RainforestNN-Train.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import torchaudio\n",
    "import torchvision.models as vmodels\n",
    "import torchaudio.models as models\n",
    "import torchaudio.transforms as transforms\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from tqdm.notebook import tqdm\n",
    "import glob\n",
    "import librosa\n",
    "import librosa.display\n",
    "import soundfile as sf\n",
    "from scipy.interpolate import interp1d\n",
    "import IPython.display as ipd\n",
    "import ast\n",
    "import time\n",
    "import sys\n",
    "from nnutils import reshape_mel_spec, RainforestNetV2, Params, MelSpecDatasetV2\n",
    "from scipy import stats\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load meta.csv to save metadata for training session\n",
    "try:\n",
    "    meta = pd.read_csv('models/meta.csv')\n",
    "except:\n",
    "    meta = pd.DataFrame()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Hyper parameters      \n",
    "params = Params()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# set here if the model should be trained on the full dataset\n",
    "train_full = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "validate = True and not train_full\n",
    "train_split = 0.8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda:  True\n"
     ]
    }
   ],
   "source": [
    "# CUDA\n",
    "is_cuda = True and torch.cuda.is_available()\n",
    "cpu = torch.device('cpu')\n",
    "gpu = torch.device('cuda')\n",
    "device = gpu if is_cuda else cpu\n",
    "\n",
    "print ('cuda: ', is_cuda)\n",
    "\n",
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "288700029d0e435a8eda9befcba7d6c7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value=''), FloatProgress(value=0.0, max=905.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "train_test_split = None if train_full else (train_split, True)\n",
    "trainset = MelSpecDatasetV2(params, train_test_split=train_test_split)\n",
    "trainloader = torch.utils.data.DataLoader(trainset, batch_size=params.TRAIN_BATCH_SIZE,shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([0.0652, 0.0631, 0.0573, 0.0728, 0.0961, 0.0620, 0.0983, 0.0905, 0.0728,\n",
      "        0.0619, 0.0851, 0.0839, 0.0725, 0.0857, 1.0000, 0.0910, 0.0506, 0.0770,\n",
      "        0.0756, 0.0812, 0.0922, 0.0724, 0.0562, 0.0823])\n",
      "torch.Size([3, 224, 1153])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def imshow(img):\n",
    "    npimg = img.numpy()\n",
    "    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n",
    "    plt.show()\n",
    "\n",
    "ms, l = trainset[0]\n",
    "print (l)\n",
    "print (ms.shape)\n",
    "imshow(ms)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Downloading: \"https://download.pytorch.org/models/resnet34-333f7ec4.pth\" to C:\\Users\\Vasishta/.cache\\torch\\hub\\checkpoints\\resnet34-333f7ec4.pth\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cd42685f76934217ab21e767cabe782e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value=''), FloatProgress(value=0.0, max=87306240.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "model = RainforestNetV2().to(device)\n",
    "model.train()\n",
    "criterion = nn.BCELoss()\n",
    "optimizer = optim.Adam(model.parameters(), lr=params.LR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "27d1a034f44f40adbadde69cb06666d4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 1/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6cf67c3e1d5f4fe0937a64f2103aa2cb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 2/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "810f57715e1541aeaf61defef47f5ea2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 3/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6c4b4cacb8c944aa843460d38c1faab2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 4/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9df9433106ce4023b35dd7f6a80820b1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 5/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7484db4a280f4c8792e6f31bc28701ac",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 6/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5c0a3a4ae4804ea9abae2d057a613cc4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 7/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "327d7ef862834726840be4ec75976946",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 8/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d77def6f16914b9badec5dd86a86dfb6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 9/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value='')…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0bd9679cab2044799d805f50136a3abf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 10/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1e8d4eabee59443ca295bfbda3f970f6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 11/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c1b79ffde1f348b4b7ad4e0413e1908f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 12/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c4e159c9571c444f9a0673e2b4a0b158",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 13/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "accc19dbc17e4934b55d6a72df2502c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 14/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2855f52b2f5049a69b85ccac9d6b3c2b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 15/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ccfd59174ecb480eaab24bd538ca2110",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 16/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bb0013a7953c49cfa37e825319645fd5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 17/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "73c90f32b1434be0921054042d8b043c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 18/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "131c446991934ee4a40170b0e1f4ec63",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 19/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "95ff0cb2d96645668ed74b259b5e812b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 20/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9d7f328e8024490fa65f81a226ec963b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 21/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3f17daba7d7d40c7b4ddd72e0ce266bd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 22/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "50625e958271425c99b1eaae38f486b8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 23/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d38dadd31dcd44faa7a5af99ae666939",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 24/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d180c179737b46048dce3632c74a0a2f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 25/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f13190e5f0da43329645cbb042e97246",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 26/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "455c00bc759d40c19f976a059bd07ad9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 27/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2a0630016cea477ba35c083af4dcc0af",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 28/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a5f1ef6aa2ba4b048feda97e6ae62a06",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 29/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6b2bf5144c964de38d0bc819cf18dee7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value='epoch 30/30, running loss: 0.0'), FloatProgress(value=0.0, max=91.0), HTML(value=''…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "-------- DONE --------\n",
      "start time: 1608874633.949333\n",
      "end time: 1608875529.2274983\n",
      "\n",
      "training duration: 895.2781653404236\n"
     ]
    }
   ],
   "source": [
    "start_time = time.time()\n",
    "losses = []\n",
    "losses_detail = []\n",
    "\n",
    "for epoch in range(params.EPOCHS):\n",
    "    running_loss = 0.0\n",
    "    t = tqdm(trainloader, desc=\"epoch {}/{}, running loss: {}\".format(epoch+1, params.EPOCHS, 0.0))\n",
    "    for i, data in enumerate(t):\n",
    "        inputs, labels = data\n",
    "        inputs, labels = inputs.to(device), labels.to(device)\n",
    "\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        output = model(inputs)\n",
    "        loss = criterion(output, labels)\n",
    "\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        del inputs\n",
    "        del labels\n",
    "\n",
    "        torch.cuda.empty_cache()\n",
    "\n",
    "        running_loss += loss.item()\n",
    "        losses_detail.append(loss.item())\n",
    "        loss_interval = 10\n",
    "        if i % loss_interval == loss_interval-1:    # print every 50 mini-batches\n",
    "            #print('[%d %5d] loss: %.3f' % (epoch+1, i+1, running_loss / 50))\n",
    "            interval_loss = running_loss/loss_interval\n",
    "            losses.append(interval_loss)\n",
    "            t.set_description(\"epoch {}/{}, running_loss: {:.2f}\".format(epoch+1, params.EPOCHS, interval_loss))\n",
    "            t.refresh()\n",
    "\n",
    "            running_loss = 0.0\n",
    "\n",
    "end_time = time.time()\n",
    "print ('\\n-------- DONE --------')\n",
    "print ('start time: {}'.format(start_time))\n",
    "print ('end time: {}\\n'.format(end_time))\n",
    "\n",
    "duration = end_time-start_time\n",
    "print ('training duration: {}'.format(duration))\n",
    "\n",
    "model_path = 'models/rainforest{}.pt'.format(end_time)\n",
    "torch.save(model.state_dict(), model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.plot(losses)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e43d3766b2a04d39b3aec57d7b067513",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value=''), FloatProgress(value=0.0, max=227.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# validate\n",
    "\n",
    "if validate:\n",
    "    validset = MelSpecDatasetV2(params, train_test_split=(train_split, False))\n",
    "    validloader = torch.utils.data.DataLoader(validset, batch_size=1, shuffle=False)\n",
    "else:\n",
    "    mdata = params.__to_dict__()\n",
    "    mdata.update({\n",
    "        'start_time': start_time,\n",
    "        'end_time': end_time,\n",
    "        'train_duration': duration,\n",
    "        'model_path': model_path,\n",
    "    })\n",
    "    meta = meta.append(mdata, ignore_index=True)\n",
    "    meta.to_csv(\"models/meta.csv\", index=False, float_format='%.5f')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b1e15e6256ec4ac487b543953867491f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value=''), FloatProgress(value=0.0, max=227.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "if validate:    \n",
    "    valid_model = RainforestNetV2()\n",
    "    valid_model.load_state_dict(torch.load(model_path))\n",
    "    valid_model.eval()\n",
    "    valid_model = valid_model.to(device)\n",
    "    \n",
    "    valid_results = pd.DataFrame()\n",
    "    with torch.no_grad():\n",
    "        for i, data in enumerate(tqdm(validloader)):\n",
    "            inputs, labels = data\n",
    "            inputs, labels = inputs.to(device), labels.to(device)\n",
    "\n",
    "            outputs = valid_model(inputs)\n",
    "            res = outputs[0].to(cpu).numpy()\n",
    "\n",
    "            valid_results = valid_results.append({ 'result': res }, ignore_index=True)\n",
    "    \n",
    "    submission = pd.DataFrame()\n",
    "    \n",
    "    valid_ids = validset.get_recording_ids()\n",
    "\n",
    "    for idx in range(len(valid_results)):\n",
    "        ex = valid_results.iloc[idx]\n",
    "        recording_id = valid_ids[idx]\n",
    "        result = ex.result\n",
    "\n",
    "        species = 0\n",
    "        o = {}\n",
    "        for r in result:\n",
    "            s = 's{}'.format(species)\n",
    "            o[s] = r\n",
    "            species += 1 \n",
    "        o[\"recording_id\"] = recording_id\n",
    "        submission = submission.append(o, ignore_index=True)\n",
    "    \n",
    "    del valid_results\n",
    "    cols = ['recording_id', 's0', 's1', 's2', 's3', 's4', 's5', 's6', 's7', 's8', 's9', 's10', 's11', 's12', 's13', 's14', 's15', 's16', 's17', 's18', 's19', 's20', 's21', 's22', 's23']\n",
    "    submission = submission[cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.4434143210436392, 0.44513283895039407)\n"
     ]
    }
   ],
   "source": [
    "if validate:\n",
    "    from utils import score\n",
    "    s = score(submission)\n",
    "    print(s)\n",
    "    mdata = params.__to_dict__()\n",
    "    mdata.update({\n",
    "        'start_time': start_time,\n",
    "        'end_time': end_time,\n",
    "        'train_duration': duration,\n",
    "        'model_path': model_path,\n",
    "        'validation_score': s\n",
    "    })\n",
    "    meta = meta.append(mdata, ignore_index=True)\n",
    "    meta.to_csv(\"models/meta.csv\", index=False, float_format='%.5f')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "rainforest_proj",
   "language": "python",
   "name": "rainforest_proj"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}