+    "\n",
+    "metrics = []\n",
+    "\n",
+    "for params in itertools.product(*param_space.values()):\n",
+    "\n",
+    "    config = dict(zip(param_space.keys(), params))\n",
+    "\n",
+    "    # Setup directories\n",
+    "    trial_dir = tuning_dir / '_'.join([f'{k}={v}' for k, v in config.items()])\n",
+    "\n",
+    "    data_dir = trial_dir / 'segger_data'\n",
+    "    data_dir.mkdir(exist_ok=True, parents=True)\n",
+    "    config['data_dir'] = data_dir\n",
+    "\n",
+    "    model_dir = trial_dir / 'models'\n",
+    "    model_dir.mkdir(exist_ok=True, parents=True)\n",
+    "    config['model_dir'] = model_dir\n",
+    "\n",
+    "    segmentation = trainable(config)\n",
+    "    trial = evaluate(segmentation, predict_kwargs['score_cut'])\n",
+    "    trial = pd.concat([pd.Series(config), trial])\n",
+    "    metrics.append(trial)\n",
+    "\n",
+    "metrics = pd.DataFrame(metrics)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "dcfa5570-ada2-4102-aae0-a3830d304c5f",
+   "metadata": {
+    "id": "dcfa5570-ada2-4102-aae0-a3830d304c5f"
+   },
+   "source": [
+    "### Interpreting Output Metrics\n",
+    "\n",
+    "The key output metrics include:\n",
+    "- **`frac_assigned`**: The fraction of transcripts that were successfully assigned to a cell. A higher value indicates that the model is doing a good job associating transcripts with nuclei, which is a strong indicator of successful segmentation.\n",
+    "- **`cell_size_assigned`**: The average size of cells that have assigned transcripts. This helps assess how well the model is predicting cell boundaries, with unusually large or small values indicating potential issues with segmentation accuracy.\n",
+    "- **`cell_size_cc`**: The average size of connected components that were not assigned to a cell (i.e., nucleus-less regions). Large values here may suggest that transcripts are being incorrectly grouped together in the absence of a nucleus, which could indicate problems with the receptive field parameters or the segmentation process.\n",
+    "\n",
+    "These metrics illuminate the effectiveness of the model by highlighting both the success in associating transcripts with cells and potential areas where the model may need further tuning.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1a89aed4-c53b-460f-8a6f-f690920b6829",
+   "metadata": {
+    "execution": {
+     "iopub.status.busy": "2024-09-12T01:19:55.171961Z",
+     "iopub.status.idle": "2024-09-12T01:19:55.172161Z",
+     "shell.execute_reply": "2024-09-12T01:19:55.172071Z",
+     "shell.execute_reply.started": "2024-09-12T01:19:55.172062Z"
+    },
+    "id": "1a89aed4-c53b-460f-8a6f-f690920b6829"
+   },
+   "outputs": [],
+   "source": [
+    "metrics"
+   ]
+  }
+ ],
+ "metadata": {
+  "accelerator": "GPU",
+  "colab": {
+   "gpuType": "T4",
+   "provenance": []
+  },
+  "kernelspec": {
+   "display_name": "Python 3",
+   "name": "python3"
   },
-  "nbformat": 4,
-  "nbformat_minor": 5
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.9"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
 }

From 7e6b4d8b0f26cba4c54b6139718aaabb7a784c70 Mon Sep 17 00:00:00 2001
From: "pre-commit-ci[bot]"
 <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed, 8 Jan 2025 00:19:03 +0000
Subject: [PATCH 2/2] [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci
---
 README.md                                |   6 +-
 scripts/create_data_fast_sample.py       |  74 +++++-------
 scripts/predict_model_sample.py          |   6 +-
 scripts/sand.py                          | 142 +++++++++++------------
 scripts/train_model_sample.py            |   2 +-
 src/segger/data/Untitled-4.py            |  34 +++---
 src/segger/data/parquet/_utils.py        |   1 +
 src/segger/prediction/predict_parquet.py |   6 +-
 8 files changed, 121 insertions(+), 150 deletions(-)

diff --git a/README.md b/README.md
index 13d7f3c..4fb0e83 100644
--- a/README.md
+++ b/README.md
@@ -2,11 +2,8 @@
 
 [![pre-commit.ci status](https://results.pre-commit.ci/badge/github/EliHei2/segger_dev/main.svg)](https://results.pre-commit.ci/latest/github/EliHei2/segger_dev/main)
 
-
-
 **Important note (Dec 2024)**: As segger is currently undergoing constant development we highly recommending installing segger directly via github.
 
-
 **segger** is a cutting-edge tool for **cell segmentation** in **single-molecule spatial omics** datasets. By leveraging **graph neural networks (GNNs)** and heterogeneous graphs, segger offers unmatched accuracy and scalability.
 
 # How segger Works
@@ -52,7 +49,7 @@ segger tackles these with a **graph-based approach**, achieving superior segment
 
 ---
 
-## Installation 
+## Installation
 
 **Important note (Dec 2024)**: As segger is currently undergoing constant development we highly recommending installing segger directly via github.
 
@@ -78,7 +75,6 @@ pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -
 
 Afterwards choose the installation method that best suits your needs.
 
-
 ### GitHub Installation
 
 For a straightforward local installation from GitHub, clone the repository and install the package using `pip`:
diff --git a/scripts/create_data_fast_sample.py b/scripts/create_data_fast_sample.py
index 54b8845..573dd1e 100644
--- a/scripts/create_data_fast_sample.py
+++ b/scripts/create_data_fast_sample.py
@@ -7,87 +7,77 @@
 import numpy as np
 from segger.data.parquet._utils import get_polygons_from_xy
 
-xenium_data_dir = Path('data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/')
-segger_data_dir = Path('data_tidy/pyg_datasets/bc_rep1_emb')
+xenium_data_dir = Path("data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/")
+segger_data_dir = Path("data_tidy/pyg_datasets/bc_rep1_emb")
 
 
-scrnaseq_file = Path('/omics/groups/OE0606/internal/tangy/tasks/schier/data/atals_filtered.h5ad')
-celltype_column = 'celltype_major'
-gene_celltype_abundance_embedding = calculate_gene_celltype_abundance_embedding(
-    sc.read(scrnaseq_file),
-    celltype_column
-)
+scrnaseq_file = Path("/omics/groups/OE0606/internal/tangy/tasks/schier/data/atals_filtered.h5ad")
+celltype_column = "celltype_major"
+gene_celltype_abundance_embedding = calculate_gene_celltype_abundance_embedding(sc.read(scrnaseq_file), celltype_column)
 
 sample = STSampleParquet(
     base_dir=xenium_data_dir,
     n_workers=4,
-    sample_type='xenium',
-    weights=gene_celltype_abundance_embedding, # uncomment if gene-celltype embeddings are available
+    sample_type="xenium",
+    weights=gene_celltype_abundance_embedding,  # uncomment if gene-celltype embeddings are available
 )
 
-transcripts = pd.read_parquet(
-    xenium_data_dir / 'transcripts.parquet',
-    filters=[[('overlaps_nucleus', '=', 1)]]
-)
-boundaries = pd.read_parquet(xenium_data_dir / 'nucleus_boundaries.parquet')
+transcripts = pd.read_parquet(xenium_data_dir / "transcripts.parquet", filters=[[("overlaps_nucleus", "=", 1)]])
+boundaries = pd.read_parquet(xenium_data_dir / "nucleus_boundaries.parquet")
 
-sizes = transcripts.groupby('cell_id').size()
-polygons = get_polygons_from_xy(boundaries, 'vertex_x', 'vertex_y', 'cell_id')
+sizes = transcripts.groupby("cell_id").size()
+polygons = get_polygons_from_xy(boundaries, "vertex_x", "vertex_y", "cell_id")
 densities = polygons[sizes.index].area / sizes
 bd_width = polygons.minimum_bounding_radius().median() * 2
 
 # 1/4 median boundary diameter
 dist_tx = bd_width / 4
 # 90th percentile density of bounding circle with radius=dist_tx
-k_tx = math.ceil(np.quantile(dist_tx ** 2 * np.pi * densities, 0.9)) 
+k_tx = math.ceil(np.quantile(dist_tx**2 * np.pi * densities, 0.9))
 
 print(k_tx)
 print(dist_tx)
 
 
 sample.save(
-      data_dir=segger_data_dir,
-      k_bd=3,
-      dist_bd=15.0,
-      k_tx=dist_tx,
-      dist_tx=k_tx,
-      tile_width=120,
-      tile_height=120,
-      neg_sampling_ratio=5.0,
-      frac=1.0,
-      val_prob=0.1,
-      test_prob=0.1,
+    data_dir=segger_data_dir,
+    k_bd=3,
+    dist_bd=15.0,
+    k_tx=dist_tx,
+    dist_tx=k_tx,
+    tile_width=120,
+    tile_height=120,
+    neg_sampling_ratio=5.0,
+    frac=1.0,
+    val_prob=0.1,
+    test_prob=0.1,
 )
 
 
-xenium_data_dir = Path('data_tidy/bc_5k')
-segger_data_dir = Path('data_tidy/pyg_datasets/bc_5k_emb')
-
+xenium_data_dir = Path("data_tidy/bc_5k")
+segger_data_dir = Path("data_tidy/pyg_datasets/bc_5k_emb")
 
 
 sample = STSampleParquet(
     base_dir=xenium_data_dir,
     n_workers=1,
-    sample_type='xenium',
-    weights=gene_celltype_abundance_embedding, # uncomment if gene-celltype embeddings are available
+    sample_type="xenium",
+    weights=gene_celltype_abundance_embedding,  # uncomment if gene-celltype embeddings are available
 )
 
 
-transcripts = pd.read_parquet(
-    xenium_data_dir / 'transcripts.parquet',
-    filters=[[('overlaps_nucleus', '=', 1)]]
-)
-boundaries = pd.read_parquet(xenium_data_dir / 'nucleus_boundaries.parquet')
+transcripts = pd.read_parquet(xenium_data_dir / "transcripts.parquet", filters=[[("overlaps_nucleus", "=", 1)]])
+boundaries = pd.read_parquet(xenium_data_dir / "nucleus_boundaries.parquet")
 
-sizes = transcripts.groupby('cell_id').size()
-polygons = get_polygons_from_xy(boundaries, 'vertex_x', 'vertex_y', 'cell_id')
+sizes = transcripts.groupby("cell_id").size()
+polygons = get_polygons_from_xy(boundaries, "vertex_x", "vertex_y", "cell_id")
 densities = polygons[sizes.index].area / sizes
 bd_width = polygons.minimum_bounding_radius().median() * 2
 
 # 1/4 median boundary diameter
 dist_tx = bd_width / 4
 # 90th percentile density of bounding circle with radius=dist_tx
-k_tx = math.ceil(np.quantile(dist_tx ** 2 * np.pi * densities, 0.9)) 
+k_tx = math.ceil(np.quantile(dist_tx**2 * np.pi * densities, 0.9))
 
 print(k_tx)
 print(dist_tx)
diff --git a/scripts/predict_model_sample.py b/scripts/predict_model_sample.py
index 9624abc..c1bc9c0 100644
--- a/scripts/predict_model_sample.py
+++ b/scripts/predict_model_sample.py
@@ -16,12 +16,11 @@
 import dask.dataframe as dd
 
 
-
 seg_tag = "bc_fast_data_emb_major"
 model_version = 1
 
-segger_data_dir = Path('data_tidy/pyg_datasets') / seg_tag
-models_dir = Path("./models") / seg_tag 
+segger_data_dir = Path("data_tidy/pyg_datasets") / seg_tag
+models_dir = Path("./models") / seg_tag
 benchmarks_dir = Path("/dkfz/cluster/gpu/data/OE0606/elihei/segger_experiments/data_tidy/benchmarks/xe_rep1_bc")
 transcripts_file = "data_raw/xenium/Xenium_FFPE_Human_Breast_Cancer_Rep1/transcripts.parquet"
 # Initialize the Lightning data module
@@ -58,4 +57,3 @@
     gpu_ids=["0"],
     # client=client
 )
-
diff --git a/scripts/sand.py b/scripts/sand.py
index 447c5e6..ab480c0 100644
--- a/scripts/sand.py
+++ b/scripts/sand.py
@@ -32,7 +32,7 @@ def process_group(group, area_low, area_high):
     cell_boundary = generate_boundary(seg_cell)
     if isinstance(cell_boundary, MultiPolygon):
         # polygons = sorted(cell_boundary, key=lambda p: p.area, reverse=True)
-            # Extract the largest polygon (by area) from MultiPolygon
+        # Extract the largest polygon (by area) from MultiPolygon
         # cell_boundary = polygons[0]
         # cell_boundary = unary_union(cell_boundary)
         # polygons = sorted(cell_boundary, key=lambda p: p.area, reverse=True)
@@ -72,7 +72,6 @@ def process_group(group, area_low, area_high):
         },
         "seg_mask_value": uint_cell_id,
     }
-    
 
 
 def get_coordinates(boundary):
@@ -86,8 +85,8 @@ def get_coordinates(boundary):
     elif isinstance(boundary, Polygon):
         # Return coordinates from a single Polygon
         return list(boundary.exterior.coords)
-    
-    
+
+
 def get_flatten_version(polygon_vertices, max_value=21):
     """
     Flattens and standardizes the shape of polygon vertices to a fixed length.
@@ -108,8 +107,11 @@ def get_flatten_version(polygon_vertices, max_value=21):
             flattened.append(vertices + [(0.0, 0.0)] * (max_value - len(vertices)))
     return np.array(flattened, dtype=np.float32)
 
+
 from segger.validation.xenium_explorer import *
 from segger.prediction.boundary import *
+
+
 def seg2explorer(
     seg_df: pd.DataFrame,
     source_path: str,
@@ -155,18 +157,13 @@ def seg2explorer(
     cell_id = [res["uint_cell_id"] for res in results]
     cell_summary = [res["cell_summary"] for res in results]
     # print('********************************1')
-    polygon_num_vertices = [
-        len(res["cell_boundary"].exterior.coords) for res in results
-    ]
+    polygon_num_vertices = [len(res["cell_boundary"].exterior.coords) for res in results]
     print(polygon_num_vertices)
     # polygon_vertices = np.array(
     #     [list(res["cell_boundary"].exterior.coords) for res in results],
     #     dtype=object
     # )
-    polygon_vertices = np.array(
-        [list(res["cell_boundary"].exterior.coords) for res in results],
-        dtype=object
-    )
+    polygon_vertices = np.array([list(res["cell_boundary"].exterior.coords) for res in results], dtype=object)
     # polygon_vertices = get_flatten_version(
     #     [get_coordinates(res["cell_boundary"]) for res in results],
     #     max_value=21,
@@ -182,7 +179,7 @@ def seg2explorer(
         "polygon_vertices": np.array(polygon_vertices).astype(np.float32),
         "seg_mask_value": np.array(seg_mask_value, dtype=np.int32),
     }
-    print(len(cells['cell_id']))
+    print(len(cells["cell_id"]))
     # Save cells data
     existing_store = zarr.open(source_path / "cells.zarr.zip", mode="r")
     new_store = zarr.open(storage / f"{cells_filename}.zarr.zip", mode="w")
@@ -194,7 +191,7 @@ def seg2explorer(
     new_store.attrs["number_cells"] = len(cells["cell_id"])
     print(new_store)
     new_store.store.close()
-    
+
     print(cells["polygon_vertices"])
     # # Save analysis data
     if analysis_df is None:
@@ -242,8 +239,6 @@ def seg2explorer(
     )
 
 
-
-
 def seg2explorer(
     seg_df: pd.DataFrame,
     source_path: str,
@@ -274,6 +269,7 @@ def seg2explorer(
     """
     import zarr
     import json
+
     source_path = Path(source_path)
     storage = Path(output_dir)
     cell_id2old_id = {}
@@ -297,9 +293,9 @@ def seg2explorer(
         cell_convex_hull = get_boundary(seg_cell)
         # print(cell_convex_hull)
         if isinstance(cell_convex_hull, MultiPolygon):
-        # polygons = sorted(cell_boundary, key=lambda p: p.area, reverse=True)
+            # polygons = sorted(cell_boundary, key=lambda p: p.area, reverse=True)
             # Extract the largest polygon (by area) from MultiPolygon
-        # cell_boundary = polygons[0]
+            # cell_boundary = polygons[0]
             # cell_convex_hull = unary_union(cell_convex_hull)
             # print('****************1')
             # polygons = sorted(cell_convex_hull.geoms, key=lambda p: p.area, reverse=True)
@@ -404,100 +400,96 @@ def seg2explorer(
         cells_name=cells_filename,
         analysis_name=analysis_filename,
     )
-    
-    
 
 
-ddf = dd.read_parquet('/dkfz/cluster/gpu/data/OE0606/elihei/segger_experiments/data_tidy/benchmarks/xe_rep1_bc/parquet_train_big_0.5_False_3_10_5_3_20241030/segger_transcripts.parquet').compute()
+ddf = dd.read_parquet(
+    "/dkfz/cluster/gpu/data/OE0606/elihei/segger_experiments/data_tidy/benchmarks/xe_rep1_bc/parquet_train_big_0.5_False_3_10_5_3_20241030/segger_transcripts.parquet"
+).compute()
 ddf = ddf.dropna()
-ddf = ddf[ddf.segger_cell_id != 'None']
-ddf = ddf.sort_values('segger_cell_id')
+ddf = ddf[ddf.segger_cell_id != "None"]
+ddf = ddf.sort_values("segger_cell_id")
 df = ddf.iloc[:10000, :]
 
 
-
-df_path = Path('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep1_v9_segger.csv.gz')
+df_path = Path("data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep1_v9_segger.csv.gz")
 df_v9 = dd.read_csv(df_path)
-df_main = dd.read_parquet('data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/transcripts.parquet')
+df_main = dd.read_parquet("data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/transcripts.parquet")
 
-ddf = df_v9.merge(df_main, on='transcript_id')
+ddf = df_v9.merge(df_main, on="transcript_id")
 ddf = ddf.compute()
-ddf = ddf[ddf.segger_cell_id != 'None']
-ddf = ddf.sort_values('segger_cell_id')
-df = ddf.loc[(ddf.x_location > 250) & (ddf.x_location < 1500) & (ddf.y_location > 500) & (ddf.y_location < 1500),:]
+ddf = ddf[ddf.segger_cell_id != "None"]
+ddf = ddf.sort_values("segger_cell_id")
+df = ddf.loc[(ddf.x_location > 250) & (ddf.x_location < 1500) & (ddf.y_location > 500) & (ddf.y_location < 1500), :]
 
 # tx_df = dd.read_csv('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep2_v9_segger.csv.gz')
 # ddf = tx_df.merge(df_main, on='transcript_id')
 
 seg2explorer(
-    seg_df = df,
-    source_path = 'data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs',
-    output_dir = 'data_tidy/explorer/rep1sis',
-    cells_filename = "segger_cells_seg_roi1",
-    analysis_filename = "segger_analysis_seg_roi1",
-    xenium_filename = "segger_experiment_seg_roi1.xenium",
-    analysis_df = None,
-    cell_id_columns = "segger_cell_id",
-    area_low = 10,
-    area_high= 1000,
+    seg_df=df,
+    source_path="data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs",
+    output_dir="data_tidy/explorer/rep1sis",
+    cells_filename="segger_cells_seg_roi1",
+    analysis_filename="segger_analysis_seg_roi1",
+    xenium_filename="segger_experiment_seg_roi1.xenium",
+    analysis_df=None,
+    cell_id_columns="segger_cell_id",
+    area_low=10,
+    area_high=1000,
 )
 
 
-
-df = ddf.loc[(ddf.x_location > 1550) & (ddf.x_location < 3250) & (ddf.y_location > 2250) & (ddf.y_location < 3550),:]
+df = ddf.loc[(ddf.x_location > 1550) & (ddf.x_location < 3250) & (ddf.y_location > 2250) & (ddf.y_location < 3550), :]
 
 # tx_df = dd.read_csv('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep2_v9_segger.csv.gz')
 # ddf = tx_df.merge(df_main, on='transcript_id')
 
 seg2explorer(
-    seg_df = df,
-    source_path = 'data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs',
-    output_dir = 'data_tidy/explorer/rep1sis',
-    cells_filename = "segger_cells_seg_roi2",
-    analysis_filename = "segger_analysis_seg_roi2",
-    xenium_filename = "segger_experiment_seg_roi2.xenium",
-    analysis_df = None,
-    cell_id_columns = "segger_cell_id",
-    area_low = 10,
-    area_high= 1000,
+    seg_df=df,
+    source_path="data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs",
+    output_dir="data_tidy/explorer/rep1sis",
+    cells_filename="segger_cells_seg_roi2",
+    analysis_filename="segger_analysis_seg_roi2",
+    xenium_filename="segger_experiment_seg_roi2.xenium",
+    analysis_df=None,
+    cell_id_columns="segger_cell_id",
+    area_low=10,
+    area_high=1000,
 )
 
 
-
-df = ddf.loc[(ddf.x_location > 4000) & (ddf.x_location < 4500) & (ddf.y_location > 1000) & (ddf.y_location < 1500),:]
+df = ddf.loc[(ddf.x_location > 4000) & (ddf.x_location < 4500) & (ddf.y_location > 1000) & (ddf.y_location < 1500), :]
 
 # tx_df = dd.read_csv('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep2_v9_segger.csv.gz')
 # ddf = tx_df.merge(df_main, on='transcript_id')
 
 seg2explorer(
-    seg_df = df,
-    source_path = 'data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs',
-    output_dir = 'data_tidy/explorer/rep1sis',
-    cells_filename = "segger_cells_seg_roi3",
-    analysis_filename = "segger_analysis_seg_roi3",
-    xenium_filename = "segger_experiment_seg_roi3.xenium",
-    analysis_df = None,
-    cell_id_columns = "segger_cell_id",
-    area_low = 10,
-    area_high= 1000,
+    seg_df=df,
+    source_path="data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs",
+    output_dir="data_tidy/explorer/rep1sis",
+    cells_filename="segger_cells_seg_roi3",
+    analysis_filename="segger_analysis_seg_roi3",
+    xenium_filename="segger_experiment_seg_roi3.xenium",
+    analysis_df=None,
+    cell_id_columns="segger_cell_id",
+    area_low=10,
+    area_high=1000,
 )
 
 
-df = ddf.loc[(ddf.x_location > 1550) & (ddf.x_location < 3250) & (ddf.y_location > 2250) & (ddf.y_location < 3550),:]
+df = ddf.loc[(ddf.x_location > 1550) & (ddf.x_location < 3250) & (ddf.y_location > 2250) & (ddf.y_location < 3550), :]
 
 # tx_df = dd.read_csv('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep2_v9_segger.csv.gz')
 # ddf = tx_df.merge(df_main, on='transcript_id')
 
 seg2explorer(
-    seg_df = df,
-    source_path = 'data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs',
-    output_dir = 'data_tidy/explorer/rep1sis',
-    cells_filename = "segger_cells_seg_roi2",
-    analysis_filename = "segger_analysis_seg_roi2",
-    xenium_filename = "segger_experiment_seg_roi2.xenium",
-    analysis_df = None,
-    cell_id_columns = "segger_cell_id",
-    area_low = 10,
-    area_high= 1000,
+    seg_df=df,
+    source_path="data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs",
+    output_dir="data_tidy/explorer/rep1sis",
+    cells_filename="segger_cells_seg_roi2",
+    analysis_filename="segger_analysis_seg_roi2",
+    xenium_filename="segger_experiment_seg_roi2.xenium",
+    analysis_df=None,
+    cell_id_columns="segger_cell_id",
+    area_low=10,
+    area_high=1000,
 )
-
diff --git a/scripts/train_model_sample.py b/scripts/train_model_sample.py
index 411cee8..826989f 100644
--- a/scripts/train_model_sample.py
+++ b/scripts/train_model_sample.py
@@ -15,7 +15,7 @@
 import os
 
 
-segger_data_dir = segger_data_dir = Path('data_tidy/pyg_datasets/bc_fast_data_emb_minor')
+segger_data_dir = segger_data_dir = Path("data_tidy/pyg_datasets/bc_fast_data_emb_minor")
 models_dir = Path("./models/bc_fast_data_emb_minor")
 
 dm = SeggerDataModule(
diff --git a/src/segger/data/Untitled-4.py b/src/segger/data/Untitled-4.py
index 51b3034..dea3858 100644
--- a/src/segger/data/Untitled-4.py
+++ b/src/segger/data/Untitled-4.py
@@ -1,38 +1,34 @@
-df_path = Path('/omics/groups/OE0606/internal/gleb/xenium/elyas/rep1/baysor_rep1.csv')
+df_path = Path("/omics/groups/OE0606/internal/gleb/xenium/elyas/rep1/baysor_rep1.csv")
 df = dd.read_csv(df_path)
 
-res = compute_cells_per_nucleus(df, new_cell_col='baysor_cell_id')
+res = compute_cells_per_nucleus(df, new_cell_col="baysor_cell_id")
 
 
 plot_gradient_nuclei_histogram(res, figures_path)
 
 
-
-
-
-df_path = Path('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep1_v9_segger.csv.gz')
+df_path = Path("data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep1_v9_segger.csv.gz")
 df_v9 = dd.read_csv(df_path)
-df_main = dd.read_parquet('data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/transcripts.parquet')
+df_main = dd.read_parquet("data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/transcripts.parquet")
 
-ddf = df_v9.merge(df_main, on='transcript_id')
+ddf = df_v9.merge(df_main, on="transcript_id")
 # df = dd.read_parquet(segger_emb_path/'segger_transcripts.parquet')
 ddf1 = ddf.compute()
-res = compute_cells_per_nucleus(df.compute(), new_cell_col='segger_cell_id')
-plot_gradient_nuclei_histogram(res, figures_path, 'segger')
+res = compute_cells_per_nucleus(df.compute(), new_cell_col="segger_cell_id")
+plot_gradient_nuclei_histogram(res, figures_path, "segger")
 
 dfff = df.compute()
 dfff = dfff.dropna()
 gene_dropout = calculate_gene_dropout_rate(dfff)
 
 
-plot_z_distance_distribution(dfff, figures_path, title='Z-Distance Distribution')
+plot_z_distance_distribution(dfff, figures_path, title="Z-Distance Distribution")
 
-plot_z_distance_boxplot(dfff, figures_path, title='Z-Distance Distribution')
+plot_z_distance_boxplot(dfff, figures_path, title="Z-Distance Distribution")
 
 
-tx_df = dd.read_csv('data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep2_v9_segger.csv.gz')
-ddf = tx_df.merge(df_main, on='transcript_id')
-
+tx_df = dd.read_csv("data_tidy/Xenium_FFPE_Human_Breast_Cancer_Rep2_v9_segger.csv.gz")
+ddf = tx_df.merge(df_main, on="transcript_id")
 
 
 from segger.prediction.boundary import generate_boundary
@@ -44,14 +40,12 @@
 
 from pqdm.processes import pqdm  # or use pqdm.threads for threading-based parallelism
 
+
 # Modify the function to work with a single group to use with pqdm
 def process_group(group):
     cell_id, t = group
-    return {
-        "cell_id": cell_id,
-        "length": len(t),
-        "geom": generate_boundary(t, x="x_location", y="y_location")
-    }
+    return {"cell_id": cell_id, "length": len(t), "geom": generate_boundary(t, x="x_location", y="y_location")}
+
 
 def generate_boundaries(df, x="x_location", y="y_location", cell_id="segger_cell_id", n_jobs=10):
     # Group by cell_id
diff --git a/src/segger/data/parquet/_utils.py b/src/segger/data/parquet/_utils.py
index 9010443..92791a8 100644
--- a/src/segger/data/parquet/_utils.py
+++ b/src/segger/data/parquet/_utils.py
@@ -55,6 +55,7 @@ def get_xy_extents(
     # If statistics are not available, compute them manually from the data
     except:
         import gc
+
         print("metadata lacks the statistics of the tile's bounding box, computing might take longer!")
         parquet_file = pd.read_parquet(filepath)
         x_col = parquet_file.loc[:, x]
diff --git a/src/segger/prediction/predict_parquet.py b/src/segger/prediction/predict_parquet.py
index e4d499b..43dacf2 100644
--- a/src/segger/prediction/predict_parquet.py
+++ b/src/segger/prediction/predict_parquet.py
@@ -221,7 +221,7 @@ def get_similarity_scores(
             coords_1 = coords_2 = batch[to_type].pos
         else:
             coords_1 = batch[to_type].pos[:, :2]  # 'tx' positions
-            coords_2 = batch[from_type].pos[:, :2] 
+            coords_2 = batch[from_type].pos[:, :2]
         if knn_method == "kd_tree":
             # Compute edge indices using knn method (still on GPU)
             edge_index = get_edge_index(
@@ -244,9 +244,9 @@ def get_similarity_scores(
         edge_index = coo_to_dense_adj(edge_index.T, num_nodes=shape[0], num_nbrs=receptive_field[f"k_{to_type}"])
 
         with torch.no_grad():
-            if from_type != to_type: 
+            if from_type != to_type:
                 embeddings = model(batch.x_dict, batch.edge_index_dict)
-            else: # to go with the inital embeddings for tx-tx
+            else:  # to go with the inital embeddings for tx-tx
                 embeddings = {key: model.node_init[key](x) for key, x in batch.x_dict.items()}
                 norms = embeddings[to_type].norm(dim=1, keepdim=True)
                 # Avoid division by zero in case there are zero vectors