clr_callback_tests.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = '0'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CLR Callback"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from clr_scheduler import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai.basics import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Net(nn.Module):\n",
    "\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        \n",
    "        self.fc1 = nn.Linear(15, 10)\n",
    "        self.fc2 = nn.Linear(10, 1)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = torch.relu(self.fc1(x))\n",
    "        y_pred = torch.sigmoid(self.fc2(x))\n",
    "        return y_pred.squeeze()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_x = torch.rand(2000000, 15)\n",
    "valid_x = torch.rand(2000000, 15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_y = torch.randint(0,2, (2000000,)).float()\n",
    "valid_y = torch.randint(0,2, (2000000,)).float()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Net()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ArrayDataset(Dataset):\n",
    "    \"Sample numpy array dataset\"\n",
    "    def __init__(self, x, y):\n",
    "        self.x, self.y = x, y\n",
    "        self.c = 2 # binary label\n",
    "    \n",
    "    def __len__(self):\n",
    "        return len(self.x)\n",
    "    \n",
    "    def __getitem__(self, i):\n",
    "        return self.x[i], self.y[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_ds, valid_ds = ArrayDataset(train_x, train_y), ArrayDataset(valid_x, valid_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = DataBunch.create(train_ds, valid_ds, bs=2000, num_workers=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.BCELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = Learner(data, model, loss_func=criterion)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Callback test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "lr = 0.001\n",
    "learn.callbacks.append(CircularLRScheduler(learn, mode='triangular', base_lr=lr, max_lr=0.003, step_size=2000))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "learn.fit(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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": [
    "learn.recorder.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.693232</td>\n",
       "      <td>0.693222</td>\n",
       "      <td>01:36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.693181</td>\n",
       "      <td>0.693235</td>\n",
       "      <td>01:34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.693183</td>\n",
       "      <td>0.693167</td>\n",
       "      <td>01:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.693175</td>\n",
       "      <td>0.693163</td>\n",
       "      <td>01:36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.693152</td>\n",
       "      <td>0.693152</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.693181</td>\n",
       "      <td>0.693235</td>\n",
       "      <td>01:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.693156</td>\n",
       "      <td>0.693196</td>\n",
       "      <td>01:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.693163</td>\n",
       "      <td>0.693150</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.693155</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.693159</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.693155</td>\n",
       "      <td>0.693149</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.693158</td>\n",
       "      <td>0.693153</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.693155</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.693150</td>\n",
       "      <td>0.693153</td>\n",
       "      <td>01:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.693151</td>\n",
       "      <td>0.693150</td>\n",
       "      <td>01:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.693148</td>\n",
       "      <td>0.693152</td>\n",
       "      <td>01:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.693154</td>\n",
       "      <td>0.693150</td>\n",
       "      <td>01:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.693145</td>\n",
       "      <td>0.693156</td>\n",
       "      <td>01:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.693136</td>\n",
       "      <td>0.693171</td>\n",
       "      <td>01:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.693155</td>\n",
       "      <td>0.693163</td>\n",
       "      <td>01:29</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.callbacks = []\n",
    "learn.callbacks.append(CircularLRScheduler(learn, mode='triangular2', base_lr=0.001, max_lr=0.003, step_size=2000))\n",
    "\n",
    "learn.fit(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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": [
    "learn.recorder.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.callbacks = []\n",
    "learn.callbacks.append(CircularLRScheduler(learn, mode='exp_range', base_lr=0.001, max_lr=0.003, step_size=2000, gamma=0.99994))\n",
    "\n",
    "learn.fit(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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": [
    "learn.recorder.plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.693166</td>\n",
       "      <td>0.693149</td>\n",
       "      <td>01:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.693159</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.693155</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.693152</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.693154</td>\n",
       "      <td>0.693149</td>\n",
       "      <td>01:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.693159</td>\n",
       "      <td>0.693149</td>\n",
       "      <td>01:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.693155</td>\n",
       "      <td>0.693148</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.693150</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.693149</td>\n",
       "      <td>0.693150</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.693159</td>\n",
       "      <td>0.693147</td>\n",
       "      <td>01:33</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "clr_fn = lambda x: 1/(5**(x*0.0001))\n",
    "clr = CircularLRScheduler(learn,base_lr=0.001, max_lr=0.006,\n",
    "                                step_size=2000., scale_fn=clr_fn,\n",
    "                                scale_mode='iterations')\n",
    "learn.callbacks = []                    \n",
    "learn.fit(10, callbacks=[clr])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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": [
    "learn.recorder.plot_lr()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}