From d44a543e6314e7b6b1dc9fdca6713b76c39e8252 Mon Sep 17 00:00:00 2001
From: merveenoyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 16:46:40 +0200
Subject: [PATCH 01/28] initial commit

---
 ocr-open-models.md            | 348 ++++++++++++++++++++++++++++++++++
 ocr-open-models/thumbnail.png | Bin 0 -> 77228 bytes
 2 files changed, 348 insertions(+)
 create mode 100644 ocr-open-models.md
 create mode 100644 ocr-open-models/thumbnail.png

diff --git a/ocr-open-models.md b/ocr-open-models.md
new file mode 100644
index 0000000000..7634a799f0
--- /dev/null
+++ b/ocr-open-models.md
@@ -0,0 +1,348 @@
+---
+title: "Supercharge your OCR Pipelines with Open Models"
+thumbnail: /blog/assets/ocr-open-models/thumbnail.png
+authors:
+- user: merve
+- user: ariG23498
+- user: davanstrien
+- user: hynky
+- user: andito
+- user: reach-vb
+- user: pcuenq
+---
+
+# Supercharge your OCR Pipelines with Open Models
+
+TL;DR: The rise of powerful vision-language models has transformed document AI. Each model comes with unique strengths, making it tricky to choose the right one. Open-weight models offer better cost efficiency and privacy. To help you get started with them, we’ve put together this guide.
+
+You’ll learn:
+
+* The landscape of current models and their capabilities  
+* When to fine-tune models vs. use models out-of-the-box  
+* Key factors to consider when selecting a model for your use case  
+* How to move beyond OCR with multimodal retrieval and document QA
+
+By the end, you’ll know how to choose the right OCR model, start building with it, and gain deeper insights into document AI. Let’s go\!
+
+# Table-of-Contents 
+
+- [Supercharge your OCR Pipelines with Open Models](#supercharge-your-ocr-pipelines-with-open-models)
+  - [Brief Introduction to Modern OCR](#brief-introduction-to-modern-ocr)
+    - [Model Capabilities](#model-capabilities)
+      - [Transcription](#transcription)
+      - [Handling complex components in documents](#handling-complex-components-in-documents)
+      - [Output formats](#output-formats)
+      - [Locality Awareness in OCR](#locality-awareness-in-ocr)
+      - [Model Prompting](#model-prompting)
+  - [Cutting-edge Open OCR Models](#cutting-edge-open-ocr-models)
+    - [Comparing Latest Models](#comparing-latest-models)
+    - [Evaluating Models](#evaluating-models)
+      - [Benchmarks](#benchmarks)
+      - [Cost-efficiency](#cost-efficiency)
+      - [Open OCR Datasets](#open-ocr-datasets)
+  - [Tools to Run Models](#tools-to-run-models)
+    - [Locally](#locally)
+      - [Transformers](#transformers)
+      - [MLX](#mlx)
+    - [Remotely](#remotely)
+      - [Inference Endpoints for Managed Deployment](#inference-endpoints-for-managed-deployment)
+      - [Hugging Face Jobs for Batch Inference](#hugging-face-jobs-for-batch-inference)
+  - [Going Beyond OCR](#going-beyond-ocr)
+    - [Visual Document Retrievers](#visual-document-retrievers)
+    - [Using Vision Language Models for Document Question Answering](#using-vision-language-models-for-document-question-answering)
+  - [Wrapping up](#wrapping-up)
+
+# Brief Introduction to Modern OCR 
+
+Optical Character Recognition (OCR) is one of the earliest and longest running challenges in computer vision.  Many of AI’s first practical applications focused on turning printed text into digital form.
+
+With the surge of [vision-language models](https://huggingface.co/blog/vlms) (VLMs), OCR has advanced significantly. Recently, many OCR models have been developed by fine-tuning existing VLMs. But today’s capabilities extend far beyond OCR: you can retrieve documents by query or answer questions about them directly. Thanks to stronger vision features, these models can also handle low-quality scans, interpret complex elements like tables, charts, and images, and fuse text with visuals to answer open-ended questions across documents.
+
+## Model Capabilities
+
+### Transcription
+Recent models transcribe texts into a machine-readable format.   
+The input can include: 
+
+- Handwritten text   
+- Various scripts like Latin, Arabic, and Japanese characters  
+- Mathematical expressions   
+- Chemical formulas  
+- Image/Layout/Page number tags
+
+	  
+	OCR models convert them into machine-readable text that comes in many different formats like HTML, Markdown and more.  
+	
+
+### Handling complex components in documents
+
+On top of text, some models can also recognize:
+
+- Images  
+- Charts  
+- Tables
+
+Some models know where images are inside the document, extract their coordinates, and insert them appropriately between texts. Other models generate captions for images and insert them where they appear. This is especially useful if you are feeding the machine-readable output into an LLM. Example models are [OlmOCR by AllenAI](https://huggingface.co/allenai/olmOCR-7B-0825), or [PaddleOCR-VL by PaddlePaddle](https://huggingface.co/PaddlePaddle/PaddleOCR-VL).
+
+Models use different machine-readable output formats, such as **DocTags**, **HTML** or **Markdown** (explained in the next section *Output Formats*). The way a model handles tables and charts often depends on the output format they are using. Some models treat charts like images: they are kept as is. Other models convert charts into markdown tables or JSON, e.g., a bar chart can be converted as follows. 
+
+![Chart Rendering](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/chart-rendering.png)
+
+Similarly for tables, cells are converted into a machine-readable format while retaining context from headings and columns. 
+
+![Table Rendering][https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/table-rendering.png]
+
+### Output formats
+Different OCR models have different output formats. Briefly, here are the common output formats used by modern models.   
+**DocTag:** DocTag is an XML-like format for documents that expresses location, text format, component-level information, and more. Below is an illustration of a paper parsed into DocTags. This format is employed by the open Docling models.  
+
+![DocTags](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/doctags_v2.png)  
+
+- **HTML:** HTML is one of the most popular output formats used for document parsing as it properly encodes structure and hierarchical information.   
+- **Markdown:** Markdown is the most human-readable format. It’s simpler than HTML but not as expressive. For example, it can’t represent split-column tables.  
+- **JSON:** JSON is not a format that models use for the entire output, but it can be used to represent information in tables or charts.
+
+The right model depends on how you plan to use its outputs:
+
+* **Digital reconstruction**: To reconstruct documents digitally, choose a model with a layout-preserving format (e.g., DocTags or HTML).  
+* **LLM input or Q\&A**: If the use case involves passing outputs to LLM, pick a model that outputs Markdown and image captions, since they’re closer to natural language.  
+* **Programmatic use**: If you want to pass your outputs to a program (like data analysis), opt for a model that generates structured outputs like JSON.
+
+### Locality Awareness 
+
+Documents can have complex structures, like multi-column text blocks and floating figures. Older OCR models handled these documents by detecting words and then the layout of pages manually in post-processing to have the text rendered in reading order, which is brittle.  Modern OCR models, on the other hand, incorporate layout metadata to help preserve reading order and accuracy. This metadata is called “anchor”, it can come in bounding boxes. This process is also called as “grounding/anchoring” because it helps with hallucination.
+
+
+### Model Prompting
+
+OCR models can either take in images or a text prompt along with the image, which heavily depends on the architecture and the pre-training setup.   
+Some OCR models support prompt-based task switching, e.g. [granite-docling](https://huggingface.co/ibm-granite/granite-docling-258M) can parse an entire page with the prompt “Convert this page to Docling” while it can also take prompts like “Convert this formula to LaTeX” along with a page full of formulas.   
+Other models, however, are trained only for parsing entire pages, and they are conditioned to do this through a system prompt.   
+For instance, [OlmOCR by AllenAI](https://huggingface.co/collections/allenai/olmocr-67af8630b0062a25bf1b54a1) takes a long conditioning prompt. Like many others, OlmOCR is technically an OCR fine-tuned version of a VLM (Qwen2.5VL in this case), so you can prompt for other tasks, but its performance will not be on par with the OCR capabilities. 
+
+## Cutting-edge Open OCR Models
+
+We’ve seen an incredible wave of new models this past year. Because so much work is happening in the open, these players build on and benefit from each other’s work. A great example is AllenAI’s release of OlmOCR, which not only released a model but also the dataset used to train it. With these, others can build upon them in new directions. The field is incredibly active, but it’s not always obvious which model to use. 
+
+### Comparing Latest Models
+
+To make things a bit easier, we’re putting together a non-exhaustive comparison of some of our current favorite models. All of the models below are layout-aware and can parse tables, charts, and math equations. The full list of languages each model supports are detailed in their model cards, so make sure to check them if you’re interested.
+
+| Model Name | Output formats | Features | Model Size | Multilingual? |
+| :---- | :---- | :---- | :---- | :---- |
+| [Nanonets-OCR2-3B](https://huggingface.co/collections/nanonets/nanonets-ocr2-68ed207f17ee6c31d226319e) | structured Markdown with semantic tagging (plus HTML tables, etc.) | Captions images in the documents Signature & watermark extraction Handles checkboxes, flowcharts, and handwriting | 4B | ✅Supports English, Chinese, French, Arabic and more.  |
+| [PaddleOCR-VL](https://huggingface.co/collections/PaddlePaddle/paddleocr-vl-68f0db852483c7af0bc86849) | Markdown, JSON, HTML tables and charts  |  Handles handwriting, old documents Allows prompting Converts tables & charts to HTML Extracts and inserts images directly  | 0.9B | ✅Supports 109 languages |
+| [dots.ocr](https://huggingface.co/rednote-hilab/dots.ocr) | Markdown, JSON | Grounding Extracts and inserts images Handles handwriting | 3B | ✅Multilingual with language info not available |
+| [OlmOCR](https://huggingface.co/allenai/olmOCR-7B-0825) | Markdown, html, latex | Grounding Optimized for large-scale batch processing | 8B | ❎English-only  |
+| [Granite-Docling-258M](https://huggingface.co/ibm-granite/granite-docling-258M) | DocTags | Prompt-based task switching Ability to prompt element locations with location tokens Rich output | 258M | ✅Supports English, Japanese, Arabic and Chinese. |
+| [DeepSeek-OCR](https://huggingface.co/deepseek-ai/DeepSeek-OCR) | Markdown \+ HTML  | Supports general visual understanding Can parse and re-render all charts, tables, and more into HTML Handles handwriting Memory-efficient, solves text through image | 3B | ✅Supports nearly 100 languages |
+
+Here’s a small demo for you to try some of the latest models and compare their outputs.   
+\<iframe  
+    src\="https://prithivMLmods-Multimodal-OCR3.hf.space"  
+    frameborder\="0"  
+    width\="850"  
+    height\="450"
+
+\>\</iframe\>
+
+### Evaluating Models
+
+#### Benchmarks
+
+There’s no single best model, as every problem has different needs. Should tables be rendered in Markdown or HTML? Which elements should we extract? How should we quantify text accuracy and error rates? 👀  
+While there are many evaluation datasets and tools, many don’t answer these questions. So we suggest using the following benchmarks:
+
+1. [**OmniDocBenchmark**](https://huggingface.co/datasets/opendatalab/OmniDocBench)**:** This widely used benchmark stands out for its diverse document types: books, magazines, and textbooks. Its evaluation criteria are well designed, accepting tables in both HTML and Markdown formats. A novel matching algorithm evaluates the reading order, and formulas are normalized before evaluation. Most metrics rely on edit distance or tree edit distance (tables). Notably, the annotations used for evaluation are not solely human-generated but are acquired through SoTA VLMs or conventional OCR methods.  
+2. [**OlmOCR-Bench**](https://huggingface.co/datasets/allenai/olmOCR-bench): OlmOCR-Bench takes a different approach: they treat the evaluation as a set of unit tests. For example, table evaluation is done by checking the relation between selected cells of a given table. They use PDFs from public sources, and annotations are done using a wide range of closed-source VLMs. This benchmark is quite successful to evaluate on the English language.  
+3. [**CC-OCR (Multilingual)**:](https://huggingface.co/datasets/wulipc/CC-OCR) Compared to the previous benchmarks, CC-OCR is less preferred when picking models, due to lower document quality and diversity. However, it’s the only benchmark that contains evaluation beyond English and Chinese\! While the evaluation is far from perfect (images are photos with few words), it’s still the best you can do for multilingual evaluation.
+
+When testing different OCR models, we've found that the performance across different document types, languages, etc., varies a lot. Your domain may not be well represented in existing benchmarks\! To make effective use of this new generation of VLM-based OCR models we suggest aiming to collect a dataset of representative examples of your task domain and testing a few different models to compare their performance. 
+
+#### Cost-efficiency
+
+Most OCR models are small, having between 3B and 7B parameters; you can even find models with fewer than 1B parameters, like PaddleOCR-VL. However, the cost also depends on the availability of optimized implementations  inference frameworks. For example, OlmOCR comes with vLLM and SGLang implementations, and the cost per million pages is 190 dollars (assuming on H100 for $2.69/hour). DeepSeek-OCR can process 200k+ pages per day on a single A100 with 40GB VRAM.  With napkin math, we see that the cost per million pages is more or less similar to OlmOCR (although it depends on your A100 provider). If your use case remains unaffected, you can also opt for quantized versions of the models. The cost of running open-source models heavily depends on the hourly cost of the instance and the optimizations the model includes, but it’s guaranteed to be cheaper than many closed-source models out there on a larger scale.
+
+#### Open OCR Datasets 
+
+While the past year has seen a surge in open OCR models, this hasn't been matched by as many open training and evaluation datasets. An exception is AllenAI's \[olmOCR-mix-0225\](https://huggingface.co/datasets/allenai/olmOCR-mix-0225), which has been used to train at least [72 models on the Hub](https://huggingface.co/models?dataset=dataset:allenai/olmOCR-mix-0225) – likely more, since not all models document their training data.
+
+Sharing more datasets could unlock even greater advances in open OCR models. There are several promising approaches for creating these datasets:
+
+\- \*\*Synthetic data generation\*\* (e.g., \[isl\_synthetic\_ocr\](https://huggingface.co/datasets/Sigurdur/isl\_synthetic\_ocr))  
+\- \*\*VLM-generated transcriptions\*\* filtered manually or through heuristics  
+\- \*\*Using existing OCR models\*\* to generate training data for new, potentially more efficient models in specific domains  
+\- \*\*Leveraging existing corrected datasets\*\* like the \[Medical History of British India Dataset\](https://huggingface.co/NationalLibraryOfScotland), which contains extensively human-corrected OCR for historic documents
+
+It's worth noting that many such datasets exist but remain unused. Making them more readily available as 'training-ready' datasets carries a considerable potential for the open-source community.
+
+## Tools to Run Models
+
+We have received many questions about getting started with OCR models, so here are a few ways you can use local inference tools and host remotely with Hugging Face.
+
+### Locally
+
+Most cutting-edge models come with vLLM support and transformers implementation. You can get more info about how to serve each from the models’ own cards. For convenience, we show how to infer locally using vLLM here. The code below can differ from model to model, but for most models it looks like the following. 
+
+```shell
+vllm serve nanonets/Nanonets-OCR2-3B
+```
+
+And then you can query as follows using e.g. OpenAI client. 
+
+```py
+from openai import OpenAI
+import base64
+
+client = OpenAI(api_key={YOUR_API_KEY}, base_url="http://localhost:8000/v1")
+
+model = "nanonets/Nanonets-OCR2-3B"
+
+def encode_image(image_path):
+    with open(image_path, "rb") as image_file:
+        return base64.b64encode(image_file.read()).decode("utf-8")
+
+def infer(img_base64):
+    response = client.chat.completions.create(
+        model=model,
+        messages=[
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "image_url",
+                        "image_url": {"url": f"data:image/png;base64,{img_base64}"},
+                    },
+                    {
+                        "type": "text",
+                        "text": "Extract the text from the above document as if you were reading it naturally.",
+                    },
+                ],
+            }
+        ],
+        temperature=0.0,
+        max_tokens=15000
+    )
+    return response.choices[0].message.content
+
+img_base64 = encode_image(your_img_path)
+print(infer(img_base64))
+```
+
+**Transformers**
+
+Transformers provides standard model definitions for easy inference and fine-tuning. Models available in transformers come with either official transformers implementation (model definitions within the library) or “remote code” implementations. Latter is defined by the model owners to enable easy loading of models into transformers interface, so you don’t have to go through the model implementation. Below is an example loading Nanonets model using transformers implementation.
+
+```py
+# make sure to install flash-attn and transformers
+from transformers import AutoProcessor, AutoModelForImageTextToText
+
+model = AutoModelForImageTextToText.from_pretrained(
+    "nanonets/Nanonets-OCR2-3B", 
+    torch_dtype="auto", 
+    device_map="auto", 
+    attn_implementation="flash_attention_2"
+)
+model.eval()
+processor = AutoProcessor.from_pretrained("nanonets/Nanonets-OCR2-3B")
+
+def infer(image_url, model, processor, max_new_tokens=4096):
+    prompt = """Extract the text from the above document as if you were reading it naturally. Return the tables in html format. Return the equations in LaTeX representation. If there is an image in the document and image caption is not present, add a small description of the image inside the <img></img> tag; otherwise, add the image caption inside <img></img>. Watermarks should be wrapped in brackets. Ex: <watermark>OFFICIAL COPY</watermark>. Page numbers should be wrapped in brackets. Ex: <page_number>14</page_number> or <page_number>9/22</page_number>. Prefer using ☐ and ☑ for check boxes."""
+    image = Image.open(image_path)
+    messages = [
+        {"role": "system", "content": "You are a helpful assistant."},
+        {"role": "user", "content": [
+            {"type": "image", "image": image_url},
+            {"type": "text", "text": prompt},
+        ]},
+    ]
+    text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = processor(text=[text], images=[image], padding=True, return_tensors="pt").to(model.device)
+    
+    output_ids = model.generate(**inputs, max_new_tokens=max_new_tokens, do_sample=False)
+    generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(inputs.input_ids, output_ids)]
+    
+    output_text = processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True)
+    return output_text[0]
+
+result = infer(image_path, model, processor, max_new_tokens=15000)
+print(result)
+```
+
+**MLX**  
+MLX is an open-source machine learning framework for Apple Silicon. MLX-VLM is built on top of MLX to serve vision language models easily. You can explore all the OCR models available in MLX format [here](https://huggingface.co/models?sort=trending&search=ocr). They also come in quantized versions.  
+You can install MLX-VLM as follows.
+
+```
+pip install -U mlx-vlm
+```
+
+```
+wget https://huggingface.co/datasets/merve/vlm_test_images/resolve/main/throughput_smolvlm.png
+
+python -m mlx_vlm.generate --model ibm-granite/granite-docling-258M-mlx --max-tokens 4096 --temperature 0.0 --prompt "Convert this chart to JSON." --image throughput_smolvlm.png 
+
+```
+
+### Remotely
+
+**Inference Endpoints for Managed Deployment**  
+You can deploy OCR models compatible with vLLM or SGLang on Hugging Face Inference Endpoints, either from a model repository “Deploy” option or directly through [Inference Endpoints interface](https://endpoints.huggingface.co/). Inference Endpoints serve the cutting-edge models in a fully managed environment with GPU acceleration, auto-scaling, and monitoring without manually managing the infrastructure.  
+   
+Here is a simple method of deploying `nanonets` using vLLM as the inference engine.
+
+1. Navigate to the model repository [`nanonets/Nanonets-OCR2-3B`](https://huggingface.co/nanonets/Nanonets-OCR2-3B)  
+2. Click on the “Deploy” button and select the “HF Inference Endpoints”
+
+![Inference Endpoints](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/IE.png)
+
+3. Configure the deployment setup within seconds
+
+![Inference Endpoints][https://huggingface.co/datasets/huggingface/documentation-images/blob/main/resolve/ocr/IE2.png]
+
+4. After the endpoint is created, you can consume it using the OpenAI client snippet we provided in the previous section.
+
+You can learn more about it \[here\](https://huggingface.co/docs/inference-endpoints/engines/vllm).
+
+**Hugging Face Jobs for Batch Inference** 
+
+For many OCR applications, you want to do efficient batch inference, i.e., running a model across thousands of images as cheaply and efficiently as possible. A good approach is to use vLLM's offline inference mode. As discussed above, many recent VLM-based OCR models are supported by vLLM, which efficiently batches images and generates OCR outputs at scale.
+
+To make this even easier, we've created \[uv-scripts/ocr\](https://huggingface.co/datasets/uv-scripts/ocr)—a collection of ready-to-run OCR scripts that work with Hugging Face Jobs. These scripts let you run OCR on any dataset without needing your own GPU. Simply point the script at your input dataset, and it will:
+
+\- Process all images in a dataset column using many different open OCR models  
+\- Add OCR results as a new markdown column to the dataset  
+\- Push the updated dataset with OCR results to the Hub
+
+For example, to run OCR on 100 images:  
+\`\`\`bash  
+hf jobs uv run \--flavor l4x1 \\  
+  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/nanonets-ocr.py \\  
+  your-input-dataset your-output-dataset \\  
+  \--max-samples 100  
+\`\`\`
+
+The scripts handle all the vLLM configuration and batching automatically, making batch OCR accessible without infrastructure setup.
+
+### Going Beyond OCR
+
+If you are interested in document AI, not just OCR, here are some of our recommendations. 
+
+**Visual Document Retrievers**  
+Visual document retrieval is to retrieve the most relevant top-k documents when given a text query. If you have previously worked with retriever models, the difference is that you search directly on a stack of PDFs. Aside from using them standalone, you can also build multimodal RAG pipelines by combining them with a vision language model (find how to do so \[here\](https://huggingface.co/merve/smol-vision/blob/main/ColPali\_%2B\_Qwen2\_VL.ipynb)). You can find \[all of them on Hugging Face Hub\](https://huggingface.co/models?pipeline\_tag=visual-document-retrieval\&sort=trending).
+
+There are two types of visual document retrievers, single-vector and multi-vector models. Single-vector models are more memory efficient and less performant; meanwhile, multi-vector models are more memory hungry and more performant. Most of these models often come with vLLM and transformers integrations, so you can index documents using them and then do a search easily using a vector DB.
+
+**Using Vision Language Models for Document Question Answering**  
+If you have a task at hand that only requires answering questions based on documents, you can use some of the vision language models that had document tasks in their training tasks. We’ve observed users trying to convert documents into text and passing the output to LLMs, but if your document has a complex layout, and your converted document outputs charts and so on in HTML, or images are captioned incorrectly, the LLM will miss out. Instead, feed your document and query to one of the advanced vision language models like [Qwen3-VL](https://huggingface.co/collections/Qwen/qwen3-vl-68d2a7c1b8a8afce4ebd2dbe) not to miss out on any context. 
+
+## Wrapping up
+
+In this blog post, we wanted to give you an overview of how to pick your OCR model, existing cutting-edge models and capabilities, and the tools to get you started with OCR.   
+If you want to learn more about OCR and vision language models, we encourage you to read the resources below. 
+
+- [Vision Language Models Explained](https://huggingface.co/blog/vlms)  
+- [Vision Language Models 2025 Update](https://huggingface.co/blog/vlms-2025)  
+- \[PP-OCR-v5\](https://huggingface.co/blog/baidu/ppocrv5)  
+- \[SOTA OCR on-device with Core ML and dots.ocr\](https://huggingface.co/blog/dots-ocr-ne)
+
diff --git a/ocr-open-models/thumbnail.png b/ocr-open-models/thumbnail.png
new file mode 100644
index 0000000000000000000000000000000000000000..1ab1346264445f2cfeb5a17b7bcf07bdffcfb2b8
GIT binary patch
literal 77228
zcmeEsRacx%)MeukT!OoW;O_1&!JXhvaCZnETpD+GcXubaySqC~=bgEjA24gp&DR%C
z_j*=USJgRZpR;RMg(=EQAb!F90ssIIr6fgv002-J004M6EI8;*pMoGb0F2I2N>oV2
zE&Xhf3ALLT&@I?)*pQP<ePoi~*A-9>G!uXTHGOKdg#mqaZvxFgU$6%_pl5(PGyvdd
zkO>9=5U2dl>;GRi>4gUXK0nWvr=$~pVh?Vcp_;HOTN=$27cBHsp+o<~(7+b9;f_Ws
zm4iVg5L8rZiww<PBVU%bR61oB!tOn1+$B0{P1=0rCz^T~qWRC6C{rNsqJ;kE(tZ}u
zZ^)g0{szBB06qQP{&`)H7ZeqO#Q)R$|BYpSomFE2K8(;=*c>|sSLe;=k8jx)D)l!{
zhtw(r%BHH2Jh?y^l;MSdIh_-gyD#m~V3C<me|vOtDi<F*=@stq)?&!Se{%r<s%fqf
z@IxKdEp1y&lTXhE_}2A3H1MjdPW$+)pa1|*(oFw%RxbVZ`IEJhSkukAj#tnM{K$lP
zAzlsgHP=1obt}R;I#@1G{zh%nK>`2}s8ijF6YJ%{4XFg;kAy@O$5su`wy<5`0KcA4
zGXw0wA}jlf%8iP6vzN^G7Tqkpg{J6+mtuKV7yOA$G*BSnL+#*T3+Ab2B%YM3ST$ZQ
zPu#D#DeG@8J|qk17kbQlR_{*kuin31Z&`r@y8V1WaRrLzbqfQP`DK`^&z3q0AZ>i`
z84U>-S}#%n6?g!AvvzW;-|NjvX@y$ZK!J<rYtqWXikjG|Rx&j;EOqS&p+7*17!Gdv
zF*qj1nVIF$U0}_hkwlQPB_2pVRpm3BzUFxm5A@j*Fx$Nv$9Bw2CW@o)P2mc~KD_IY
zQ{}v)`(qv0(r{RH)AWEaweb-uE2Jv_DPt1JG>B~^i-@0zW8_^pLt@V!#~^-&QFLiT
zCA18da3Qb=y~6fvUMVElEMoL`;9oOPkCat=H`B;_p*zvYjEEU5!Nc>lC|!c#)qrwm
zPNmhN(_8l4fK`00xI@FCuAg<Bk2;_+K;2XZwt-A8F54rL=}}CM1_%x>8-Ecn3iEzp
z(VBj{7=7(Db6db+d;!Ne`_&5g-(BYV{&o31%>?R_&vk`p|FRKoy=NnSt*62LdVAC;
zo<MNKMd!8Vg*5fwO2#`ld07VKzl05?a7m^@bSZ*XBJfA#H1?5C3lCara2t`Y&vBgm
zDN0!8UC`6?zdq$%pg!z7&r!lR1Cp)SLQL;19n>((o7umbEblQcg*?%{WRar2!(aDF
zl=21qvsWb3H;@6W^O>JWEs)~qFQ)930l3nN&3MnNGe3%4OuDx(8Fx)XlIF(#&E{Ww
zj-9LVCD>h#FJX149AA~tUmQI>(H%Ipht*hBOLUhl9N0l#(esf^uQ+&O|7R;~qsc8$
zmthosOgI)4j*2YDIxvh&u}SEJ3-yNbI&&-0ZCmn57jxN;$UYXM$S5ji)L4-S@*tFT
zhA?6ty_sF`L2PBz5VOJG3>_SXlw`KlJ4JnCRaqHxUrFs)zPNO+6<XRK#=px=UEyc^
z_Jd_rN4CTPjjJ<7i4KU`i}6U~rrjS}ElcGdOXDt0<v!7#kRwQvkEoa>6#@7G+MRKU
zIw+Tlk#8Tvncas{+Vf;LX8boiqMS;JC=B4i!j##JtPTFz^$c==^xxnlhF+gM7B=0n
z74%t|-bkuk=CmS;<swF-!e>64eD{o?>B51D=8I~AbJI0?7Yw;#5v?kF@HQ0|6wLnJ
zFgdMWYAvrl-!nmK%3}Y!wwpX!RUgNAL{5~kAW;SP_?U|Q^_P<26%?G1RGt?cN;R^Y
zW0Gkkm&~+)t&=NaWDVFXE(#x*{VXa+fj|gLmZ5FmOa^D|S2l?`n#!&pn1&l7NOTA6
z?G58{^Ps7m3Xq~mq6PRJAdMxxz1(v(ypH}^Oh2+VXVbPO4|>N1qfnvwJVgeBwXa1N
z6odr>4sXg}9hik8vriJB*K)G)(oqZIoEC17avMcvgn=9+qG`Zt_>Iiviy<XqaaJV~
z{<?p!Q-WEk3J|v*n${BnI!|QxcUQJSQQ8;7NTvgPAZT=0XS$42q(!eZ<uu)89f$G>
z(%F(?*`t<fO%ec=)C>6uoqPo0A}5ku)+GOe9C;A2Vzp8h{s{*U$x~_=D%KyQ1u6oi
zsTu1*Y@XnMQ;6DtsfkBTWS<r?1tcMXaK}hcfJ6PYs1ZKd@P-YEE-|F$1&Twp58J3*
zA>oM{{hXfv!Nhd_5egBQC|B#XP)`dP<VQ;rIQRfKJ0%mAP0a>;Cs$;Cmhutbwn6cz
zm^Kg6gGob3L&T&3wiS28r)uvHQnuu1+FCafSjEw)Nq&1(60Y$Q&*k^7=UOVx%0Q_%
z7Ve+7K;8b(8)f7yG9T1H1|=9gQqn;ZQ|8=nq`dCgy1-S%T4X35MHj)SQJoP14A-|h
zvt5bkUAJJK(w~t`F7(pm7qufi@X;0A8ZDG<rKKfxi``z-$zS~ZiWMhPG;9p(U33EI
zNrdq5N}>=JAaN<|0*Tj;k^P8Yk&Aw(NjZHYfIBN$eVb<=NxX0T>u@?_C?f=H$e#P5
zV_{{R82lsus8a2_Alka)z#(N+tGngh&fHZqF%e;eX=KJ9kXZ?78F7_LLV-a?Qqwrl
zE84I2AejE-rmmf%vaQy6W<@jQtjGUA(%$@~%m%`jN9~Z3u&K$c84UHN6hnpO1%Q~?
zn3N)8$JhW9JWfz0#pTWXd>rmKd?8LD<!NsKTYhj17)`8~L-UePha#ocV-5oMxfzO(
zBuX4HU67zU-oCnq?=X9jSxOx|t%P*NH!YJ8f$cp}dns5TY>?P)-$mxvCUY;1?C3wp
z-G-6=pc!Y1SqZu^_r{^}V{RLH(m$SOEkJ{!fr?<>o%)XbDR=f))Gn`$71l<FZ*}w}
zC>&j&P1YS8L-tqGR|M>+m-KyfH2uzEfQd`ADCmW91bMp5n}?$j?K^o5(}0ZX5%P?r
zcwNm>J{|Z9Kf(hw>8T5?juP)zen%l-LBsmT>IDX(WL!*`m(xm{+0Z)2HxjzP`Xw>S
ztVU#*pnSD9fe}_s9uzI6fhLTma{8cwBmIy=MQ|@xdVfsc)VX#$UaEV!y@mT<pdvsH
zGpn*Ym#=BI8yfi(sV+$N9yC{L#APprHN#G*-nL(M&2VsO!r&RRgb7KF9g2(AB1l2F
zNgc4euP^*b{Ir=EZ<=FWlBY{jg9QT<5=GC0P}32(?w#q;;|_z##lZvqOJPTE;v}3%
zJv9%grVIk}9*s`fyzu)lN^$;;;`6P}Z8f*Qjur5=wD7JrM9f-JLFEx1RR}tft?|=K
zb>U>3+QjYQ&ZvMDjTIM_pwRuwbaUwPYW7)bjuIj!zf-=O&{#TV;Rq5(RITti*q7CA
zjFF>hbad@fIpKH>y5jC7x@d!wVpO7P8=?{$&jK``14C|xpnft=B!4AXoTx3%qYhP{
z{+b^N7ZS^0D*MH2b%)KTa!)DvTAj~i;g78)xP#buVF{5sD@#6oc*ig_sr7NyqYt-0
zFiCJg<i?J#?Y-wmEPzO0WQE#_)lXX@<y9W;fgR6+6};2%v2rr&t&@po)3R=+gsfIR
zGn2wNS3rkn+YCO=3AYC|wI((Y6jGMAc)?f3piIw6*i&<bw;xxOvZ@#xTpLI=A?O(|
zL%6L}xZ$VS#8t0i{BuIaw)8jCpg1N4p*p*(&h1L&c(^IauNMTHD9E%(P4s)aPfAYr
z_RJMy5+Q`B;Wc>A@FXeoAi1qXS9#t?OUELhe1>bmn=UR1;O!z6PJvGPz_9`wj|60y
zLc&1QWd86c3wke6Risg*Y*KjVV0y1`80II#WxVMV0^H|u>OpKrK2p|tpq&Scx4KfE
zEcqB8lEq-%<Ll?y2Ou7cyw<LeO~tW}%mkSUr`l4~rSUoQ5Dx^dQ4T0h_rXMQG#}{m
ztV)G8JNW?F{=Df(L_1K@Ql4U)BJ;?7StqO#1@R`cTxuA#x|et4uUMNA^)v)pNEJhk
ziT|zpw*Up=#;wrcH=O+pOyC&|i3qUtykMv+2}ZYV>8L716#T0HejK7m)EwNHxNG!b
zEi0{aX^uWD9#w1k8Ntvg^m<~oXi^S$;82)_JrYDiWGP6lo85ygJLU<*oCz&&l@eJ<
zO-&xf1UYSe=W~Lm%0UQTRD@36TvL5}TS;%lWXa$a)eM`+125<jz;l`}*x1k_5tmg{
znQV=tu&-Zv`KzDGuepgN$WJGH%KEoGUx{dq%_yO&ry`MgQ0HGIp$7hfI@nnJTveFM
zZ3Hwt7xRU$c-@cU>R}2*J>ekH@oWkmXgy+p3Q3>3lE`Yt<;A-4Uvr+8_{LgwwQ8xC
zdj;?$-NVy7m=SFMx`Z%Tl>DCUAE&pz{GA?-Frtsg)55TJ8FzOkM!wDioCf5b0{_Rx
zBX?<xw@H~%^X{;bC-01<l}K|3dF-3~!{dEL-l)KUQ+y6M9;FCmLi>NImVFH+XV&ty
z3-P(@gIj`YZK-pt-fu>m*FO0(Pv%`#Txyuir2g4j(%s6*@DS{lL7nw{k`<!Jg#Zgc
zDF)N-{#yT?sWA4PAM{7&>$(7=zTk?o$A{T?#|hNeD+h1XR4eGB%AW<)Fo+beVVAbo
z$RfS67?<|iQlO!AP_`AFZ)+bPOw?F+_h?O|{QS}wZlAmKA;m+if}(=Fu{}i5g#qN&
zDJ()=yMfh@U6!5~i3j1hWwH;CbY->fFGL(2)tvh!!6SOy6R@9^{xv@XB{G*`3vY!5
zIp2i0iGKSrJ_YxVA#*dl`WA(3ICn`Lf85@&KcMNn%<8Q;a^zJ48YE^E%T2Gz)NXWW
zz2<TXxUuwk4+|}^jjTC|yA2cdSC#Ib%z$wY(&*XbhxU2SRrJW|uKrlgvE=rUXUB6c
zh81eGzAeZ1pN(3yPL@eLe#bTbpipoer_%Si#m0l?<gJVC=YT}rMa2({frK3D!Jb>h
zv%0N0S-#al#|YH^{44Xy^LYJ7pIZAtd}`{VOCdL>pKpM1a0%rb`g}vmrI+jF0oi@g
zWl_a^=uq;@$HR!A4VF}#qP&2MYYpMap0N2SL*vHV?eBEq2wBAqsVH(lyZ7PUYT=M%
zJ9KIdiTv$O9ER&+cS=MsCF#AxxaRL=S5kXS(m^s}a1Xn09-OV{`@qAN!d*y8U8tx^
zGf!<YJ3wLQ(g+g6JwKJStMYe{tAa*EJ*EHX-`za6u+cGR;QvZuWY<fs<9<7lX=kfN
z0T+gqpFA}1^)!Zjl|iPvNAI3|*R8G9OBd`u89(X$-59*vd#6R{Z)y#@8*9eyH6Sb8
zi$Ff^WEj*Ea>AQHBsxRLnsk5sLgV8<gLP+J&J#Z6vpfZ^QC06js|^=8lXKkxGvhHX
z2gNkF6hn^Q(O$BJrEe@FOckO82No41ae1YEt%K4qDsAoJ3KR)r=C`X;Y6JeLDyuB7
z8>#gUZbl*HUox9ghL_#oM*L<z_?vQrF%pqykT_rOcS7(L(|uh?DgK_F#5R4AYj@@F
z2?cM+A0a`=)UsDi1fKW?_>j&4)**%uO8TD4$2Aocqew(PcHqIM&FV$25kQ{eFdJYo
z^=<q5C+>{=*4`QKK0!P=c8BQl<dJ8Vxtpgm9xNcV_)f#KpXIcXm*3Sz7-ee9)S7As
zWQs`a$P{~cO8Se#^9Cr8E1-B@x<r>T!dCT;3%obYmTv6@UX43$kh<6Hxayspw=!N0
z*xE{nLB)#Mn3SZyxTw?eq<Xmaosn9~phvcE(Jn+vm)zc(kkJG-HPnURr1kECJ0jS)
zHtUjtSg(~@j0-lEiM{@pn`vKwT6FzDiih`9NxQd_wqLFj#+j&9;FrX;hcB|Re0bXb
z3N^LUq`d1$prr2JbnUK<afRc5DT8&%xYUmATUT>u!<dJq9V6+RmHSpl)02e>I44)x
z2<2=TQme7Wc}^bYE7n&(c{1Htw8obn_~#5$YxCGOP?CU1YK3*}1jQj|aypOvR(G=7
z=g)EvXc)D!_;fIghen%3JJ{a1`Qxk%hpmWAy7(%42?9_M8ZU*qU=rZHzcm%b)sgqC
z61(`|e{{=rYT<0{m9tBA?~%E@bG&upwJMa=uI$+%Ey}*Ui`S{TSUE>B9bVb4T2n<}
zN0%iVt$`mQ$gioo0FFO<4tzo7PxqWU+1Ns2G<Rc#N6|-d-z@6a0eea=3jXSP?xT=5
zQ)=m&I=WG$J3a8IuVa6)t8<ll%+GtA%cTAyM4@+R=$M?u&681GapOk^ip+s~NQh<U
z#tI1}EU~b+)naL2V$)P8+Y#$qCks3IRlN;6E7ZVNZ$W6rPCs46P)r+KOPWsSx7)vq
zTX=u#YC7O9*A^EE5jq*6dxs#6P`qTX;DVyzHo&G|1rJR=rwoD$(DiAFW)AUShPIOg
zc>XW4z>9_YQQNW=HdUA4Bnt_q!#DN%E&%m<OyM{6%Y}{<To7%cD6Pu9Se}xq?H9a%
zzGBxz^h6Zw)ef>B)6`I#O<j(-SvRP9?6-F6ZT<+!B~rYDD81FuT(J0aabogO9SyfV
zQha>geW;YjM2$*v4I3811nZOc@<N<}Rj3n!=u4p3AhH7DSOiX-w_F~i;Ca8=|9Z25
zsNN7sZx;K(r-O4NAi<<5uDgQ+Ti=~zNst^8fMi=~@@XqvIL2@_s_~V{&cenqf)-tn
zs@SUIY5%f`mz{%hI79$?kQYD1j6FY6-%r!Ed-d~8eStvi*eBsk#5?wM%wf*T4Vuv-
ziI4ZR9R(Or3>S(WU7ZicS9UP$XG;YA;Cq3R2`1YNzPiahHV+D<9h7b4&3(Gh6UJ<O
zAY=t}SNqWCz>r+rd`CePr!Xu^0p-!ng7|oIL1Fu#I3y`G*Fxu3q75EV2Hwy!kZEdg
zws9E+73GhNZ}?TsVQXOyWC0Ep52IV)xY_LK`1v2oJzF6sT;U;v;@c#TIeZ~G{?UiL
z7E?i8L}Y;%GGX*KeXX{uhmLBe<|+`7Z!wAn)9oF`m$Unb2wC`ingq(XmRD?rL?Jb7
zQk;w?V$sG%S^%UZOJ*$50xSp1pn^iObMO^ZO$$1>u@Xa+)csC)Q&yqdO6!2$A?WC1
zueldgrMHUlAt&u0L?V1T15@2QMFd)ZKLg**&ITs1jmRQp?-nN{vffB#mNoGFgc#I#
ze|qe(3kW>~hOX=ENEXogW9pIGNZoJuUcB*g_cz~$x###yS_h6|<B9Uxq#>eec|T5N
z&~diHQ@kp=bbCW(DBJjd$*3n3ai<(*QSzA%?8Lefr0}$F2lt;@r;D!y={neW+t$q9
z9@#sf@Py>#^28^BL>(bW=%6P(FFom#pc9P=`+}#%;BB%x`R(A@{rGA3=QzamFbf{3
zFgnw-7Ar>Hncv|B!})%Q{VkJ^zetRunkrKGV>T~()c%Po`&t2Qe|rds&3|G!u($kU
z^NuWOvae%lrd3O%G|E9>8JND_+*jM$TwI<gq@)3Rbd<Culi%y9w6jGFek52-Dc?Hd
zj;up54|;$mn}Q44SUaPlnLXaRMBspSG3_t|p7R189zo5I1lV3}`h->8647l{|AxT>
z7Uzv&zRjuYvf}_y$>C@+8TmTrvf}HCY7Wg9zF+?xjiWA;Me!pKi$D|U^!2K2D>I6`
z@1JeN!AqGFOYx!u?|pTwnvcwD1Hn!Dp+t9KIKBmw#qQh(2Ro{mRmHUJUz9i^lLiUY
zU8p!<f=>zTeJ(Nkk~lg%fP&R;PyYGp<f@;h%<9jwF;1;4PXsRE)2~mwCOKzU|41en
z5br+43PeRuHt7B8&CPKt`*j(lLFPJXw9Qpo=Z}6sD_4eG)XS++(ry3qk`nAu?I=^%
zOKZnip*0$%AB@l5>scS=Pl)S{6+)?zDUXb$`=3ieR%NR^l9U=YcMlhF?oxTSK+KBR
zCuO5?Wl(7ucSo=6hHB!LMRSpUfm!-2D23t@vGVh!iWItn_h-WH?74y$bHZ(WLvKl2
zSDWt!uEsPV&DtK>hGijrGNu<h>&t;aS{!;LK{JcD+Ln+wDQqyJQ<d!dwKx&(A!d*_
z*5Sj^srX@)LxiEVLT!0>f^rr_pELNlspy9#cBcL?O?6vk;y8-Gv5vREgP}s4<o{!J
zML&cN(XoSLW?;jG6j|T;wz0rF6fV`-)prF12=YsH`VIPP6Or)h%qsf+G_9-utMY`v
zqxp(p0sTaZF;PGApR$oIQ$^b!&^<No#W6xec1O*-BhV(D9e!7jf0I5Qq<tAO+GLC(
z!FX@RHT0oG4KBPxfqQGZ=W-y1VI7H$+@)AQSE8z=z-<?q8FpnKngwm7sDF90#5evr
zx%OLR6C^OL)9o@0WJb@-Z+E7GI4C2^Plb`|A^r&!y)R%D;?`a~<5S&6Ng`pyZ%@ow
z0U;22FY0x&e%Y9H&f7GrP48!Df%1J><VLnd?)FmQZB_>3b;sU}r!KS}Yhk)QsiiU3
zTp+W51!8w)C&P26DP@!yMy(Hy_v{og+VTsd=R97_(jBOsd0@3|x=Qhx<*^Za(YA6M
zC_B%)8gWtTufPAu6B~k?J1BdG1!XQq(PEopovS9Bb6xgy1P$#^^dz(*^!7=2Nu>k3
zsZ`sgv}IM5pA_J*^WVldxF-g}NZzEF4_rXzS>h_7?S&G!IIT3mWchO;dLcfSE^B2I
zWnOtq;T)Y25;-+r&O%Nog++2q`t+OA)o76V>=P}i2RH2=QRt1eUqRvw15mAbwaj2&
zuF~+aiW}PmwR_;&B-{E0Xr+3Z#Wctp9d-19{J2=}L@mK>MgDdSH+zQBk)Mt`{u>vf
zU&0q&N>@jB)qo5f6;ubV()(jtj5tB*)f5C5o``C4of~4|omVG&Dd;aQ*~ZAU2bo=?
z>yesG_tvapq=xvVx6pc&JB2?7<zNYR@1k1LY7UL+*-91rT7t=md2zOPQ8E!e8AJ3f
zoI+ExP2HfYKjA~J&hP?eu(FR}*ob<tjR*r_tLzwFUK9~w0YTk&D(?!H4_HY;)PhNH
z!hQ*SBaCnCl*&7&CCR-5GQ+_rq8Qm&Jp8aS?d@Ou06{2CAFm;OV^~nQ_kE}9Cf1=K
z=q};ZdyKc>l|)SXnyf4B+4j!pGB#!0r;=?xqGf{R)8d=E{lSh7i-_f1J;sZ+9@tR$
z)oT(_2?jYt;z64GjhOi>cU7I9nhW50RmI0i`K$B;sgJnpSbiB7NS*&pL5g2u^}Jk4
z8^G5FotA$$MOv&3z18~D%!I=(E=ZK~r<hf;bxn-?w@B{os&9glto4tNl%%9FkZ`%W
zh3FZ^jl!@`M+VG-Xu{yA%SII^3hI=0L%%K5I<fDr>~JgC0V<1zK6kU@k4bVwt(VkB
zeE^Wkg)bJ_*DJA&<xy$1f@^O~<G_*Aa_~mb$d>CQuv7S8O~eATx^Z&oWuZ)d@QF-1
z$mP8oTe0+Lku`b&lL1zve6CzO#WTa_%EH_wY}6t{J@V8`!YdIeVorMUMfZjllvRn!
zXgA+K_|s$ou+!2@S8Bx#x(kx@aL%IuS}<}&HHNOP()*4ot%y+@KfWS#5_Vz7A++5N
zj^{y9b5>FM@Z`!o{2MWFMAH{r<)PE8#zHe-g_?;Xf#?-SIE7<}Z|59?N&AKjpz=9Y
z5H{y)^C9J?x`5@#XcLORAkjHZm+g-|#CS4yEY88T;OJ7rGuFDx?u64xXCWh;2VeiU
z^GWA4PWWVjWtrE;M9%t+mP#+#=3n3A4!K>ETM)VMV(uGcC@o%p6aS>#DZ00yad5<|
z?Gk!eJmSppp(twm;0VRue6KZtVSk%%P%w_H5qey;yVcW3BSQy^fh24mdDI_!^%@R{
zPuR0Y&n^DzuugX0BBmAvNjkju3y6XOUK=^Alp19h{3l{o2*7ImKvy?9SJLyYK=AsM
z_Mq%5gN%!4%?|62>_g-=t12Fapl&^0*uLFM1ttI8&Nfj9MHtfUl-j3hNqdUNj^%(5
z1T6?a3aS0_)Ko)n%LUbg_wP9G_p{H;rtSU3x@(Q^{O-MB^5A*)>oVT!9SDNzFx0g%
zk9^V(+}8>G;nE^&82$1>in3Xic2=SXTKIaKW3|83L({@3rnu|dABnZ15eM^djK=Sc
za|c~y=2%W?YZ1U92M4Vj*+s)#xIZT*l?$0<plj|*ej_h5KH`4&R9q06%CcB``pLOs
zj7$_jNlFrg@Z#cmI4@9wg=XD&yGa4r8;~D4AG_Tt9H+xPXV<X(8aX*eR4m|r!{WnA
z@v`elf3@`(SF;^iWWxeYJ~<eWC1`f^-V}U1ti>@MUvSgNnJ!Qc`$t3)95?)RMQW$~
z_S@kBE<TNz7$$Y?jLN=K#vvxC%B%FtiaV{es4+iIs{{vH#ndV78%|a@38a=9grC_;
zY4a`(*@I~G7%WMpr%keIqn!x_oY#G{qGck<+*Gl&LnByxRSCb2ISKeZNe<1T^%NtF
z9DUUijtMW0;(Uc1A2=GJd&zafXCs&(Q5*|^bdGH_4@;%e(|azUM6dMIUKln5GlKq3
zk8{k*sMVX}eu{;rYphI{lU-#upm`FThl9}Flq^j=gv<43?|#Q8yA=VFPw)ctS3(Sw
zKdbuZ1@F@y*4oR9^1?gz^A1lbfb5bl=W^9fJ8n>^Wiu*KNgnCNsyTCn0fs2R@M0Cw
z!fu}K7SA!3*cMC^n7OJ_2GAMkcHTNe+sgIido|K%Glx-ROPpNc6q3ru^MziS9Hk(d
z+nIcqz-KSyGQ|iE04&~DC0^OQCd01#Syq_1pPXPB+v2bE)v#kxVn+|*yY4Osj9Ral
zX`WZtXjz{|bGa7KIS<KLqpzFt*p3#eHhFTu_6&@ZwAVKi!bg5)c5o1FAxrVaPT~#h
zKFT~#Lc-th3$&Wv$UoSv^afM1;KQ4+aXaK+f<6_E?Ti2h1>s1hpF0_-?b#0YuyN|;
zxM<32JwQhm(S5F7v)3-P{9}g7a(i{Bmhw&k#E@u44KRzMZf(|DtjG5E5civBN1T@`
z=XJ4=g#|~hKG=o#?JcXKqH%7n&Z~T5uU+*Lu#ivE%T}bM5ESZrW;kxHHmmPbY@@Z_
z5tsYwk)GG37|5OsICvkd4D?d=pp(0nK=F`MpLNF$TQm3vuK+|}F$-wg3~GTH)8l=O
z?-hh{iJWSMH+3%)2BM>+v>piP$;3{P=#F(?+2N6VL+)u}x)3wyH_*x%jPZ5~ow^Vr
z5B_n%lhCU=RbX25T%QmB*Ntb9t0(LV8qBXW5<fbwHygV;H~%J=teBm{6kAjW2b&iu
zOna^z{1l}sZP0yJ^cH}gS6DR3c@Js5XdO_vvpRdh4?>AUiC4Z)b8${mU43MqDmV8r
zw$0TWQ*Oe+Q;BPXHyD@u!Qd!4nS^xqjD0B(6lgUUJrSPQHoXiI$)2F--P4utHY776
z%Ch=8E_Zf3QBW?NkPq1LpU^J({m$*&r#p38@Cg2FJ>UiRYhILr^mFKNK;3+!gIP`O
zg$Aq+&!*>yj@_co5XEu7ot_n;n}Dd>%hb?mnhgZpb}1R@QNJSthqKEJ06rdW^&xH1
zYtb~`BHKH|KeRaF<5@UH?;8MRTeQe|+0%1d%8t^A3=Fg<{n0SD>}zh-pBa6<A+q!c
z9ODSMaGotMuKx&9Z6F`--&fe_e5|>e;m=@mF8qs7uJ7NJ^z#v>!<T1Ee1YEBXMdXc
z*pef$7u$%MIQMi@^0Izd+VZc(z$d>2X++tegGuk$WqNUmI>|NK^n%c(?*?T(A*-5n
z5LrQz!a`G7oQu>TyszDYTiIuo6-f|z*c?+UjNMH`7AN8hWl>{pNnc)Pq=EI)F58qf
zB)pWLc*IITk7uqOY*OcfjOT&+((}buB3L&@SUpWFFBY_sfTww>@P#S<v)+E<b}0*X
zjWtF6%SDpv6v9c!BxsH#!ZtEycU7H9Ca=}Eb&t<ocPUc1OJ2Km0)K*{6F9qSi+f9@
zL*jyFW>(T0P_=Fn3Z}@nIX<m@uCpdADHJ>8Rcyl%ZjyEE9TH_#T!rOze|WbK_?+F2
zKXtDwOsc;NMb^}9gM_gFHVkB!P)i#@3vq_?U1+O=M)8pRh+pq^_Nz3KZM6a;EoBfb
zY1Jp1UQbj=UdP8HIDRdNA&WfqS6eJb*5+|xmA*zzH+3ji_o&eu=g4L75`=Bs#f3eE
zYF`peFp{)yr7LR95R%YP$-O!{M2a7$ex^mK`D`5I&KO9AD#jlizrKeF0#Hp7*i^HQ
z_Ik_)p%Foh1bQ*&(!+ukn`v@j6JH{-uRX;6v8(3DUNrJl0eK&|bj8u`{;ly1zTw$&
z@lYBkt8Zog-C;cV3eiO$B(^UQK_;*^7GZI$7<;O{Y-Rz4-^drEwnmvB2w9uT2qusY
zoZJVCSpfC*iU+7NU9PAC@v<6C4Z3-b>qWxIn<sdCcAslD%(~YnNa}nMpcBUm5N51{
zzE22RL3oLHG%7UXyKB$+rf=ZwVu7}=va#=HIn<z?t77wNU?Z5Lni)J&l`HnIOfNc@
zw^eVDn3^(c47_+Z$Pz~zO}Am_Xg=4{-wp`OXv>n6oMN{kNBD(zGK-p%Jxt09UJ10k
zzyK;L*|BNQy*c>7?z6FDLvrgd<I@848ESU03OzHeAO%I2RVUcKag%|k^AZFyEin9}
z`q^V2jmd-4Ot4|<x8G5BLC1g~n=zRIXGfO*hV67LJ#n<RVCeHlk2i>La?v1_yI_0C
z>q0UJJpMampCl2MGJ0JIfL<M$pY-)*pWc+|PHdx-0atZqpC_7Llg_-EQs})_`|;QY
z1Ex_<|Mu8`0?ur!X^n986&$qb?r*5N+ySmeA|ioGnDz9&ZSNV6hlN51zP7%4&gV5e
z<Fgi|IG1bkeB_Fk3Zb|drseOs^0Q8JJK+n;d|eCQl9``7XlKt$6(AW)Lw`iHOD{QZ
z9yU!@l(<)J5nGr0%(}?s5uEY)OR+baLze|F`5@1TOmr9TSGTF#D1bHdIUS##cP<%o
z_v*Sc6Z_l03o+Bqja$QNksbMFo;L$a_wt|7V_(7~Zgj|XAwk<gJBv#30TR%WO()9G
z=*zQ?^Km;050t+W)q3;`DVryMb0H5egk&4qYnR{rKuA1D5cfmPsV&*<^WJ9%0wmG&
zK7vPz{T=?+oo3RY?~p+|RVJWJOUA$zVSd$ToxeJR0E>aktiI6f-uqU<V=CjPce)~V
zEDRfWJUB4;s%M>G0t+S;de}Lt>HeyMRq1CI2?n}hlj3Pyc}@XR*J7Md9oV_n)0ki#
zlYWk;^XslD2E34}9Ojaa1yUNph?BT{*(oI9p4ec%Rcn9DPW{?AN9M~D$I#dX!i*1C
zZI@DoFNk&{g7c43eG5P#4xrtw)PAQ}98!?#T`T$hLFbkPX)GKnmC7=}NTQS+S1j{q
zR;=Gj#5+h0_9jwlFs?gAQcgCguyoQ?)q+t6HuBF*_t}boBB0ZwW+$xs+lNy+EPJl2
zFV*`xTglH4y>yTa@EvsPsYOJD$?_I8X|0ZLzxE^O@EMuRWmMC5b>2Uf7c_frWM>J2
zq>t9*4VW2=(2Rhiuq)*Ab8xb3%l6GrF;|edXdedOS2yW*%t2;><R$XUoX_V8<$(Pk
z3fa9-r%f9_K32&9p~O<nuUdIZKWc$ifpO%FBx0ARWGgEf@NL3dnRt;?Z~Yi}RB%xM
zVt|lnZHFN;*PbS94p@ag&dadOP}3!cwmi?B%=fD~EX>KfFg%9a8ZF5(`|VAtAi(wm
zUX;M+to?qDs7Ac1_h&jJ)@N2gP3*-d=msw6&RmA88~*(K;3i9*ZikFo%^p2R%VU7&
zkU_P-A_N?^q-5S>oFe!(JYNC;$so0+hUsFx8LMSPH;H!jq{AkIYm=z1japBN#+?Gk
z(?{Wkx6Sop&C|@q_@JYxWJgo1z$WZ0V0aQ%Az)*3c4iV*#=H@xp}6dJ-d23$`(CD~
z66KevT-u`jvU@HqK0^=EMl3^5vRawIb?Td%TPuv$mH2i}cr%xrS;fP}Y;$rX(ZZt9
z)nv)!SZsSwIr;ky<!N_LwN4Kx==ieY;(GqmKZmLsXQO7|dcMq}<}~l$En`qBxI6|g
z-oJ~ca17kd$o#a1@h@p4?m&Qu#97~Mu=>Q)w1_RMi0J{36|Cmw?9$k1oW1!QViHU-
z!Pt3P@AN7wkV99Q&?nnVH!HVUS*n)al%<>L@;VI}*vO{Z7Cx$qeC^3m|A?!iQt9;8
zH*G<+^8;lSex)#6D!-!VAUc(!^E+hVVQwuf)G<|yUw8m!Reb}jf@m)Qv>2yHN)NR?
zqf1i!)oDu_sowTQt4=-6fP$%4hqq!|Z0BT{EN}o4jatW_a){1!c`>nl=@1<B&E?i-
zWB-2HpbG*vi!q*<ejBn(iZHCo($?D%!^?^)EyO<`b0puzF@p8uLzU^L2}CcPXt9p1
z{aFq5Z1rOICJ2aot2<lIH^u83=zNHQ&pNRQoc!8A+{g)E9G+5{CllvQZ4P>Uv4RPu
zfx2+%vTdyJi^Xh^X5(ZzsEI_w|4Cqdpr);+oioGLQ5*l;EwQhh7OO<Y-pQtn!Ft6=
z5?ANed1iN5;5_gr1S!*wixc67XcOXPgS3#{*RP2*sK94qACsx%1-~9YDuceFSTg0i
zfNAk0Gt=LCN4(p@rj30v_e*7U57P=v)eW)jb~eUa=T$TxFP32($yM_PvjN39e9nex
z8ooeHG%A;~=Kw<>m{PVj^Z-M%(XX~kEvm<GShWxY!6MWYAcmRaBFWc8{qLo#jp5-{
zGe;l~c|=x?csn!gyAQ90oS-4J-RI8UAF<(-$Y?u9RY*$WqrH&j5bY?p^HF@V{_quj
znOx9OqkDK)l;C~rLMq#A&IfZAN9ct6&eCzh`yqr4;XBY(6&@~)CoCHAgQ2+#`WfCx
z9H&LvRz`Oe^I1O{dg|3GC;59r>o*I*WEqt@<e2A$V;_ZfO-l=Em=wVj_bB@(03Tc5
z$omhHuuWfTrwy`TS`pI^+s|i?%1d+{1M{lYD)U_e95MLE%vGn_Ew2(BLF*VMF79jv
zYsJfy?j-{wolqub!75Z(0BS6a&0FG4f3ePXF$jMHu1@%&m3h+^u&f@)66b2hMcZ#b
z(|I(|C-Be?NzPq1ak`l83s^|`p=!KAjbuqCIIqX)@%8-SEh-wB751{kCz<-mdAeK(
ze{UgCP1ueoa-+ZKMs%`d->T@=Z6P%J`Y-8WI~ip>Y&Z4U4#cX`0hK^uYAGD&zMV-6
zK4Ku$eUBdHK;c;xfw5`Sl0@<&CgaPrjJ=GgXbRtm(<*HjluTJ_eJ69CV>+}=F`+h|
zP<Hh%+zs_E_bM^lr<*Vo^zZa;6p=Ovm$nQnETC`#1JuwIr7mhCi<0|QrR^7a=MGyM
zF`gm0pI^UehmgCuJ9h@-KqW4$Vx4Jv@=lqUnfl3V(8%wp)AGCZ9QRGYvLDCgK|uGz
zJVqm5U>+0U7=&IeLgChrI`c4m-cXqp=Kerd<t;1LVgA@YVuKq&H~8|fMskVqr+hF>
zju6ScTyq~Qe6laLv#)9Fj&EB^p3E^MbAo3^E7<7N-p!;f*sO@Y40Q_d!}RBE$OK{j
zs(EDd>}|B3W{@tty&(%ghZDh(OMFwes4r!+$M*V7vKaIM06+=TOa_#_{J;c>+wX=9
zx@la^#ga8~%xdwqA2p&2R&mzq;%)B?yo>(e@`0UIXtS}z6J&MbCneI^tXRe3$z_TK
z<#V3z{}^1xVq}qG^(pq*C6p2I4n;ZoR{q5KH=p80n5%Bgl`BtirD3S{Go@l>T*^LG
z3F3Za-5E>^7J<Rn91xHQbN|;yz`*1;)pUTanH@p;mNrY}I2#Xm?YOjw#axt`>EQCt
zNxqr7C8QyAcV!2(__-VmG?<C$YvuTkE^wgsZ<t_qv%#)k&{k|l?~k1Tbm7PzO5GTt
zyv6WwS@l2P0!OO||FbT5Sm3lHDH@DBzm^vgPMf_4PxvG|r|?f`2L(6Dym{9_6Y5Mh
z5Ar*a9bzYZfRSRI0>hNV30nHn+T_<d=ErPSydHvdeNVaG%HV1hH&2hp8f>rl!y7FV
zE=k?+ywej_LOn3j$QPT8K{#K(J`do>C^a>8G8p)b>+2LvR#*&nSy<TOFI9A47ErSW
z%Wlc0<(eV5;43pK#%kTT!SAjx%ugP}9Y%;g8aO%8@o^WuPtCo^XDAw9CJsYiN@6A@
zvN4o_LQJ^09!NoA#%@{&a4+StQC$fwXl%tF`eO(^Cdl_bUZimZybi_w8#K3W+W~%n
z`0Df3;yuo&t@X~|2*!KTESF;fvBv=7FVZ~5zvUy;KQX*MGr&A>&ep5QxSl3L0z=l5
zU<6I|TzNl*OSssV+>RZTO_Ok*zb?=feOVx!5>QrwBBo#Btkmw$4vIMv43zqYTc0aQ
z&Z54IpOK|af%d&*9-9~~bjxRa$-+|I=Nwn;++p?Lw&385`BovsksO%lqLi;U<ImzT
zw8q4mquiV&esus{R$NbS*~|Df@n9D#TrWPBcf*m8*6Qn@O0gBKghO{Wtz^`cH+AIm
zO}@z9aT(1*hM89~dz98c*5%nwyL%bE9|OZnyR#;`;CzEW-vZh6AH72OHrc$@tcegh
zmK`UrZbR(dD41P1+%B$Gr$1X?qQ*Y`%T??a%5rlZ8hRr4k{8lPs}g`vy?I^mZ?6zS
za)n=v09vx2x*w%BN#d3m3fQ^&x9%9r+a2HW0H;S$(W6-&Z3#GW-@YmYIpP@_2}9=i
z=c|TWn^x!N6qJ>2eio9_t;>t9bB8O7<JzSgOP`g_j?gY^W*@*=G#Mme=L*L8)9&9?
zs+_iFB=r?p#|7CEusyDAZPqo(GuAy;&WsMLvj4&L&RO|!Xp83Wm-D^VIZNUg%&s<q
zrB3wjI5+&6ArYk~4@Vy*20M?dvow&(y+Piz5I|D?*%IfV`Caz?+75byg>+`D?3S03
zc&vTfxF-YtVJ|^D;YPP1TAIGDwo8wt*=V?4Z|1dD8nZwz1_&g94WP2TczJ(g+NA{N
zbl!Q|hImCLkmbut{611kD^*Ji9{>=v?T+v9o&iS%Ljtymzu3LpE*%MWP-m;IVEscG
zYsh3@3@<9&J1-I4#9&XBmZ4h5Eg3TR+mVO$OEyKs9f^kkIigS#Qxt`ORb1U`Oe@B+
zgm)7yk^PXbr8JW^_0_Q*-YI@&DAOb9iZWv9>}s`dm^u$Z{AU0?*=fC`3+*QiJ^r=r
z00>*^@XWU}>EQ{<i{BTxqe2}|4ax8@3v98Q4=URULdso2>IyRVHHFH0t}xaQrt*Lt
z_gBuR8_Y1D58J&3cys_!u-Y?LQcKZDf@1wTVy$WpOuUI8=hX=oLRu$DS`_1C<<?qZ
zB~j68x6g@>$IUaM4WYhy+GPzC%Q=;(6yqY1YQwtuP)Of6n*B?4^{$reoo=``5kphh
zy~sK!A<&K08UM??HcHD>RT^!~>wY{6JKEwHlL0S8DRGVJ>Vv-F1U+t{01i)-@p15X
zSEYS#hJ}O=Qb80jKa*yo(l8V(ip+P4Kc#7t*AwI9(?Om$92Yf=BE+d1JRZy5WLo7+
z6u>f@zttS<nZsoUE-%*l2wb0+b4|5nMZ~Z*_Scs_Hv(;a+jkd9IFLdp_1qa$q4V>J
zT+7Qtvq(ge>#k?R+qu1|0D2)5u*#z15%*94?aO-gvNuohjb;JCKOPp2^%RnGiExJv
zBjR_6gyR|0Sh!Ew54Pwj*@n0kzq(pssw#_om-)MrEbU?$CFu*kb~ITT!%*w&myRNX
z@k^^t+Yc0r%r7QT;)Yv@l9^lE(<K#4&ur8;X-voIzH=Npt8^MS2wzxInL@)sw6?o$
zEc(M}-rf{CM<R~h#u-M1BgrfcXwn;fI&u4<cBDoG*)JYeH-rO5yK2mzrLJ6+0t>GX
zpT;v!{_A$U8X&dx1um^AMc}Q|VMxeIlLA6A6t#&V@(<iske;ooGuoxC>pT=M*7!R<
zoNJz+WLW<wHvf4-w(E2#jl17MLuAiD^2})fGfXTdaVjfW&4$Cbsg(P^xuQLe0Lhfq
z+yRF8i^jE3fB&}l&7`k(jm0X(__O17>iNN*{K~57%R2We1GNhIWn|wku^giDUkDvc
zLmgFM-Ni*|g}~l`ps2Qr*XU(SEXP#dwd$KZNy}aZA^<*`-I`T~hh675N~kP1(_tcx
zkB1yQYa#h<?AcB1=ZdktqyybL?H74{2&G>aYOKAFV~gRrq2h!PSfZljl5u(Y%};B~
z<LE<wkqDN2*>`1CRSVhzUol9{p?<#hE#5^VtDR3#p9fQCs;b~p@WMe<mf6vjPt4p)
zzN4COC9_6sAiH{lD`|ImuO4zXzr~5$(&O^AMhj(Gz;*=bWVnB`H=d{>!&}fLTFi%G
zm>BF0Il)H^?cQ=fVw&s=Gkg5`iMEGt_YnCTFz$^A=Db@lv9Iw)KI)LhmkM&}*Dnx#
z`kT#%)?wqaL9YidU7t>{xT-DV#W9>7PwsW|ILxEnD^p}c4~sM|x~6Xq!|WFiyvZb(
z(0u#W5%6>DJpD^Jozwuo{_5dh!?4paOXk$frj^wWJN;*WzZB6rYG;Q1s==!>Tvch$
zgLn@H2r%P7#)FCel4R6gvfjUuu4Ek>H+UdU`ASfkvSu|^usCx)Pk<&IWO{}RU|yQc
z|CW@qcW0?*-B{jp_6o0N%|?^_9fjY6{%+y)tsWeJqHLOJH0TZWjrn{u*qm*66&K4m
zzi=Hf#e_f%flptHga&zJ$1`N&1btQ~_&ZAxmfF3id&V+7m|}W25Y@0(P9z2mIrbMX
zmbSvqFjUGpeFXnJITSkN&!u7}WP$7QM~<STB$tH>!+^dEMs@Z5!+ipQ)W)aPDrp&+
z1Y2+=;bNxU0gH)N`mbM!aw!p%l`Q=7dyZUfGSbHzYP5fOkv%X7sn#GL)T(Xwo8F~s
zbagE5HKI?&X2kNMt`7>HmP{sbBphzvv8TU4iD}BYcFHa_Vdq`i`ygHT_4^O6!B7>X
z9g>(Ng(&}B*y^rLg4dOSO6rQl(<@P4AH@#UoHD$QFQ&KIuKt&7Q4msWerU<JwC(>n
zeKUs3T`I;U;ZHDSS)`U|zb-3g@^<Q<#BD_&oK<gne?3BVbj+yJaCa>=*-?Z@1!e#(
z3j`>;^?y;m{*ZoTchilW@-26uF#~YOu}x%Hga3L;@Lrc6n-pOW>A#qicKKf5Xc4TV
z1$MwaP-r#QxDpkP^i5c{nX|U5Qyr^9yg&6kiSD(0beNsT_d#x;&KLG!u=2E17Byd|
zAoO5^ahN%4B1!FiM`cpfZ5vt0b^LmfvCTK5pUS?^EOw;MVU)&%qV#RsH_7WQz37Y@
ziv*2(z~X_R;ror_bL<)@1+sc;=|bj)+1WKLnKUOj{=mjuix85U4xgk^Yt$`A!qAM-
z|2>Y6&mH|~E$kC~n~ckA%$ZtOq1g2eHAPJ64jC2!_RGae$56dy-RI*bvaeRHq!d?@
zs`baz0!tTBB6;bcS>l7>*C}dhXVCu82>*vEDb=83NV*YcE8+2~cRIt%mILw*S2M)U
z%g!5#zPt*uGGlABSW^wGRY^(99`9dQVsA4esR0IUq^kaIF1Eo#q1za5jE=(W&GMI)
z{)$A-_Aenmw&Fi7ep7Kp&)XR+usGvFkTKnUr*+45IW728m79A*69QBxn5bx!3SZ`;
zmc?I_U|_xY4nHg^P+6w7W(5T%;&Cc|zo0~C)DUOx@yqpmORv>i!iHG{rP?9F<n349
zQN!<c>`MxS<D=vbwI=PGyP4lr8wep8LMeR9cXtp%1&KG?$`1|N<1*|i(S@sD>xyXb
znnJ)bM7{@%&10j}%j4OkaIY*?MRyuC6I*5s5q5GMeH2AtzNt`*84M;1s0-W*LO@Y^
z+t*-agz<&ggdehi^?>!x`hF32mXl-HUkpRS%4*@bisg_28)n9=VDo)rX4B+I3pIwH
z0$NM+*~Xi|incL_w*bMjaFD&yJaXpF?x=}%7KySpLIhC(#TV}!utx^MFz7F?-Df#V
zzL)|;E+6<56a#+hCB1kp_>V2pmNjjMpWD`hz`&AMY{1h5V$nhL1$h}oxylN}Kfcq<
z2$>D4kePoC3>+Vnto8lg?$m%Cd%TlG4>8GQQ}qjIQ@Ro)?z}rMjaPST@xyBoEIdl7
z=d&gaT%K6-qWlsrg9JqLHE<A&$)Hq<uX`$^>3Koi=%w>*2UpMyt&lfh-}iBu76ia9
zdC1%W7@g00<`<XQaINeTU;j|9`xGnnWP6ZeM5?F@CMF8mX;=B#!GKBg)fGA2ysy`v
zbd=ge3=N&pqgtr3nhFh@989p<U;J?<iZI*vI%g|VXJ}I8v@m0#!(ss)h!hd^_1n+=
zlZ3G!T5nNv-{o1Vi)_b<`?^0`32ShLB#hNb%QMJW>X4om$w=UnRKX0y51=FpG?7^A
zm9A*v65JPZhGn49a>X=SGJ91<4H5ASe>G{8(}ody&N8R?YZs<wdKRfbV|$Tyk#_as
z9#`27&ZA<UansNUop&!?Je`0YrNg%s+f~3skl#fNO@#miO+Va^cTNvCdTI5;eJa^d
znd<g_t^K;K2thhiPj6yKNX0tiPQxdd=2{b&%Qxawtsh@YXFZ{&rW5*yEkQ*DXr1W2
zYeW~S@$Nqi_7?`H^t;plfMvl5p4VRCv5z<?cnAEW7ylE%`k{rLg&tq%Krek6%Bo;0
zX|5ArGfIeK0Xxfz#>^4tNAk{asalD&w=r9QVGMJfypga#q$dDx^YPt@x(ic$1FA#c
zh`_}Q%+cM5K*dYcM%fIn>kmte$Hv!wA#8#1ZZ**T2~-0|@iAf3r&qS6%_$9|;sy?r
z`8x^(n+kF@D6t<MRzCtfmwNzlr2YBd3pdB_jD$^q>9ME7!?8EF1>s+of0}h<1_TTg
zOEP^Imkpgxc?*LJAgCP|C5BmIOAW~FiS&_Hp#w-Vmm}3vT-?Y*Vk6EeLbf$7w6)aH
z)?&d1^^?nmNM7qy_9KM!pGF^U@>2b2XAu}1C$i=<a^z$w!mEFJB0SPjdmkjhQGVH`
zpcsy9X)gXXHF+Kw5_MJKbCI$%@|^LrQOeb~2H8(Me~H*=D#dT_s9TT8U+Sk%2H9IK
z_+74`r8rZfnlQ)#>Kw5dOa@5Yh&I)IErFcX<?|+W!Uy~b`=nH@KAxip*6(vLyEH#4
z9urKn+&)kyiF$(jhI#e_O>};ry@MI~s25Q@k|WU$0Cg*PE*qy%F@pPhoN?oa@<@q1
zu^KMssFUe*C3@zHX^uOIRIb-bsT>CA#|~x2_c7a~io%fQUQgM;l^ov<v?JF^vS><N
zU~c7cEeseAlEPY@<}NtRAwYz#bJ^>Z_S54;jERO`&(6ha*fz}OiQ4>-{(AD*c(CzZ
zjTuGz^ps}5tZo7f{0(uAM`^V!Z$bAUbvdzKR%*h=%s&*k9<ZEeN<){pK6%MaC)>(C
zr_MYiYJpM41A&7@jvZRI_tN#p5B=NuMHF5XAm@*QM<5F7*imfp#Aqyl(QTrpeG}o&
zqk}V~hsizMApyl2+ZU;Aly;Yc%PhZQL9-<mmaezAc;t`gSf<@?xCqdITH3a{fCdd2
znc6*CcBJOW@oJtYYkOxksrzuU7J4(t@T<kB3pAhG?5FP|N^~+@mAQXmOWo93v@Cdu
zu>In-V_NGgcHNMxx{KRRIVcsnFYJC+SZ=4_62yJW9^wRPuB~-xy$I<%p|>Hi-WWf$
zaVb{<$;uedFpqQ&I&VX&&j#32Jtn&B-(~K7Q%N}TO0kF7>t{fEw2BU6UcdFLvGxJp
zsXV3{*L%L0KjmZS%WN|}8BYk%sp5m4p5=0swz5__e9_)5H0H0^H#f4Zlqv$Nx3?i`
zKwRYIV#=o``9Mf`fQ$eOOQ6H({^g<yO_3yXVI*UWYvY8BL=vNI{`PtBS3qb?l5JI~
zaaY6_xn`yQI9ikla^^hww?-J8g=WTh|4TL1w%y{kZ<g-}b1sXMlE)Q}zI0Tv>^4l%
zgxOe%r;9&T#@ya#9VnPMQGY+o!(rOd*w-ZvA+h~002V>%zO3D8VFrN?XP0@u-D7n~
zIb{~NSNu)G0I3;!MFtTI@5hAKYFgfZ>y6Joe!oZ0-krO4OX}1mHa<QiB+O_sX*3#F
zPHiyD&PamBuZx*LCyyUhOwg#C;+xsE%fyNsKNnQmmK@CfA@zb@BYb@8z?%lfYt-Cg
zD!xM%&nj0RT}NWT9E5dju)~qn*uMmog>ASd7sobW=x?Ak)M?b{(Q^_~e{A+>A#q5V
ziB3ZsjN-0G!5QQnM|K)6q+K~)*<P~edyKrx`8lpHOd+T)X^VLsR{Z=jZhQ^?VRhQa
zJpgk#GWWK{8dOwN001)cvD}IYZ*3{4v;j5lvXacJkL9gQ%MCPYzqxzJxNea;O;fdl
zE&&29Ka%UV4EOD-BLl(Wjd-RJBMk0B!2)6Ooq1apMnwcDDYw8tVWm9&)?9bdCPBc+
z5k^RiR=l8ED<n)EW(W-uipykXR8{j|Un~CLnNZ;(=U&AEY6p&;aV+0pE~}L0PBabZ
ztrHjkyF?Vp#7Zu48B9(e;|8l_$wbe-SLQbA&KP54!5I<V1_*?Howa?sYLA?<%M4k@
z<a-MauZ-y&r4g7xU?Q{3^Pes{a@y|BsrJ@mA>MmFGcaj{;hDFbPi|&$wao0LY&=(D
zRw{}E5Crs$ZHAMD@~mFd4G&Gk`y1ep0f3p!)zt^~?MdCc+ii9hM6p9`TyoDoUAlKq
z>eMAVCdO#;6GhQ|ffpO7*_9OTn=d|h&lqe%unFa6R9N7UK?IiF3fd{B39w4?9Sion
zw<UeW1EYeB+BSK{Rjzzg@vL$+kvstd=O8eu!4AhaV)wW0BbycM793fP{<G`1_nqEs
z@f@Cnd9OFor?L<^Cy;vr1*egFvU%fQVWh3az#D3I8B%6o$Cqt1b#*CLyn;C|L!+zH
zKILy%@(l7%x6v|GRIYBAkzZjQ^3M%cyR3MG)~Hp6%nU-g#XjTx?ISyft$1LhNhdZd
zsFE41xaBrwX(Pg>Qw2&jCua2Tp(VSQ=1~qe(w}b&cza=mBtd4RpL4{Fy70(d0XI+c
z1MuDI>epOWO#q^RhkoN99Z24nYq~@PBAp)drkFnXwi1BM_;OkGzy22DZ=~81QL&I$
zTIHGZ3%2gJ3Id4m`ezlx`sg-(5v?auY>kCg(%YX^ELm@sWWIT--=FRc3=koDlA6rS
zgoA0;vP#+2f*WV}!NF}B&MxuGe-57YV1e?@?%*DivTvMZdgxaFupnXcUh6aOl@*sd
z{mO~ZF-Ft3uil$|69s|LzneDDUnr}TW#*D<+wt6r<miBU4=TG9soQujGrzYbT@=V6
zvp^Wp8DXaS>9re28RxCvKoIcuTBmC|J00t5kR|EDxvcE-=hv>PDZ^<p`6YGk(yMPu
zkDk3dCUpu42^B?+%reW8^7AnB(~tZ)^NcfHt46>V_hQB%5KvNrgXgf}I5r)}sXSEM
zV6nsQOcCeIC$}%FG~wwL=`T!;Xjb|(6_o}ce(#3%UW1BC>(l)j47$-nII){EyRiFP
z6ki4fiv7cI$A=!;*S>{QTbnUXk8v3PVEx9*Lgc5T=qw5|ke}|GBGp$618>04Tb!S*
zRakH@9Bo)Ns?$(ReWZS)wF--$M9GD=TZT%zbJr(L#5{lXhnANaS7{gq2EMxf;Q2B|
z;?^#bBoN#ch8WNY6kyPvcy4m2pUxYzp3KaQ;wszTtP+RJy<-C7Lrnsq5$zrS^ok3m
zW@Q-*|1W&*HAc}ps~I!w8u<0il4ZDqdj$bXZDIEZ00Kb}P@bA#Xj}a#%Y3d#>NGLi
zy}ZyMp>TgZRXI~nD}t-c%w48rmsQD9Ez7V*gYF%*hnB=Zz<(B2Jos{nC^&VRKw!q2
z4<p77F^Cs=bTh-k2pW{-a5z1FV<N<JYhwvMNN43e%*-0P^Rc3@S5~>NOBA7e7F^-b
z0El4FQ{k3agN{5ps5dyiQ5bSvPU<n6%n$|o_l+qJjOpZl(v*y4MOC)c^QBh1+%qO1
zG0aaSl9)puURG&yCIf%(R?MFOp_vZt%Ht}@$g9A}e;}(E%n+0x=%QNOwQ+^G3$k4E
zV~2%D^yt|uGCJmX+L2>vhif_m$sKUypCGGqSeXHaMg(+#2!{o~9LGanAuZeaCqM*|
zsAH(<#B)=;{5w)nX#~~sta7E(X()!?f{?fdJDk{zeLq}A^_D7&aB?fUjjrF`e|Gc4
zbNWui^+nPU6r`g(4`=t|!Vw?TD83Y;yCbR_qProwJN!a%d?Plz*PiRC$DtJ%a2>Qp
z0MPrP$2c5W)poNh+c!gQs?(_V!>nhq=rL56HgKAXii!`AnN!b|9=cGYoFhYp<?lu$
z_tINRrB8pXdg9-u45#vxn9D8p&d;qn^TO0nKV2glSQ+Y(C5PYNa>nYAl)T5xtkVeB
zr^GI}W01B1>vd_hoj{7>Na?2Y(4#?E>p(?X$Iz54o1GQS?-tDN526TJM$!IwgVQkB
zc8Wp}C?QJo{DVQyzFVfWlvT>3=H&eHbu{eGh*m+Qp*QE0RmxIbleoiyV;S~WKQ40s
zFTGy|z}-P&0GN1h!LoP4r;Rbzvp>rM{`hQ><o4`keEnJ&5qp`5+E)+}9Qgd@u!3S~
z^>&MM<u1Wkd$U8Uq1|6c>9x&J&dw<u+$BM?@2Cxc%)BkV=nrE%yN~{B{izpM9y8k{
z1`~*xwW2VtTf`T44X&^{$W{GoR4+ho@S$LU7%T|R9?XDiZwU|x(E&(*9Tvu+bZk6<
z4acxM1LbDd3gJ}L?(21iF>@UU`r@L(9lvaGA9datLVy7~p8{Ky`w2o#zyVtUKm<m2
z#eo-*TZ-7fLGiPYz!^o=Pb@p~>g?WX7H*YGM72DtT)D*d!SLG<*}1_EnR~JA^Y*mi
z#PmSUaS!bcd0a1k3b`j*dr`i`KO9kA5#1fp-4WH*Lx=y(Z%-%riy3}<jn8~pF24Fh
zW1SoN`7VX&kJV4U4WQ^O7C-6BD$^>LGpgmeow)V0-QVmxuQ*Ew`3q-ObqMk&!Rgcr
zyN15}q4I&3iXD<u5)&vU(2()cRE^;IZCqx~DYFcIZ9_q&RgxL5P-G=QR1hdYuUYxv
zs9{N=wa;QkR!Mczvn!QZ0{tm}bDUOehJsSWhFPEoo+?_ftV$r%n_6bvHq-B;SHtuY
zHxeY=dN8w%F?3$;k#xHv9Yqlnh8k9U5-HdLE$(`}=zq(cIWh?_0uI3La-<hU2n4&U
z%9P-gZmzEa(pPrq7+n)aD+ze@qq3*pE^~M2*Im2on`p6}t{or;`18e8_r6dJKondC
zQUC;5#-N_s6(2^11`0y+S12>T{&~d{|1MShCi=z(r#>-3NgLMb*;VICs+C7kNwwut
z%cR!`7Q3W8CmI1IAHYxa?2q7ljvSD}dOBjKlT<C{Zxc$>$Z7`xB3N;<XIZ&aS^0qo
z7NE+Cl42Y>j}6DL^9+vVz~=f96-6y`o%(R~AVQcQc0Y@R5D3(SyM$ttt`ZPEA7+~~
zLAOQ}4n8@tM^u0+pH(geRXnR)iG;*q<ef<9-(ZKrO#JdGvX8dmnrdMT!hq|LJP|r0
zD)RCDV;&5`v!6rkMKzz(cVP9u+GdR{$0?P&A*Opx@nnx=AH|Q)w$-w{!x)6${<r4q
z>Nk+S({s<se?^xOuvKHltG;}YG&)@WcMo~dvr@7AA3mE!<#I+9&+SC=i)&8gS19l9
zcV6rF*{fl&*Q7xPNFqL$CylrzFRw^)>*tiXp#6_exOg;X<W*Suy|TWb!s?FuRNmNY
zJ`)1;+NJY{kLeQb*53}9zuKL3&sX~-nFRrZdg^w37ftr&c^4}SX5y@3dFjgPMQf{1
zowWnhB|&q;MC0|-{o=y}feCHvv|M|63_4rj=sq=D*^B@n3h3Kix8aMZ(;4=Wx8~Vw
zN>DC30egT_pmPHSkDmf;z#Fck0wFO@bM&Ve9T78=jvo70>FfWkQ{63RM_iPtJq?gb
z40f@F*KRa_xxDJoNxRv~2{Gc75r$i4`S(oHhyYro?IoF4?Xp}uzd&XPL~$X;%$KH_
zZBnOaR~A&-YKI_G-A9pK3FKz>#gacl@VIPO_v)ED<j{doG6RqZw!mIS)?%SFT`b%q
z7VHv>_ma5;G9WOE1PBZnU}7*q1~h<0MyVa870Aj%Y8G~#!Tz(zDs~EcxD3m1+2kJ_
zhZk-{h~BHx({L6uLy3%ze?+y-c|Y$PmGs_C{Z;X-aw({8+bUNUrVtFCi(Zo(>`<JA
zJ>MgJM;orG7KQ-ynTGzeq0u{A0wNI{gWQY6^MRi{i0973;U9i^nQ8$9L?F5wqPiif
zE26r2b&Ac5!z;1>hgM!$^D&p<=sNVA==?fhPJ`k(x*GuKOql*SmOhVyGrn1X!-k(<
z!MuOg-*-spkMR#)US94hDpwX&HizOVO@u+cb*QCRl@B5#F;d*|P4uW+bF=d$#ntrS
z`Lfv`ZvXM#5%pik3AcT|r?9qiCg^n}5X@F*CNN@DSnc!vyDRPKN#TCZR7L_}dq$Bm
z1~a;K)S#J;yY2;pkq|8W@wR|J-s;ax00d$tg2d#Eb8a6EK}2Bg7^&I3FnYwiTqV3*
zl5ya;t@rfo06%go=P=+GAP5KqJ_PQotDgyY3wQvKfdJq+I3r-soO7(+W}Y?LR8+-x
zJXqj%kOm-%*!p=?q>)_Jg4+k>4ut82o2L8Co9;(4lnT`{L68}jU=k-FIZ5lvJIy6k
zHk(5p_}coyDx158GZ<Pe^g61naydk}J*Z{eJ*wfs8IZVYPO2Q#2P`O80st^Tt7mPL
z6p$>%PX~xuB6|^8@`Z|XLeU|yC{-*;rOFJFWFiOxGm%c9NF7Iparbac><tD;MyVYI
zRmdyG@pCwyi&8UUf-p1*1G=JvzY_M~ZHT_oEm*{Z6Y$0=C>Hlc2QIvMV~RoJadSXL
z#Vb_tta7Db48p)U=sTmq4%H?2<ujb#-iB+cMN~KRnuH$X>bIXe=sC^x+5IRv?;%vX
z&ou1#qE%MdEHrw=^+v(~B=kpMWHU}Qm*L>gIJ^>ebJLyMh5buv#B)R^#P-64G!NZU
zmtpkX0H8DBx)-qY1)rtGt}emymvO_(5bG8{?J*8zIoP+@H?ydysMN(F*XT0;0D&)h
zM^|!Lb&L?!eG)nHw!D&ZXFlweX}SOT<;?uio!s3S_;Od)#*+nZxf0+fEP5w=WFNgi
z*nPrw&*MdB&N-AsM@5ykyZ*O#`2!=Ba|NQkXG;{p&5W)cwJoltL`qt3XIf=f8>4Ni
zQZ<gE9*1_<ZuvTT%q_V}CNpMa7f766)9r6SxKe-$xCclB-f=l;6YvJ`2gTzTcog_A
zkOgP*u)lv$e)6p2wU5h-N@cgpx<<rjFNF<B(Gj<U!*)HKS+d;5U)KiBC1!6ULVSc6
z<S&#}%FHm^rKf*NJDFFhJc<Ir&qN>nJ#5+tgF#D2G9C9nQ?w_|=02T)f!!M1SYRSK
zq>CKf2Q1WCCsC~vMCa9MBUu|M1@w_(r`eUE9i@W#0##)R#mB_rBSJ}<P<fneRZwb8
z2>?;BmV<-2TQEj<0lS7I*}2(NXCCkFhQN4!4&L8@Dl3?|sM5ACt7Ldms2cgK(qdHc
ztn%weqeH)I(f@jA^$qqpunfDuZPVn*ZaNP`|LZ-6UhiFm-QP5L;{GL=@W=X%y(VGL
zckOc_07BxB&=2wbkkGH$=T;UWbt#Ung=BAqmHCeHJe=5!?qi)_Q)YP<&xcnac|5`s
z0YGcStf#T~X_TDz%^C_bu=-6*f2@A{(BHsPiL^DonMFlKr53d-H&;~19$&2${oJji
z=8IRuW<5}#lumX?{Fg-sXY`7W3o!~Vd+xv1pLCsqKtEyU57FJCoSC^t_0;eGF=ptE
zxyR4gm2y67kLPbaT{x<9m~wh{nMHA2B|_(p8c1!M4!f;T^z+D`+O(fz#@~^jUo3Im
zkXHc+2P%LdmtuMYm<g-~48R;91%OLSYk(hsAui+do+GxsX*QrH*xRV5|2z|N%S=D2
zwM%kEqR~@alvrLVGvJW<U+Yi0@2W<`&Tpc-chnLy0sT8^R(=*a{KnkV=N#@nVq->G
zvh{faUMc+ALCLN}WSU?(AQ%C?q?qY;W!u1EC3BWgaYiUTDHb0VN>5ScX|m>!L|`Vx
zQ33?WbP2iLj7G!+<5Zq=gtVOU5uL(XI+%)zDxOt-?Q|QBk$1ru)L@779oY4CJ0#QA
z=+J#EQf48zgXa$G-onX7vuB^!jG?!{KfF$(&eYafpaBF%p~pCM9o1N^uSO`&Lh3S{
z*yQ6?wI2I^s1eVJgAf#hvK$ZHb}ht=C!L@5{<!W1EPmQ^&lZrm2RpyQuv_c5kG&VR
zYMlDTH?ydysJJj{MMbhQ<79>%BC+HlBv_GrQ%4(L{%i0PZ<Z>xWbG0UdTqn`m#6DQ
zl9_j979Y+ocbc7n$L<N}7F}DLvx%_y-H4%cb1JHs0f#I<y!g=WN5?CNE%q8)bB7L^
zi*4Xn42o&Id#v`{x(+uzT=>%_vn1E(X9>6$_#-d|7zd;PNkD&KwCm-S0UJ;T<OAuz
z?z&Fh?s13^8!0aQSNNzDJvkF)U7<G=AR<|$*hukcx^th}UGA6&-@hHvJwYSXq$1}K
zJ$?69ME~n^T&H8{PTYHulN^*0Bsdc1wA|gDYUy%Wu?rj6ud8J0D#c8-DL&Z@!JJLy
zC&kh<q4=;+dPb-^OXl;CtPBFNdva>5`c+*V2=l`!=i|Y7<(4b<=T~VHRPn6xt0f=;
z*UU$JN`oDW&SJ~QzR!o;Bw9UsPeZ?14N7{46)RrD`9n<)wSNi5{Jwr;zgcZ3p1Y4l
z&xwdhZsF;twjphGyCr5*2#my_xd5>1Yj5jXmuA<g7M3y-TR!&C?d*PJrPk!7GX~-M
zm$2k%pCxnJzXYN2=r*Q)`=md^Zbs%l-^`+-aupym3q+F40x21b2p1W$lEVE?7gQ-R
z&pQs<AOa2eOA`?!e)O)uU-nv8Y&FX=04}b!{ptII@679`5$LPk=M)<%5h6l`M}HSY
za*ZP_5s4zYC1`#-$N%jwoCz-uT`0>dw?qaQh+wZ(peZI?^!@F)%n*nL!Vm97eEW0t
zlW&%0W;@`Lz6@A^b-)_e?IXYs2mnOD4pacuu5m8bQbf>cY2H-R2hWBYH6R9a`{YP#
zOGx6F2(h+(U3nBc$7!aHG77c!Vw9brKAp6EyJ?S{w#zaIcy}Xy?@2rpg9RPy{2}?`
z4mvY^lRKt>WCjUfW<v+o$H<Y_u;4mmtq{tO3nd4{qC-N-F`+V>Y`JVNQ@q|AUyL)k
zeXUN7B=lBM@di~qtNc0{cmoFB08IncFRM$jYaxzpXxlXb8a?_>N55-f2xzcRc|KOY
zhN81x4R~ZV2Hyl@V4cRmC?pL*#_mg6YU3CK(RU_#O=|2y8o_Qx+FBf1(ayQe;!}`3
z9$lP~gb*}X_m;Qk?)kPxJolc0T?=8Y^3ZL|$GGLKn&4zpFlN7iCC{KD-$UmXu<mUH
zMmv*})n@t=`0?3x-jz^Mx$IG1k`JCQ{cOkCO{WS9E3F2tm>eBAwP)10ZV_D~{B@$B
zyt>QG^9IDNJ(f=ZGxi^|ZP;NMH$dM=K5CZulUKrePtPtYlbIngfB4Ip-wx|II5Bij
zRtY(k8iw`J3u4`3N=i0t2Oj%f;D_H;TCB_rhs-aoJpRtS6lT~ZcIPz?4Y=a9p3@ng
zY?wR2c++&#>mQfD|4rq&T!%aGTe*3zBv_~fDqZJ#)Io}iuU1Qgdg<POHl%lV9ogAE
zSL<HP<$#RX$ohLO%=qsA!ZjD|9S{+LpT0NnrbmhZ$c(fLD6t__TR$|OiKOV^kXtQc
z{$e@JN~kkK6Cj5SltTtOx`?beh)66J?J?|lP$)@d$W1IsgQ6<H6+9mqpl{D)aVo!l
z)O2$yzcv#3<JNaD^p*yd^O4n9_)t42=fuId?OhDJwK2h&xfkF5&0FQHaA28-#(uM!
zy*w`oOTabv;kTb+;GC8y=ls*y^Z`D*2iyPGZpt|(8KdvUogZP=Gd0Q?K<R~M3{sYh
zlUr)OHk632JRdt3)@kt%!*$O$QEI4J{QM8kQhdf!2u<+WEGjBj91K)h9pnDJdB7{{
z-`kR&mQ$WvZp|#JUUfA0Pv0Nx^Y^t$&#e4%PgYsAU9n65c35JtN$Zvte|xn=(h)%H
z$xx-_;)?X6pFfCTm(V3(+*@19&Gzh4XF^1QaLq7-Z1td61c9O=#o6PG$`P6Qht%@|
zQE9bZa-|vb_oG1LRq>vt+#i7e8$Wz!K*q+{<?lxhPtgbY3q%kEm)Y5wM7-vIsMW6Y
z6a++tiMPx!rLX9)_3P-~Nm{{CGsRVb#2uo<dRME_pjS_w$G~A?3?F2u@p1<WD?M&Y
zkgexmNtW6&ih!W(g%AW0f(U>d&Io8ZbWqiGd!(2NU~0sc%nVsZUb%bK)IBP|;~%w(
zikGOCXO&+S6Cd*6#p<T$EH-_JJWt81TNS@hjQu?l2RG4e-(u`q*hHU;IkE}^u7}>Q
zPGd|mB0Bl3Jd#59vFI}mkzHDHuFZ^7TXATG&oX8;7g38&!;mx>3H_lndgze46nhsn
zZ_Is*YaX^f(~$Zj95xT#4*rC$qv|At3y8$br?KQI*nN@L&02+JFJbOK>tB~vk87XB
zqQ@`0(q$?tmjMh^TOHAl|7??F2ISiHZCx3p@+z!%{ckUTe_r3`fiazoT5(QFhyU)#
zRE*DuPuU-RwdBo*gBwazq}WRb_SXOT_JDW3s!&|5Eq3{-WoZ_N>?&1@9-TbDXfZRa
zm^0P%-D;=babAUWTY6ELzfN*_+y?s#dTj&N)hiKKg)N3fo;*}PZA=u4*qdto<cF%Y
z+s(ygY_+o_L2{M}b0(z{1=4Du(^71tICqNa?(6*GJ8R`C88Tp%1%Yr?DQ7@t1c!<>
zb({fFz-=@9L>rSQvoSM_K{b0c3{+Nv>Q+l+fMlH`erEfpTXUu<CqRHi)l*jo{4a{6
z!iN4%WXA53D63Y|u#-tA4(S+jRXq<XK1UVLD!&pmI!yWtk{S@G<}&PDh-2&9c1?gr
zhk-Ys-?jCt3f04A#=5tWxu+#l*ey7`5`%86-`?+9to?UO2CRqP4?QO#Wk&tN!d`NE
zJ5F!InO#0uNeeIqBd!k;`Xgy*gSpo8Pve(QJ=X(koRafM--*t{0KgD{UXyWfx#yl6
z-@{FB)@cb(#MDRe(@V{s)FWj%Sn&$3d$E4IDHyY##}7}ya#5Xd6_u+S0;atCi%pUp
zN<=b45CDcCKxSuEBqh8Vz~et2et7YrW%EaTeCxoahjNOlY+#ld|NdXa4U<g6dh0~N
zv-%<;LB`7uhrGM6LXq4Ies4>9E!l$b5YeOhCL#u3H^vwoAe2_fN>-=qKHR?KzTpCq
z%uLF;0zqnTxze^ct7?m2JHw6vJ#_CssS^#M&=VMPrCeFX<yErT%6>*6$d61Wq5uIh
zU}cAu1p=y=7#@Y<RYGWB9ft)XeDXqQBln|T*<;C5v_I>EaZvz5d!lS#NhC!L5HGCp
zD2J6ltN(0-QcIi&BRhwgbegO8$x!hHs(4oUl`#8xL|qj5sc?KFwten7B(+ruiN%zM
z5gOk_x1ubpd==%F5CnVZXY{+azMFd2QP}wvDhgUYY(NAC&+|}vx=Ab*ICl^kyKy$P
zeaba=9Egqs5Z@O;F^zSrEX1}2NZ--&iBf-bie~`*XE!LGi?VR&=lZVQ@hO=67p(ce
zmd{?B?4#K59>zadzda}#*FK9SPy5Eot)g;8z`&-H1>4RP)o#3bs%g@2Lr$Ue!&>vc
zqc*pV6MzIJ!tDQS?-*u!X;!cMzuB*xLWF7e<rn=DM+~9C;Ab`f8LMOCC)Nap)E2uH
zRSl{Q0<=1M?CyZa|6QhhFRr$4IbG;J92z9BfEL$z{<Y&&(aK%7gKaVZ0{s2R-%kVp
zcBX^ka#`uGad|6WwWdX8CLpAiC7B3`u^LfFZ1=niLl^DO2p7i+u;7TFM0UBh&$g0%
zO_pW%-EQmKPDCurrVfo{)f0?Zn+C;RF3UVQIqGVCG*o<sDxOu^35_0M3CKI;Rp&9k
z^Q@f9^Re-LTsY#}rFlj2c#OUWnuaS&E2Qth`gc6L`+AS10>{>&&-D83{bph7r!5&U
zECKzmYs>-OBNx(e{vggBM1FcpCcQ*V!AKg6jsp;v(qOo|8(TH@{eaY^mpW<wX=EQo
zhn@gn3PJa=II-Dt&s|@m%gFkn&Rs@Aa$x;Cm%0pv;~Npy5h*k4w}&TT%EMUx58uq9
z;zMRWh2Yp`pH(kG;Kf|~T(i@~vV^ZLDjI*~!pskTo2n$%R!mR(m&f#pcfRmEc6VU8
zgl*d`5B#I}T&_fLzE~3>tJwUPMF&NJ9FntURh5~Oret5(9>>+P$3)GF^(!hum_N_|
z<3*KHJ*7-Ib=EPqpXdKl5RokLoi_wL`Bs_CPQ&$c%a1B9<^%-#6JJ7k+bfJ3g;vwl
zPD{CrAtGq7WM#Gc9|?i5?45Ag#zMoUXG=xRKPd$G`JsM+G?H!I?+3VEvLqtDaN$B+
zhZuuSQ*DzK)!lXynwbSbEIo!sqMq$2sRop+1Vp1cg(*1>+GvF;S0bu-R%s7RA(;9w
zqPwFw3*Y>;iOza|^q%UWJ!3c4z6D2{8{7;581s8{8PRmlEg$2^YTrzny3B)k?l}p2
zzHhe6ZjW&u444;VtwLcYiq4`q3xyf&BbWii^gzb}Na&C7B(M6L%dme5j;{5YB{p@b
zQ#=FcKf6KkY`0+ZN0|0#{r2u-VK-yTC%#$Cw$BmV0UZa_Z%^otX^&#%KYcTcim&hw
zLtv!uhYg5m_9=Y%xLUS4%s5k6bv(BM<lJO<^O4{_@mja(n3;n_OdDf3Jt}tR0qac<
z7v>a7a9W|+;*e@@O)w*;Q0hAV!qK&{8f%04Sp&THV(1q?RaKemyu`~gw(YgvcZ0vj
zlF?9d|Aq*U-W~Ybr{&7L4p&Sm115t2zIu(QD$RwNnNe8I?|oDAV$dm0liWkcmImb|
z6Hc6{+2Us)#IMfA1cM+1Nl`;ds`Dsn1w8%2D;qa$5k)aRK5p)u*|V>m?%H@_w2g`u
zAe3aFc7%5=WX7rku(>kEh5G6G#Rsd(S>+N@#j{GA(_<V)-R;bN6q<m*D3s?l*55zW
zLt|Aj*1y{h%DMA!Tr<CZ`mT#9ID>U>`!ai$xeSGw2v4ljC~DDn26lhbyrGgOcy6yQ
zMcygookCG23NyVeiQF2PLXgl89S0z;Z_~HhP04v2{0T>W9ru2AAM#H-^QeU;Ahr)K
z9QNEZV>dE(BWXze_TE!bQQ!kJxWei;aQ)vA(W!oW(hy&IiL0pmY9T^Sxdo_sm;UW<
z0)#pp2u?E;kLs_>S{M7}^2+;PEU`IQmOZ?IGb8=HW5k@CU%rplIT|Pw%*+<7`#5s+
zoq3L0Ue-j|eb~x6g1I4&S%Sn5+#c}if6Ff#69&C%TUBWWM8q0A|89wU6GtHY_oXn&
z%whv(XJ%%?2j5kQ0wftkhzUf5q27bXvVSZ{sQ2d~F?uF<|7F{b%F3$D%(Jh(@z(nv
zesaV0v*+J^yDX4MkeD5^(>p_!nFxX)__WkE3l-<dQRp$>HZugmn`@k337DK5skw?H
z;Z=NvDxOu^mTTst=fs+?MQ58(&HyS3;jq?skoF73?B}ucdDyCbzsQCd^LzB3(sZ|7
zUt`}7KASWs8r{aA*JRJtkCZIXc1tq`Jh2%uJ>ak*|1=~!%5qU~2H9yS%k|ZgTS!DF
zBn?L5;QD(6O;VhNU0)-kMfupgVaHdP{haf2|Je<SX8@c2gXkU}+-`^8j*3E@-sXeF
zIBfXoCEV}|f}-mm@~xM+ipsARyQ9vVv_>zmO}^M?)frL?zH_$Ulxs|X`$zFNtEvGH
zPVq|R=Nli$|L&a#1JlKC|3oB!hYZk97-m?z!y?NyC#Iis?A~V`l&ozyY8tU&;&bzZ
zUi`4Ej?o*?s)>zMnu9EJX0G(VWmWDv;r)|!qlfC*+E}1Rv6+8ZXI36s0&W`)66*Rv
zv*1XWNe&rQ$7h>CH_e?hd)D-`85wIfZe6~7Wkp5hXA8dExoc0ao;`N%-rK)l%0u(-
zF$M%yS61xZv#)z{_t*|G4sKuvswJ@Y<cCb$&{(n_+fTa!o|&ijidN0eDwhuNdpETA
z8dR<hjDeW;7^1t^v{V#e(c>_eHGOWk(U|xr5A6kK+DJe9g<{&HO_y=X*o_@uqBPq#
zlLmLdh&zxtsEKad7U1Ad%^TSig6fhMugk~8^+n=94*|p8aPA<EtZGHs>}Jl(h;~KY
zfAd!qXEoR}J_WO$@qB2>)5tmQoAm@lVD3LX7ves=5?lZ4n^{yYk(=MVTxxN%j1T6u
z_afQ|2KJmS9`^c1xT-~6*b*BPETVz;b)^q8pv8}Csvmf%q@Y;xSdK>^+%nzt>C53-
z02e7?rFc@gos-62C@rrUfC%FT>(?)cW_yFL_{=D%mg7gCSNu!eC(fVi|L*glf}?@c
zR8(rg0JJpo-n^B+Sd?8Dtp=MHL=EYw6YE!fcAdh^0=@fX#h?FK;@-GPdj&lrP<yY5
zS**V9pd8o(R4+3Zy9=Rf1XNX97A;x!pASFLXf(2G11T~h;@02X^!fk3Dkv<_Yjv|`
zPMd%4-F|+4GJ^;Nh+qOFA|+qIMSrBqvOK5KW|90Cb*0MFo>TcT%Iye!05U7}qH+A$
zDP1G{Rr#!PiHOh7=-G5H6_raQJPC9DfsnX5jVoXC*0H%L3)U(m^snFU7lQZ{oZ1Sx
zHD4~Gx?=Y8O(g!YR$|TnVQ=#zkUbLI0i*85HS-bNp-#Ka?4c%BSVFVKbF(bftwiEL
z448un525EoM0JDK(2R2{3vu8lta}?rR{LzLqH(H<YVxb;{c&bzgFVahpflAd?b>u1
zfwTL(+AV4lTQx2mLAS9GE=n>Q)fHL;&K>l@EGm~s?<p__wf7<v7<<2O?=`d)4817)
z--c7lOSeEcdB!n!x}VTM1SUCO%$Z~)C+Y6I!T(f-{Y0k2gKIXBcG_N2Doq+?R19#n
z2UG^pizJ9xx3kW_vFCE7;eGXq9Yks{3Y3jH3J(@nZL_#F3=szP)=j_0NLNwBx=IUS
zW|^^Zx8?cw%bYf3!n9GwNA3+08ci>Fu0Wc3e}Vgy*kC+=Jv6Av7RDgfM_Y%yLQJUt
z^tx+klg!X)H2wPaDlREKcKn3VXuSERIcLsfmX?)m+p)`PwHovWo6UCeRQh*|mY7VY
zu3b85HJZgse)@XRlJ%Rno;rOdD8Mf~EX=Kw6-_P@1ZTmii!PY}f%Lnc1UWqZQOUqJ
zJF)D5v%+&ii0RqsJsWagS5fgERXnS-K1qWy^GWFa>ojis02#ZRF-YDikWiDl$&FtK
z;`-q9w${?meWqjTBhVTf>yx$yEB}eY3?EFF&>y3Jhf%*pSjYN@3eMoiXAqu<z^FRy
zIwPuzP~eN%+kg%|F<=fRK7{1)2v2IFKK8}rp2YSA*!TgmkG5%*tg>7r4uW3@0E8vr
z#OB6Iqn$f|uA@Adg+&cI55wv0zVVf)DnVfex{da5D7rg{a3ReHv#4AOs(5Y>h+x!-
z$MY(W<yHV-#;Ghvug=<DU33i@p8=2)7`3>0mS1pyux-E9X0P9kfxT(A@(OvvD5C(X
zXMa{k7%@b@WKDH`agF4a8Ot}D=id>ir~0SVoXFI>m+phFDlN8}Xw?D9x*6k)Xdu~`
zib@Sc(9wju@=MB{1-(p0`tkjU5QAIa)<Kb`)iUNjRCqSqS*eN$KmHk=LY0gFHJ#<~
zA@;6!5aS~8Tr;^Mp+Pie<d6yD#@;{wcS8mZnlWWUZhnE?X7AIx=l{L&N|&U>Lr0EQ
zRaNcUy=T!6KOQ`s_T8ezSy@?`nOTPpA6~Nb=NZ!`n@lDpMjAkR`sv4?dU4~HZKFpF
z)o8W#E`(SXOOESKybVuo+spvru1}C(0RVzPcMVIN+A~W1Q&PEfRPn6R$_%*~WABAf
zH@g1da_nEy?4d3kfhib~o$I&zhqab|9(^|kH|pznYy;N*A5Lt#)M8WsIu5{;zoFl?
z9%5Mm99@qUufS$TWub@4%wZjI=;upaK&uiGhki3L{z3GefymBH*P5z_WXFk3*z_Ol
zTZH0st+T|Ys4T28J~QLofyO%LovIVt??!7td|w>j2-z2vTg!4#UE<lXze6uLY{>PI
zm$*vHP{ngQK}7xHg5KVE+HG?CVQqCvvbKAY4&EdImNB%ie$FJ*@=cbCYPntrGy}U2
z+e*viNuxdJXJ(e|=+#5JaCuejvQ;)4f4!{g!8-#*_N<=))&OIN7(V@>%5@bm8R)jz
z{-Ua#RT_nvCD7kqD_Ogv#&hBRn*)9`!%qM;pq$Ar&|m&uvUt6jT>;Wb;duRK2=z3w
zEZBSfQ4SkYFW~uNh``JY3J3@k1c3p)!8mFBnA_%FH*vx^ox#+(OXu6?U2k_dPMkbt
zv)Qw=b7WZ_f6dt5y?a`1b|RX4!)$}m=uAc++o#RGsjRFlJ11xT#w}AOj??Q6b+3hW
z2c8y+Q^`}FZ3e0dfBw?FF#P@6-krj|PC=@o(lDxcR%tbiftc~Q2c?&N6l+_W6Zq^t
z1b0B#MPY=Bem?b!Z#0R*)JHrlUcqk0;gwkPCQfegfuD9le@y%n`d`;z(C?OyYf@@g
z6d=AI{KM*W(EFkIoG+7m`-*@F^qz{bzsHc95R(j@so5hGoyFcoSob#4ceHbyc3Cbu
z4|8gv(0CkP+1O@HRWX<m+uOq-V-SL4k-ptGt137HSwdVN4~OFW!cu|!)4rKVr3I+s
zx%~(>X|<xT<#eH2OMJJgIwC~q*GI2}Iy7#A26I@TF#nE#9s8_qyPO*a_8hU5mdlez
z8~>lZ^A3-y+S>TL_nDSSAqgQNB!mbA2tA;55b3?B6af(xKM?^ztbkMnMHB@K7Ovil
zqM~5vU1~sj6$pV4LP!s!mzg<deSgfPP0pmwK)~~ReD0I%y=U(=<{a-?XRY-vR@}2?
zjB(ED)m2@*YP@vaSo9;A+4P_FnJ<oY)i4<Y%;);~s{Y(#iCPm79WU1N6<Rk@D-Ej@
zY(Yc>+)1UeZ>5MLV9vhw5w?9*i(49)GpqFgav5HocW3p6tOCzjqeA>=5b}GTTNVRg
zEU)7#sB72?I~T<wWnYlU0Y+d(2~?2a&4rMV))Su}LjV!pm@#eq*ijujb{PHCkY}G6
zq0;DJ>76cx{eAM_p)evMqRh-JjaK_;o0f(7+e9Rg(}6J{gDtaSlkxeU{IA2Xct~>>
ztxCC#6snFQ7^dJDT)i=TAzZwy$FuHW-J4Y{`yB8t0^7=r$6dkZMRw?4rFFsJIr8#g
znaMb_377ZSA=C`Ow+T9ogKra?3-YqCbur?@iw(+3QkAD;_B}(RN5jr&{CH_!t#nLz
zxON2p>_);>yINK?iVD}AScenq%2@x@68HqkPaXXUXF}~}ShtDtz0d*dUxiD1>~@es
zg-`^;?kf^e=TG*ZioR=-i1J;6ew}q2zVcQ}HaqteV2;i)0Zadr{_gVBk`iK^W9k^!
z<?nhh(fWf#NY0?TJz}g9zs=o%Zv5pJ;K}ndU9C<8(=vUlTe*4Wv{QiL#_l?Gy@ZnM
z6%+svKsr{Vb#!KSe!@RT=$EwycGSsUV-mq2q3gJW)0Yheua-E&=GV}>DO9|ek4#{s
z56T>Qg^4zEwkvEf5dcY&)M~+Gk_!Coq7vLhBnYfymtKM(FvfCna#nu+al6MJEBrb{
zU^%X<T}ds;nk51sZo#~b5G4qlHS%`3^j;sk*sq|VIL`_Wz_Tuf&X*6l{PB)`LSaJ}
zz^|Fic-8%9L5(Lrkm;kFnT!+baQUyw?_3nmy67}sMgpxUDbd*eA<`3NrbaB4v2gRE
z`!5p}5P~L8pkb%V?+|?!5r+}^cePM8E6KR8;N}eg#vJ@Gv5YIt#S5c8vDvD6<9yt{
zXg9-pO)JbQw_@)~gdecmK?>!k$E)&H7^UMAWL@WCMZ(nzr`SjmHqI-1*+YjYQN#H=
zVp4NV`N3n(pm$;i*7&q;tYKmqDct6#d6PspZ<`tqjuT4=cL^MCzv%Y<YwlK>UPJ`W
zmVBEwd!_Z8VzWPd=trN9tu-a}!-&Wv;`<F*Qx@MbYdCR^w!xae*VpC-J2v=Hct|ii
z!!ORg^Yhj$bAxftFuK3~f6Kgt5~)$<#i^N6!~SvUnNq>~AkHyw2<8j~FP@ieo;L-K
z%IW@#9i8X$U=BDttJ#&vTiH3eO&SI=mAbg*wkRZo>32R(a!TApAcnv<5px#+IAcqO
zG@sckNa<6hU>6kUS;5|TJ%XY0<sF8<zQmP%_BlnR#jpkP!Qr?o*t7s*38j~E()LNT
zA5&~3Egok=?H0b%xuX3TH0xc)vP-+M{~MdLfo(D1UF*U4aBP@eRr4x`)&<RaqiJu8
zV_6B}t{~zNBEpcDX;;haO;Bg_o?-d<z_<8kR~hRALNLh2A0;;fzs^8*s@-fGG~4Fu
zm|e^5(`8!W9x&niGTOHNJmvm(Wq&|0*DH_w{@D7nw+jXtgTOJom(#kH-li-`CCKz`
zE(#Hege(b<bibXNVO;{PR^gqAZVM+{|H+8RB%yx4*xQL>$;Q!Y6%PF9)1ie1vfjIf
zH5(XvInr1VF#o?1E?>{{WcDjdQg{d;A_+KhCa>EB%S(21oqY6OKM!XaoiOvYTf6)1
zlvJr`CdG#%@U~p)5&;l-?i2qr1&o&I?BHN5=_18jRTcS0qHgbSRqQrKGIxbUNHU`K
zE9Q@Y2p4AeZeGXD(KajmODLoIUqPW_cm-nELV3^m(r&wX&Hzk#*t`gt$ud*EO)zLy
zg=L%z$SG38eDxqg=i=uncJrKD4Z^cum+d+4U0Isv3?S;1jJS90#xiQ;jwT=keP-ar
zAJJiKWjyEPC>;F>|C@wA=iuVb>h3uMhzLV^g5~Fy1Iu0?ea2>bw$8QC54Q{3m)h(j
z-Dd{swXx?}3JTR5iAW*}yPopy#)|@H9(ql|LpKpjBL3Kw<I~~R*26iZVMGN_>M}6K
z1O_(&Q3w6n2CJ>2b4C+B{62lv`plBOp#_dlXL-u*X%f+YQc`M;%&9~q<w-|=tL@=t
ziIWlGhYgu)w`NJQ?WYy~m6)&CxVuvCmy(LQ$eUJqyUTgb0GN(HHaa1>i090JX?^i#
zKl6q%r!bYu<lFaN#+(a}w!j(VMT)2an7_P1eQP`9JZAumAtJiih`)!Ea`PzI3q>$Y
z!G6@LkKqfUk?ZOdcLh5>w&z)DEjm9hpScad%?G}XaqS4G;{7p`B9U|h2@%-$HNy8J
zBdLm(TBzZRL9-Ct6GEv8nOS$RX#t|ol)Yh2IvRGi9(3`#M+v&_br3QXJ*S{~-^x^V
z#$Caczi{9ioLZ0gEB0yKbkC4P)N2a>PHsquC_4ria~}0OKrdIt;N}g+?UuYv5+eUb
zy*Bc3%SK(16nT%@Vk=aI)<eo@dwyZt1Mr&Z{$WnzT(ST9)8GC5#evA9H||`D%g8m!
zJY;9*(jPmscO1-py04Q~WA67XAA&XyY8jg_!u58X>1=qh%qB^~j>9?K+Bwy&UD(Ol
ztd<n?h-!COZfufF@Y!Ibpa4~yU=1stc542(nxk<8^>5oP|7{c{>^z)1w2!`~t4y)9
z!oLl3UHu&c+763L%ajTlP0gC<HeslXY)f-VqSSQhtxxV2tv+DDim~u@125Ua834HC
z9%5)Q4XAd_k(nV`d+uWhat)vXiQ&^7IC&F95IE})==n;w`by5Wg1t~8&kA;;R(%Ye
zFW(P2BN5w{*z+vkV2u0(!97Y_<JSTM->q;<Br{PbaDFSY?o`e?8Cnj&*p={URK~I!
z$MNfo^45e$9>bkz>p}k(s9n}6Rl(U4t%qR5QjA%FkYR9^530zMk%)^sux$yx8;d{Z
z;KaHI)segWS7FLShcTrt&1dH_o7K^eJ`F!Pt?bHS$i}9HHvh#Acn9^{+wCj`g=&c;
z(KA0BpSj_De0m;%M2RFKBAISB<`DGJb9uhqW8eEKEssMYOR;6iZ5d+#sR`dNths2K
z^}8@pM6VYTVv|Ke(I7Udh<!54Q#PhYgv-~B5afMzfwN&f^i#*W76iJpbLolk3F&zx
zNu|Zf6&`wsh=^rN==WTFT(af84Ci>Xsb<#X8r((=0ue5T8;ZU}VhpuwpiQ7{&zL09
zw<`#&$n|EvK{TAy%={xsk|aq|!4Kx29ejyOs+c`wLphrWqKIFQgCsCOB6_7;i58~{
z3ROUHo)zqY*>jd{L}bjvmc_`+vfEjmCt&!ZVlq6p|B(+}`nSZ;`SOwQdxonwhQE(4
zFO;#&n2Y^iWBbyI{fjyw*L&vC5f#{>{50r*L9_APx9BupMysq8Zk@y7AFy^B)=b91
z?-6wh#wyF9-(N%{POi87?DYuzTbBKi)L88N#O7?D88$`5DnU*fHqMu?XaLafO$3!l
zSWr-SSRld&JFonHI!4O(nK`rkQ%cw=*FYbBo8EeG?2f;4p|Y99ZT6ob;oX;NteoS4
z5(-L>j)>1P6<g0h9h+)qzvRZb^<vJ@wuP2VwtWl{pT6zUt&O%IdHYV9)N*KCu7uL_
z%oQGD%wB08(dg;PXRcVP;yHtpmhJrB+lk9`bvDB}S~b&X)y1T=LjoWOHg63kF_V99
zCP`K+ILD-EZ%v;5_T|f$jd?lSw{1Ut>Qr`4c1m*6*;A*^o;ocs$6lPonEL9sw(KTC
zbUKnVEbq&R(6h0(5_VRwBc*1LDkzi(4}Y7UGl0D-?ACwS#S49BTKE1r^dDT<2JeO#
zwg4)v^+$XfVfgz9ooi3;z5E3CM3?8w$iM^;cLjS_7FN=hpX*0)HwGU5#RdaH;N1X;
z*DJi6EG}MX+7rP&D;8u<Pe9B?+&qa}=k3rrtQr!&-_nT7m{Z=%`{rrv{Q^Cw$V@r8
zq32X=`_OLYW!*vOJUq2nUh%&73~-gAzpe1_L4<3`nIG)9lHZyNj3LlZ)i6M1G*MKn
z7#(LaOIVk5JS9TJP1E34l7>CuwCrt<KtB~kl+5C`{3_Dq(XOt}eByg4qG;KT2q~%3
zOS4mcURtvt`o|a}!rUqDuFh=kmuW_0VE}yMQ_fHKaw_$So;cyRm9@H#j*o~o5kVrv
zCW(#u#YSxR)u}lY@$G*Z=Jsg@iAK*z-hL=Y%5Tw3z{w51ZhD?ywIBmy%tgm0jduBb
zO-BChfnLo)MtLWcz@X8KwK|pD@NR1Q=`&|lD%IqvGyVMP+_(|N7}ILCCX>lz5((6~
zWBX+vEh6F$xju=QlrAKMfy=w-jN?p1{*A_<V$4%tSrHrs1vwP8ZH21g?1_=f;ABI?
z@XSUWU1Ohv>bFOaSBtsm*{Rs}VPTT9UvmtdFaO!QF$kSk)#lO_LgR!UQ_4<}6^%H$
z2Isd`{-B_)=>3}Y;LTImTID?XN7|uT?=qTc6(#l(60RaH90}pbNw4bZ)gP@hyaJJt
zT(P$v?Z=?)lk!vhzrm&5b}`L@XI%_i2%W2q<COl-)fGi;y9&(TdTHVI%ODV-QsLv*
z-KR}(Gw0Yb5!a$k8+K>^u`}D!;Z`n}n*mNbHg~eyt7F}qRe<eYRwBdUQ+cCiCMTwd
zMfbV5)wij?x|p^R7&z{xOTX{QzH!s!>caXz?iA8kBkP}AArW1R7rT#3h%b_$8(2rU
z@TaenhL`zqg@*|u5Xbm;l7HEjT_F0UQL}B|*6JRj<z*Ga6X=<@QhwiQ`K-WjaWR6t
zAjl=561eEoGrQjoM%L4u(eYy^zghj?%U8lV=bUk!PMedP%Q<&-alV_HmXninZ2xw(
zPH$Eb5G|td<+@PI?#D*N)6YAvZ%ry21<xE~zs9<gkR%A4^=s<0c~WQPE>W-}iu0^c
zHDntOn~A-I(7ARvLZwC57p*ItV=rOXr`BzL{94%5*-lSH=v<q*toFsd4hFmf_c~=P
zt8~EGim@wE!`FIn-J3FU+~q+~qvZg!8eGQXI{?v$xXZYG5ph=#d#N&K{X6M1Lq_Qh
zfM~>TGwoal?iGmPi{zCEp!m<#6~%e3EJW!3*@3Xg6tnPkn_$h!KWih;(#+Z1<d$;=
z4bEINEc`NU*OA;@gM9ikXQ)?4_0!Ut9a?Ic1X&-|oIsnLK|?1e9=~8H=&1AgJ06qA
zxLQTU3W`8E&$omD<?|?r=+aG7w`UXXrdt|i2?8E#rrG<WkCR|i<V@k=gE(GadUwT7
z8B)IJmsZ18FY+AIPhaLWKt#-$*X<k=lOO`%43E@8#7BT=BmJ3>``9vL)7Pyc&t~UV
z6h*aK`0cm#*KgeH*QZCzmd#UA?&x$nH?6u&m)=fJP9Y(!NFqTH8Z~OzuV?rA4I41d
zi7bty1dzlSQQ`G(zCAc2lgY#d6$d2PdN(AgFD%oZc?Tfb4$nCve5WHL2Vh`eJEwH%
z(YTh9wyj`C6t!)ID$sd?&ERU*9fZz>XsGHD-VHEh9{h_dDxKSc-OFLjwO*W&i1@2$
z(p_dBtuq>S!i{4N&)=@!9)o6=k>s40g(LsP!SC#2j3Ha*ui#P>5r-?Y+pq-)?pa!@
zdBRnMAHa!q*!LB}4<PQc-2}<2IsPrts4Gl{sw&A{{~NU%7c1H~!v**Tm)$tJLS&~R
z_8$aw0hg=L349cW93)(|=ZOkcM^W3ZGz{-;xs;J>B7k$udf6?cp@#E9`D<=|au-qC
zQy4wO`Q@iwBV$aFw@f8FKoO9ZDXm_gbt~4?wX>5>1(3C)YyMTQVG~EYh6HQQUN#y`
z@N(x%X1IGeTbG^|?X|;-^xi4wH}i7m1G_n`*_>rG0RRzhCyEJ4V!tQ!Lb+pl3J)$s
zB!PXiK6CEMw1U??It^Pn$K$y{F6F!c40$wnWm<t|*wEG(+7`0q$pkQw<vqHPbafzi
z*_-PTnl=sY*0qzjw>M+l$w{x#XdtMsU5|{5i%(2UN=!;hNJzYN>B_qGp{GutZ5G_v
z!^7j;`HPcYogQ)hhF49mn(nTQ3k7=`^Kw60w&MN8A8*{Wm4N2WgJr%K5g6(Ip3Dr%
zOeS#zL^z*-FL#3g#=sd{^-PFbHny&yP+647vqEJF2*IG)HfMi-3rUexJ)&uEbbr};
z@Zfj2xU-Um<)k5Wp51D&OVafPv=~s@g1ju8+l+HtVa&1Txnowy1iJrtyE2u}|F_m!
zqC7nT$x%p(MBHU0L_jp!_~b*1)&(sGphf?}{O4b1;Lfco?J1}+bUtb~mYF*KD^C7V
zrOiv?-vUGC*_=K03r?-KIjeB*Q6kTkMIx&G&eoJHgG9tRe)-tz$sSHBIXP?Ohl07l
z(lVs}uOuG7V30_r7Fys?tJ&V~z1z3aaFPl0<_F@4yKrVCQZr;ru(v}?L=tc{%GBzq
zIFktl72Q8Asrk$ReM!e&;bFpjtwzO)!9(MWCQAx~T7|{0xzBvgt(<>&lSu1!W<NVC
z#T>O}z|XH>R2y*loNWfRd&a0(#<Ue}@~jbP)7IU$Vq$bUoxxx@a40M)Dq0YPoSdBC
zCXK_dMrgGffaT@osnqK4zFyI|Q4kmt7%ltk%Z;117Nl)=?b2!C{Mn2ZvyVt(hIrSX
z{bX~ti2#l-_F&pt0ALKx8dcgSLzJ+yf_+h(XN5}P?1|CKZAt{&B^|WmID`x<He$@d
z&QHtL1GyCcIWG&F7TCeDw?gNN0q>U9CQ3Bo{8pUW;E<lijk=<johqM$x}sTch(@H`
zM0_}I|MTwv5kc<0jd8hnq17NX?`vHM@XKqsd%LQet92N<0N!%_Gwkx>$%=YyFmRU5
z*>hVE_M@FoRH!tH^ITbo&~(wRtBIKs5oZ`XSpUPKnzB7_i+qF#-~ziDXX^57LR_L)
zv<q%g26fZ@{<XKpM9lVv1u5W;aAnQO-X|^^I*uzW_F)W{ch(8=Rxzd|Z$He4PN?7Q
z_O1B*x_5>t&$xUx+mp%kFSkjjXXF1qZy-wrGH!kfKNnlhGdBf{%6;M|lqh&Eor0w9
zf)SlsNK8te`ObUc;SpM`R+1!l_Zqdmz0X~^sMBg0=R3BpSF5!ehOsYBii(OZxQVi}
zvrmWZHgQd{3!6wtJfPqDge~WpROq`De_e(|P;oY{UBK#bZ58KP!M-S&3<{N^|64Xa
z=ZHg9^_-pDFl3&!=bU{P8)lcwa|RH723tOmpVB&G*dq7_SII&vyaF-u<I+6mm<#x2
zI*$J25YHLF)k8A1vh~_ltf!Qjh(idSi_MF0<bQ}f_Am9EJp<5lD#ouut3ky)=Zr+`
zTxJK)8Gtbtn-|JQu=~CNbqU@0DsuBQwlB3g+iDQHy=3PT6%_6VkJj^mS$~-D=bjwF
znWI?aWoyhC)FhIBE#c;Vzl|$v>9kC??d^`RT-R2&ZYPV9?G0H4ld&rVV~jJjZLV4V
zmPdiZ!x_fUOopN*{P03Vl0+}PkrI<27WlpWy6BeAt;xiSbS!ga7q1nzwrX7$e(tt(
zw!tLHRBs5R3SY_Qi(klxb8eJS!^88x-+dXn;g{jVhrB+0>h?d^uKIe#*|6Q-KHdZR
z_0eiHj4@(p+^A8}^F&b;O(ot9FjC#zOltYo)I`Y0#({811jb;JXl75_4`vDql|^x$
z6)MhSBjMdpekwj3e|=S@&Al67<TC3b=fvyy?G2<vS74v$GuZZ#{FGXY;ft(~s5A{a
z;>i!-;w3+u8jIbZVbcOT2!oZ*iFGoAkBxHdp${KEjWFPCjQ$jjA1^kZ8jHWa#G1*t
z{<q!EGABWl3?aICqw7Swol?M!<Jh&_=4|u6GRX@H3J(+n8v9rP00Ice%BFxWG2~i?
z|BQUUlo`;kvo7U?|C`Ud0oD;YlL(WEg8Ie%x-DB0*nMfU%otO1_S&<q%^RrA=7|zc
zTr~U?nhD$=lN1~c0XYWxd0VzQ-I_DhsU`feycY7LvW2Phk%-b0#Oxf({}p|jmHq)H
z)(U}KQE`PRD3CT0^Qu{M%A|>d2lmaAI1yz_eBCcUyz%-}0taA}Fy)nrf=b2n@%Q)R
zy5i~&L_|<Uua~K>FwX<WZ^D$Hm+qofHLv5Qgq;=aj^aEkR1g<0JT_8(DlHyc7gwn!
zHXsB~e`Kw;6?F=sb1TzmD(Zwy{~5rLd3fZp3h!w}r}5}JQ>MQ(fV(l+yAo@sRZ_UC
zLe#8>psq6Gfo<XSh{I2RI0$?cqdvt`ixF7-s>n#P{2j0Cv;U#lsn|GQMuDu!6BTF@
zR(`I9VgD+dvyHl<|J(LIR6*fh!5Qn-#Jg)lFM%`0KtN1_7}V`{E`wxg#avnlngN`_
ziP+-lHLmZj)2Wq~G3(Hr!ARq0r>tC)e!u)@0FZDr)K{-7^qb~=nO^Z2p~3?I5i!ns
zPE0hKEEgLA-!G}DQL|FY2jup_cfV&AT(69wXG_~U+hCHbA2~O))iS!G<R@fs#)Sff
zFvb`WN0E|d7gx7Uf36+>?3mWAT8@2YWa#f}B*`j|gLu|8o)OLDZapG!751Lb@4F6o
zyV^U4OF;pEqPDG2VR}u4yPwQ>UKTdYw(cEOA$)@|Xco9oZ1D15*u5N5Map`Lk{XMY
zoARB300ebGdIC~zR%U0l7K7%XQ8$^Xs1w-#4UVp<<jxf$AOu~WN9PGPJ7ekHaV^Z@
zW<0oPozZdt`piJHKG4@FHgz`!N7vx5FC6=KG-Tt(anx^LtcN)O|CYG252h;X)mxDi
ziQEj-Ya>7F;RoMfTt8a*J5HhMC~DhPfPr;xQ1goe*BKZXh>)5teZMX%=n++uAT^ot
zm)UX}%+Kls4`*TQFz0qHv^x&v=H^*Hsqa3L+p?iLxV{GE4*l5`5=7K+<vR}Nnk`Gq
zl<N5kAx*SO+p33;%v^eH@m(oDVS7w}{Tna1aj61-Xd;4&jhdO9kwpL)pcb%f6r2Pb
z&NJht`V&O|zF_w*z-InBYPF_Ks}}wG_G}x{%G@NP*w<uG-QKB=+CW^!t!E72nDqzZ
z(?Ed0+2n5Zy9d=$?g$0DqePw+DuRCt>$qG2+m=+RWl_y~GA)|VZNa{(C4mEoJcga0
z*qrS-6|DwUX4@W9WfH%`58#*S*!I!A=zAR86Jx)?pxLO`wzOrBv_s7Rha2-C;^u>1
z)A0O%(D7OMto!(@4v99GBs~EeW?ScP>)p|Nx<ih;upI}#vpMVE0z>E9zvx;);hw-4
z8rE{&I=QpCA+kCCoOnlk>b2w^&nBF`W`u?jndOU1i$VjW0S0!_MeVLL|CJi7NESB{
zi16OZG-CHh<C-xha5m*RSKnGf!TGD#XF<7ZA1=%i>u-NjBql+a^{P8e6-{|yxeT`x
z3rnWEG=hh-&GydBB$?_Bsi;&~)v<)587FYc@8(u~1*xKUkRV_1ED;h?aqgC-;2FS+
z9hJ;G1&5$G&k7Zx`y~1CL#q*YrSjWqT`*`i)K0|)FKoA)0`T>t*!8K++0GNtVN8Yh
zY|x?bXJZadt;c_!$KI6=?1NQ6P**(jIi7e$rdO~mkB)G#K@S?f!5Hv1#(jxK-K;Ym
zibh<6dtmtc)|aJ&$V|bXbFBU6x~=f&(+)ZA;?B}uOZhayuthc%0TmRgH=MCP!QNY5
z=}cs)*)+Sm`%mVz8yP?N{}PY<V}Obk6bZ|E(Uvq*Fc%$r@6{Stchwo*L$4BGer%Tr
zkuj#69O-^#ZWED|LsQ4OGRv<x@lW2(Xp>U){E%Vf_-=isSplANjO?p#QcuMzC`w15
zn-PWebP{1$YcR5*7RCUA=+UI=1<!Vd2%=FHzssg!gni)!1<%aOMH^UCuM|8hI0D6a
zRwy?u2EffnX8c+hE^M#T&i&r7u5%7QfJ3XRw5Ju<j$rp^HfP&BY4g2+Ey8|;F$YPJ
z2>TKL{XZQ01-Thjvamcf?2KnVNAK6*A(uH1;MO_BoVOlq&=DT~jyCPVAh0clzmMUI
zWD+Q@AI17vIJUOfpl=gA`2n12STA-2*{S$*j<u9`yV3A(>5$_t|5ciPSgraP{=W5H
zsi5$HkeXxsHY~E_W3wn>+u_`9V-vcJiC?=TE6V^$FuT^aqtE~oslK<cZl%}0H9oyM
z=+r83j>lSP^e+5<WrUi&rzvAxo%I$AOd>+}<=huz6dXs&9m$kmYamIOG}e`g74~Tf
zSoM3RkpH~TfX>Z8Hd?X)T=Zy6T5o$5JlhSDc-ncEon*^-W@_v{pZ~rL6jV!j|Eu6A
zRQKBeg$D#djkZt9j3-B7?-y0tx%CkEH7_=j5P`ibtF)&q5r@H;O-bRVy`Xh2Ey3H0
zb6aq3i@i)M8=oL_d#Q}pQxRd<y&PVF7`e=Ps@<pxe6@#?pw0*x4zEC&sl;p8^(oR5
z0U#v`A;X|^Eu5_xfTx!rbUrc?9d2ZHDmKi*;P;BD+4q@&wbLB(YUWxP4B0k2qI=fG
zh^5%P=w6iWD-`AIrZQ@t@4<Vz2NnXd@=QZl9Sw_2CX(+23*hK^!_o68?yh%V8R;@=
zxJ&Z@6^KAGY1sKA#Ej_JRI~jDpDPiD^*ghN_te8^BY1bO(7W+Q_0*>?832HA=y>jH
z<6Th&$xel8g~V{|bRNrh7F7b8H>z+`G9r=~TeCS!%6B<+8Z`1M%`-MNe4e3%R^I)E
zh&UHw4}#gy*(Lxua^2jg4xF)(ZR(1WtrTwsg({&q&kE(E)nMq|tq1cmvF*bu$!lw!
zt+l(2c@A{*t{#FUqR;E{vkf|8=saxu5XL-58BhtFYM{#t2(qDhR=}wZmZ)iR6s{gZ
zqvGP*4LYJ$eI!LX%E)SqT8E~+(0Yi?AnJ*A#kzc%3^@HKI*l*3#?>27Ek@{EnP%E{
zCoc=3b1`r>{PG)#=v>k7O@z*M$Z^qUuxSAXybF~^W>)Wxrxs(!$A~`bkkb`pIlJk8
zTkv4B{I7^3ktt@+nE=ipktN-l@*lFz|9dxGTCzHQ$?EjBP1WyCtnqk9org1nL?v~M
z&A)O6W}qf@RdZf)yXRj^j$8@}R-e9X009iq@gf8$&htY9f$ck<%LRxcI73i9RZV>*
zd_k6yE!|DG1e!U+z!q@Q$UC6~Fq32o7z|lSF0=ag{c@@qsc!Ej&KYEa=0v!Wh=>Fb
zfiZB#MnCG$70+40F(_%<3gtkfN1I_X<J*@aGr3ATH|}m7V>$6#Rrf5NYe%r{Lz}aH
zEii0RvB2Z~rEzzRSy6`PEE=)%lVbVC$9|Dn*8Ukg+g8nSt%XkGG2uIOo?z2+-v5nE
zf9$iHWSVm8-7#z-JnxD0n#q8zA0YZn;YgjP==PGMj*Gc~jq`23Fj8qTXbzh8a@6Sx
z3igA5l+3)vJHyFRxe2uj)1Py_vf20gjyhXb)p~BIbAYd)Dk3s$E{8sS+4$6q<XRnL
zrZ2e@5ht1eiKwKgn8k(Wtblv}9g7+xNz^n*Q((lMG^x5}+A3590d(bt$sBiP93f3K
zQpKOyiP8D9h5vb%2&3DBm)iAQa;YObPb3+$(jiHu>+LxRX37(i!yxB50|3M3vz8D&
z1FPq$_j1uHcY}h1P@HFla?re=bsrr7M}M+|rnGkh>yg-hs=8;HM4iOuMK+rN)vAxD
zmdML5-%B)l^m`LMr<9@FlopS5Z{qsVVuKlpIJen)Wk3jggY9f&^~A3Q2E2o3zeejJ
zrOACA_!gIT%TFErUS?45j;EHOZY%p3Svn>IwtRrgdkaUJ_eINrjyf*kDnj2wUY5-T
z-6vT`;}sODCENaqH<(0_CFZ>Ad+(24b8lEz72qTEX{Y<o!kRY^`XB$bcKenZjhYq2
zr6dUei16jFnN9k|HX9Om<eb64nfW>WzV&l16GAP|!lyWcp}N|sP+8J53$m@@Q&T8y
zsSOBNy&+2|^3m2orD%b1(Yq}P?lNvCNuPcGRoA}5mVNexDLWHXNU>GLsf58ckc6be
zEH6VoU`}9a>?rK;OfCItD4w%|<4~Mug>ulmkM+pSQ#cc9hn-8lxqtd2M=1;!<<>cD
zS^#6N{A`WdczQ{BwX7XMt@;@KDeAW`W7)MZtb0RV-}~erGQLo!XYFHN^+c^j(_VPy
zGYp-Fz(=iTt^U?0f5$(&Y;F~O22m&Ej~M(OYB#E~t>t6i*Tn?qOMWRkk`##zvyqc#
zb3yAN81SxjGdTr?>I4JM#-tT^8-6}Q_s&i#fe8$pGeKY=LJqYIQXg4cJ7!;<_AND>
z7xth+z88Hx+SKj21mDgv`;O;IDxezs&jw==!xVSnp`+-<nOE_bBN9=b2^;oiiy{Cp
zz)gp`E~S0M5Il{MC|mHHn3UwC({0<nvty3|3*P@QD>GdukRX6D2%IzmX_=&Fk`qJ~
zOioPX43cCIMS3D)T!`60Tu#QovIZAoELJnH5pDdW>Uz&AJOn5;&kAMnNIUB=?A_a?
z>GW2J_^Sx&YCYI)H0rm<o)x%r%RZ;bA03X3^Dua>d}vqifu}#j=J%0!&4D(^65In%
zOfGxb!S8T!r_I^CESy?z9WC@}guu3lI#pE*s~hfh(4s$@JRvW$l%9z6_+pyXq7f<4
zW$bxywau31eO||JGfRup+L3cxaOW0!PlwJGHy!+kyBmY`@0KRA9oQC6et^x36u~Nm
z>Inf^1{3n*YA%9=vvJhKm%$vIvLspyoRNuYy7I$oeC}pZ=P?Q4H;jy-;9=YBey3(g
z{a#KA^iy5iQAaco=l8GL(-sU-F@@drT=l%Vbr({o3IzChs~EEg!d;Is*-*<WZHt@9
zqL_b;aE>t@Ajp?9=XZFn+fP2O#`s4|=AS%uZpO@aoSd8v9z48j_g+Cz`TP5+)vBzl
zthBVWygY-!XcR?JV2l_GY1Q(xrHgbfuI8Ih6h*aKkVFzCNf3mh<<sV$MNwpoan5Bb
zt2tD$VPIS~ZCfIozlpe15WpC8YCgDyuhOJS!Eq>&XN9t8+249(`$slowiV^Vc6krk
z>k*7z?(h(F;&uEv$EGKsb-~aD)=J<GMNp&X)UrM2<S4A2R+{H*=IkaJf#~*QD{o#k
zq(MgvTZl0$(4xPL=Y0Djwts~9aO-YQf*O5iqVLSodQfF1<L{s4r(C>jYPLIqTj%iK
z|HGei?uEL0dLq`(vIvjMQmZ~jehl|I4!1->q2e$g&{NOyL$D?zA=`jU@s=l@B9SP@
z7-x(TJoW7IM&C;teQSGh6wiesf+!&}#^l*CMx<)zmJ*SmW@j$t6-3&Lrf@4f1Q5_B
zSZ%&+CBo@T22reFF0{ao9LX*CfGFX)PPSvW4B#NG8`+QwlOz_>y5+$=o4@$%V;>(M
zqtPTuVq9F@ty?k4$tihx29rqy5d%a?5;zaPdadt}k!#nk)2WzRz>b}}Uz#-S?OAg#
zUi?QRfU)9@)=4^Z`b=D0oFMa-idgPlo_5Yw!82p{?U?1ZXJAchyXiGbXi&j{F#7VT
z`)z|lHAe4_3B~I_!+x~$L%g#m2F*cD8&XVnW3Xq1V`ZW}{4jJr^d2@B?EVZ9haG5u
z89|Lfv*6dFjAf_R%j<#364V*JXIKyJU5W4m6*jM0q0%C_CqjnUEKR?>7iTvkBMJTA
zDy>S{kb?uO5D{i`!KhDdUYSS!_t2^$JJrCj1*P={&rZcJ(~*~X4{cIVC>tV#MJ7MK
z{Gdd{0B0>Sz5$s4or*R01q00S<>HU#L_`cG!uEr?6W>coNs~~_(FS7(s4Yb9@#7+@
z4@3k#Z_?{_M6}6l?W7SdU(fZdc3!?hWs!(ZTrhMPXA!Vu43Yc%0^E7|MWiK$o-ZUE
zJe4Qqd!31^;H8r9LPg+W?H)OOc9s%XGdsErM4*g2Dc7Q7qoZ$`MA2EVuTi6hS4~eh
z7oAqERtZA<owQlA=O-j40r2o}cXxM>jEoWlZr-F}{rdm?ZiNwo*@Y&Oo*DN-e0+i=
z(e$Y=4<9~^7ptZaRnhBo`^SMxvfrFEX!jmYMVbFKVQTk=pFe$1)JYWnttjs)6-p(j
zOR-<$uGl~Pm30Tdy^c$J<YztnF=8otPIaJy*WKG#KMNU2HW&1sjyA&`b-Ee7JBG|H
z!*jlI0_)zWfaeS#B1|Uj{DGF;*1MzQI868+U0;yDBra^jk1t{0*T_o2(0Ro?=d&BF
z7i*l*dj{H#wt3v%6*i|jjW5l;e4xom!-hGfy|L1}JN%ANP^dVJp?#odaBWv}bJwgq
zbo~IN0dh@r<~GuDY<!9_f*Kgy-6?Xf-#7C;z1%s9l$#PEHc1lItXdUO65)2VDKf_5
zIdg_yU7bWzbuv+*YA~>%;KC+?L^yIRmzDdmU6SZ-mK1iz;yE*hIyK;FL$-ndMr3I5
zHcGw9Eh6)b1Kh>Kt4&C&;X?+F7(Qe`-`<@%whIbs=;P~Oqo%jBd(EJR4SxN3^~*0k
zPega_+>MI5X)>7x59;6j@h%2~;rfjmb3gb5`2u6d{yui=)@=ZSz(O}~cTyLANdidV
zs_2bi@?mEpkP+9D^1lVxpk}p|f@cMXBFrDy^nTl*P>ty{9<D`Pluh$17H)c^9lE`U
z)Hr0Nl(91EBvRuL5CTCi*~Tjn&HF;*gv1DlrV1EPP7K+&eiZfFL2E-v!@nh5yyQis
zs!r`j82&yy{A@0`aU457#)a+3POb3fcWxoLXR#5D9)@frT&?g{_m_ay=r9&NUWIQH
z8QH4rRGiy_U7zCmQ5bWfbH%WQs8vMEeg7()4uyw5JnLG2woX&%J>)Bjvs00qfqHG^
zO{(3<re61hj(cr*_}h%S17Hl7|GI|;DJYZ+&RBgf=bukR6TpC&ySNdLLG8drLP|Ev
zHo#2>fm_9&EiM6(h9jh@=H=0@!S&TKabjGu$QT&IkSCl*_j0N}Rg^>o#@<<;dg31g
z00y||+1Il@wCelZKuf`{kVeJ6T$_<=AOM^pE=hdhNmrD++4AXd<LdPmc@EAny*D0j
z3|nz)CaQBE{}v?32d1qSTNDHsSpGj?MIi#?LaP?d1`p^HP_JH-#zB)Oz3}9aflqYn
z8q_Fg&)$8PuY|w!>?rg9n`fq{hi={`2;5*Wb?oq1*Djq4l))GgICRI~W2QT@W=VvD
zSMluuOTn|dlX}I7)@nfs4JtU0{|5j7|NrcrcYM_K{l}l5?~Ux802z!B2q9q#G6h)*
zqJVYPT5D~^w%XRw;of!BYOA%a{dH=qbt`UAQTE<D5LN<7Adp4o<?j3a{QmgfT?Uuj
zWfI`?eDI*T&-XLlclX8S^?fgu6`*LI&=?Xk>R+z@FXhKtv3%<744wuM7RQ?b<cQ|1
zplCm{e(n2Lc8iUXS2B1y8{cF5XRR<JK&ndk;6dg->O)+a44y_<42zz})l5g+w$Xby
za~^K?>WPD_e1*b2ZQtMJ;M8#<do~&zbv3)c#_8J+)4d3dA#*yJmk=K3dqU9xwtmF^
z)%7E%M;&_u@3(!zfC)Y}$-0D7$Jlb3YZ}<Gkf3nJeA~yRg#HYh&Gyd&eluqS{YEi-
zj{j7D^;}=kEfIoQnZ540WYF`A_nC!$Utdc}HE;ZcFdLNCaH>%BiXbr(;jSX)71C}&
zV#vrKapUEo-<=b-|EOzMPMysvuDB>z_{;=7b7*!xSnT<4xM14Y;IK%GaH@Zp=MF*$
zVaO+c9`)0wN&!Q%W^3K1y>%B1v<0G{rA2(bvc5cxA-VaYmcmVCP?!|fj=IZD^9)H`
zkDl}9TqXp%ff*PvX~O8JsHlX5IIMO+)3mH1Sug(n=}lX<XQU69K7Cqa7qir=j)|Is
zKKpA)iKg+YC$CN*M)!#gvbLqCmqHOZD(6|zJpIQq>(}m|t4n$H)<73FQ7<PC@y@SX
zT+o<{e@xaTEloI8#M=L{zxh<vHWL)i+($_0-|U2whxqiVws?SjGGGGJeu_27$ED&t
z*1yG}^=;faG=_{Rq>XR(y=?iAHLtgE^RA)q1!T@3bxgAp3U;&UeU5BvG#DPwyvG`~
z8T;lhWG|_IdG*uuY^K-L+rOkZ*Y~B#^BFSB$GEeG*KR)-s!!WP7&Mh3mj)V(zWyyX
zzSqe%DHLr%f>SrX_v(iCx8@t3`bsG={OuMl%S0Lo5LS((ND{&^JygCDp=t1p$lU^`
zdpLEal!oNR*DCJ$Lz!7EOb9-DIpOm0!GUMPD$WX|lunBXA92L#l7J9gGb8lv7vpjH
z5zkUesfi1&F5I!-=}ma9ewAQXGYQX9bESQ^cH*mUXL6_E&YBiN6Vj0GD){E0Rm0^m
zwf@~n-M077LF;}Ye1&aGi3q0uT^v4XF5{^=L++b7Nc}TVbSf%rThTlNCX$@x{(1Fl
zoH!8Bs;ETfJk;n{@MV9ete_R=hDMRvOqU>&pfJ+L({D7Vj<r-7xDDua>|aT*3?dSm
zu`PxH6Uf_2?Wy*jHDo5!euj2h^v9tMto%2t|4UhZTegi!VeEIf^e$3{5!QSaw8RXu
zmrz&TmLocquvkXSVaCrHGPBuG_Pnhueu+)*aq4)Z@t738^+coZZsjXY`#F~txC0_i
zC$b0oR`|Y@yP1$EdJJefZm|)a)G5z#x{T0hF1(IQ?k4p@LLviNa(o}l|AEm72^0!N
zyAhyi;#<R$iz*zOkCXvIP+7|xYuI;;5vjz4fi5Yp6-5<D2(h^Xxy{ucoqRfH@?AAR
z2qEp_@}Hb=INX9NE6w(YdtyYlk@t!YL_=)YS+_sWozA}JkaN~W!O000TKu}D<Vdk`
z{~yZSDe*9%2X{?#SIuiW89>xd{I?A5*{HR2C(xihRsBnMq&KJiTa!|1TJWX^MAhNu
z-fEqbM;74Hk)R3j^t>U7QNijTf}&GVVcUx4iSOtB!Rct^yK8~I7cldeja;cEM>y1)
zRN*N<A#+ARGZXqV_fZOVv-01ZJlu*STBfds4<BLXFG(BkdqQ|TS3k|?zolSzyAB$6
zW22XqCt3Fv2Ug>%YvYD}Ml$N^Mv?X{GvWJu@q8Q4>I~xgGIAd2lUtgTvw|%jbMlan
zOYy19eT3kM`jPc-`-t!_+{3=*zE9~hlBCRLzt-jdpe&z@e%N$8bqoV0o|XDJsuv^Y
zk=bID$LT1av!dH0z+w?EUq5nUTHI}KZK-s)q(s77>v(HDzr2(?XOkF?rc3h6#nEzN
zLrIP##O6-4lxD_dcl`ZZN~t4$@?ddAm8VCQ<i7cl!cYO*=OAV({+rLo#ZSz4)VaUn
zUp`y5<h7nQ-}SS+E7`EaQQy5>@Qay9!?!dR2r0t!9v6r%DLB_UWh<m8+1uPj-Qbh0
z)H>bDxo2e1$llT2r#6A&Tt#`F70ol|+ioe6y(FOc;eMmI{DG#f)V2TfEp>cy#e)o(
zSpRa_-`V~d$ywM#n_Uu-K-Q(iq;T?3NA}J+um*c5y_$*dwb;m*N?9Q#d2K%;t`Bj2
z0S;|o<GU<*nc~A3ZR&**9>?@M7<+xA>TXTQ-PE)_Z)^%jcTjn<Ehly?$wQg?6DHi`
z{y<Krtbh&gviL>zt?(T)pP0eCM+pk6AKCB@8{hT4+p)c*PxjFl8i?)FY~gp{cn*cx
z^dH-_tX)zTI~TR_yE`K!Wisi9Oun^IO*Ow9+Q2t|A*UJVyh72PFn@i{N{qaJ=Ach@
z6rQZAlSl+e)*R%IUr<uXh=GJ#L4d=_@k&ut2@q^W6Xs0MUftz)aWe?Lu*|smfnsk7
z<-tk3e1jD6;lfuVF2y;3-KK5YRkttK9qf~TTu+GAMrYYuh&7Nl@zQ@PHt%sii`O*X
zz8#y^T$3tfc%QngJCNP&?vdK513|lgYj&YN9iegO8yq?bAcVMLNa8gkQ@Y!GP@H?H
zoM%Pz%=$I`#{#^2kN-}~J<_=HH)i>VdK^`}Iv?HX|14_|v+pCNL9eEb@3H<Z09z2_
zZ(_(y|C{XkiVbgfSVA%|{0b)C+|q=#Z?O4;w(lR`kILfqX3UPhmQh#vC_vqirT@q7
z#a#D%eKdi|$-}(!t9G5<9&90`Uqsf-rdBoN<Y6|w&w({9?b9$h-^6bd=J0stKGO7Q
zg2|%iIJmB*wXuD;;(?}(vbui`xyitb7%{h{tbU*5Zf4yZ9U0%DP;?chocwan@$dh4
z<H<^gZb-9DttK!eGl%f6o9Pt?y1NRo&>_e|QY7(V_2JyzL6=G?5w#joBagUr>9uR=
z{an-^5f>^3#3Oy=Z!6Arq?E!YV#XaUuW<KZ3Jwzc7bYdfYC`zX&t~iLv{{9@h3-yf
zk-?Pz8=5H+Dy3F8^w)JGp7fDNc4jg3pluJ^wm<7L;Ta%=xLYYT3yKR~{od%C#y0Z^
zUZLnXlz3J&Pf$3$ha+_kto1W9ES773M`(1@@y!cZ`#=8&*@Bt#FuexVk8FIG^=~(t
zla$GnpZHcg4bh$KT*T%NsVVn=ZXl$MXXY<kny}{^mb~0CXSJ4~FlOA@Y$GKTXDy4K
zb63^sHHfR8YC2*0KiRpcvn*_mq)Y}+W5C3gG=I)2b}uG>ho7xRUdi~z0q`sS#g2vk
zZLo)O**!kWN|)sE)-U|54US;uFPiq{e(M+QibUvCB6~7qCK*#3g|+)dcd~am+diT6
zXs21GP;>>lAuXEt=c+?LduOXtH-orAq6u_Kh@IExGjBYiu0FI?=wJ~fgpn9Yhz+y4
z5nb+xmfq-W(8Q`wPfUKG#vHR~d~ZDeoR6j>fS%#>jrGh``uSWzN(n}_>*c=`-u(u;
z0iX#6_O<R>l%%=y-Wz@0x*_)*bq$+S=<NZ1(Lk2n+syVLq^P_qM~3%9G^$fAr4&LK
zo`PeZ9#nxg%4$E-LfCuhT1Ae}^Q6S76Lvc;-JNEU{m9G#31KR1ThZAl@vP_wu6o*c
zg!3yuq{d%iEqe%a9--&J`jIuSv*knIlSj<u!fUaG_}SCtVCNTX`KV(gbW$dlKj1sc
z*W}1HzIdLxuI>Dk(4WihY3l!Mg44k#k8`X+OXnGPl0K>NoLc_(vyODvw?I%hgQt;o
zX`^b`O{ghn+o$Ya9FRl)^up}R|G~~LT4B;fKV;bKrh^NA7f_zUnClsRP5sNF{k(Ty
zD~|3OQZHoa>_#yiflzabor~D9@LVV)u26U|q;%=>{jK>wd2>sJ!zmDklwPN@-O8ka
z{OP-7^u?$JiKdy-0;C|sLR=ULVFcR%w}@#-XZ*_UR>?xTBzdLI6IJQ=$Ri~f9#M4S
zFT~hAWzHldN0A(5u2lN@T)?~mbc4+W(k{n8#<~VkfWrsbwtSi-tBi*JXR`<R%Wqfw
z{Hap&xr-)v=Qiezz-J35iNM;k;(G<c)#x5JV;eM#lClc(RtgOci3tlj<#1`5rfFKI
zt7s{uO+@@VT5!r|pDLHZzgKYE>lg-_5UFvYIgia!!dcM;DDkXlH_gR1?_SK(|MRof
zVq@;3jcQ!2`VZSaX=zDlG!wqxQ~~<Ll+c?ObbtVe?Zezhe7M%Vl;-pKGp*}`-5yde
z<g#WSBYUYX<<qB|CY*;x^W8r;eGssB8DIaYgDq@A%22W{ZMFuRNktKx7O?xPK=+(|
zAN@u*8Y#}@<0o4E#3XqLGwvWPwtl4YByZk@v$pv)9#V!edA>U!qNbczZfV`ookwUi
zS(lPIqtW{cg#2CX{>t4JTcJ1?F{G?>8Jmuj{pf$2_Z&NA(S+9(Ed)Xk7tR&Kcx)ai
ziBRkACF@qv0?@Dv!h?tjB_fC*D}a=oUb0JwKt=57jng%O5Y*|MsFX#OMy>9$$cay|
zHLu?Hcws-*JOth_ZS$y<C=!)^J|{5m3Mtu`Z{DC44hGyuai#lS6Oh)&y8EL<0alCK
zCoH9uLE2Tf6@9$a9ib3n<JA9QX(pUsN{k*8t1e$DblF5ayIjt1UGv?FiVAP%w#dlv
z?_77y)TtNu?~`O^U^|_Lw_<ijzkOXR-5&bMC?R}zct&IJ{p>sLt`>I3l(auwJzR-r
zMHisNv!Z?2L%Heyd>8b3;|~8Fs%GC$zftus_pjp17XmuB*C3|e7En)Gr-Pjf+450)
zC3JWkb02LfZoICB&!1^;3GP4`IEhQ{^fTvp4xj(lw~QP*n@OjYj9&0Cg?l>2yoU5h
zW8fqPOz<P&^0u(^3l47#bX%*P+4uYQraCHF^`AhC!+QyiVEl~?o?bul!2^vv_AN8$
zV$vtGV__@aLg$e@lp+3}Q=E>ohTUJYb5Ykl08%K<9Of_Mh9UPARldCZz{|@IN+~r>
z=!T~*k-#G86UVHf+&+!Weh4SJ%aa%FEhi>GXjnCZEri+$3nJ8p-GbFqnn?m~W$r^E
zyZc9bevlHEL5-7&TFM<%yQHq09>DxcW<10GqNWOc^@EH{Z6JBft0_?=MY%UrV&W{t
zEcxy-*ivYACpVw)tj}fAQ|JwH2qHqoAMTI&;dS9gwX|qLO6fF6o|<1=?oN>I9ZTMy
zkTuPAcy^5Wy>rmd(N2rm#<>0GzuLZIr<5RYIb9(k!RDou<mA}c7$L;9SI@h6@<iQe
z<oY&)z3o=5&ec|FEhWgH)wkFV{8!++!*iXT@SAZN?wIY*ZksqOtCzan6<vT5&x-aj
zVlESI@-cpJ9gCjxv$lzfD=*~jU*U4J;`E`j8TTFkQQUy5jvWixyr5kbx$U7`@ldlR
zxJ_CcHr!e+yp9X6^E1b98MAAirAOm3%jJdq@8|8c%}ZD;gQt-`*-wso%_(+&#jdYf
zSMl2(!kmZc+2Xg5zmqRtXtfJED2$=A8G0FI`7C(2b!*Q!wjk0cGklK!x&Tcm&1d_k
z>|NIO9%%~2S&NhyQkK;?m+m|9hi`J$9x9PYfsoQ&n%j(s?w`Q;G;X|zse`cE!D&W{
zdUo+fO`DibV~d7G!z!?9?w>8feF;EffNtP2aOpTDE(6^_H%wJzig;-tY`nafM?a*b
z(p-kO|B&f?UiA=~Y^<LJBrE-VHe=p?bvoORVHn;30x2;BpY7o4mvDI$r5P@9!C?Cz
zABY)0D%e$Htl3pJ<$K4yH_3O#^56LlZr6tA0N5(8%_X=eqG=l|DWx=Ab+zswtcp{o
zKKlHtHETEQ+P&Luw*x{5m&^6?OV1A-Hmsp|4zY+27k;^H`RY+4hFyErTtl-o*D1Z(
z*l#JNRYbhr6XyxxD`wk}6qM8Fm+pTNNRB);J1Hte-Rz1kK#6BX`?&EhME3MC{{H<O
z&uM94SS;WDbEE3qZ{F3Gx<tVdTzD-*PfJ_%iL;jNpR@Jjc1h?KT0ooZ`if<LYy18!
zFzFVCUe?lt!<$(BAO8;j_*AYt?a@Yn)!kaa9zyC^GNzK8<!5I8PIfPDOKfmZ7;_(O
zS}gkTrbh7#CY8l}`XnVsT7AS>M`TZiU(TSZ*aLWq)R2PRZ2h=fc*_-vGm7_j^S#F_
zKG{+D>e}2r$4+7Pzm=W<S&0xthcYaM@dKEe#n`@tha)7oFl4>|*vGH$^0c2D&S@Zx
zObqGv0=~bQXFg`j5%)qNuv)aAPELDqNe&)`^}(IY9foM4kqt<Upihi@UnMEdBFyir
z*2%8p=th&!dnu6?Mm@szqm9B4Oz-RHEc^H7h3z@${J%FV-8Y(G$vsRQfYn!rXNiy@
z11hiI+a%$+!Ev5?brK=42npiw!GkN;Z>n?Dtyr~s|NetRhh+WjFVABjpwdyda>dH`
z7JRyW$4-mIqU%O-QsR?OJT!3N04W<MW`m|-g*tcBwWjY=T14c(gQe#1dGG3yCqLmg
zpP(C9G%+kWYU91rRH-mU*Pz6+q8%h>G4F97<AvEQc*xJ%=|5x8l=_!`dkeKpR4>Ne
zNZPo7CfC)l{WG?Gf}^U9$6fRThF#v$g#2ABe3rWEHtz2eZ|5+dII39tI=cc+7@qij
zhR<m_7)VdD(+LWr-)II*^i%g6aMqB$lpPEG%X^=M#xU>Erm^~qU*gzaZv2a{<7)Zf
z0gAh81?={dI)))L={>yF3v*Vm<>OW~U{@&456tjXDdo1K<sWS?_+VSX&Vq74LrC)%
zevjK&2!d^7CNaD>WBN0!AA=Hz3Ih#cAPjeAvJe5K9h(uknufF>bhZ`p{yN@V%fXZN
zVp|{*B7)xd!I+B&#3w)SSy83KbTapkXWLVRp5DS&LR={Q<J>Qmt~iURbdX)-ek9WX
zDUpKL*7AdYx*wg?PmmrTY=fsTLJ$?giN9dcn@xCj^}VKM>bt(uihX11UbBhYo;}&O
z-gc*^X}wdDQ&Uq<oIH8x;GxQ@Dy!A%bUH7cF@4pVHBP6?YPG)b`)7s@$uiCLWKheu
z-hOY>=B>Z_#m^HHdo|6^m%8Qn(vUANZ7ycp(CB?P$I9Ia&(F*o^6MGt>K<2g4N5#K
z+QI3yj!pb(u<&?pYEVn8tbjM}=m2|qr88-MpgnBss@buK?VtLu2kw*18BF;}OB2cp
z`1py|1thn?<@eL~H0P>G{tou7AZKOk!fvfWT>k=*JsJ%b=kd0`E=^u+L8Ojl;KXJP
zYe-oE+dm_FN!w$CBYN<yCz@6lH~Ux9{F8-}BfNW0K%K_VB7!5xyoAgdO}{?BR2IAA
z<U6X7n?iApV*1$(DR&oEe7L>f^IgX_=ao8j>8{CyJFMFsK`j6*g0vnCNoG(l2K6Gj
z2R$Q62qWB1h!uPlItI|Ga&oeo+!At*vU)$??BPgheJ5%m5QreFcH@{nzrT7|xZNs|
zs}Gh;`{Qz#ZkT%Mxo>m#r3j<3Kr-b}WH1Bc-C9R!i?f2V8gfqJD^^)bDh-C-PhN@p
zdXI>pI$f64)S0E@o6Hw5ybqh7LNxP#ywcEWr@ZFs^X(=H&jF$#-}~UBKfU~Sm&@g@
zRVHOrROHw(7hZqOm6?MF6&4=*)%}m=<{b$R4qouy|AK=<%xN|&SAYBF{QLr&&6d``
z|G)nBl57|>E+xp2#j`AhUpFV7OLgLY?0%2HhI^-tGUF^1iq1!gXGJ^s&L4?MZaTR0
z3zq-G&%!1>e!hHx{i{06s<d&8yRoHaNKM$im@OZ+rA|_x5zP8E)}Ur5)K>7>(|*F&
z17YIL48Nl3preWdtJ(Hh+k5c#9LUwrG#$)d!Z*IWxL)j`^dC#V(e%Bbr7q1T)n(+Y
zWdEx6eh6`WnENOpQH=&0_H*u;&ec!*Uc2i{mi@h5r*|KE4<~a*tBZjFN4Bx+OAfB<
z1e+9!a~V@|N=aEw-N!o$7w$T?BmdN);wp#BpuQAXeG)Qw{DmGLV0aKQVMK)x6H0h6
zHVak_hl?r)rPY*FQFIE23vb6PLA_=Mf^FKRgL^(TCu?x8NQBq(&B$jK@5n!8NGTvP
zggsBwGvc(oR#Vv9tz`6YR|QvePrQ#diYhr=;xk-9N~Fep@ACX2x7*of(H7q^>4HAd
zcf7mp!)*oS)lLbrKGYgb@Qax|d$qfDdD9I6Ex7W!0&LN3d$1y<)C5IEC%#(z&9M_F
z<6~n-4$B%ge7HrEbuOuCny%}5UG1&6-(6i(^V+L_3knJmLP*2dyKnzbetM_Z1^wxQ
zHzT5A8U=N0&d66nq-OZ;@C+LcGWIFJEH5C6etSi9s9ilAQgjVUJS*BtP#CxTr|F=p
zj@N$T-<SKX-;g-iv&6yQzuWQV4V}&CYY2%9bpFA0Z2o|v1FbtHz8{xA;HQ-IS1-1=
z@z3}EN<>2a%iPWE`MP6-kk9gCOa989uj^mhLWt`_Y#-wK5Z|x<i))!<d)f902iA6g
zkv{uupWY0-g|POI9wmQQ2Uv2Z2#F$NDw#7{@kq#|s)Xz%>|RX8iO#W1p*UZVZmA$7
zRMa|`?mPL}uEO<)OOKtZEvc&0r34H^n(@_AHgW>`tkvY*)+@n<XRW3Q0e$1bt{Bqm
zfteXSB7&N@jin^7v^xFqFPyqz>Zof+^VUxhr+ewVq2_~n5E*1fr7IV7_d<FK)8&<N
ztfE=v47nv_JxHz79jiQTP{Nl#pX8nX`>%8UyDoQMQDv=DH%BzVrpFkXgvGab){Ll;
zRo~hm^maw{8-@W`ESh0RZvim#Z}Z=Q)haO1J^v)6l(jWAm(IS@tRfZ~8uH<L|FZ{$
z)Tcd5iCz-%p9HK;1J_M-`1$MneVIGqdGdhxukV<wZfZrBp~SPIt@IecHNR~-xbhXY
zf98Mf4KFpS3H<7Os>?dkHWRCzA+s2HWh*3fVK$o=kh`fBr-a6E#e>B6YiZ7gci8Yw
zEB0(c`Xny;A=Ra9`;@~Q+Zbow99KU>&w)(`KYkp|LR>%kjc(M1+b_D4oR#cc)au6&
z&Cz!hm*3xX(C2Ly9>)zY`3?jq&F4*j&AqxOeMd2PT1&a^&2eZ0d%ofDhEA|ap*Y`>
zo+`oSuj?gMb*l~*f3fHIx<e($D{8A8dUc(vPL~a}01av{dGW{$_Z*u=M1|OTL<C(m
zJbC`a)ZQ^6URCAOhU@>e?!7Jfrcf7xg}-FRpa%84n_xgn6v<KbJ&qLJ0qIF<&pu9t
z!~D_rL@Q$-Wy4{2Ttkr6a^Ue<iIKtHZ_<!b2=Usw+#kHQ$qY&86-E9Z!PQK{Gw4ZI
zRA2gWJH>O$c)L$GbSz6Li;9Zgc=Mf;#U=B9`2GI<`!#yoQnwy_BY4&Keb(8Q5~rlc
z9h|ChKSX%-rqMTE*hk&biY`NmXGL2{8^_FFH5xf_kazD1XzjP2Y$VBaCp*7j%ZHsK
zp@SH9IV0y16dvf3;yku2VBd;X95L(H^cm68oI@M<>c!SI;_4bA61eVpg2LLn&yj8H
zU(Np2IBPn>EC5n3<g$C44la3_J>U4AKV%k@=lj02^l$7AxSHA-C$cAlr;{<oPf)|@
zlohb&8?u+EtY(Fx%P{+034yoRuOa2(;_7`x6*~$}?K@tXUtV+kRBcIBU9HQ|r8Ep_
z)r8%u#fI6FqC!&RLNDkYGp1k6kfbQ#SuO(WjN9|RPft`j%mQZp;>msvo8Fw20EF7e
z>IF(XcN?U{kZdi$@R8i>ouW(rwTj#S<E`s1emx`ow^t51ee(7HUiZ%CqoyREk;#|$
zg0ngCER5I*RdZK#s(AK^sh>e;nlKEbQ6noU5rlp=Ml0DYr0<01+$vHVJP$wc_^j0U
zFm*#Kx(p?r6>VkE6sG^I(a77sq&PRAwaqmv-u@YzKcu>}1I%lJJ%r(N7%?ZHrqCu8
zMQr(qJ>LXcZfeR;$ehvAgpxcyd6KG<Kn>_JQid}3(KfEl-$~924y<i^>GI}CpTvyQ
z+{It~o&#%IS~dC_#$4}Xyt<6nZaWJqhow&<^AeIWTfH!66}uOczoQduQYaLy#%K__
zZD}EH8+1bg4YhaRNBnnP-u(Y<G7Pu#`G&E){v)5IrW!62Wc46C2virVGl%z+FRvl{
z1ir$my%PapA@VjV9cHC>F(4r<`_Y*`d!h|~;>(<(N<&Ib@Z9y>HO+UzGwz<@){1Kj
z2#SN|A{M$FX&@^jUWt;nW@9cs*~+{>fk*OJy>Ml)&C)%rSDY7E9=p8tzojV@K(F*h
z&2<m1W5?%#u03&(^ox9WJA0-xVlE+(6z5XcaSlvCcTup5U0<R*iSO5_zjSj1h0%8u
zS(jq5Qk?5wcOBk{ql!Kwo1G96#h{BR+(T7y>x}PgP8H#-ZAHf;XAOB<*zq|_{*RrD
zIB}@sMQP@Arv0qx;KFD9XL=XxMt71jtm(Kd7)KSy_jQ<s9ZO_SF1U(I?<Re6qx^Qi
zR2H-K6TW(py-TSm>NNWl3PtM>LTH{pEzdOP1mtKw?TF#Y(aUm9<d)Q!xy-xsnKhiA
zk@ZcdeRHykDhE9x+@;ABU5vNB+}<JzPC>Jw+L+&?oviyQDD&pVi??qX+b=G($vq>X
ztlIg|C%eoHd<lQOiRh4KzhMCqY0Y2i$oi4c^?D=D^erK(^6k4{Li7Ep7=lL^u=f~J
zV%5Zz!;`-=y0_}^tmsOV=ULHKhR$Zv59?ok@BqhJqfDZcxbg`?qnn+Oy@bsR+Sm}d
zW$YmgyPOeov4;k_#97O(FWLIB|G@HoqqyW3*qSfF4SfA4_AU$b=q_XYjf}jqr3oc@
z9Nod8jh*h^96pDM-)}nj`7<02I2S#kKQr!V+7Y_CjMsnK?IP#X#xZz?pO}B2<g6rT
zC5Jb5f=vpA;_N_5PF6Y69$8db=Q3-9^@^n6&qy<Qwq-#fG%^wh4^mmsU5D4<T<c_a
z5iY&uKHz4+_TQFq*BiJzox^XRoc5>dMx4IRkaG3G;)`EcZYB}*iRIw$5zQn#yZ1N$
z4(VSz@z1WL%dkiQT?j)55E>xc@LZ=|==L4Y*|$FEGiKY6RJ(}#DRrJt?&~`zPZ^k?
zZeB%KqQtYJt)z`-<}ckp7w7W!FI$}Li;zgB-cG;KEzLQ+fo-36idr)z^zb>YOZVOR
z1)CT6>E0aMhbtav+L5>k+dpUJzgux`*D&xR(kGM9-~CXdrkt_@3ipt|i{tw`I2NEe
zCfvk`xlIQ@eTuv-t=?+v4UD>~(a4U4toT>kZR<cnqsh9ItV;-OWoyd3t>moa;98uu
zonx6op*Wi`r2PBJLqC0ct2>JnE*``;4<KCrR+`EwCYZDYH0n!!D>@5lhDvj!lw+0t
z!t31?wo7a1c_+Go08JAq(IGn?yd*r>diwXIOZo2Rqc^^~K{uo(7}b|`Pe7fYuT+AN
zLL$&Yr7c#5#T(HXdi*FOVZ0taRNBMLNy3mG<uBY7+&kk|kx~jHe8FH*vEOI4FhjEA
zI3pjdzgGO*oY>F+VhI(BjzfuOMO$guC3yW?Y<#bc`=ktI(tIENq8oDJAX`6fZ^I{_
z*n-KN!LZpx^k~JZeaqSOK}!vmgTk2gE0Tw{w8_aseEzH-9e%zs^OruF7=Q8`j%;h=
zhRH)1djq}F-7hPOc<uJKo7W-qy@1Raq>gFbs**f*FJ|8gYAZU&EQLaGwxS#ITYq2k
z>CQqhV_Et8_ql08^Od(7BEe#zcO*R`-0g!~c7`h2fY-NQRmZ_%YMkKbV>VaRXbk)X
zh2`}^d)eI=PfUyR8KtaC`N5X_@4T|!6wep*W&IO~fOD!mU7;jWxMRAtAZbsKA&Ev*
zUp;oX5qE(RGt3A{LX#kbAux1PPP^Y-`;i9DqKMZMrL~!aXTe{;;is<yXhMYAEv3)R
zwW@q)Mc1Olv!bnpMsxE&-9NwcD^4D6*KR{EW6bq_oWNBjZ2yeC%Q~ef_kanEm`jfV
ztypqsJ)7QdsV9@)K<K(^7XL~069TY>F#A4|vl<QRPCk9Q1L6+)UckjaB0QcQpR?i&
zIs8MS$ecmu^wvG5aa57BlI$fE?e8?p6bi-Jjg%a#s2%b2;**tihLnINIPg4uqWmR4
z*KYx#cKXH;8Ke@SI})#KHY9mvoCx&kr<53i8~?`J8@wsf;<3vIKQwD_b6<`j<;L97
z@y{<c!?qJ6Ir0L+X#1CGn(nqn3J^j=+T)FoJ~Ap*kIB+w1{={sWLSS`4MIqyXy~9u
zw7#jhl-h|EVV_+rEzKr8U-BGFvoR!AP0XK|`j>ByPzle9E=GxGMLU@HD9J-9+RuCU
zb--@+P%gZd;dA`YJi3E@%Q>*7<0DrS2Q&PN*0t5m+sgLOnhh#W8_$e8u-IDKVdo;2
z|FfkD=RCm?%ze~Huk5x(Yd1$?28G!jwp-xz9!|zI22O0<s-xS<Udq1Zonw<ip*W|I
zk}tE5U-svfU^c%cE{p@eC)^UScorZbHk97c1Y0pwB6K^E^^x$!Rpgc7^poLSzfvLv
zudd?OSKN;>G$BT&L@&Gh;t+dFZr18Ly~n+u)Hn@OKd*V1(Y>)WU&_0+c;vJILm0xm
zZKN&S2<s;!)Ag9aMofkgm2QOf#~zE20EY0U1o(>}mk1+h&ChH(e;4>pcrL9a`7RtT
zcfYQk4_%VcGeX^@iY`WpXGJ$7suyFv&A>?kP2RtXeJeV?W>RENM$9E+3f30uel_I8
zK{hY&-9;&`FIPP1r(ti=0T%s^%5GMLyK4xGW!_^%Cp8^k_$-IcW}kdeIGHoZoIzBt
zR;(&3AbTm<-*ifNg+ie?caf6EKi~c2qHMEsl_r=rn2&!!aH}M+1Oz?9Ns7d7L8^9E
zZNmJ}1XMaWTuN14;IZ&#s`J`hCjQ1fA_O5e%l^k^B}N7NSqmedUc95=lqsIC9nD*}
zgWic>U;{ZVh3PJrNSjrLq!>||dfYH0Cd-KEE5rI4!M%iL0g%G&Om8TsrIbjm>QLDG
zgAlrJx3gr~ey0D{ojGo^XjLy<X>G9&oI=qVDDkZ5mc;d8?04vQVL+1|RqR{N-lZ)?
zP`81gFos^nkXbF21aCrlAzMCV_t#CwgCn@?9ufz)w6?Z_FaB`e_2~_d=c*@t6j%H5
z1@^Bx69#nz$wL`(sgLCC7O1P?z-sn<!?C@cW|l&sIG2%7=Q1Y!Vd<v4GI#MX!2_4`
z__xqnrGMS&wSuGwdPQJ01&@-@f#Ln2>U45TDXDIuhqKSsKz=z{4^ruHyZ?cucU?Sb
zfbUf2M$6&Jg?k?RY>yeeZ?mxcKF0JzbX6B;i5b3)Cv4jgx<G=4435>KvW&P<dfZ4O
zy1xubltDeQgkn0GrO;083i;|PQGL+#RyW;Qf`v!k=c!NK;yF1gB=4!&>ZVk5JxV+)
zx-mV{nD_${GXkAoa)j(9>|I7pc}JR&F_n??n(Vz*R9sK+E{eOmySuwPBuF5*ySux)
zySoPn?g4_kdvFQv?wp<9|Ezl+?!$S$Yu%TTp4oePs;jEM`l`D7LC`AjgssJI;wN*q
zRI1{y%pLV7&CUbY7M~R?@@c6ewDucEVxi}Y)v<AA(H)?KZ;4WZvc+g|C-zMDfZt5C
zP>Dt>I3(#u$K7x@O-*Hg@t$5!>Sau5OQb$H8o5lA4N2iQ%t4>M)JcV0a#IXd!|Q>5
z=wF}?W25`!E1Cg@B)^Lyq0LMM`1sjTkgFiVRW4tz`T*yoEC&ILYU7i%KNHZn#Sh&Y
ztOZko5jHGJh+!g8cIs#h8`qknVN0t+#h8^4Hm&$ot|XK-zMc|U+-P<nxRH&h-KOqe
zd5UfvkovGw9Ag>UnCVc~YtJ6lo9hjt@T3tpZzdz`_zC2V>d#ev$eln#>kTZdy-5TU
zZ~j(i-Z@L^Rx(d$6PobO*^VgN`!A<E6(V|0!W%vDMLT^_$0fjbe5I2>+O#~gREcJa
zzog{qyKR2B5O;7ZD5-`EGCS0qTWX=W7)@7kz4iv$H{PPt>x1avFG<>Ck!s^+0%NO%
zj*IXNe|Gd3LvU94z1{^?v{lhhr^k`jpD(!BUqE1*Z|`ey=ZR_YJ$eT-eIRxnm9nKQ
z+On&UIX*LSp&(Q5y75`acO^gO8@;8VsydZs?*Sv=`Etabgz7A7S^unpp3@A(W4*5X
zb|)<CeMSJvn82-x_&cT9djN@X*seeHaM0lWzTcaMdG%~hOK*{0xsf@5@Y4fDTW0>c
zar}?~PyWRuY=Vq!8|+W&;F<2&rs?5LdKK_*X*x*L_jB6rN-^_(DO1>ZCKU>TTkE;<
z=~<#%mHHE)ir!xj95#m7JrNX64fYz15!fb79w@*dz;ky^04JCr`H@N(eZQJ5H*4+v
zN&ngM(x^M^a=8rAW$bwjRc3-p5Tt70Scre|%Vsh;cs;l0tN6_YR`+eg>unrtGxjG_
zz)J(1BsIXK?^Xi^L*zY@UIu>$el08}bS8pZkFG_H&$6Z=WQ-(aQkX*V3Hwv83GG}_
zfp`PS-=eLGv;%%?%@=j2sXxEAu#VhG8!zu!KPL4@^ydC9H4$l}ybmV|Wp4I^>nf{H
zesG>?OZ8q%HGKut9K=OjsGa5&Fjp5^`Fg{~x=(is$tG3d`I&w?8H93o1Th1TA7t+4
zJBz&@V;10Q@RX_qG`*F@&5~Sui`!TiPt45Bm6sS;=vKnlXPCA}{S=_1N9H}7Y5G6$
z=icT*Uj0!F8WS=Hnh==qhcZ4SGP*&a!{Eh>Ebr}$UVW<O`r<z)h10VK_XmLjEd0w{
zx3XB}obrET4i6h|{6xTJZg8vO*+9z#?WxUMm$n|e;3}cx@|_t<k=hupkQH>2nciTH
z`b;;hDh~dq(K@2v!R1F5=BI>v%hudQUjgmc&&>GiT0I{4Jd6za@|XDQdGBh{Xa3cD
z%2##wGEKNpv7O+m&-#bGbI6YJCpRCS*F4JgC8T$oUA5UD1Ng4uNcjG<3u~*>i{e&3
zf@9Q4EW~mJJz8pO3o2>uE5=X}D}gJfY5%}+72Gw3+6{KL3S0S@$rHz|&&x1(E@Dg=
zm5XsN;ilCOIPoIbplXQKBizv^jxrM#QxsZQ3$B?&1(yw7`!fW-ib*TO<431$2xY+5
z^K0=PPY1rIURCpu|C5DveRu+Aa$%#-f)!H8FGd2L@ej@gRrdhacd2An&Y75_KX7h-
zy)B9|GJA8KzWeW7CHpY?0*aP?&^Vq$xV-V#Teqy;Po;JnGdwgo)n7(_m0<3yHe;GF
z8y;fk@NH(M(E_nwBaX6yr{nVp_+Xd|aF95Tn~-Eh^Tx|+G(;h)9XV~%5sJGtJ&*fM
zQCI!yc3L(y9c}gR!sSOxtHaG8bD_W2@fEi;=7Ab+kAr(E#%r`lxgGT{2BWPX-DTAs
z3iBZ;9ybM+1;y7-C8$}iVima&C`MwqNfS??IvfcT6k>g#A?W>o-AuEGTzk`PWSkU|
zJbTs{?OQFV^K#KqZO-q4G2!(8$>Q}`Fb6RCQ-p<t{_K^WDDl&5tYJIw!dZkIi5CjA
zm)HDi{?*Qut}pFJSi-luNNR7<f?G|?zu3q3wQ<?zho5$Z8|)?1wb5~8GZDI~U0c<9
z?c982r^XZ-WdVZ#6O*&jeObUxw;y3lC3;`yPa{-xYx9GG#ez5F@(MvEj8Zk=xN^;N
z<?O)svSh?#%H6e6`2I~doRojR#>022=RIXus6+vJnmye8Fuv1DLV3++N?s8(TxH9p
z<V?Q`*}wZ%10X29zZbxgQkE59<&9DuESqXo1cH)s8Hq*47w8KVtKg=NL3SOZB4W+-
z{{8qKawTAO)D~oNVAt~;GA!gWYy%WX^WF`Z5;k>-%~}B^TRHR)(kvc)ne1;1yH-dW
zvKM9#wFg&2NF&j4vZTgv!DSHN)z6NQSY)o#MLPXCzkGPLsO{u^I78{48UONr$pNr+
zhTPD1qa?sXxCDS#jcUr8I+|;hDFF<Nh}@WUj*6a6OL4*am4{+-(drcrY$0KJK0YCA
z7Ri8u3rl@r$O;Mv9m`U#;GEuMH7vyfA8ln>Q(6|%`~>23B9Pq(*z${;FV0DPI}d}m
z%bckdPjMDUphX;wl#>?KoSbdQI)DCqkl#<asIYO?6oVoLgK(^Hk{0WZl9#nxW|<}B
zgComYGQh$RTIh?rnZ2Ol&X?<8h(q+e5qG)XRH{Y2rw2E-zWe+#OIy1oP}aFA9_367
za8-u{wEZ}2jhKR^3WvY(9vBa2!jM>~e9`)U7D#7(W;SE4h|g#ti2tB5+PUQMYy(F)
zWn<uC>d8+r6k>7}&v*Xg0i{?xJtA_5ec0g7$xF|px{zu1&sOjglaf{AGX1sMIeRAX
zeF;z}(za%(@xD;YqSRhBXu>q36G}+1kaD%sq#`yoc|T+(h;0hkqR+SRVPhhVSyfaN
zPyMZLERKgvAM&-o9<H6~iu!&7kOAvLyfwFZj2tnQ=`y4emae*bviAUGc(#Yh53gWK
zT+8E7)bMNRcrwL}zoImy{M_u1C7dy+G8&=OC64mms0>9FcW{zSJT)CcB`LzphT}3Q
zSFW3|QJqu2zNdLtCrtask8{7LxrMemyH^M)$Ejjqx*-@Ol6#?mk_QUPxGuJxfa->I
z#(i76S`N7hgDV&1VHAu5^RmKX&9@27d%86t+jK$QFu@dF<XQHO;ZnQ5&KJB1ccUg>
z5t|@%pIM#Mk{XldigRs%NMET5o0eDexb1G5zV$HOd#@F7^oKOgGPM%rJkvzhI#(<T
zKdM{bdB{q=_Ig)2zRUL$7ZtpRbhgk&JEhZ-hB!EARaYJz89WY&3k-BC3SrOC&zj2G
zn%I5CEh5L)5{2(m-Laswr6L~0<UB5dt!-ZR-%J(04L-c7Kr5;d3TSG6-(l&Yq%B1O
zvq%upjeius@5nmG@&ZoaioVlBIF+{|v%J286JG%diQ`yj-IU*KJ9b{;u=CID@oHG&
zAU!~kf4dsCm7iCk;RE}pOnkGlCSNIi*>)Ulw@Ru5$k#^J#4ZlpaFf|$K+$noh`cR2
zA9x#LnLh)Hf9Phra23w#JVVUN0;-h!=MU*OgpQ1u%7F(Y3qMoVSM_>-Oi5_IRW=|V
zIg%+UX&ZA~)mO6>QZ_p=R}t&>hu*^5q#<Z^j<etXl}e;vD^I>6Rguer-^+D|Rh`A3
zqT_+bv{=5=GQ{JVh4(1Or@K2>>H1AYIyL!(kD!Cva5=r1Fv?P0lW^;uKHANLPbM(s
zZLhP~4tJ|I+d~EM_?ma`qXABF@r<c@MN_GpvFB}j^hdEN=di-!65mGb7{%O#BbU!p
z`DuPa8~YYz<TnI46Aie2@8m2QIR;r&h&8+cP?(V36P=4U<m}sZA#(^vLs5RBoLP3V
zwrn4zs9W@Z5<{dF$K?q9*sFtV_D$}vLbBiejd#=1R294c6v?th55%-wD(Q}4_ms1A
zp7XoV&B_9i?IU>fMM2JMtQnwt%m@|@#lV&pwH+Mv7*UN_w3R=cC;dDZPl5--r2oS7
z5_eD9rW**0>qdZ3;E;%bXG3<T8sPV0ovEE+G2p4~TnI5pbm0j3qJQ--9RTV7ZCZf8
zaLnUEz+{!nqtk})s9DsA;na8uguIZZ6CdFGx2LI%R0FTAaoOw~_5HUw{3z}j+<m7l
zwJvQB1XU{;>K*46t5ba{)N}-rm~kr2Ot<nl0DRy0`pcahx|xC^qPGiYF+N65g98b>
z(Ez=j@}gM&FM(NWH^2W`^#s>q^EVnkq<@0JZ6Sfu83=&Adr&W^GSxwH;&wJ*^vqt{
zWaFa>6-}1yX*<`uo+Pe&h~@fuF5p$_JSg7XbsC+wx81?SJt8iF%x0dz8d|6EnVgG+
zRAav7bkpL}Z3>{7*<+c=d6u8g!+%0M<Jtv!=foWLc=Wn6S)yP6X4X~X3ZA*D>}48!
z)MSQvk>!<AKWHm|a+^z0<j2txBd$1vWKJt<h8wm)8YV0@++btJ%R~yDYGGwxZ+Z|e
zkdz8IG+6cT%r%EcbML2Nsq&0j<zO~5Z6w@lF2UqqZ_dz!=+BFE_;cBhT#=mO3*$I@
zsu^+n^%d~wEA47JcB5fP{ow)1ICRH<sP12qxudsdujPSe^JIu7<MPnX<gA5Z99}O*
zr|dUhA0d@b!<cOZFJA1;sWa!*6DfdGv3PhDD1Jq<r!a?vUT&pt69J~{4&pMo_v1C~
zkCvMRLm78zvJn}RRq%l@7cJQO2Zu1YC|4LWRW{wLe9Q^|!CYG=5g1fVBf+=q@xHUA
zMg?B(aMIEg3o#E$JI!WiU%Sz{9CT}>)H8m)Z@T5`{P%q6hTW$28Lf)EwtS_W@)S~#
z{=UQefv`Jry!0L5KomOV9n>d_Km8k)0@?Ob4O4R8&h66@zh=$@QS%m0TgYm!kWl6%
zRL8CM@{2izR_CDzTN#@{(WkPlh#{AEx}$-syCRi34MvpP?UrjhcTdago2Fdr)@Al<
zX3oLSoazv{SJo=MW4E7PdDHG5I+X>MMbZ*Vz4Ph#%5H}i=MTnw6{Q2O0lyQRfWwq(
zAF&vxn6%v6O2&4aaFx@QZ)xl<i_yf<Z07a4{q`S+9|RADX$Cbz(5B%+!?$6ezMA}o
zNMc*-Oug7Y`lo>|{HBmKdUnXApBgXmh581v?Nq}L^-u8tLOj{yi@KAYXq$g5Wm20m
ztiKRi6B<wDEUX3s#!)@FzoS#F|DuTwNG5S2K26TM%G#U`mvj_xd56=%UjoFMD*5`=
z7T^c-(CJgO(3;4OZ{NF|7mFgVE9<{pc0A=c^0Z?0lb4rDq|dSJU&9TLfm36fW1rLi
zaKLSF|9W&*t+yQa6{%hu56(Wch#!cN#2oQ@QSXSQKkI?K9cZU2hNB6-bc>@2zw2U+
zwl#d=r^XyL<Vo~WZnbxE4RLGl#6uEqE^X^`_uXIkvrDwU1X$yM@eyMyQMP2M5a@h)
zK#M#4-v)HfQo>N{80YbF`gVh756dbKtlc>!qUwtqEYg{=?S20Sb^WU7dF7ka(aN31
zQsQG}x73x6pd3^O+d&E{kKT9N_!5D!@z4E59n^=<*mJxO&zeIlxh)($I0=N4ukkcA
zFRC{5?~P{2Eg|%Q(J;E*ux!0$CWo(?7-a*wN$mIEfRL4@06?1|^2Y-W`ZZriwXfHe
zK%-dBoE`%XQt8-X<6$X{n~Q)IUp$kId0<G~x!s<qL9M+&JMg%dn*vy6No2mosuR}P
z&_pGnk?^;lSTTc-MbtMB;7X%`(U&v;=)(NOJ6zGPa?oc7)jK0@?t4c+lGGT^I3uF{
zC3N&Z*<1Z4URoU%n^%WKaD5Ste?IyiM{hYB$eZU{!&BJ`94CX(J*#=8tWOf?3<DIB
zX*@9B=Y(chJqw>Y`QHthr86PYZa~&>hCQ0?Yz#<~Oc{_@H9|KkVsU-kVaEH3cp`dO
zhrqM8=%4Z54pT`&R;}xX?p6aNX=qVu$nd;5L|1q+8$$uxcR$)Ga(j4B?a?jNqof7a
zukEo3Fv9Z+nDzU$H7hNMF`q=_htJ&$o@4F*)Rx>gDc{oUz;hD|$yVuk;-5zZ3rq2i
zSKquEdjUYE89gaGJm1JrSGw~S!P#6D_;hvY1_n@L_kiUv-lOS0L)gyPTLddUMtx|q
z!RcyHvEfuU`JD#iT=HMf2r2HUO>sswDNNl&bXw;{dllf^R5H%DMDfr@7uLXj(6qSI
z((50mz|FB=*r@foc?g?F$NCk^;!AIR`@r`i+>`*4oi&$m!=3-6MgN^oN6|m=eU14s
zYqdSk!<a4E<HE}fN%W*bVj-+evZ3HtoVn0$>EJ@Uz<m*<h?dV&)fT>Bp+CXy*?B8K
z(42hR)hs+BL1UL8k{g>rqFF*bQiys`X$$YOfosgJ;ox#W1M<`E0n2K2H<%BMmmvLL
zt3@YXTpJ-5v4qA85eePYE_Y16lDjs24x>YMuC|KZX8}3Ys1`%44lVZzlaEiXb6z!4
z*4!BjFxpJ+r1Dy0CFYq)g+cR$CA)y<own^}PJuU}(7zX`jj`+wrvnQ>oi)QsS*4O~
z=x&baKZzCB|2a+-t#$E2C^-ZXE8;Vl-qpGJm-Bqn<)OpZg^M(q^8sq5dw|-pw9n+Z
zw+B3Qf?iDIS@OwY%8gO;X$Kc?@Ax{X#1z@f9RNUWV-?STx#4MuE#@EHiX5NabjAeI
zaKDBVcwF}a<K4TTCjBOax<Ft8CHLZv*EzHIWkPm{1dy*&s1CCQ*Wao5<Abh%AXUWV
zF6L=Dws@#)C@-BjUd)jR0K#IkqY`%{q^Sc4Xmfl+w7NM!<vV0&WsL*!1L5GmL_~nr
zIzLv!#iSbd<PF-tm7x(QfA^A#qZIwbXFs?u5C$HLFKN7xA3sGg3AX`mwZ=lXofE)!
zN)TTG7~lM<)%q+`X(vOW2TN%4cZYW(U%`gukEhy~iCdCD#P;=TzEfnb)Ja3PaY7a7
zH#3DaLzzVzpl;(4m^%xPw9sDW*>j03?+8{vLp#63D`!kz`(l`>s7_>D0t=izs|Zx9
zy@PZ0n_k+ED1Q0`%~%kpTa7(GIY!-Tb(J%Y5MJbQ<<+gEnekckwxqx1o@p1t1^f$T
z%#CGeLp*$SHwB3=Xnr_&z(;~+{Vuw`!hEy9x^@c@7zS_eYM%@<Q3vII4?dTkfsg`U
zwP}u!wY<JuXXE0U@H4AXrNPJz2mH!lW<S=~X&1$SM-efywdU;Nq&^!3!tzU+t-Kwo
zl!L}I`x*=Pb^6?KViKD5;nE*ySONo;MQR#r<`lBQvJ(`n{nHbDHoye$Z{F$i2e92j
z3-o`BciIm73H>$Qe($g$6zhp;bGQT(>kTAtl}T4<i#gj>TBva0Z(~%C9+dezHkBmW
zKyi@142O5U_t%pv7xA`kb_4eb`semqNd@vkJ}63(X7WQmLWY{Gxh+XJKIf6_4eqT|
zFrPBhxZ*N<ZhR|?B9MtocO~d;vR`Rv#kI2v_<396`|(FKWnDGt-PC9RsZo<qcBE{*
zPKUX36Ju_(tv9z_%N!Zf{<<b-dkO5<O?<6hN-%r1)+OU`@%UgT(hDF(<w%%BgV=F9
z4HwR1u~iBpEu}*!Q_8soI1MEibulto8FHc2r|y9gl;u|RX}({9^281T!;g#q(OTWi
zsl>*;{om7Db{Vu$T0f?TJ>+>IJ84(`DFLCyVAwxl9ibu6j$7*pdl}DrYv0sl%}k2e
zDN5RrtPC-m-E8Xw1v@Mn&2CpK&z6~SA%uv8)Oj+ub`7Noxb)X#UPD;oD{FDqZx8R8
zn?B4z$Z^qQ9Kr$!2g4j#yclu*#RkstJwg_VhO1L`-_~?odjHabgXjGap=0IotpQBP
zr-U(Y2dMmTCY;h5i+8+1p+<5P2%jsqt=9x_0fue_1h6v!Jz>0G%i$$GvQ-v~i7zj!
z2>MD*$&mB3ja@ek!%?$^R%_bK#fnX2nJ9}ioW~Kk(3Yqsr}EOc+3zCQjRe-hhgLrx
zvzq{qPSxf@U`YfsKzz_lnHeQTo=Muk&B{tpqb_56TX;WKSCP82hcA6onM%+siG38T
z`IP_}-!yet4fG_+8BB#azoOdPS$wmL_}rxfK_1-;Jk@du86KXDF;Khg>fJF!T8uC(
zkg6jz!HqwYQn_d~A!ri%`PL}SG+BK&Z)Q6(0Hh>k?{&D#WAX}>_M03hAS>#BvO2^n
zWF2^L`;1f|V@RvRsTL*lk<$bv7A*TC0hrq0a|~@EJ>gILhdd9fZx3bUrC^{q7hwt+
zp1oNlBXVBMROQb1I78)z@J{fS_OTaC=tX7JnlK@l9B|Rk9{9xeMz7y)zHFGaX)A&j
zSoX1)Xe77-#aY@zkN8Oo-E(<>Cu9H$Ktp#0fQokugu+<~N?_%?tfoe_?bi8|kdQB)
zwCV6_`9C2vV%B=QohCN4auIr(%3qAqh0|u24rr@SRK*$hkIEUj2veD}38jY6l!zS3
z5FSs_T;MRCUocZ{fDzFv7x1rm#$di`WKKR569@#_k(dU%O$LadUpU;!g74YqCh+lB
zsk>Cj!~C@Y#>f4?Q4+3p3!5<77Z5Ab>f&-|kIoZ!cMJMQzZUXDvkscqsC;_a&#q<1
z`!y8T6)jOFf{~WqS=p%M8g=+X0`*REMlx)gzBn_<3mQh^C|uJf;0O~J8uAituQO2I
z9qvmBW!&1z=LH2hIq3WQjg;|Sp=bOGmd$HgJE#*EjSygldF*VIFCFm82Z)Mz3*>zQ
z%A;nXVuAmBmgnTmIF1+YtZd-a^u~m_qI;R>f8FZ5ZVCM9daQpY^Z`^(RUu&<rZ3|t
zbHSc)v5vcB7cK;#+{I8(W_4dvNY2wI4bYJF!y=PXL>Tb9!HXFzJK*AMFH|~z+bj4H
zHioB=w`)O5O~<P&rpNkh`ki-1q6;`QSDOVsWReWn>>x99_4ZF9?Xh4#CuzwU*6}J_
zXwmC`D0ed%SPYcTrHwN&{xGfaESHW`@}CZaip#Mtc&g|f1BRadH)ljhPg8FXVx7p+
z#F!Q_Z^RZa<;)4-Ks}^QgB>;fC2tY%%5CFXU$B+Y#n6lFZH}(?E9k?hrf{=x@U6Hv
zX64ucV445V6VZUc#2NgCm>WtIXgV&P^HKZACSRKh`Ja(GWi43<N!Xr?<qVe-+SIGt
z6>;>_oQ#AEUuF2iQlbu(&s4q^L!GmBu8jq@k&xKiQc|%UqlcPqdfy>XrLOD}VNq}k
z%-2Pb|5O5_95qoyDdW{#8WWRxs)F-Bp}Vu}waVvesqN{w7I^Gj7@s1i6A1sUaB2H-
zqT|}JV+&3xzvbe)RBu7^v+IY3*W1)V_h!**7RibkEOLVsly*3Uq&yyqu&<Jsno&t*
zGQL(3Ag;iFqWtHD00WsQz*mO(WwmNLl2bIM8(Q9Sm~Kv7H9}uA#avbCcoLQpHjtn!
z=k92O4w{9MB{Cl(84(guAjC>q>1ZgPTogtnjd+2Hifv|p0TgHyS}^{z`hPtKVjuXh
za<0~;B@wHQgpBt2v06Zwd?{{sUcl`5Ki}ZPnAB#iuF>G2%*2F~EHB8l1pnWk{P%Qp
z@!tUapC;n}#PmPc0<C?>|L4d4eQ5_8h%x@3PsIwaf&q2@uPgs^i>IAW(Es*P*{HV!
z|AryiYhUNMzt9H3Sv&$+ftI?Xv_Y$G5`t=>+b7z4qt2qzD~~MTv5N}=0utl_1p*R4
zk_TK)5@8NB!2JKs|8JSpHbP&8Y;Fz@AD*FtfYb$zdoqllq1q&R960KC)(cM&0pg?Q
zAspJe{MXuO`+t?Ax~jm18)5qG|J?4?_W_LS#Ps<&uM4;l2-oB)7w`mejQ~7~3V?%v
zh@eaZFHr{A|8LDZ#Q&oP?mpW0?ZQ{_6ypuQ_PZWi_68Jk1akyoa4ASd#%Nl*9CjpW
z8BLr2DtmolW;{n`DxLY_y{1iL3l1TJLuzsfL0!MM;&|pUK)&fc)?#z|817?xdc>=h
zMoBMliS3_o$(t4zAMY}COE}#R>}D3jI)no}n1Ml0|L>PW8L$Kb?Ekd_4^$3U6aT*O
z?|tU)|BqhK#`6O5-=D~d_2(dCFSrWn09S24vT?WrT_wy1J>1=mT6w;T4RC{C<W}+o
zjx{Kbk7PY0pY3k%pB?^<V+<1Tn<u0k{hZtBsXm64OIL<M;~GX;O_U7R78iG@G~Qtc
zJe+80U*qoH`%7M+wzdY|k`Hkn=pLi(;?s$0(^k{p&Vh+wMVeW^XBF*kcw?kN^68j*
zOL7LpXkwB-{P;W`W-;>E@NT58&3e(_!#v#!t6(XI37ut<&1aV{89kxw3@?OMt~l0z
z-Qm3#)qO~Ne9CIa2~#smzljy)2AjLV?`&so(0wn2Q3%N9UKe0xi}|o?S}rPt=*N>L
z#x;&vO$jFk+uI?82!*ClMnEj8B_wRNpc@{ki{1M2Q^Wzyr*HcJvJyTFYGyv@0H2Gf
z&WUN1{^%j)?^eo8;&gzstOi+3j_Vs4YFgYMuuftWV}+Pt^Mn2Gg(Gc*v(DR?5;lZ*
z67QTT`QoFnaYeYsQ<MBS>`P>3Og{afu=V`~c&?H(f}suU$%OJrousCiyVztrsBFCT
z;^yFEQ$^cf?wX^!M~2K%QLJ}ko5*Jn3@ku=bC0UUzhM4gD(vxAhOT9Ir=C;GUNQH8
z5eJ{SD=x<*bhT&?{Gl_?4PPo(vN4>s=OX0;l@lFadqkxu%ufVu+zN)2mpSjGYIGe8
zEM}_w>OzdIE@4K(27Y@XwyB5VGkVO5V>~5Y+1KZt*c20IJlsZynjD(JFwqIG{|Wy0
zu*Q?Q5+PGNp&nPx-ng?;ETNh3_|);3eBWQnzq#Y}q^6#O5gtS;OvxS5)B9Vp{pd6v
za7^3N1xSD>=a1WjS5_@L$;V^c(oSpdH3mO^>Ou40?JnHWb|+RPYnXW&A9Z#02Er%N
zwr<~Jc&UM|I^V?p9x^X1x=~ILc&oePm(hqK@7Y^yjA1oLON~_B3htyd_7XDG_{5M?
z*dAEJ!E4%v_U*4ffQ1;!?Cj|Sk0uuBLnH7}y`x=MR>qc?kQn&-UGI0|mT|!{7qc~$
zIwcwwxgBT}(%zl^k>%BLoY99MRE-t}VcyGLHqjFTPx{>a$RAMTV9|pzr&)s)W-!?3
z(AZRLUO+cb1=y)e;CjpQwNQBDVP*ST(owS5AcqHVj8#Zz@$vH;8Jvice-UKs2UN|i
zzxU78om0Y-9kPmEuROp*P;0&g4-HQAFXf{vzPXX4lKKkTeZu-aFkOiD`dLU1X@_z0
zJi4dkH?D#5;Q+?SHWLlrgBnM`p}sw1T_n6qD6j}Un324jXnqi)=8LQssP<8vjlFNj
zC5ShKla2SSvHpbBm|o>8oP6&~`PK86GWBW*YNNe)#)u}!eK`x;+$sbS5d<}%sA+%8
zS9LCF6R0%d81Y*4Y=w77NNAKySZ>9;=3+WzK}I)Wq3~VyZ5_NX*VEJ}?AU&ZIJlx;
zAb&99ujL1u4@m&7Fh~ARmMkJ+{%2<-K9q5cJT#G?#`{wVnR)IA!bB%NZ)faz@pr@b
z`0DtaW^$enXpU6v=e;4xNvSb84C3r0Y;Iad+Og^4U27v1X+pk}Ge41e*nj_-Eheh=
zIC?c<Q5+*?;{@;rQ&gsy8$v7Lr_|rPF*ha?I(z4d4cSprE4lyH?n7YXyU3}q77ea&
zwC;-<({`RjDftndbF;2M*MfwUll#>leKfWfy%A_tN$4}ak{Y)+|FH+&X3@<oLzTjP
zu*zt}1Y}$jGWi5FG?w}0f6i~k_M;Cunf`RR4Y7wHGdu1&kigBh#T6j-brukBTSt_`
zd@Di`%$^E>gl%xkz_({{8?Pf<&BIYx7_(ypri=33o$zgmBHfM9c>ky=udbma^n_8?
zTds+2a5;O9)SZE~2i~va5CbBZaNNul{IiP|41-m&Xs|(O+qfX5+)!RK%YkWh2zM(t
zO5jGM?&AK)C$*WJvWl2ssxT>ZZ8S4mu@_X<VdXFP{*d`B5JO@+k%gjnt_6iS!B)}w
z=mG19AK#Q-M2;aNG2S<h>_k;EIhiSkirCBG0{uDG1xOPO3!rslTX4s2V*}_gh=hvE
ztBR*I4i+s(R3m@$w7rV$1b*1vwG`GPkc-C&G2CigC(gn~0q)gnUW8fU16fWnwv{8b
z@J)cpCn7Z-aBKjBz<aP6S^Gi!9`vnIa_h@{JUlVXD?yc0;w!7Ny5Wwml2`E^2Ga%W
zJAN$gQiQ^%tX>>uJPPjMd+|29S)=GwnNyQ0JU%6wiQQnIy{J(FJQWL4H>!&E@}I~H
zFcpbPF$b{qcMy?{$OE`j>cGwpauzHizXbZi3kiGvEc)uVAtetI9#}d1whm(t6hT{y
znwK&YjK{my)h~z-$L}g+q6Pv!{iNzfb=Vt>7BmMS*}SNLmr=Qn4hi{Tr{$%E!+VNg
z?eMIUk~dWJdUbC%TiqCaA^n6<<U8#wVeEjG<{NIo{1%0Szxrg{cDmgE%L`^X4kVxo
z4e*I|5JHAt{4PtycL+B^fo1_hTFNqPW_BVbRKOJVC!Di_S!xtEg@F5E?*~ftdtlx5
z2);!v;rr(RWkw_T$G1K0X02bJ6o0Iauo0r9|L}#Bqk3+<{Q2gK>-oyyfurgU79bIX
zd&=qz?n8dpf`nMahkW>9W=G`OLsL>MVgUzc6ghnNVYh%qW}hEP^C!%lfd$FZ8Z}vx
zLh9s_E3LQfP2lDs%I+RP<N!-zn(GPOV%7(on~CI8W&9^=ZUO~E>CaE#I3kK<<MC+{
z_O`wKGpyL+61I{W(KoS>5}41g9ack>!;@nQO~>VZ$dCaocm%%5E4;pf%4%rm!yhCB
zEzQr;QY6OcsDq$5OWWI3E{N2w)JSho`1#@sHb&O#U>?x(>GQvG?^rEft{3ekKSh?~
ziZ1XP05C^<Ojxbt1lEn3(HOgek;Bu>9@4C}e;yi6Y-UT*5;i^a+AXIO6f-$sXTBw=
zn^-ric>#vS`iYQ`pHgfe`=~E5(U)Tj^JQMUuz*n17}v>d?*Q{loE2_TBPtg-(@(E+
z@UDtc=^fDBT5%^;GFQ%|22>RCHyo>Giz?A)O`TQd7}(g`*xx6YsOOOSMf(u5dClPU
zw;{3RA2CD9Mu*St==m?!(hm1CPr`_`;y6C&Q>#dqA~XU!2a>g(dEVt$;L1J-q57!a
z3LI%H`a6-;=Q3khW+G>oL-)xetG1@XJjM@$z3k3(MzrR!Tm4>Z4o~hwZdb74N<r<z
z1rFuz55HtMPF{MbPptZy$z{d><vU_%_@Ok>g$^6-c!pfjWA@?!Oo01q5&9CsR^$7{
zlKOK?+xz(vhqhxnk@{oQ1n12O9R`cWGFnvYd;&7U<1f<2y{#yM;TH!W+ZWL(sWMt&
zMg2UlKQ&!lyw*@~!v7-An{N(cI>Mu{!s~zFHHQxAQ9II_Rf}8vII0JAxD|UN8(;*E
zjVcPYGAL#*yZT)OozRi=v386!#BzmC&c(!-YPhMJ(TwWk1X~3W=m--dej=CSIlTK^
zA^-R4ikxX<ht#;&`8!qO4+Dz?KNN_BHjLZLK)R4kKDEiV$`~YdD!7Pd64hpN5Z_o_
zGu);f>7j%wG*+nTNxpPpPPKOr6CCNy-#?cl$EC<5Pw0H!6%D;DepsZPv0rA4pv-D2
zK8k*oTE*q_XqkqasDeJ;)<ltMd!1X3>`S(AN$U)doDc@zr>11VvR!m=9ZrHyY)E<%
zOOruO3Yq+jbM|Mqu2+>oOB0~+0%CxKI-KihB`-MY&*1@Z@d*JFe|49+1_`iFnkQ+)
zrSrBM$eRa44da(95SOKX>Kqw*luS|RX-D{oUyq5nckX5rIuHYJ_{@GKkBa@LGcyy2
zR1c_kHy~}A2D_YrvysTKUx1;zJMHuYZWfUh>&xt?0`-u|zNf4@QEBmV{O}~XRNhhF
z_?yQ>ID|7kyIXSO!#IQ@+oL%^!^R&L_KQSuEo&j|$_<#{v2M)8vEey#1Ph3WHzTSB
zRqby8cqE%L@GT~Gkobb9`7PfWS}Xu6)RO+y#i5O^{zJORoYM>G;*E3Z1H{m>EFLjH
zBI@pb<tpY_kh<Y0hE{du#kr)0>%FxxA{(1aFut0QEb$gXPgLWPVs1&};f1`PMo_*f
zoB~&y96M8%WAm3-%2ldN-M(nGuNa3@9t{m>_9uV><u5IX)qG53W}T0UKzs>D-UkMz
ze{?!LvwNZQlI1o8ZMOe}kqkeQSBc6IhxjdN81HW1BReeilc>c9o4N!3D{8*OleF<+
zZDPmR3U-4;kmO^e8%3<FRv4o&yDqIfnW!>X%P#|lRFN?`8;`N-zuk6v8UDy_-9|6S
zCG2cpr_FBSN%3@3ui~lxj6NoHz?d6@;^2zq&i*v<#ygXI5Z8DVpT~adpgwo~8lxEx
zAocGZSx0VL4mR(+Vk+m}9BSV}Nqhwjq}~or;t&<_J13zB3{IoTWbJ6=Fu6Wff{H5h
z{4%;!U06$Y&gNI$4elQqjs}(7cXq?e+?^f!qohOTuVjEexnp-(9@xS}aX<bZ8ZaTX
z?u2~J_8+1^v++^m5KLO!w_CYkmzdyi_pd}4Czf(aGre@WPtfyKCyc~^jh6He(ZLc`
z=+JxTVdKiY2c1nEO?*c&+cLcPP#K$xkIA@l>3AQ4oJAUljEbN8<|V?wN+vJ000Q@r
z?N?k$f>oBq+s*9>oM8Lv#Sx9Z0a+$<rn7$SZ1gRjG!g^e4R?*y6;n;vPIOC735-iX
z-X1QZ_a>Zi+}>V+cnik~Ezq84Rs}0hff#Bvm>@@_TG*;GOG>4n-z;No5H~BR<cwAb
zZ-^@TpYXKoV$5YRVPg7QK7XfSOXTb1CUzoYZA#lKdl+maD{1f$72Qca^kj89D+x4Q
zU~JH3yr#E7!1tJr7Fsyo#G93du$mWm>>7?#mRH<<vbfJE8cZP>qnATY2peZ=OAgC+
z;LWNT13Lu!EbIp8kV!tc@0{56#skyq1KlhUfBvRECfJsP;a&kA=3!{8bBpHlev?7^
z@eVD2-4VHQ2r=goM8Qpvrts!+;1F=GngrRe6Q8~#o*%7~hR>z_6@FMCCrsYIIy;#K
z4=#W64Ai>^((YnQDmqd>OOvt%Z(`&`lx{V9+$dPkv9+R{!*@$c6b|KnLgUFzY_!L{
z{7D&$xHhRz1mR{T^P{q-lqaJ%BIqj;PAE+-#A)OQFhwJ+AbGcZP=qEBn_PIq-(EQR
z9Cq+>fGfG<j<aVr1Wm2gKY<$EqXq=(0brvv-qBR}9fXj7>+X&#_bPKJEHT?Y=Cdc9
zI8E5-42pLpJdW?qJ-!u#NTOZzjKkB3(x?LE=ie|rH=|rbL;%Nj<1lhd3^(t?!CwH?
zi1l}_14tY4Py3#~9&y}0Q7#3gYAIFRi)y^?ghi#5Z=5I?Nc~Jqa8%nz)}qJri%!rI
zU6<%#j`1asvn7@g(X!hK+0$`Ms@fU8hV5fVK0iiu8~bzfok$qtR3gw5mGu*OvwecG
zOl1?uEJQJ(axYjV$ZHKN`g@hHAZ(Vg4O8@_jwLwqca2ifyUIuDOSEQZ#?0`|KQpn&
z4}A08MT1onl6y>Z<19R4Ve$<r>c88vdr&jXss2vtgjH{D4h~>&@}A-2o`R<piwC*f
zsM<tkz$1KTR>tX@@X#1`(2ClzGLMLH;tx$Qfp7TEasZRV62qJQ%U5QMiz^l}z#Zja
zBb0YdDK)yNzhbI!xmObi`>`Kmh_cmD4$e#F-{eYUG$V61ykdn6??80_QWAKp*hf&g
za;gcC5&X)nPxC8c6XOrcqOV{KkwOH5gIO=awcAUqSlp9n$Ls!U4Cz_tx^3+Q8HWeL
zaU>9tS9u-5hyyqTYZOj==ayNO`=H%O9$b;B*9dKxcKjytHqC(D;)-7!La=Jbv8JDL
ziUM&{Y+W-835o1jLctkkm5aMo>ciNB_f6Zs>DvxD4SD`-1PM-~*;6NySVPZlehUw-
zEi@Hd!AcRKC{WOgTaTao{Hx@K6-|sHw0mS`hi~=>TV9rk@0L5!nVna;vKFS|Pecfl
z+CgH=iM63PFnZBFvK|Caobk^8m35C8m%N1O?or#mQ|~d^(O9+z5<)l1L<Xv$y>Fen
zfz+M2+9M?|<Ik3@=y2eo062>`eV1jF80sP5q#<3gqmT+)P~JCSA_f5{hvDKRCVxWr
zB#GN(<#<i5^XA}l6KCfrcVgAml|o0sw0H`Epb)UIo#<f!;;ldP#gJ0n;NU!?5m+ck
z@EUxq??ATEVBT-a8FZw65+p)Lv;-?3wjDX-;M(K{!Cj+u<@0nuD-;@`dAI#GC}=^m
zWPOZL3vF0IMNBz7s|T-6nbZ*Kk3m?8&>PuixXFM&L=Xu0CYzT(fQQj|6`zMsHgOtY
z6j=vrQX{oz{;2|*TD<Qc#*SVzd_hIvi;hN{l#b3nJcTX<P?S;=Ox6s8fm<-cn6-oe
z=G|!+qM`FnM$JX!xd}Vk=TL=7QlH6|Q{1~c^&+QJZEw>Aao2B5vQhl*0aE_S_p!#A
zDn1`iiHzo*B+d%bk^RjB#P8e4s{8=d2@`=gdN1riF@9#-AkUt8jK?;(mL1IeD8i3#
zenhz0cX+|O1>cWj{f<hiDg3Sj^?V-CI_}vafi0Ooop;#8wj3Y^G?aAROkhmSBYzFl
z--h-!eh*^_ri}paYD+9$-X)IZu^Rh51M=``^>)e07SdNcvn79+n(2vu^PuHpDG#}a
z9pGrEcSnH~eX<*|c#k<T?fzKmetKHB-NiteOzzl+0h>v`?u&{OUGa66!JXTw>{o;P
zIv9grz)#yXHFnCq%_&#x3^-Rwp102HL4L+mi^_p1ee~rX<Yqp;<dfsJo6;+ZSOkDY
zLCK{&JB!Y&!WKv_&OK2uyfL2yaT1n#)@<w0RuSy{z{3<6K6+<om|Q`-AbDRUbAt3i
z?0#Oej`R-U9a`Fps42RvU{4IxpO_(0_{5b&r(;{t3$23gnwnxx3O47;bp+@JrDxB!
zU!j=s8Jz;7!g$tsA~$dZ&s|4JmfluBXS~huZtsN?+d_`s9%2L5ZGB?CQ<Y<<ADort
zGuINL3URf+ieVaENDSm*oq5l$?CzJ?bdVPpBMono9grbehwS9YrVEkDona6U??Oim
zGmw8tneH+_)8zoK_!W@}1zK2z){Tnwy_CyyXr#R9!mKFq314;_T}}XpQu#O+zNbjh
zwbyT=N895>EQt{&4-|v2Z(=Q^U$Y}~4c-4SLUSPK<cH3Vz;lVI%YcSY&e1Uqe{@sx
z0DDC?w$b7I=<oB<eMkcWUPR2@W+)k>+cn6@$AOzGTr<&DQf}tz_d>{5>z7bTcPd(I
zXpT>JKNBaKt{q}qo6LAg73ax)9rwUWhenY6Pi2oLM(!Cc`55+2FA=iY&g_tBk`8$2
z#Szvv6_6Q(g173cPF<FcFMcM!f>_M)S#L*9Zl##KACSW4ynPx13R>%b!WdfaeY!l(
zo*p46kp0ABt7$bp|HS;((78)MDA0+Ut2L5Qe4S-}kRzvHuoaC3kojcnb$h~*1CkHs
z8I>C4<&d!pf{JwW+Mt+eQYI0{w&*n<0859!)~Pjm{K<(56=4cB^|@|l9AF~47f9sg
z^OfNNg*=gj1S)cz<x$~DYVNhIioruP**xZJAXRC1?5I?~#`lvT6sy=A4IJJIDg{DK
z*1?DtJR_zm!zo**fU39ZI!Ol6Dc2uhbOCIM`ms4=cIH&fAbS$a-C9Y!iVTx^+s%Uq
zml3|wyaSn^&262NZ3C1t+Uzg-H%=0hTv)tY48NO;TYTIJ|M|MmB|5$->OJL1i4237
zucT_p%ztc6csf_;l`c86>5t+!`@HM-OlO6Rz9f$4QKE&E#C)pMZLd1atq*uJwNDu0
zF?n#s<bXCsVURd@Jirab{A25y@Z_#OR5CpA%{Hl$^oIk~Nw_|6h=q(-A!NL*s-!I7
zW-kbNK0X7#k5Xn5XzQVPemh~+vebh*9#Wo{b$_o2P~kU<2NiK*n5x6iTagZn&;;U%
z1d1oAthm<6+8e7W+|x2LdS<J8T21KEWrGI^0|wSf#j3$MXvz3q2fym<)lvK56oO@Z
zv{}@Nb*>n*#oUT9OfI}7T1b`(so;@@3A~eU2m!*Jl$&`I?I1De_=YV<fyxZfy>)*6
zX&QL~(G9hVgf&=)pFg=5=coAa9o~A*-^=FDnGDSF>SfX44EnPCojB!1VvcZ0%qV2H
zFW80dI~qY<LtQSl9ms;v0o3zdPh>~53j-|;P5XfWL%1VJk_>gIzXNgyJzkgs_c=fQ
zLpYOhq62dJftu@5cc>mVd+7<xbjrpA<~e_7>=5ATnwnv<4Lhn6=^&yn`Xhz^{C#3r
zo|kJ?MA-ZV7D>4N3q{lv4%65y6D2vC3zO)M<9G_n`%WnV9DAr1bQ;9OMr>Dk{Khj&
zILmg5Z8=-LU=c2UUKp*Awjw_vdQdfT1+p(6;O@=4>SnW}>M7C2nPBO6OHa$|j{J5Y
zQIa5#duBs}i2NkB<Wzswp``aEGqs8XOB~ipzM<=aAWhfXZ}1&PzI3v%eMNBA{GKRd
za~-s8smh?l+9R;GkeNX<Mf%5n_x7{tx}cS}Y)$v~)I{~%p=>UwYJ_QMI^^<mrhEN4
z<*dE8J8?;k<R82REMOjRJ{`g(OBaM5iv3VjD%zyhEtWDD{9}FpqC!PA%6%qrWaaW(
z(wcB~_q?U1YAPf4y2<w7tw}2-ueTffdS{-7&!mIIT=}8NgM%Uxkx!mLI*Te;jL*ot
zVe95Aom)aXWT&3^5Y;%n;mV!f=4t?5e-1rbwWrY=>?k`3eje!@_97Mvc4Qw`dWhtG
zlWQ0!4SPlaV<H)q_1z~qdqg@Tdj%J@8z%0hKdJC>>u(?7nly!ny$j=B*DrDV{Uj{t
z(end->vmSPyTn`W8v6p(qMJfKPcg<pDCM2-JRwz#LdQY&<!SglFJYC7QfML50PSCY
z&5N!oJdiM+H!S2B1UXH34>I#pxNDW&peg?MV5&^1l)|fd%qmj6K%=(I7s|MvSm#Sp
zTSkBaD2q<%BxOtfX}0fR)R-DZ<u>!jh_`5zDNHs?&6>(pO;9R5&@qDDU(zATfnk;&
zWlc{^ojeR`2VE*g(5ktuA>V^D8z|+zNTT5jxpVEC3LW{XBM6~HpXt38VN$a{XZPD6
z^rr#2Yb-7;*M$Th{t@ZjHicoD6RU}oJBaOz;tnBHBaTb;2#bApqM^%Iji6I>DTZbs
z5!D&LF|TPks~3LL*ygsdQXO9DHogv7mLflvfbTY)nlG&nttMd9@kwLGbnyiF&}HF>
zr3@gX-aa@*{}Id?1Q_c5Q?yQ;GA9IiNptYG;z)Y|QCixej2d8GwBrTu0^mHozQM;v
zMpr5kJ3>ruS#RS?+v>o}&x2(z&OWtL3kh=^g~q@{PxnZvDLhUq_;30_2To6tM#t`F
zjsoY+Jl9-m!U`ra*z{fS1xGBoC?izq0|R|^2iBDm;*~7Ani_vBtaSQ0%uGPhW*26p
z%*T_7(+WY2g?AaW=pzUBL5vQvFa^AyNHgTfIoKGg)7F)~R+E6B;P8i5W-R8e2o}z*
z+=R0l=Q<E$Lt^>BL+j@zfw%ct30Tb`87svG>bY<hb+3$p3Ag-paR&{@IRc#+cQt~@
z^aBE%OW{!5Qx@l7iyd*oEr)7Pg8YwgQ+qik?=+bm?9idP>*qDTj!mY1vcuj;8X+m^
zr)GtQ?M&8kfOzYaM6`Ce5ClqYC?h6wS{PF^J3q$M2EfV#Bje>4Vvf4qbY511xXH}p
zGz~SxU5db!LPbv4{+4P^?qfz&pG&;X03fC}m3b(bo1J}nRbo{$2En+0{@}v@oJJw4
zdJ)?jM(8mQvbSXCHZxN$4nEj~n>nPcs0&L7YV;iO)&~5@wtt4?8`tOvm>D;-7h+xe
z_3t}975ByfnoKAXsxP&Ai5Uy3cQ8KA?G6?48=);$Mmd^ffa#-`ZrJ?YC8A&=?-t+1
zCS^c$m2B}JURRt^vwcl(#{6eH5UajQ`d_QzP}v12vg)i%U?5+=)e&DuCv@PUvP-Ji
zN>7NnVEtgqd@U;;mSyyQh!s&!zN)RoceZN>mG9B3jba+&iaGaQ3u*j<W85+Xu#^hl
z8fP`=ybkeIJR_0vAP2}yLc>%09o5*Y?JL>q)qo^TeXw%qjdL8X?2#QAQMaxRYVzL^
z6JhkyYKnu<IE1}%bqV^;mB3D5LlDEIJln;#E6}m&=!=<e@aHh7%=ay5rV2I}FR%>1
z)N#h7CW1O^9b_6iVVlXOPje9^ifr78XRoL0IQT%g-dqwII|onj710;dL!Vt;Q|N%)
z36M{EukJ%{v@7RURgQm3O+S+TbKF{Z1e2JnAVL->!BuJ3u*s$K9`(ck<fL7Mg<k6K
zYtKq@u9*z{b)ibjLE1Wsx4&q2e|hFj$q%N(z-=Y>q|E)o;8zG}%Bt-SdejF+*X2aw
z&MZWT7NZv3y;In;q!=A}ka<IOF{eb?pmri&aJ}{vW4%C(?|ex4UW<9%?Q&7G;c!GO
z3p^HWWb9O;b5T%bPzM((hGOmaGn*Oh0QFpHfgeEBxy7OiZI-ig6}p1{>m5?%@7F3P
zLR{W2rT{zD1#Vr_H;4=n{xyq772+zqbj-WEL_RkHZo-xfPEZ9GV(hAdl&;OO$k`>h
zBuLB*h@_)^QchDX3dJ(AKL$!&ymLf}oN~Q_@BQ}6*VqI6;4mgKT~Z7YqvB_?*|_Ij
zV}`0dLOwR<D;)1hX8q1#2w%o9HC{s^@3Gl<xdehSKh122o%j93+*P2c6^aEQA9hOI
zL0`#Fp<bm?$vcvlHRF>%gX8mxPh0UdEy%;FAFAOn-AG!N;}kJ1=M=#YibM<qzC*3{
zQRScOOLJInaA+qCW|03&5XRC$So?!y5^VOi?HvL?X<-T14#vJTm#uYw$cgwc*C{!T
z+CpJ!e36Sq$Sf)q>m_Td@n-Aw`C{$$NpMs;Z)s*z^hHRa*(wo*g8|XGVss9~r@zK?
zyACh2IKxMSdOZJYXj5~aWR3Zz=+x`e8zqoP`MCBy^Tc}f56#(-;E*9(ZVxRdm)|Eh
z(3T7jVHAap;HbzlF1d4m=I!tu;v*}y6+zg=M&*x3K%MV{u1MMVDJNkDhu=snW)gOU
z`e!xu-GvYv8*)ilQoY#Kaq!15N>74c@7(;|*X5s5vp+KnAZ)z!Tvp8u2wFbT>TeSo
zlJu|)y8}0`JDS06FB%Gza?mrw+(GL*MCTii?CwqZ7$4>?Bd~b6hhtp#2EicySS5Ty
z&e}<a`2+4A^wQwz_;HzA5l4#aOb<bUWC?fGitZy|M0}RxSkGjl5Xglp05}CvF@9JK
zjTSB4aM|DRw#Rsg4`Lv0ld}sfu>oyOzl@li!|MA9^#<w}-fTboqy}l}igjNA!&HCd
zA<jL&NE?n%-^Ixv$091u(*_j~%IgRu8digeT!c+dM{hoX8`bStN=&w#M8y7>lJsWm
zvfN=}@Km~7RsZ=rNO$EKpWz*QjW{v`)oRyH<Hx3WyhPCRBZ+l*E)uP2Xx-qxnA%<x
z((H8i>Sxrd`%H25g^<P#$-xF^<>R6y9l8RxizBAO`cVaYOLYfTexAe8PB^0CDWe}!
zMDGhRYUC1#j<Do}CS|!hn8w4#wI;>K?8+0v+;TKg-ur!1!4pr{=mvCl{_x2#Y9?qY
zcL@`aOc|m&dyB2F_DB~-_pQy@n2DI^2J-S#>UD2hD4DE$$Xon4j_^RC$jZm~HT5yj
zS;NTH9(>AAcD~+bbrLiC>;-y7#rZ2Hk_Su*uGsZ@j(3+O#U2?O&qNyLTJ+>lM&V#(
zNqa=O-*f)Gw$p}-##aa=ZdX^7cs1H<lt+l5+k^}{BCOZ&jSoNSeu$lc8LkV>82g^U
zf!2r4U&8~wz~1N#-&o6yhK<T>@-gzSCs2`K7K4nNIBHDY;ivUfvck~tRp#gKtV{A&
zQUvRQy9U#w!idb3XShFBGm9|oV7;<MUCXY^5ln|k=p*ZL)u^gJ>aKYg(hK;IZ8>Ku
z3a>)^Ir`>Nvk+vqF(Xa{vfr`zv>0z5j24X-xM1110PYu0Ze|PMvwCHV2?<!8QCoj2
zL>ImO0SSKXQp9ol71O|eWE&n%3>d<p^9jz<H|;uj>dh9R9!GO<xj6K_31wWYjqoPU
ziYL5D%%jVob;KV^A*G~j;B34c?`>`=d2tqrN$6BWKfablc3%kr6%DgQSyA3wA{(4{
z?!EoQ|Es<442LV~!kt8rsA+l-M1%y<MHemFNHFSzAVinwZItLSAtbudhG8&D7>uY<
zqqotAh~9f|_xQehpXdI%KksvY-1F-^Gy9ygc3FF`_3n4AoyqOGla?tDom42V)MGOf
z=r1B`jJgc2CKpVwk)YdNlf0QOPkp9l`N||}Za#ov+7r}sjxgaY7#s^oB^a;^E^vD^
z!Li6P&*mw0_LdHn{y{iN%<|!5O1p2BbYs+aQUo1U`puZ2S%clKzsSQOiyfmBo41GP
zkOBGM7v<<=U;Yzh_m_Vy<zDxBMQZYyqCSyw=*&h?gQucplwUHibFJe~d+`Grh@WO!
zk5e~{2yV%1m&bFK>&X*{sZE=^TDm5}o2)VA3x6rY!f9B3cip*{%Z!!;j&3}3xGke_
zL?&vyMd2!jEN0#yK@dB00w$cAR_6NS`z$xX&0`+@MwVrVmR$^2Gos2cRdbh6ZOZoI
zD>pjl3vPS+s;D&D>Rbk3FFCPU__@H{TW3g(8}=Lu;{35&2VIx{4$@!GI^7LMg%^FH
z0QTB=ppv%R{fzp~54fKywQ7L$J-iDo`p@%vkhxQ1dIXV^j_t_iGau9$L6s5vQPjCv
zWAfBz&%+`^pWm>v{dp*fz-ZWpRXIL`$D@2FH#3zQAbs%LF5q}>lvNJMa(XnK{ra@0
zx#}1FiL<TrT|()t@5&4ItGe@9YZTG}%UN<i8lTw(PDU@wFEG6@C{f?brC1EsVommY
z;6rX~`e=Zl`Cibus3zby(c^<ZmlX21UI~kM+t1a`IU1b9&SpQY7DSm!hbL9uNqRh|
z_G~nS!^z<I=ipd*HF>$N`TNABq}V?(CG=?*gM6MFj=yuFm$WK1qDGIy5n7k;#u@Lk
z+y*Ba?J(Ec^(qBsTfQ=~`bS^R@kB(j{!sBf_R9APc`=jHq4}fdC;syXM7~sZ5&QUC
z=>AWeU=+~$bvB;kqAn~gZHi@NNjBD?Fw<Nv!w6p7BkS(Bp%PxW^zf*;$!85?^@h92
zE}P}Bd@~F!scD&~^(&gsg{u4y#Z>;>G7Dh|6M}A`W*3MdH+3^-FOQWUKBt;l25{vT
z>UY3=e9FK4>Sr-Csp&n0zn!9X1*P6k%7dBEBdO2BEB(n&TbX^xyJ?1Ne8vpDM;YXd
zzY<R~F^U-F>fAw|m_LBoX%p`<7R{D;nJvbM!7W{dY~Itq*feH+LDXlfbQ!MN8J;{d
zecSyPImH1(WF0)rA<B7~<HsBPCj9%K5CRG?|CEWpJbo)1Ld#=_RGs^|7(r2gOB4L3
zNGOp^hC!ctAcyS6=e*b*IAP$5Sc9yjOs@&3fdRNY)!x6p)vR0b{4k9Vg95G>=~a(*
zf<G9yv;l-Z3(5OyZDSt1e0OJQg0t8DJfOoAGn>}f+uJ_jelgLUm#mp5+0E-eKCPLs
zTZ*fJ4J+a#*#1<VbG46Z*xS}b5fvLi3zT`vG;Q~?nzB;h&G>sZBqf*nw9)CnCP@D@
z(*)ZhtiR)__vTbIQ9ZV5rAzmh-(6A@fx`*ajDL2%2doPR557{&G(2s;>?-843Jf%P
zaKsB7>BF9ryBz+Ky(R0E>-1kwdymI84!5+)8aV!~M+UOK67|q@&9`<Vpw7L&i&+_<
z-t3t^yFd$scBT>Y^G!}$ce`{4M9`J%>M)Si3)ZG64EzvWczwm-5flt|@o!4|@W$)U
zPp+C2;?dp0Qb@u%LHtaz-xVp5d@c3tp0g~f^LkS8UU@|6YK$55;=anXur%D|SjTt!
z=dw|ug}2eiHC_w(p7-T?GsyIevl<*y3rl=77orLY7creO8T^iCDxO8A|IKHJCjUMH
z4QwhlmSy3KG4F>d@;(8H93LI!y6!l{qF8aPvvcW%Qn3*$kxHQnSk$6@^s+-!9o?Iw
zI78U`Alp<Q>adH>=DEvc-#PlPav)bRv&+uBKXeG1O}V&)tr?-J7z;hR0I(@3;42f=
zmp;3L(2Eq>{JU#vBuZWOnAImuY5tKihO5IX-H}?824lPcM=ie1ID631%*CeF``pI*
zc9>~fQvH|5;i5wZUts(zZu`S@vrQkqRgQe7<Ccu<HTn6@U!hgVf(a~UjTRn`7}hpB
zO<7K=Pi7fCjI0vboe_E*H3AUlhO{Hj`J2S9+t}TfZX0zK{K_as{&AjsDEUo&rUZ5R
z`E~Vr1ui<w)8EVW@K%x)aGs=+kb5$!rH(EJN1T)rLSu+E623n>=|f9Vho75Gmz7)L
z=`tX_r1yUFul@E)apsY@TCavfi9}yev0vo3_g!OA;YoXEAJ=a>(q?QmWG{pp&5Rmb
zV-$y}{!x$M!)stj$j8%P-5m|WUa<Ao9!BZ;YCH-j-X{{vw#{{Pcaj3TcyFpB@98K+
z4%<RyL;fN4tRgu@;_tGiqsTgy0X#(fnCm`^+%O_MO@W_W*CTU{!dX_{DaEkZYg=It
zYj(kx<v7;<9N0eSGF4&KXc_2|(DM+g^Cf0eeN|5>;PRC>&K|9}%zfX_5_WNr>|1lm
z(aisHg(o;S5-QtJLL#F7+Owhc#m$1<kGo{-6f3FMfX#27I??`?Wnt9n+rL|zeG^eh
zOyMu&{bB|kO<EJrdf)gyD64$<i906h&4<@t^Px~lcnWdzaHc#kOmOJ{SKuyf6aN>P
zuj6-ypz@6gqT)2l@zT*0gH7$4^I2o8_P_pmtpV(-FV*bt9q14qebYpnw@nVZ>fSsW
z7h8uUhm>;vV}|$X9tAfx(eaT{P-oSXp1-wzrbH02GjHj;uV~dkpe|Cg9$Ly%>9#D>
zvu8XxV2v;a&vUN%Gk~DbJCdx85wE>1GUy21dF`d|K#kwRShK7YXeh*J<{atQLaOg|
z)8z3sXN;dDV56Xc?HV_l8@iduw5(71ry~}a@_5Ugsq!&VJl=et?`7(m83aw~4bk-X
zYU}VgP{zab`JdTfLZR1k$OMm%6jHXUtr7c2r?ZczUTuWtX&foZqd)(?r*OWssXU=M
zS!3=vY7_Izl=CMm!iU<+2jeTB^dKhbCOqU9Ka?QQ7gXmCSXo7V$@5sj7g>t{#cYG;
z)ja)@ys?NQG%;33z%85arl!*C2Q{BFK87iVW?lnmct|R2NY;z*nAv<qYsTy5e`!Ft
zyLWF16D`>QkJ;`pZ(;ty0jKmYLjb_DrVRUik6zxC%g8Zf=U+C7(RwRCSxnjA8B#I+
zb|^sfo1z*rRhV@nANKVkdB}5bO8)gpgHm9l%$^4ldl##pxslp|OpkvA0B4E@JU?V!
zS6O)845|g1ULhjRkcZyI3Yg7A<J#L$)$D7q#ycuasIbIm_MXc81r82uGN;7b?-p06
zdO-je*zVUyPqGF%{)=bYChcrlf~Ct|W=Q3)9;_{L(HvCar9hgz>u>>pU2QV~A47}^
zzdE2S{cB<-ok-5wr@$Trvo26J7MUZ<%8!S|1BbzPldE5ZSo2GE#t2!w^rd?T7g%ft
z81wEQ!uHGU$F+g%_$fX{7%y&hmz0V4$fq?y{dz>yvcWlK8{nf*@<WE<ha=;+Cg6nM
zM|=S3ye9(SoYI$Z^vY#@S|CaFQrS~E%lcNAW?g){m&J+HOsHfzw=as`IPtBsvvz?P
z!15MoMct!HW3cFo^3L-GGo)maF=-M>#y>LTQFtX7`!lBMzO&d2Kr5I)=f@}Z={8ua
zl}(AJ`u949HC?geMe$cWbOFf~dpTpuGxm^m0mJY=pWSb)@KI38ZIFoZ_xI@zMh6uq
zJ#m!6f4=wCfqD4v(L|I+LC@9GR(8s(8P5TMagjcXozGOWKOP2pJpaMd-;S%KJoTUd
zd!6m>t?RRjl~i^Bj-KnJYVe;BgMzJgAP8Yhe~iKN$b6=j_QNDT3RwP9VI!?rU(D?v
zH$zBk(Y3T~mjNf{r!2RxDmnUN_HR-t+V%X={j}e$MUPb;kfnfkWK|Sy{oGL0MU~jo
zpzE0+xF=EptpjLTaiMlXpZ2Z#_Wq!esElHwuQ^pQL91L!ZFq)Ll7GKI7I4B#K+8SE
zPw43_)>1lzK>aSaV<|S@d7db2kR&~5VzeHj*F~0HbHH2kD7anh<*yWS!~og}8bU<L
z+`p<sDR<ld9vHzzaV6z{;te$Z0;t@7Z|Q}A{o4O|Yta>aN&Vkwk7XmKk@(haeH}Jo
z&~-Me=cC)-0HWalj0@%f(?|5UfK!|Q-gzFbdi@q#CQde-{?WQ-9Ep9&MLm}(^nT8D
zZt#?mPqETfBoGMHKs<957X$*)1z&{&fk4pPSHVFb&`0_I2MXLd8x{j(QPvPFD79T|
zG?EMZBk8|vt*YbIwEaD}qX$th0s70v0mRG%6x$uY-hR2*_+?k*_wF1{)SaR&((Jbo
z=&w6au?EW0EOyQpkYgs>6J6>&Vbn`Sd|zc95MLAE6tP7C`WBn3W+|McU*af}DPGj|
zV}Lu~rD<<gvW1VLjr`8}anMMl)Y$M|(u}7-;YjaEJs?zH6!09<;Y4Hg*B5AYDlIp+
z_P|ac)%TOCLdMHudXn7y(l0Yd21+wO`PwFn|2jR1d(*Jjj%1<VI{;dp(shC);LW_#
z(z41qkoWe5|BRjOg~8>q(f36UuMgiAz*2~KDG@H9OP<>U^3$sZhMG8fX<7q$@l70f
zyd>6qep8kiYUSn#diVk{K!3@AUQ`~4tR#X&!!)J+wa$;JzMAqlj#I&O@~qOTZ-77>
zpsOOi<7sz^!ix0#sD?v7DLnu5mhM|e`^Xtj<#iCqnw|uNe5^g7HhzM)ebqK7fp|Uk
zrlhDnC)Uu;YY@G%T_ifPxdc{uyQ90QkKY^9+Nvl&WnteT0<k;<8XBI6mA1nP3guw-
zE5PHps(~`Kw3RTe1@>Kg<zKrzQHL&EbtrSCkj@=3U3XoI0>Op{g>nl0S*c$>yFp&&
zi&xXv^LicF7G{lxyc7{R%Mbn$jsDbDQ27#~x%eYV&|d^PtZ3HGPlu~v+<toty|#B;
zF*W4vPviy(st$+tzF<<TJg|N<^W`N3|2aR{Rp3H|5iSW)iX_9|x6Y#ls>0^Hn0^D)
za4zQwH2qyZyH??Q2HNgw{?Y=u-QR=r<!dKn>-&#6O?5g+$5(h{q9J?#V2h0y6@4%J
z!mKzLrxuMZ_1tmd05@~D=c(C6ycB*i0=0vv)F)qYed*&O5smXb-H|^PBiTDy-e+9|
zm%HX!m#(LVnvSW$3~DDDN6`1|JU5ERj4R*0RVPjEVR#HqU`A+KJAvt9rivKyZ7fa8
za-GLXKh0a5lovvw_-hx39PXJp9GpE8vQm`k&i5oVRy?aOBPYaZcv);j3qIm@)hW|=
zJ~2P`xtP2TVqv`Mh1g>YmoXgU&_b(&+c*@OzKE@MIvVIgpV=W=#QYVui>jC2%jcBn
z9b#QG<n*Z$gx~Zj%A?_;BM?g5*|S)4I`0|_-Ubr8<ZfHhf~M@LVzeVSwFIOILt>Xt
zOvAUjtm0xS?62@ghjYjKT=&?x>G&bag>0ys7@z4XX-XIw$aI@MF;H^hQnPf8?F)lI
zq-r=~wYkPe*EK45?-8G$DWAH)^wJbW9M-Hw?Jg)zQ`F{uwA;+Qa$-f_-&o{?$5WM8
ze7I<?F!A6iM%U;KjGGLonp+o>gQ`9lSl<94AH@i<Ue;tJPsEN8h+TZihHEQW)#vt{
z)eBCCAMDJR>l<%i+G7pR&sqlu=5hCGy^-(ucfTPyyHl4FWT6sYNLNm_F)#5fMZUiV
zBPwK1m==;CrQ}L#4k|s5DRK$y27{fNn04G`8ceg>YQI;O@XKG_NOg5g#l`u;eEaoo
zoAjiudUp2Z(V$Lr`5%|_Gn7k?+;p0~oyExfy@|Wr%Pqgy{8>Pt!&_Ghr(9B$$o)nb
zCZSR7@6+Uj$=5U#^a@MTDMrR+DM?>`oVbOCr}=N)w=w!h2go(@ofEUZ+#*RVHMjm+
z(+>67*S}gnk%{X?uN(6k`h?kfX;G}!4eG(O5zRvzwYOAO$+R>tADaD2_0rSAK+s`U
zci-0TCY$?(0l7*oOuf9B)*5tVvVABCg2Ojj{_q|rj!5S*sC$Z9?Ofw@UdSrQwtk}}
zvoDonvjdF`lTKYbKS9II#611;<=;kOnUkJa7Y=wCwI#1dZckmGC_DTTKdH6)*J-Kc
zSC2!NDLesNH?b>HdsBEX>oC(gGe6Cqdat?;#j?JCz)M+w;e94@5`PU;M|?Ff;q5uc
zM<I)-j*1@RDfQ1{_|JSkruw}7^GP`izAsR&YLby^xw?!(sU@*>-y7Qlt5LQ1skdJt
zl9Iv4$%xv<ySSbkDXR3>E4t50u{NpMUyuWdjPZm`Fhthi?s9971Nk-E#*36-%%*G6
zn@i_zB*Xf%mE9WLR+<#^Qfbd#63!hFxq6;IB{dudvApXL%`3xKa2ud(Tmv04Q8t0z
zy%&#%6ms^?;5PKf1m=Wuc}&Z~aIeEuVHZ9PkixvSiG}@!7<JA25-x96nxP^D6Ce%g
zzVuPDcMT@wpcV^;l#NYQvzx-7BDC4^L$xAO-J0Cb%*7GF3^%~di`8%cZn*x;<Wc|+
z`6ip(UtPIBP=|8s@Q^`yL>HxH-&oZLj+fO;XPEl&N_8+^&g8+!FOm6wvG?<N;`(<2
z_mQK#P|-u0UJV;Nqyy5A2l?Tx`nhp;(}~68EQ|$TpKf~j$q5XG6r3i@>RG7)N>)b;
ztc@8v6d;bKiY;tzYzC${u?>Q~>X$6c{wl-okZ?bLD~w;SyTgWvRekU7xS_rJre-=$
zeoup6Gyia5Z$e8g*OdjYq?k^(XVc<~Dit0l&CT~MbeVTAcO_hOjI6=YA>nFf#h+X+
zMr$-O#HSAUhOVEKWJ#;vUvsRkvmT>|lC7Hd7RQ!l+rIMIJ*iKZ@NY0K%%0ygdiE`P
zxtI&V+Ja=T{MG6`#$HR;YHH(gZ&U5n=J4L;VrZZPC}5NPKRV=b)26ihqhPxItzH%%
zkDL#bG~=Q5n4jJ{HL$UD#m17O8rr^t*cl%tKV}b_q}t+Hg5gjYzQ_4w299f(T8)CH
z=%uf0jP;cHf4D&VF4|%J(pge0#wuc*_!-<ao%mB&3pM5?Pi-bn8fuM&3cHF!rMlU^
z&)Fq&Im3)Z>&chLdN{z?FyeJ)cyELixSMixSD-?IsKFx|umP4r9BeAuL?$W*1#h{8
zM&F-s=VnrIK!A-VEZsJJsS;`{yhR<67}BoB;*o=XQ4EEz6v69Eofzx$%X#%$V8Y&p
zU2+1ue$ax*qFa4t{6uw6clnyGu;Vz`l<ZJSuAoV=amxjgx~lDd_|$z@(5!Qb-CUgB
zdI^t8t3MRxpagBf%zCf%tD3XwKo~UU^unuQTS_yXuPJn_CNT6Rb`?L;I#XMkW#X8`
zZ}&iaZAh0f<COt<`?7>}!OCHN98IqU&sUqX){_-r`2?&L30BtfVb9a#w5ew6{z;m%
zcouzJl~?DdLQlljY`nRj)?gVNjvE;%sUry1lEbJAC6b0uie@;57=5KvAfKRZgKD`a
zn;|41&(}w*EI_m)28$Ii8>iQDJSp_atbHoEVE>@zPFfR+vuZYBp(9tb<gwS3%fvd>
zxSQ#&*t`>zSA28SD79<FKFg(TCuZ?y*-(6|lc4)!C+=ihTx_c&v?Ctugez3BKrvC_
zyU1_bdwYwz>JW;fpL(D4d%w)i#!6$d-D3Ly^%V63HF$inoB%x+>>v>9s18&uTIn}d
z<07?1MimMim?nl-Q)WJh7a%u<FQ^K!K4;IzynGi5U5Gvs{Atpp9mOlp4;;q|?R@?^
z@6tH*!?vUwV2e|tL^Xsk)m6tc{xixP!ISyuw6K?e74)C=m%`Zc*}@B{Hr`XIFl`qm
zHCZA44KIt_;f|Hg(jG+ZHhIptPxOb}B0t($^EF4<ntmm0c|qBAHFiHRJ(psJjHsS4
zp*}wO(KK=0&eqbjcba<72o!Mo6&Pt8vr@J_FH6abgQDzR((Hp!-sTgUFri8cOBXce
zy1$n1pB(1hJXm*6Yd-<MJTb<iFsa>JiT86d>8LQDJ_z;JXKuBK>S%B}?1x_*#<|l?
z%KDWVVqCkouY-K~M|XiQAZJFrhWa>1M}{UN-07fogz4wy-i|L24}PyREK;jG5HMd0
z!T~z}E1s0l%{u+Gdhz7`)*_Lycto22dPOctC4~!_vFP_{jqHjX0<1Uu<oD=k&E0`0
z8^NTn7!eOGTX~q#a8K3(SlB-+u{P*X&FFNu$&~LJsEYJvFX*|Npw7Y_Np;e3EnIfT
zJke_C;x(?js$}@C?gq7)3R^E54T4bIXSb$+=}oz&;x64q?REa@HbY!>c;2#|@h5_F
z=bGI_YHkYoK0oi*#;mkTqEUJktq@xYu))1kgzZrrm>V&mD6dYsz%UdTY6X_mrv!oO
z$fX#F1Kv2lQ*F;oPV=i?XPTum0vi{-kPLP-L&MpI&~}qCrNfN8wmZIkOjYTQhKYpL
zWCvAj0M5s~%APl8e3)_47ivC1bz4l+X5-It_Dn||iZ+_0dhPFy!1=zS|Jl2XMxT{)
zDv+o6e`;xajoHZ7{7<3Z(R*+SU=tk;PY&Q_MlPV#V()6gb?{#}lRSpXdmncZ<*+*W
zI3(|vq-rLvRxkF;q?B7$lOM)ri|FJo0frvcGsW@N*CMb5&CLtc@y~5?r8VqE%8K7^
z$<ll|`*}Q;DV|iiq8G6G-J12?)sncIwrBQ&NkNm6S5AdMI$dFB*<$p5C##W^Zq+kW
zoyO|Nz59DeKR=80C+>+gEWg)Ue#fwEmQzMX{n)b|lzLiqcLka{xs}r7U!I*lZSd4D
zBGtr^%82+}58hdJYC>agoGObLYz0pszb6SftR?dQ6r1*Hj{id!U5$dIujKL;mth~}
z=!6CV{)~fBuC$7p_cmC^YggZa)o*ruKZFCbv6^0tO%=_x+f@!Vx%dhKNj+HeA_=&o
z<GyF6B<l<K0ontz%+eLvD5{(inZi9^qR22{#be7-dj*DUle^%I<B4Y^Ba=JnV>QlR
z*4OB{t9`hqFPVeLH>XQ}@rw*!aC%ZQzDw|PAuU5#9*3H$)E199qC?Bm1xJn(s6<jA
zGN@3~m{>QCJ*DQ`1Zpe5e)eM>IVk^_y_@$)NqxBDqiEZpk?PZNHz6>Dzw$@qQ%Q7W
zIHMo&>K9s(xO$@|)}p0KR!X{>a#u}b@KdOsj63rP2PbAzO*<oVzt>1&XdC|XJ5LCH
z>nU{jcML8_ZZ(V;<m+)YGQ6p*L=a^?h@HHd!G%z-JJ_fGO{o%&)ei<g=n!pvL9Q!d
z{6*hZpy$xIuivvnEY^Dse;E^u6D&H~#{J%osjj*{;diGKaj`NvSkuMS*T#oxZE8|K
z<Y8m(vTgiHGln)g1p93JLHNbG2eY|)F%tv}IyFm*7xmW5$qP<Aac`=8X#l%d`m|4v
z%_HAPqHiFulMs(!+v_C)O{zZweAbw9oY|D-%A3Jl%vZ_65)s(AQDx8RaZO)uwWYs8
zv_{P7#p;pSDW7EJ(D})(YC0ibw$PAE76xVct!Rma*7R}`+4+QigC0(c)~s}PT<#Ml
zlD|#=)8V%$2Z5-`;;wvMw|YN+R4%JX%KEX7oqX*WeUJSIBudMP6$f7=7$x|sA@<zo
z=g`0B!}NW@x#J|jDHZOH+#&lXXg^=Bw%sQSC@offN2XhP)Jwf>-ng8MFn7a{@Yei=
zuN%tCKYTSE){OLzm_j;%nmhbCTr8B#QV<ET%_o6`=hLovrTa>VEFmdD4bsZ<fB_F_
zDi({z;--w-ioXZ@oue=3YRd9<F#CV>1U(Q<LwhzZZ#PH0^D-7JPDUcb5>kzA16Tb+
z|4HRf`wgKhiCYsa05{3U(E;CUaWaeUJbk+S-ssZ1fA?p|`HS<5E~(?Oyfy2=9r+Fo
zLWOfC<fBY-e=)zC)VdpmaZ~X+XRYzBw+|?QG#mh#;1$ShFbeyzP`6cLO<$WgW9`Sm
zf-b#1+I_6wN!G?eYqt#nX%02zx&?zSvvqCex{4#`g9-@wm6OMfh3ecsHy|&)&oy+s
zFxetrj(jWeztwWf(|3~Db_LE)w4-I7#tHn$;BUp&`ie}UZ}+=KGiADb0{%h6X`>6%
z@hB}^6_u4+Y}I;N<uK$^x$C<lG)GRs2eP<tfK6?n)Jd8wQ|8yO;>X;(wChXV$O77W
zew=){J3W1J<qieu;B`%A6P)vR#2%&FUV9sbzWqjKK|}@%H(jas6MJx3;w-ZC#9(j@
z&CW@@njfr7@$@@=UZI2^TG0fol$m^{2_ecC?`VO9@3}7`xefH$b6=@~Ar~kf`8Z|-
z_<;2$;IO4~8`Ur_3{{Wc7W)%st2QU>hkmUYSZ#X^q975?L${e3fzZKsq1hA2h99N#
z$kSwXfRn@~RMe6s5_Dp8;<WNfbSb%6SC37B{tr?D`hTlZE${N$Zx$F@dH8dsL&mxS
zx5jd@KQ_QC!*x1Y2Jy(%Jn2%Z%J$>Ubi7)-l*)LXc+tP3jN+Rh&-SYo_2>(nC=7tG
zHl(kxC_DBmsEYIoassXZa^SbCix1HKtE=kaRrVJE&iVYm3M>FWY{>Cir)8u>!}g5<
zWRCv&)r+WfgS;ybFirq$^Um6!0$9cVPp|v;-*iFiBZ2=Z_t1a!4(*f3xhvo=^y-TZ
z->$GfdcaixffTObzjy6df>C%~B@1G^lI{8bFZ_S8HI$R(bJ{FOB0d2v4^mYED;B+c
H_xXPSYyVmf

literal 0
HcmV?d00001


From 5c9d6d6cd063eccce832e6b51ca783dcf501fb8f Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:09:45 +0200
Subject: [PATCH 02/28] fix links

---
 ocr-open-models.md | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 7634a799f0..8f47506075 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -298,29 +298,30 @@ Here is a simple method of deploying `nanonets` using vLLM as the inference engi
 
 3. Configure the deployment setup within seconds
 
-![Inference Endpoints][https://huggingface.co/datasets/huggingface/documentation-images/blob/main/resolve/ocr/IE2.png]
+![Inference Endpoints][https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/resolve/ocr/IE2.png]
 
 4. After the endpoint is created, you can consume it using the OpenAI client snippet we provided in the previous section.
 
-You can learn more about it \[here\](https://huggingface.co/docs/inference-endpoints/engines/vllm).
+You can learn more about it [here](https://huggingface.co/docs/inference-endpoints/engines/vllm).
 
 **Hugging Face Jobs for Batch Inference** 
 
 For many OCR applications, you want to do efficient batch inference, i.e., running a model across thousands of images as cheaply and efficiently as possible. A good approach is to use vLLM's offline inference mode. As discussed above, many recent VLM-based OCR models are supported by vLLM, which efficiently batches images and generates OCR outputs at scale.
 
-To make this even easier, we've created \[uv-scripts/ocr\](https://huggingface.co/datasets/uv-scripts/ocr)—a collection of ready-to-run OCR scripts that work with Hugging Face Jobs. These scripts let you run OCR on any dataset without needing your own GPU. Simply point the script at your input dataset, and it will:
+To make this even easier, we've created [uv-scripts/ocr](https://huggingface.co/datasets/uv-scripts/ocr), a collection of ready-to-run OCR scripts that work with Hugging Face Jobs. These scripts let you run OCR on any dataset without needing your own GPU. Simply point the script at your input dataset, and it will:
 
 \- Process all images in a dataset column using many different open OCR models  
 \- Add OCR results as a new markdown column to the dataset  
 \- Push the updated dataset with OCR results to the Hub
 
-For example, to run OCR on 100 images:  
-\`\`\`bash  
+For example, to run OCR on 100 images:
+
+```bash  
 hf jobs uv run \--flavor l4x1 \\  
   https://huggingface.co/datasets/uv-scripts/ocr/raw/main/nanonets-ocr.py \\  
   your-input-dataset your-output-dataset \\  
   \--max-samples 100  
-\`\`\`
+```
 
 The scripts handle all the vLLM configuration and batching automatically, making batch OCR accessible without infrastructure setup.
 
@@ -329,7 +330,7 @@ The scripts handle all the vLLM configuration and batching automatically, making
 If you are interested in document AI, not just OCR, here are some of our recommendations. 
 
 **Visual Document Retrievers**  
-Visual document retrieval is to retrieve the most relevant top-k documents when given a text query. If you have previously worked with retriever models, the difference is that you search directly on a stack of PDFs. Aside from using them standalone, you can also build multimodal RAG pipelines by combining them with a vision language model (find how to do so \[here\](https://huggingface.co/merve/smol-vision/blob/main/ColPali\_%2B\_Qwen2\_VL.ipynb)). You can find \[all of them on Hugging Face Hub\](https://huggingface.co/models?pipeline\_tag=visual-document-retrieval\&sort=trending).
+Visual document retrieval is to retrieve the most relevant top-k documents when given a text query. If you have previously worked with retriever models, the difference is that you search directly on a stack of PDFs. Aside from using them standalone, you can also build multimodal RAG pipelines by combining them with a vision language model (find how to do so [here](https://huggingface.co/merve/smol-vision/blob/main/ColPali\_%2B\_Qwen2\_VL.ipynb)). You can find [all of them on Hugging Face Hub](https://huggingface.co/models?pipeline\_tag=visual-document-retrieval\&sort=trending).
 
 There are two types of visual document retrievers, single-vector and multi-vector models. Single-vector models are more memory efficient and less performant; meanwhile, multi-vector models are more memory hungry and more performant. Most of these models often come with vLLM and transformers integrations, so you can index documents using them and then do a search easily using a vector DB.
 

From e2c20201c1fb8ed73d72b6d8b1d074a991d898b0 Mon Sep 17 00:00:00 2001
From: merveenoyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:13:20 +0200
Subject: [PATCH 03/28] blog.yml

---
 _blog.yml | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/_blog.yml b/_blog.yml
index d759f5659c..275f696b64 100644
--- a/_blog.yml
+++ b/_blog.yml
@@ -6819,4 +6819,16 @@
   tags:
     - datasets
     - open-source
-    - vision
\ No newline at end of file
+    - vision
+
+- local: open-ocr-models
+  title: "Supercharge your OCR Pipelines with Open Models"
+  author: merve
+  thumbnail: /blog/assets/aisheets/open-ocr-models.png
+  date: Oct 21, 2025
+  tags:
+    - ocr
+    - vision
+    - multimodal
+    - vlm
+    - open-source
\ No newline at end of file

From e2469f1e7e9f90e17c11e893c08d504b7aeb632c Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:30:51 +0200
Subject: [PATCH 04/28] Update _blog.yml

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 _blog.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/_blog.yml b/_blog.yml
index 275f696b64..4f0f211be3 100644
--- a/_blog.yml
+++ b/_blog.yml
@@ -6821,7 +6821,7 @@
     - open-source
     - vision
 
-- local: open-ocr-models
+- local: ocr-open-models
   title: "Supercharge your OCR Pipelines with Open Models"
   author: merve
   thumbnail: /blog/assets/aisheets/open-ocr-models.png

From cafb7b9abc1f4335cfedd3e78f2caf5e70738486 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:04 +0200
Subject: [PATCH 05/28] Update _blog.yml

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 _blog.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/_blog.yml b/_blog.yml
index 4f0f211be3..5d017322b4 100644
--- a/_blog.yml
+++ b/_blog.yml
@@ -6824,7 +6824,7 @@
 - local: ocr-open-models
   title: "Supercharge your OCR Pipelines with Open Models"
   author: merve
-  thumbnail: /blog/assets/aisheets/open-ocr-models.png
+  thumbnail: /blog/assets/ocr-open-models/thumbnail.png
   date: Oct 21, 2025
   tags:
     - ocr

From acedfc2c6cd97662554c77f0cc569069e9a3925b Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:12 +0200
Subject: [PATCH 06/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 8f47506075..31add53574 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -24,7 +24,7 @@ You’ll learn:
 
 By the end, you’ll know how to choose the right OCR model, start building with it, and gain deeper insights into document AI. Let’s go\!
 
-# Table-of-Contents 
+## Table-of-Contents 
 
 - [Supercharge your OCR Pipelines with Open Models](#supercharge-your-ocr-pipelines-with-open-models)
   - [Brief Introduction to Modern OCR](#brief-introduction-to-modern-ocr)

From 58b7b66d13577e153ba03c75479b1383340ef61d Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:20 +0200
Subject: [PATCH 07/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 31add53574..0c4361ea36 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -52,7 +52,7 @@ By the end, you’ll know how to choose the right OCR model, start building with
     - [Using Vision Language Models for Document Question Answering](#using-vision-language-models-for-document-question-answering)
   - [Wrapping up](#wrapping-up)
 
-# Brief Introduction to Modern OCR 
+## Brief Introduction to Modern OCR 
 
 Optical Character Recognition (OCR) is one of the earliest and longest running challenges in computer vision.  Many of AI’s first practical applications focused on turning printed text into digital form.
 

From 0e984ae68269f1466c8675bc13228be86ee8f179 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:28 +0200
Subject: [PATCH 08/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 0c4361ea36..0ed18ba259 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -58,7 +58,7 @@ Optical Character Recognition (OCR) is one of the earliest and longest running c
 
 With the surge of [vision-language models](https://huggingface.co/blog/vlms) (VLMs), OCR has advanced significantly. Recently, many OCR models have been developed by fine-tuning existing VLMs. But today’s capabilities extend far beyond OCR: you can retrieve documents by query or answer questions about them directly. Thanks to stronger vision features, these models can also handle low-quality scans, interpret complex elements like tables, charts, and images, and fuse text with visuals to answer open-ended questions across documents.
 
-## Model Capabilities
+### Model Capabilities
 
 ### Transcription
 Recent models transcribe texts into a machine-readable format.   

From 9c735ca6da1c5007818bb5e03f4f046c4f76b52f Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:36 +0200
Subject: [PATCH 09/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 0ed18ba259..4e716e5ce2 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -74,7 +74,7 @@ The input can include:
 	OCR models convert them into machine-readable text that comes in many different formats like HTML, Markdown and more.  
 	
 
-### Handling complex components in documents
+#### Handling complex components in documents
 
 On top of text, some models can also recognize:
 

From 58acc606e9b1de93a4c20224840c6cab6fa83853 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:45 +0200
Subject: [PATCH 10/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 4e716e5ce2..307c864e67 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -60,7 +60,7 @@ With the surge of [vision-language models](https://huggingface.co/blog/vlms) (VL
 
 ### Model Capabilities
 
-### Transcription
+#### Transcription
 Recent models transcribe texts into a machine-readable format.   
 The input can include: 
 

From 2ddf2438e0b1ab596c4ff7fc5316eb7d2225ded1 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:31:53 +0200
Subject: [PATCH 11/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 307c864e67..a7790598f2 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -92,7 +92,7 @@ Similarly for tables, cells are converted into a machine-readable format while r
 
 ![Table Rendering][https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/table-rendering.png]
 
-### Output formats
+#### Output formats
 Different OCR models have different output formats. Briefly, here are the common output formats used by modern models.   
 **DocTag:** DocTag is an XML-like format for documents that expresses location, text format, component-level information, and more. Below is an illustration of a paper parsed into DocTags. This format is employed by the open Docling models.  
 

From 244bad3e64eda88a6b8f42a52d202381ad453d27 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:33:14 +0200
Subject: [PATCH 12/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index a7790598f2..127d302cde 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -108,7 +108,7 @@ The right model depends on how you plan to use its outputs:
 * **LLM input or Q\&A**: If the use case involves passing outputs to LLM, pick a model that outputs Markdown and image captions, since they’re closer to natural language.  
 * **Programmatic use**: If you want to pass your outputs to a program (like data analysis), opt for a model that generates structured outputs like JSON.
 
-### Locality Awareness 
+#### Locality Awareness 
 
 Documents can have complex structures, like multi-column text blocks and floating figures. Older OCR models handled these documents by detecting words and then the layout of pages manually in post-processing to have the text rendered in reading order, which is brittle.  Modern OCR models, on the other hand, incorporate layout metadata to help preserve reading order and accuracy. This metadata is called “anchor”, it can come in bounding boxes. This process is also called as “grounding/anchoring” because it helps with hallucination.
 

From 6b950a8e1612b1f3db604eea3b261e0508a8e743 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:35:37 +0200
Subject: [PATCH 13/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 127d302cde..1ad6c6958b 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -71,7 +71,7 @@ The input can include:
 - Image/Layout/Page number tags
 
 	  
-	OCR models convert them into machine-readable text that comes in many different formats like HTML, Markdown and more.  
+OCR models convert them into machine-readable text that comes in many different formats like HTML, Markdown and more.  
 	
 
 #### Handling complex components in documents

From 105020cb3c7c2c49a8405444c9b72b6ba21c9c84 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:35:45 +0200
Subject: [PATCH 14/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 1ad6c6958b..13fbfe083f 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -113,7 +113,7 @@ The right model depends on how you plan to use its outputs:
 Documents can have complex structures, like multi-column text blocks and floating figures. Older OCR models handled these documents by detecting words and then the layout of pages manually in post-processing to have the text rendered in reading order, which is brittle.  Modern OCR models, on the other hand, incorporate layout metadata to help preserve reading order and accuracy. This metadata is called “anchor”, it can come in bounding boxes. This process is also called as “grounding/anchoring” because it helps with hallucination.
 
 
-### Model Prompting
+#### Model Prompting
 
 OCR models can either take in images or a text prompt along with the image, which heavily depends on the architecture and the pre-training setup.   
 Some OCR models support prompt-based task switching, e.g. [granite-docling](https://huggingface.co/ibm-granite/granite-docling-258M) can parse an entire page with the prompt “Convert this page to Docling” while it can also take prompts like “Convert this formula to LaTeX” along with a page full of formulas.   

From 99518831ac6584042081cebc7bd4729524a3d22b Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:38:01 +0200
Subject: [PATCH 15/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 13fbfe083f..3ad8f1f05b 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -110,7 +110,7 @@ The right model depends on how you plan to use its outputs:
 
 #### Locality Awareness 
 
-Documents can have complex structures, like multi-column text blocks and floating figures. Older OCR models handled these documents by detecting words and then the layout of pages manually in post-processing to have the text rendered in reading order, which is brittle.  Modern OCR models, on the other hand, incorporate layout metadata to help preserve reading order and accuracy. This metadata is called “anchor”, it can come in bounding boxes. This process is also called as “grounding/anchoring” because it helps with hallucination.
+Documents can have complex structures, like multi-column text blocks and floating figures. Older OCR models handled these documents by detecting words and then the layout of pages manually in post-processing to have the text rendered in reading order, which is brittle.  Modern OCR models, on the other hand, incorporate layout metadata to help preserve reading order and accuracy. This metadata is called “anchor”, it can come in bounding boxes. This process is also called as “grounding/anchoring” because it helps with reducing hallucination.
 
 
 #### Model Prompting

From a1ddcde7e6c03decd8ef6caada37bcb7bd413b1d Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:41:05 +0200
Subject: [PATCH 16/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 3ad8f1f05b..1b319e7b51 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -194,7 +194,7 @@ And then you can query as follows using e.g. OpenAI client.
 from openai import OpenAI
 import base64
 
-client = OpenAI(api_key={YOUR_API_KEY}, base_url="http://localhost:8000/v1")
+client = OpenAI(base_url="http://localhost:8000/v1")
 
 model = "nanonets/Nanonets-OCR2-3B"
 

From a7c5928e79c4d3f581c1a9574ffe734429a07bed Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:41:19 +0200
Subject: [PATCH 17/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 1b319e7b51..765dff6653 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -169,10 +169,10 @@ While the past year has seen a surge in open OCR models, this hasn't been matche
 
 Sharing more datasets could unlock even greater advances in open OCR models. There are several promising approaches for creating these datasets:
 
-\- \*\*Synthetic data generation\*\* (e.g., \[isl\_synthetic\_ocr\](https://huggingface.co/datasets/Sigurdur/isl\_synthetic\_ocr))  
-\- \*\*VLM-generated transcriptions\*\* filtered manually or through heuristics  
-\- \*\*Using existing OCR models\*\* to generate training data for new, potentially more efficient models in specific domains  
-\- \*\*Leveraging existing corrected datasets\*\* like the \[Medical History of British India Dataset\](https://huggingface.co/NationalLibraryOfScotland), which contains extensively human-corrected OCR for historic documents
+- **Synthetic data generation** (e.g., [isl_synthetic_ocr](https://huggingface.co/datasets/Sigurdur/isl_synthetic_ocr))  
+- **VLM-generated transcriptions** filtered manually or through heuristics  
+- **Using existing OCR models** to generate training data for new, potentially more efficient models in specific domains  
+- **Leveraging existing corrected datasets** like the [Medical History of British India Dataset](https://huggingface.co/NationalLibraryOfScotland), which contains extensively human-corrected OCR for historic documents
 
 It's worth noting that many such datasets exist but remain unused. Making them more readily available as 'training-ready' datasets carries a considerable potential for the open-source community.
 

From 7413fcee0b184f03a8d6dc35328316565353c0a1 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:41:29 +0200
Subject: [PATCH 18/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 765dff6653..12fda2fa10 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -165,7 +165,7 @@ Most OCR models are small, having between 3B and 7B parameters; you can even fin
 
 #### Open OCR Datasets 
 
-While the past year has seen a surge in open OCR models, this hasn't been matched by as many open training and evaluation datasets. An exception is AllenAI's \[olmOCR-mix-0225\](https://huggingface.co/datasets/allenai/olmOCR-mix-0225), which has been used to train at least [72 models on the Hub](https://huggingface.co/models?dataset=dataset:allenai/olmOCR-mix-0225) – likely more, since not all models document their training data.
+While the past year has seen a surge in open OCR models, this hasn't been matched by as many open training and evaluation datasets. An exception is AllenAI's [olmOCR-mix-0225](https://huggingface.co/datasets/allenai/olmOCR-mix-0225), which has been used to train at least [72 models on the Hub](https://huggingface.co/models?dataset=dataset:allenai/olmOCR-mix-0225) – likely more, since not all models document their training data.
 
 Sharing more datasets could unlock even greater advances in open OCR models. There are several promising approaches for creating these datasets:
 

From d7cc8a0fa12dd82a4c532a68acd47dfe1f5e7a4f Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:46:53 +0200
Subject: [PATCH 19/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 12fda2fa10..9c1e70f9b8 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -138,13 +138,13 @@ To make things a bit easier, we’re putting together a non-exhaustive compariso
 | [DeepSeek-OCR](https://huggingface.co/deepseek-ai/DeepSeek-OCR) | Markdown \+ HTML  | Supports general visual understanding Can parse and re-render all charts, tables, and more into HTML Handles handwriting Memory-efficient, solves text through image | 3B | ✅Supports nearly 100 languages |
 
 Here’s a small demo for you to try some of the latest models and compare their outputs.   
-\<iframe  
-    src\="https://prithivMLmods-Multimodal-OCR3.hf.space"  
-    frameborder\="0"  
-    width\="850"  
-    height\="450"
+<iframe  
+    src="https://prithivMLmods-Multimodal-OCR3.hf.space"  
+    frameborder="0"  
+    width="850"  
+    height="450"
 
-\>\</iframe\>
+></iframe>
 
 ### Evaluating Models
 

From cb5a45171f4348aa7c3c607327a70965fde8a719 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:50:33 +0200
Subject: [PATCH 20/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 9c1e70f9b8..1b65fbbf01 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -320,7 +320,7 @@ For example, to run OCR on 100 images:
 hf jobs uv run \--flavor l4x1 \\  
   https://huggingface.co/datasets/uv-scripts/ocr/raw/main/nanonets-ocr.py \\  
   your-input-dataset your-output-dataset \\  
-  \--max-samples 100  
+  --max-samples 100  
 ```
 
 The scripts handle all the vLLM configuration and batching automatically, making batch OCR accessible without infrastructure setup.

From ed0e841f36faad5ad75ea550bef704d559f0d798 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:51:04 +0200
Subject: [PATCH 21/28] Update ocr-open-models.md

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
---
 ocr-open-models.md | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 1b65fbbf01..1d08ab5ffa 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -310,9 +310,9 @@ For many OCR applications, you want to do efficient batch inference, i.e., runni
 
 To make this even easier, we've created [uv-scripts/ocr](https://huggingface.co/datasets/uv-scripts/ocr), a collection of ready-to-run OCR scripts that work with Hugging Face Jobs. These scripts let you run OCR on any dataset without needing your own GPU. Simply point the script at your input dataset, and it will:
 
-\- Process all images in a dataset column using many different open OCR models  
-\- Add OCR results as a new markdown column to the dataset  
-\- Push the updated dataset with OCR results to the Hub
+- Process all images in a dataset column using many different open OCR models  
+- Add OCR results as a new markdown column to the dataset  
+- Push the updated dataset with OCR results to the Hub
 
 For example, to run OCR on 100 images:
 

From d2a41ad73c6549b213dd831a788473467934fcaa Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:51:12 +0200
Subject: [PATCH 22/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 1d08ab5ffa..03af7907b7 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -15,7 +15,7 @@ authors:
 
 TL;DR: The rise of powerful vision-language models has transformed document AI. Each model comes with unique strengths, making it tricky to choose the right one. Open-weight models offer better cost efficiency and privacy. To help you get started with them, we’ve put together this guide.
 
-You’ll learn:
+In this guide, you’ll learn:
 
 * The landscape of current models and their capabilities  
 * When to fine-tune models vs. use models out-of-the-box  

From d70bccd96fe06c301158076adace4f29bfe5f433 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 17:53:43 +0200
Subject: [PATCH 23/28] fix

---
 ocr-open-models.md | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 03af7907b7..1dd16ac82e 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -298,7 +298,7 @@ Here is a simple method of deploying `nanonets` using vLLM as the inference engi
 
 3. Configure the deployment setup within seconds
 
-![Inference Endpoints][https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/resolve/ocr/IE2.png]
+![Inference Endpoints](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/resolve/ocr/IE2.png)
 
 4. After the endpoint is created, you can consume it using the OpenAI client snippet we provided in the previous section.
 
@@ -344,6 +344,6 @@ If you want to learn more about OCR and vision language models, we encourage you
 
 - [Vision Language Models Explained](https://huggingface.co/blog/vlms)  
 - [Vision Language Models 2025 Update](https://huggingface.co/blog/vlms-2025)  
-- \[PP-OCR-v5\](https://huggingface.co/blog/baidu/ppocrv5)  
-- \[SOTA OCR on-device with Core ML and dots.ocr\](https://huggingface.co/blog/dots-ocr-ne)
+- [PP-OCR-v5](https://huggingface.co/blog/baidu/ppocrv5)  
+- [SOTA OCR on-device with Core ML and dots.ocr](https://huggingface.co/blog/dots-ocr-ne)
 

From 3e6a3654a7cbe775370a93541130b005fd14acba Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 18:00:49 +0200
Subject: [PATCH 24/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 1dd16ac82e..540ab8e1e4 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -270,7 +270,7 @@ print(result)
 ```
 
 **MLX**  
-MLX is an open-source machine learning framework for Apple Silicon. MLX-VLM is built on top of MLX to serve vision language models easily. You can explore all the OCR models available in MLX format [here](https://huggingface.co/models?sort=trending&search=ocr). They also come in quantized versions.  
+MLX is an open-source machine learning framework for Apple Silicon. [MLX-VLM](https://github.com/Blaizzy/mlx-vlm) is built on top of MLX to serve vision language models easily. You can explore all the OCR models available in MLX format [here](https://huggingface.co/models?sort=trending&search=ocr). They also come in quantized versions.  
 You can install MLX-VLM as follows.
 
 ```

From 74b59765c0d7eef76510e06f074761bc3ec49b1c Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 18:01:04 +0200
Subject: [PATCH 25/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 540ab8e1e4..20f28a4bef 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -137,7 +137,7 @@ To make things a bit easier, we’re putting together a non-exhaustive compariso
 | [Granite-Docling-258M](https://huggingface.co/ibm-granite/granite-docling-258M) | DocTags | Prompt-based task switching Ability to prompt element locations with location tokens Rich output | 258M | ✅Supports English, Japanese, Arabic and Chinese. |
 | [DeepSeek-OCR](https://huggingface.co/deepseek-ai/DeepSeek-OCR) | Markdown \+ HTML  | Supports general visual understanding Can parse and re-render all charts, tables, and more into HTML Handles handwriting Memory-efficient, solves text through image | 3B | ✅Supports nearly 100 languages |
 
-Here’s a small demo for you to try some of the latest models and compare their outputs.   
+Here’s a [small demo](https://prithivMLmods-Multimodal-OCR3.hf.space) for you to try some of the latest models and compare their outputs.   
 <iframe  
     src="https://prithivMLmods-Multimodal-OCR3.hf.space"  
     frameborder="0"  

From 9dd85fe0c0710c9424ab2d97cc4ab9f378b75ae3 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 18:03:39 +0200
Subject: [PATCH 26/28] Update ocr-open-models.md

Co-authored-by: vb <vaibhavs10@gmail.com>
---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index 20f28a4bef..a43693c2af 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -115,7 +115,7 @@ Documents can have complex structures, like multi-column text blocks and floatin
 
 #### Model Prompting
 
-OCR models can either take in images or a text prompt along with the image, which heavily depends on the architecture and the pre-training setup.   
+OCR models can either take in images and an optional text prompt, this depends on the model architecture and the pre-training setup.   
 Some OCR models support prompt-based task switching, e.g. [granite-docling](https://huggingface.co/ibm-granite/granite-docling-258M) can parse an entire page with the prompt “Convert this page to Docling” while it can also take prompts like “Convert this formula to LaTeX” along with a page full of formulas.   
 Other models, however, are trained only for parsing entire pages, and they are conditioned to do this through a system prompt.   
 For instance, [OlmOCR by AllenAI](https://huggingface.co/collections/allenai/olmocr-67af8630b0062a25bf1b54a1) takes a long conditioning prompt. Like many others, OlmOCR is technically an OCR fine-tuned version of a VLM (Qwen2.5VL in this case), so you can prompt for other tasks, but its performance will not be on par with the OCR capabilities. 

From 947b862abc1a46d52c90d3bb7ec9154e4d804559 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 18:11:44 +0200
Subject: [PATCH 27/28] Update ocr-open-models.md

---
 ocr-open-models.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index a43693c2af..d477659263 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -90,7 +90,7 @@ Models use different machine-readable output formats, such as **DocTags**, **HTM
 
 Similarly for tables, cells are converted into a machine-readable format while retaining context from headings and columns. 
 
-![Table Rendering][https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/table-rendering.png]
+![Table Rendering](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/table-rendering.png)
 
 #### Output formats
 Different OCR models have different output formats. Briefly, here are the common output formats used by modern models.   

From 04c835b263120460c9404601f616ba932dce9360 Mon Sep 17 00:00:00 2001
From: Merve Noyan <merveenoyan@gmail.com>
Date: Tue, 21 Oct 2025 18:14:22 +0200
Subject: [PATCH 28/28] Fix image links and improve code formatting

Updated image links and formatting in the documentation.
---
 ocr-open-models.md | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/ocr-open-models.md b/ocr-open-models.md
index d477659263..5b0d1644ac 100644
--- a/ocr-open-models.md
+++ b/ocr-open-models.md
@@ -298,7 +298,7 @@ Here is a simple method of deploying `nanonets` using vLLM as the inference engi
 
 3. Configure the deployment setup within seconds
 
-![Inference Endpoints](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/resolve/ocr/IE2.png)
+![Inference Endpoints](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/ocr/IE2.png)
 
 4. After the endpoint is created, you can consume it using the OpenAI client snippet we provided in the previous section.
 
@@ -317,10 +317,10 @@ To make this even easier, we've created [uv-scripts/ocr](https://huggingface.co/
 For example, to run OCR on 100 images:
 
 ```bash  
-hf jobs uv run \--flavor l4x1 \\  
-  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/nanonets-ocr.py \\  
-  your-input-dataset your-output-dataset \\  
-  --max-samples 100  
+hf jobs uv run --flavor l4x1 \
+  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/nanonets-ocr.py \
+  your-input-dataset your-output-dataset \
+  --max-samples 100
 ```
 
 The scripts handle all the vLLM configuration and batching automatically, making batch OCR accessible without infrastructure setup.