{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "afe0f590", "metadata": { "ExecuteTime": { "end_time": "2023-12-19T23:28:37.651807Z", "start_time": "2023-12-19T23:28:37.650256Z" } }, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\"" ] }, { "cell_type": "code", "execution_count": 2, "id": "a62a5a27", "metadata": { "ExecuteTime": { "end_time": "2023-12-19T23:28:40.392479Z", "start_time": "2023-12-19T23:28:38.642489Z" } }, "outputs": [], "source": [ "import random\n", "import json\n", "import numpy as np\n", "import tqdm\n", "import torch\n", "import funcy\n", "import time\n", "import gc\n", "import tempfile\n", "import collections\n", "from joblib import Parallel, delayed\n", "\n", "from suno_utils.audio import Audio\n", "from suno_utils.tasks.data_loader import load_audio_mp\n", "from suno_utils.utils.text import write_jsonl, read_jsonl, write_json, read_json\n", "from suno_utils.utils.s3 import read_from_s3, check_s3_file_exists, open_from_s3\n", "from suno_utils.audio.conversion import convert_audio_files\n", "\n", "SAMPLE_RATE = 24_000\n", "EMBEDDING_RATE = 25\n", "N_CODEBOOKS = 8\n", "\n", "OUT_DATA_DIR = \"/app/suno/data/mert_25hz\"\n", "\n", "OUT_AUDIO_DIR = os.path.join(OUT_DATA_DIR, \"audio\")\n", "OUT_TSV_DIR = os.path.join(OUT_DATA_DIR, \"audio_tsv\")\n", "OUT_LABEL_DIR = os.path.join(OUT_DATA_DIR, \"label\")\n", "\n", "os.makedirs(OUT_DATA_DIR, exist_ok=True)\n", "os.makedirs(OUT_AUDIO_DIR, exist_ok=True)\n", "os.makedirs(OUT_TSV_DIR, exist_ok=True)\n", "os.makedirs(OUT_LABEL_DIR, exist_ok=True)" ] }, { "cell_type": "code", "execution_count": 3, "id": "49ff7b9a", "metadata": { "ExecuteTime": { "end_time": "2023-12-19T23:28:53.971758Z", "start_time": "2023-12-19T23:28:41.470982Z" } }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "25.0k hours of music\n" ] } ], "source": [ "def load_metas_simple(filepath):\n", " assert(filepath.startswith(\"s3://\"))\n", " data = []\n", " with open_from_s3(filepath) as f:\n", " for line in f:\n", " line = line.strip()\n", " if len(line) == 0:\n", " continue\n", " m = json.loads(line)\n", " _id = m[\"id\"]\n", " duration_s = m[\"duration_s\"]\n", " filepath = m.get(\"s3_filepath\", m.get(\"audio_filepath\", m.get(\"filepath\")))\n", " assert(filepath is not None)\n", " data.append({\n", " \"id\": _id,\n", " \"filepath\": filepath,\n", " \"duration_s\": duration_s,\n", " })\n", " return data\n", "\n", "metas = load_metas_simple(\"s3://suno-data/datasets/bundles/v2/music_sample/metas.jsonl\")\n", "print(f\"{sum([m['duration_s'] for m in metas])/60/60/1e3:.1f}k hours of music\")" ] }, { "cell_type": "code", "execution_count": 5, "id": "0d4a196d", "metadata": { "ExecuteTime": { "end_time": "2023-12-19T23:30:48.498024Z", "start_time": "2023-12-19T23:30:46.718347Z" } }, "outputs": [], "source": [ "with open(\"/app/suno/data/mert_cluster/metas.json\", \"w\") as fp:\n", " json.dump(metas, fp)" ] }, { "cell_type": "code", "execution_count": 5, "id": "6943cab3", "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "30.4 hours of data converted in 30.0s and written in 33.0s\n", "30.4 hours of data converted in 33.0s and written in 34.0s\n", "29.8 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 36.0s and written in 35.0s\n", "30.5 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 32.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 34.0s\n", "29.6 hours of data converted in 33.0s and written in 33.0s\n", "30.1 hours of data converted in 38.0s and written in 34.0s\n", "30.5 hours of data converted in 67.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 30.0s and written in 33.0s\n", "30.0 hours of data converted in 36.0s and written in 33.0s\n", "30.6 hours of data converted in 32.0s and written in 36.0s\n", "30.9 hours of data converted in 30.0s and written in 34.0s\n", "30.4 hours of data converted in 30.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.5 hours of data converted in 32.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 34.0s and written in 33.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "30.2 hours of data converted in 36.0s and written in 33.0s\n", "30.8 hours of data converted in 38.0s and written in 33.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 34.0s and written in 33.0s\n", "30.6 hours of data converted in 30.0s and written in 33.0s\n", "30.8 hours of data converted in 34.0s and written in 36.0s\n", "30.1 hours of data converted in 30.0s and written in 34.0s\n", "30.0 hours of data converted in 31.0s and written in 33.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 34.0s and written in 39.0s\n", "29.7 hours of data converted in 34.0s and written in 35.0s\n", "30.0 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "29.9 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 34.0s and written in 34.0s\n", "30.0 hours of data converted in 34.0s and written in 33.0s\n", "30.4 hours of data converted in 32.0s and written in 34.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.8 hours of data converted in 34.0s and written in 34.0s\n", "30.2 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.1 hours of data converted in 35.0s and written in 35.0s\n", "30.7 hours of data converted in 33.0s and written in 34.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "29.9 hours of data converted in 33.0s and written in 33.0s\n", "29.9 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.5 hours of data converted in 32.0s and written in 33.0s\n", "30.1 hours of data converted in 31.0s and written in 34.0s\n", "30.5 hours of data converted in 30.0s and written in 33.0s\n", "30.7 hours of data converted in 34.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.2 hours of data converted in 43.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 30.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 32.0s and written in 33.0s\n", "30.1 hours of data converted in 37.0s and written in 33.0s\n", "30.6 hours of data converted in 30.0s and written in 34.0s\n", "30.4 hours of data converted in 35.0s and written in 34.0s\n", "30.4 hours of data converted in 32.0s and written in 34.0s\n", "30.4 hours of data converted in 33.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.4 hours of data converted in 33.0s and written in 34.0s\n", "29.6 hours of data converted in 30.0s and written in 33.0s\n", "29.7 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 35.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 34.0s\n", "30.5 hours of data converted in 31.0s and written in 34.0s\n", "30.8 hours of data converted in 36.0s and written in 34.0s\n", "31.0 hours of data converted in 31.0s and written in 33.0s\n", "30.8 hours of data converted in 34.0s and written in 34.0s\n", "30.4 hours of data converted in 30.0s and written in 34.0s\n", "30.5 hours of data converted in 33.0s and written in 34.0s\n", "30.4 hours of data converted in 34.0s and written in 34.0s\n", "30.7 hours of data converted in 35.0s and written in 33.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 39.0s and written in 33.0s\n", "30.5 hours of data converted in 36.0s and written in 34.0s\n", "29.9 hours of data converted in 30.0s and written in 33.0s\n", "30.6 hours of data converted in 30.0s and written in 34.0s\n", "30.6 hours of data converted in 37.0s and written in 34.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.7 hours of data converted in 34.0s and written in 36.0s\n", "30.2 hours of data converted in 33.0s and written in 34.0s\n", "29.8 hours of data converted in 30.0s and written in 34.0s\n", "30.0 hours of data converted in 32.0s and written in 33.0s\n", "30.3 hours of data converted in 36.0s and written in 36.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "29.9 hours of data converted in 35.0s and written in 34.0s\n", "30.1 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 36.0s and written in 33.0s\n", "30.2 hours of data converted in 31.0s and written in 37.0s\n", "30.2 hours of data converted in 35.0s and written in 36.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "30.9 hours of data converted in 34.0s and written in 37.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 36.0s and written in 33.0s\n", "30.5 hours of data converted in 31.0s and written in 34.0s\n", "30.7 hours of data converted in 31.0s and written in 34.0s\n", "30.7 hours of data converted in 33.0s and written in 34.0s\n", "30.0 hours of data converted in 31.0s and written in 36.0s\n", "30.4 hours of data converted in 31.0s and written in 34.0s\n", "29.9 hours of data converted in 38.0s and written in 34.0s\n", "30.7 hours of data converted in 33.0s and written in 34.0s\n", "30.3 hours of data converted in 34.0s and written in 34.0s\n", "30.0 hours of data converted in 31.0s and written in 34.0s\n", "30.4 hours of data converted in 34.0s and written in 34.0s\n", "29.8 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 34.0s\n", "29.9 hours of data converted in 31.0s and written in 34.0s\n", "30.1 hours of data converted in 39.0s and written in 33.0s\n", "30.0 hours of data converted in 37.0s and written in 36.0s\n", "29.7 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 35.0s and written in 33.0s\n", "30.1 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 33.0s and written in 33.0s\n", "30.3 hours of data converted in 32.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 38.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "31.0 hours of data converted in 35.0s and written in 36.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 33.0s and written in 33.0s\n", "29.8 hours of data converted in 38.0s and written in 33.0s\n", "30.5 hours of data converted in 32.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "29.5 hours of data converted in 35.0s and written in 35.0s\n", "29.9 hours of data converted in 32.0s and written in 34.0s\n", "30.2 hours of data converted in 33.0s and written in 34.0s\n", "29.9 hours of data converted in 41.0s and written in 33.0s\n", "30.0 hours of data converted in 31.0s and written in 33.0s\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "30.2 hours of data converted in 36.0s and written in 33.0s\n", "29.8 hours of data converted in 32.0s and written in 33.0s\n", "29.8 hours of data converted in 34.0s and written in 33.0s\n", "30.6 hours of data converted in 32.0s and written in 34.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.7 hours of data converted in 32.0s and written in 33.0s\n", "30.2 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "30.5 hours of data converted in 41.0s and written in 35.0s\n", "29.8 hours of data converted in 34.0s and written in 36.0s\n", "30.2 hours of data converted in 31.0s and written in 33.0s\n", "29.7 hours of data converted in 37.0s and written in 34.0s\n", "30.6 hours of data converted in 37.0s and written in 39.0s\n", "30.6 hours of data converted in 33.0s and written in 35.0s\n", "30.1 hours of data converted in 30.0s and written in 33.0s\n", "29.9 hours of data converted in 37.0s and written in 34.0s\n", "30.8 hours of data converted in 31.0s and written in 33.0s\n", "29.7 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 33.0s and written in 33.0s\n", "30.2 hours of data converted in 34.0s and written in 43.0s\n", "30.1 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 30.0s and written in 33.0s\n", "29.5 hours of data converted in 32.0s and written in 33.0s\n", "29.6 hours of data converted in 31.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.8 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "30.4 hours of data converted in 39.0s and written in 36.0s\n", "30.1 hours of data converted in 35.0s and written in 35.0s\n", "30.4 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 35.0s and written in 33.0s\n", "29.6 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 33.0s and written in 33.0s\n", "30.8 hours of data converted in 39.0s and written in 33.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "30.4 hours of data converted in 35.0s and written in 33.0s\n", "30.5 hours of data converted in 34.0s and written in 33.0s\n", "30.9 hours of data converted in 41.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 33.0s and written in 33.0s\n", "30.1 hours of data converted in 32.0s and written in 34.0s\n", "30.3 hours of data converted in 33.0s and written in 33.0s\n", "30.9 hours of data converted in 31.0s and written in 33.0s\n", "31.0 hours of data converted in 35.0s and written in 34.0s\n", "30.3 hours of data converted in 35.0s and written in 34.0s\n", "29.8 hours of data converted in 38.0s and written in 34.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "29.3 hours of data converted in 31.0s and written in 33.0s\n", "30.1 hours of data converted in 40.0s and written in 33.0s\n", "30.7 hours of data converted in 36.0s and written in 36.0s\n", "29.5 hours of data converted in 33.0s and written in 36.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 37.0s and written in 33.0s\n", "30.3 hours of data converted in 30.0s and written in 33.0s\n", "30.2 hours of data converted in 34.0s and written in 33.0s\n", "30.1 hours of data converted in 35.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "29.9 hours of data converted in 44.0s and written in 33.0s\n", "29.9 hours of data converted in 31.0s and written in 33.0s\n", "29.8 hours of data converted in 33.0s and written in 34.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 35.0s and written in 34.0s\n", "30.0 hours of data converted in 32.0s and written in 36.0s\n", "30.2 hours of data converted in 33.0s and written in 35.0s\n", "30.5 hours of data converted in 35.0s and written in 33.0s\n", "29.9 hours of data converted in 32.0s and written in 33.0s\n", "29.9 hours of data converted in 30.0s and written in 33.0s\n", "30.6 hours of data converted in 34.0s and written in 36.0s\n", "30.3 hours of data converted in 30.0s and written in 34.0s\n", "29.8 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 35.0s and written in 34.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.9 hours of data converted in 34.0s and written in 33.0s\n", "30.6 hours of data converted in 34.0s and written in 36.0s\n", "30.3 hours of data converted in 35.0s and written in 33.0s\n", "30.2 hours of data converted in 35.0s and written in 34.0s\n", "30.3 hours of data converted in 30.0s and written in 33.0s\n", "30.0 hours of data converted in 34.0s and written in 33.0s\n", "30.4 hours of data converted in 34.0s and written in 33.0s\n", "30.3 hours of data converted in 32.0s and written in 36.0s\n", "30.4 hours of data converted in 36.0s and written in 33.0s\n", "30.2 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "29.5 hours of data converted in 35.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.1 hours of data converted in 33.0s and written in 33.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.1 hours of data converted in 34.0s and written in 33.0s\n", "30.7 hours of data converted in 36.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.5 hours of data converted in 34.0s and written in 33.0s\n", "29.9 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.5 hours of data converted in 34.0s and written in 33.0s\n", "30.0 hours of data converted in 30.0s and written in 33.0s\n", "30.6 hours of data converted in 31.0s and written in 33.0s\n", "29.8 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 31.0s and written in 33.0s\n", "31.2 hours of data converted in 63.0s and written in 35.0s\n", "30.4 hours of data converted in 35.0s and written in 33.0s\n", "30.2 hours of data converted in 33.0s and written in 33.0s\n", "30.1 hours of data converted in 33.0s and written in 33.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "30.0 hours of data converted in 33.0s and written in 33.0s\n", "30.5 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 40.0s and written in 35.0s\n", "30.9 hours of data converted in 38.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 34.0s\n", "30.0 hours of data converted in 31.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 31.0s and written in 33.0s\n", "30.5 hours of data converted in 33.0s and written in 36.0s\n", "30.5 hours of data converted in 32.0s and written in 37.0s\n", "30.7 hours of data converted in 36.0s and written in 36.0s\n", "29.7 hours of data converted in 30.0s and written in 33.0s\n", "29.9 hours of data converted in 37.0s and written in 36.0s\n", "30.3 hours of data converted in 41.0s and written in 36.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 32.0s and written in 33.0s\n", "30.8 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 34.0s and written in 33.0s\n", "30.5 hours of data converted in 34.0s and written in 35.0s\n", "29.9 hours of data converted in 35.0s and written in 36.0s\n", "30.4 hours of data converted in 32.0s and written in 33.0s\n", "29.7 hours of data converted in 33.0s and written in 33.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.3 hours of data converted in 34.0s and written in 33.0s\n", "30.4 hours of data converted in 33.0s and written in 34.0s\n", "30.2 hours of data converted in 33.0s and written in 33.0s\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 37.0s and written in 33.0s\n", "30.5 hours of data converted in 36.0s and written in 36.0s\n", "30.2 hours of data converted in 34.0s and written in 33.0s\n", "29.9 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 32.0s and written in 33.0s\n", "30.4 hours of data converted in 33.0s and written in 33.0s\n", "30.2 hours of data converted in 35.0s and written in 33.0s\n", "30.1 hours of data converted in 37.0s and written in 33.0s\n", "29.6 hours of data converted in 34.0s and written in 33.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "30.6 hours of data converted in 31.0s and written in 33.0s\n", "30.4 hours of data converted in 34.0s and written in 33.0s\n", "30.6 hours of data converted in 34.0s and written in 33.0s\n", "30.1 hours of data converted in 33.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 30.0s and written in 33.0s\n", "30.1 hours of data converted in 30.0s and written in 33.0s\n", "30.9 hours of data converted in 84.0s and written in 35.0s\n", "29.8 hours of data converted in 31.0s and written in 33.0s\n", "30.0 hours of data converted in 34.0s and written in 33.0s\n", "30.3 hours of data converted in 33.0s and written in 34.0s\n", "30.5 hours of data converted in 33.0s and written in 33.0s\n", "29.8 hours of data converted in 30.0s and written in 33.0s\n", "30.4 hours of data converted in 31.0s and written in 36.0s\n", "29.7 hours of data converted in 33.0s and written in 36.0s\n", "30.2 hours of data converted in 38.0s and written in 34.0s\n", "30.4 hours of data converted in 35.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "30.1 hours of data converted in 30.0s and written in 33.0s\n", "29.6 hours of data converted in 30.0s and written in 33.0s\n", "30.8 hours of data converted in 31.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 33.0s\n", "30.0 hours of data converted in 32.0s and written in 33.0s\n", "30.0 hours of data converted in 35.0s and written in 34.0s\n", "29.5 hours of data converted in 30.0s and written in 33.0s\n", "30.7 hours of data converted in 38.0s and written in 33.0s\n", "30.1 hours of data converted in 35.0s and written in 33.0s\n", "30.8 hours of data converted in 34.0s and written in 33.0s\n", "30.2 hours of data converted in 30.0s and written in 34.0s\n", "31.0 hours of data converted in 34.0s and written in 36.0s\n", "30.2 hours of data converted in 32.0s and written in 33.0s\n", "30.8 hours of data converted in 30.0s and written in 33.0s\n", "29.7 hours of data converted in 37.0s and written in 36.0s\n", "30.1 hours of data converted in 30.0s and written in 34.0s\n", "29.9 hours of data converted in 34.0s and written in 33.0s\n", "30.3 hours of data converted in 30.0s and written in 33.0s\n", "30.9 hours of data converted in 34.0s and written in 33.0s\n", "30.1 hours of data converted in 32.0s and written in 33.0s\n", "30.3 hours of data converted in 35.0s and written in 33.0s\n", "30.7 hours of data converted in 31.0s and written in 33.0s\n", "32.1 hours of data converted in 34.0s and written in 34.0s\n", "32.7 hours of data converted in 36.0s and written in 34.0s\n", "32.5 hours of data converted in 32.0s and written in 36.0s\n", "32.3 hours of data converted in 32.0s and written in 34.0s\n", "32.6 hours of data converted in 34.0s and written in 34.0s\n", "32.7 hours of data converted in 34.0s and written in 37.0s\n", "32.9 hours of data converted in 31.0s and written in 34.0s\n", "32.6 hours of data converted in 32.0s and written in 34.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "32.8 hours of data converted in 36.0s and written in 34.0s\n", "32.3 hours of data converted in 35.0s and written in 33.0s\n", "32.4 hours of data converted in 32.0s and written in 34.0s\n", "32.1 hours of data converted in 37.0s and written in 36.0s\n", "32.6 hours of data converted in 32.0s and written in 34.0s\n", "32.7 hours of data converted in 32.0s and written in 34.0s\n", "32.8 hours of data converted in 34.0s and written in 34.0s\n", "32.9 hours of data converted in 32.0s and written in 34.0s\n", "32.7 hours of data converted in 32.0s and written in 35.0s\n", "32.5 hours of data converted in 38.0s and written in 34.0s\n", "32.7 hours of data converted in 37.0s and written in 34.0s\n", "32.7 hours of data converted in 37.0s and written in 34.0s\n", "32.6 hours of data converted in 32.0s and written in 34.0s\n", "33.3 hours of data converted in 34.0s and written in 37.0s\n", "32.5 hours of data converted in 31.0s and written in 34.0s\n", "31.9 hours of data converted in 37.0s and written in 37.0s\n", "32.5 hours of data converted in 34.0s and written in 37.0s\n", "33.5 hours of data converted in 34.0s and written in 34.0s\n", "33.1 hours of data converted in 32.0s and written in 34.0s\n", "33.9 hours of data converted in 40.0s and written in 34.0s\n", "32.0 hours of data converted in 40.0s and written in 34.0s\n", "33.0 hours of data converted in 37.0s and written in 37.0s\n", "32.4 hours of data converted in 36.0s and written in 34.0s\n", "32.8 hours of data converted in 38.0s and written in 39.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "33.2 hours of data converted in 31.0s and written in 34.0s\n", "32.4 hours of data converted in 31.0s and written in 34.0s\n", "33.2 hours of data converted in 36.0s and written in 36.0s\n", "33.3 hours of data converted in 32.0s and written in 34.0s\n", "32.4 hours of data converted in 40.0s and written in 37.0s\n", "32.9 hours of data converted in 32.0s and written in 34.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "32.9 hours of data converted in 38.0s and written in 36.0s\n", "32.7 hours of data converted in 33.0s and written in 34.0s\n", "33.0 hours of data converted in 32.0s and written in 34.0s\n", "32.7 hours of data converted in 31.0s and written in 34.0s\n", "32.4 hours of data converted in 34.0s and written in 34.0s\n", "32.7 hours of data converted in 31.0s and written in 34.0s\n", "32.2 hours of data converted in 36.0s and written in 34.0s\n", "33.4 hours of data converted in 34.0s and written in 34.0s\n", "32.6 hours of data converted in 32.0s and written in 35.0s\n", "32.6 hours of data converted in 35.0s and written in 34.0s\n", "32.9 hours of data converted in 33.0s and written in 34.0s\n", "32.7 hours of data converted in 38.0s and written in 37.0s\n", "33.2 hours of data converted in 32.0s and written in 34.0s\n", "32.3 hours of data converted in 35.0s and written in 34.0s\n", "32.6 hours of data converted in 32.0s and written in 34.0s\n", "32.4 hours of data converted in 34.0s and written in 34.0s\n", "33.5 hours of data converted in 36.0s and written in 35.0s\n", "33.5 hours of data converted in 32.0s and written in 35.0s\n", "32.8 hours of data converted in 31.0s and written in 34.0s\n", "32.8 hours of data converted in 37.0s and written in 34.0s\n", "32.2 hours of data converted in 78.0s and written in 34.0s\n", "32.1 hours of data converted in 34.0s and written in 36.0s\n", "32.2 hours of data converted in 32.0s and written in 35.0s\n", "32.0 hours of data converted in 33.0s and written in 34.0s\n", "32.6 hours of data converted in 33.0s and written in 34.0s\n", "33.4 hours of data converted in 32.0s and written in 34.0s\n", "32.6 hours of data converted in 45.0s and written in 34.0s\n", "32.5 hours of data converted in 38.0s and written in 37.0s\n", "33.4 hours of data converted in 33.0s and written in 34.0s\n", "32.7 hours of data converted in 34.0s and written in 38.0s\n", "32.6 hours of data converted in 33.0s and written in 37.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "32.2 hours of data converted in 36.0s and written in 33.0s\n", "32.8 hours of data converted in 38.0s and written in 34.0s\n", "32.5 hours of data converted in 79.0s and written in 34.0s\n", "32.1 hours of data converted in 32.0s and written in 34.0s\n", "33.3 hours of data converted in 39.0s and written in 34.0s\n", "33.0 hours of data converted in 32.0s and written in 34.0s\n", "32.6 hours of data converted in 47.0s and written in 35.0s\n", "32.9 hours of data converted in 35.0s and written in 34.0s\n", "32.4 hours of data converted in 32.0s and written in 34.0s\n", "32.7 hours of data converted in 32.0s and written in 34.0s\n", "32.9 hours of data converted in 38.0s and written in 34.0s\n", "32.8 hours of data converted in 39.0s and written in 34.0s\n", "32.7 hours of data converted in 41.0s and written in 36.0s\n", "32.7 hours of data converted in 32.0s and written in 34.0s\n", "32.3 hours of data converted in 36.0s and written in 34.0s\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "32.2 hours of data converted in 31.0s and written in 34.0s\n", "33.7 hours of data converted in 34.0s and written in 34.0s\n", "32.9 hours of data converted in 37.0s and written in 34.0s\n", "33.7 hours of data converted in 32.0s and written in 34.0s\n", "31.8 hours of data converted in 31.0s and written in 34.0s\n", "32.8 hours of data converted in 35.0s and written in 34.0s\n", "32.4 hours of data converted in 42.0s and written in 37.0s\n", "33.2 hours of data converted in 32.0s and written in 34.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "33.2 hours of data converted in 37.0s and written in 34.0s\n", "32.4 hours of data converted in 31.0s and written in 34.0s\n", "32.7 hours of data converted in 34.0s and written in 34.0s\n", "32.7 hours of data converted in 93.0s and written in 34.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "32.9 hours of data converted in 35.0s and written in 34.0s\n", "32.8 hours of data converted in 33.0s and written in 36.0s\n", "32.4 hours of data converted in 35.0s and written in 34.0s\n", "32.9 hours of data converted in 31.0s and written in 34.0s\n", "33.2 hours of data converted in 36.0s and written in 34.0s\n", "32.4 hours of data converted in 36.0s and written in 36.0s\n", "33.3 hours of data converted in 33.0s and written in 34.0s\n", "32.3 hours of data converted in 35.0s and written in 34.0s\n", "32.7 hours of data converted in 39.0s and written in 34.0s\n", "33.2 hours of data converted in 32.0s and written in 34.0s\n", "31.9 hours of data converted in 31.0s and written in 34.0s\n", "32.1 hours of data converted in 33.0s and written in 34.0s\n", "32.6 hours of data converted in 38.0s and written in 36.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "32.5 hours of data converted in 37.0s and written in 34.0s\n", "31.5 hours of data converted in 35.0s and written in 33.0s\n", "33.2 hours of data converted in 32.0s and written in 34.0s\n", "31.8 hours of data converted in 31.0s and written in 33.0s\n", "33.0 hours of data converted in 31.0s and written in 34.0s\n", "32.4 hours of data converted in 32.0s and written in 34.0s\n", "32.6 hours of data converted in 41.0s and written in 37.0s\n", "32.2 hours of data converted in 36.0s and written in 35.0s\n", "32.7 hours of data converted in 32.0s and written in 34.0s\n", "32.8 hours of data converted in 32.0s and written in 34.0s\n", "32.2 hours of data converted in 34.0s and written in 35.0s\n", "32.9 hours of data converted in 32.0s and written in 34.0s\n", "32.4 hours of data converted in 31.0s and written in 34.0s\n", "32.4 hours of data converted in 38.0s and written in 34.0s\n", "33.0 hours of data converted in 35.0s and written in 34.0s\n", "32.5 hours of data converted in 34.0s and written in 34.0s\n", "32.7 hours of data converted in 32.0s and written in 34.0s\n", "32.5 hours of data converted in 35.0s and written in 34.0s\n", "32.6 hours of data converted in 32.0s and written in 34.0s\n", "32.6 hours of data converted in 31.0s and written in 34.0s\n", "32.6 hours of data converted in 32.0s and written in 34.0s\n", "32.9 hours of data converted in 36.0s and written in 34.0s\n", "33.2 hours of data converted in 35.0s and written in 36.0s\n", "32.9 hours of data converted in 32.0s and written in 34.0s\n", "33.3 hours of data converted in 35.0s and written in 34.0s\n", "32.8 hours of data converted in 32.0s and written in 34.0s\n", "33.2 hours of data converted in 38.0s and written in 34.0s\n", "32.7 hours of data converted in 32.0s and written in 34.0s\n", "31.8 hours of data converted in 37.0s and written in 36.0s\n", "33.7 hours of data converted in 31.0s and written in 34.0s\n", "32.0 hours of data converted in 37.0s and written in 34.0s\n", "32.5 hours of data converted in 32.0s and written in 34.0s\n", "32.5 hours of data converted in 35.0s and written in 35.0s\n", "32.1 hours of data converted in 31.0s and written in 34.0s\n", "32.4 hours of data converted in 38.0s and written in 37.0s\n", "33.5 hours of data converted in 37.0s and written in 37.0s\n", "31.0 hours of data converted in 36.0s and written in 34.0s\n", "31.3 hours of data converted in 45.0s and written in 34.0s\n", "29.8 hours of data converted in 34.0s and written in 34.0s\n", "31.0 hours of data converted in 29.0s and written in 34.0s\n", "30.8 hours of data converted in 33.0s and written in 34.0s\n", "32.4 hours of data converted in 41.0s and written in 34.0s\n", "30.4 hours of data converted in 33.0s and written in 34.0s\n", "31.6 hours of data converted in 34.0s and written in 34.0s\n", "30.9 hours of data converted in 37.0s and written in 33.0s\n", "30.1 hours of data converted in 31.0s and written in 33.0s\n", "31.4 hours of data converted in 40.0s and written in 34.0s\n", "30.9 hours of data converted in 29.0s and written in 33.0s\n", "31.3 hours of data converted in 36.0s and written in 34.0s\n", "30.6 hours of data converted in 30.0s and written in 33.0s\n", "30.9 hours of data converted in 33.0s and written in 34.0s\n", "31.1 hours of data converted in 41.0s and written in 33.0s\n", "31.2 hours of data converted in 30.0s and written in 33.0s\n", "30.4 hours of data converted in 29.0s and written in 34.0s\n", "30.8 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 37.0s and written in 34.0s\n", "29.8 hours of data converted in 33.0s and written in 33.0s\n", "32.0 hours of data converted in 34.0s and written in 36.0s\n", "31.1 hours of data converted in 35.0s and written in 34.0s\n", "30.7 hours of data converted in 43.0s and written in 33.0s\n", "30.6 hours of data converted in 35.0s and written in 36.0s\n", "30.7 hours of data converted in 30.0s and written in 33.0s\n", "30.3 hours of data converted in 35.0s and written in 36.0s\n", "31.5 hours of data converted in 37.0s and written in 34.0s\n", "30.5 hours of data converted in 34.0s and written in 33.0s\n", "30.4 hours of data converted in 29.0s and written in 33.0s\n", "31.4 hours of data converted in 34.0s and written in 34.0s\n", "30.6 hours of data converted in 38.0s and written in 34.0s\n", "31.8 hours of data converted in 43.0s and written in 34.0s\n", "31.7 hours of data converted in 32.0s and written in 34.0s\n", "31.3 hours of data converted in 31.0s and written in 34.0s\n", "30.4 hours of data converted in 34.0s and written in 36.0s\n", "31.3 hours of data converted in 29.0s and written in 33.0s\n", "31.4 hours of data converted in 34.0s and written in 36.0s\n", "31.0 hours of data converted in 35.0s and written in 34.0s\n", "30.0 hours of data converted in 31.0s and written in 33.0s\n", "31.8 hours of data converted in 34.0s and written in 34.0s\n", "32.5 hours of data converted in 37.0s and written in 34.0s\n", "30.9 hours of data converted in 31.0s and written in 33.0s\n", "31.7 hours of data converted in 30.0s and written in 34.0s\n", "30.8 hours of data converted in 32.0s and written in 34.0s\n", "31.5 hours of data converted in 34.0s and written in 34.0s\n", "30.6 hours of data converted in 44.0s and written in 33.0s\n", "31.8 hours of data converted in 39.0s and written in 34.0s\n", "29.4 hours of data converted in 35.0s and written in 33.0s\n", "31.7 hours of data converted in 34.0s and written in 37.0s\n", "31.2 hours of data converted in 42.0s and written in 33.0s\n", "31.4 hours of data converted in 32.0s and written in 34.0s\n", "31.4 hours of data converted in 39.0s and written in 35.0s\n", "31.6 hours of data converted in 32.0s and written in 34.0s\n", "30.5 hours of data converted in 38.0s and written in 33.0s\n", "30.6 hours of data converted in 34.0s and written in 35.0s\n", "31.5 hours of data converted in 41.0s and written in 34.0s\n", "31.4 hours of data converted in 35.0s and written in 34.0s\n", "30.7 hours of data converted in 32.0s and written in 33.0s\n", "31.4 hours of data converted in 32.0s and written in 34.0s\n", "31.2 hours of data converted in 43.0s and written in 33.0s\n", "31.2 hours of data converted in 33.0s and written in 34.0s\n", "30.8 hours of data converted in 36.0s and written in 33.0s\n", "30.7 hours of data converted in 34.0s and written in 33.0s\n", "31.4 hours of data converted in 36.0s and written in 34.0s\n", "31.0 hours of data converted in 37.0s and written in 34.0s\n", "30.3 hours of data converted in 31.0s and written in 33.0s\n", "30.6 hours of data converted in 44.0s and written in 33.0s\n", "30.4 hours of data converted in 32.0s and written in 33.0s\n", "30.4 hours of data converted in 39.0s and written in 33.0s\n", "30.4 hours of data converted in 85.0s and written in 33.0s\n", "31.6 hours of data converted in 45.0s and written in 34.0s\n", "31.6 hours of data converted in 41.0s and written in 38.0s\n", "31.8 hours of data converted in 33.0s and written in 34.0s\n", "30.6 hours of data converted in 37.0s and written in 33.0s\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "31.1 hours of data converted in 35.0s and written in 33.0s\n", "31.3 hours of data converted in 33.0s and written in 34.0s\n", "30.3 hours of data converted in 40.0s and written in 33.0s\n", "31.0 hours of data converted in 37.0s and written in 36.0s\n", "30.4 hours of data converted in 33.0s and written in 34.0s\n", "29.5 hours of data converted in 36.0s and written in 35.0s\n", "26.6 hours of data converted in 30.0s and written in 30.0s\n", "27.1 hours of data converted in 30.0s and written in 30.0s\n", "27.3 hours of data converted in 32.0s and written in 31.0s\n", "27.4 hours of data converted in 28.0s and written in 31.0s\n", "27.7 hours of data converted in 38.0s and written in 31.0s\n", "27.0 hours of data converted in 39.0s and written in 31.0s\n", "28.9 hours of data converted in 33.0s and written in 32.0s\n", "25.7 hours of data converted in 30.0s and written in 31.0s\n", "25.7 hours of data converted in 32.0s and written in 30.0s\n", "25.4 hours of data converted in 33.0s and written in 30.0s\n", "26.4 hours of data converted in 33.0s and written in 30.0s\n", "27.2 hours of data converted in 33.0s and written in 31.0s\n", "26.7 hours of data converted in 30.0s and written in 30.0s\n", "28.7 hours of data converted in 33.0s and written in 31.0s\n", "28.2 hours of data converted in 34.0s and written in 33.0s\n", "27.0 hours of data converted in 32.0s and written in 30.0s\n", "26.7 hours of data converted in 27.0s and written in 31.0s\n", "27.0 hours of data converted in 34.0s and written in 32.0s\n", "28.0 hours of data converted in 36.0s and written in 31.0s\n", "26.6 hours of data converted in 33.0s and written in 30.0s\n", "27.6 hours of data converted in 33.0s and written in 31.0s\n", "25.8 hours of data converted in 35.0s and written in 30.0s\n", "26.9 hours of data converted in 29.0s and written in 30.0s\n", "26.8 hours of data converted in 35.0s and written in 32.0s\n", "26.9 hours of data converted in 33.0s and written in 30.0s\n", "27.3 hours of data converted in 33.0s and written in 30.0s\n", "26.5 hours of data converted in 29.0s and written in 30.0s\n", "27.5 hours of data converted in 28.0s and written in 30.0s\n", "26.7 hours of data converted in 32.0s and written in 30.0s\n", "27.4 hours of data converted in 33.0s and written in 30.0s\n", "26.6 hours of data converted in 34.0s and written in 31.0s\n", "27.3 hours of data converted in 33.0s and written in 34.0s\n", "27.3 hours of data converted in 36.0s and written in 30.0s\n", "26.4 hours of data converted in 32.0s and written in 32.0s\n", "26.4 hours of data converted in 33.0s and written in 30.0s\n", "26.9 hours of data converted in 40.0s and written in 30.0s\n", "27.1 hours of data converted in 35.0s and written in 30.0s\n", "27.0 hours of data converted in 32.0s and written in 32.0s\n", "27.1 hours of data converted in 34.0s and written in 30.0s\n", "25.5 hours of data converted in 35.0s and written in 32.0s\n", "27.8 hours of data converted in 31.0s and written in 31.0s\n", "27.4 hours of data converted in 31.0s and written in 31.0s\n", "25.8 hours of data converted in 34.0s and written in 29.0s\n", "27.3 hours of data converted in 35.0s and written in 30.0s\n", "27.0 hours of data converted in 34.0s and written in 31.0s\n", "26.7 hours of data converted in 39.0s and written in 30.0s\n", "27.2 hours of data converted in 28.0s and written in 30.0s\n", "27.0 hours of data converted in 35.0s and written in 33.0s\n", "25.9 hours of data converted in 39.0s and written in 30.0s\n", "27.2 hours of data converted in 34.0s and written in 30.0s\n", "27.0 hours of data converted in 36.0s and written in 31.0s\n", "27.4 hours of data converted in 28.0s and written in 31.0s\n", "26.3 hours of data converted in 29.0s and written in 30.0s\n", "27.1 hours of data converted in 29.0s and written in 31.0s\n", "27.7 hours of data converted in 36.0s and written in 32.0s\n", "26.4 hours of data converted in 29.0s and written in 30.0s\n", "28.0 hours of data converted in 31.0s and written in 35.0s\n", "27.8 hours of data converted in 36.0s and written in 31.0s\n", "25.7 hours of data converted in 32.0s and written in 29.0s\n", "26.2 hours of data converted in 41.0s and written in 30.0s\n", "26.8 hours of data converted in 40.0s and written in 31.0s\n", "27.9 hours of data converted in 30.0s and written in 31.0s\n", "26.2 hours of data converted in 35.0s and written in 31.0s\n", "26.1 hours of data converted in 32.0s and written in 30.0s\n", "26.3 hours of data converted in 31.0s and written in 30.0s\n", "26.3 hours of data converted in 33.0s and written in 30.0s\n", "26.4 hours of data converted in 36.0s and written in 30.0s\n", "28.2 hours of data converted in 34.0s and written in 31.0s\n", "27.5 hours of data converted in 34.0s and written in 33.0s\n", "27.4 hours of data converted in 30.0s and written in 31.0s\n", "26.6 hours of data converted in 34.0s and written in 31.0s\n", "26.5 hours of data converted in 33.0s and written in 31.0s\n", "26.2 hours of data converted in 31.0s and written in 30.0s\n", "26.1 hours of data converted in 33.0s and written in 30.0s\n", "26.6 hours of data converted in 35.0s and written in 31.0s\n", "25.0 hours of data converted in 41.0s and written in 29.0s\n", "26.8 hours of data converted in 28.0s and written in 31.0s\n", "27.1 hours of data converted in 30.0s and written in 31.0s\n", "27.1 hours of data converted in 37.0s and written in 30.0s\n", "26.9 hours of data converted in 27.0s and written in 30.0s\n", "27.0 hours of data converted in 39.0s and written in 30.0s\n", "27.3 hours of data converted in 32.0s and written in 30.0s\n", "27.9 hours of data converted in 37.0s and written in 31.0s\n", "26.9 hours of data converted in 39.0s and written in 34.0s\n", "19.7 hours of data converted in 28.0s and written in 29.0s\n", "17.4 hours of data converted in 24.0s and written in 29.0s\n", "17.6 hours of data converted in 19.0s and written in 29.0s\n", "18.1 hours of data converted in 22.0s and written in 29.0s\n", "17.2 hours of data converted in 24.0s and written in 28.0s\n", "18.5 hours of data converted in 20.0s and written in 29.0s\n", "17.3 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 23.0s and written in 28.0s\n", "17.5 hours of data converted in 25.0s and written in 29.0s\n", "16.9 hours of data converted in 18.0s and written in 29.0s\n", "17.4 hours of data converted in 25.0s and written in 30.0s\n", "17.9 hours of data converted in 23.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "17.4 hours of data converted in 23.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "18.1 hours of data converted in 19.0s and written in 29.0s\n", "18.3 hours of data converted in 22.0s and written in 29.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "18.1 hours of data converted in 28.0s and written in 32.0s\n", "18.6 hours of data converted in 21.0s and written in 29.0s\n", "18.4 hours of data converted in 22.0s and written in 31.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "18.1 hours of data converted in 19.0s and written in 29.0s\n", "17.4 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 25.0s and written in 31.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "16.8 hours of data converted in 18.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "18.5 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "17.3 hours of data converted in 21.0s and written in 31.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 21.0s and written in 29.0s\n", "18.3 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 21.0s and written in 31.0s\n", "18.3 hours of data converted in 20.0s and written in 29.0s\n", "17.4 hours of data converted in 20.0s and written in 31.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "17.7 hours of data converted in 26.0s and written in 29.0s\n", "18.4 hours of data converted in 22.0s and written in 29.0s\n", "18.6 hours of data converted in 22.0s and written in 29.0s\n", "17.9 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 22.0s and written in 29.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "17.5 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 21.0s and written in 29.0s\n", "18.5 hours of data converted in 67.0s and written in 29.0s\n", "18.0 hours of data converted in 21.0s and written in 29.0s\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "17.6 hours of data converted in 19.0s and written in 28.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "17.4 hours of data converted in 25.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "17.5 hours of data converted in 19.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "18.3 hours of data converted in 22.0s and written in 29.0s\n", "17.8 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "17.5 hours of data converted in 22.0s and written in 29.0s\n", "17.9 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 23.0s and written in 31.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "17.1 hours of data converted in 24.0s and written in 32.0s\n", "17.7 hours of data converted in 25.0s and written in 29.0s\n", "17.6 hours of data converted in 19.0s and written in 29.0s\n", "17.6 hours of data converted in 25.0s and written in 29.0s\n", "17.6 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 21.0s and written in 29.0s\n", "18.6 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 19.0s and written in 29.0s\n", "18.0 hours of data converted in 24.0s and written in 30.0s\n", "17.7 hours of data converted in 19.0s and written in 29.0s\n", "17.5 hours of data converted in 24.0s and written in 29.0s\n", "16.6 hours of data converted in 19.0s and written in 28.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "18.4 hours of data converted in 23.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 28.0s and written in 29.0s\n", "17.7 hours of data converted in 21.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "17.7 hours of data converted in 25.0s and written in 29.0s\n", "17.5 hours of data converted in 19.0s and written in 29.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "16.8 hours of data converted in 19.0s and written in 29.0s\n", "18.1 hours of data converted in 20.0s and written in 29.0s\n", "18.5 hours of data converted in 22.0s and written in 29.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "18.1 hours of data converted in 19.0s and written in 29.0s\n", "18.8 hours of data converted in 20.0s and written in 29.0s\n", "17.5 hours of data converted in 22.0s and written in 31.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "17.5 hours of data converted in 21.0s and written in 29.0s\n", "17.0 hours of data converted in 20.0s and written in 28.0s\n", "18.8 hours of data converted in 23.0s and written in 30.0s\n", "18.0 hours of data converted in 24.0s and written in 29.0s\n", "18.4 hours of data converted in 19.0s and written in 29.0s\n", "16.9 hours of data converted in 20.0s and written in 29.0s\n", "17.1 hours of data converted in 19.0s and written in 29.0s\n", "16.7 hours of data converted in 20.0s and written in 29.0s\n", "17.2 hours of data converted in 19.0s and written in 28.0s\n", "18.2 hours of data converted in 21.0s and written in 29.0s\n", "18.6 hours of data converted in 24.0s and written in 30.0s\n", "17.0 hours of data converted in 20.0s and written in 29.0s\n", "17.0 hours of data converted in 25.0s and written in 29.0s\n", "18.0 hours of data converted in 25.0s and written in 29.0s\n", "17.3 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 23.0s and written in 29.0s\n", "17.5 hours of data converted in 23.0s and written in 30.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "17.7 hours of data converted in 19.0s and written in 29.0s\n", "17.0 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 30.0s\n", "17.9 hours of data converted in 22.0s and written in 29.0s\n", "17.6 hours of data converted in 24.0s and written in 29.0s\n", "17.5 hours of data converted in 24.0s and written in 29.0s\n", "17.5 hours of data converted in 28.0s and written in 29.0s\n", "17.4 hours of data converted in 22.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 23.0s and written in 29.0s\n", "17.6 hours of data converted in 21.0s and written in 29.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "18.6 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 26.0s and written in 29.0s\n", "18.1 hours of data converted in 25.0s and written in 29.0s\n", "17.7 hours of data converted in 25.0s and written in 29.0s\n", "17.3 hours of data converted in 19.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 19.0s and written in 29.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "19.0 hours of data converted in 20.0s and written in 29.0s\n", "18.3 hours of data converted in 22.0s and written in 29.0s\n", "18.1 hours of data converted in 21.0s and written in 29.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 21.0s and written in 29.0s\n", "18.1 hours of data converted in 19.0s and written in 29.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 25.0s and written in 29.0s\n", "17.2 hours of data converted in 24.0s and written in 29.0s\n", "17.7 hours of data converted in 19.0s and written in 29.0s\n", "18.8 hours of data converted in 21.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 26.0s and written in 29.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "16.8 hours of data converted in 20.0s and written in 29.0s\n", "17.3 hours of data converted in 23.0s and written in 29.0s\n", "17.2 hours of data converted in 26.0s and written in 29.0s\n", "17.3 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 20.0s and written in 29.0s\n", "17.4 hours of data converted in 21.0s and written in 31.0s\n", "18.3 hours of data converted in 25.0s and written in 29.0s\n", "18.0 hours of data converted in 24.0s and written in 29.0s\n", "18.4 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 24.0s and written in 31.0s\n", "17.4 hours of data converted in 19.0s and written in 29.0s\n", "17.2 hours of data converted in 25.0s and written in 30.0s\n", "18.5 hours of data converted in 24.0s and written in 29.0s\n", "18.4 hours of data converted in 19.0s and written in 29.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 25.0s and written in 29.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "18.0 hours of data converted in 20.0s and written in 30.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "18.1 hours of data converted in 20.0s and written in 29.0s\n", "17.3 hours of data converted in 23.0s and written in 30.0s\n", "18.1 hours of data converted in 20.0s and written in 29.0s\n", "18.8 hours of data converted in 25.0s and written in 29.0s\n", "17.4 hours of data converted in 24.0s and written in 29.0s\n", "18.3 hours of data converted in 19.0s and written in 29.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 28.0s and written in 32.0s\n", "17.1 hours of data converted in 20.0s and written in 29.0s\n", "18.4 hours of data converted in 19.0s and written in 29.0s\n", "17.5 hours of data converted in 20.0s and written in 29.0s\n", "18.3 hours of data converted in 53.0s and written in 29.0s\n", "17.4 hours of data converted in 20.0s and written in 29.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 82.0s and written in 29.0s\n", "17.9 hours of data converted in 20.0s and written in 29.0s\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "17.9 hours of data converted in 19.0s and written in 29.0s\n", "18.1 hours of data converted in 20.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "18.4 hours of data converted in 20.0s and written in 29.0s\n", "18.2 hours of data converted in 26.0s and written in 30.0s\n", "18.0 hours of data converted in 21.0s and written in 29.0s\n", "17.3 hours of data converted in 19.0s and written in 28.0s\n", "17.6 hours of data converted in 20.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 32.0s\n", "17.5 hours of data converted in 20.0s and written in 30.0s\n", "17.6 hours of data converted in 24.0s and written in 29.0s\n", "18.1 hours of data converted in 21.0s and written in 32.0s\n", "17.9 hours of data converted in 20.0s and written in 29.0s\n", "18.0 hours of data converted in 25.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 30.0s\n", "17.7 hours of data converted in 24.0s and written in 29.0s\n", "17.7 hours of data converted in 19.0s and written in 29.0s\n", "17.3 hours of data converted in 24.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "18.3 hours of data converted in 23.0s and written in 31.0s\n", "17.4 hours of data converted in 22.0s and written in 28.0s\n", "17.8 hours of data converted in 21.0s and written in 29.0s\n", "17.7 hours of data converted in 19.0s and written in 29.0s\n", "18.1 hours of data converted in 19.0s and written in 31.0s\n", "17.5 hours of data converted in 21.0s and written in 29.0s\n", "17.6 hours of data converted in 19.0s and written in 28.0s\n", "16.7 hours of data converted in 19.0s and written in 28.0s\n", "17.9 hours of data converted in 21.0s and written in 29.0s\n", "18.3 hours of data converted in 22.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 20.0s and written in 29.0s\n", "17.6 hours of data converted in 19.0s and written in 28.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "17.4 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "17.0 hours of data converted in 25.0s and written in 28.0s\n", "17.3 hours of data converted in 20.0s and written in 31.0s\n", "17.8 hours of data converted in 27.0s and written in 31.0s\n", "17.3 hours of data converted in 25.0s and written in 29.0s\n", "17.5 hours of data converted in 19.0s and written in 29.0s\n", "18.3 hours of data converted in 24.0s and written in 29.0s\n", "18.1 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 21.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "18.3 hours of data converted in 19.0s and written in 29.0s\n", "17.9 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 19.0s and written in 29.0s\n", "18.8 hours of data converted in 20.0s and written in 29.0s\n", "17.6 hours of data converted in 23.0s and written in 29.0s\n", "19.0 hours of data converted in 21.0s and written in 29.0s\n", "17.7 hours of data converted in 25.0s and written in 31.0s\n", "17.8 hours of data converted in 20.0s and written in 31.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "18.7 hours of data converted in 21.0s and written in 29.0s\n", "17.7 hours of data converted in 23.0s and written in 31.0s\n", "17.9 hours of data converted in 25.0s and written in 29.0s\n", "18.2 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "18.7 hours of data converted in 19.0s and written in 33.0s\n", "17.8 hours of data converted in 21.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "17.4 hours of data converted in 19.0s and written in 28.0s\n", "19.0 hours of data converted in 20.0s and written in 29.0s\n", "18.3 hours of data converted in 19.0s and written in 29.0s\n", "17.7 hours of data converted in 20.0s and written in 29.0s\n", "18.3 hours of data converted in 19.0s and written in 31.0s\n", "18.0 hours of data converted in 24.0s and written in 29.0s\n", "18.0 hours of data converted in 26.0s and written in 29.0s\n", "18.7 hours of data converted in 20.0s and written in 29.0s\n", "17.5 hours of data converted in 19.0s and written in 29.0s\n", "18.2 hours of data converted in 20.0s and written in 29.0s\n", "17.9 hours of data converted in 19.0s and written in 29.0s\n", "18.6 hours of data converted in 22.0s and written in 29.0s\n", "18.0 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 28.0s and written in 29.0s\n", "18.3 hours of data converted in 21.0s and written in 29.0s\n", "17.4 hours of data converted in 19.0s and written in 28.0s\n", "17.8 hours of data converted in 20.0s and written in 29.0s\n", "18.1 hours of data converted in 21.0s and written in 29.0s\n", "18.1 hours of data converted in 22.0s and written in 31.0s\n", "17.6 hours of data converted in 28.0s and written in 31.0s\n", "18.3 hours of data converted in 22.0s and written in 29.0s\n", "17.8 hours of data converted in 19.0s and written in 29.0s\n", "17.8 hours of data converted in 25.0s and written in 32.0s\n", "17.3 hours of data converted in 20.0s and written in 28.0s\n", "18.1 hours of data converted in 20.0s and written in 29.0s\n", "9.4 hours of data converted in 12.0s and written in 16.0s\n" ] } ], "source": [ "CHUNKSIZE = 500\n", "MIN_DURATION_S = 5\n", "MAX_DURATION_S = 8*60\n", "\n", "# set up files\n", "for dset_type in (\"train\", \"valid\"):\n", " with open(os.path.join(OUT_TSV_DIR, f\"{dset_type}.tsv\"), \"w\") as f:\n", " f.write(OUT_AUDIO_DIR + \"\\n\")\n", " for codebook in range(N_CODEBOOKS):\n", " with open(os.path.join(OUT_LABEL_DIR, f\"{dset_type}.codec_{codebook}\"), \"w\") as f:\n", " f.write(\"\")\n", "\n", "n_items = 0\n", "tot_duration_h = {\"valid\": 0, \"train\": 0}\n", "for n_iter, metas_chunk in enumerate(funcy.chunks(CHUNKSIZE, metas)):\n", " # convert to wav\n", " t0 = time.time()\n", " out_rel_filepaths = [f\"{n_iter*CHUNKSIZE+n}.wav\" for n in range(len(metas_chunk))]\n", " confirmed_list = convert_audio_files(\n", " [m[\"filepath\"] for m in metas_chunk],\n", " [os.path.join(OUT_AUDIO_DIR, out_rel_fp) for out_rel_fp in out_rel_filepaths],\n", " n_cores=32,\n", " force_threads=False,\n", " sample_rate=SAMPLE_RATE,\n", " byte_width=2,\n", " n_channels=1,\n", " debug=False,\n", " )\n", " td_convert = int(round(time.time() - t0))\n", " # collect codec tokens and keep only valid ones then write tsv entries and codec\n", " t0 = time.time()\n", " codec_archive_fp = f\"s3://suno-data/datasets/bundles/v2/music_sample/dac_25_8/part_{n_iter}.npz\"\n", " codec_archive = read_from_s3(codec_archive_fp, read_f=np.load)\n", " # keep only valid ones in terms of duration and has both and then write tsv entries and codec\n", " tsv_files = {\n", " \"valid\": open(os.path.join(OUT_TSV_DIR, \"valid.tsv\"), \"a\"),\n", " \"train\": open(os.path.join(OUT_TSV_DIR, \"train.tsv\"), \"a\")\n", " }\n", " codec_files = {\n", " codebook: {\n", " \"valid\": open(os.path.join(OUT_LABEL_DIR, f\"valid.codec_{codebook}\"), \"a\"),\n", " \"train\": open(os.path.join(OUT_LABEL_DIR, f\"train.codec_{codebook}\"), \"a\")\n", " } for codebook in range(N_CODEBOOKS)\n", " }\n", " chunk_duration_h = 0\n", " n_segment = 0\n", " for m, wav_fp, conv_success in zip(metas_chunk, out_rel_filepaths, confirmed_list):\n", " if m[\"id\"] not in codec_archive or not conv_success:\n", " continue\n", " duration_s = Audio.get_duration_s(os.path.join(OUT_AUDIO_DIR, wav_fp))\n", " codec_arr = codec_archive[m[\"id\"]]\n", " codec_duration_s = codec_arr.shape[0] / EMBEDDING_RATE\n", " if abs(duration_s - codec_duration_s) > 0.1:\n", " continue\n", " if duration_s < MIN_DURATION_S or duration_s > MAX_DURATION_S:\n", " continue\n", " dset_type = \"valid\" if n_segment % 20 == 0 else \"train\"\n", " tsv_files[dset_type].write(wav_fp + \"\\t\" + str(int(round(duration_s*SAMPLE_RATE))) + \"\\n\")\n", " for codebook in range(N_CODEBOOKS):\n", " codec_files[codebook][dset_type].write(\" \".join(map(str, codec_arr[:,codebook])) + \"\\n\")\n", " tot_duration_h[dset_type] += duration_s / 60 / 60\n", " chunk_duration_h += duration_s / 60 / 60\n", " n_segment += 1\n", " for f in tsv_files.values():\n", " f.close()\n", " for codebook_files in codec_files.values():\n", " for f in codebook_files.values():\n", " f.close()\n", " td_write = int(round(time.time() - t0))\n", " print(\n", " f\"{chunk_duration_h:,.1f} hours of data converted in {td_convert:,.1f}s\"\n", " f\" and written in {td_write:,.1f}s\"\n", " )\n", " time.sleep(5) # make sure things close\n", "\n", "# total should take ~15h for prep\n", "# 30.4 hours of data converted in 62.0s and written in 33.0s" ] }, { "cell_type": "code", "execution_count": 6, "id": "3a57184b", "metadata": {}, "outputs": [], "source": [ "# only needed once\n", "for codebook in range(N_CODEBOOKS):\n", " with open(os.path.join(OUT_LABEL_DIR, f\"dict.codec_{codebook}.txt\"), \"w\") as f:\n", " for n in range(N_CODEBOOKS):\n", " f.write(f\"{n} 1\\n\")" ] }, { "cell_type": "code", "execution_count": 4, "id": "3611adfa", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "4.5T\t/app/suno/data/mert_25hz\r\n" ] } ], "source": [ "# !du -hs /app/suno/data/mert_25hz\n", "# 4.5T" ] }, { "cell_type": "code", "execution_count": 84, "id": "23ed99a5", "metadata": {}, "outputs": [], "source": [ "# # (optional) overwrite train with valid\n", "\n", "# !cp /app/suno/data/mert_25hz/audio_tsv/valid.tsv /app/suno/data/mert_25hz/audio_tsv/train.tsv\n", "\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_0 /app/suno/data/mert_25hz/label/train.codec_0\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_1 /app/suno/data/mert_25hz/label/train.codec_1\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_2 /app/suno/data/mert_25hz/label/train.codec_2\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_3 /app/suno/data/mert_25hz/label/train.codec_3\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_4 /app/suno/data/mert_25hz/label/train.codec_4\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_5 /app/suno/data/mert_25hz/label/train.codec_5\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_6 /app/suno/data/mert_25hz/label/train.codec_6\n", "# !cp /app/suno/data/mert_25hz/label/valid.codec_7 /app/suno/data/mert_25hz/label/train.codec_7" ] }, { "cell_type": "code", "execution_count": null, "id": "8bfb957a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "a09ff456", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "fbb4916d", "metadata": {}, "source": [ "## (TODO) make a second version with 12s segments" ] }, { "cell_type": "code", "execution_count": 1, "id": "4f9e0143", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\"" ] }, { "cell_type": "code", "execution_count": 2, "id": "47a20175", "metadata": {}, "outputs": [], "source": [ "import random\n", "import json\n", "import numpy as np\n", "import tqdm\n", "import torch\n", "import funcy\n", "import time\n", "import gc\n", "from scipy.io import wavfile\n", "import tempfile\n", "import collections\n", "from collections import defaultdict\n", "from joblib import Parallel, delayed\n", "\n", "from suno_utils.utils.s3 import _apply_mp\n", "from suno_utils.audio import Audio\n", "from suno_utils.tasks.data_loader import load_audio_mp\n", "from suno_utils.utils.text import write_jsonl, read_jsonl, write_json, read_json\n", "from suno_utils.utils.s3 import read_from_s3, check_s3_file_exists, open_from_s3\n", "from suno_utils.audio.conversion import convert_audio_files\n", "\n", "SAMPLE_RATE = 24_000\n", "EMBEDDING_RATE = 25\n", "N_CODEBOOKS = 8\n", "\n", "IN_DATA_DIR = \"/app/suno/data/mert_25hz\"\n", "IN_AUDIO_DIR = os.path.join(IN_DATA_DIR, \"audio\")\n", "IN_TSV_DIR = os.path.join(IN_DATA_DIR, \"audio_tsv\")\n", "IN_LABEL_DIR = os.path.join(IN_DATA_DIR, \"label\")\n", "\n", "OUT_DATA_DIR = \"/app/suno/data/mert_25hz_short\"\n", "OUT_AUDIO_DIR = os.path.join(OUT_DATA_DIR, \"audio\")\n", "OUT_TSV_DIR = os.path.join(OUT_DATA_DIR, \"audio_tsv\")\n", "OUT_LABEL_DIR = os.path.join(OUT_DATA_DIR, \"label\")\n", "\n", "os.makedirs(OUT_DATA_DIR, exist_ok=True)\n", "os.makedirs(OUT_AUDIO_DIR, exist_ok=True)\n", "os.makedirs(OUT_TSV_DIR, exist_ok=True)\n", "os.makedirs(OUT_LABEL_DIR, exist_ok=True)" ] }, { "cell_type": "code", "execution_count": 204, "id": "d8930637", "metadata": {}, "outputs": [], "source": [ "# !rm -rf /app/suno/data/mert_25hz_short/audio/*\n", "# !rm -rf /app/suno/data/mert_25hz_short/audio_tsv/*\n", "# !rm -rf /app/suno/data/mert_25hz_short/label/*" ] }, { "cell_type": "markdown", "id": "31351aab", "metadata": {}, "source": [ "### make val" ] }, { "cell_type": "code", "execution_count": 3, "id": "b97d6573", "metadata": {}, "outputs": [], "source": [ "MIN_SIZE = SAMPLE_RATE * 5\n", "MAX_SIZE = SAMPLE_RATE * 12\n", "\n", "n_audio = 0" ] }, { "cell_type": "code", "execution_count": 202, "id": "4ab5ccb0", "metadata": {}, "outputs": [], "source": [ "# load current data\n", "tsv_data = []\n", "with open(os.path.join(IN_TSV_DIR, \"valid.tsv\"), \"r\") as f:\n", " for line in f.read().strip().split(\"\\n\"):\n", " if len(line.strip()) == 0:\n", " continue\n", " tsv_data.append(line.strip().split(\"\\t\"))\n", "assert(IN_AUDIO_DIR == tsv_data[0][0])\n", "tsv_data = tsv_data[1:]\n", "codec_data = defaultdict(list)\n", "for codebook in range(N_CODEBOOKS):\n", " with open(os.path.join(IN_LABEL_DIR, f\"valid.codec_{codebook}\"), \"r\") as f:\n", " for line in f.read().strip().split(\"\\n\"):\n", " if len(line.strip()) == 0:\n", " continue\n", " codec_data[codebook].append(line.strip().split())" ] }, { "cell_type": "code", "execution_count": 236, "id": "0624d598", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|█████████████████████████████████████████████████████████████████████████████| 23066/23066 [01:08<00:00, 335.09it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "374219 lines\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\n" ] } ], "source": [ "new_tsv_data = []\n", "new_codec_data = defaultdict(list)\n", "for n_row, (audio_fn, audio_size) in tqdm.tqdm(enumerate(tsv_data), total=len(tsv_data)):\n", " b_skip = False\n", " for codebook in range(N_CODEBOOKS):\n", " if np.abs(int(audio_size) / SAMPLE_RATE - len(codec_data[codebook][n_row]) / EMBEDDING_RATE) > 0.1:\n", " b_skip = True\n", " if b_skip:\n", " continue\n", " _, audio_arr = wavfile.read(os.path.join(IN_AUDIO_DIR, audio_fn))\n", " for n in range(int(np.ceil(int(audio_size)/MAX_SIZE))):\n", " start_idx = n * MAX_SIZE\n", " end_idx = min(int(audio_size), (n + 1) * MAX_SIZE)\n", " if end_idx - start_idx < MIN_SIZE:\n", " continue\n", " out_audio_fn = f\"{n_audio}.wav\"\n", " # write tsv\n", " new_tsv_data.append((audio_fn, out_audio_fn, start_idx, end_idx))\n", " # write codec\n", " start_idx_embed = int(round(start_idx / SAMPLE_RATE * EMBEDDING_RATE))\n", " end_idx_embed = int(round(end_idx / SAMPLE_RATE * EMBEDDING_RATE))\n", " for codebook in range(N_CODEBOOKS):\n", " new_codec_data[codebook].append(\" \".join(codec_data[codebook][n_row][start_idx_embed:end_idx_embed]))\n", " n_audio += 1\n", " \n", "# if n_audio >= 10000:\n", "# break\n", " \n", "assert(len(new_codec_data[0]) == len(new_tsv_data))\n", "assert(len(set(len(new_codec_data[n]) for n in range(N_CODEBOOKS))) == 1)\n", "assert(len(new_tsv_data) == n_audio)\n", "print(len(new_tsv_data), \"lines\")\n", "# 374219 lines" ] }, { "cell_type": "code", "execution_count": 238, "id": "ca06e378", "metadata": {}, "outputs": [], "source": [ "# write metafiles to disk\n", "with open(os.path.join(OUT_TSV_DIR, f\"valid.tsv\"), \"w\") as f:\n", " f.write(OUT_AUDIO_DIR + \"\\n\")\n", " f.write(\"\\n\".join([fn + \"\\t\" + str(int(idx_end - idx_start)) for _, fn, idx_start, idx_end in new_tsv_data]))\n", "for codebook in range(N_CODEBOOKS):\n", " with open(os.path.join(OUT_LABEL_DIR, f\"valid.codec_{codebook}\"), \"w\") as f:\n", " f.write(\"\\n\".join(new_codec_data[codebook]))" ] }, { "cell_type": "code", "execution_count": 239, "id": "4b9f2d43", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "23066 work_items\n" ] } ], "source": [ "# write new audios to disk\n", "def _write_new_wav_files(work_item):\n", " from_fn, to_work_items = work_item\n", " _, audio_arr = wavfile.read(os.path.join(IN_AUDIO_DIR, from_fn))\n", " for to_fn, (start_idx, end_idx) in to_work_items:\n", " wavfile.write(os.path.join(OUT_AUDIO_DIR, to_fn), SAMPLE_RATE, audio_arr[start_idx:end_idx])\n", "\n", "work_items = defaultdict(list)\n", "for from_fn, to_fn, start_idx, end_idx in new_tsv_data:\n", " work_items[from_fn].append((to_fn, (start_idx, end_idx)))\n", "work_items = [(k, v) for k, v in work_items.items()]\n", "print(len(work_items), \"work_items\")" ] }, { "cell_type": "code", "execution_count": null, "id": "2dc2f78f", "metadata": {}, "outputs": [], "source": [ "# # 40mins for val with single core (~20x for train)\n", "# for work_item in tqdm.tqdm(work_items[:500]):\n", "# _write_new_wav_files(work_item)" ] }, { "cell_type": "code", "execution_count": 269, "id": "c11b7333", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|████████████████████████████████████████████████████████████████████████████████████| 24/24 [39:11<00:00, 98.00s/it]\n" ] } ], "source": [ "_ = _apply_mp(_write_new_wav_files, work_items, chunksize=1000, n_cores=10, joblib_backend=\"threads\")" ] }, { "cell_type": "code", "execution_count": 270, "id": "7a9e97d6", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "423" ] }, "execution_count": 270, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# delete cached stuff\n", "del work_items, tsv_data, codec_data, new_tsv_data, new_codec_data\n", "gc.collect();" ] }, { "cell_type": "code", "execution_count": 25, "id": "c57529be", "metadata": {}, "outputs": [], "source": [ "# # verify data\n", "\n", "# from collections import deque\n", "# from suno_utils.tasks.gpt_v2.chirp_v1 import codec_decode, preload_codec_models\n", "\n", "# _ = preload_codec_models(\"s3://suno-data/georg/trained_models/chirp_v1/codec.pt\")\n", "\n", "# def tail(filename, n=1):\n", "# 'Return the last n lines of a file'\n", "# with open(filename) as f:\n", "# return deque(f, n) \n", "\n", "# data = []\n", "# for n in range(8):\n", "# line = tail(f\"/app/suno/data/mert_25hz_short/label/valid.codec_{n}\")[-1].strip()\n", "# data.append(np.array([int(e) for e in line.split()]))\n", "# codec_decode(np.stack(data).T).play()\n", "\n", "# Audio.from_file(\"/app/suno/data/mert_25hz_short/audio/374218.wav\").play()" ] }, { "cell_type": "markdown", "id": "a5cddda6", "metadata": {}, "source": [ "### Make train" ] }, { "cell_type": "code", "execution_count": 6, "id": "61490c93", "metadata": {}, "outputs": [], "source": [ "# !tail -1 /app/suno/data/mert_25hz_short/audio_tsv/valid.tsv" ] }, { "cell_type": "code", "execution_count": 3, "id": "ebd3f56c", "metadata": {}, "outputs": [], "source": [ "MIN_SIZE = SAMPLE_RATE * 5\n", "MAX_SIZE = SAMPLE_RATE * 12\n", "\n", "n_audio = 374219" ] }, { "cell_type": "code", "execution_count": null, "id": "41f31eb6", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "134c35c3", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "6c159241", "metadata": {}, "outputs": [], "source": [ "# load current data\n", "tsv_data = []\n", "with open(os.path.join(IN_TSV_DIR, \"train.tsv\"), \"r\") as f:\n", " for line in f.read().strip().split(\"\\n\"):\n", " if len(line.strip()) == 0:\n", " continue\n", " tsv_data.append(line.strip().split(\"\\t\"))\n", "assert(IN_AUDIO_DIR == tsv_data[0][0])\n", "tsv_data = tsv_data[1:]\n", "codec_data = defaultdict(list)\n", "for codebook in range(N_CODEBOOKS):\n", " with open(os.path.join(IN_LABEL_DIR, f\"train.codec_{codebook}\"), \"r\") as f:\n", " for line in f.read().strip().split(\"\\n\"):\n", " if len(line.strip()) == 0:\n", " continue\n", " codec_data[codebook].append(line.strip().split())" ] }, { "cell_type": "code", "execution_count": null, "id": "a3205b95", "metadata": {}, "outputs": [], "source": [ "new_tsv_data = []\n", "new_codec_data = defaultdict(list)\n", "for n_row, (audio_fn, audio_size) in tqdm.tqdm(enumerate(tsv_data), total=len(tsv_data)):\n", " b_skip = False\n", " for codebook in range(N_CODEBOOKS):\n", " if np.abs(int(audio_size) / SAMPLE_RATE - len(codec_data[codebook][n_row]) / EMBEDDING_RATE) > 0.1:\n", " b_skip = True\n", " if b_skip:\n", " continue\n", " _, audio_arr = wavfile.read(os.path.join(IN_AUDIO_DIR, audio_fn))\n", " for n in range(int(np.ceil(int(audio_size)/MAX_SIZE))):\n", " start_idx = n * MAX_SIZE\n", " end_idx = min(int(audio_size), (n + 1) * MAX_SIZE)\n", " if end_idx - start_idx < MIN_SIZE:\n", " continue\n", " out_audio_fn = f\"{n_audio}.wav\"\n", " # write tsv\n", " new_tsv_data.append((audio_fn, out_audio_fn, start_idx, end_idx))\n", " # write codec\n", " start_idx_embed = int(round(start_idx / SAMPLE_RATE * EMBEDDING_RATE))\n", " end_idx_embed = int(round(end_idx / SAMPLE_RATE * EMBEDDING_RATE))\n", " for codebook in range(N_CODEBOOKS):\n", " new_codec_data[codebook].append(\" \".join(codec_data[codebook][n_row][start_idx_embed:end_idx_embed]))\n", " n_audio += 1\n", " \n", "# if n_audio >= 10000:\n", "# break\n", " \n", "assert(len(new_codec_data[0]) == len(new_tsv_data))\n", "assert(len(set(len(new_codec_data[n]) for n in range(N_CODEBOOKS))) == 1)\n", "assert(len(new_tsv_data) == n_audio)\n", "print(len(new_tsv_data), \"lines\")\n", "# 374219 lines" ] }, { "cell_type": "code", "execution_count": null, "id": "bef29f7d", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "252b379f", "metadata": {}, "outputs": [], "source": [ "# write metafiles to disk\n", "with open(os.path.join(OUT_TSV_DIR, f\"train.tsv\"), \"w\") as f:\n", " f.write(OUT_AUDIO_DIR + \"\\n\")\n", " f.write(\"\\n\".join([fn + \"\\t\" + str(int(idx_end - idx_start)) for _, fn, idx_start, idx_end in new_tsv_data]))\n", "for codebook in range(N_CODEBOOKS):\n", " with open(os.path.join(OUT_LABEL_DIR, f\"train.codec_{codebook}\"), \"w\") as f:\n", " f.write(\"\\n\".join(new_codec_data[codebook]))" ] }, { "cell_type": "code", "execution_count": null, "id": "d269853c", "metadata": {}, "outputs": [], "source": [ "# write new audios to disk\n", "def _write_new_wav_files(work_item):\n", " from_fn, to_work_items = work_item\n", " _, audio_arr = wavfile.read(os.path.join(IN_AUDIO_DIR, from_fn))\n", " for to_fn, (start_idx, end_idx) in to_work_items:\n", " wavfile.write(os.path.join(OUT_AUDIO_DIR, to_fn), SAMPLE_RATE, audio_arr[start_idx:end_idx])\n", "\n", "work_items = defaultdict(list)\n", "for from_fn, to_fn, start_idx, end_idx in new_tsv_data:\n", " work_items[from_fn].append((to_fn, (start_idx, end_idx)))\n", "work_items = [(k, v) for k, v in work_items.items()]\n", "print(len(work_items), \"work_items\")" ] }, { "cell_type": "code", "execution_count": null, "id": "86cde2be", "metadata": {}, "outputs": [], "source": [ "_ = _apply_mp(_write_new_wav_files, work_items, chunksize=1000, n_cores=10, joblib_backend=\"threads\")" ] }, { "cell_type": "code", "execution_count": null, "id": "f5293786", "metadata": {}, "outputs": [], "source": [ "# delete cached stuff\n", "del work_items, tsv_data, codec_data, new_tsv_data, new_codec_data\n", "gc.collect();" ] }, { "cell_type": "code", "execution_count": null, "id": "dbcf28de", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "7680198b", "metadata": {}, "source": [ "### TODO: copy the same weird index files into labels" ] }, { "cell_type": "code", "execution_count": null, "id": "8b94812a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "f254c8cf", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "c9fadcf9", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "afb375fe", "metadata": {}, "source": [ "## run training" ] }, { "cell_type": "code", "execution_count": null, "id": "742ed56e", "metadata": {}, "outputs": [], "source": [ "OMP_NUM_THREADS=6 python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz' \\\n", " distributed_training.distributed_rank=0 \\\n", " distributed_training.distributed_world_size=4 \\\n", " distributed_training.nprocs_per_node=4 \\\n", " optimization.update_freq='[2]' \\\n", " distributed_training.distributed_init_method=\"tcp://127.0.0.1:39683\" \\\n", " task.data='/app/suno/data/mert_25hz/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=true \\\n", " model.label_rate=25\n", "\n", "# sample dataset: /app/suno/data/mert_25hz_test\n", "# max_tokens defines batch_size so might need to be lower?\n", "# need to change below because 4 instead of 8 GPUS?\n", "# optimization.max_update=400000 \\\n", "# lr_scheduler.warmup_updates=32000 \\\n", "# could also just do\n", "# optimization.update_freq='[2]' \\" ] }, { "cell_type": "code", "execution_count": null, "id": "9f61c33c", "metadata": {}, "outputs": [], "source": [ "# ~7h for warmup steps -> 4x24h for full train" ] }, { "cell_type": "markdown", "id": "6f87dda5", "metadata": {}, "source": [ "#### 8-nodes" ] }, { "cell_type": "code", "execution_count": null, "id": "cd004d3f", "metadata": {}, "outputs": [], "source": [ "# rm -rf /app/suno/checkpoints/mert_25hz_8x/* | pkill -f 'spawn_main'" ] }, { "cell_type": "code", "execution_count": null, "id": "0052c4bf", "metadata": {}, "outputs": [], "source": [ "OMP_NUM_THREADS=6 NCCL_DEBUG=WARN NCCL_IB_SL=0 NCCL_IB_TC=41 NCCL_IB_QPS_PER_CONNECTION=4 UCX_TLS=ud,self,sm \\\n", "HCOLL_ENABLE_MCAST_ALL=0 NCCL_IB_GID_INDEX=3 UCX_NET_DEVICES=\"mlx5_0:1\" coll_hcoll_enable=0 \\\n", "NCCL_IB_HCA=\"=mlx5_5,mlx5_6,mlx5_7,mlx5_8,mlx5_1,mlx5_2,mlx5_3,mlx5_4,mlx5_14,mlx5_15,mlx5_16,mlx5_17,\\\n", "mlx5_9,mlx5_10,mlx5_11,mlx5_12\" \\\n", "python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz_8x' \\\n", " distributed_training.distributed_rank=$NODE_RANK \\\n", " distributed_training.distributed_world_size=56 \\\n", " distributed_training.nprocs_per_node=8 \\\n", " distributed_training.distributed_init_method=\"tcp://compute-hpc-node-192:2345\" \\\n", " task.data='/app/suno/data/mert_25hz_short/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz_short/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=false \\\n", " model.label_rate=25" ] }, { "cell_type": "code", "execution_count": 2, "id": "29115c51", "metadata": {}, "outputs": [], "source": [ "# TODO: loss scale??\n", "# TODO: is tcp an issue??\n", "# TODO: try with disable validation??" ] }, { "cell_type": "code", "execution_count": null, "id": "fd76d556", "metadata": {}, "outputs": [], "source": [ "# [2023-09-05 00:34:27,929][fairseq.tasks.fairseq_task][INFO] - can_reuse_epoch_itr = False\n", "# [2023-09-05 00:34:27,929][fairseq.tasks.fairseq_task][INFO] - reuse_dataloader = True\n", "# [2023-09-05 00:34:27,929][fairseq.tasks.fairseq_task][INFO] - rebuild_batches = False\n", "# [2023-09-05 00:34:27,929][fairseq.tasks.fairseq_task][INFO] - creating new batches for epoch 1" ] }, { "cell_type": "code", "execution_count": 19, "id": "610740fc", "metadata": {}, "outputs": [], "source": [ "# !ls /app/suno/data/mert_25hz_short/label" ] }, { "cell_type": "code", "execution_count": null, "id": "0c849b15", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "02aaabbb", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "650f7f17", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "10eeb15a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "fdb194cf", "metadata": {}, "source": [ "### Inference" ] }, { "cell_type": "code", "execution_count": 1, "id": "2bfafe41", "metadata": {}, "outputs": [], "source": [ "%matplotlib inline\n", "from matplotlib import pyplot as plt" ] }, { "cell_type": "code", "execution_count": 2, "id": "46bf6e89", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\"" ] }, { "cell_type": "code", "execution_count": 3, "id": "2ec447b0", "metadata": {}, "outputs": [], "source": [ "# # convert to HF checkpoint\n", "# import torch\n", "# import re\n", "# from_fp = \"/app/suno/checkpoints/mert_25hz/checkpoint_last.pt\"\n", "# to_fp = \"/home/georg/notebooks/dataprep/models/mert_test_25hz.pt\"\n", "# sd = torch.load(\"/app/suno/checkpoints/mert_25hz/checkpoint_last.pt\", map_location=\"cpu\")\n", "# new_sd = {}\n", "# for k, v in sd[\"model\"].items():\n", "# k = k.replace(\"mask_emb\", \"masked_spec_embed\")\n", "# k = k.replace(\".self_attn.\", \".attention.\")\n", "# k = k.replace(\".fc1.\", \".feed_forward.intermediate_dense.\")\n", "# k = k.replace(\".fc2.\", \".feed_forward.output_dense.\")\n", "# k = k.replace(\".self_attn_layer_norm.\", \".layer_norm.\")\n", "# k = k.replace(\"conv_layers.0.0\", \"conv_layers.0.conv\")\n", "# k = k.replace(\"conv_layers.1.0\", \"conv_layers.1.conv\")\n", "# k = k.replace(\"conv_layers.2.0\", \"conv_layers.2.conv\")\n", "# k = k.replace(\"conv_layers.3.0\", \"conv_layers.3.conv\")\n", "# k = k.replace(\"conv_layers.4.0\", \"conv_layers.4.conv\")\n", "# k = k.replace(\"conv_layers.5.0\", \"conv_layers.5.conv\")\n", "# k = k.replace(\"conv_layers.6.0\", \"conv_layers.6.conv\")\n", "# k = k.replace(\"conv_layers.7.0\", \"conv_layers.7.conv\")\n", "# k = k.replace(\"encoder.pos_conv.0.weight_v\", \"encoder.pos_conv_embed.conv.weight_v\")\n", "# k = k.replace(\"encoder.pos_conv.0.weight_g\", \"encoder.pos_conv_embed.conv.weight_g\")\n", "# k = k.replace(\"encoder.pos_conv.0.bias\", \"encoder.pos_conv_embed.conv.bias\")\n", "# k = re.sub(r\"^layer_norm.weight$\", \"feature_extractor.conv_layers.0.layer_norm.weight\", k)\n", "# k = re.sub(r\"^layer_norm.bias$\", \"feature_extractor.conv_layers.0.layer_norm.bias\", k)\n", "# k = k.replace(\"post_extract_proj.weight\", \"feature_projection.projection.weight\")\n", "# k = k.replace(\"post_extract_proj.bias\", \"feature_projection.projection.bias\")\n", "# k = k.replace(\"feature_extractor.conv_layers.0.2.weight\", \"feature_projection.layer_norm.weight\")\n", "# k = k.replace(\"feature_extractor.conv_layers.0.2.bias\", \"feature_projection.layer_norm.bias\")\n", "# new_sd[k] = v\n", "# for k in [\n", "# \"label_embs_concat\", \"final_proj.weight\", \"final_proj.bias\", \"encoder_cqt_model.spec_layer.lenghts\", \n", "# \"encoder_cqt_model.spec_layer.cqt_kernels_real\", \"encoder_cqt_model.spec_layer.cqt_kernels_imag\", \n", "# \"encoder_cqt_model.fc.weight\", \"encoder_cqt_model.fc.bias\"\n", "# ]:\n", "# del new_sd[k]\n", "# torch.save(new_sd, to_fp)" ] }, { "cell_type": "code", "execution_count": 4, "id": "13c2242c", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/home/georg/anaconda3/envs/nb/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", " from .autonotebook import tqdm as notebook_tqdm\n" ] } ], "source": [ "import re\n", "import json\n", "import torch\n", "from transformers import Wav2Vec2FeatureExtractor\n", "from suno_utils.models.mert.modeling_MERT import MERTModel, MERTConfig\n", "from suno_utils.audio import Audio\n", "\n", "with open(\"/home/georg/code/glockenspiel/suno_utils/suno_utils/models/mert/preprocessor_config.json\") as f:\n", " processor = Wav2Vec2FeatureExtractor(**json.load(f))\n", "processor.do_normalize = False\n", "with open(\"/home/georg/code/glockenspiel/suno_utils/suno_utils/models/mert/config_small_25hz.json\") as f:\n", " cfg = MERTConfig(**json.load(f))\n", "model = MERTModel(cfg)\n", "sd = torch.load(\"/home/georg/notebooks/dataprep/models/mert_test_25hz.pt\")\n", "model.load_state_dict(sd);\n", "model.eval();" ] }, { "cell_type": "code", "execution_count": 18, "id": "560fcb11", "metadata": {}, "outputs": [], "source": [ "# fs_sd = torch.load(\"/app/suno/checkpoints/mert_25hz/checkpoint_last.pt\", map_location=\"cpu\")" ] }, { "cell_type": "code", "execution_count": 15, "id": "dc1ab4e8", "metadata": {}, "outputs": [], "source": [ "# fs_sd[\"cfg\"][\"model\"]" ] }, { "cell_type": "code", "execution_count": 16, "id": "cdffe49a", "metadata": {}, "outputs": [], "source": [ "# fs_sd[\"cfg\"][\"task\"]" ] }, { "cell_type": "code", "execution_count": 19, "id": "1ce651fd", "metadata": {}, "outputs": [], "source": [ "audio_arr = Audio.from_file(\"../samples/halo.m4a\").get_segment(20, 30).convert(24_000, 2, 1).array_float" ] }, { "cell_type": "code", "execution_count": 20, "id": "a39e6824", "metadata": {}, "outputs": [], "source": [ "inputs = processor(audio_arr, sampling_rate=24_000, return_tensors=\"pt\")\n", "with torch.no_grad():\n", " outputs = model(**inputs, output_hidden_states=True)\n", " \n", "inputs = processor(audio_arr[:len(audio_arr)//2], sampling_rate=24_000, return_tensors=\"pt\")\n", "with torch.no_grad():\n", " outputs_h = model(**inputs, output_hidden_states=True)" ] }, { "cell_type": "code", "execution_count": 21, "id": "ceb72be3", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "13 layers\n", "torch.Size([1, 249, 768])\n", "torch.Size([1, 124, 768])\n" ] } ], "source": [ "print(len(outputs.hidden_states), \"layers\")\n", "print(outputs.hidden_states[7].shape)\n", "print(outputs_h.hidden_states[7].shape)" ] }, { "cell_type": "code", "execution_count": 22, "id": "2aca37d1", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(outputs.hidden_states[7].detach().cpu().numpy()[0].mean(-1));\n", "plt.plot(outputs_h.hidden_states[7].detach().cpu().numpy()[0].mean(-1));" ] }, { "cell_type": "code", "execution_count": null, "id": "c0a4fdc9", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "41120112", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "3cdb7c53", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "ccdc845a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "ba090008", "metadata": {}, "source": [ "### compare to default" ] }, { "cell_type": "code", "execution_count": 1, "id": "52961bfa", "metadata": {}, "outputs": [], "source": [ "%matplotlib inline\n", "from matplotlib import pyplot as plt" ] }, { "cell_type": "code", "execution_count": 2, "id": "35fbfa23", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\"" ] }, { "cell_type": "code", "execution_count": 3, "id": "b9b38335", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/home/georg/anaconda3/envs/nb/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", " from .autonotebook import tqdm as notebook_tqdm\n" ] } ], "source": [ "from transformers import Wav2Vec2FeatureExtractor\n", "from transformers import AutoModel\n", "import torch\n", "from torch import nn\n", "import torchaudio.transforms as T\n", "from suno_utils.audio import Audio\n", "\n", "model = AutoModel.from_pretrained(\"m-a-p/MERT-v1-95M\", trust_remote_code=True)\n", "processor = Wav2Vec2FeatureExtractor.from_pretrained(\"m-a-p/MERT-v1-95M\",trust_remote_code=True)" ] }, { "cell_type": "code", "execution_count": 4, "id": "e2e93864", "metadata": {}, "outputs": [], "source": [ "# processor.do_normalize = False" ] }, { "cell_type": "code", "execution_count": 5, "id": "87edb7ca", "metadata": {}, "outputs": [], "source": [ "audio_arr = Audio.from_file(\"../samples/halo.m4a\").get_segment(20, 30).convert(24_000, 2, 1).array_float" ] }, { "cell_type": "code", "execution_count": 6, "id": "cc50502c", "metadata": {}, "outputs": [], "source": [ "inputs = processor(audio_arr, sampling_rate=24_000, return_tensors=\"pt\")\n", "with torch.no_grad():\n", " outputs = model(**inputs, output_hidden_states=True)\n", " \n", "inputs = processor(audio_arr[:len(audio_arr)//2], sampling_rate=24_000, return_tensors=\"pt\")\n", "with torch.no_grad():\n", " outputs_h = model(**inputs, output_hidden_states=True)" ] }, { "cell_type": "code", "execution_count": 7, "id": "2a9d906b", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "13 layers\n", "torch.Size([1, 749, 768])\n", "torch.Size([1, 374, 768])\n" ] } ], "source": [ "print(len(outputs.hidden_states), \"layers\")\n", "print(outputs.hidden_states[7].shape)\n", "print(outputs_h.hidden_states[7].shape)" ] }, { "cell_type": "code", "execution_count": 8, "id": "ec1acef3", "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkkAAAGdCAYAAAAGx+eQAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAADlrElEQVR4nOx9eZgU1dn9qarung1m2BlQVFwRd1EJhixGIkSTSBYTl0TDh0sSjQuJ/j6NQaMm5tNoXBNi3BMVJTFGjUERo6hsioqKGyiKyL4OM8xMd1fV74/qqnrvrXtr6WWmYe55nnmmu5Zbt6ur654677nvq9m2bUNBQUFBQUFBQYGB3t0dUFBQUFBQUFCoRiiSpKCgoKCgoKAggCJJCgoKCgoKCgoCKJKkoKCgoKCgoCCAIkkKCgoKCgoKCgIokqSgoKCgoKCgIIAiSQoKCgoKCgoKAiiSpKCgoKCgoKAgQKq7O7Ajw7IsrFq1Cr1794amad3dHQUFBQUFBYUYsG0b27Ztw9ChQ6Hrcr1IkaQSsGrVKgwbNqy7u6GgoKCgoKBQBD799FPsuuuu0vWKJJWA3r17A3BOcmNjYzf3RkFBQUFBQSEOWlpaMGzYMG8cl0GRpBLghtgaGxsVSVJQUFBQUNjBEGWVUcZtBQUFBQUFBQUBFElSUFBQUFBQUBBAkSQFBQUFBQUFBQEUSVJQUFBQUFBQEECRJAUFBQUFBQUFARRJUlBQUFBQUFAQQJEkBQUFBQUFBQUBFElSUFBQUFBQUBBAkSQFBQUFBQUFBQEUSVJQUFBQUFBQEECRJAUFBQUFBQUFARRJUlBQUFBQUFAQQJEkBQUFBYUux+JPt+Del5fDtu3u7oqCghSp7u6AgoKCgkLPw4m3vwwAGNC7Bl8/eGg390ZBQQylJCkoKCgodBuWrm3t7i4oKEihSJKCgoKCgoKCggCKJCkoKCgodBs0rbt7oKAghyJJCgoKCgrdBg2KJSlULxRJUlBQUFDoNiglSaGaoUiSgoKCgkK3QXEkhWqGIkkKCgoKCgoKCgIokqSgoKCg0G1Q4TaFaoYiSQoKCgoK3QZNsSSFKoYiSQo9B+vfB+b8Hsi2dXdPFBQUFBR2AKiyJAo9B7cf5fxv3wyM/0339kVBQQGACrcpVDeUkqTQ8/DZa93dA4UdEP95azWWrt3W3d3Y6aDyJClUM5SSpNDzoKlnA4VkeHnZBvzkAYdcf/y7E7q5NzsXlJKkUM1Qo4VCz4O6KyskxNufbe3uLigoKHQDFElS6HnQje7ugcIOBru7O7ATQz2yKFQzFElS6HnQFElSUKgWKGFXoZqhSJJCz0MpnqQXbwD+9HknnYBCj4GtpKSywiYnVBm3FaoZiiQp9DyUQpJmXwWsfRv2nz5fvv4oVAfatwAv/QHYsqK7e7LTw7QISVIcSaGKoUiSQs9DGTxJmpUrQ0cUqgpPXgQ8eyVw99e6uyc7PSylzCnsIOgSknT77bdjjz32QG1tLUaPHo2FCxeGbj9jxgyMGDECtbW1OOigg/DUU08x623bxtSpUzFkyBDU1dVh3LhxWLp0qbf++eefh6Zpwr9XXnkFAPDxxx8L18+fP7/8J0ChuqBSACiIsPQZ53/Lyu7tRw+ApeKXCjsIKj5aPPzww5gyZQquuOIKvPbaazjkkEMwfvx4rFu3Trj93Llzccopp2Dy5Ml4/fXXMXHiREycOBFvv/22t811112HW265BdOmTcOCBQvQ0NCA8ePHo6OjAwBw9NFHY/Xq1czfmWeeieHDh+OII45gjvfss88y240aNapyJ0OhOqBIkoII+U7pKlvNbysrKElStdsUqhkVHy1uvPFGnHXWWZg0aRJGjhyJadOmob6+Hnfffbdw+5tvvhkTJkzAxRdfjP333x9XX301Dj/8cNx2220AHBXppptuwuWXX44TTzwRBx98MO6//36sWrUKjz32GAAgk8mgubnZ++vfvz/+9a9/YdKkSYEfZP/+/Zlt0+l0Rc+HQhVAkSQFEUoModq2jWuefAf/fF0pUVGgnqQ1W9u7sScKCuGo6GiRzWaxaNEijBs3zj+grmPcuHGYN2+ecJ958+Yx2wPA+PHjve2XL1+ONWvWMNs0NTVh9OjR0jYff/xxbNy4EZMmTQqs++Y3v4lBgwZh7NixePzxx0M/T2dnJ1paWpg/hR0QKk+SQkLEiQ49//563PnSclz08OLKd2gHh2X5r//y4nJMX6jM8grViYqSpA0bNsA0TQwePJhZPnjwYKxZs0a4z5o1a0K3d/8nafOuu+7C+PHjseuuu3rLevXqhRtuuAEzZszAv//9b4wdOxYTJ04MJUrXXnstmpqavL9hw4ZJt1WoYqg8SQoVwMa2bHd3YYcB70m6/mmVUkOhOrHT125buXIlnn76aTzyyCPM8gEDBmDKlCne+yOPPBKrVq3C9ddfj29+85vCti699FJmn5aWFkWUdkSocJuCQrfCVMZthR0EFR0tBgwYAMMwsHbtWmb52rVr0dzcLNynubk5dHv3f9w277nnHvTv319KfChGjx6NZcuWSdfX1NSgsbGR+VPYAaFIkkI3wbZtbOtQ6SPU7DaFHQUVHS0ymQxGjRqF2bNne8ssy8Ls2bMxZswY4T5jxoxhtgeAWbNmedsPHz4czc3NzDYtLS1YsGBBoE3btnHPPffg9NNPj2XIfuONNzBkyJDYn09hx0SnqW7QCt2Di//+Jg668hm8tbJnF8ylniQFhWpGxcNtU6ZMwRlnnIEjjjgCRx11FG666Sa0tbV5JurTTz8du+yyC6699loAwAUXXIAvfelLuOGGG3DCCSdg+vTpePXVV3HHHXcAcKaLXnjhhbjmmmuwzz77YPjw4fjVr36FoUOHYuLEicyxn3vuOSxfvhxnnnlmoF/33XcfMpkMDjvsMADAo48+irvvvht33nlnBc9Gz0NHzsQHa7fhoF2aqmaq7xsrt2F0d3dCoUfi74ucmW/T5nyI2089vJt7031Q4TaFHQUVJ0nf//73sX79ekydOhVr1qzBoYceipkzZ3rG6xUrVkDXfUHr6KOPxoMPPojLL78cl112GfbZZx889thjOPDAA71tLrnkErS1teHss8/Gli1bMHbsWMycORO1tbXMse+66y4cffTRGDFihLBvV199NT755BOkUimMGDECDz/8ML773e9W4Cz0XJzz10V44YP1uPrEA/DDMXt0X0fITbk1p27QCiEwagKL7ISDum3boQ8FNUbPDvlaKuW2wg6CLjFun3feeTjvvPOE655//vnAspNOOgknnXSStD1N03DVVVfhqquuCj3ugw8+KF13xhln4IwzzgjdX6F0vPDBegDAvXM/7maS5Ov7uqFmtymEIFUbvU0ELBswQoTTmnQPJ0kc6VSUSaFa0bN/qQpdBr27Q22mb5Y19PJc9papjBVVg8UPA8//rjxtpTIlN5GPMN3UpHo2UVdCksKOgp0+BYBCdaC7ORKsvPeyXEqSaeWhG6UPqJZlY0NrJwY1lq5g9Fj882zn/95fBXYtsbSQINyWFGYEC6hJ9ezn06jzo6BQLejZv1SFLkO3K0mUJOlFPhtwIQIzn5dsmAw/feA1HPXb2ZhTCE0qlIDtG4vbj363AiUpqc84LyAB2byvLmV6OElSKQAUdhT07F+qQo+BTUiSUaxp1jLZt1Z5SNLMJU6m+L+8+FFZ2lMoArS4bRk8SaYgzcT2rH+9ZHq6cVuRJIUdBD37l6pQNmxtz+GFD9ZLZfTunv7f3ukPgim9yL7YLEkyTVOyoUL3ocjBN0+KrBrFFbmmA79ISWrLquvFBX+fSDp7UEGhq6BIkkJZcOpf5uOMuxfinpeXC9d3tyWpvd0nSUYEYevISQYzXklSxu2dB8TYX2xtPzqtXfSw0NbpK0kiEtWToDiRwo4CRZIUyoIlq1oA+MnyeHS3JamDKEka5OTm1tlLMeJXM/HiUoE/yOZJUnnCbQpVAEqSBGpUnDE9Z1ElKXiNsSSpZxNsZdxW2FGgSJJCyXj1403e63aBCmPA7HaS1N5JKrTb8gHqhlkfAAB+9djbwZUWH25TJKnqUKxEYZLrwwoPi8lCQyZRFsVKkt9uvoeXxlEZtxV2FCiSpFAyvjttnqfObOd8F5ONp/BuzY8wMv9ud3TNA/Uk2SEkydtGuNDi3iqPyU4DasIXDOB0kWx8z1tRniT/GLkeTpKUB0lhR4EiSQol44b0n/BizYVoQDs6OJL0q/TfkNFMnN96czf1zkFHR4f/psgbtM3NZlPG7WpEGZQkO/x7lc3MMiM8SdTrpsJt3d0DBYV4UCRJoWR8x3gRu2obcLyxQBhuAwC7m63b+TzxnMRQksRtsCRJeZKqBGVQJfK5+Eqj7GiMkiRQimg3e7pxW6UAUNhRoEiSQtlg2br05t/dJMmmhKZcJMmysK0jh6eXrJHPiKsQtmzPYmt7LnrDnoAyDLht2/0UALlc+HktVkmyQUlUz5ZSVIFbhR0FiiQplA0WR4TYG2E3kySrdJLEG7UtM4+f/O01nPPXRbjm3++U0r1E6MybOPSqWTjk18+oWUIAGG2n2FAqmd2WzwcJLyU48TxJwWuMUZJ6uCeJv2x79tlQqGYokqRQEqgBkydJWfK0bHfj9DbLspHLEs9JjFuyaCC08qzCYFt5vLRsAwBg+sJPS+liImxo9T+LLLzZo1Ak6aWw8v45zUWEUWUkybTCZ7fRRbkeQG4XfbIZUx55A+u3dQbWqdltCjsKVIFbhZKQM224la5sjnNnTQu13rruIUnZXB6v/24c9syt9B8JijTN8goDTSbZlR4TmgxTKUngSFJx58MiZUlMkZJExSrJMaJmt9EHip4QbvvOn+YCALZ15PGX049g1qlwm8KOAqUkKZSEHBlQXCXJtm1sbO1ELk8Hgu4hSVuWv47R5uvYTafJIaNv0Cs2bcelj76FHM19wykMZplTALy4dANufnZp5PRonfxq1WCDsniSqJKEkGSjQDBU5ILWaxN7knz0JOP2h+taA8t4X5cSlhSqFYokKZSEfM4fXFySdNk/38aoa57FU2+v8daZ3USSbF1QYiKmkvTQwhV49DU/gzhPkiqRJ+kPz36A+R9tit6wgFwPn0oOoEzhNj+UqkfNbpOM6EpJEkMUWpMpoL996l1c9s+3VB4lhaqBIkk7MTraWrD8N6Ow6K4LKnaMHJk6/cfMLWjGRjy0cAUA4P/+8563rrvueZZeI1gavzOrt/r5lUTG7UpgQ2vQw0GhDMA8ymDcJkpSWNkaIERJYma3RRi3e5CSFOXPotvdMecjPLhgBZYJ1CcFhe6AIkk7Md78958wPLcMoz69t2LHoEoSANyRudFfRwYK3tTdVbBFxUoTKA+mZaO1M4/xf5iDe15cxqyzKpRxO8rjzlSbVySpPEoSmd2mCdqzpW985BmSFFxPd8v1ICVJFBIWpVGgZGoVeThRUOhOKOP2Tox8R1vFj2FyJOlgfbl/fNMGChylu5QkUWJA0SAow63PLYMG4P2121CzbgNAhKlKlSXRIgilxSgSPWewlaIMxu1kSpIsT5IlfO0do4cqgHHDbfS8imbEKSh0B5SStBPDNrPRG5WIXD54DHeQYZ6su0BJsm0bH61vZW7AwgEtofJwy3OOgmRwg6fdTWVJrAjvy06LnERdKAMDtxklSXTNkDxJkjaiPEmMAtiDvjeRaBalJCmSpFAtUCRpZ0a+8hmZTUF24i/qb+H29E0YiM3eMtuuPEm69j/v4Ss3vIDpr6zwj1tGtUfnSJJl5QHYGIoNZTtGHNABpseEbf51HvCbwcCGpcF1ZQi3USVJR2Vqt7Gz23rI9wbx+RKSJLJs3TYVblOoDiiStBODPh1XCny4DQDuy/wfTjAW4pr0Pd6ySg8JHTkTd8z5CACb2FE4S6aIQfVA7SMM0rZwzdj439RDmFt7Pk43nk7cpgzRniT/dY8J27z+V+f/vNvCt5Omw+4Elj4LZMUhaKq6ipQkuiRWxm3R99JDvWTC0JrIs0WWrVNKkkKVQHmSdmLYVuVJUl4QbnOxi+YrLJWOLtBZaIcO6+O9tqPuxjFwoPYRnqy5PLDcsvL4cepJAMCvUn8DcEuidmXQ4ITUdF3MltiwTc9RJAAAuuCWFef7fPqXwCt/AfY7ATjlweB68kDBK4YAe85l09Oj8iQxGbd7igIIsXGb9ynZts0s68z1nPOjUN1QStLOjC5QkqwQkqSR5+9Ke5LkeWeCN+goYy6PMbqkLhvxJKW18oX1Xlq2AYf8+hk8vniVcD0ddHI9SJEAICZJ7VvIG8n5eOUvzv/3/y1czShJQpIUeYRkeZJ6kidJFG4TEScrmogqKHQ1FEnamdEV4bYQ3xN9Ircq7EmSDVCWUEmS34AHYCu+rc9BDfxBU1ZSpVIpAB5YsALbOvM4/6HXxccl3e9xZUl4ktS+GbhtlP++WH9ShHGbKklFz24jr3t8uE0YjSQPVYokKVQJVLhtJ4bWBeG2MJKUIiSp0vc8ZkYb89ifLNz2j8wV2F1fh1/b9+Fl60BckDtXqhzYdjfNbuuJxm0XGvdc9+lC9n0ZSJLIuM3UbovjSYogBj0pTBpl0gYcAkmX9Tjyr1C1UErSTgzNqkxGaIqwcBujJKF4Cf21FZtx87NLQwlB3rLRH1txjvEE6nIb/eMKjqmF5NLZXV8HAOittWOC8QpOMZ4LFO71G+9+ktSTFAkA4nAbRbFsnAm3hYeHip7d1kO/tyjVyEXU+VNQ6A4oJWlnBlGSbNuGFjVtqphDxCRJOiyYlo2UkbwP3/6jU028V20Kk8cOF26TNy38MXMzRuvv4aMVbwN4qdDB0ozbQ7WNWG33E66zuknFsXuoIgEAMNLh66XfrYbQRJMWVZJs5yST30vi2W0Rg3xPUgBFhEdMIsPXKyh0B5SStJNi6dpt2LzNn+5sVuimbOXlahVNvpiChfZcacrLh+vl9Zzylo3RulMrbs/2t7zl4hQA8W/AvbBd6knqrnAbHUCSGrfzpoV3VrUIjbM7BCKVJMl1HvGA0Kvlw9B22Nlt4jailBDW1xTanZ0eUcRJptYpKHQ1FEnaSfHH5z9Emngr8hUqxmqFZPU2NEqS8tjWUVwfGtGKQ7VlEFRh8yB78hQlk0wyu62X1iGvO1cN4baEStLUx5fg+FtexK3PLYveuBrBe5L4wTRUSZJg44cYsPFVdhn33TKeJIkiFZUnibbR00mASGlTniSFakSXkKTbb78de+yxB2prazF69GgsXLgwdPsZM2ZgxIgRqK2txUEHHYSnnnqKWW/bNqZOnYohQ4agrq4O48aNw9KlbCbePfbYA5qmMX+/+93vmG3efPNNfOELX0BtbS2GDRuG6667rjwfuArQtz6DFCFJlapYb4UYt3klqViSNLvmF3isZir2a10g3UYW3hAtFZadkKAX2rt8dlsUSkkm+eACJxv5H579oJxd6joUqySFYd27ke3EUYGoaikKp9HdejxJEpwfqm72IMuWQpWj4iTp4YcfxpQpU3DFFVfgtddewyGHHILx48dj3bp1wu3nzp2LU045BZMnT8brr7+OiRMnYuLEiXj77be9ba677jrccsstmDZtGhYsWICGhgaMHz8eHR1sKvurrroKq1ev9v5+9rOfeetaWlpw3HHHYffdd8eiRYtw/fXX48orr8Qdd9xRmRPRxRjcWIM0fFJiVogkhWX1ZkmSiW0dxc22G6i1AABGbH1Juo0sT5KowG2S/N8NWoc83GaZyJf4EyrGzG7bNsboS3Ck9l6PyrcDIEiS+DBaMeE2XaBRhobbxOc8atYhQ7R6jiVJCFGYmF7KO2w4WGGnQ8VJ0o033oizzjoLkyZNwsiRIzFt2jTU19fj7rvvFm5/8803Y8KECbj44oux//774+qrr8bhhx+O225zyhHYto2bbroJl19+OU488UQcfPDBuP/++7Fq1So89thjTFu9e/dGc3Oz99fQ0OCte+CBB5DNZnH33XfjgAMOwMknn4zzzz8fN954Y8XORVcib9lMuM3MV0b1iGvcNmAWrSS5sEWDWQFsMV3/shbWbktATHbRNshnw1kmzBLnPhQVVuhowUOZ32BGzVX4+4IPd/7Ee/Q7LFpJCiFJhVmgb1l7kHbY64YZwGWXAzl0VqQkqXCbB1GYWM1uU6hGVJQkZbNZLFq0COPGjfMPqOsYN24c5s2bJ9xn3rx5zPYAMH78eG/75cuXY82aNcw2TU1NGD16dKDN3/3ud+jfvz8OO+wwXH/99cgTk/G8efPwxS9+EZlMhjnO+++/j82bNwv71tnZiZaWFuavWmFZNnbR1vvvK5VYMiQXE1WS0jDRUqSS5MIuDJDzP9oYMHHTm2peS2P9tk7cMnspNghqQIWlAOCxq7YB4/TXxP2xTeSJU6oYslLMWKBl/c/+wWcb8NKyri2w2+WgvrcQogyguHBb4beRBZk5x7XDfrXFKUmybXsiAmFimz0niiQpAMCF01/HhdNfx9b2yuf8k6GiJGnDhg0wTRODBw9mlg8ePBhr1qwR7rNmzZrQ7d3/UW2ef/75mD59Ov773//inHPOwW9/+1tccsklkcehx+Bx7bXXoqmpyfsbNmyY9LN3N4ZsWoDddJ8kXfqPNyqiONghs9sYJUkrXUmCnsKyda04+Y75OPaGF5hVjGkWKZzz11dx46wP8Idn3gs0oyUcSHfT1opXWCbymq9sFBP6KmawtMjP1oCJjzduT9zGDgVK8Csxu62gJGVtQpICxu04niT/dS4flWtJ3p2eAHG4TWXcVmDxr8Wr8Ngbq5DNd198eqfNkzRlyhTv9cEHH4xMJoNzzjkH1157LWpqaopq89JLL2XabWlpqVqiNGTrG8z7Rcs34uON2zF8QIN4hyIR5kmqgb+uFOO2dywthffWiNU76knKaym8tmKLtJ0kShIgr8tmmybyRH3ImRbSRrLnjuJIkj/gp2GiIROhrhBoWuWzn5cdNCkqT5ICs9uK+HAuSaK3Q66dOCkA6DbCcBu/fUgR450dUeE25UnqPnTkTNSm499TKgXbtr3fWnf+TCqqJA0YMACGYWDtWvZJfO3atWhubhbu09zcHLq9+z9JmwAwevRo5PN5fPzxx6HHocfgUVNTg8bGRuavWtGh1zPvdVgVSWAXlgKgRvMHNycFQKnhNiNeIj9CXISEKOFASg3wbDMWoySJ1IMoRIUVhLOkCGlII4/6TPxnndpU99/8EoMS8aiEqMV4krxwGyVJck+SLAVAJEniduvJITelJFUnHlq4AiN+NROPvrayu7vC/OaMbmRJFSVJmUwGo0aNwuzZs71llmVh9uzZGDNmjHCfMWPGMNsDwKxZs7zthw8fjubmZmablpYWLFiwQNomALzxxhvQdR2DBg3yjjNnzhzkcv4NeNasWdhvv/3Qt2/f5B+22sCVJDFgVeSmnM/KSRIFVZK2bM9iyaqtyQ+mpaQaEOtJ8gc7XbBHkjxJQJAkrcsU1EPLhEmOJRoYoxD1wNwhSMBJi/ZmtBxq0/F/xkm2rRpQ31vUNVxCuM2EgbytC9uJMzONDbeFz25z3su7tLNDNBuVLlKepO7BpY86iXinPLK4m3vCXgOVqBYRFxW/Y06ZMgV/+ctfcN999+Hdd9/FT37yE7S1tWHSpEkAgNNPPx2XXnqpt/0FF1yAmTNn4oYbbsB7772HK6+8Eq+++irOO+88AM7JuvDCC3HNNdfg8ccfx1tvvYXTTz8dQ4cOxcSJEwE4puybbroJixcvxkcffYQHHngAF110EX7wgx94BOjUU09FJpPB5MmTsWTJEjz88MO4+eabmXDajgw+G7ShWRW58eRC8iRRpGBi2TrHcDz6t7Nxwi0v4Y1Pt0Tux/io9JTUV8UqSQYm6AtxpPaeUElK6klKcwVPV9XuXeibiTxRH4oKnUV8Jx05QV8ZJclMFDys2dGVpKhPW8LstjwMP5QZYtwuWkkK2b6ngffv2eCJaM89NwoO6PXQneG2inuSvv/972P9+vWYOnUq1qxZg0MPPRQzZ870TNIrVqyArvtc7eijj8aDDz6Iyy+/HJdddhn22WcfPPbYYzjwwAO9bS655BK0tbXh7LPPxpYtWzB27FjMnDkTtbW1AJyw2PTp03HllVeis7MTw4cPx0UXXcQQoKamJjzzzDM499xzMWrUKAwYMABTp07F2WefXelT0iXQOOOpDqsiRTXNkBQAFAZMLPx4Eza3ZdFZeMqe88F6HDqsT3j7lu1dpDbnR/nXG5/hxEN3AQDkSYqDXtY2TMvcBAD4YfZ/A20m9iTx4bZC1mfbsliSVETym6iBUqQk2VRJQj7RgFKzQypJ5PzzJChunqQY7eehF4oZm6HGbbknyX8tDG2H+Jx6GoR5pBLUvlPY+UF/Ht0ZbusS4/Z5553nKUE8nn/++cCyk046CSeddJK0PU3TcNVVV+Gqq64Srj/88MMxf/78yH4dfPDBePHFFyO32yHB3eQNWB45KethcsEp9iKkNROmZWNZSP01EfJ5QkP0FPPDuWD6G9h/SCP2HdybMZAbRPk5VBOV30h2A85wJMkwHDUml8szKQDMXCcA1gsWhSjvRacgvxXvSUoynuyYniRCxCsRbitcO3mk/BxbgXBbdBeYmnqi2W0BT5K8Szs7RA9s9LfQkwmkggOTUZJ24nCbQjdBoCSJBtxSYeTaojeC4w3SYIWqHqu2tAfW5ygJ01NeqOPL+hv4jj4H763ZVtiw3dvMsv0f1NEpeQqAf76+ErPekUzvp33X2D4ZhkPblq7dgg7bf86wYxJGCtFYcKi2DLMyF+OL+mJhuI0qSWnkE4VRqSepkkkobdvGJX9fjCsfX1J6YxUPtznt52wabuON29EDOD2fncJwW89VkvhrLRBus22GNCpPkgL9fXQjR1IkaacF70mCVZFcEykzHkkCHP+MTDl5YvEqHP275/CLGaxhkDGGk9lt92auww2ZacivdgZhLd9O9vKPMQibA8fSYOPxxatw0cOLcdb9ryb2P6RSDjHa0tqBtpz/67XzyUmSaKC8P3Mt9tE/w/2Z/xOSKOo3y2i5RINtDZnaWwll0cXKze145NWVuHfux6Vfd0y4LWLKf1HGbed8mtD9HFQ2P4iT17JmIozbsWvx7oTgSQ+fAsC0bbZ2myJJPR7092EoJUmh7OBI0iH6hxUZFDMJSFIGOWagoJf9zbOdAsWPvv4Zs08+59fjs7lwGwAYG993XhAlKWX7g6oumMmmwcaMVz/13suIm6mJQ1OukmRwbRdT8FY0GDRq/mcR9Y2WWkmqJNWk/J98e7ZyBXqpl6pkxYQqSTyziHqfoP0cUr6SZCVXkqJrt3GH3cmVJDqu8Z+VD7dZNpcnaec+NVWP7lRuXKhwm0JFwRu3r0/fgZffWlb2EEuN2R69kbst5KqHbKA3c76SZNuCGUKtTlZx6tOpgb+PAQERsG3k8iZ+ajyGHxizpMfOG3XC5am0k4w0reWha/5gWAxJivo6RH2jJCmDfCISQqfSFpOyIC5oHpySTbghpW+CJEl2rDgpAPR4s9tiGLfFs9t6criNfc+TSMuymUFRKUndi+4kJS6Y2W07a54khW6EHRywF7z1Dp7/YL1g4+JgWTZq7PgkqRZZ6ROi7KbIFNC1rQDJ07dvdLYzfZJkEA8Rr/YAjro0qnM+Lkk/gmvS9yBvismNaYhN2EamQJKQZ2bK2UXMbhN9buqpEg6kNvUkmYkGW7uMA1FrZx7nPvAannprdWAdHQTNUmdVmiF5kuIqSfxNv30LcOdXgQV3oLXduYbDjdt0dlu0J0kUbuMZfk8iSfxn5YmzZdtlvTYVgnh3dQumPPIGVm6OLmPUneEtF+41091J6RVJ2kmhWcEs0TpsLPhoU9mO0ZbNoxc6ojcsoCbEPyNVc4gZ2ratwjjjb5vqdDxHlCRRyMJtX97+jH/swoDGD35mKkJJgskkqzQTkqTNbVnc/fLywHLaC6GSRD5rBjkkEYTKWUT0jjkf4d9vrcZPHwgWAKaeE1EJikRgruVwkpRf/ZZE6uHutC/fDKxcCPznYqzc6Jj//RQAAObfDiwjSXBjeZL8NVkBMeSX9CCOFPQk8UqSzSWT7Eknp4vw9VtfwqOvfSb8vfKoAo7kJW3tblVLkaSdFJpASTJgoa6MNXm2Z000IFm4TfYULiuZQpUkV7Wh6k2T5WTuloW6UoJwmwYLDZafiiBfUCoCnpGUWElK1zj5uIJKUrJw24UPv4H7530SWG6TAV1kKmdmt2nJ8iRRvlJqGGxjq9yoTmfllawKhKUA4EhS6v0ngff+Hd1m5zbvZUPKaTMPw1eSXv8b8LdvA58tKhyW+GVkqicNtwlmkvL79SS1hP+oorIklEz3pHPTVXDP6ftrtkVs2f3EBCBKUjdLSYok7aQQkaQa5FCXKd9XnjMtNGgJlCTkmEGaMXZKPUmskgSbDaE12oXyJgLlDJB7kuhgaxbCbbzKJSJJlq0hXQi31SDHKElJw20vSEKflCQJjds260kyC6GKvy9aGVnuhVWSSlN4wm6k1LhdsicpiXEbABY/FFhk812l+xWUuTxNAQAA+x0P7DIKABduk3STNW5HK0k9KdwWTAEQ/N74c6aybncfujvEBfhjQnf3RZGknRWCwaNWy5ZVScqbNhqShNuQk8rosoHUyrEqgmXbDDGptZ3jWxJfUUroSbJBhyyXJPFEzRKE22wANYySRJ9+y2OEjgq3adzsNsu28fwH6/GLGYtxwi0vhbdNmiuVvIRlwaVK0pqWDjz62kph9vBY+Psk8ibCkwQIYwWBmXxkP5uWJbHJLXGvr3gv6amS50nyXwsLE/Nd70EcIBhuEyhJ3DlTIbfKII5IVA1Kkvv1d3dfFEnaScHPbgOA36enIWXHq7UWB/l8FnVavLIkAFCjZaXhNrlxm1eSbFyZujewnZ1ASdIgVpICVdpl4bZMgSRpecbpUoxxWwQm3CY4X5agLMm7q1titd0vvwYXGP9AP7SUHNIIJ0n+ef/2H+diyiOL8YdnP0h+EP6cxko2pAHv/wdo8Q3lAY8QoyQ5v4lMJsMqSRpJvMnsK+kq6VvesgNKSLDA7c5LAmzbZr6qQLhN8Fv5bAsbulcht+5Dd4e4AP/30d0mckWSqhTrtnXgyTdXSb06URCF23bVNmCvj6eX2jUPVme8EiN22iEbTgoA8TYycy9TG862MWjjfJya+q+/qDCoJVGSNFisoiIJt1mp2sC+FnToqQwAR8WhxnC+qHAUUtIbEQm3iU6LIE+SFjbFneDKTf+Li9L/wE3p2ytLkgSenFlLorObB2DyvqcYJOndx4GHTgZuHUX2Cqnx5hJsPe17kgCA1ApkPElcF9o683juvbUBpSwqxcLOTAKCqlm0knTrc2wJoZ35/MjQmTer4nN3Z600F66S2N2iliJJVYrvTZuH8x58HXfM+aio/UUkCQBqOtaV0i0GVmf0VFIAsAthKz5PkqZpaM+a+N6f54mr3QOwabjNMpHJbpF0Rvx5ZbPbQJa7M+N4kuSSO2YZABgOScrws9sSEtq0If75RYXbaHqHtJaHace/kTRbawAAY/W3Sw63hcngwkSVxdzsctzEgCTJI0nJnMAnZWQOR0my9RRMUouPkiTWk8S29tMHXsP/3PuqwJzM51rilSR513d0BPx93HuRZ4vHjhZu68iZuOfl5fhkY/wEu/z+o387GxNumlPmniVHEo7UnjXx80cW45kla8raB/f30t2qliJJVYqPNzoE5J9cBuq40CUkKW+Xz5Nkxq1VViAbtVowT9LCjzdh4XJ5WgKTmHbnf7QRrR1sWM1tTqbipDVBuI0zbrsqVqAAabpB3CmPJHHG7aRKkuH8+I/X5+Oh9DUYWCihIgq3deRMf5AVhNvofSROwlBds8ugJMnXlS27e57zvJWjtodtC43bmpFCXqIkUaGTP20yAz5fjoXfr5K187oSrZ15LFy+iQkvRhXzjZMWYkczbt/63FL8+ol3cMzvn4+1/dbtOUy+9xUvz9jSta3Ysj2Hpetau/2zawnkmztf/Aj/eG0lzv7rorL2wT0FKtymIIGNRrRhW0dxHiKZkpSzy/eVW2bMvqV9JYkfGDa3iT1Nd8z5EBNumoPWNl+tyuXz+Pebq9gNC83ZkjxJIgSM2/mCksTfmASeJBs6YKQBOKEu2o6dMGmiqyT9MXMLxhjv4FfpvxWOQcNtNlZu3o6RU2fi5484de342W2WbTOqTtxM2iWTpJiz21wUdavjlaRnrwA2EXU1Jkliwm2Wye5n03AbVZL813GSSfLg1ZJgxu1YzVQ9Tr5jHr7353l46JUV3rJA6JpPfxBHSdrBTtC8DwuJbWN2+w/PfoDZ763z8hbRmcftxU5yKBOSiDertsafvJME7vefhLBVAookVSmmpGZgUc2P8ZWOZ4vaXxOEmQAgh2RK0i2zl+LcB18TPtm45CIKdoZ6kth2Nkhy7fz2qffw3pptmPW2X2NNFzpmXXUl/k1FgwXdpuE2sSfJrgkqSTYApEhZEtInK6GqwXuS+qDVP4bXpo37530Cy/br2lGTerqQAoBCFrrkUbqSJL99CElSMTc7UdHgB77nv475vTMkyWZJ0m4bXwYA6CleSfJ/K3HKkvAIXE/cfjsaCZDh7c+cSQN/X7TSWxZlUhcZt3nsaOG2pL3l7300/N6Wjf/QlxRx/IvJ1JvKfE8q47ZCKE42nkdaM3Gt/qei9peF26yEM7BunPUB/v3maswtPCVRMKbqEGjEuM2LHOu3hYfsaEhPQ1CNcMlgkkSOGmzoZNabS24C4ZBML7Qd+INgAwY1bhcfbpN7klglqTbFbqeRwSNdCLfR8STuVPvSUwD4r/kBPy5Ri0RekKx041L/dRHhtttmvyfeL6AkEeM2/Z5jDgqBlBKBENSORQKiEDabLRBui6EklWmyaNWCf2ig5297Z/cqSUkeaCp1Gbvff3ebyBVJqlKssfv6byQEIJsP1jJzIQu3GaJBJwZETzbxw20FkqRlAwNDFEnKkAK6InXMIwwJSZJGBkk/BQDbN01PIbvXeGaZRcJtmUJJVBd2QuO260niYXEkqYbLbUUJYV+tFabFhtjikqSSk0mSmxdvUi46JxKPbRFm0CLu0NP++4GQJDmeJP9c/+vNtfh4g2PCZfIkxTxtwa6FK0s7OtgZgHIlybZtj6AfuEsjvn34LsL2Si5n08VI+n3yv356jiqpJMVBiEgcQMVIkqckKZKkIEAO/lOsKOSwZmsHDrjC96nwkClJullc/FhYHiOfkCQJPEnrQ0pbAEA9KR/Ce4mcZQUVKAFJcip0EZKUd8Nt7HaakYZupJhl7Ow29vMn6QMQDLf5ChJr3K7hlCQ6u20/bQUs20YnISUdOSuW8bPI7BIeqCQfIEnlMG63rgemnxq+TRFKkgFLuJ9upGESz94/F6/BN251knPGybjNIyrcttMpSeQ1f/3R3z1VMB+Y/Dmce8zewvZ2MI4UGx05E8vWtQZmpNJTJpwd2oVIEm4TpfsoB9zfh0oBoCCERWV/MxjW+tv8T5Azbc+nwkOTDB4pszglSeQPcOuqdWi1QMMg4As/F/clI8+TtLE1PGTXAN+4rQmGJ/dzigr6yqDDYkikS274z6gbKeiBRyrNI0l766uwl+4nLEw6WylOCgDLBmrT1BtjM6rZ7to6GPk2ZjbZZf98C6OvnS31e7koVUmiMjg/kysnIEmJ73XLX4jepgjjtpQkpVglyYSBbZ1uegh/uyhy497U+espalr8jg4aXgzUQaQ1A0mozTA05vpm9tnBzk/c3p7yl/kYd+ML+O97bDoWm1GSgsTj6SVrcN3M90qe+VbOjNsz316Nf72xKnrDIuAlk1ThNgURmMFZQJKifBHCmmUAjGJJkuCH6Ybb1huDgF98ABw7Fdj1qODOGccALTJuR83EaiQkSYcdIEqeJymBH0gDmz/JNPNoz5r4BzGeAg5J0nReSfJJEo+kBW5je5LIIJIzbWaA1zUbfdo+ZkjSok82Y/22TvxtfrB4LkXJNdUIvvOnucx70fWZ+Imwtil6m5gkiXrHDJjC/TQjwySTpITJZsJF4cdKF367PGmOSrC4oyOMSDI1A8nrlK5JSyXtcMb2mN/n6yu2AABauHQm9Lxs7ww+9J3z10X44/MfYta7RSRlTYi4v9Uf/+21ivXB/fpVuE1BiJRObsqiGT62jR8Ys3CgJk42KZvdlk5AksI8BgBgF0iSpaX8X9X/zMTK789i+5Jh8yQZMNEHTiXqqKeiRs1PzKZrgbzJPtlJoIposANK0m+eegc3zmLLZjjhNvYnQsNtPJKa4vknpOCnc857bZqbGswdRzfbhXmJ5Bm9HZQ6CNH93bxeLsoy/tf0jt4mJjmmpNhRkoIdrNVNVkkioTd2dlvEA0rhvAcmAnDb7WgcIAr0vGzZzoWiKUkiH9wIIUk7WzgyCvRnzStJLSQVzLqWyk65B+ITk0ryF3dsULPbFISg5TSy2eCPYt+Ns3FN+h48WXO5cH9d8oSdtmImgAT7o+nMWVj0ySbGe2IVPEmmRtQW3QDq+rANZZzBrg6dME0Lj2auwBu156BP+4pINYMqSRAqSc77JCqODgu6RgigaeLJN1cHtjMMA0YSJSmhP0ZGUngliZKdjpwZJAaWiU6BL0CmVEUdPy7CkiOKWo5bOsXfIUa6ipjn3CC/p5QmDrf1ttuY2W1UVaID9vYIv4j7fQVnt3HvdzKWRD/euBvZUCklAAxJ0rSg506w3Y6AUnvLXmO+ktSRM3Hwlc+U7Tgy0Ht73BBXJVUeV1lTSpKCEEy5i2yQ2AxsWxpYRiELt6Wt+EoSlX+vfvIdfOdP83Dl40u8ZYySRKAVZn95KCgC9ehE3rJxiO6oX3uvezryRtioseE2Hi4Z1OxkySTpoGlZllD50Iw0NIMbqDXNm93GIylJkhFEutTkpvdvz5rBmYu2KVaSIkhSqeG2MGOySG1JfK+Lcz6LCLfpEk9SPtPI5EmihImeqp/PWBzwYFG4sxYDSgjv09nJlJIw5UemJOm6Bl3XGLVUtF13Y+nabRj/hzleduxKgJ6+NpICYOXm4iwSMsh+hvR+EDcFQCXpi3s+FElSEMLQ/JtwTqAkRUE2uy1txW9LJP8+sMDPquuSJJt74jd4YuGSJK0jcONz3gdvhsO0tThen49GkHAbrICSNBgbgOd+g/7mhngfCo76REmSKVGhDCMFnfssYeG2pCkA8pLt+bIk/KwX3n9lWyY6BXmJMpIUAy5KVpL4GUz0dTnGt7KSJD7cFtxvRf/PM8SIEiae9K3bJv8duUk2+egrf0p2Mo4UqnBYgtA9VStEIbdqIkkXPfIG3l+7zcuOLUKp36dMSUpH/I7LBZoFPW6Iq5IExksB0M3xtlT0JgrdAcZYLCRJ4b9IUWFXAKhJQJKinnRlSpIeoiTxbdZarXgx83M8Zx2KK/KTvOUv1lwUOJ7op5JBHphzHc4I7Snfjs0obZaZL6gcXCjPSMHQeZKkeRm3eSTNuB0v3MaaoEWeJFim0AAfFW4rVUkSzd4yCn0vy/AWiyTFOxIlxeP1VwIhy5vz30KGy7jNKknctREyOKR0sZIU9X5HR1wlyb3ueJK0mUupUU1KW1uM5I5xk4zKwORJIscLeBcrdFpoFnRd02BaNt7+bCsOGNooVaUrKfKYypOkEAaqBOUFhWRlU/y9/SUkKYX4teCinuTcemmWzpIiPreQT5I6mKcVDRrGm3MwTF+PM1KO2TvMbKzDZrxExSIQbisQDD6cpxtpaCKSJAm3xUloua6lA62FmSt8aQab+w84N85ANm2RJ0mQvDEq3GaWmCgpLIN0WQhAhZSkC1KPBkKjlq0jk9Jh2lRJ8l/zg2SYn0gWbgvmSfJfm5aN7/5pLi6c/nr4B6lihH3lbDLOAkkiI6woDcDO5tmKAv241B/UVeEmmprBsm3838z3cOLtL+Oaf78r3aeSfVPhNoVQ0Ju6iCRF7i+b9RNz8LJtG7967G0MxGbsra3EQdpHmJG5EodrH3jt7L7+v87LgJLEk6RGAEC91smoFzbsQMHdMGlVg8WQm2Khw2LOr5snKdC2bjD1uzwUadze0NqJo347G4f82jFhyot8ssZtOtjOfncdU5bEObAl9CRFcCQkrMcbQDBhoPi1i8S128pFktZ/gLTm/x502NjAhctM6KhJ6YySRDOfb97OpuHgk2dSGBIliT8l9PwtWbUVr36yGY+9sWqHTQ0Q1m/6WU2BkiQiSdUUbotz5Zb6tdHzZ4Y8cJR6fch+hzTDuWUDd8xxvKP3zv0YedPC4k+3BCwClVR5PCVJ5UlSEIENt4lIUvgNyZ0dlx8yCgCwbp+TAYgTMorw2orNeHzxKrxSey6erbkET9RcjiP1D/BI5ipng3f+hT03FEgSRySMVIiSxN342m2fcGSQc0iB5OaoAUhJDOlJYGi8kuS0yatvmpEGNPYnsh11RacAeHPlFgD+jz8Xc3Ybxd0vL4fFJ860TKGROCojQanJJENLT5TUstuI37/Z5mGSTsS4Hv44mnm7AY1o4UiPBR2GrnGeJOd1R84MzGjLSRimrvlPvlG12uh7ncleXj3kIAnCOI0pIAB07KvLCEjSDkoWiwWTjJM+TIb7/8sGRknivszfPuWoSr9+4h1meZd4klS4TUEEOoXfzLFPve+v2YbNbfJM1XnLhl4wfnd89VrggsXYcMCPAESH6Vy0SmLwKddQTrIh21y4LZUKmd3GPInoaLd8QvVB7Rk4u+ZZqVeGV4BKASVbLrnhCWR9TU1gGvpLh90IaBrsUx4OtBn1hMdPgZd7knzw4TYgmO6gI5vDqq3BGTBRg0z5PUn+67KoIe612nwQPrKHhG8Tp50C6tAZUP1M6LBtBDJuA8GcP0Aww7gLXdO8m3qA7IeE2+g0+PZy1b3rYoSFWOkq97qn4WCRcbuqypLEGKhLN27T13IlqVICG6sk2ehd49+b7355OQDgr1yC2ormSVLhNoUwMEpSzidE27N5jL9pDlZuahPtBsC5wF0SoKcyQN89kCqoPXGVpMjaPYZvXra4XEJGINzWC4BDsO6Z8x6zik96+f/su6SFLculJAFAjearMZZlwrYFnqRUilGS/tl0Or77ja87fek1KNholKrBnVJZyIZXkgKDD3ectz7dJBzIn3xzNR6TlK0BwsJ98RBWi0wYbiv2AJouTLTpbJN8JK1HZ+AcWtBg2TabcbsQCt4keCCRZYp3SFIhT1Ig3CZXkmgIROQv2xEQ7kkKhtv0CE/SKx9vKl/ndgAw4TZyefGkKCzUGwey32F7low5to1etdHzuioZCvNmQSqSpCACVZIsoiRt64jOB+QM+AV1xCVHhYs5LkmKrAJNZnjxnqSAYTjd4L2sgx861DQgZQcHIJnCcbi+1JnNVmbIPUkpxpPU3Kfe91EIzNulzm6zvZlhhCQJlSTOFyBR1+Z8sB4XPvwGNpIabqbNtl0s1m/r9DwLfr/I63KUJSk0mLc1xh8kPWhM1Go5gAtZWoVbYV6QTJL3IwHygUrTfK9NgERyu1z08Bue2kRJRFIlybJsfLppe/SGFUaYkkSvdd+T5K+n4ba+9c5v69HX2TJB3YlYniTu/aOvrcQP71qALYLrRwSZEsufV1FdxHKgjaQdsG2gdxyS1AXhth5R4Pb222/HHnvsgdraWowePRoLFy4M3X7GjBkYMWIEamtrcdBBB+Gpp55i1tu2jalTp2LIkCGoq6vDuHHjsHSpn1zx448/xuTJkzF8+HDU1dVhr732whVXXIFsNstso2la4G/+/Pnl/fBFQqfhoLzfb2f2VzDzNIWjJBVmbBVUHS2hkiS7+L0n7VStv5AjDIFsrUYKOc3x8dQTkmTbQMoOKiAyheMwfRkuTT8U1XUh3mr8EjBFPEvDDV8FSrnorJJk05+LLpjhliA+YFk24wEAgK8Yb+B841E23GZawW+MM+W75G6fQb2Ex2oldaAs8hlMy8Ylf1+M0+6cn3gm0XkPBvPFMCGCMtzH129zQohLVm2DdJgqgiQBwaSqbriNIagFwpSEJOmF+wggMm6z77dnTSxY7qgllEQkJUmX/fMtfOG6/2LGq58m2q/ciBtuc7dL6TTc5r8+Zj9Hpd0qUEerGXyIecoji/Hi0g340wsfAgBeXLoen//dc9L9ZfXt+Haj6l0WC5qbybJt9KqJQ5Iq0pVCH9xj7ORK0sMPP4wpU6bgiiuuwGuvvYZDDjkE48ePx7p164Tbz507F6eccgomT56M119/HRMnTsTEiRPx9ttve9tcd911uOWWWzBt2jQsWLAADQ0NGD9+PDo6HMXlvffeg2VZ+POf/4wlS5bgD3/4A6ZNm4bLLrsscLxnn30Wq1ev9v5GjRpVmROREMzsqzwlFjYezlyNs1JPiXZztrd9Y7KvJOmBdkOPL7kwsyiQA5oriPPtiGZP5PQ6AE5CSRc2bKEyVM7Cqy5MLQXU9xeus0wLmibI6B1GkgRKUtTsNnpW8pYtDCtOSf8dQzU/zGBZZmS4zfWfHbhLk/C4jB+EhpMsG4+8uhIvL9uIt1dtDe07D3dwZ7rFGLdLV5I+WtcCADDDqH2RalitgCRZts2QJHemm8j/J/ck+QNHQCkUdNUtJ8OQpIiyJzymv+KQoxue+SBiy8oibrjN/X1TtZqG2/rUOw9U1eTNSjwzk8BV/39410J8tkWcPdviwuqMiZs7r5UjSf75Ni0bDTFIEv/wUs60DV6qiJ19dtuNN96Is846C5MmTcLIkSMxbdo01NfX4+677xZuf/PNN2PChAm4+OKLsf/+++Pqq6/G4Ycfjttuuw2AQxJuuukmXH755TjxxBNx8MEH4/7778eqVavw2GOPAQAmTJiAe+65B8cddxz23HNPfPOb38QvfvELPProo4Hj9e/fH83Nzd5fOi3JgdPFoGTGJkoStq7EaP09wR4+LBpuKwzmWmGwj68kAaJ5FK4ixITb9OgfU9ZwlKcG+CTJsp0ZbTwqMvVX06W1wNxwW4AkaToTbrPpjVJEkiLkE42ZwWTFM2BaJmAD/dDiXxMcGXMJ8f5NOZygzw+cU0aZ4pQkb5synHI2XFB6ezWpgiIDuSfJ5mf6RSBbyINUa7PhKQs6du/fwCxzlSRReoWcaeGlpRvw7DtsRXZd06ThNtEpcdUUOkB2CLKnx0GpXpVSETq7TTBzi3pNqHF7SJNzr8iZdmj5l65EKcN0bSq6BiEfVhcpby5KPieSD7OdTNaJH25j3xcbwheRqx4Rbstms1i0aBHGjRvnH1DXMW7cOMybN0+4z7x585jtAWD8+PHe9suXL8eaNWuYbZqamjB69GhpmwCwdetW9OvXL7D8m9/8JgYNGoSxY8fi8ccfD/08nZ2daGlpYf4qBZYk+UqSFaPop02N28UqSbomrJXmkSQyDT6Omdos7JcGK+mKlaRK3Bg1cc4j+OE2oSeJnG/GFyMIt0UV2aW/ddHAK+ybmUdjy/t4rfbHmJ65WngcA44S9u0l5+H2zC04P8U+DDBhMPKTp56ickja5S5wW+smZYTck7SuJVldq1Y4imadze7Xp6EWx4wYyNbNg46UrglJe2fewg/uWoAz73+VWa5pPhkOKknBdlw1hW7bUaSCUimFIT4K/irhgOe/FuW/oUrS4CY/lE9DQN2FYmZq0u8zIyngy28vVZK4r7VSZJh6kkzLFprpefAqTzFJZGe8+ikO+fUzWPDRRma5yODfHagoSdqwYQNM08TgwYOZ5YMHD8aaNWuE+6xZsyZ0e/d/kjaXLVuGW2+9Feecc463rFevXrjhhhswY8YM/Pvf/8bYsWMxceLEUKJ07bXXoqmpyfsbNmyYdNtSwQzYlCQh+sKlSpJbeyypkmRomjBxY15zyMGmTv/CjUOS7MLxDXJ8ywYyWnmVJFuW6NEZvZhwk9cPSQqAxOG2iL7R3/qLS9dHbF1o0zax+0rnmjxKf99ph/Mk6bBQk9IxYJujMH5NX8i1QW6+ErIRadSPgagUAEnvde492nHgiXfO5ZMNotvhDMA8Of/8PoMCN2MTBlKGJgz/ykJiuk5SAERk3AbESlKxYaZqUZJEA6UoUSLNrk+N2/0bMh6xaEsYeiw3tmzPYsy1z2HputbIbenH3tbh39dqYpMk/31YCoBcvgJKO9hwW1yyw/8qi3m+vfjvb2JbZx4/e+h1ZrnbhZ0+3Nbd+OyzzzBhwgScdNJJOOuss7zlAwYMwJQpUzB69GgceeSR+N3vfocf/OAHuP7666VtXXrppdi6dav39+mnlTNKMp4k0w+3WbJM2gTUuK0VFA89IUni++DC9SS9+skWb5kRY8aZWwRXJ4V7LctGTbk9SfUDxMs1P3QT6JtlwrZs7Ktzs2m42W0WHUQFIcYoJYnigulvxNpOs/KB2nh8WM+AhRoi6W8rqCUu6PkUfX6gPE9rUSkAksLN6WXZIk3TO1KiNrfb4rp7mm4EzkEeOlK6LiTtMiKja5oXRopIkwTAHwAov0nqSXKRzVt4YvEqHH3tbC9xaaUhmoUlCrmYApVEl4TbGmvTaCiQpu2d3askPbBgBda0xKt3SX14NCWH+3nDfmKmbXM+JPlvqVKKYRud4GGLFUGesPBerVJmzPIlqXpEMskBAwbAMAysXcvG7deuXYvm5mbhPs3NzaHbu//jtLlq1Socc8wxOProo3HHHXdE9nf06NFYtmyZdH1NTQ0aGxuZv0qBpgDQqCcpBlW3LMuvccaH22JexNT8TZEthM361hGvjhk9C8VXkkguDssuvyep7+6SFXKSZFkWTsV/8LfMtewK3WDCbUObCPkQ1W+LMG4X87FsywyQJNHstloyO2ibXc+sd2fR2bYtDVsl5UiiG1elMm5bIUqSFuP30GqT8A3EJAm6EfhMdiELt4i08xm4vWaYjNvsfqKnc/e8M+G2fHEkybKBnz30OlZt7cCP/7qoqDaKOaYL9+OJbjGicBtf4NZF79oU6jPONS87z9WOre3+fc0l1GEFpy3L5vIkyZWkUkmS7KfOK0mi+xVPkvj3pdy705zi5mdm34mVpEwmg1GjRmH27NneMsuyMHv2bIwZM0a4z5gxY5jtAWDWrFne9sOHD0dzczOzTUtLCxYsWMC0+dlnn+HLX/4yRo0ahXvuuQd6jHjCG2+8gSFDJJl9uxiUTNhUSYqhVjCKhqsg6cmUJMsWeHQA5ODcvGpT/oW7vT36SctTkqhCZgdJUs42MPne8BQRYdCGHCJZ4Xx+UxSutEycgxmCfTSGPew7iBh7RWG9CJIkLgMS/n3YZh4m70MT5EmiSlKrREmy7PIpSaLZPmFmUyD5DdQmJEl+a48eMNbZfQAAV+V+iHZCmCg0zRB+JseTFDyGzDfkpBIp9CxGuE2YJ6kMxGB7F80MEw3mou/ZJQB/ev5D/OieVwCwA2ya3E8a69KoLyhJbd3sSUrys6DfL00b4ZKkTAhJygfCbfR1mY3bEvApAES/YT6xI39+bNvGc++txSl3zMfKzclyd/Hnp1pSAMSZ41cSpkyZgjPOOANHHHEEjjrqKNx0001oa2vDpEmTAACnn346dtllF1x7rfMUf8EFF+BLX/oSbrjhBpxwwgmYPn06Xn31VU8J0jQNF154Ia655hrss88+GD58OH71q19h6NChmDhxIgCfIO2+++74/e9/j/Xrff+Hqzbdd999yGQyOOywwwAAjz76KO6++27ceeedlT4lsUDJxC6bFgJmHjBSkTOoAMAyyY1FZ/MkxTVu27YtDrfZ6cJ6f13f2uiL2FUCWJIU9IZ0IIPPNm8HxGNZNCTT/EOVJGkIU2PMOlpkuC2KJAWXhQWSnEZjKkmERwWVJDdsZQs9WUBySVu0eVS4LTFJKpxPGxosO3A3BjQtWOxXAJfsv2HthbH6W+Jjed9zMJyQVEnyC9xyxxBs756Tchi3mXa7qP6b6DsXhtsKn+//ZvozcylJovXqetemUF+Yfr5dUh6pq5B0soELSpI6PCVJ3hafAoB9zW77+orNWLJqKw4Y2lRU32Ro41IAiD107GcI+PgsG/9zrzOZ4dJH38JfJ7N1E8PAK20usS6HX7IUVJwkff/738f69esxdepUrFmzBoceeihmzpzpGa9XrFjBqDxHH300HnzwQVx++eW47LLLsM8+++Cxxx7DgQce6G1zySWXoK2tDWeffTa2bNmCsWPHYubMmaitdUbWWbNmYdmyZVi2bBl23XVXpj9U0rz66qvxySefIJVKYcSIEXj44Yfx3e9+t5KnIzZSxLvTr305sP5doPmgWEoSk/nZVZLccJcW7+ZpQ6wkeeEaMlDv2hSdNsFVkthwW9C43YGM8LixoelA3z2AzR9zxy8YZDU9MFpZphkM52R6OzXncmQWFL1raJrjWaGKX4RfTEQSIj+rlYeVDidJOiz0TflqXhvHMPNkIJaH25IqSYKu0tBLYK2dXIr3lCRBCgDLBIxU4FwwKCQ8NQq/pTwMdEBs7PfyiXEfrD5jCAlHmCfJC7fFmN3mEgpKLIpNAUBRiVxjUcdxP59IUBWGbriUGC7Shu55knYoJYm8prUvO2KE20xbTpL462ZDaxYn3PISPvrt8WUtC0L9X7YtuV8Z/EMEu55exxta42UadxEIt1XJ7LaKkyQAOO+883DeeecJ1z3//POBZSeddBJOOukkaXuapuGqq67CVVddJVz/ox/9CD/60Y9C+3TGGWfgjDPOCN2m20AUiVa7Fr20Dn+GWwzjtm0GSRLzYyo8hYd3QexJ0t3jkz4aMXLVuCTlrswN/jFsK6AkmdBhlFKfTTeAs/4LvPMv4MkLvcXetGxBuI03XGeRRubiZQVPEv3hch4T6GDCPRGqhki+vugrewJz5fvYlhnssyAFwADdr+XHq2W+J4ldp8FiZ+wlgHM+5Z4b27ZRhw5Y0PEt4yVckpqOy/JXAvhS7GO4pNMSBYmtHGCkglnSmQacde51bEJHJ8SEXuPVOvcwti1URjqkSpLm/dbihdvc//7K7VkTbZ35mMn8xCjFQJvoOIKkh/TYtWkdHTlLeO3Te1KakwuqxZOUZHimZIbW33PDp2EkKW+yyk1YMklvH8tGpgiSJHsgagt4kqLDbTyBKSV7S02Vhtt2+tltOyQIEXI9QO7AGCejqdiTRAbaGDdQyxb7l7xyKfRxMRZJElT5toPJJNfZfUorYqungPp+wJ78YFxI8KcJwm0WW/ojq9cC6YIaE3LeeDJiR3w3oiez8QcICuUSaFYepkYHdjs4u02z0M/wpyjzJDNn+eE2GrZiVbDQbgT7JVhGBwnDyuHd2v/BqzU/we/Sd6Kf1or/bb8p2TFcv47IuF2YLBAabiusc0mSBd0LFweOVRik+ad2yxIrYNJwmw6SAoDPHRWiJJFj3P3ychxwxdNYKMhqHhcVScgqgMUM5sHPMnbvAYHtXNAB9xuHDMWhw/rgZ1/ZGwDQUFOY3VYGkrSxtRPfvO0l/HXex4n3LXZ8puZqVxkMy5fEkxL6E5dNx0+SkyhOrie+vl4c4zbP0cIUsChQXxrdf6ee3aZQJCwBSSoQpzgZhhmS5IURdLpBdBuS2W261w+ybo+xke1BQE6c2W3O5+nY70SnXWixfVPi4xjs/wLcbNniFADs8ZhcVEw77I8+oPBEqHyim1qNHn4j4We3GbACYT0dFppIuReeZLrhIt6TRMlU0hua6OmONjHYWg0A6K354cqkw7bre3PccfzduPA7CLuWC+v0GEqSa3zgj2PZ4tCVzBhNw20z316DUdc8i+ffd0owiZ6yRcZtF9c/7fh3Xlq6Aa9+nIwwdRVJygtIknstGbpfxy4n6E+KhG7qMgYeO/fz+Plx+wHwlaS2MqQAuP2/H+LNlVvxq38tKbmtMNBPSEOm7TE8SUnyJHnHS/AVx9mW95eJ7glxPEnFgjdui5KOdgcUSapCUON1llOS7BjTP62Q2W1OIzHasH0vB4WrJDF1ysZeFNmeUMGxgZqCJ8lO1RfatxiFIzFc5Yc/nutJEqYAYAc8pvwIQy45lYEnYpGz24I3kFrBZDsKzWbDbRnkhOG2Bi3LvKfIe0oS+/nTlCSFdyPYrwhPkigEK/NDyRA6u81TkpKF27ISh4EmKa1j2rZHMt3q9EBEuK1wcp59dy02tWW92VwiJen6p99He9YUmvo1TcPG1k784K4F+O60ed6gtakti3dX+9n+i0m2Z9s2Fn+6peSZdHRgdc3XJlEAXLXoV4+9jSsfZ0lKWBjFLYlBkzIWi6xZ/Gcs1rjdSdI4xPIk8Rm3Y+QcS6IkxQm/UrWP90i5CHqSxLmNigF/flS4TUEKiwwwWZtVkqw4BIeOVp4nieQ1iuFrgpXFvzOXBhYbNhtue7nXcWyxWxkE4TbT9pUkLe1MWzdgl+ZJcklNgJQVlCRBP3hPEuPTCZlaYYMnSdHhth8Ys/Bl/XVvWZTtxLZMWGQATyMfIGMGbDToPknilaQckydJpiSF94NH1Ow2wxaQJH6GWhQ8JUkQbrPcwTOs4264rVBENkxJkpT7sW3bU0vO+dJeXjhIZtzWNHl4QHSO31ndglufWyocxDQAG9toIlnn/xHXzMLXbn4R76xyiFKYQiHDjEUrceLtL+MHdy1IvC+FqPQKNdzSn8+9cz9mtg0jd70LP4zWMihJYeQkCsWmAOgUKElh4TbTtrnQJchr/82hw/ow+8QF/Z5kn4k/pmiCZIq7HwZqtyVO8+Fvzxu3e0QySYXiYJrBcFs+X1Bc4hT0pNsUfhFUSYqTRqBxzXw0acE8F36R1YKiFPMSEnuBSJ6kAknSyqUk8XXatDCSZIEO+7b0zhiuJIWqGgAat7yLa9L34N6Mn9U9ExVusy2G2GRgekqSWQjD9cZ29LW3etvwCqBomjnAeZISakki8ydLkoIKgCxHkwwuGexVl8Heg3txjTnngC/RImiENW7b4bPbeJiW7eVJSukaGmsdkiWbdUWVJIqt7TkpEX1vzbZYXkPTC805719etgGAnARs2S6fXfTQwhUAgEWfbI48bpw+uXDCNM5rGm4TIY6S1NLRvSQpCei5YD1JcZUk/72MMN11xhHe6xjPy6SNOD5UVr0SXZM8YeG/wqRZ92kNy4BxW5B0tDugSFIVwhZ4kjYUCnnGIThmYf88+XrpDcmKk2tJcoXzxm0R+RFBbNy2/YK3henaOmxhmC82NEm4zTVuxwm3yerjRYXbItIzZNrXBZdFkCRYFvu0hbxHxlxD92mp2fjG6lu8bYJKkh9uo2Z8qiQlnZUiGt/o6UkLlCRZ3TgpLJ+Ia/z3ySmaUtiWb9y2Q8Jthvg7t2x4T9SGrnmqzUfr24Tb0zxJFC3tuVAiKnoCDxuAAN8PJEtS+MO7ik/KGhd8vy1OSeJnQ1Hw/haKXgUyuq0MJCnsOFFIkhqDnguqJLnJH8OSSYYWuC0sP3CXRvStzwSWx0EcgYffRlRonFeS+C4ktSTR8yQLtyVNT1JuKJJUhTCJJ8kuhFraO52nwijfC+ATKUoINI2G22K0IRnQDHdf99cRkyQJjds2CElyQnZ6qeE2NzQl8ySJwioWlycp5o8yQBATliWpSemR6pNDaP0dM1rOU/NkxXz58+cOprZtMyQprVFPUkIlSdRXcsdM2Z3B9UlvN64RWNOD36dblDhqoLAt73zlYYSkAHDaX26xpY0sTkniQwI8dJJxm0KWnA9wzqUodKJrrCNGREgAeRjnrc+2CpcDycOrMvCmdodU+mGSMH4SGm4royeppHBbgm0ZkkQ8Se7iKOO2qA4eQGd5sddWIk9SnFnRXHuifXgTNb9N0nAbLcETaLtKwm1dkidJIRnsAknK2zps3QBs4K9zP8LPDs3GKqLqkiRqlNXIjSKekiTJzOwOwG4/YitJIuO27Q/UqUyhfavEFAAy47Y8BQCvzsnVsXAlKQom58mpSekBE3bgiLbJjGhp5L3zo6VqICh9FwhX0rIkMiUp6aCp6xoa0I7r03/Gv83P4d/W5xgSmLGDoZ5SPEmB4cojl9FKkns+LOjolCWTNJxb4WPWWOyS24BX7BGFPttenilD1yP9P5ok3CbLO+PsIw5thCXqA/z8V8WQgDJxJGF9Oj9TsvhcuAibteSTpHKE27pmlM0zJIlUFnAfUkL2DZ3dVmjKKXnjECXbTqba0OtLdjb47zInMCXxqhx/SVOSFOfBi2aXp31sz5q4ZfZSAMHcTF0NpSRVIVwSY0L3BuzVm1vx6yeWxAq32V6Ywr+4GON2nDYkRMENXdjMABYDonCbBTLgu+E2K7pUR5zjBPrv1m4T50mikPqs+AFBZ1WJqPNK12qwUJs2ItMGaJbJKH8Z5P3Za4bYMB9QkkhZEnpuSyGjGoBzUk/geGMhbs/c4rXv9zOoJCUNtzHXGL+r7XqSokmSW+zZhI6Ddh8o3EzzUgDouNX8NuZbI519SPgopWuh5ltAHm6z7LAhQ5MYt/lEffxTu/PZiyIBZZKSRJ4k145jkMSaIoR1u9ELt3WzkpTg1NLrn9ZXc5eHkRo+3CYiTO6plBVQjts32V78dykKt8nUHuFxYnSPpkqg+8561y9gr8JtCgG4SpIF3Rv0DVh4f22rUEniZVJfSSKeJJ0qSfLB8ZWPN+Gj9a3SGl8pmM7xknqSBMZY0/I9SVra9yRVUkkSZtzmiUrMH6WlcaGbyHCb364BCzVpPfJOYtsWE1JiSZI4dMQrSb7hl63HxxRRTqokaRr6YRvXV9JPS0CSJOqkFN4jtI7ArcqK6Uky/UHWhO6RcR66NOO2rxDouhY54MrDbfJzHKYk0aXBQax4Y2u5lKRAuIUoZo6SJN/XCJk56ipJreXwJJHvLGk+sCRnNk/M2jTc5hIJ99if3ztYX9I5b/57kXHbJUdGESSJkhnZJAG+uXyIkvTnFz7EeQ++xnxm2lfAv8ZMy8acD9YzBXRdUCWJydROHka6mSOpcFs1wk0BYEL3Qjo6HAOvSK2wbXCxarecAyVJBlkvHlg+Wt+Kk6bNAwBMnyC+MnVYMC3be4IPGGplkIXbXEJU8CQ1a5tRj47AtrHh+nQks9uEAzWXcVtkMi+sYXfTeZISUbuN7G7AQk3KiAy3wWJH1zTynlJkS1IvBDJuu8kkLfamT/MkyRPW2cInOU0LqnL0Blkj9CQVGW7T9OCd0lOSIgYKjiTpaUm6CsnsNppxOxWLJIlnbfGeEwoNMuO2JjXz0vfF5JEplycp3Lgd3rewU9mrkAKgLWvCtOySZjhRpS1n2sik4rdVtHGbUZLY9bv3b8DLyzYG9qXXB5snqUCG3ZnKGttuHNBhI27eJVESVfd7uPY/7wXWAeLr+Pqn38e0Fz7EDz+3O66eeCCzro2rF+eid61/b+3uIsdKSapCWKZPcmhhWNOyhYNqoEYUqZ7ugWFRYpL0wVpfGZANaCmYyFu2n+ivJJLkh3z0woBfo+Xwj5pfx2tTBFddkRi3heG2QMhLluiGe8uTpKhwG9k/BdPxJEWE22xYTGw/reU9pUiTGLcDniQSbtNIgWPGkyRoJ5u38LWbX8T5D70uWKsFjNj0CTUDgScJGp57b21guRRhniRPSYoimf5N2CFJYiVJPrvN9sIOdHabDJpkRldYuO2Zd9ZiTUuQVPLkKeBJKiG7cVKjvgx8n2zLvx9FhttCPUn+b6tUNYkS22yMZLwUSfgnPRdZgSfJM9oL2GEw3BZUkty+yAoox+2bLL9SYHab4FxF+YNEhvNpL3wIAPjr/E8C229pJw8xpAP0d9ZShpBrKVAkqQrhhtSoJymlmQVTpEBJ4heIwm2a5oV7ZMZt+tuRkSSPrNk0FBINkTpjWjZSbrgtTkJKANlC+RIpZCQJbrhNpiT5n9cMKERi2Fy4KzLjNhNuK5Cktg3hx7BshnxlkIehseobj14cd8qTmzQNt6UY43bwxjn/o414b802PL54VWBdUUqSreF///EWs+ydVS14WzYLK1RJstj/Mlj+DdaCDiNTJ9xMlifJspHYkySKIjm/Xfl+7kAS2I98vFJnElFUUknywkOR4Tb5ykxK99bTGVDFgJqNc/mEJEmwTKYIypUkX8kFxB4yR4Ej7y12HeCTI782YAJPkiClgKgPFELjdsRDAn890LDjvnyuMzj5w7x9JZ6scpj3S4EiSVUIS+BJ0mEXErWJw23Me9e4TQYWXfOJj8xgzDQj2caA5Qy6ESSJn8kkLAdi28horHE7CtkjzgnfwCU4PCkrnIu86JLn1IhgGM3bkH3HbRc1a5Dec1KwcHx2JnDf10P3cZQQqtDkfLO7REnapZHtFy1LQr+VVISSFHYL1jVnSr2LnxiPs54kAUkyoaN/L5/YZfMWjr/lRXz91peENbr8a11Akor0JMnCbTxJaqrzz2HOm92mhea6AeTJJHmlIA50jfeSsOv97zU540myy5JVW/HppmBiWSA6T1JouC1ClXDJTS6h+sODhsySKkkiyM5dWAoAm3i1RCHbvBmdJ8kl375xO36fZcVzmW1ikPCo0C5vEG8jobIhTcEHFEqSbMnnL4d5vxQoklSFoEqS60n6v/Rf0MfcKDRu84Oz0JOkaR5JilPaRKaKGDBx23NL/VlFEpLE/7xEBm+NJBzUZF4RDrokLOJ3MCPsl3ujFHmSeNJo8oZsb0PuU3EkRSRPs7v7350BE2duuSVka3cfPpmk6alBsnPGZ6HOk7IkjHGbJO1MOs5qXLjt/6Wns0qSYHabDQ2fbGzDd/40Fx+ub2XqhonKT7ifwyH7Yk+SbvrHedL8XLCjcY3bHEnac2CD99oNnaSMeMZtWQoA9xz/79dGhLbhwkZ4uI3PwJ0EcXdZs7UDJ9zyEr5w3X+F6/mB1bJJuK2EFACATyZEBuIkoL+fbEIlSQRRuMrmjNc0SaKzHqEkiaZOcN/TfQGiJOlFGLdjKUnse9HsNqc/8uMGiDy5J4rI7pbt4nAbPUY5sq6XAkWSqhAuEbJIuA0ALspOEz8GcIRGnEzSf88TrY2tnbhv7sfMBSt7Qs9oJj54+TGvH7Lp8nxqAEswq0wn/XBrt0VBNyLmGrghsEDMQ+5J4kmSXEliUVvLDrhuORnLsnHxjMW49+Xl7HFIuZm4pVdsy2KmuTt5kgokSRJu03mSRAZTWQqAqFk//HpdGG7zt6kR5UmCju1ZE4s+2YwLp7/B3ISFoRdCxGWJO428k/n67OxF+Gv+q8E2Cp4ky9bwxM++GFtJ2nOAHxpY39pZ6KMeSZJktdtMyycmI4c04luH7RLaDhCuLrjrgfDBUhaSizvL6+ON4sziXh8CKQD8GmSRxu0oJakQ2pEN1nFBe5hUSRKdJdE55U8nfxzT9muhiUK2psWSFJE/SePCbUlm6smK58q2AcSeN564u3BVP75teh5e+XgTfjFjMd5cuQXzP3KM61RJYj+//1opSQoBUOM2DRsNtDYIi9MGfixM9XQHGlGS+O3P+esiXPH4Elz2z7dIE/KbyX2Z//OVJqmSxN4ARfXQmPpektARD5l3xG/HJzh0EHePnxcpSTbnSdJkfWHP28A+vZn3Lkl6Yel6zFi0Elc+8Q63O1GStJg+C9tijpvRcp7hmiFJTcOA464BAOhcSRD3ac602IzbYcbtmW+vxj2E5PH3RU3TgiSJXDNRs9s2b89GG49tn4gHZhkVCHYq74SB1tp9GVO6v51zjZnQUZ8xvFQTPNzrao/+9QCA7x2xq7du/bYCSdK0yJlRusSsTGcvRYWhXGRNK9RL4g5WYWNlh6QQb1zUpv3fm0iFCebJoZmSIzxJEf4WtwRGNl+akkQv7qRKksgcLSKl/HnglSSqFIk8STnTCsmT5Pzn8yQl4XvM1HxbTLCCxm2xYib63boPD8z5stk2cqaNvy9aiW/e9jJOvmM+Fny0ES2UJElU030Hs/fZroZKAVCFcIvYWpoeMDyLyEsgGSKpecUs9zxJ7I3zVUGRyygT8pufbsJhKYhdqhCE2wR8PEVJUkzjduxwG1ii5g6yeYEVkw9PWbrkZ8GdE94T5JJb+sO/4Zn38fPj9ivsLjZNh2FTawfeX9MBVyOpQc5TkvQUOf4eY4GhhzvLeZJEjNtMWZKQjNs//ttrXBsWDI6g8qFLeuN1DfnM9pyyKUq4x7Tn9lvTIQu3uSSpDbWoE8yog+mn09A1DSmZcbugUD7+s7FY19KBXfrUB7ZxPEnh15+ui9UTGm6TqU08cqbFzkoS5CRy25ahI2eioSZ4PccVIWqI6tGeMwMqSDDcZntit6FroebsKCUpUyYlSZbkMd6+wWUiJYX/Djo5s7lFZv2JfG0XPvwG9hnkq5dCT5KXAqC0cJuzbzCZJ0+cROEx2xYrTClDA3LB44T5yV5cuiHEuO2/vuF7h0jb6AooJakKwYbbokkST2i8ZJKcyuMbt+U/rtONp/Fl/fVguG0Ie6F6YRvp7LZo43amUFPDgu7XXIuAEbUdWc8c000BIFCSNG5CdF6Pp2rxfTYF382tzy3DsnWF1AqEnBqwAmqb8BCwsGZru/e+Dp2+AkRJWqrW64/ck8SG29hkkuE3XP7mJ5rdRsm3LggnMhngNY2R4oXjoPc5RMbtQrjNdEjSdruWIYB+xx3iZMKZLaVLSJJLABtr09h7UG8h908ZGtJxlCTBJk4KgELYBPESQOZMK3R2GyW/MrQXlKTNbVn8Y9FKL6Ff3BQAtJ8iVUo8u80f1MPyDEWdAzcJpGiWVRKUEm4TnVvR7ZPfjN+Gzl6rSYuJ9tJ1rf72An+Oe7pcjiX73l9bsRnrWthcc7K6f2HLxOE2W+i99JQkzrgd9t2150xsafcfbFi1y3lzyK5NQsN3V0KRpCoEQ5KYJJA2WtqDiRZ5I7YbkguEvFzjtiSB4SHaMlyVvg/3Zq4PGsTPeIJ5G0WSAp4kQQqATEFtyGup0IQkHbWDvNeyfDYeGCWJkqTCTBkhSbKY2XiBTNr7f9P5f/jp0mMBQGcuhxc+WM9M/wXIzcbmSJIkyzOFDlb9qdc6fXJD1bd0nUeSeE9SjsyCkqYAINuLCFOgkKllB/Mk2cxdLtAGP5EgSknyEkXKlKR8FkYhnNYKyezITR8BKJi2NcDINAg30zhWJFKDjBjJJLXQcFv4NjyyeSvwdE2/Gy//Vci475Z9+NE9C/HzGYvxf4UkgHFFCLrd9mzwvhHI0UbDbXq4YhYVcvQ8SSXOSKNdTJoCQEgmBOQhKh2DSb67OglJkh3XItcN4J830Xf4xqdb8O0/zsVRv50tbU/W3zgpJixLTHzcEGKQaMnPd0fOZMuS0HCb6+bo7nTbUCSpKsF6kvyvKG8BDy9cEdieV4Zs4uVg2nWN25IEfM3aJtKmf/FmbQNIsWzeH2zjhdtEeZcymusXSQlruwEAvnc/3jno/3lvtSgliXiSGB+UV5aE+JQyjrzN14oL5En63v3AL9cCTbuyyzmSpMPGGXcvDOT18G5qhHimYApLtfAIkCR0+rPSDJ4kOe1pFnt8WpaEfgvUkxRmEAYE04NtO/CdXvL3NzDz7dXSz8J65NjQh9gM6+ZJ0oLGbcsEcr6puD1AkgrH+tdPAQCNWjt0TYORkcxu4yYEiEJBpWTcpgqLpsUr2pkzbea85y2bOU++Jyk83AYAi1c6uahmLlnj7BN5dL/fLtoFJInnLxYxbhtaeLiNL5bKI10grqUkzQRY1awzsZIUXCYyPkeFvmiSzbpM9O9eFHoKepKCx5z34cbAMqcNrj8x1DARbNhC4uP6x+jptW07NNzWkZN77ugMye6GIklVCOpJ4smDIZoVxV/dZjBPEkA9SeILl25tMaETOxBaatAKipbkZh9UsURKUoEkaSl52G7kiTANMrBpOjBopHhbgCEuonBbjuZvqm1yduHOaUDh0TRAZPjlkkm6LfM5ZVwyYHPhtjghRr7gbz06fAWIHj/lK0mycJuTJ8lvK03N4/RpW/CkyKsIpgUY3FDb1pHzvEyi0BdDksD6FUQ3aA0+2Q9cZbYJZB2S1GmnkOPslaIZioauIZMSD1D8hADRZR0/T1JwOZ3dpiFeuCubt1hvisVOM4+TAoD3xvSpc34fcWdGMSRJEG7jyYGTAsB5rUWE26LUNFdJKjVPEj0/yT1J8ZSkKB5HCyXXxlKSgq99T5K8b9LjS/xs7DGj27Ns4LdPBUuSuEoS325YuK0jb0ofzvgQY3dCkaQqBC1Qa3MDKa96AMHwmVxJCs5uo6/pwEbN4LqGgEG7EQ4RkKkhwXCbQEkqDPaWHkKSwJE93QB+8A/gmF+KNybni02B4P6IyXEKJIkf0OMW7Q2SJOecLV65hVnu3eR543YIScoOOsTrG+1fnUY8SUy4Te5JcsN/oQVu6bEFg9LRv3sOV/zrbe8935bbVx8iPwPnSaJKksi4TZSkAGtZ8GegowUAsL2gItHjd5gCg74G6ew0XkkSDe5GzIzbIpXIyVJf2EbXYj21i4zbdFCJ40nq4GZZ9al3rtm4wyttWuRJCpIknwwYJWTcBsrnSaIfIinhEp3awDT3vIWT75gf2g417tdGXEOAzJPEhtuKTSYpeh+3vfdWt+AJQQZ+V2FlxhaEn++OrCk1a6twm0IoPCUJeqDwpiY0xPL7+ySL3S6oJP3ysbchAlU9RKpAb62glkgIhWHwBE2uJFlhShI4wqUZQONQ4EuXiDemShLzA3OVJEKcahoLa7jnemmBW/mx3HYA4M1CaMNFzjNOs8kkQ2fqFdo2uKBWPYgniR6/rq9H2jRudpurJlgWS11TktltMg/IffM+YbbhZ+hRAq9HsABNY0MfwoE+LOP2xy8CL/4egDOzDeBIviC8q2vy2WmRqSXghIeiBva6TEp4Y6cZlx0lKRo5LgXAxxvbGEUvjpLUnjWZgatvfeGaiTnA0u9F5EniFQqbeG8MSR07F7Fnt5XqSSKvy5MCgH3/1Fur8e7qlsh23HOZMqLJNptp3Q/TAj65FKmBslMamIUYM7UBD1HSV8AnSawaBNz78sfSttpzJld+JdiXOGHpSkORpCqEZ9zWdGgceRCG2/jwGX0Cp4sF3pgHF/geJzrIUCIlIkkDUCACEnLDhzXEs9sKZFBPR5Aksi5BniQmxFf47GP28k3gqOvr7BIIt8VVksQkSTYNVqOeJM0K91cVPofjSfL714AOPxElVZJ6NRNPkkxJAqckUeM2fdqOJ7vrWlBJ8jlEsA16bF3TGBOt6KbtJdHUdQhvVW//AwCw3a4pHJ/0T0KSZLPTIlNLwEkmGYVeNSkhkTJt/wxrWjz/R860mcHy//3jLXzt5jnee1dJCvUk5U20EXIjUpLCCqXSVeJwW/C922dNCw+p9aoNDze7PpdcqZ4ksnubgOiFQXRo/nyJzouoHf+8aExqBRFs2/9eZeE207KxobUTv35iCVOcXARRPqtgHwskrogYlxduIw2v2LQds99bJ92nIycPt/GlWLoTKk9SFcIvS5JiSIENDbogYV4wT1K4cVtWloT1JIU/ce2uOxc/T+L8trhwW4hx25aRpHOcASGRysOdL2+3wp1lt/4NwPLCwoYBzi7cdPxiw22iae8ACV+R8z6sKR3uSSrkQNI19swx4TZK0no3S2e3eZ4oW54nid4044QkTMsOkEsdNlFRRCSJhHYLKQAa0A4NtjgLcGGZLVKSCLYLZraJ0isYmtxTpMfwh8UZPBprUxJPEju7LY6UkzWtAOFe2+In6aSGfB5pQ0POtNGRs7Bhm7+Pq2Aws+QsGxnJZ6Ntd4hmtwmmltNwW5gQQOvjiVCu2W30M9AcZnEgui6D6lm8dtxbqq5pqEkZ2CbIJUbh5jKSGbcvfPgNrN7qeEPvn/cJPvzt8fK2uFMozvXk/E8bOvKSGdCahOGLUgBEoSNnCX1ItJ04SVcrjSrgaQoBeOE2wwlFEYhUHRvuVGAbM99ejW3tzk2RH+zdgYOWuThgaKOwbVGNOCFkhGLECc7//vs4fQvJk2TrqaBCpKcDuZmc5TFDYQCbJ8p93UmeuAokKXBOi1SShHl6QEmKf04vm7Bv6GfRCsZj3pNUj05S0oTcQHoP8UmXxYfbqJLkt2VIypLEIkl2kCRRJUl0Luj2GoBsLocltZPxdu2ZsHKCRJBhniQCV0n6wPJnH4quN00X180CAD3GI2ucmTa9asThNmZ2G+IqSVbodrTcDA93mnlHzvQGUroPoySFHIReFyLFJKBQWP5nM3TNm6EmQmNtOEkqX+02//WW7YLrLHRfAUmK4e/hQcNtRgwlibbrNu97kpz39HuNk4Igqs80HCiDTLV0yXeSmYiOkiTuIyWU3Q2lJFUhbDdLsGYE1CCRcdu90i5/7C289srL2F1fj3Hp4EDhK0liiZPpQ6kk6YTfA8OOBEZ8wzlmSJ4k6OngIEjfU+UrLoEBp6R5JIl4BwpT6HWwAz6fwFOKwQcwb4XfDQjpIJ+jb60efLyjKBiJddhMu0wyyQ7ifarvB2x3pv86xm0bLonyPEmckpSSGLdjhdssG7oeJEmugigiSXSZrgPo9JPn6ds3ABjA7sCUvpF/7x1wyOo69MVxnf+HbXY9nqy5LLCdHjI46anyKEm9alOBPFlAYXYbUZJiTbe2wwmrSfJf8ajPpNDSkcfKze24nPgOTZMdeIHgwGZaNn76wCLs19yIL+zjfydx8iRd+5938fWDhwBwPqeoBIeLSCWpcL5zJWbcpj1k6lPGgDATPLcsDi3gU0DEIUmmZSNtgOzHGreTICpRpE2M5WFpLmQcyP08SfJQBcNt/jq/tE3s5ioGpSRVIfwUAEZADRJ5ktzwWWrRnZhZ87/4pXG/s0KaAkCcH4dRkkQhufNfxz358ewyGaGo6Q0ceSbQe7DTN4Hng1GSQsjPPoP9dP1JlCTqyfKUAkosCsc0YLGqSlwitsdY4Ft3APtOcJrjbpeu8uDPbiODjJUHzGB9M6+/hfPqGLf9dhu0DqRcLxDdX9OYc0OvE7eOlGnbUiWJTQEQT0nii/TqsDxxK0pJaus08evHyWw50Y2fGLfDxgU6/f8DexhWo7803FYryVGjx7iu4ipJwozbZHTRtXgpAAAICZeLvIDwuKgvfM5pL3zI7uMpSWRw4kjxgo824ukla3HL7KVMv4WeJG7UfHHpBlz/9AcAnFBROoQMNNaFE9MoJWlzWxZ/nfcxtkqIz+srNuPqJ99BGzEbb00YbhPmSSqCs9HUCLoWbdx29mGVQj7cluj4EcZteg2FEVsZagoe1CRKUparVyeadV0NeZKUklSNMP1wm8mREKHvxbbQkTNxVfo+AMBu+npnsXR2W5FKUr89cdgPfgNMf9pfFvMHe9iwJoC9X/t5eoxMKEkaUE/CWnFVHrCKkJAkFZbpTmCT7Jng2eGQ7wNbPgE+mIl9BzUAZHZsfcbAto48cm6BTpsnSQXpf+hhwKrX2XYLA4TGDfe94JcowaGnAu/9G9j/G4Vu+z/nFEyYBeO664mybZvxtKU0MVmOIklWwV8TTAFAAoA2fVPoHhmYl29owwBC0kw7eB1pVEkKuT74HEnO4cUpAOolOWrizKKJc8PuXZsWDsRsuC2ekgSEk6RsXq4kyRIWuuoTqyRZsG0bU/+1BCOG9Mawvn7dOnp7iFOWBAA2tBYKAkfklYpSktIReZLOffA1zP1wI555Zy3+Onl0YP23/jg3sGxLYpIU7UmK82WaFkmyqWvS0iTssdk+eOG2kNsT6yu1PeN8VDZt+jlTRbila9POPknLvoiSozr9cf6rFAAKYriVyzWDmaHVD9swUvsksLlt28KblTSZpC027NKtbUtsKjx4WH92QcwfVEbwdFJTUJI0Q2TcloTbkihJXF4eABxJcm7SX96nL6uKJDgGAI8o9qlj92vIOIN3VjC7DSvm+3057jeCJp22dFjs7DaNqEf1/YGfzgOOuYz5PACQJqbQzsLgZnFP5LIUAFHhNlcKFxm3vRk4EbPbnOP7fRTOsIrpScoKSJJsdltKMmjHEQ95kiQKv/WuTQmf9E2bLXAb94GbTwbJrgsLt4mv4facicn3voKVm32ybVo2FizfhL/O/wS//OfbjCeFZlcWkZWwS0WLUEx6R3iSovIkzS1kl35x6YbQdihkqpMMIv4jSqAZ3Q6taRc/3ObuC/i32rhKEg1T8n3OmRY+WLvNa5v6gUpRkrwHwpig3RLZQKpASFIkqSrBhNv8q2SYvh6npp4Lbm7ZsPJBQyKf5doLI5EfD0uuiNwp8QEEpkonUHZ4ZBiSxLfLULaijkfDZkIlqWC8bshtRqO2neyX8Jfphu24mYfuQCXyJGHBn/zXtY3goXkqV0gZXH5GVqrG+0xN8Et2uIPp1nY2vGfAwveM/+KnxmMMpYmaTeTNXhIYt91TJ8rnxZOktOaTJDEpd3ulIyBLEeRsEUkK3tpKqUgP+ObUeyYdif/5/HDc+P1DA9s01qal4TavwG2ScFtO/l34XrPgOj6r8y+P3x8A8PSStYFp2XnLZvMHkfZoKRIRkQ1LP2Bo4WVcopS5dJlmt1HQgqpxEKfGWZzs5aZtM+pIHJLEpwBwVZUwdYWuEikzLq58YgmO+8Mc3PbcssKx/HVRpXdEqEm7xu0k35Um7WM1hdu6hCTdfvvt2GOPPVBbW4vRo0dj4cKFodvPmDEDI0aMQG1tLQ466CA89dRTzHrbtjF16lQMGTIEdXV1GDduHJYuXcpss2nTJpx22mlobGxEnz59MHnyZLS2tjLbvPnmm/jCF76A2tpaDBs2DNddd115PnCpKAwYJlKIYwu0bUs4O4j31rhP1zQFAL1I6dM/X//LX8G2WYoc6hq3NWG4jSpJ5BwkUpIEJGligZwcd403hb9x5X+Z/QY21iMZnL7yJMkNeaxt6cT98z5GR1byFFvI18S0qFGSJLkGeJKkaUBdPwDAeanH4F47nXkLcz5Yj//39zeYzVMwcV36L7gk/Qjqti7zlkdJ5t4TsXB2m1xJ4suY0BQEtilQTJhwWwhJEiQqFZ2xUoqtAvBCR8fsNwhTvzEyMNB9eb+BOGBoo7TArTfYQYvn9kV4uM0lUKJB2lUxAYdshOUkMi2b+SwtpPYg9SGJ/TkhJClGGZcwlCtPEkVyT1Lw2IGiviH7u4O8ZYENt0nK41DwKR7cyyqu0EP9Qfz39PIyR4W7YdYHzDGA4kiS+z0nCbdpGgIZ5fnXPSLc9vDDD2PKlCm44oor8Nprr+GQQw7B+PHjsW6dOMnU3Llzccopp2Dy5Ml4/fXXMXHiREycOBFvv+2bPK+77jrccsstmDZtGhYsWICGhgaMHz8eHR3+lMjTTjsNS5YswaxZs/Dkk09izpw5OPvss731LS0tOO6447D77rtj0aJFuP7663HllVfijjvuqNzJiAnLq71m+JXQw2DbwtxHwfppQU+SdPqqlCRxGcBLyPblKUkpQbhtwm/918zsNvKZvvVnYPexwB5fkPTVb9NwydXIbwKXrgSO/llgCr+L3fv3Ei6XwjOAs3AHqmkvfIip/1qCdVu3Q4iaoJIEjyRZ0llzQlWt3iFJp6T+i2P117zFVz35TqAdGm4z8r7yFBlukyhJsjxR3nou+WSaCbcFr7dSPEm8kvSj7CXeDfdZ87Bg32JcxvwNmxKrYf3qcO+ko6DrmiTcBibcFnfYDw23mX5qBx403FaTMkJn5pmWzQxum9r8By6qJAlzBoXcn7QQ4/Zpo3eT7uciKk9SFAEb3FgTWNaRs5is4FGIE24La8pVw6gnTdd85SUMJqck8WVJIvc3Y9znvfXBPseFoWve95wk3Gbb4txItD/VkAKg4iTpxhtvxFlnnYVJkyZh5MiRmDZtGurr63H33XcLt7/55psxYcIEXHzxxdh///1x9dVX4/DDD8dtt90GwDmpN910Ey6//HKceOKJOPjgg3H//fdj1apVeOyxxwAA7777LmbOnIk777wTo0ePxtixY3Hrrbdi+vTpWLXKcdY+8MADyGazuPvuu3HAAQfg5JNPxvnnn48bb7yx0qckEma+8LSjx1eS3LQBzPJAniQ3kVywtAHABTRkKQD4NvlisPJeBpa4yovOeZI+bjgYOOJ/QvcFABxyMjDp30Ilxumb36ZGB4ma3oUOiElSck9SgdBIlCSvWRmJSAmK5zKeJJmSJOhnQUkCgF0136vRnjUD7VAl57dPvY9fFaaKR4U3XEXd0NhrRAN5AozhSaIkCaLZlJ4nKVxJEnmSbGIEvzp3Gl6ET4x+nvsJrst9n+1bETdjeknR/UVtOYOBb8CNO0jPfleesTibtxivC0UtQ5L00Nw3ectmwnrrSfJJRkkShtvkfZcpSV/YZwB+862D5DsW4M1ukyhJUUSjPiO+N+VMG9+dNg9n3vdq6P5/nf8J/jo/6AHlfx5hBIQmWaQDf00MtcZtVla7jce8Dzcy5ypMSeLBGLcTKklpww+rJqmNRxOPAuz15eeUStSViqCiJCmbzWLRokUYN26cf0Bdx7hx4zBv3jzhPvPmzWO2B4Dx48d72y9fvhxr1qxhtmlqasLo0aO9bebNm4c+ffrgiCOO8LYZN24cdF3HggULvG2++MUvIpPJMMd5//33sXnzZmHfOjs70dLSwvxVAi5J0owUk/hRBtuyhBmyZckkYbGMvQ+24WTjOb8eGyBXkgLV0mNexWFPnJlaZhBsyQyOvW9hA8lSQbiNwpAYR0XKThgiPEne4WQkSdQPGm4TZFk3ZVmoyefcYDd5r7dn8wKSRJQcaN6AECWZy43bpCCr4DvjlSwm47cg3OapqJoeyOBOIVaSiPkY7LT8reiFP5onsn0riiT5+1BPk2iMYcJtCZSkFZsk6mMBnXlxwkk6i68mpYeWVDEtmwnrubPTAAhrxfH7yuBMdQ+e17hlL7w8SZLrsS5ihpjM9/Pu6hYs+mQznn13rbQky7aOHH712NuJPzMPWtPMy/2ja/GUJC7c5vn9JKfvlL/Mx7Tn/SnEYZ4kHvQ8JFWS6jMpv85eAk8S/U3wfewxGbc3bNgA0zQxeDA76A0ePBhr1qwR7rNmzZrQ7d3/UdsMGjSIWZ9KpdCvXz9mG1Eb9Bg8rr32WjQ1NXl/w4YNE3/wEmGaBZKkp+QqAoENm1GH/OW86iOa3WbjT+mb8bv0nbg6fa+3nC+S6q/Qw98XASNdx/3yuR9GFFFMNwgXU5JoiGpziZSkXY5wptYngSTcxitJhib5HKIbgfvUKDFumwIfDgAg6w+qZ35pH6/p7VkzQFLcsjCAT6A/29IeahYG/BvhIG1LoMverJSIPEmAjTNSfioJoXHbC7dpoWQ8awdJJj1rOYizYDN9K8IgSpukr0WExCQhHg2xZo3HgmhaPsBeezXp8HBb3rKYdhglKSrcFkGSRP6WODXwAESqE7w5nYfsHNP2ZIkqw8b64sJtVBFCLE/S9IUrvH2d/Qrex5DvkvrJKGEJq89Hj+H0Odk9vS5teOpTNkG4jVesmYzb7mfuKcbtnQWXXnoptm7d6v19+umnFTmOlVhJsmGJwmOScFu/j57wlpmWjTHGO4JOxAu3leJJcmFkuHBTYECL+OGNuwJoPgj4xs3sXnR2m4jMiUjSD//JFo6NAy/cxn5XDZzcLyS8DQND29S4ArcupFnBcz5JOmyXBvSqcfrQmQ+G7TKMkuQc7/O/ew6X/ONNcdvuthZwmvEsjtbZ60aDFRpuo8uO0d/ARMPPYyPO8B4v3BalJOVgVKSaOKMkkZv5ocP6BPtj0QK3cR594qFDQmhpqClj6KEDq2U5hXBdrG9NEm6TfxJDhzAFQFwlyZ/dJgm3RcwQk6kaNAwl89+FzT4MGreTh9tkeawobinMPJPVbosCY4SO9CSRcFtCYlKXMYoKt/Hnfv22Tlz66FvozJuBz9ydqChJGjBgAAzDwNq1a5nla9euRXNzs3Cf5ubm0O3d/1Hb8MbwfD6PTZs2MduI2qDH4FFTU4PGxkbmrxIw8+6srwSz20Q3sIBx28HAZY8ALasByJ8w5LPbNJi03dhKUsgTZzqCJEU9djcOBX78EjDqR1w7hCSJno5EYa6kfiRyHF49iRVuu2BxaJuGMONPIdwmQr89yUY55omV75+bpwoQ5xWSIWda+E066Cl023euqXAlaU9tFbNOlHKChtvCUgBkhbPbSLjNNoq62T75s7Gh62U+pL0H9cJeA1l10ySzm5waoeWhSffP+1i4nDFup/XQge/dNS345T/9iTE0QzUTbhN0OWzwlStJMcNtEXmSooiGTDyhpEtWRiNMeAkmYpRv63qyLMv33+iahr0Hxp8cYhNyBYQ+LzDIM+G2eCRJ15KrqnVpwwu3JSFJorD+QwtXYMFHm7zfyk4fbstkMhg1ahRmz57tLbMsC7Nnz8aYMWOE+4wZM4bZHgBmzZrlbT98+HA0Nzcz27S0tGDBggXeNmPGjMGWLVuwaNEib5vnnnsOlmVh9OjR3jZz5sxBLpdjjrPffvuhb1+xEbirYBVM2LqRijW77Z6XPhJ6OnhPEvOufROAkJscDbfVNjGrmJlDcY3bIR+DJ0kBw2WRAwoNNxpxlaQY1eADcGXwCON2gCTV9wcy4lAhVadEuoOoFh4A4Pjr/ddmjnna5hWpIQ3+DSgJSZLNuHL7adriECH9/Lz6I1RCCyqqpumhYeeo2W05iBM8RuHAXZpC11MRlW9/cCN7TZs2UZIQ35MUhT8SDwpFfcC4Lb/VX/n4EuY9Db11RChJYWOiLjFuxyZJerjPpTYiZCXbL0vulbJBPSyMmKTSvftZaXu6Bhy0a/i1xRyPmw5fjJIUFW6jRIz/flw1Woa6jEEIbek5rdZv6+xZ4bYpU6bgL3/5C+677z68++67+MlPfoK2tjZMmjQJAHD66afj0ksv9ba/4IILMHPmTNxwww147733cOWVV+LVV1/FeeedB8C5UC688EJcc801ePzxx/HWW2/h9NNPx9ChQzFx4kQAwP77748JEybgrLPOwsKFC/Hyyy/jvPPOw8knn4yhQ4cCAE499VRkMhlMnjwZS5YswcMPP4ybb74ZU6ZMqfQpiYRdKFehG2lhqIXHf99fh/kfrg+2E6j7RgYizTcUisBkh/7Rv5l1dADSynAR8yRpz4G92Q1SklloEWA9SXFJUngWYCGIfyhD1BleSQrMbqvhPiezsbOvxhW4dSElSY1DgRFfd17n2tEr5X+P/De1X3//5heSsjKA9qwFS1BGxO2naYlDhJTo5HliExZugx7qKYoTbgut/WYXlxCVUZK4y4sPBdm27TEjPYlzu0jQkh9RKQD4mmz0/fas/7AkuleEEQZDknE7fritMLtNoiTRGXyiGZl8TToX7Vl/W9kkhbDPxe8i2zRF0kHkGJKkYa8ESpIfpmP/R4Get7iz20QkaeaFX8D5x+4j3bcu7YfbZHm99h3cC0fuEU982NjW2XPCbQDw/e9/H7///e8xdepUHHrooXjjjTcwc+ZMzyS9YsUKrF692tv+6KOPxoMPPog77rgDhxxyCP7+97/jsccew4EHHuhtc8kll+BnP/sZzj77bBx55JFobW3FzJkzUVvrD7YPPPAARowYgWOPPRbHH388xo4dy+RAampqwjPPPIPly5dj1KhR+PnPf46pU6cyuZS6CzZRkhoy0V+RBhuzlqwSrOCVJHoBu4ZCmZLk3Cjf7P81x+9DV9EBtZjwFAeeJAVmou13PDD8i8AXfpGsYSYFQES4re9w4KJ34iXMkRyn9/L/YEnN/+BYfRF0LWgsDShJmRCSFJFM0pIZtwH/c/3nYvyz9Yde+Q+epBmfhSd1laE9ZyIvuHVoDEkSdItRkrhs8AKSRPMkJVWSmHAbDObGf/5X9o7cP9AX0URCyew2IOjFMS0b2wvkI0nG7WJBSVLKCA58FPwtgJq1t0cYt8MUCl0Tm4DDEltSpCJCOLXkHG8X1ZWT3NsoCZSF8sJJEh9uE2+r65qnhFASp+vO9/HnH47C7v3DE9dalo0HFzqzTuMYt2X9jEh9RpI3BpWqAb1q8KV9Jd5JuCTJ2UdGkn78pb3wjUOGxuk2NrZmqyrc1iUFbs877zxPCeLx/PPPB5addNJJOOmkk6TtaZqGq666CldddZV0m379+uHBBx8M7dfBBx+MF198MXSb7oAXbktlcNDgeiDCH67DRmdWkCcJYUqSKwNLyo+4niRBmIr6YbQyeJKCRmnuh2GkgTOeQGJQT1JKcKlTJWngfkDTLsmPwR0nrZn4Q/pPOMK8OxBqCChCgnIkfJu6JBAmVZIARg2rRzuGaBvxqT2YPX5dX6B9s7yNELTnzIKayOdJctrPW2JiR0kaX0pEaNwm2RfDbpVZQVkSfnYbvdleMG5ffHnEIKBgq8qHEc4CROoHXcQrXRkuFDTz7TUwLRu6BjTUpHDJ+BF46i3xLNpyoKnevwbypp3IjEvN4JHhtjBPki5WkqLCNy6i8iTR3E/bO000crXgZOoJS5KSh9veXLkFtz63FOceszeOP2iIdFtD07xrJM+F2wBg/AHN2NiaxWX/fEu4f9rQ8K/Fn3nfh7tf3LQrNNwY5oGzSV1BkZLkECf5ceozciXpvGP2xndG7YrhAxqEOadE2NCaxZCmWq8/3Q01u60K4SlJqRQEaUYC0GDDFM5uY3emT/LuoFRjs/W8vF1tN+t38BKxiyFJYXF8PplimX4YFpNxWxBGoyRJlNAxLjKsdL7FbkBaD5pWA+G2/b4mbu/0fxHjtjh0FUqSOEO6S44YkjRwf/n+EWjPmhDVRnPbz5tiHxX9/DwxiVSSQi6z6HAbmwLA0DUcvltfsj6aJIme3vk2KfhcMx9tcDKaf2nfgRjQqwZ7DGjA9LM/F3ncYkH9Op15M5EZl4agSlGSnNptweNGFbZ1kY5QkujzHR8yBOTkqp08UGZlxu0Ql8Mfn/8QS1a14PfPvB96HEP3CQcNfdGBX0Ref3HcvgAclevqJ9/1lict0SEq8yFCR87ykzfqQZKkR6TgqCUkiT+fdRkDwwc4vsu4M0w3tnX6OaUUSVIQwc0ZYxjp6BxBcEhSXmCm5aeJ00Hq043bYFk2UwiVadPNpSQYnVhPUhkuoQqRJMa4LTJ5UjKRTlqvjYBLPrkBjikzNNw24uvAkWf67wfs5/zv1Qzs+WXvHGgyJSlsYOfM5z45Ksw+gw5k2M+bZFJ6R0S4LW+JfVR0WSAgJ8jz5ZFDTYcgn6YHUcZtJpmkbQgTPLqIE25LCa5zNuM2u05WMmPU7j45q2TxTvoT6sxbRdXjAljyIfYkyffVdQ1pwXnrHTfc5tZui+EbEpEdqZJEPElx2pbB9RzK1HhD98kFVXXo9y4yJlMSSUvEuIQh7sxIJuN2yD5t2TyT6JQnM1QRE6GeCbexv2PqzYt7uW9szfYsT5JCESiQpFRaUNNMAA028nFmt5GBKJ/LwrRtNGlikqQXZreJlCRm+nkZkklGhtuKBB2IddGsNaokpeuKPxBnwF5v94GNkBQAx/wSOPkB9pinPQIcfrofVvTKkohDV6LvxT8Qa0h3ybFLUmxogVl1Sc64H25joTPhNkG3SB4pgwvV2aLByn1A0HVh1nEXUZ6kqNltfOhPBBGhCStFIiMllDiXK6GkCLQ/nTmraEJG/Uki4hCVAkBMAuKG24IqDAXlQIlIUgxPUlReIcDP+B2qJGnB47CJR4P7yc6PeyrjZvyOO7stZ1qhxm1NQ2jG+7qM4T0U8Iloabg17ky1zduz3m+jkg8ScaFIUhWCUZI+fyGQCh/ANQCmqIo6d2Ez4R4rD9Oy0UeiJG3e1u7tFehfUcbtkB+2wZGkcoXbGCUpgiSVEm7jSNJGuxG2HVK7TfT5+u4BfPNWYOC+hW1843ai2W1AINymceE2G1ogS3lUcVqK9qwpzNPkGbdNMbHTYCOFPPbXPmFKkgCALcjwXq6yJDkY4SSpSE8SowjEJEn0ybpcuZKikDWtQP+/MmIQTjhoSOS+kUpShHFbhMRKkuQY9PxlBfc/GXnpiOFJipoy7xzf+S+bRadr/uw2xrjNhGlFSps4HOmSjCgTtgs2T1LIdqZNjNIsmdE0J8wXdkuupRm3ufNJFdW4obOcSZLSqnCbgggaVZJ6DQSmBDNid8Af4GVKEq/y0Kd3zc7Dsm00aa3CPnhP/aJwG501FldJ6rO7fF3FlCSaAkBEksjNSLQ+LjgDdh4GTMsO1JbyyE4YwXHBGLdFSlL8cJtLSNx2bE0vg5Ik8OgU2ndKPZA+j/8tAEdJuy19K/5Tcykmp/7D7iwk+aQsSUh/xMZt/7vPwwidtBgr3Cbw1rCKAEeSJGbCmohSGuUCHeg6c0FP0m796mMZqKlC40aMXv14Ex57/TNnWWjGbfE5iOtJShlBgkFBjy0qpSMjOjStgTwFQHT/3OPLlST/e6BEj6nzJyABMhLpfp7Y4Tby2aLyPtFs4CkB+ee7ScmPY9wuhNs4b1imiHCbU9PN9UjF26eSqIIuKARQMLGmUoWbSW2fwCaLdvkh1tvO4KzDhpUXzG7jBlKD+j5ME6YlD7e5hCrKuC0sHCvCwd8DvnixeF3Ak1Sey9Iiv2wtKtxWTBJJF5ySlIKJrGkFw20u8YyjvkWQJFG4yz8QG25LFb5LXaPhttI8SaLacWyepMK2pzzqpG8orJ9gvAIAGKmzM11EN34/kaqRONzGG7dFT7FvWXsAAGaYX5K2fcU3RkLTgBu/d2hgHW0zbBChiCqlUS7QAakzbwU8VWlDi6XoUJLkhqC+O20eLnz4Dby5ckt4MkmJChBXScpE5Emi434n1xHbtuXGbUKo3Izbm9uy+GDtNm95nJCWu4ls25Su+yEy0r+ocJuMvLqlY+Ims4ybcZuSEp3kdgJ8Esc/ptDLKSxPEiVJcUNnOdMOZBnvTiiSVI1wlSSXJImIiGZ4ZEU2uy2Ycdu/gJvnXgEr2yk1bqcQoiTRyyYuSdIN4CuXi9fxSlIFwm3CflIyEUfdkYEzbrveI9647YfbkpAkcegqlEhy4bZ0IU+Sa4S2NT0QbotLkizLjgy35U3b+6yapnt95QviUgjDbTGVpGLDbadmL8dp2UtxjyWZZQhg0ueH4/2rv4bP7z0gsE5Wuw2I6UmSHjUeLpmwH0YOEaeR0KD5BO/7hwaUsLShx8pXxJh/OTLw8cbtoYOvnCTFVZKKN26HcZwOgSfpiN88i+P+MMcjSnGISJSSpOv+dZH3wkf8rMjgdVKbFl87biqAYjxJYfswJEljr2W3q3w3aZqI2jCSRH4HsVMXqHCbQhQG1jtfS11tSKFVUiRTkw6kck9S3eb3kF50h1RJctUH0UVKByCtFHLhHYz3A1WAJIn6SRWdUpQkThlKac65kxq34yhJerhx2wojSZJwG+NtKjLcZtk22nOmMEO3b9z2rxBN0+KRwpCyJI4nSQ7R7Da2dltKKPVvQz1etg6CHvF9iHL9APzsNvYAsn3Kadz+6Zf3RmOd3OQ76fPD8e5VE/ClfQcGPEkpQ2cUC0PX8I+fHB16PJ44mJYVEW5z/vMFf+tjhhy9ZJKS2WMykrS1PYe/heTkoWZ0l4C5g/L8jzYG2pbBVbiks9uIJ8klY4GZY4LLpEZSbsUld2HpCZj+xfQkmSF5knSJksT2VxemegDY8LJMSOKvzRwNt1UBSeqSZJIKybD/4HrgI6Bfb/m0dFszvIFAgyBRISDwJLG/rrXL35YqSX64TVA8VDO8x2CtDBm3A2VHyhRuY5JpikgQ/QGW43MU4J47uXE7xudjUgCIlUR5B7hwm5YHbOJJEqQAiKtrmAWSlLeNALNyPx+TTFLXY31eYZ4kj1Tq0DRxvTgg3uy2sDGv2Bk09AGiKON2GbJui1IT0L65pCygdOkak3yxf0MmMhTIqxGmFa5QuH6cf/zkaLR25vGTvy2CoWvoUx8zT5IeEW4jtzOqYFz08Bt47r11gj0cUDM670lyz2cctcad1i8LORp60LjNXycitU32PXhKUkx2TclbGOnLm7Z0dlucUihUSeJBlSQZ4UkZGkPo8qYVKMXSnVBKUjXCHTBC1A1N0wlJssSzk0LLkgAfrd2KPjLjNg3N8N0rKuN2CCqVTDLJLLwykqQa5HBT+jbUvvJHZvn+gwtJJ2ORJDq7LfjdhqcA4MNtrJJka1ogAaaQZAtg264nSX78a558xyNJGrR451aQD0yLqSS120HFlapvORheMkcRZEQjCuF5ksQ9pkrSLn1KSDtRgIzg8T8h/jOmuHBb/141iUmSRZ74RaBlNJrq0njwrM/hr5NHxw6h+GVJZJ4ksZIURpAAsZLEH5P/XDd9/9BAWY1IJUn38wvlSLiNOZ7g2pMqSQVPUtxwW9zabZZtM2VJDAH55/t96ujdSH91OUmiBbYl3zvv37NsQiqrgCUpklSNcEuCEJL04Z4/ZDaxdb9ghUxJCha4ZX/MW7Z3hHiS5Mkk2RQAFSBJZYIdFW6jKCXcBjC+pK8Zr2CiMRf6rF95y47YvS/2GlAYFBMYt5tqDSFB4EvOMAiE21xPEk0BUJxx27IdT5I43OZcX6+t2OKTJF2LR3pN05EGVswHss41adj+7yCshVYErx96rUeVHRHNXIsDSlDie5L85bv3b8AdPxyFfQfHL3Ya1gcKXqEwAp4k1rg9oFdGGiJ0EQi3kcFV2LcSH3a8PEkSEkK7kxUk05WB8STlLWbSQNojSew+Dnlgl7nqh9STpNGM285nCGSzFoXbJJ4kdwZf3NltTJ6kMCXJYme3UWLiXjeU4Bx/UDN26+ffP2rSuvQ6rIlh3M6kDNx26mG46sQDvGU5ifLWHVAkqRohIEl7nXoDNmR29d6zSpJYbYhSLFKw0EtrF64zPOOtwJNECEdZlCRO+ShXuM2k7USRuVJJ0k/mAvtOkK7OpHRfIUxg3D5y96aSk0mm+BQAEKUAiHfjff799WjLmoFkkE4bwfYc43YcJckEXrkTuHs88LfvAoCX/NTS06E8qx1BJYn2L2qKf5K6ZhTFJJPkVYLjDmhmSqQkhZwkse8DniRdYzxJ/RuiSVIw3GaHel1KVQHSZHbbZ1vaMfH2l/H4Yr+QN6MkhU2z47A9yxq3aajOPSb/WUWJMfOcn4kHm3HbD2fx7fKQzYzs9Ga3CVcHwHiSQnZqz5qY9+EGr8/0WvFnt7F9pgS4JmWEEKDoFAC6Bnz94KE4adQwb1nWI0nSbncZFEmqRghIElI1+GjAMd5bOkjKEg5Gfb0GLF8x4uCaj6OUpCjTaywE6qqV55fBqB1RA/XgA0s7WJ9hwP7flK5OG7pfeiOWkuRso9lF5EmKCLdB04o2bv/0gdewcPkmpAUeIXoN+sbtmJ4k2wReLVScXTHXac/L+i5W03wE16bIQ0NUssjiPUn+a34A3aWvOJQmm7nkYkCvkMkaAsgIHm+0FRm33ZpagDPQyQZnF/w4a9l2Uckk44LObpv62Nt449MtOP+h15njuxDlSZKB9yTRfd3wF/+5eGIA+CE0aQFe3d/HVUbiCCMycukqYEXNbgtRki6esRi/f+YDr38ihZSvU0j7WJPSpd91nGSS7jGootuZU0qSQhg8ksQVAaUDo27Asp0L6K+ZazHWeDvYTsTgpIeQpFBPEk0mWY5wG6/iVMSTJFETzn4B+PadwPAvlH5AAfnpje0ACk9UxGMTCc1P75BYSeJIZ1orFExmwm3sIJ4kTxIA4XWjk7IjvpIUz5OkWVagfpsfbksnvia22j4BCA1NohRPUtDg6mL08H64ZMJ++PJ+A5nlIr8JHb/mXPJlvP6rr8YmGFJPkh69XZ/6DP542uHYd3AvfP+oYbGUJBrqMS07dPAttaREWvcJxgZSw8wFU5YkgZLUzmXc7hCE6ngeYujBATtKSdJ1zROwXZJUbA09AF4aimLyJIWdnlVbO7zXNEs44JNr+lXytdxq04bcbxSjLIl7PErks8qTpBAKiXHbJu81TfOUkt5aOy5IPRpsJ2JwOt5YiF4Qh9vCPUlkGRfaKQp8PwcfIN4uIdg8SZJzMfRQ4OCTynI8ERF7q/ZM9MJ254mqiHCbkW/Dd4050vVCcNnDazRBxm3ueyuaJB11Djr67hdYrycMt9m25T8cuH0qkCZbNxL3715zfOxtix3MRUn3XGiahp9+eW987cBmZrlISaKz3OrSBvo2ZHDa6N0BAMfsNxC/+/ZB0j7IlSTuvabhnz/1p/i7qsTxBw3BMxd9CYfv1jcWSaJ8IDLcVuLDjqskWba4NhslC7c+twwtHblY7dK2cpyS5LbJkz9NFG6LUJJoCgD3GLKp8lGYcEAzfvLlvZg+RoFm3I7rY3LyJJH3rpJErihDZ2fAhSlJcQrcukRSIx4uNwRaBRxJkaSqhCjcBjCDja2lhOZZdvvor3d3XTwTxAgjSVRJ4v1ECWFC91WCs/7rlLA4+OSS2nQRmSep3JAQsQO0T/DdI3ZNGG4rGLdXPIuUFhwgwsNtLAGq0dkCt4Amre8WF5mCGRxHnYXadKrQPs0qTEJ7cZQzywzkSvIyxOtppn835r4b2dwsaxQuzp2NH2WdLO8TDmiWbls8SfJfy56keaIgm7nEt/PLE/bHnacfgdtOPdwjCyKIkhGKjgsAhxHvE5/0D/Cn3MtgcUbt6HBbqSSJhl9Eag977DvnfJT4GJvacnjuvbXee/fzCcNt3OmJmt1Gs1d35EtTkn44Znfv2okbbsvFLEtC4XxOv48pL9zmb2PoGnN+alLy2oisJyn6N5LiSJIqcKsghll4IuJJEnlvpuqF9bMYlGCAToUkPmTIRyk1zwC2vMUuhwNjzi3PjDlwpvIyTvGXQhLSu+7bI3HMfoOIQhiHJJXw3fLhNs+4XQihQgtsU7SSZKS9vjyQuRanG09z/YwXbhORJL1AxGw9xTzJ3mJ+G3k76hrRMMP8Mp63DsMvjtsXt592uHTLcuRJko19dKBPG8EK6zLUpg2MGzkYDTWpUPVBdtwogtIhIB26roUei2Zmdt6HD76l1t2ipK1dRJI4brKlPZ6SRPHQwhW48gm/NqZHkmyeJAXVQnfWnSyPU4ooLi7Ji/J9uaB+MYBVIOMmIe0UlJSJguO9Iu8FJEnXNSYtQ01aL4kk0fPqkki3XIzKuK0ghjTc5g82llEbObDZJVxgoUoSE8YqTUkKrWZfIg4a1s9/U6YZc6GQkKTd+xaUncLUdr4kiLit8PMSriSx/dBtZ/Dw8yQZgnBbMqQg9gtdlb6v0B5RreLcoG1T4Elyw22pwMQEkbomw9A+daHkpNh6akz5BskZ7FXj/z5qJSpS1OkJ67tMSYr66dNcQRSyPgJuIVRb+j7YhxI9SWS03i7oL3/svvWlh/5dksSTP9sOfh6nxpg8DYIzu815nVRJ+vuPxzDvqQIZl/CwxYltr09hoKVUAKoksaFlmgU9Y+jS6401bsuOSY5X+M7V7DaFcEiM2/S9adSiXusMbycBAbG4GUD76p+5jQSbJcVGNT5bdkKICqWWC/VkgCp5in8cyI6RL5hOs4XEnZkYJCmK1CVIAWDYeTSiDb9K/xVAgTyXFG6zkdGokiS4RuhMunSMpIm2QEmyycNCCTdL2cD0m28diIG9a/D7kw4pqt04N3CaiyhuYVceYcZyqSdJ0re+hWzXX95vkHA9nyWewrKD9cBCy5KUSJLoYC1SkvhD92sIvxfR3D4yyJSkzrwlJBiWHZEnyfMkOf1Pp+Kdk/69apiZjlSRjBs6o0qSu0uUJyqQJ6lw7THGbV1j1DO+KK6LxVOPC8yKEx/Tf+2H25zzVQ1lSRRJqka4JCmQP4gYtzP14txIBCkjPgHJG5KpxwKixUz1LlFJqiRJAjcbsOKQHcNVkNz/NTGSB0aQpCSz2wzbxK9Sf8WR+gfuBgKSFB/MzDYSbmO6QJWkTD1wjKS4cQG2HT67LbzEbThkA8Npo3fHwsuOxf6SIrFRiKOU0NIfTRKlI2rIC8vjFDeZpIvnfv5lPPrTozF2n2DBXiBYb5BCZNwOD7eVNsBpmh/+izJuA9Gk9XtH7Bq+AXyVhv9YHTlT+Hlyprx+HS1LEldJOuUoP5O1iDwApYXboo6vaVyeJNdUzRm3+Uzl/KkZ0CuDJq78TBzfnvtAkPWM291PklTttmrEj54E8h1Avz3Z5WQQNmrqAhm0eaRT8b9eU88A5vbActGFTcmZnkpIkjQyFR6VJkldbdyWnO+sU1kcna6SVDpJCv08nLqX1vI4RP/Qe29petCTpNmxy9IzJElPQ6g2UiUJcPxmIdAsMzC7TYcfbuOVrresPXCQ/jEWWftE9jdsYCglJJRUSWqSFKONGvTCMoJLUwBItu/bkEHfEMWlNqT4rGWxRm3LtiGx4zh9KMP4ltJ15ExxaJAnJ7LyJS50XcPIIY14Z3WLdJu8JNzWkTeFA3besqWeJJpPyFOSuGuReo/e/vV4NBCSaggUHVHfZBCF26I8UbrGzdqUKEl8ygX+3AjHDVm4jZIkjhRXAUdSJKkqMTA4pRoAS5IyDZEkqSadhCRJSoMI1BHmuElnt+lpwPTDhGYlvUKM27AbSVJnqzMSZstHkuywz8OF29IwvdIkXtslhNvSMZQk78y766LOv22xTlzL9D1Uegpb+rBpIc7M/gLfN57HQ+ZXovtbqoNYAnpzl93MG+v881yfKe52G9Z/mcpU7BN4qJJks3mR8lzeJB7lmJmUMjRA4sfmuYKsfIkLDVpkMk+zMPjzBKw9awqN6PkwJYnkE3JVHZ6kDG6sxePnfR69a9NMBnRAPOsLYD1JT1/4RYy/SZAiBH7ICvCJVdRvgS9w6/Fzjd2GV5L4y0301cuuB7rc7V81zW5TJGlHAhmEU7UNkUVJMwlIkiXLdyR6IiBKkJ50dpvBkSS7guSF3ry60biNbCuQa4cn1cTyJEWdl5CbBxc6TSHPZMi2oQdISxKSlKKES5eRJBJuA6I9YbySZJKRUU+htWFPnNh5FTbYTQCAteiHW8xvx+pvV5AkGejAVyxxCRsoZMn2in0CDyNyvAcpb1oVrd0GhH93SZUkTYtOwWDaToHmO19aziyvz6SwVTB7Lh8ScqT5hHwlKXhODt61j3R/0Wuq5g1ulGdoz4o8SRGeKIMjSSlPSWJDcNuzPElyTOruVyK61qUpAARG8WoKtylP0o4EMrA5JCki3JaJXzjWlHqSwgbAIpg+N1jmK3oJkptXlxi3JTfgzlZfRRIUlxUiitSFKTOpYO22DCE2ovxaxShJlmY402GijNtAJOnT+NltlDDpKWgasNjeG59hYGDfw3brgwOGNuLsL/rhafrknYlplk0K+rFl93J2oBNvY0ec+zCzrdy4XdxnDg232TYj9rV2mliySh66Ksf07TA/lkuSRjT3BhAdhtIQXRbGtKwAQTrh4CH4zqhdxOE2U25ep56kdkm4LQyiWWYASw7DMlLTkJgV25PEbuOVJaH90jQhIQ2rZei2LQLjvXLr9RW+yyrgSIok7UigpChdGyPcVscOxm22/KnDkoXbhMZtfzCLm/fDAxfmqagniaI7w23ZbezMtjh5oCLuDkmSSWaQZ8JtItO36GjNjbX4+Vf3hQYLx+ivYyC2OO1pfv4iZ+filSQv55ZtsbPbLFZJCnuiHLNnf/z7/C/gyD38lA9sfqLK3OaSPiBIP0OUJynkehHN2ColQhFl3KahnocWrghtqxyhEtF3d+eLH2FzW9YjbG4Kh3xEaRJd00JJIBAs3/GVEYNw+6mHS4u4OsZtcVuGrqG5ybmvbt7uXM/pBOkm+ASOLppICFes2Dj/aSZxM7YniTduF0gSJUC6hl0FtQnZ5KrBtqXhNrIxn8RThdsUEiFFnq7TtdGz2zK17A10YvZqzKq5RLhtq5VCf8FyUW02Ss5qIp7MAhiwH9C23ntbUSWJCbd1AUmSHaOzlZi2Y4TagLKmAEhxniRLcM5FStJvv30g+tRnsHz2S7gxMw0tdh0O7rwLNXBSGlhGbYHihilJ4Z4kU09DtzodksQoSf5rTU+HckZ3IKX307Sho8MrBdF94bZStncRNlB8+/Bd8dGGNuwzqBemPLIYQGkKThhJsuzwavI8ynHaRSraNf9+F3OWbvD8UG4ILRfhkXLCbdFKEgVvWOYRFm7TNQ1H7N6XWZbooZI0S4nyn34wCr+YsRi/OG4/YUizoSaFbR35opQkXRcrSQwBgobJY4dj/bZOHDdysL+vpnmdThRuI8v5VA8q3KaQCAZxMNbW1EYqSbWckrTU3hUvmeK6aE29e4sbiQi3RcX4A/j2n4Eaf8p1vqJKEg23daOStG010L7ZeR3HtA2UlyRpJjLk2hElGeVJ0i9Tf8Mhc85CGja+YrwOAGjUnDp/NYW2LDdEG2rcdpUkGUly+qpx0/9dT1Le1iOlEXcgpaEHOhhVjiQl2/6IPfoKl0fRjrD+G7qG/zdhBI4d4Q9WcWc/iRCWJwmQ5wQSoSzhNslnn/PBek/ByRAlKap/IiVp1751OHW0M/WeV5KiQkimJfdlGTpw4C5NjDKTpHYbbdUg++0/pBH/Pv8LOGbEIOHDg+uDYz1JLkkKP76uacz15iWTBD0Pznm88psH4Oi9BzD70m1EbQuPSb5iRZIUSkLK8ith16YNGFr4DSFdE1QtZGbvPo2SXDGiAZAYt6OKYgbQtCtw0j3+cRtiJBosFl2tJMlI0op5wP3fdF7HVZKiSF2oJ4kNq2aQ95M/gsuYXsD9mf/DxanpAAANFs5KPYX+q15Ar81vBbb1SFKB4AjN/R6Bjwi3aYXQAU+SCqppHkagMjkPX0kiHg5DTJjKCZYEyPv3zEVfxBXfGIkffm73oo4TJ+RgFFk4lUddRDiKn9UUhnIYt+N4ktx7UM60hfmUXGiScFuvmpR3nKCSFD7w50I9STpq0wYaiHk/CWGnqpjsPIiuDVcNFM1ui7pfp3SN+e14ZUlokYUYSSHDwoA86GfoyPIkKbS7XQJFknYguOUlAKA2DjlJBT1IuoxYCbYFAF0QbqNP/UWVdCC/uIFNMUlDUaBKUlfMbotBxMqmJIUciyMkTF4jiEkSAJybehwA0B/bvGWGgGjXaLySJFKmwK6T9Ndy8zXxA03Bk+SQpHCLlizc5r2ukHE7LvYd3BuTPj9cqoocvZco0O0jjvoQRiaSICzcBoiTOspQDj9J2KDOqyOmFU6SdEm4LW3oXl95JUo2w8xFuHHb+U/N4olIkqQfFCIy0iBSkgovMxHKf1NdmrnefCXJh9yArYVuEyfctr0KlSTlSdqRQKZFRxkQAQjLQUhnMaXEio4oISX1QhV1cyYEQOuKWWddhTifJU6JDiBGxu2Q75+7sTBT9lFIJhmCwdpm77Uh2NZVkuyQcBsCniTxuZGG2wqepDwMaBor9/Pwwm2SvDKVCrdRlHIv/87hu6KhJoVDhvURrqfk6vxj98FLS9fjoq/uy2xTLoNrXUQup2ThtlJ7w5qUefjhNue3kLesQJJDpj8AagT3TV3zVS++LhpVTUQDds6yAiE6F26b9F6dRHmn5Eumyom+9oaMPNzWuyb8++1Tn2H8T+7nZwiQ5LfIpKUT9Jdeo7Vp3zNIt+VDl8q4rZAI7YYfEotFklLsjLVpPzgctf+QZFaWKEkZUdZuEm4rzndAf00VvATj5u8vF+J8Flk+Kh5RU+YTKGMZjiSJUgBQDCIkqfnxU/B1Yw2z3jduu9dMiJLkQtJfV0miIVwAvifJC7fJ++uSINml2BUkqRTouobjDxoiXU8J3/7NvTGFI0j8NqUgKtx223NLY7dVjnDbwN4hM3LdcJtbOT5GuC0tOE+apnnhSpOb2h5l3A6rX+cSDHpOE3mSGCFcRkxESpIbbguWJYmqH9i3PsOE2/wCt6QvMiUpglDSRXVpg5AkeX+qQEiqbLht06ZNOO2009DY2Ig+ffpg8uTJaG1tDd2no6MD5557Lvr3749evXrhO9/5DtauXctss2LFCpxwwgmor6/HoEGDcPHFFyOf9weCRx99FF/96lcxcOBANDY2YsyYMXj66aeZNq688spCAiz/b8SIEeX78BXAO0O/g8fNMTg/e248hs2RpAkHDsEhu0i8RylxCoC0ICFlsmKoAjAB7i4ybncFYpGkmBnKo0hQAo9VINyWRElqWxNYH6UkabCCeZJkniSpkuR7ktxEdTJEkaBKeZIoKnkvpwRIdh7KYZIG/AFWhmffXRe7rbAcPnERSpIKqoM7wzZvWmjtZB8IDO7ciZSwMCWJEj3R58mFJNR0v7cahiQl8STF3pRBvUhJsuKRpKa6lCRPUjgB4pdHhdvoQ37YWFYNDzgV7cFpp52GJUuWYNasWXjyyScxZ84cnH322aH7XHTRRXjiiScwY8YMvPDCC1i1ahW+/W0/s65pmjjhhBOQzWYxd+5c3Hfffbj33nsxdepUb5s5c+bgq1/9Kp566iksWrQIxxxzDL7xjW/g9ddfZ451wAEHYPXq1d7fSy+9VN4TUGa0I4Pzcz/D49bn4+2QjpdM0rK1QAJCrwlRuI0f0JKCeSzZmZSkOOqe/KbPthVxXhIoSSmN9ySFD16DsTl0fdCTxPbFgMUWuAVCPEnOdWdbPEnylaSgK4KFqLYZnXWZ5Om9GkHDbZW+pAc3xk9AG4Vy+EkG9Zb3xz0XLgnOW3aAJFGCrAnKaQDOIO0bt8PCbcE+tLTnpWkR3H2pf7RY43YSuJ6kTlOkJIU/pDUQEzvgk8Q4iVOjjNuUDFF1Lew6KZdCWgoqRpLeffddzJw5E3feeSdGjx6NsWPH4tZbb8X06dOxatUq4T5bt27FXXfdhRtvvBFf+cpXMGrUKNxzzz2YO3cu5s+fDwB45pln8M477+Bvf/sbDj30UHzta1/D1Vdfjdtvvx3ZrBMGuOmmm3DJJZfgyCOPxD777IPf/va32GefffDEE08wx0ulUmhubvb+BgwQV8auFkSl3Q9ApA59/Q9AbR/g8xd4i0zo4Aueuking8uj8jNFglGSKhnx7WYl6dgrgtuUKdyWREkKhNsi9m3UgoWOKYIpANgbmcEoSeGeJNtwSRIfbisoSbYTbgsbb0VK0SHDmvz+VMGNthSw2ZYre6yhTclnm579xT2Fg1k5znuscJuXAsDGtg62dAj1AGkQz87TNL8QbYAk0XCb4CJ8d3VLaO02gPMkFWncToJeBTUwm7c8ouV+bL4+HI+GGlZJch9AovxGzjZUdROt91/XxiVJVfCAUzGSNG/ePPTp0wdHHHGEt2zcuHHQdR0LFiwQ7rNo0SLkcjmMGzfOWzZixAjstttumDdvntfuQQcdhMGD/bwg48ePR0tLC5YsWSJs17IsbNu2Df369WOWL126FEOHDsWee+6J0047DStWhGeQ7ezsREtLC/PXlehMMLMEgJgkNR8EXLIc+PyF3iJNg3QQS9cIMqvy/pGkoGrKTqUkkc+SrgfGXhTcJna4rXxKUnMvlhSFKUkarEgSHBVuMxKE29zzEVSSHJJkQoemRd1I3azd/rKvjBjkvS5XKKq7QAcKPhxUbgxuiql0EjTWpnDp8fsHlpeDmw4KJUnOf19JsrCtg1OSKEnSxA+augapkkSJnijctmRVi/Q7cfels9uSDPqlhtsAvzSJq3aFGeGdfQ2mj34yyeiQLz09IjJmMOG2YEgPCM70DMs231WoWA/WrFmDQYMGMctSqRT69euHNWuCPgd3n0wmgz59+jDLBw8e7O2zZs0ahiC56911Ivz+979Ha2srvve973nLRo8ejXvvvRczZ87En/70Jyxfvhxf+MIXsG3bNmEbAHDttdeiqanJ+xs2bJh020ogSY4SAPKZVLrOEBUDFiApVMtn7QbKoCTR2mVdkeSxq8AQAU2Smz9uuK18StLgXux3K0sBAAApWAEPEw+PJHkknP2cKZhkiUuSJMcsXIe2xQ5ubrgth1TkYCsKp31p30H4yZf3wpXfGBm+c5lQSR5GB4piQzBxkTg5LBzyMHnscLx4yTFlKehL0a9BrryagjxJAZJEw23QsNegYAoOXdNg6H7Ijl/ngg7yLnlbum4beBHUhZtuoLZYT1KRWhL1HX39lpdw3B9ewPtrnXGtMYIkDR/QwHqS3HAb2SaOJ0kU1qPraT9oc386bRSzTzWowIlJ0v/+7/8GDM/833vvvVeJvhaFBx98EL/+9a/xyCOPMKTta1/7Gk466SQcfPDBGD9+PJ566ils2bIFjzzyiLStSy+9FFu3bvX+Pv300674CB5OGrUrgOi8Kh7CBmSyToMtV5IqQpIIedtZw22yASK2khQcrGg5ES0BudRMNgQhyrjtHRZWdE1AzQlry5UkE5rbRoSSpBkyTxJVkrRQciAKYeg68P8mjMCPPj889LPsCGAqwHehOiqqCyeCO/gN61cfORssKcLyNtl8uM0KGrdpXiRdA04+chguGrcvfnT0HmS55uU04v1FTDJJcpn1qXd+x505i1GSaLsu4axNFRluK/KrrssYXv+WrmvFB2v9yVKNEuN2/4YMpv3gcOzevwFpSsoL//nabSKwJCn8vj6glz/+UELVVJ/2ChYD1WHcTjxC/fznP8ePfvSj0G323HNPNDc3Y906diZEPp/Hpk2b0NzcLNyvubkZ2WwWW7ZsYdSktWvXevs0Nzdj4cKFzH7u7De+3enTp+PMM8/EjBkzmBCeCH369MG+++6LZcuWSbepqalBTU1yObpc2HNgLyyeelzkBeghTKrklSOJJymVCd4otVJJEs3JVEklqcvDbfR8y0hSTE+SgFTYmu6lX7CTlHMxO5m3otptLlIw44fbUmJPUgpWUEmSKF+aO2FA4knKecZtOdxwG5N8r4tDbGF5nMoJmWpRTlz33YPx8Cuf4tSjdsPPZyyO3F42A6wcSlJDSN6mQLjNtNEaEW5LGzouGLcPnli8CvfO/dhb7ipJubBwG/k8NSQ3kxuiW3jZsXjj0y1eu+6sOyaZZILEpsXevdK6jmF967Fl+9bAOlGeKAA4eu8BmHCgk4aChtvcWyhbu00M+nWL8jFRlY6SpH4N7NgTlcCzq5GYpg0cOBAjRowI/ctkMhgzZgy2bNmCRYsWefs+99xzsCwLo0ePFrY9atQopNNpzJ4921v2/vvvY8WKFRgzZgwAYMyYMXjrrbcYAjZr1iw0NjZi5EhfWn/ooYcwadIkPPTQQzjhhBMiP1drays+/PBDDBkiz1dSDWiqT5dlam0AMoVD4GsqOQUAVZJ4BaGs6GKSRGFmxcvjzm4TkApb8288VpIBKNfOthPyszdgRitJEcZtPYFxWyucD93izpeXAiAFXQu/4kThtmq4uVYCXaEkfe+IYfjHT47GsJhKkiYZQMshAtSHpCQIGLetoHE7zYXbXDBhNF2DewnlON8nv50LV6Gi0+wNnS2f4ypIRYfbivyuU4aGYf3EVgvZwwM13lOS5J5jRkkqMtzmqlsAsHt//9rq18DeE+l5roaZqRXTsvbff39MmDABZ511FhYuXIiXX34Z5513Hk4++WQMHToUAPDZZ59hxIgRnjLU1NSEyZMnY8qUKfjvf/+LRYsWYdKkSRgzZgw+97nPAQCOO+44jBw5Ej/84Q+xePFiPP3007j88stx7rnneirPgw8+iNNPPx033HADRo8ejTVr1mDNmjXYutVn1r/4xS/wwgsv4OOPP8bcuXPxrW99C4Zh4JRTTqnUKalu5CQzmgRpBIxSjduUJMnIRDnQ1UoShavefOcudnkJxm2LEKcwNSiALPvdhuVJSsGCrpVm3HY8SbxxWzLYFYod19nc9VfwJJnQIz3q7sDDlOrbwc3aMnTlJR2VWNKFTG0ph5IUFp7yUgCQArfb+BQAnJIkeq1rGgwvISVPkvzXlGC4KhHN8G3oGnOtuttQ9aZ/Q/xIRLHfdcpwlCQRHCIXXE6/KxpuE82mlCaTpEqSINqRNnS8/evxWPLr8ehPvGb9Od/ZDq8kJcEDDzyAESNG4Nhjj8Xxxx+PsWPH4o477vDW53I5vP/++9i+3b9B/uEPf8DXv/51fOc738EXv/hFNDc349FHH/XWG4aBJ598EoZhYMyYMfjBD36A008/HVdddZW3zR133IF8Po9zzz0XQ4YM8f4uuMCf9r5y5Uqccsop2G+//fC9730P/fv3x/z58zFw4MBKnpLqxeZPxMsF5Uqk9d/ighIF3rC7s+Gg7wIn3u6/jx1uEylJ5OaV5Keba2PbCQkNGbE8SYWndYlx29BE4TZJfxudB6YB9kZmcUenQ55ztoHhA3qFDhjV4FvoKnSlJ6ku45/XK78xEl/cV3xvZBM2lpckxSG7NcS4zYfbahiSRPsG5rWrpPBlTWSz21zyRpUkXdeYY7jHpn3YtW/8FAvFh9s0jBjSW7guZYiL/KYkn1PUC6mSRPaT5WPqVZNCQ00Kfer9+2BfjiR1dUmhKFS0LEm/fv3w4IMPStfvscceAUmxtrYWt99+O26//XbJXsDuu++Op556Srr++eefj+zb9OnTI7fZEfFm36/i4F4twD5fTbZjb7FPLG5CyqJRSSWpO8NtFDRkWQpJ0mm4rfibh7fvQd8D3mInKsQLt3HGbe48G7B8Iu3eUGW5VZqcyQgD7c0M11q2ZjMOBJDJZDB8QAPeXyOfdVoNknxXCVdheZJqUnryNCEhoHXcdIkCAcjJR1epADQFQCCZJJcnyXutscTOVYn4siYyj5XrSaLfh1M+h6pNzjYt7X4IcEhT/Ptp8eE2Hd8YORj/fH0VXlm+Ce2kaKyha2iqS2N7lrU5GJLfkNADJ1WS4hu3aeiNV5JkIc7uQvfTNIWyIms0AJOfAb54cbIdP/cTYMx5weUCT9IWO1gZvmhwM6/Kiu7gSHV9g8uKIUlCTxIJt9nF/3Q9T9KJtyGvsf05P/UoRmoSVbGAgHGb85UxKQQiyJzeZxcAQJPGql3btnc4x6oREzEKL9xWLaS4gtglRIloTjAAxwENt9m23M9iMITDX14R76QA1JPUxg3+bMZt0jeO2LmDcWi4jVzKokK1hsYSSVdBWt/qT5xIdUEyyZShIWXouG/SkXh96lcZJSula8Ks6rLvVvSbknuS/NeRJImkAKCEiUd6Z86TpNBNSKIw0IG4pjcw/jfI8eKiYFC/uu9v8Ia1F84xrgqsS4xKkqTuGDTLRZJK9ST94FGgaTfhKi8YlqrBysZDmXWnpv6LvfTVoU2n3Qze7mfhPGppJsN3+ECZ7jPUKYvDo3BduGb1sIdq94YcllOn0qg0HXjwzNG4/IT98cV95FUByllSBGCn33fmTWnoS1aOoqtEgDSZ3dbBkyQmBYAs3KYRkhSSAkAQSmO21cVq096C3ExxUGxk1SUWmuaE1lKcx6dZRJIkX1YyT1ISJcn/rdKZboF+VYFKXNFwm0I3IInun6oJGLZTms1yC0F7v5h0Mv4yZzQuO3r3IjtJsLMZtweNBDZ9xC6jM9pKSgFASFIUGd77WODs54Hr9wy2QwhWVB03Ydf4umycr4xVksLbz9TUYRN6YwDY7PW2S5Ji5NFyExgeMLQJl35tBIb2SV5ao9px9N4DcPTe4WWTyk2SKBHozFmJq793VRqGGmLc3s49GMmM2zz5kSlJ7Vk2VOUdMy1Wkugx3H79z+eHwzRtHLv/4MA+YSjWf8Zn9XbUK+dzpHRdqDjKaqSJuhCnLElUUtJMSse8S78Cy0bAI0UPWQ212xRJ2tmQREkyMgGSpMWYuTakqQ5Ty5XJeGdTkk64Ech3Akee6S+js/kkhYQDEHiS6mtrgMJsfjNOuE2SRZ1JJlnEQObOXNPcay0s3BaBVCqNjXYjBmgsSWre/Eqhr855oN/kb751IH75z7f9NkgI45wv7RX72DsbBoeU8CgGdNDryJuJq7931QxDL+O2ZQeM17LZcXxY0DNuc54k6nEyIkhAIAVA2k8B8LNj94nzUVjEvH1lUjrTb96jl+bKjAjDbTKSJOhEnLIkccqvDJHVCSSHrAaSpMJtOxnCpncHIKrt1tXEwqogSdr1iOhtyo3eg4Ef/B3Yb4K/rCglKXgT1omqYsZRgCTH6ttAb07FkyRvvrPNkqJ0Ak+SZqSR14KehD3XPwcAsARK0qlH7YYBvbovtFatOPkop0zSocP6lL3tzpwlTcUgs9l0mXGbKklcuI0O1rKZd7rmq2E8yaIpBUQz1/x1hUK5ESG5JIh7J15w6bH4y+n+vY6vd8Ybob92YHCSjpQkCTohD7v6y0shN5SYVYNxWylJOxsS1PSKrWpUEpUMtx3+IwAasPvnK3eMOKBpFOLmSRJ9j4Q4hSWE9LcXH2tYf2K8T/y0bxMlqbCIV5I0En6Lal83YGry25At+AyapqEj1wWppxOgGvIy7T2oNxZediyaQoywxaIzb8VSkuig2vXhNpuZyQWwhEHunfKVJD7cRmvBsUkOxUSECbcJQnJJEDfc1rchgwN3aZT2jfbJ0DXsMaABL15yDOZ+uAH/7x9vectFEM2wi5MnySiT4boafldKSdrpUKqS1MWoZLjNSAFHTgYGjajcMeKgTJ4kuqyuNkY7kgSOGqPuJLsJ6cTF5LXDhWh/maJpP6JIUgpWCElyySJ/r+7MVzJT+46LQY21RRWpjUJHLl64jaKELBUM/vnTo3HMfvL8de6AvD1nBq4TJocTk3Gb9JMqSXy4jWTw5k3gjCG6cA5M4nQW5SNKgiSWJCYZJBfqop/b7fOwfvUYSMKz8nBb+LGY45TJj9adOYBFUCRpZ0MSBh+3REYlMbCbCUxXoEx5kuiyw3YPN/ECkKs4ZPSKpUgRGLTkiMS4PVL/JLoPLvQUY0jn4SpJvDeCn4WkUFk4SpJ4HaskkVBJmVSAw3bri3smHYU9+ouzSLsDvymYikWJDNMdTlXylSS2jbZOn4zvTxI0vvXZViaU5952qRJVergt/jXOhLpC0gxQMkRJnCw8VmzG7VJmpVXbL1uRpJ0NSW5MRjeSpHPmAIefDnzz1u7rQ1eBJuSMW9JFpCQRclOSnE0JWMJ7mU6K13pPjmH196LkBM0Q+o78A0anAKgG7DmwjLnDqhB96tNSBaGrPEmivEtUBRIdl1WSSFuSPEm8J+lbh+/iva5JGdilMHPysN36MDl8XEJISVJYSZU4CEscyoMxTXPnQ5Oso3mwZDmtRCG/inuSquzHrkjSzoYkGvfB33P+999bvP4LPy+9PzIMOcQhSI3VXVC4LEgVQZKivkdZLbTEbScNt1nQ+Gzadljoi7Q/8kRBg0ZouC1OCoDuxCPnjMH5x+6DU48S56Ta0XHHD0fhqyMH46Jx+0oHUjpoRmQPKQmigVcn2bJd9CIV6OOUTNE1zfPxuOE2XQP+Nnk0zhw7nGn7sXM/j2u/fRAmfX4PRi3xw3X+GSjZT5OAKxgyxQx80WFCkjLRShIlLCOHOL6nz+3ZX7gt730qFtVFkZRxe+dDEpJ05JlAvz2BXUaJ1x87tTx96umgZu24T0nCGyw1UpSJJCW8kdNwWzwlibT/3XuBebcBs35VWGcAmhbuSZKQpN61KWzryHdrAkkAOGp4Pxw1vF+39qGSOO6AZhx3gDMbSjbuGUy4Tby8HBApWXyWa8AhSVsLpUBk5IExcetagCQ01qUxVpC4c2DvGpxSIMSsKVysRJWCJOE26jsKIyiUuFEliVema9M6OnIWvkTq9T3xs7HI5i2GXMnQVab9roAiSTsZwvwdAehG8hpvCqWhTwmKAyU33akkeSSp0IdQdYwZjYBdDifvndtPeLjN9SSxeOisz+HGWR/gkgn7xey5QqmQh9v85TQ8U44CtxSi3Dvu1HsKOojTmV6MJYkLt/HT5uMM8nwmawDYb7C4sGwxSBJ1CjvvMkWLIUncJs//4hi88elmHDfSTxdg6FooQaLhwZI8SVUmJSmStJOhM19lV5iCgx+/DLRvAvqWkKW8tsl/XcoAVIKS1Edr843bnpKUl+/AK5sM0SvcfsKUJEN8Uz5wlybc/aMjI3qrUE7IBlu6mJqny127TeTDo9myXdByKvJwG9sGT8DihMoY43Zh+/2ae+PBM0eXpY5ekjs5TTfQUBNvWK8l54n3PzU31WJCUzIrBA3NVUMSyHJBkaSdDOtbK5nBWqFoNB9YehsNZBp0KeE2gqSz2+bUXOR3IWm4DRCqYWFKkje7rdoeL3sg4hi0yxlu4iH2JAXDgHWSWVtheZL4afNxPNdUpaLbR5WPiYuDd23C6yu2RNZBA4D6TAr3TDoSsFlPFiB/DqLnqRzfG/2FKk+SQnXgzOeAV+8CBh8APH0ZAKCupvyJ5BSqBA3k5ltKuI2iBEXKI0lxjduAkCSFmbO1Kjdu9yTEyZPE5xkqJ3gi4x6bV33kShK7H11eTLiNmUVXAQ/OH087HH/874c4I2aNzGP2GyRcLusaJXm5MnxvVI3iz2ciVNkDkboD7cjYdZTz987j3qIv7Cv+oSjsBKAkKa5Bf/iXgOUvhGxQStiuVCWpML1fUJbEg7FjpADoCYhFkiqoJGUESTI1Laha1Gf8YY3JkwQ5YeLbiBVuoypVBcJLQ5rqcPXEMijQMcBnGi8GNuOLKqGdkntSXqgUADsDSFLIXnEyMSvsmKgnJCnbFm+fH/4zfH1JSpI44za3lfx4LkkKU4vilnFRqDjiFDatJJnNCJQkp6gsu4yai4cP6CVsi69nxpfyiBMuovuU26ReTmgxHoSyZUjOSr/7aignUi4oJWlnAM3iXK5aAArVB5pvafvGePuUKywngO9JCjNu8ySJ9MclRyEkyS3qu2tfScVwhS5DnNltlURGkME6beiBftVnDPzzp0dj1ZYOHDC0MbAPwBYm0LTgzLk4n0mWtLLaEIevlENJiltrLgrVphorkrQzoJ4k96rr2339UKgsbAuo6+fMktvtc+VpsgRS7SlJJNxmH3IqtMUP0o24nQSepDC1qLDuiD364eoTD5AqAwqVh4wIdJVqwKs9gEOceJJUlzFw2G59cdhuwNbt/kQWmneIreOmMdmzgXjEIs3MbovevppRHpJUho4gWX6oroAiSTsDmg8CvvVnJwRz0HdLa6vPbsCWFcCAfcvTN4XywbaAC990VKS+e5Sp0TJ4kohxW8v45TksaMF4PpN+wCgsCjFuG/66H47Zo9ieKpQBMuLQZUqSlCSxy+rT/jVDLzeqULAz3YJ5fRIbt6uYJcXpWZwZdFHYWWegKpK0M0DTgENOLk9bp/8LmP8nYMx55WlPoYywgZrezl/5Wiwamjsw0HBbmobFBLdnUZ4kXa4kacqTVDWQG7e75vhpQbgtY+iheZLY4rv+NnyJEj69wE7lSQrp280nH4onFq/COV/aq+TjqHCbQs9Avz2B46/v7l4oiBC37lsClObXdMNtpF9ESbJFN2cRSUrJiZCuUgBUDeTV37tPSapJ6aEZt+NUrNe0oHE76ey2HVVJOvHQXXDiobuEbBEf5SI31UaSlMtXQWFHQQVIkmUXf3P3xgWaJ4lRfqJIUiHcFpYYUylJVQNZCKqrCIIomaRr5qarpEoS2Y9NJul8BrZAa3R/DEHttqpEF3WtXErSSUfsCsBJplkNUI9pCgo7CirwqFaSkiQwblMSZMcNt4UQIT2lblHVApm6IiIIosSPpUJExlySZOgarMLFTEkS7ZotqW/mkr+UriFXaCMO6WGM2zuoklROlOv2dMaYPbD/kEYcuIsiSQoKCklQkXBbmTNuM6qQiCQF8yRpKXluL91Qeb+qBfIUAMFloplopUJIkgrHca5FZ5SuI8kkZUoScxnqLknSkTPNwH5x+lMBTrjDoVxKkq5r+Nye/aM37CKocJuCwo6CPb9cnnbIAFDKtF3NzcHketjGTmHyMkV7kgrhthAlic5uU+heyMQSkcLUZSTJVZJIH2ThNgqq/Lib0FxJSY3bVe1J6qJQYJVZicoGRZIUFKodgw8ELloC9Bte3P6Hnsa+L1u4rfD/iP8Bfv4BcOzUGEpSMNymhxAhXXmSqgaykJLIqyRK/FgqxCTJud7oKlq4VZYNXFTslp2tFt0ftnhuFZOkLjpOuZSkaoMiSQoK1Y50HdC0a/H7f+MW4CdzhatKUpLowNB7sPNITjN8l0FJ0kNCcQpdizi121yIZqKVCiEZM1zjtr+ujvEk0X3EniSX6xgJSc9hu/mJe7dnw4o89wzspBxJkSQFhapHqXcfIwUMPkC4qhQlSRdl62YGl3hKUihJUkpS1UA6nV5wGVRESRIYf7zZbaRzfPZsEWhLnpKUcEr/KUcN816v3Lw9cvvuQlepXIokKSgo7HQoqRqBUCnyn+JTgqrtjNLkhtvC8iSFrFPoWiRRkioxu02YAqBwHLpKRKYAeTJJ93UqYXJI2sZmUv6k2tBVkcCdNeO2IkkKClWPyt18SpndJlSSCAkSFCURliUJm8FmKJJUNYhTlsT1A43de2DZj28IFCKaAsCFiEwBfJ6k4OvUDjKlv1pRrtpt1YaKkqRNmzbhtNNOQ2NjI/r06YPJkyejtbU1dJ+Ojg6ce+656N+/P3r16oXvfOc7WLt2LbPNihUrcMIJJ6C+vh6DBg3CxRdfjHzeL43w/PPPQ9O0wN+aNWuYdm6//XbsscceqK2txejRo7Fw4cLyfXgFhVKx11ec/0edU7FDlOSkiFCShI5RJtxWCJUoJWmHgLzArf965oVfwK++PhIXj9+v/McXHN4lSXyZERFY43Zwm1QRU/oby1DzrNK44htOqP28Y/au6HGUcbsInHbaaViyZAlmzZqFJ598EnPmzMHZZ58dus9FF12EJ554AjNmzMALL7yAVatW4dvf/ra33jRNnHDCCchms5g7dy7uu+8+3HvvvZg6dWqgrffffx+rV6/2/gYNGuSte/jhhzFlyhRcccUVeO2113DIIYdg/PjxWLduXflOgIJCKTj1EeDchcDB36vYIfrUlWCMFg1GeoLZba6SRIjQImsffLfT/y0bypNUNZDmSSLLd+/fgMljhzPm6XLBEBW4NYLHkSlJFGKSlDyD9pf2GxS9UTdjzF798e5VE/CLChBXCqUkJcS7776LmTNn4s4778To0aMxduxY3HrrrZg+fTpWrVol3Gfr1q246667cOONN+IrX/kKRo0ahXvuuQdz587F/PnzAQDPPPMM3nnnHfztb3/DoYceiq997Wu4+uqrcfvttyObzTLtDRo0CM3Nzd6fTn4EN954I8466yxMmjQJI0eOxLRp01BfX4+77767UqdEQSEZjDQwcL+KmgoOGtZHvvIrlwM/fkm6Wmzc1sWvvWXks9Q2AmBDankY+NT2Bx5DZdyuGnR37bawVAOMSiQNt/kbMZdp4X8xGbSvmXggfnT0HnjivLGxtu8u/P/27j8qqjr/H/hzhmEGiWZGFBhQQFl/IP4OVhzT2l0m0ThblsdVlzpiZGVQllbqbpuc9rNLp7Yt65R920zbU62r2+aW+QvBcDWEnCQVk7QoXHSgIn5Yyq95f/8wrjNyB2bQYYY7z8c59zjzvu975/2e8cy8eL3f9329EbR2pcwoyWtBUklJCYxGI1JSUqQyi8UCtVqN0tJS2WOsViva2tpgsVikssTERMTFxaGkpEQ67/jx4xEVFSXVSU9PR1NTEyoqKpzON2nSJERHR+Omm27CgQMHpPLW1lZYrVan11Gr1bBYLNLryGlpaUFTU5PTRtSfyf0lLpnxCGAa73K3Su6HxOnXp4f9g0cBAIKCL2WzBFRoxKWb5Krtzn/4kO+4vC2JH9y7zXHSsMs5ST0NtzkuDulm4GcYEIy8W8ZivJ/cZ8yXmEnykM1mcxreAgCNRoPw8PAuc4Mcj9FqtTAajU7lUVFR0jE2m80pQOrc37kPAKKjo/HKK6/gnXfewTvvvIPY2Fj84he/wCeffAIA+Pbbb9HR0SF7HldtA4D8/HwYDAZpi42NdVmXyH85XebjulrnvgcPA5oQuQpdizwZbhs8+qdDLmWSOoQaF6DFWRGORhGKtmuvYH0ouqpczUnqq9Wm5YIxKUhyoz2uJm7LHefGKgJ0Gc5J+smqVatkJ0U7bidOnPBGW902evRo3HvvvUhOTsa0adPw+uuvY9q0aXjuueeu6LyrV69GY2OjtJ0+ffoqtZjIR+SGxC4XngDETulSLD/c1sNikmqH4bOopJ+KLmWS7FABUOGGlufx6pTtGDrY2HP7qE+4Hm7rm9eXW3tJF9Q1k+RO0CaXSXI8TNVn61Qrh12hqSSPB/xXrFiBrKysbuskJCTAZDJ1mQTd3t6O+vp6mEwm2eNMJhNaW1vR0NDglE2qra2VjjGZTF2uQuu8+s3VeQFgypQp2L//4vyKwYMHIygoqMtVc46vI0en00Gn07ncT9Q/9LDgoxx1168K2eG2njJJwQOAm/54MTj7aRVxx0yS+OmY2RPj8GjGZPfaRn3C5XBbH81JmpkUhXFD9DhWc2mag2wmya01jhwfX3xS/d2lBSFP2DiVwlPKDJF6kUmKiIhAYmJit5tWq4XZbEZDQwOsVqt0bFFREex2O1JTU2XPnZycjODgYBQWFkpllZWVqK6uhtlsBgCYzWYcPXrUKQArKCiAXq9HUlKSy3aXl5cjOjoaAKDVapGcnOz0Ona7HYWFhdLrECmXm8NtjuSCpJ7mHLnKUl3/IDAtV3rqOHHbjq7r3pB/8GQxSW8ICQ7CtgdmOE2w1nkwcTt24ADpsVym6EzjBenxF9/8cKXNDTgKHW3zPJPkrjFjxmDWrFlYsmQJXnnlFbS1tSE3NxcLFixATEwMAKCmpgZpaWn4+9//jilTpsBgMCA7OxvLly9HeHg49Ho9HnjgAZjNZkydOhUAMHPmTCQlJeHOO+/E008/DZvNhscffxw5OTlSluf555/H8OHDMXbsWFy4cAGvvfYaioqKsHv3bql9y5cvx6JFi5CSkoIpU6bg+eefxw8//IDFixd76y0h8kPuBklyl+L3tE6Se+dWOwVJKk8OpT7k6nZsfR3QDggOQlvHxXXxEiLCAHQ/H+af90zF1/U/Ot1rracm33tjwpU3NMAodU6SV6+vfeutt5Cbm4u0tDSo1WrMnTsXL7zwgrS/ra0NlZWV+PHHS2nO5557Tqrb0tKC9PR0vPzyy9L+oKAgbNu2DUuXLoXZbMY111yDRYsW4cknn5TqtLa2YsWKFaipqUFoaCgmTJiAPXv24Je//KVUZ/78+fjmm2/wxBNPwGazYdKkSdi5c2eXydxEiuZuNBIk81Uhu+K2Y5m7QdKlOUkdnZkkRkl+x3UmqW/bMThMh6YLF4OkEZEXg6TuxnpSEwYhNWGQU5lcX26bPATvHq7BrZNi8MhM764ppEQMknohPDwcb7/9tsv9w4YN63K/l5CQELz00kt46aWXXB4XHx+P7du3u9z/2GOP4bHHHuuxfbm5ucjNze2xHpFiXcFwW88rbnueSeqck8ThNv/jak6Sq3JvmTXOhJc//AKh2iDp/4mnP89yTc6/fTzumBqPybFG3pakFxQaI3k3SCIif3eVh9scJ267c+UcAI3MxG3+SPkff/lIHrKMwoDgINw09lLW39MshvMNbi/+GxIchOT4gS6OoJ6EBAehpf1K7pjtn7gaBFGg0cdceuwqC3B5UOR2JunKhttUP+UEONzmf/pqgnZPtBo1HkgbiUSTXiq7kiyGf/Sq/3s9KwVDjAPw/+5M9nVTripmkogCxW83A5/vvOyGuS5+IrShzs/l5iT1OHHbvWY5rritxsW/RP0la0GX+EuQJEco9gL0/iM5PhwHVv3K18246hgkEQWKUekXN0eufviCLwuS5IbbenODWxlBwZfOrf7px47Dbf7Hnz8Spc6HId/jcBtRQHM3SOpNJsnNOUkax0zSxV+7pGi9q+rkI/6dSSLyDmaSiAKZq0Dm8iCpN0sAuPmjGuRwk11tkMAf0pMw9zres83f+PP9zC6/SproamGQRBTIXAUyl89J8uLEbY1DkKTXqpE9fbhbx1Hf8utMEmMk8hI//tuAiLzP1XDbAOfn3lxx2+nGoh1uHUN9z6+DJF83gBSLQRJRIHN34nZQLyZuuzknyXHNGrVQ3jorSuHXQdIVrQHgv/0i32OQRBTQXPxATFzg/NzpqrVujlV5fnWb8xmZE/BXfn112xUc68fdIj/AIIkokMn9FZ3xLDDmFuey3iwB0Iu/0FXMJPktf16WgXOSyFsYJBEFMrkhsZjrugY4sksAyJ3P8bjeZJIYJPkrfx5uI/IWBklEAU3mh08umyM7J0nm66MX6yQ5UgtO3PZXcokkjR9nl4iuBgZJRIFM7jdOLkhydwmAKxxuUzOT5LdUMp+nPw/BEV0NDJKIApqbmaTerLjNiduKwkwSBSIGSUSBTC7bM2hE17JeLQHAidtKIjcnKUgB85TGDTH4ugnkx7jiNlFAc/iRW7QN0McA1wzuWs3tTJLD310ancet4XCb/5LL8fXn4bY9y2/E6e9/xKRYo6+bQn6MQRJRIHMMagxDgXAXtwSRnaTdQ5Aku0p39xgk+S+5BRv783DbiMgwjIgM83UzyM9xuI0okDkGOkHaburJfVX0MNwmd1PcHnDFbf+ltEwSkTsYJBEFNMcgqZvMj2wmqYclAHqRSeI6Sf5LbsHG/pxJInIHgySiQGZvv/S42yBJbmitp0wSh9uUpWuUFMQgiRSOQRJRILM7LN54NYbbnOYk9WK4jUsA+C07M0kUgBgkEQUye9ulx90Nj6lkbnArm126wkwS5yT1K5yTRErHIIkokF3JcFtPE7d7dXUbb0virzgniQIRgySiQOYYJHW3MKDbSwC4ORHcBc5J8l86Tdf/A7zpLSkdgySiQNbR3nMdwP1MkiMPgqQ2cTEDValKcPsY6lsThhowL3kolt80SirjxG1SOgZJRIHM7m6Q5GYmyZEHw203t+ZjY/tM5GkedPsY6lsqlQrPzJuIB9NGSmUcbiOl44rbRIHMceJ2d3oTJHmQSTophiKvPQuR8PxWJuQ7nLhNSsdMElEgczeTdE2ETGFPmSSZK+J6ag5XAOhX/CWTlJkaBwBYOCXOxy0hpWEmiSiQ2d28mixqLDDz/4Ddj18qu4rDbZ3k7g9G/kenUaOl3Y7pI+SC576Xd8tY3DppCG9WS1cdgySiQNbh5nAbAEx7AKf+uxkjzh/5qaCn4bZuFqd0wc4gqV/Ys/xG7D/1LeZeN9TXTQEABAepMWV4uK+bQQrEIIkokLk7J0lOj3OSPP964XBb/xAbHsqhLQoIXp2TVF9fj8zMTOj1ehiNRmRnZ+PcuXPdHnPhwgXk5ORg0KBBCAsLw9y5c1FbW+tUp7q6GhkZGQgNDUVkZCQeffRRtLdfmluRlZUFlUrVZRs7dqxUJy8vr8v+xMTEq/sGEPk7d+ckSVQuHsvwYLhtzqQYAMDSX/zMw/YQEXmPV4OkzMxMVFRUoKCgANu2bcO+fftwzz33dHvMww8/jPfffx9btmxBcXExzpw5g9tvv13a39HRgYyMDLS2tuKjjz7CG2+8gY0bN+KJJ56Q6qxduxZnz56VttOnTyM8PBzz5s1zeq2xY8c61du/f//VfQOI/J276yT9xCnRI3s/NwceXN32l3kTsWPZDNx7A9dJIiL/4bXhts8++ww7d+7Exx9/jJSUFADAiy++iJtvvhl/+ctfEBMT0+WYxsZGrF+/Hm+//TZ+9atfAQA2bNiAMWPG4ODBg5g6dSp2796N48ePY8+ePYiKisKkSZPwxz/+EStXrkReXh60Wi0MBgMMBoN03q1bt+L777/H4sWLnTuv0cBkMnnrLSDyfx5nkhxcxYnbmiA1xkTre98WIiIv8FomqaSkBEajUQqQAMBisUCtVqO0tFT2GKvVira2NlgsFqksMTERcXFxKCkpkc47fvx4REVFSXXS09PR1NSEiooK2fOuX78eFosF8fHxTuUnT55ETEwMEhISkJmZierq6m771NLSgqamJqeNqF/zNEhSuTHclpwFGOKAiQt62yoiIr/gtSDJZrMhMjLSqUyj0SA8PBw2m83lMVqtFkaj0ak8KipKOsZmszkFSJ37O/dd7syZM9ixYwfuvvtup/LU1FRs3LgRO3fuxLp161BVVYUZM2agubnZZZ/y8/OlLJXBYEBsbKzLukT9wqArmAPkKpP067XAQ0eAEGaGiKh/8zhIWrVqleykaMftxIkT3mhrr7zxxhswGo2YM2eOU/ns2bMxb948TJgwAenp6di+fTsaGhqwefNml+davXo1Ghsbpe306dNebj2Rl6U9Afz8buCu3b04uLsb4vrHIoNERFfC4zlJK1asQFZWVrd1EhISYDKZUFdX51Te3t6O+vp6l/OATCYTWltb0dDQ4JRNqq2tlY4xmUwoKytzOq7z6rfLzyuEwOuvv44777wTWm33a7YYjUaMGjUKp06dcllHp9NBp+NtE0hBQgxAxrMeHOAQ/DAQIiKF8ziTFBERgcTExG43rVYLs9mMhoYGWK1W6diioiLY7XakpqbKnjs5ORnBwcEoLCyUyiorK1FdXQ2z2QwAMJvNOHr0qFMAVlBQAL1ej6SkJKfzFRcX49SpU8jOzu6xX+fOncMXX3yB6Ohoj94PokByjc7x7yoGSUSkbF6bkzRmzBjMmjULS5YsQVlZGQ4cOIDc3FwsWLBAurKtpqYGiYmJUmbIYDAgOzsby5cvx969e2G1WrF48WKYzWZMnToVADBz5kwkJSXhzjvvxKeffopdu3bh8ccfR05OTpcsz/r165Gamopx48Z1ad8jjzyC4uJifPXVV/joo49w2223ISgoCAsXLvTWW0LU75kMAy496WkJACKifs6rK26/9dZbyM3NRVpaGtRqNebOnYsXXnhB2t/W1obKykr8+OOPUtlzzz0n1W1paUF6ejpefvllaX9QUBC2bduGpUuXwmw245prrsGiRYvw5JNPOr12Y2Mj3nnnHaxdu1a2bf/73/+wcOFCfPfdd4iIiMD06dNx8OBBRET4x72IiPyRWsXhNiIKHCrBO0r2WlNTEwwGAxobG6HX80oeCgBv3AJUFV98nNfo27YQEfWSu7/fzJcTERERyWCQRERERCSDQRIRuY/zkIgogDBIIiIPMEgiosDBIImIiIhIBoMkInIfh9uIKIAwSCIiDzBIIqLAwSCJiIiISAaDJCIiIiIZDJKIiIiIZDBIIiL3ceI2EQUQBklE5AEGSUQUOBgkEREREclgkERE7uNwGxEFEAZJRERERDIYJBERERHJYJBERB7gcBsRBQ4GSUREREQyGCQRkfs4cZuIAgiDJCLyAIMkIgocDJKIiIiIZDBIIiIiIpLBIImI3Mc5SUQUQBgkEREREclgkEREHmAmiYgCB4MkIiIiIhkMkojIfZyTREQBhEESERERkQwGSUREREQyGCQRERERyWCQRETu45wkIgogXguS6uvrkZmZCb1eD6PRiOzsbJw7d67bYy5cuICcnBwMGjQIYWFhmDt3Lmpra53qPPjgg0hOToZOp8OkSZNkz3PkyBHMmDEDISEhiI2NxdNPP92lzpYtW5CYmIiQkBCMHz8e27dv73VfiYiISHm8FiRlZmaioqICBQUF2LZtG/bt24d77rmn22MefvhhvP/++9iyZQuKi4tx5swZ3H777V3q3XXXXZg/f77sOZqamjBz5kzEx8fDarXimWeeQV5eHl599VWpzkcffYSFCxciOzsbhw8fxpw5czBnzhwcO3bsyjpNpHjMJBFRABFecPz4cQFAfPzxx1LZjh07hEqlEjU1NbLHNDQ0iODgYLFlyxap7LPPPhMARElJSZf6a9asERMnTuxS/vLLL4uBAweKlpYWqWzlypVi9OjR0vPf/OY3IiMjw+m41NRUce+997rdRyGEaGxsFABEY2OjR8cR9Vub7hBijf7iRkTUT7n7++2VTFJJSQmMRiNSUlKkMovFArVajdLSUtljrFYr2traYLFYpLLExETExcWhpKTEo9e+4YYboNVqpbL09HRUVlbi+++/l+o4vk5nnZ5ep6WlBU1NTU4bUUDhnCQiCiBeCZJsNhsiIyOdyjQaDcLDw2Gz2Vweo9VqYTQancqjoqJcHuPqPFFRUV3O0bmvuzo9vU5+fj4MBoO0xcbGut0uIiIi6l88CpJWrVoFlUrV7XbixAlvtdXnVq9ejcbGRmk7ffq0r5tEREREXqLxpPKKFSuQlZXVbZ2EhASYTCbU1dU5lbe3t6O+vh4mk0n2OJPJhNbWVjQ0NDhlk2pra10e4+o8l18R1/m88zyu6vT0OjqdDjqdzu22ECkPh9uIKHB4lEmKiIhAYmJit5tWq4XZbEZDQwOsVqt0bFFREex2O1JTU2XPnZycjODgYBQWFkpllZWVqK6uhtlsdruNZrMZ+/btQ1tbm1RWUFCA0aNHY+DAgVIdx9fprOPJ6xAFJM5JIqIA4pU5SWPGjMGsWbOwZMkSlJWV4cCBA8jNzcWCBQsQExMDAKipqUFiYiLKysoAAAaDAdnZ2Vi+fDn27t0Lq9WKxYsXw2w2Y+rUqdK5T506hfLycthsNpw/fx7l5eUoLy9Ha2srAOC3v/0ttFotsrOzUVFRgX/+859Yu3Ytli9fLp1j2bJl2LlzJ5599lmcOHECeXl5OHToEHJzc73xdhAREVF/5K3L67777juxcOFCERYWJvR6vVi8eLFobm6W9ldVVQkAYu/evVLZ+fPnxf333y8GDhwoQkNDxW233SbOnj3rdN4bb7xRAOiyVVVVSXU+/fRTMX36dKHT6cSQIUPEU0891aV9mzdvFqNGjRJarVaMHTtWfPDBBx73kUsAUMDZnMUlAIio33P391slhBC+C9H6t6amJhgMBjQ2NkKv1/u6OUTet2UxUPHvi4/zGn3bFiKiXnL395v3biMiIiKSwSCJiNzHidtEFEAYJBERERHJYJBERB5gJomIAgeDJCJyH4fbiCiAMEgiIiIiksEgiYiIiEgGgyQiIiIiGQySiMgDnJNERIGDQRIRuS92iq9bQETUZzS+bgAR9SPJi4GgYCD+el+3hIjI6xgkEZH7gjRAcpavW0FE1Cc43EZEREQkg0ESERERkQwGSUREREQyGCQRERERyWCQRERERCSDQRIRERGRDAZJRERERDIYJBERERHJYJBEREREJINBEhEREZEMBklEREREMhgkEREREclgkEREREQkQ+PrBvRnQggAQFNTk49bQkRERO7q/N3u/B13hUHSFWhubgYAxMbG+rglRERE5Knm5mYYDAaX+1WipzCKXLLb7Thz5gyuvfZaqFSqq3bepqYmxMbG4vTp09Dr9VftvP4q0PoLBF6f2V/lC7Q+s7/9mxACzc3NiImJgVrteuYRM0lXQK1WY+jQoV47v16vV8R/RncFWn+BwOsz+6t8gdZn9rf/6i6D1IkTt4mIiIhkMEgiIiIiksEgyQ/pdDqsWbMGOp3O103pE4HWXyDw+sz+Kl+g9Zn9DQycuE1EREQkg5kkIiIiIhkMkoiIiIhkMEgiIiIiksEgiYiIiEgGgyQ/9NJLL2HYsGEICQlBamoqysrKfN2kXtm3bx9+/etfIyYmBiqVClu3bnXaL4TAE088gejoaAwYMAAWiwUnT550qlNfX4/MzEzo9XoYjUZkZ2fj3LlzfdgL9+Xn5+PnP/85rr32WkRGRmLOnDmorKx0qnPhwgXk5ORg0KBBCAsLw9y5c1FbW+tUp7q6GhkZGQgNDUVkZCQeffRRtLe392VX3LJu3TpMmDBBWlzObDZjx44d0n4l9VXOU089BZVKhYceekgqU1Kf8/LyoFKpnLbExERpv5L66qimpgZ33HEHBg0ahAEDBmD8+PE4dOiQtF9J31vDhg3r8hmrVCrk5OQAUO5n7BFBfmXTpk1Cq9WK119/XVRUVIglS5YIo9Eoamtrfd00j23fvl38/ve/F//+978FAPHuu+867X/qqaeEwWAQW7duFZ9++qm45ZZbxPDhw8X58+elOrNmzRITJ04UBw8eFP/973/FiBEjxMKFC/u4J+5JT08XGzZsEMeOHRPl5eXi5ptvFnFxceLcuXNSnfvuu0/ExsaKwsJCcejQITF16lQxbdo0aX97e7sYN26csFgs4vDhw2L79u1i8ODBYvXq1b7oUrfee+898cEHH4jPP/9cVFZWit/97nciODhYHDt2TAihrL5erqysTAwbNkxMmDBBLFu2TCpXUp/XrFkjxo4dK86ePStt33zzjbRfSX3tVF9fL+Lj40VWVpYoLS0VX375pdi1a5c4deqUVEdJ31t1dXVOn29BQYEAIPbu3SuEUOZn7CkGSX5mypQpIicnR3re0dEhYmJiRH5+vg9bdeUuD5LsdrswmUzimWeekcoaGhqETqcT//jHP4QQQhw/flwAEB9//LFUZ8eOHUKlUomampo+a3tv1dXVCQCiuLhYCHGxf8HBwWLLli1Snc8++0wAECUlJUKIi4GlWq0WNptNqrNu3Tqh1+tFS0tL33agFwYOHChee+01Rfe1ublZjBw5UhQUFIgbb7xRCpKU1uc1a9aIiRMnyu5TWl87rVy5UkyfPt3lfqV/by1btkz87Gc/E3a7XbGfsac43OZHWltbYbVaYbFYpDK1Wg2LxYKSkhIftuzqq6qqgs1mc+qrwWBAamqq1NeSkhIYjUakpKRIdSwWC9RqNUpLS/u8zZ5qbGwEAISHhwMArFYr2tranPqcmJiIuLg4pz6PHz8eUVFRUp309HQ0NTWhoqKiD1vvmY6ODmzatAk//PADzGazovuak5ODjIwMp74Byvx8T548iZiYGCQkJCAzMxPV1dUAlNlXAHjvvfeQkpKCefPmITIyEpMnT8bf/vY3ab+Sv7daW1vx5ptv4q677oJKpVLsZ+wpBkl+5Ntvv0VHR4fTfzgAiIqKgs1m81GrvKOzP9311WazITIy0mm/RqNBeHi4378fdrsdDz30EK6//nqMGzcOwMX+aLVaGI1Gp7qX91nuPenc52+OHj2KsLAw6HQ63HfffXj33XeRlJSkyL4CwKZNm/DJJ58gPz+/yz6l9Tk1NRUbN27Ezp07sW7dOlRVVWHGjBlobm5WXF87ffnll1i3bh1GjhyJXbt2YenSpXjwwQfxxhtvAFD299bWrVvR0NCArKwsAMr7/9xbGl83gEiJcnJycOzYMezfv9/XTfGq0aNHo7y8HI2NjfjXv/6FRYsWobi42NfN8orTp09j2bJlKCgoQEhIiK+b43WzZ8+WHk+YMAGpqamIj4/H5s2bMWDAAB+2zHvsdjtSUlLw5z//GQAwefJkHDt2DK+88goWLVrk49Z51/r16zF79mzExMT4uil+hZkkPzJ48GAEBQV1uXqgtrYWJpPJR63yjs7+dNdXk8mEuro6p/3t7e2or6/36/cjNzcX27Ztw969ezF06FCp3GQyobW1FQ0NDU71L++z3HvSuc/faLVajBgxAsnJycjPz8fEiROxdu1aRfbVarWirq4O1113HTQaDTQaDYqLi/HCCy9Ao9EgKipKcX12ZDQaMWrUKJw6dUqRny8AREdHIykpyalszJgx0jCjUr+3vv76a+zZswd33323VKbUz9hTDJL8iFarRXJyMgoLC6Uyu92OwsJCmM1mH7bs6hs+fDhMJpNTX5uamlBaWir11Ww2o6GhAVarVapTVFQEu92O1NTUPm9zT4QQyM3NxbvvvouioiIMHz7caX9ycjKCg4Od+lxZWYnq6mqnPh89etTpS7agoAB6vb7Ll7c/stvtaGlpUWRf09LScPToUZSXl0tbSkoKMjMzpcdK67Ojc+fO4YsvvkB0dLQiP18AuP7667ss2/H5558jPj4egDK/twBgw4YNiIyMREZGhlSm1M/YY76eOU7ONm3aJHQ6ndi4caM4fvy4uOeee4TRaHS6eqC/aG5uFocPHxaHDx8WAMRf//pXcfjwYfH1118LIS5eSms0GsV//vMfceTIEXHrrbfKXko7efJkUVpaKvbv3y9Gjhzpl5fSCiHE0qVLhcFgEB9++KHTZbU//vijVOe+++4TcXFxoqioSBw6dEiYzWZhNpul/Z2X1M6cOVOUl5eLnTt3ioiICL+8pHbVqlWiuLhYVFVViSNHjohVq1YJlUoldu/eLYRQVl9dcby6TQhl9XnFihXiww8/FFVVVeLAgQPCYrGIwYMHi7q6OiGEsvraqaysTGg0GvGnP/1JnDx5Urz11lsiNDRUvPnmm1IdpX1vdXR0iLi4OLFy5cou+5T4GXuKQZIfevHFF0VcXJzQarViypQp4uDBg75uUq/s3btXAOiyLVq0SAhx8XLaP/zhDyIqKkrodDqRlpYmKisrnc7x3XffiYULF4qwsDCh1+vF4sWLRXNzsw960zO5vgIQGzZskOqcP39e3H///WLgwIEiNDRU3HbbbeLs2bNO5/nqq6/E7NmzxYABA8TgwYPFihUrRFtbWx/3pmd33XWXiI+PF1qtVkRERIi0tDQpQBJCWX115fIgSUl9nj9/voiOjhZarVYMGTJEzJ8/32m9ICX11dH7778vxo0bJ3Q6nUhMTBSvvvqq036lfW/t2rVLAOjSByGU+xl7QiWEED5JYRERERH5Mc5JIiIiIpLBIImIiIhIBoMkIiIiIhkMkoiIiIhkMEgiIiIiksEgiYiIiEgGgyQiIiIiGQySiIiIiGQwSCIiIiKSwSCJiIiISAaDJCIiIiIZDJKIiIiIZPx/M/xZxomWXv0AAAAASUVORK5CYII=", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(outputs.hidden_states[7].detach().cpu().numpy()[0].mean(-1));\n", "plt.plot(outputs_h.hidden_states[7].detach().cpu().numpy()[0].mean(-1));" ] }, { "cell_type": "code", "execution_count": null, "id": "922c5f43", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "561e5d65", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "0d82fa5a", "metadata": {}, "source": [ "### compare to default" ] }, { "cell_type": "code", "execution_count": 20, "id": "27916f9e", "metadata": {}, "outputs": [], "source": [ "%matplotlib inline\n", "from matplotlib import pyplot as plt" ] }, { "cell_type": "code", "execution_count": 1, "id": "38b37b8f", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\"" ] }, { "cell_type": "code", "execution_count": 4, "id": "fe89cdf6", "metadata": {}, "outputs": [], "source": [ "from suno_utils.tasks.mert_v2 import encode\n", "from suno_utils.audio import Audio" ] }, { "cell_type": "code", "execution_count": 5, "id": "3c28bb6d", "metadata": {}, "outputs": [], "source": [ "audio = Audio.from_file(\"../samples/halo.m4a\").get_segment(20, 30)\n", "out = encode(audio, do_clustering=False)\n", "out_h = encode(audio.get_segment(to_s=5), do_clustering=False)" ] }, { "cell_type": "code", "execution_count": 8, "id": "df5faa4d", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(out.T.mean(-1));\n", "plt.plot(out_h.T.mean(-1));" ] }, { "cell_type": "code", "execution_count": null, "id": "f41fbede", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "18d6411a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "b4bb4b97", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "315dbbe9", "metadata": {}, "source": [ "### user mert repo" ] }, { "cell_type": "code", "execution_count": 1, "id": "288daf89", "metadata": {}, "outputs": [], "source": [ "%matplotlib inline\n", "from matplotlib import pyplot as plt" ] }, { "cell_type": "code", "execution_count": 2, "id": "cf294dc0", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"\"" ] }, { "cell_type": "code", "execution_count": 3, "id": "a94a40cc", "metadata": {}, "outputs": [], "source": [ "import sys\n", "sys.path.insert(0, \"/home/georg/code/MERT/\")" ] }, { "cell_type": "code", "execution_count": 4, "id": "f066c6a9", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "2023-08-31 00:11:46 | INFO | fairseq.tasks.text_to_speech | Please install tensorboardX: pip install tensorboardX\n" ] } ], "source": [ "from mert_fairseq.models.mert.mert_model import MERTConfig, MERTModel\n", "from mert_fairseq.tasks.mert_pretraining import MERTPretrainingTask" ] }, { "cell_type": "code", "execution_count": 5, "id": "70f6303d", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "2023-08-31 00:11:48 | INFO | mert_fairseq.tasks.mert_pretraining | current directory is /home/georg/notebooks/dataprep\n", "2023-08-31 00:11:48 | INFO | mert_fairseq.tasks.mert_pretraining | MERTPretrainingTask Config {'_name': 'mert_pretraining', 'data': '/app/suno/data/mert_25hz/audio_tsv', 'sharding_data': -1, 'load_random_data_shard': True, 'fine_tuning': False, 'labels': ['codec_0', 'codec_1', 'codec_2', 'codec_3', 'codec_4', 'codec_5', 'codec_6', 'codec_7'], 'label_dir': '/app/suno/data/mert_25hz/label', 'label_rate': 25.0, 'sample_rate': 24000, 'normalize': False, 'enable_padding': False, 'max_keep_size': None, 'max_sample_size': 120000, 'min_sample_size': 72000, 'single_target': False, 'random_crop': True, 'pad_audio': False, 'store_labels': False, 'numpy_memmap_label': False, 'augmentation_effects': '[]', 'augmentation_probs': '[]', 'inbatch_noise_augment_len_range': '[8000, 24000]', 'inbatch_noise_augment_number_range': '[1, 3]', 'inbatch_noise_augment_volume': 1.0, 'dynamic_crops': '[]', 'dynamic_crops_epoches': '[]', 'cqt_loss_bin_dataloader': -1}\n", "2023-08-31 00:11:48 | INFO | mert_fairseq.models.mert.mert_model | MERTModel Config: {'_name': 'mert', 'label_rate': 25.0, 'extractor_mode': default, 'encoder_layers': 12, 'encoder_embed_dim': 768, 'encoder_ffn_embed_dim': 3072, 'encoder_attention_heads': 12, 'activation_fn': gelu, 'layer_type': transformer, 'dropout': 0.1, 'attention_dropout': 0.1, 'activation_dropout': 0.0, 'encoder_layerdrop': 0.05, 'dropout_input': 0.1, 'dropout_features': 0.1, 'final_dim': 64, 'untie_final_proj': True, 'layer_norm_first': False, 'audio_extract_type': w2v_conv, 'music_conv_nmel': 80, 'music_conv_hoplen': 40, 'conv_feature_layers': '[(512,10,5)] + [(512,5,3)] + [(512,3,2)] * 4 + [(512,2,2)] * 2', 'conv_bias': False, 'logit_temp': 0.1, 'target_glu': False, 'feature_grad_mult': 0.1, 'conv_pos': 128, 'conv_pos_groups': 16, 'pos_conv_depth': 1, 'mask_length': 5, 'mask_prob': 0.8, 'mask_dynamic_prob_step': '[]', 'mask_dynamic_prob': '[]', 'mask_dynamic_len_step': '[]', 'mask_dynamic_len': '[]', 'mask_selection': static, 'mask_other': 0.0, 'no_mask_overlap': False, 'mask_min_space': 1, 'mask_replace': 0.0, 'mask_replace_type': in_sample, 'mask_origin': 0.0, 'mask_channel_length': 10, 'mask_channel_prob': 0.0, 'mask_channel_selection': static, 'mask_channel_other': 0.0, 'no_mask_channel_overlap': False, 'mask_channel_min_space': 1, 'latent_temp': [2.0, 0.5, 0.999995], 'skip_masked': False, 'skip_nomask': True, 'checkpoint_activations': False, 'required_seq_len_multiple': 2, 'depthwise_conv_kernel_size': 31, 'attn_type': '', 'pos_enc_type': 'abs', 'fp16': False, 'audio_cqt_loss_m': True, 'audio_cqt_bins': 336, 'audio_mel_loss_m': False, 'audio_mel_bins': 84, 'feature_extractor_cqt': False, 'feature_extractor_cqt_bins': 84, 'mixture_prob': 0.5, 'inbatch_noise_augment_len_range': '[12000, 24000]', 'inbatch_noise_augment_number_range': '[1, 3]', 'inbatch_noise_augment_volume': 1.0, 'learnable_temp': False, 'learnable_temp_init': 0.1, 'learnable_temp_max': 100.0, 'chunk_nce_cal': -1, 'pretrained_weights': '', 'random_codebook': -1, 'deepnorm': False, 'subln': False, 'emb_grad_mult': 1.0, 'attention_relax': -1.0, 'do_cnn_feat_stable_layernorm': False, 'wav_normalize': False}\n", "2023-08-31 00:11:50 | INFO | mert_fairseq.models.mert.mert_model | train the model with extra task: reconstruct cqt from transformer output\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "CQT kernels created, time used = 0.2464 seconds\n" ] } ], "source": [ "import fairseq\n", "ckpt_path = \"/app/suno/checkpoints/mert_25hz/checkpoint_last.pt\"\n", "models, cfg, task = fairseq.checkpoint_utils.load_model_ensemble_and_task([ckpt_path])\n", "model = models[0]\n", "model.eval();" ] }, { "cell_type": "code", "execution_count": 6, "id": "2c4599ce", "metadata": {}, "outputs": [], "source": [ "import torch\n", "from suno_utils.audio import Audio" ] }, { "cell_type": "code", "execution_count": 7, "id": "0f944611", "metadata": {}, "outputs": [], "source": [ "# audio_arr = Audio.from_file(\"../samples/halo.m4a\").get_segment(20, 30).convert(24_000, 2, 1).array_float\n", "audio_arr = Audio.from_file(\"../samples/covid.wav\").get_segment(0,10).convert(24_000, 2, 1).array_float\n", "audio_arr = torch.from_numpy(audio_arr)[None]" ] }, { "cell_type": "code", "execution_count": 16, "id": "d15db597", "metadata": {}, "outputs": [], "source": [ "with torch.no_grad():\n", " x = model.forward_features(audio_arr).transpose(1, 2)\n", " x = model.layer_norm(x)\n", " x = model.post_extract_proj(x)\n", " x, _ = model.encoder(\n", " x,\n", " layer=7,\n", " )\n", " out = x[0].detach().cpu().numpy()\n", " \n", "with torch.no_grad():\n", " x = model.forward_features(audio_arr[...,:audio_arr.shape[-1]//2]).transpose(1, 2)\n", " x = model.layer_norm(x)\n", " x = model.post_extract_proj(x)\n", " x, _ = model.encoder(\n", " x,\n", " layer=7,\n", " )\n", " out_h = x[0].detach().cpu().numpy()" ] }, { "cell_type": "code", "execution_count": 17, "id": "746e5120", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(out.mean(-1));\n", "plt.plot(out_h.mean(-1));" ] }, { "cell_type": "code", "execution_count": null, "id": "25fa1606", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "88bf8d94", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 10, "id": "1b2ba991", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/home/georg/anaconda3/envs/nb/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", " from .autonotebook import tqdm as notebook_tqdm\n" ] } ], "source": [ "import re\n", "import json\n", "import torch\n", "from transformers import Wav2Vec2FeatureExtractor\n", "from suno_utils.models.mert.modeling_MERT import MERTModel, MERTConfig\n", "from suno_utils.audio import Audio\n", "\n", "with open(\"/home/georg/code/glockenspiel/suno_utils/suno_utils/models/mert/preprocessor_config.json\") as f:\n", " hf_processor = Wav2Vec2FeatureExtractor(**json.load(f))\n", "hf_processor.do_normalize = False\n", "with open(\"/home/georg/code/glockenspiel/suno_utils/suno_utils/models/mert/config_small_25hz.json\") as f:\n", " cfg = MERTConfig(**json.load(f))\n", "hf_model = MERTModel(cfg)\n", "sd = torch.load(\"/home/georg/notebooks/dataprep/models/mert_test_25hz.pt\")\n", "hf_model.load_state_dict(sd);\n", "hf_model.eval();" ] }, { "cell_type": "code", "execution_count": 36, "id": "d5aac8d3", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "torch.Size([1, 249, 512])" ] }, "execution_count": 36, "metadata": {}, "output_type": "execute_result" } ], "source": [ "with torch.no_grad():\n", " x = model.forward_features(audio_arr).transpose(1, 2)\n", "x.shape" ] }, { "cell_type": "code", "execution_count": 37, "id": "07e76ccf", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "torch.Size([1, 249, 512])" ] }, "execution_count": 37, "metadata": {}, "output_type": "execute_result" } ], "source": [ "with torch.no_grad():\n", " inputs = hf_processor(audio_arr[0], sampling_rate=24_000, return_tensors=\"pt\")\n", " x2 = hf_model.feature_extractor(inputs[\"input_values\"]).transpose(1,2)\n", "# outputs = hf_model(**inputs, output_hidden_states=True)\n", "x2.shape" ] }, { "cell_type": "code", "execution_count": 39, "id": "6681f41b", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(x.detach().cpu().numpy()[0].mean(-1));" ] }, { "cell_type": "code", "execution_count": 40, "id": "6605b983", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(x2.detach().cpu().numpy()[0].mean(-1))" ] }, { "cell_type": "code", "execution_count": null, "id": "65e5ecf3", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "36041079", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "b1e9fc05", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "1ea023c5", "metadata": {}, "source": [ "## Playground" ] }, { "cell_type": "code", "execution_count": null, "id": "37d21a6a", "metadata": {}, "outputs": [], "source": [ "MIN_DURATION_S = 3\n", "MAX_DURATION_S = 8*60\n", "\n", "def _convert_float_audio(sig):\n", " dtype = np.int16\n", " dtype_info = np.iinfo(dtype)\n", " abs_max = 2 ** (dtype_info.bits - 1)\n", " offset = dtype_info.min + abs_max\n", " return (sig * abs_max + offset).clip(dtype_info.min, dtype_info.max).astype(dtype)\n", "\n", "out_mm_filepath_tr = os.path.join(OUT_DATA_DIR, f\"music_24khz_tr.bin\")\n", "out_mm_filepath_val = os.path.join(OUT_DATA_DIR, f\"music_24khz_val.bin\")\n", "out_mm_tr = np.memmap(out_mm_filepath_tr, dtype=np.int16, mode=\"w+\", shape=(1,))\n", "out_mm_val = np.memmap(out_mm_filepath_val, dtype=np.int16, mode=\"w+\", shape=(1,))\n", "n_offs_tr = 0\n", "n_offs_val = 0\n", "dset_duration_h = 0\n", "for n_iter, metas_chunk in enumerate(funcy.chunks(5000, metas)):\n", "# t0 = time.time()\n", "# audio_arr_list = load_audio_mp(\n", "# [m[\"filepath\"] for m in metas_chunk],\n", "# target_sample_rate=SAMPLE_RATE,\n", "# min_duration_s=MIN_DURATION_S,\n", "# max_duration_s=MAX_DURATION_S,\n", "# num_workers=32,\n", "# force_threads=False,\n", "# # debug=False,\n", "# silent=True,\n", "# )\n", "# audio_arr_list = [_convert_float_audio(arr.numpy()[0]) for arr in audio_arr_list]\n", "# td_fetch = int(round(time.time() - t0))\n", "# chunk_duration_h = round(sum(arr.shape[-1] / SAMPLE_RATE for arr in audio_arr_list) / 60 / 60, 1)\n", "# t0 = time.time()\n", "# to_write_arr = np.concatenate(audio_arr_list, axis=0)\n", "# to_write_len = to_write_arr.shape[-1]\n", "# if n_iter == 0:\n", "# out_mm = np.memmap(\n", "# out_mm_filepath_val, dtype=np.int16, mode=\"r+\", shape=(n_offs_val+to_write_len,)\n", "# )\n", "# out_mm[n_offs_val:n_offs_val+to_write_len] = to_write_arr\n", "# n_offs_val += to_write_len\n", "# dset_type = \"val\"\n", "# else:\n", "# out_mm = np.memmap(\n", "# out_mm_filepath_tr, dtype=np.int16, mode=\"r+\", shape=(n_offs_tr+to_write_len,)\n", "# )\n", "# out_mm[n_offs_tr:n_offs_tr+to_write_len] = to_write_arr\n", "# n_offs_tr += to_write_len\n", "# dset_type = \"tr\"\n", "# out_mm.flush()\n", "# del audio_arr_list, to_write_arr\n", "# gc.collect();\n", "# td_write = int(round(time.time() - t0))\n", "# print(\n", "# f\"{chunk_duration_h:,} hours of data fetched in {td_fetch:,}s\"\n", "# f\" and written in {td_write:,}s as `{dset_type}`\"\n", "# )\n", "# dset_duration_h += chunk_duration_h\n", "# time.sleep(5) # make sure things close\n", "# print(f\"done. collected total of {dset_duration_h:,.1f} hours\")\n", "# # should be ~4h for 5k hours\n", "# # 270.6 hours of data fetched in 373s and written in 214s as `tr`" ] }, { "cell_type": "code", "execution_count": 25, "id": "cd111fdf", "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", "\n" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# # test output\n", "# mm = np.memmap(\"/mnt/data/georg/data/raw_audio/music_24khz_val.bin\", dtype=np.int16, mode=\"r\")\n", "# idx = random.randint(0, len(mm)-1-24_000*10)\n", "# Audio.from_array(mm[idx:idx+24_000*10], sample_rate=24_000).play()" ] }, { "cell_type": "code", "execution_count": 23, "id": "141f9ab4", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "3.9T\t/mnt/data/georg/data/raw_audio/music_24khz_tr.bin\r\n" ] } ], "source": [ "# !du -hs /mnt/data/georg/data/raw_audio/music_24khz_tr.bin" ] }, { "cell_type": "code", "execution_count": null, "id": "47041ed8", "metadata": {}, "outputs": [], "source": [ "# !aws s3 cp /mnt/data/georg/data/raw_audio/music_24khz_val.bin s3://suno-data/georg/data/raw_audio/\n", "# !aws s3 cp /mnt/data/georg/data/raw_audio/music_24khz_tr.bin s3://suno-data/georg/data/raw_audio/" ] }, { "cell_type": "code", "execution_count": null, "id": "3e67e882", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "12c2b806", "metadata": {}, "outputs": [], "source": [ "# node 0 (1x8)\n", "OMP_NUM_THREADS=6 NCCL_DEBUG=WARN NCCL_IB_SL=0 NCCL_IB_TC=41 NCCL_IB_QPS_PER_CONNECTION=4 UCX_TLS=ud,self,sm \\\n", "HCOLL_ENABLE_MCAST_ALL=0 NCCL_IB_GID_INDEX=3 UCX_NET_DEVICES=\"mlx5_0:1\" coll_hcoll_enable=0 \\\n", "NCCL_IB_HCA=\"=mlx5_5,mlx5_6,mlx5_7,mlx5_8,mlx5_1,mlx5_2,mlx5_3,mlx5_4,mlx5_14,mlx5_15,mlx5_16,mlx5_17,\\\n", "mlx5_9,mlx5_10,mlx5_11,mlx5_12\" \\\n", "python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz_8x_0' \\\n", " distributed_training.distributed_rank=0 \\\n", " distributed_training.distributed_world_size=8 \\\n", " distributed_training.nprocs_per_node=8 \\\n", " distributed_training.distributed_init_method=\"tcp://127.0.0.1:2345\" \\\n", " task.data='/app/suno/data/mert_25hz/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=true \\\n", " model.label_rate=25" ] }, { "cell_type": "code", "execution_count": null, "id": "c5de053c", "metadata": {}, "outputs": [], "source": [ "# node 1 (1x4x2)\n", "OMP_NUM_THREADS=6 NCCL_DEBUG=WARN NCCL_IB_SL=0 NCCL_IB_TC=41 NCCL_IB_QPS_PER_CONNECTION=4 UCX_TLS=ud,self,sm \\\n", "HCOLL_ENABLE_MCAST_ALL=0 NCCL_IB_GID_INDEX=3 UCX_NET_DEVICES=\"mlx5_0:1\" coll_hcoll_enable=0 \\\n", "NCCL_IB_HCA=\"=mlx5_5,mlx5_6,mlx5_7,mlx5_8,mlx5_1,mlx5_2,mlx5_3,mlx5_4,mlx5_14,mlx5_15,mlx5_16,mlx5_17,\\\n", "mlx5_9,mlx5_10,mlx5_11,mlx5_12\" \\\n", "python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz_8x_1' \\\n", " distributed_training.distributed_rank=0 \\\n", " distributed_training.distributed_world_size=4 \\\n", " distributed_training.nprocs_per_node=4 \\\n", " optimization.update_freq='[2]' \\\n", " distributed_training.distributed_init_method=\"tcp://127.0.0.1:2345\" \\\n", " task.data='/app/suno/data/mert_25hz/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=true \\\n", " model.label_rate=25" ] }, { "cell_type": "code", "execution_count": null, "id": "a88ea965", "metadata": {}, "outputs": [], "source": [ "# node 2 (1x8, +val)\n", "OMP_NUM_THREADS=6 NCCL_DEBUG=WARN NCCL_IB_SL=0 NCCL_IB_TC=41 NCCL_IB_QPS_PER_CONNECTION=4 UCX_TLS=ud,self,sm \\\n", "HCOLL_ENABLE_MCAST_ALL=0 NCCL_IB_GID_INDEX=3 UCX_NET_DEVICES=\"mlx5_0:1\" coll_hcoll_enable=0 \\\n", "NCCL_IB_HCA=\"=mlx5_5,mlx5_6,mlx5_7,mlx5_8,mlx5_1,mlx5_2,mlx5_3,mlx5_4,mlx5_14,mlx5_15,mlx5_16,mlx5_17,\\\n", "mlx5_9,mlx5_10,mlx5_11,mlx5_12\" \\\n", "python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz_8x_2' \\\n", " distributed_training.distributed_rank=0 \\\n", " distributed_training.distributed_world_size=8 \\\n", " distributed_training.nprocs_per_node=8 \\\n", " distributed_training.distributed_init_method=\"tcp://127.0.0.1:2345\" \\\n", " task.data='/app/suno/data/mert_25hz/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=false \\\n", " model.label_rate=25" ] }, { "cell_type": "code", "execution_count": null, "id": "9f33dc85", "metadata": {}, "outputs": [], "source": [ "# node 3 (1x8, previous style)\n", "OMP_NUM_THREADS=6 python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz_8x_3' \\\n", " distributed_training.distributed_rank=0 \\\n", " distributed_training.distributed_world_size=8 \\\n", " distributed_training.nprocs_per_node=8 \\\n", " distributed_training.distributed_init_method=\"tcp://127.0.0.1:39683\" \\\n", " task.data='/app/suno/data/mert_25hz/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=true \\\n", " model.label_rate=25" ] }, { "cell_type": "code", "execution_count": null, "id": "af2119b0", "metadata": {}, "outputs": [], "source": [ "# node 4&5 (2x8)\n", "OMP_NUM_THREADS=6 NCCL_DEBUG=WARN NCCL_IB_SL=0 NCCL_IB_TC=41 NCCL_IB_QPS_PER_CONNECTION=4 UCX_TLS=ud,self,sm \\\n", "HCOLL_ENABLE_MCAST_ALL=0 NCCL_IB_GID_INDEX=3 UCX_NET_DEVICES=\"mlx5_0:1\" coll_hcoll_enable=0 \\\n", "NCCL_IB_HCA=\"=mlx5_5,mlx5_6,mlx5_7,mlx5_8,mlx5_1,mlx5_2,mlx5_3,mlx5_4,mlx5_14,mlx5_15,mlx5_16,mlx5_17,\\\n", "mlx5_9,mlx5_10,mlx5_11,mlx5_12\" \\\n", "python -u /home/georg/code/fairseq/fairseq_cli/hydra_train.py \\\n", " --config-dir '/home/georg/code/MERT/mert_fairseq/config/pretrain' \\\n", " --config-name 'custom_95M' \\\n", " common.user_dir='/home/georg/code/MERT/mert_fairseq' \\\n", " common.wandb_project='mert_test' \\\n", " checkpoint.save_dir='/app/suno/checkpoints/mert_25hz_8x_4' \\\n", " distributed_training.distributed_rank=0 \\\n", " distributed_training.distributed_world_size=16 \\\n", " distributed_training.nprocs_per_node=8 \\\n", " distributed_training.distributed_init_method=\"tcp://compute-hpc-node-5:2345\" \\\n", " task.data='/app/suno/data/mert_25hz/audio_tsv' \\\n", " task.label_dir='/app/suno/data/mert_25hz/label' \\\n", " task.labels='[\"codec_0\",\"codec_1\",\"codec_2\",\"codec_3\",\"codec_4\",\"codec_5\",\"codec_6\",\"codec_7\"]' \\\n", " dataset.num_workers=6 \\\n", " dataset.max_tokens=1800000 \\\n", " dataset.disable_validation=true \\\n", " model.label_rate=25" ] }, { "cell_type": "code", "execution_count": null, "id": "5689c798", "metadata": {}, "outputs": [], "source": [ "# # rack 1\n", "# compute-hpc-node-192\n", "# compute-hpc-node-666\n", "# # rack 2\n", "# compute-hpc-node-537\n", "# compute-hpc-node-861\n", "# # rack 3\n", "# compute-hpc-node-5\n", "# # rack 4\n", "# compute-hpc-node-433\n", "# # rack 5\n", "# compute-hpc-node-699\n", "# # rack 6\n", "# compute-hpc-node-750" ] }, { "cell_type": "code", "execution_count": null, "id": "bb45a18a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "94672412", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "a9ab5ce0", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "aec26269", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.12" }, "toc": { "base_numbering": 1, "nav_menu": {}, "number_sections": true, "sideBar": true, "skip_h1_title": false, "title_cell": "Table of Contents", "title_sidebar": "Contents", "toc_cell": false, "toc_position": {}, "toc_section_display": true, "toc_window_display": false } }, "nbformat": 4, "nbformat_minor": 5 }