06_callback.training.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#default_exp callback.training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training Callbacks\n",
    "> Callbacks to help during training, including `fit_one_cycle`, the LR Finder, and hyper-parameter scheduling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "# Contains code used/modified by fastai_minima author from fastai\n",
    "# Copyright 2019 the fast.ai team.\n",
    "#\n",
    "# Licensed under the Apache License, Version 2.0 (the \"License\");\n",
    "# you may not use this file except in compliance with the License.\n",
    "# You may obtain a copy of the License at\n",
    "#\n",
    "#     http://www.apache.org/licenses/LICENSE-2.0\n",
    "#\n",
    "# Unless required by applicable law or agreed to in writing, software\n",
    "# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
    "# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
    "# See the License for the specific language"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hide\n",
    "from nbdev.showdoc import *\n",
    "from fastcore.test import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "from fastcore.basics import store_attr, patch\n",
    "from fastcore.foundation import L\n",
    "from fastcore.meta import delegates\n",
    "from fastcore.xtras import is_listy\n",
    "\n",
    "from fastai_minima.optimizer import convert_params\n",
    "\n",
    "from fastai_minima.callback.core import Callback, CancelValidException, CancelFitException\n",
    "from fastai_minima.learner import Learner, Recorder\n",
    "from fastai_minima.utils import tensor, defaults, params\n",
    "\n",
    "import functools, math, collections\n",
    "\n",
    "import os\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "import torch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Annealing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "class _Annealer:\n",
    "    def __init__(self, f, start, end): store_attr('f,start,end')\n",
    "    def __call__(self, pos): return self.f(self.start, self.end, pos)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "def annealer(f):\n",
    "    \"Decorator to make `f` return itself partially applied.\"\n",
    "    @functools.wraps(f)\n",
    "    def _inner(start, end): return _Annealer(f, start, end)\n",
    "    return _inner"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is the decorator we will use for all of our scheduling functions, as it transforms a function taking `(start, end, pos)` to something taking `(start, end)` and return a function depending of `pos`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "#TODO Jeremy, make this pickle\n",
    "#@annealer\n",
    "#def SchedLin(start, end, pos): return start + pos*(end-start)\n",
    "#@annealer\n",
    "#def SchedCos(start, end, pos): return start + (1 + math.cos(math.pi*(1-pos))) * (end-start) / 2\n",
    "#@annealer\n",
    "#def SchedNo (start, end, pos): return start\n",
    "#@annealer\n",
    "#def SchedExp(start, end, pos): return start * (end/start) ** pos\n",
    "#\n",
    "#SchedLin.__doc__ = \"Linear schedule function from `start` to `end`\"\n",
    "#SchedCos.__doc__ = \"Cosine schedule function from `start` to `end`\"\n",
    "#SchedNo .__doc__ = \"Constant schedule function with `start` value\"\n",
    "#SchedExp.__doc__ = \"Exponential schedule function from `start` to `end`\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "def sched_lin(start, end, pos): return start + pos*(end-start)\n",
    "def sched_cos(start, end, pos): return start + (1 + math.cos(math.pi*(1-pos))) * (end-start) / 2\n",
    "def sched_no (start, end, pos): return start\n",
    "def sched_exp(start, end, pos): return start * (end/start) ** pos\n",
    "\n",
    "def SchedLin(start, end): \n",
    "    \"Linear schedule function from `start` to `end`\"\n",
    "    return _Annealer(sched_lin, start, end)\n",
    "\n",
    "def SchedCos(start, end): \n",
    "    \"Cosine schedule function from `start` to `end`\"\n",
    "    return _Annealer(sched_cos, start, end)\n",
    "\n",
    "def SchedNo (start, end): \n",
    "    \"Constant schedule function with `start` value\"\n",
    "    \n",
    "    return _Annealer(sched_no,  start, end)\n",
    "def SchedExp(start, end): \n",
    "    \"Exponential schedule function from `start` to `end`\"\n",
    "    return _Annealer(sched_exp, start, end)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hide\n",
    "import pickle\n",
    "tst = pickle.dumps(SchedCos(0, 5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "annealings = \"NO LINEAR COS EXP\".split()\n",
    "p = torch.linspace(0.,1,100)\n",
    "fns = [SchedNo, SchedLin, SchedCos, SchedExp]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "def SchedPoly(start, end, power):\n",
    "    \"Polynomial schedule (of `power`) function from `start` to `end`\"\n",
    "    def _inner(pos): return start + (end - start) * pos ** power\n",
    "    return _inner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": [
    "for fn, t in zip(fns, annealings):\n",
    "    plt.plot(p, [fn(2, 1e-2)(o) for o in p], label=t)\n",
    "f = SchedPoly(2,1e-2,0.5)\n",
    "plt.plot(p, [f(o) for o in p], label=\"POLY(0.5)\")\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"SchedLin\" class=\"doc_header\"><code>SchedLin</code><a href=\"__main__.py#L7\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>SchedLin</code>(**`start`**, **`end`**)\n",
       "\n",
       "Linear schedule function from `start` to `end`"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(SchedLin)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sched = SchedLin(0, 2)\n",
    "test_eq(L(map(sched, [0., 0.25, 0.5, 0.75, 1.])), [0., 0.5, 1., 1.5, 2.])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"SchedCos\" class=\"doc_header\"><code>SchedCos</code><a href=\"__main__.py#L11\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>SchedCos</code>(**`start`**, **`end`**)\n",
       "\n",
       "Cosine schedule function from `start` to `end`"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(SchedCos)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sched = SchedCos(0, 2)\n",
    "test_close(L(map(sched, [0., 0.25, 0.5, 0.75, 1.])), [0., 0.29289, 1., 1.70711, 2.])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"SchedNo\" class=\"doc_header\"><code>SchedNo</code><a href=\"__main__.py#L15\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>SchedNo</code>(**`start`**, **`end`**)\n",
       "\n",
       "Constant schedule function with `start` value"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(SchedNo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sched = SchedNo(0, 2)\n",
    "test_close(L(map(sched, [0., 0.25, 0.5, 0.75, 1.])), [0., 0., 0., 0., 0.])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"SchedExp\" class=\"doc_header\"><code>SchedExp</code><a href=\"__main__.py#L19\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>SchedExp</code>(**`start`**, **`end`**)\n",
       "\n",
       "Exponential schedule function from `start` to `end`"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(SchedExp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sched = SchedExp(1, 2)\n",
    "test_close(L(map(sched, [0., 0.25, 0.5, 0.75, 1.])), [1., 1.18921, 1.41421, 1.68179, 2.])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"SchedPoly\" class=\"doc_header\"><code>SchedPoly</code><a href=\"__main__.py#L2\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>SchedPoly</code>(**`start`**, **`end`**, **`power`**)\n",
       "\n",
       "Polynomial schedule (of `power`) function from `start` to `end`"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(SchedPoly)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sched = SchedPoly(0, 2, 2)\n",
    "test_close(L(map(sched, [0., 0.25, 0.5, 0.75, 1.])), [0., 0.125, 0.5, 1.125, 2.])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": [
    "p = torch.linspace(0.,1,100)\n",
    "\n",
    "pows = [0.5,1.,2.]\n",
    "for e in pows:\n",
    "    f = SchedPoly(2, 0, e)\n",
    "    plt.plot(p, [f(o) for o in p], label=f'power {e}')\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "def combine_scheds(pcts, scheds):\n",
    "    \"Combine `scheds` according to `pcts` in one function\"\n",
    "    assert sum(pcts) == 1.\n",
    "    pcts = tensor([0] + L(pcts))\n",
    "    assert torch.all(pcts >= 0)\n",
    "    pcts = torch.cumsum(pcts, 0)\n",
    "    def _inner(pos):\n",
    "        if int(pos) == 1: return scheds[-1](1.)\n",
    "        idx = (pos >= pcts).nonzero().max()\n",
    "        actual_pos = (pos-pcts[idx]) / (pcts[idx+1]-pcts[idx])\n",
    "        return scheds[idx](actual_pos.item())\n",
    "    return _inner"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`pcts` must be a list of positive numbers that add up to 1 and is the same length as `scheds`. The generated function will use `scheds[0]` from 0 to `pcts[0]` then `scheds[1]` from `pcts[0]` to `pcts[0]+pcts[1]` and so forth."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/mnt/d/lib/python3.7/site-packages/ipykernel_launcher.py:10: UserWarning: This overload of nonzero is deprecated:\n",
      "\tnonzero()\n",
      "Consider using one of the following signatures instead:\n",
      "\tnonzero(*, bool as_tuple) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:882.)\n",
      "  # Remove the CWD from sys.path while we load stuff.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3Xd8lfX5//HXlZOdEEYSVgYBCULYEJbbgooLUByAuFtK1To7bKu2xfptq3VWHNjiVkBaFUfFgSgKSBLZO0AIIYywEiBkX78/cvR3xGgOcJL7jOv5eORhzn3uO3nfgO/cucfnI6qKMcaY0BDmdABjjDHNx0rfGGNCiJW+McaEECt9Y4wJIVb6xhgTQqz0jTEmhFjpG2NMCLHSN8aYEGKlb4wxISTc6QBHS0pK0oyMDKdjGGNMQMnLy9ujqsmNred3pZ+RkUFubq7TMYwxJqCIyFZv1rPTO8YYE0Ks9I0xJoRY6RtjTAix0jfGmBBipW+MMSHEq9IXkZEisl5E8kXk7h9Y5woRWSMiq0XkNY/l14rIRvfHtb4Kbowx5tg1esumiLiAqcA5QBGQIyJzVHWNxzqZwO+AU1V1v4i0dS9vA/wRyAYUyHNvu9/3u2KMMaYx3tynPxjIV9XNACIyAxgNrPFY52fA1G/KXFV3u5efB3ykqvvc234EjARe901805xq65SvC/dTtL+csiM1lB2pJixMiIt0ER8dQWJcJCmtY+jYKob4KL97BMQYg3elnwJs83hdBAw5ap1uACLyJeAC/qSqH/zAtilHfwMRmQRMAkhPT/c2u2kmCzftYc6yYj5as4u9h6u82iYpPpIeHRLo0SGB3iktGdK5DW0Tops4qTGmMd6UvjSw7OjZ1MOBTOAsIBVYICK9vNwWVZ0GTAPIzs62mdr9xOaSQ9z/7ho+XV9CXKSLn/Rox8ie7cnqmEBCdDgtoiNQlMOVtRyqqKHkUCXFB46w/cARNpccYs2OMl5YWEBVTR0AXZLjOPWkJM7JasfQLolEhtt9BMY0N29KvwhI83idChQ3sM5iVa0GtojIeup/CBRR/4PAc9v5xxvWNI/q2joe/WgDzy3YTFS4i3su7MHEoZ2IjnA1uH5UuIs2cZGkJ8YysFPr732ttTvKWLx5L4s27WV2XhEvL95Ki+hwRvRox9gBqZxyUiJhYQ0dHxhjfE1Uf/zAWkTCgQ3AcGA7kANMUNXVHuuMBMar6rUikgQsBfrhvngLDHCv+jUw8Jtz/A3Jzs5WG3vHOaXl1fzi1TwWbtrL2AGp/Pb8k2nbwnenZSqqa/li4x4+XLOTD1btpKyihpRWMYwdmMrEIel2CsiY4yQieaqa3eh6jZW++4tdADxG/fn66ar6gIhMAXJVdY6ICPAw9Rdpa4EHVHWGe9sbgN+7v9QDqvr8j30vK33nFOw5zA0v5rBtXzl/vbQPlw1MbdLvV1Fdy0drdvFGXhELNpYQHiZc3KcjN5zWmV4pLZv0exsTbHxa+s3JSt8Z63aWMX7aYhR4duJAhnRJbNbvX7DnMC8sLGBW7jbKq2oZ3r0tt4/oRu9UK39jvGGlb7y2bV85Y59eiAjMmDSMzklxjmUpPVLNy4sKeG7BFkqPVHNOVjt+O/JkurZt4VgmYwKBlb7xyp5DlVz+zCL2HqrkjcmncHJ7/yjXsopqXviygOc+30x5dS1XDUnn9hHdaBMX6XQ0Y/ySt6Vv98yFsPKqGq5/PocdpUeYft0gvyl8gIToCG4dnsn8X5/F+MFpvLJ4K2c+9CkvLyqgts6/DlSMCSRW+iFKVbnnzVWsKi5l6oQBZGe0cTpSgxLjo/jLmN58cPsZ9Eltyb1vr2bs0wtZu6PM6WjGBCQr/RD1Rl4R/126nduGZzK8Rzun4zSqW7sWvHLjEB69si+F+8q56J9f8MiH67998MsY4x0r/RC0fudB7nt7FaeclMgvf5LpdByviQiX9E/lkzvPZHS/jjwxL59LnvqSDbsOOh3NmIBhpR9iyqtquOnVPOKjInhsXD9cAfgkbOu4SB65oh/PXj2QnaUVXPTPL3j+yy34200JxvgjK/0Q8+AH69m85zBPjOvn0ydtnXBez/bMveMMzshM4s/vrOHnL+dxoNy7AeGMCVVW+iEkb+t+XlxUwNVDO3FK1ySn4/hEUnwUz12Tzb0XZfHp+t1c+MQXLC206RqM+SFW+iGiqqaOu/+zgg4J0fxmZHen4/iUiHDjaZ2ZPfkUwsLgymcXMzOn0OlYxvglK/0Q8dT8fDbuPsQDl/QO2glO+qa14p1bTmNIlzb89j8r+cObK+3uHmOOYqUfAjbuOsjUT/MZ3a8jZ3dv63ScJtUqNpIXrh/M5DNP4tWvCpn4r6/Y7+XEL8aEAiv9IKeqTHl3DTERLu67KMvpOM3CFSbcfX53nhjfn2VFB7jkqS/ZXHLI6VjG+AUr/SA3f30JCzbu4fYR3UiMj3I6TrMa1bcjr/9sKAcrarjkqYUs3rzX6UjGOM5KP4hV19Zx/3tr6JIUx9XDOjkdxxEDO7XmrZtPJblFFNf8ewnvr9zhdCRjHOVV6YvISBFZLyL5InJ3A+9fJyIlIrLM/fFTj/dqPZbP8WV48+NeXbyVzSWH+f0FPYhwhe7P97Q2scyePIzeqS25+bWveXnxVqcjGeOYRm/jEBEXMBU4h/o5b3NEZI6qrjlq1ZmqeksDX+KIqvY78ajmWBwor+LRjzdyWtckhvcI7ou33mgVG8krNw7hlte+5t63VrHnYCW3j8ikftI3Y0KHN4d/g4F8Vd2sqlXADGB008YyJ+qp+Zs4WFHNPRf1sGJzi4l08ezVA7lsYCqPf7KR/3t/rQ3dYEKON6WfAmzzeF3kXna0sSKyQkRmi0iax/JoEckVkcUiMuZEwhrv7C6r4MWFBYzpn0L39glOx/Er4a4wHhzbh2uHdeK5BVu4561V1Nn4/CaEePOUTkOHiUf/X/IO8LqqVorIZOBF4Cfu99JVtVhEugDzRGSlqm76zjcQmQRMAkhPTz+mHTDfN/XTfGrqlNuGB84Ims0pLEz406iexESG88xnm6ioruPBy/oE5OBzxhwrb470iwDPI/dUoNhzBVXdq6qV7pfPAQM93it2/3czMB/of/Q3UNVpqpqtqtnJycnHtAPmu7YfOMLrS7ZxRXYqnRKdm+vW34kIvx15MneM6MZ/vi7iN7NX2IxcJiR4c6SfA2SKSGdgOzAOmOC5goh0UNVv7oUbBax1L28NlLt/A0gCTgUe9FV4831PztsIwC0BNE6+U0SE20bU/zk9+vEGRODBsX0IsyN+E8QaLX1VrRGRW4C5gAuYrqqrRWQKkKuqc4BbRWQUUAPsA65zb94DeFZE6qj/reJvDdz1Y3xk697DzMot4uqhnUhpFeN0nIBx24hM6lR5/JONhAn87VIrfhO8vBp5S1XfB94/atl9Hp//DvhdA9stBHqfYEbjpSfn5RMeJtx01klORwk4d5zTDVXliXn5xES4+NOonnbXkwlKwTncYgjafuAIby7dzsShnWibENiTozjljnO6UV5Vy7++2EJsVDi/DbIhqI0BK/2g8dznmwH42RldHE4SuESEP1zYg8NVtTw9fxPxUeHcfHZXp2MZ41NW+kFg76FKZuQUMqZ/ip3LP0Eiwl/G9OJIVQ0PzV1Py5gIJg4NzXGLTHCy0g8Cz39ZQGVNHZPPtHP5vuAKEx66vC+lR6q59+1VJMZFcn7vDk7HMsYnQncUriBxsKKaFxcVMLJne7q2jXc6TtCIcIXx1FUD6Z/WittmLGPRJhuW2QQHK/0A98riQg5W1HDTWXbu2ddiIl1Mv24QnRJj+dlLuazdUeZ0JGNOmJV+AKuqqeOFhVs4rWsSvVNbOh0nKLWKjeSlGwcTHxXO9c/nsKP0iNORjDkhVvoB7L2Vxewqq+TG0zs7HSWodWgZw/PXD+JwZQ3XTc+hrKLa6UjGHDcr/QClqvxrwRa6to3nzEwbr6ip9eiQwDNXD2RTySEmv5xHVU2d05GMOS5W+gHqqy37WF1cxo2ndbYhA5rJqV2T+PvYPizctJd73lppY/GbgGS3bAaofy3YQpu4SC7p39DUBqapjB2YSsHew/xzXj5dkuPtNlkTcOxIPwBtLjnEJ+t2MXFoJ6IjXE7HCTl3jOjGhX068PcP1vHBqp1OxzHmmFjpB6AXFhYQERbG1fakqCPCwoSHL+9L39RW3D5zKau2lzodyRivWekHmLKKambnFXFx344kt4hyOk7Iio5w8dw12bSJjWTSS7mUHKxsfCNj/ICVfoD5T14R5VW1XHdKhtNRQl5yiyimXZPNvvIqJr+SR2VNrdORjGmUlX4AqatTXl60lf7prexhLD/RK6UlD1/ej7yt+7nnzVV2R4/xe16VvoiMFJH1IpIvInc38P51IlIiIsvcHz/1eO9aEdno/rjWl+FDzYL8PWzec9iO8v3MhX06cOvwTN7IK+LFhQVOxzHmRzV6y6aIuICpwDnUT5KeIyJzGpj2cKaq3nLUtm2APwLZgAJ57m33+yR9iHlpYQFJ8VGc38tGfPQ3tw/PZO2OMu5/by3dOyQwtEui05GMaZA3R/qDgXxV3ayqVcAMYLSXX/884CNV3ecu+o+AkccXNbQV7i1n3vrdTBiSTmS4nZXzN2FhwiNX9KVTYiw3v/o1xQdsjB7jn7xpjxRgm8frIveyo40VkRUiMltE0o5xW9OIlxcX4BLhqiHpTkcxP6BFdATTrs6un9vglTwqqu3CrvE/3pR+Q8/4H3216h0gQ1X7AB8DLx7DtojIJBHJFZHckpISLyKFlorqWmblFnFez/a0s/lv/VrXtvE8emU/VhSVct/bq5yOY8z3eFP6RUCax+tUoNhzBVXdq6rf3Kj8HDDQ223d209T1WxVzU5OtsHDjvbeih2UHqnmqqF2lB8Izslqxy1nd2VWbhEzlhQ6HceY7/Cm9HOATBHpLCKRwDhgjucKIuJ5ZXEUsNb9+VzgXBFpLSKtgXPdy8wxeOWrrXRJjmOYXRwMGHec043TM5O4b85qVhbZE7vGfzRa+qpaA9xCfVmvBWap6moRmSIio9yr3Soiq0VkOXArcJ17233A/dT/4MgBpriXGS+tLi5laeEBrhrSCREbTTNQuMKEx8f1Jzk+ismv5LH/cJXTkYwBQPztYZLs7GzNzc11Oobf+P2bK/lPXhFLfj+ClrERTscxx2j5tgNc/swiTumayPRrB9kw2KbJiEieqmY3tp7d++fHDlXW8PbS7Vzct6MVfoDqm9aKey/OYv76Ep7+bJPTcYyx0vdnby7dzuGqWibaaJoBbeKQdC7u25GHP1zPwk17nI5jQpyVvp9SVV5dvJWeHRPoa+PsBDQR4a+X9iYjKY5bX1/G7rIKpyOZEGal76eWbTvAup0HmTAk3S7gBoH4qHCevmoghyqruW3GMmrr/OtamgkdVvp+6vUlhcRGuhjVt6PTUYyPnNy+BVNG92LR5r38c95Gp+OYEGWl74fKKqp5Z/kORvXtSItou4AbTC4fmMql/VN4/JONdn7fOMJK3w+9vayYI9W1jB9sT+AGGxHh/jG96JIUx20zltmMW6bZWen7GVXlta8K6dkxgT52ATcoxUWFM/WqAZQdqebOWcuos/P7phlZ6fuZFUWlrN1RxvjBdgE3mHVvn8B9F2exYOMepi3Y7HQcE0Ks9P3M60sKiYlwMbqfXcANdhMGp3NB7/b8Y+56lhbavEKmeVjp+5FDlTXMWV7MxX072AXcEFB//34f2iVEc+uMpZRVVDsdyYQAK30/8s7yYsqrahlnF3BDRsuYCJ4Y35/iAxX8/r8rbWJ10+Ss9P3IjCWFnNyuBf3TWjkdxTSjgZ1ac+c53Xh3xQ5m5xU5HccEOSt9P7GmuIzlRaWMG5xmF3BD0OQzT2Jolzb8cc5qtuw57HQcE8Ss9P3EzJxCIsPDuKS/TSEcilxhwqNX9iPCFcatry+lqqbO6UgmSFnp+4GK6lreXLqd83u1p1VspNNxjEM6tIzh72P7sHJ7KQ9/uN7pOCZIeVX6IjJSRNaLSL6I3P0j610mIioi2e7XGSJyRESWuT+e8VXwYPL+yh2UVdRw5aC0xlc2QW1kr/aMH5zOtAWbWZhvwzQY32u09EXEBUwFzgeygPEiktXAei2onyrxq6Pe2qSq/dwfk32QOejMyNlGRmKszYFrALj3oh50TorjzlnLOVBu0ywa3/LmSH8wkK+qm1W1CpgBjG5gvfuBBwEbLPwYbCo5xJIt+7hikF3ANfViI8N5Ylx/9h6u5Hd2G6fxMW9KPwXY5vG6yL3sWyLSH0hT1Xcb2L6ziCwVkc9E5PSGvoGITBKRXBHJLSkp8TZ7UJiVs43wMOGygalORzF+pFdKS+4692T+t2onb9htnMaHvCn9hg4/vz30EJEw4FHgrgbW2wGkq2p/4E7gNRFJ+N4XU52mqtmqmp2cnOxd8iBQVVPH7LwihvdoS9sW0U7HMX5m0uldGNYlkT/PWc3WvXYbp/ENb0q/CPC8wpgKFHu8bgH0AuaLSAEwFJgjItmqWqmqewFUNQ/YBHTzRfBg8MnaXew9XMW4QfYErvm+sDDh4Sv64goTbp+5jJpau43TnDhvSj8HyBSRziISCYwD5nzzpqqWqmqSqmaoagawGBilqrkikuy+EIyIdAEyARtS0O31nG10aBnNGd1C57cbc2w6torhgUt6s7TwAFM/3eR0HBMEGi19Va0BbgHmAmuBWaq6WkSmiMioRjY/A1ghIsuB2cBkVd13oqGDQdH+chZsLOHy7DRcYXYB1/ywi/t2ZEy/jjwxb6ONxmlOmPjbnQHZ2dmam5vrdIwm98hHG/jnvI0s+M3ZpLaOdTqO8XOlR6q54PEFRLiE9249nbiocKcjGT8jInmqmt3YevZErgNq65Q3crdxemayFb7xSsuYCB6+oi9b95XzwPtrnY5jApiVvgM+31DCjtIKxtsTuOYYDO2SyKTTu/DaV4XMW7fL6TgmQFnpO+C1JYUkxUcyvEc7p6OYAHPnud3o3r4Fv5m9kr2HbFJ1c+ys9JvZrrIK5q3bzWUD04gMtz9+c2yiwl08Nq4fZUeq7Wldc1ysdZrZG7nbqK1TxtmpHXOcurdP4NfnncyHa3bZ07rmmFnpN6O6OmVGzjaGdUkkIynO6TgmgN14WmeGdG7DlHfWsG1fudNxTACx0m9GX+TvoWj/EcYPsSdwzYn55mldgLveWE5tnZ3mMd6x0m9GM3IKaR0bwXk97QKuOXGprWP506ieLNmyj39/YQ+6G+9Y6TeTkoOVfLh6F2MHpBIV7nI6jgkSYwekMLJne/4xdwPrdpY5HccEACv9ZvJG3jZq6pRxg+3UjvEdEeGBS3qREBPOHTOX29y6plFW+s2grk55fUkhQ7u0oWvbeKfjmCCTGB/FXy/tw9odZTz28Qan4xg/Z6XfDBbk72HbviNMGNLJ6SgmSJ2T1Y4rslN55rNN5G21MQ3ND7PSbwavLt5KYlykXcA1Terei7Lo0DKGO2ct53BljdNxjJ+y0m9iO0sr+GTdbi7PTrMLuKZJtYiuH5StcF85f/2fDcpmGmal38Rm5tQ/gTt+sD2Ba5re0C6J3HhqZ15ZXMhnG0JrvmnjHa9KX0RGish6EckXkbt/ZL3LRERFJNtj2e/c260XkfN8ETpQ1NTWMSOnkNMzk+iUaE/gmubxq/NOJrNtPL+ZvZzS8mqn4xg/02jpu6c7nAqcD2QB40Ukq4H1WgC3Al95LMuifnrFnsBI4Klvpk8MBfPW7WZHaQVX2RO4phlFR7h45Ip+7D1UxX1zVjkdx/gZb470BwP5qrpZVauAGcDoBta7H3gQqPBYNhqY4Z4gfQuQ7/56IeHlxVvp0DKaETaEsmlmvVNb8sufZPL2smLeXVHsdBzjR7wp/RRgm8frIveyb4lIfyBNVd891m2D1eaSQyzYuIcJg9MJd9mlE9P8bjr7JPqmtuSet1axu6yi8Q1MSPCmjRqatfvb0Z1EJAx4FLjrWLf1+BqTRCRXRHJLSoLj4tPLi7cS4RJ7Atc4JsIVxsNX9ONIVS1329j7xs2b0i8CPG89SQU8f19sAfQC5otIATAUmOO+mNvYtgCo6jRVzVbV7OTk5GPbAz9UXlXD7Lwizu/VgeQWUU7HMSGsa9t47j6/O/PW7WZmzrbGNzBBz5vSzwEyRaSziERSf2F2zjdvqmqpqiapaoaqZgCLgVGqmuteb5yIRIlIZyATWOLzvfAzby0t5mBFDdcMsydwjfOuHZbBsC6J3P+ujb1vvCh9Va0BbgHmAmuBWaq6WkSmiMioRrZdDcwC1gAfADerau2Jx/ZfqspLiwro0SGBgZ1aOx3HGMLChH9c0ZcwEe6aZWPvhzqvrjCq6vuq2k1VT1LVB9zL7lPVOQ2se5b7KP+b1w+4tztZVf/nu+j+KadgP+t2HuSaYZ0QaeiShjHNL6VVDH8c1ZMlBfuY/sUWp+MYB9ltJT72/JdbaBkTweh+HZ2OYsx3jB2QwrlZ7Xho7nrW7zzodBzjECt9H9q2r5y5q3cyYUg6sZHhTscx5jtEhP+7tDctosO5c9YyG3s/RFnp+9CLCwsQEbuAa/xWUnwUf720N6uLy3jik41OxzEOsNL3kUOVNczM2cYFvTvQoWWM03GM+UHn9mzPZQNTeWp+Pl8X7nc6jmlmVvo+Mjt3Gwcra7jh1AynoxjTqD9e7B57f+Yyyqts7P1QYqXvA3V1yvMLCxiQ3or+6XabpvF/LaIj+Mflfdm6r5y/vr/O6TimGVnp+8An63azdW85N5zW2ekoxnht2En1Y++/vHirjb0fQqz0feCZzzaR0iqGkT3bOx3FmGPyzdj7v35jOQfKq5yOY5qBlf4JyinYR97W/fzs9M42mqYJONERLh69sh/7y6v4w1urbFC2EGAtdYKe/WwTrWMjuGKQTYdoAlOvlJbcPqIb763YwZzlNvZ+sLPSPwEbdh3k47W7ufaUDHsYywS0n5/RhQHprbj3rVUUHzjidBzThKz0T8Czn20mJsLFtcMynI5izAkJd4Xx6JX9qKlTfvXGcupsULagZaV/nIoPHOHtZdu5clAareMinY5jzAnrlBjHvRdlsXDTXp5fWOB0HNNErPSP07TPNwPw09PtNk0TPMYNSmNEj3b8/YN1NihbkLLSPw47Syt4bUkhlw1MJbV1rNNxjPEZEeFvY3uTEB3O7TOXUVkT1NNfhCQr/ePw9Px86uqUm8/u6nQUY3wuKT6Kv4/tw9odZTzy0Qan4xgf86r0RWSkiKwXkXwRubuB9yeLyEoRWSYiX4hIlnt5hogccS9fJiLP+HoHmtvO0gpeX7KNy7NTSWtjR/kmOA3v0Y4JQ9KZ9vlmFm3a63Qc40ONlr6IuICpwPlAFjD+m1L38Jqq9lbVfsCDwCMe721S1X7uj8m+Cu6Up+bnU6fKTWfZUb4Jbvdc2IOMxDjunLWM0vJqp+MYH/HmSH8wkK+qm1W1CpgBjPZcQVXLPF7GAUF5v9eO0iPMsKN8EyJiI8N57Mp+lBys5J637WndYOFN6acA2zxeF7mXfYeI3Cwim6g/0r/V463OIrJURD4TkdNPKK3DnvgkH8XO5ZvQ0TetFbePyOSd5cW8tWy703GMD3hT+g3N7v29H/mqOlVVTwJ+C9zjXrwDSFfV/sCdwGsikvC9byAySURyRSS3pMQ/R/vbuOsgM3MKuWpIJ7tjx4SUX5zVlexOrbn3rdVs21fudBxzgrwp/SLAc2CZVODHBuiYAYwBUNVKVd3r/jwP2AR0O3oDVZ2mqtmqmp2cnOxt9mb1t/+tIy4ynFuHZzodxZhm5QoTHr2yHwLcNmMpNbU2t24g86b0c4BMEeksIpHAOGCO5woi4tmEFwIb3cuT3ReCEZEuQCaw2RfBm9OiTXv5ZN1ubjq7K23s6VsTgtLaxPKXS3rxdeEB/jkv3+k45gQ0OkqYqtaIyC3AXMAFTFfV1SIyBchV1TnALSIyAqgG9gPXujc/A5giIjVALTBZVfc1xY40lbo65f/eX0vHltFcb1MhmhA2ul8K89eX8M95Gzk9M4nsjDZORzLHQfztinx2drbm5uY6HeNbby/bzm0zlvHw5X0ZOzDV6TjGOOpgRTUXPLGAujp4/7bTaRkT4XQk4yYieaqa3dh69kTujyirqOaB99bSKyWBS/p/74YlY0JOi+gIHh/Xn51lFfzhzZV2G2cAstL/EY98uIGSQ5U8MKY3YWEN3cRkTOgZkN6aO8/pxrsrdvBGXpHTccwxstL/ASuKDvDSogKuHtqJvmmtnI5jjF+ZfOZJDO3Shj/NWc2mkkNOxzHHwEq/AbV1yh/eXEVifBS/Ou9kp+MY43dcYcJjV/YnMjyMW19faqNxBhAr/Qa8uLCAldtLue+iLBKi7UKVMQ1p3zKaf1zWl9XFZfztf+ucjmO8ZKV/lI27DvL3D9Zx1snJXNSng9NxjPFrI7Lacf2pGTz/ZQEfrdnldBzjBSt9D5U1tdw6YxnxUeE8eFkfROzirTGNufv87vTsmMCvZy+3SdUDgJW+h4c+WM/aHWU8eFkf2raIdjqOMQEhKtzFkxMGUF1TZ8M0BAArfbfPN5Twry+2cM2wTgzv0c7pOMYElM5JcTxwSW9yCvbz6Mc225Y/s9IHCvYc5rYZS8lsG8/vL+jhdBxjAtKY/imMG5TG1E838dkG/xwt11jpc6C8ihteyAHgX9dmEx3hcjiRMYHrT6N60r19C+6YuYydpRVOxzENCOnSr6ypZdLLeRTtP8Jz12TTKTHO6UjGBLToiPrz+xXVtfzy9a/t/L4fCtnSr66t41dvrGDJln08dHkfGzHQGB/p2jaev15af37/oQ/XOx3HHCUkS7+iupZfvJLHO8uLufv87ozuZ4OpGeNLo/ulMHFoOs9+tpkPV+90Oo7xEHKlX1ZRzTX/XsIn63Zz/5heTD7zJKcjGROU7r0oiz6pLbnrjeVs3XvY6TjGLaRKf0XRAcY+tZCl2/bzxLj+XD20k9ORjAlaUeEupk4YQJgIv3jlayqqbXwef+BV6YvISBFZLyL5InJ3A+9PFpGVIrJMRL4QkSyP934LzRFXAAAO1ElEQVTn3m69iJzny/Deqqyp5aG567jkqYWUVVTzwvWDubhvRyeiGBNS0trE8uiVfVmzo4x73lpl4+/7gUanS3TPcTsVOIf6SdJzRGSOqq7xWO01VX3Gvf4o4BFgpLv8xwE9gY7AxyLSTVWb5Uf+gfIq/vP1dl5eVEDB3nIuG5jKvRdm0TLWBlEzprn8pHs7bh2eyROfbKRfWism2m/Yjmq09IHBQL6qbgYQkRnAaODb0lfVMo/144BvfpyPBmaoaiWwRUTy3V9vkQ+yf0dlTS3LCg9QuK+cwn3l5O8+xLx1u6msqaNfWiuev7gnZ3dv6+tva4zxwm3DM1m+7QB/fmc1WR0TGJDe2ulIIcub0k8Btnm8LgKGHL2SiNwM3AlEAj/x2HbxUdt+71YZEZkETAJIT0/3Jvf3lB2p4cpp9d8qTKBjqxguz05lwuBOZHVMOK6vaYzxDVeY8Pi4flz85Bfc9MrXvPPL00huEeV0rJDkTek3NNTk907MqepUYKqITADuAa49hm2nAdOgfmJ0LzJ9T1J8JC/dMJj0NrGktI4hwhVS16iN8XutYiN5ZuJAxj69kJtezePVnw4lMtz+P21u3vyJFwFpHq9TgeIfWX8GMOY4tz1uIsIZ3ZLJSIqzwjfGT/Xs2JK/j+1DTsF+pry72uk4IcmbdswBMkWks4hEUn9hdo7nCiKS6fHyQmCj+/M5wDgRiRKRzkAmsOTEYxtjAtXofin8/MwuvLK4kNeXFDodJ+Q0enpHVWtE5BZgLuACpqvqahGZAuSq6hzgFhEZAVQD+6k/tYN7vVnUX/StAW5urjt3jDH+6zfndWdNcRn3vb2KzLbxNgxKMxJ/u282Oztbc3NznY5hjGlipeXVjHnqS8qOVPP2LaeS2jrW6UgBTUTyVDW7sfXs5LcxxhEtYyN47ppsqmrr+OmLuRyurHE6Ukiw0jfGOKZr23imThjAhl0HuWPmMurq/OvMQzCy0jfGOOqMbsnce1EWH67Zxd/nrnM6TtDz5j59Y4xpUtedksGmkkM8+9lmMhLjGD/4+B7SNI2z0jfGOE5E+NPFPdm27wj3vLWK1NYxnJ6Z7HSsoGSnd4wxfiHcFcaTE/qT2Taem175mvU7DzodKShZ6Rtj/EaL6Aj+fd0gYiJdXPf8EnaUHnE6UtCx0jfG+JWUVjG8cP1gDlbUcP3zOZRVVDsdKahY6Rtj/E5WxwSemTiQ/N2H+PlLeVTW2IP8vmKlb4zxS6dlJvHQ5X1YtHkvd85cTq3dw+8TdveOMcZvXdI/lb2HqvjLe2tpGRvBA2N6IdLQiO3GW1b6xhi/9tPTu7DvcBVPzd9Em9hIfnXeyU5HCmhW+sYYv/fr805mf3k1T36aT0JMOJPOOMnpSAHLSt8Y4/dEhL+M6cXBimr+7/11xES4uHpYhtOxApKVvjEmILjChEev7EdlTR33vr2aqHAXVwxKa3xD8x12944xJmBEuJ/aPT0zid/+dwVvLi1yOlLA8ar0RWSkiKwXkXwRubuB9+8UkTUiskJEPhGRTh7v1YrIMvfHnKO3NcaYYxEV7mLa1dkM7ZzIXbOW89+vrfiPRaOlLyIuYCpwPpAFjBeRrKNWWwpkq2ofYDbwoMd7R1S1n/tjlI9yG2NCWEyki+nXDWJol0TuemM5s/Os+L3lzZH+YCBfVTerahUwAxjtuYKqfqqq5e6Xi4FU38Y0xpjviol08e9rB3HqSUn8evZyZuVsczpSQPCm9FMAzz/NIveyH3Ij8D+P19Eikisii0VkTEMbiMgk9zq5JSUlXkQyxpj64v/XtdmcnpnMb/6zgulfbHE6kt/zpvQbevytweehRWQikA085LE43T1Z7wTgMRH53g22qjpNVbNVNTs52cbQNsZ4LzrCxXPXDGRkz/ZMeXcNj3+8EVUbsuGHeFP6RYDnfVGpQPHRK4nICOAPwChVrfxmuaoWu/+7GZgP9D+BvMYY8z1R4S6enNCfsQNSefTjDdz/7lqbb/cHeHOffg6QKSKdge3AOOqP2r8lIv2BZ4GRqrrbY3lroFxVK0UkCTiV717kNcYYnwh3hfHQZX1IiAln+pdb2H2wgoev6EtUuMvpaH6l0dJX1RoRuQWYC7iA6aq6WkSmALmqOof60znxwBvuwZAK3Xfq9ACeFZE66n+r+JuqrmmifTHGhLiwMOG+i7JonxDNX/+3jj2HKnn26mxaxkQ4Hc1viL+d+8rOztbc3FynYxhjAtxbS7fz69nLyUiMY/p1g0hrE+t0pCYlInnu66c/yp7INcYEpTH9U3jx+sHsPljJ6KlfklOwz+lIfsFK3xgTtE7pmsSbN51Cq5gIJjy3mFm5di+/lb4xJqh1SY7nzZtOZUjnRH4zewX3vLUypKdftNI3xgS9lrERvHD9IH5+ZhdeWVzIlc8uZkfpEadjOcJK3xgTEsJdYfzu/B48fdUANu46yIVPfMG8dbucjtXsrPSNMSHl/N4dmPPL02iXEM0NL+Ty53dWh9TpHit9Y0zIOSk5njdvOoXrT83g+S8LGP3kl6wpLnM6VrOw0jfGhKToCBd/vLgn06/LZu/hKkY9+QWPfbyB6to6p6M1KSt9Y0xI+0n3dnx0xxlc1KcDj328kVFPfsnXhfudjtVkrPSNMSGvVWwkj43rz7NXD2Tf4UoufWohv/vvCvYfrnI6ms9Z6RtjjNt5PdvzyV1n8dPTOjMrt4izH57P9C+2UFUTPKd8rPSNMcZDfFQ491yUxXu3nkbPjglMeXcNIx75jHeWFwfFcM1W+sYY04Du7RN45cYhvHD9IGIjXfzy9aWMfPxz3l62ndoALn8bZdMYYxpRW6e8u6KYJ+fls3H3ITonxXH9qRlcOiCV+ChvpiVpet6Osmmlb4wxXqqrU+au3snTn21iRVEp8VHhjB2QwhWD0sjqkIB7PhFH+LT0RWQk8Dj1k6j8S1X/dtT7dwI/BWqAEuAGVd3qfu9a4B73qn9R1Rd/7HtZ6RtjAsHSwv28tGgr763YQVVtHd3axTOmfwoje7anS3J8s+fxWemLiAvYAJxD/Xy5OcB4zxmwRORs4CtVLReRXwBnqeqVItIGyKV+snQF8oCBqvqDN8Fa6RtjAsn+w1W8t3IHby7dTt7W+mrrnBTH2Se35ZSTEhnYqTWt4yKbPIe3pe/NyajBQL57YnNEZAYwGvi29FX1U4/1FwMT3Z+fB3ykqvvc234EjARe92YnjDHG37WOi2Ti0E5MHNqJov3lzFu3m0/W7uaVr7Yy/cstAJyUHEfPji3p1i6erm1bkNYmhnYJ0bSJjSQsrHlPCXlT+imA58wDRcCQH1n/RuB/P7JtyrEENMaYQJHaOpZrhmVwzbAMKqprWb7tALlb9/P11v3kbd3PnOXF31k/PExoER1OdISLqPAweqe24p/j+zdpRm9Kv6EfQw2eExKRidSfyjnzWLYVkUnAJID09HQvIhljjH+LjnAxpEsiQ7okfrvscGUN+bsPUXzgCLsPVrKrrIKDFTVU1tRSUV1HauuYJs/lTekXAWker1OB4qNXEpERwB+AM1W10mPbs47adv7R26rqNGAa1J/T9yKTMcYEnLiocPqmtaJvWivHMnjzcFYOkCkinUUkEhgHzPFcQUT6A88Co1R1t8dbc4FzRaS1iLQGznUvM8YY44BGj/RVtUZEbqG+rF3AdFVdLSJTgFxVnQM8BMQDb7jvUy1U1VGquk9E7qf+BwfAlG8u6hpjjGl+9nCWMcYEAW9v2bSxd4wxJoRY6RtjTAix0jfGmBBipW+MMSHESt8YY0KI3929IyIlwNYT+BJJwB4fxQkUobbPoba/YPscKk5knzupanJjK/ld6Z8oEcn15ralYBJq+xxq+wu2z6GiOfbZTu8YY0wIsdI3xpgQEoylP83pAA4ItX0Otf0F2+dQ0eT7HHTn9I0xxvywYDzSN8YY8wMCsvRFZKSIrBeRfBG5u4H3o0Rkpvv9r0Qko/lT+pYX+3yniKwRkRUi8omIdHIipy81ts8e610mIioiAX+nhzf7LCJXuP+uV4vIa82d0de8+LedLiKfishS97/vC5zI6SsiMl1EdovIqh94X0TkCfefxwoRGeDTAKoaUB/UD++8CegCRALLgayj1rkJeMb9+ThgptO5m2GfzwZi3Z//IhT22b1eC+Bz6udmznY6dzP8PWcCS4HW7tdtnc7dDPs8DfiF+/MsoMDp3Ce4z2cAA4BVP/D+BdRPOSvAUOArX37/QDzS/3aidlWtAr6ZqN3TaOBF9+ezgeHiHug/QDW6z6r6qaqWu18upn6WskDmzd8zwP3Ag0BFc4ZrIt7s88+Aqaq6H0C/O2lRIPJmnxVIcH/ekgZm7gskqvo58GPziowGXtJ6i4FWItLBV98/EEvfm8nWv11HVWuAUiCRwHWsE8x7Tk4fqBrdZ/eMbWmq+m5zBmtC3vw9dwO6iciXIrJYREY2W7qm4c0+/wmYKCJFwPvAL5snmmOO9f/3Y+LNHLn+xpvJ1r2ezD1AnMjk9IHqR/dZRMKAR4HrmitQM/Dm7zmc+lM8Z1H/29wCEemlqgeaOFtT8WafxwMvqOrDIjIMeNm9z3VNH88RTdpfgXik781E7d+uIyLh1P9KGMjTNB7r5PSj9P9PTh+oGtvnFkAvYL6IFFB/7nNOgF/M9fbf9tuqWq2qW4D11P8QCFTe7PONwCwAVV0ERFM/Rk2w8ur/9+MViKXf6ETt7tfXuj+/DJin7iskAepEJqcPVD+6z6paqqpJqpqhqhnUX8cYpaqBPNemN/+236L+oj0ikkT96Z7NzZrSt7zZ50JgOICI9KC+9EuaNWXzmgNc476LZyhQqqo7fPXFA+70jno3Ufu/qf8VMJ/6I/xxziU+cV7uc4OT0zsW+gR5uc9Bxct9ngucKyJrgFrg16q617nUJ8bLfb4LeE5E7qD+NMd1gXwQJyKvU396Lsl9neKPQASAqj5D/XWLC4B8oBy43qffP4D/7IwxxhyjQDy9Y4wx5jhZ6RtjTAix0jfGmBBipW+MMSHESt8YY0KIlb4xxoQQK31jjAkhVvrGGBNC/h8A+QlHc5forAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "p = torch.linspace(0.,1,100)\n",
    "f = combine_scheds([0.3,0.7], [SchedCos(0.3,0.6), SchedCos(0.6,0.2)])\n",
    "plt.plot(p, [f(o) for o in p]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": [
    "p = torch.linspace(0.,1,100)\n",
    "f = combine_scheds([0.3,0.2,0.5], [SchedLin(0.,1.), SchedNo(1.,1.), SchedCos(1., 0.)])\n",
    "plt.plot(p, [f(o) for o in p]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hide\n",
    "test_close([f(0.), f(0.15), f(0.3), f(0.4), f(0.5), f(0.7), f(1.)],\n",
    "           [0., 0.5, 1., 1., 1., 0.65451, 0.])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "def combined_cos(pct, start, middle, end):\n",
    "    \"Return a scheduler with cosine annealing from `start`→`middle` & `middle`→`end`\"\n",
    "    return combine_scheds([pct,1-pct], [SchedCos(start, middle), SchedCos(middle, end)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is a useful helper function for the [1cycle policy](https://sgugger.github.io/the-1cycle-policy.html). `pct` is used for the `start` to `middle` part, `1-pct` for the `middle` to `end`. Handles floats or collection of floats. For example:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": [
    "f = combined_cos(0.25,0.5,1.,0.)\n",
    "plt.plot(p, [f(o) for o in p]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hide\n",
    "test_close([f(0.), f(0.1), f(0.25), f(0.5), f(1.)], [0.5, 0.67275, 1., 0.75, 0.])\n",
    "f = combined_cos(0.25, np.array([0.25,0.5]), np.array([0.5,1.]), np.array([0.,0.]))\n",
    "for a,b in zip([f(0.), f(0.1), f(0.25), f(0.5), f(1.)],\n",
    "               [[0.25,0.5], [0.33638,0.67275], [0.5,1.], [0.375,0.75], [0.,0.]]):\n",
    "    test_close(a,b)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ParamScheduler -"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "class ParamScheduler(Callback):\n",
    "    \"Schedule hyper-parameters according to `scheds`\"\n",
    "    order,run_valid = 60,False\n",
    "\n",
    "    def __init__(self, scheds): self.scheds = scheds\n",
    "    def before_fit(self): \n",
    "        \"Initialize container for hyper-parameters\"\n",
    "        self.hps = {p:[] for p in self.scheds.keys()}\n",
    "        \n",
    "    def before_batch(self): \n",
    "        \"Set the proper hyper-parameters in the optimizer\"\n",
    "        self._update_val(self.pct_train)\n",
    "\n",
    "    def _update_val(self, pct):\n",
    "        for n,f in self.scheds.items(): self.opt.set_hyper(n, f(pct))\n",
    "\n",
    "    def after_batch(self):\n",
    "        \"Record hyper-parameters of this batch\"\n",
    "        for p in self.scheds.keys(): self.hps[p].append(self.opt.hypers[-1][p])\n",
    "\n",
    "    def after_fit(self):\n",
    "        \"Save the hyper-parameters in the recorder if there is one\"\n",
    "        if hasattr(self.learn, 'recorder') and hasattr(self, 'hps'): self.recorder.hps = self.hps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`scheds` is a dictionary with one key for each hyper-parameter you want to schedule, with either a scheduler or a list of schedulers as values (in the second case, the list must have the same length as the the number of parameters groups of the optimizer)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hide\n",
    "import torch\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "from fastai_minima.learner import DataLoaders\n",
    "from torch import nn\n",
    "def synth_dbunch(a=2, b=3, bs=16, n_train=10, n_valid=2):\n",
    "    \"A simple dataset where `x` is random and `y = a*x + b` plus some noise.\"\n",
    "    def get_data(n):\n",
    "        x = torch.randn(int(bs*n))\n",
    "        return TensorDataset(x, a*x + b + 0.1*torch.randn(int(bs*n)))\n",
    "    train_ds = get_data(n_train)\n",
    "    valid_ds = get_data(n_valid)\n",
    "    train_dl = DataLoader(train_ds, batch_size=bs, shuffle=True, num_workers=0)\n",
    "    valid_dl = DataLoader(valid_ds, batch_size=bs, num_workers=0)\n",
    "    return DataLoaders(train_dl, valid_dl)\n",
    "def synth_learner(n_train=10, n_valid=2, lr=defaults.lr, **kwargs):\n",
    "    data = synth_dbunch(n_train=n_train,n_valid=n_valid)\n",
    "    return Learner(data, RegModel(), loss_func=nn.MSELoss(), lr=lr, **kwargs)\n",
    "\n",
    "class RegModel(nn.Module):\n",
    "    \"A r\"\n",
    "    def __init__(self): \n",
    "        super().__init__()\n",
    "        self.a,self.b = nn.Parameter(torch.randn(1)),nn.Parameter(torch.randn(1))\n",
    "    def forward(self, x): return x*self.a + self.b"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 4.129218578338623, 5.213221073150635, '00:00']\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/mnt/d/lib/python3.7/site-packages/torch/autograd/__init__.py:132: UserWarning: CUDA initialization: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx (Triggered internally at  /pytorch/c10/cuda/CUDAFunctions.cpp:100.)\n",
      "  allow_unreachable=True)  # allow_unreachable flag\n"
     ]
    }
   ],
   "source": [
    "learn = synth_learner()\n",
    "sched = {'lr': SchedLin(1e-3, 1e-2)}\n",
    "learn.fit(1, cbs=ParamScheduler(sched))\n",
    "n = len(learn.dls.train)\n",
    "test_close(learn.recorder.hps['lr'], [1e-3 + (1e-2-1e-3) * i/n for i in range(n)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 3.896852731704712, 3.411391258239746, '00:00']\n"
     ]
    }
   ],
   "source": [
    "#hide\n",
    "#test discriminative lrs\n",
    "from torch import optim\n",
    "from functools import partial\n",
    "def _splitter(m): return convert_params([[m.a], [m.b]])\n",
    "\n",
    "learn = synth_learner(splitter=_splitter)\n",
    "sched = {'lr': combined_cos(0.5, np.array([1e-4,1e-3]), np.array([1e-3,1e-2]), np.array([1e-5,1e-4]))}\n",
    "learn.fit(1, cbs=ParamScheduler(sched))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"ParamScheduler.before_fit\" class=\"doc_header\"><code>ParamScheduler.before_fit</code><a href=\"__main__.py#L7\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>ParamScheduler.before_fit</code>()\n",
       "\n",
       "Initialize container for hyper-parameters"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(ParamScheduler.before_fit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"ParamScheduler.before_batch\" class=\"doc_header\"><code>ParamScheduler.before_batch</code><a href=\"__main__.py#L11\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>ParamScheduler.before_batch</code>()\n",
       "\n",
       "Set the proper hyper-parameters in the optimizer"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(ParamScheduler.before_batch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"ParamScheduler.after_batch\" class=\"doc_header\"><code>ParamScheduler.after_batch</code><a href=\"__main__.py#L18\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>ParamScheduler.after_batch</code>()\n",
       "\n",
       "Record hyper-parameters of this batch"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(ParamScheduler.after_batch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"ParamScheduler.after_fit\" class=\"doc_header\"><code>ParamScheduler.after_fit</code><a href=\"__main__.py#L22\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>ParamScheduler.after_fit</code>()\n",
       "\n",
       "Save the hyper-parameters in the recorder if there is one"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(ParamScheduler.after_fit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "def fit_one_cycle(self:Learner, n_epoch, lr_max=None, div=25., div_final=1e5, pct_start=0.25, wd=None,\n",
    "                  moms=None, cbs=None, reset_opt=False):\n",
    "    \"Fit `self.model` for `n_epoch` using the 1cycle policy.\"\n",
    "    if self.opt is None: self.create_opt()\n",
    "    self.opt.set_hyper('lr', self.lr if lr_max is None else lr_max)\n",
    "    lr_max = np.array([h['lr'] for h in self.opt.hypers])\n",
    "    scheds = {'lr': combined_cos(pct_start, lr_max/div, lr_max, lr_max/div_final),\n",
    "              'mom': combined_cos(pct_start, *(self.moms if moms is None else moms))}\n",
    "    self.fit(n_epoch, cbs=ParamScheduler(scheds)+L(cbs), reset_opt=reset_opt, wd=wd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The 1cycle policy was introduced by Leslie N. Smith et al. in [Super-Convergence: Very Fast Training of Neural Networks Using Large Learning Rates](https://arxiv.org/abs/1708.07120). It schedules the learning rate with a cosine annealing from `lr_max/div` to `lr_max` then `lr_max/div_final` (pass an array to `lr_max` if you want to use differential learning rates) and the momentum with cosine annealing according to the values in `moms`. The first phase takes `pct_start` of the training. You can optionally pass additional `cbs` and `reset_opt`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 12.811972618103027, 17.15632438659668, '00:00']\n",
      "[1, 12.450505256652832, 16.679840087890625, '00:00']\n"
     ]
    }
   ],
   "source": [
    "#Integration test: training a few epochs should make the model better\n",
    "learn = synth_learner(lr=1e-2)\n",
    "xb,yb = learn.dls.one_batch()\n",
    "init_loss = learn.loss_func(learn.model(xb), yb)\n",
    "learn.fit_one_cycle(2)\n",
    "xb,yb = learn.dls.one_batch()\n",
    "final_loss = learn.loss_func(learn.model(xb), yb)\n",
    "assert final_loss < init_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Scheduler test\n",
    "lrs,moms = learn.recorder.hps['lr'],learn.recorder.hps['mom']\n",
    "test_close(lrs,  [combined_cos(0.25,1e-2/25,1e-2,1e-7)(i/20) for i in range(20)])\n",
    "test_close(moms, [combined_cos(0.25,0.95,0.85,0.95)(i/20) for i in range(20)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "def plot_sched(self:Recorder, keys=None, figsize=None):\n",
    "    keys = self.hps.keys() if keys is None else L(keys)\n",
    "    rows,cols = (len(keys)+1)//2, min(2, len(keys))\n",
    "    figsize = figsize or (6*cols,4*rows)\n",
    "    _, axs = plt.subplots(rows, cols, figsize=figsize)\n",
    "    axs = axs.flatten() if len(keys) > 1 else L(axs)\n",
    "    for p,ax in zip(keys, axs):\n",
    "        ax.plot(self.hps[p])\n",
    "        ax.set_ylabel(p)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 22.827058792114258, 20.066909790039062, '00:00']\n"
     ]
    }
   ],
   "source": [
    "#hide\n",
    "#test discriminative lrs\n",
    "def _splitter(m): return convert_params([[m.a], [m.b]])\n",
    "learn = synth_learner(splitter=_splitter)\n",
    "learn.fit_one_cycle(1, lr_max=slice(1e-3,1e-2))\n",
    "#n = len(learn.dls.train)\n",
    "#est_close(learn.recorder.hps['lr'], [1e-3 + (1e-2-1e-3) * i/n for i in range(n)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 3.069950580596924, 3.042041540145874, '00:00']\n",
      "[1, 3.059396743774414, 3.0282907485961914, '00:00']\n"
     ]
    }
   ],
   "source": [
    "learn = synth_learner()\n",
    "learn.fit_one_cycle(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.recorder.plot_sched()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "def fit_flat_cos(self:Learner, n_epoch, lr=None, div_final=1e5, pct_start=0.75, wd=None,\n",
    "                 cbs=None, reset_opt=False):\n",
    "    \"Fit `self.model` for `n_epoch` at flat `lr` before a cosine annealing.\"\n",
    "    if self.opt is None: self.create_opt()\n",
    "    self.opt.set_hyper('lr', self.lr if lr is None else lr)\n",
    "    lr = np.array([h['lr'] for h in self.opt.hypers])\n",
    "    scheds = {'lr': combined_cos(pct_start, lr, lr, lr/div_final)}\n",
    "    self.fit(n_epoch, cbs=ParamScheduler(scheds)+L(cbs), reset_opt=reset_opt, wd=wd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 20.80571746826172, 20.53799057006836, '00:00']\n",
      "[1, 20.7146053314209, 20.456912994384766, '00:00']\n"
     ]
    }
   ],
   "source": [
    "learn = synth_learner()\n",
    "learn.fit_flat_cos(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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_sched()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "def fit_sgdr(self:Learner, n_cycles, cycle_len, lr_max=None, cycle_mult=2, cbs=None, reset_opt=False, wd=None):\n",
    "    \"Fit `self.model` for `n_cycles` of `cycle_len` using SGDR.\"\n",
    "    if self.opt is None: self.create_opt()\n",
    "    self.opt.set_hyper('lr', self.lr if lr_max is None else lr_max)\n",
    "    lr_max = np.array([h['lr'] for h in self.opt.hypers])\n",
    "    n_epoch = cycle_len * (cycle_mult**n_cycles-1)//(cycle_mult-1)\n",
    "    pcts = [cycle_len * cycle_mult**i / n_epoch for i in range(n_cycles)]\n",
    "    scheds = [SchedCos(lr_max, 0) for _ in range(n_cycles)]\n",
    "    scheds = {'lr': combine_scheds(pcts, scheds)}\n",
    "    self.fit(n_epoch, cbs=ParamScheduler(scheds)+L(cbs), reset_opt=reset_opt, wd=wd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This schedule was introduced by Ilya Loshchilov et al. in [SGDR: Stochastic Gradient Descent with Warm Restarts](https://arxiv.org/abs/1608.03983). It consists of `n_cycles` that are cosine annealings from `lr_max` (defaults to the `Learner` lr) to 0, with a length of `cycle_len * cycle_mult**i` for the `i`-th cycle (first one is `cycle_len`-long, then we multiply the length by `cycle_mult` at each epoch). You can optionally pass additional `cbs` and `reset_opt`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": [
    "#slow\n",
    "learn = synth_learner()\n",
    "with learn.no_logging(): learn.fit_sgdr(3, 1)\n",
    "test_eq(learn.n_epoch, 7)\n",
    "iters = [k * len(learn.dls.train) for k in [0,1,3,7]]\n",
    "for i in range(3):\n",
    "    n = iters[i+1]-iters[i]\n",
    "    #The start of a cycle can be mixed with the 0 of the previous cycle with rounding errors, so we test at +1\n",
    "    test_close(learn.recorder.lrs[iters[i]+1:iters[i+1]], [SchedCos(learn.lr, 0)(k/n) for k in range(1,n)])\n",
    "\n",
    "learn.recorder.plot_sched()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "@delegates(Learner.fit_one_cycle)\n",
    "def fine_tune(self:Learner, epochs, base_lr=2e-3, freeze_epochs=1, lr_mult=100,\n",
    "              pct_start=0.3, div=5.0, **kwargs):\n",
    "    \"Fine tune with `freeze` for `freeze_epochs` then with `unfreeze` from `epochs` using discriminative LR\"\n",
    "    self.freeze()\n",
    "    self.fit_one_cycle(freeze_epochs, slice(base_lr), pct_start=0.99, **kwargs)\n",
    "    base_lr /= 2\n",
    "    self.unfreeze()\n",
    "    self.fit_one_cycle(epochs, slice(base_lr/lr_mult, base_lr), pct_start=pct_start, div=div, **kwargs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 9.032106399536133, 7.739522457122803, '00:00']\n",
      "[0, 8.974329948425293, 7.714879035949707, '00:00']\n"
     ]
    }
   ],
   "source": [
    "learn.fine_tune(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LRFind -"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "class LRFinder(ParamScheduler):\n",
    "    \"Training with exponentially growing learning rate\"\n",
    "    def __init__(self, start_lr=1e-7, end_lr=10, num_it=100, stop_div=True):\n",
    "        if is_listy(start_lr):\n",
    "            self.scheds = {'lr': [SchedExp(s, e) for (s,e) in zip(start_lr,end_lr)]}\n",
    "        else: self.scheds = {'lr': SchedExp(start_lr, end_lr)}\n",
    "        self.num_it,self.stop_div = num_it,stop_div\n",
    "\n",
    "    def before_fit(self):\n",
    "        \"Initialize container for hyper-parameters and save the model\"\n",
    "        super().before_fit()\n",
    "        self.learn.save('_tmp')\n",
    "        self.best_loss = float('inf')\n",
    "\n",
    "    def before_batch(self):\n",
    "        \"Record hyper-parameters of this batch and potentially stop training\"\n",
    "        self._update_val(self.train_iter/self.num_it)\n",
    "\n",
    "    def after_batch(self):\n",
    "        \"Set the proper hyper-parameters in the optimizer\"\n",
    "        super().after_batch()\n",
    "        if self.smooth_loss < self.best_loss: self.best_loss = self.smooth_loss\n",
    "        if self.smooth_loss > 4*self.best_loss and self.stop_div: raise CancelFitException()\n",
    "        if self.train_iter >= self.num_it: raise CancelFitException()\n",
    "\n",
    "    def before_validate(self): \n",
    "        \"Skip the validation part of training\"\n",
    "        raise CancelValidException()\n",
    "\n",
    "    def after_fit(self):\n",
    "        \"Save the hyper-parameters in the recorder if there is one and load the original model\"\n",
    "        self.learn.opt.zero_grad() #Need to zero the gradients of the model before detaching the optimizer for future fits\n",
    "        tmp_f = self.path/self.model_dir/'_tmp.pth'\n",
    "        if tmp_f.exists():\n",
    "            self.learn.load('_tmp', with_opt=True)\n",
    "            os.remove(tmp_f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#slow\n",
    "import tempfile\n",
    "from fastcore.basics import range_of\n",
    "from fastcore.xtras import Path\n",
    "\n",
    "with tempfile.TemporaryDirectory() as d:\n",
    "    learn = synth_learner(path=Path(d))\n",
    "    init_a,init_b = learn.model.a,learn.model.b\n",
    "    with learn.no_logging(): learn.fit(20, cbs=LRFinder(num_it=100))\n",
    "    assert len(learn.recorder.lrs) <= 100\n",
    "    test_eq(len(learn.recorder.lrs), len(learn.recorder.losses))\n",
    "    #Check stop if diverge\n",
    "    if len(learn.recorder.lrs) < 100: assert learn.recorder.losses[-1] > 4 * min(learn.recorder.losses)\n",
    "    #Test schedule\n",
    "    test_eq(learn.recorder.lrs, [SchedExp(1e-7, 10)(i/100) for i in range_of(learn.recorder.lrs)])\n",
    "    #No validation data\n",
    "    test_eq([len(v) for v in learn.recorder.values], [1 for _ in range_of(learn.recorder.values)])\n",
    "    #Model loaded back properly\n",
    "    test_eq(learn.model.a, init_a)\n",
    "    test_eq(learn.model.b, init_b)\n",
    "    test_eq(learn.opt.state_dict()['state'], {})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"LRFinder.before_fit\" class=\"doc_header\"><code>LRFinder.before_fit</code><a href=\"__main__.py#L10\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>LRFinder.before_fit</code>()\n",
       "\n",
       "Initialize container for hyper-parameters and save the model"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(LRFinder.before_fit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"LRFinder.before_batch\" class=\"doc_header\"><code>LRFinder.before_batch</code><a href=\"__main__.py#L16\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>LRFinder.before_batch</code>()\n",
       "\n",
       "Record hyper-parameters of this batch and potentially stop training"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(LRFinder.before_batch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"LRFinder.after_batch\" class=\"doc_header\"><code>LRFinder.after_batch</code><a href=\"__main__.py#L20\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>LRFinder.after_batch</code>()\n",
       "\n",
       "Set the proper hyper-parameters in the optimizer"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(LRFinder.after_batch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<h4 id=\"LRFinder.before_validate\" class=\"doc_header\"><code>LRFinder.before_validate</code><a href=\"__main__.py#L27\" class=\"source_link\" style=\"float:right\">[source]</a></h4>\n",
       "\n",
       "> <code>LRFinder.before_validate</code>()\n",
       "\n",
       "Skip the validation part of training"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "show_doc(LRFinder.before_validate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "def plot_lr_find(self:Recorder, skip_end=5):\n",
    "    \"Plot the result of an LR Finder test (won't work if you didn't do `learn.lr_find()` before)\"\n",
    "    lrs    = self.lrs    if skip_end==0 else self.lrs   [:-skip_end]\n",
    "    losses = self.losses if skip_end==0 else self.losses[:-skip_end]\n",
    "    fig, ax = plt.subplots(1,1)\n",
    "    ax.plot(lrs, losses)\n",
    "    ax.set_ylabel(\"Loss\")\n",
    "    ax.set_xlabel(\"Learning Rate\")\n",
    "    ax.set_xscale('log')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "SuggestedLRs = collections.namedtuple('SuggestedLRs', ['lr_min', 'lr_steep'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#export\n",
    "@patch\n",
    "def lr_find(self:Learner, start_lr=1e-7, end_lr=10, num_it=100, stop_div=True, show_plot=True, suggestions=True):\n",
    "    \"Launch a mock training to find a good learning rate, return lr_min, lr_steep if `suggestions` is True\"\n",
    "    n_epoch = num_it//len(self.dls.train) + 1\n",
    "    cb=LRFinder(start_lr=start_lr, end_lr=end_lr, num_it=num_it, stop_div=stop_div)\n",
    "    with self.no_logging(): self.fit(n_epoch, cbs=cb)\n",
    "    if show_plot: self.recorder.plot_lr_find()\n",
    "    if suggestions:\n",
    "        lrs,losses = tensor(self.recorder.lrs[num_it//10:-5]),tensor(self.recorder.losses[num_it//10:-5])\n",
    "        if len(losses) == 0: return\n",
    "        lr_min = lrs[losses.argmin()].item()\n",
    "        grads = (losses[1:]-losses[:-1]) / (lrs[1:].log()-lrs[:-1].log())\n",
    "        lr_steep = lrs[grads.argmin()].item()\n",
    "        return SuggestedLRs(lr_min/10.,lr_steep)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First introduced by Leslie N. Smith in [Cyclical Learning Rates for Training Neural Networks](https://arxiv.org/pdf/1506.01186.pdf), the LR Finder trains the model with exponentially growing learning rates from `start_lr` to `end_lr` for `num_it` and stops in case of divergence (unless `stop_div=False`) then plots the losses vs the learning rates with a log scale. \n",
    "\n",
    "A good value for the learning rates is then either:\n",
    "- one tenth of the minimum before the divergence\n",
    "- when the slope is the steepest\n",
    "\n",
    "Those two values are returned by default by the Learning Rate Finder."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Minimum/10: 3.31e-01, steepest point: 1.10e-06\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": [
    "#slow\n",
    "with tempfile.TemporaryDirectory() as d:\n",
    "    learn = synth_learner(path=Path(d))\n",
    "    weights_pre_lr_find = L(learn.model.parameters())\n",
    "    lr_min,lr_steep = learn.lr_find()\n",
    "    weights_post_lr_find = L(learn.model.parameters())\n",
    "test_eq(weights_pre_lr_find, weights_post_lr_find)\n",
    "print(f\"Minimum/10: {lr_min:.2e}, steepest point: {lr_steep:.2e}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "jupytext": {
   "split_at_heading": true
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}