chainer trainer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "using PyCall"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PyObject <module 'pandas' from '/home/regonn/.local/lib/python3.7/site-packages/pandas/__init__.py'>"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chainer = pyimport(\"chainer\")\n",
    "math = pyimport(\"math\")\n",
    "random = pyimport(\"random\")\n",
    "np = pyimport(\"numpy\")\n",
    "pd = pyimport(\"pandas\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:1\n",
      "└ @ Core In[3]:1\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:2\n",
      "└ @ Core In[3]:2\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:3\n",
      "└ @ Core In[3]:3\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:4\n",
      "└ @ Core In[3]:4\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:5\n",
      "└ @ Core In[3]:5\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:6\n",
      "└ @ Core In[3]:6\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:7\n",
      "└ @ Core In[3]:7\n",
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = top-level scope at In[3]:8\n",
      "└ @ Core In[3]:8\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "PyObject <module 'chainer.training.extensions' from '/home/regonn/.local/lib/python3.7/site-packages/chainer/training/extensions/__init__.py'>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "const Chain = chainer[:Chain]\n",
    "const Variable = chainer[:Variable]\n",
    "const F = chainer[:functions]\n",
    "const L = chainer[:links]\n",
    "const iterators = chainer[:iterators]\n",
    "const optimizers = chainer[:optimizers]\n",
    "const training = chainer[:training]\n",
    "const extensions = training[:extensions]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(PyObject <chainer.datasets.tuple_dataset.TupleDataset object at 0x7efdc0af2470>, PyObject <chainer.datasets.tuple_dataset.TupleDataset object at 0x7efdaa132b70>)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "const mnist = chainer.datasets.mnist\n",
    "train, test = mnist.get_mnist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "128"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batchsize = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PyObject <chainer.iterators.serial_iterator.SerialIterator object at 0x7efdaa1326a0>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_iter = iterators.SerialIterator(train, batchsize)\n",
    "test_iter = iterators.SerialIterator(test, batchsize, repeat=false, shuffle=false)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PyObject <class 'MLP'>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "@pydef mutable struct MLP <: Chain\n",
    "    function __init__(self, n_mid_units=100, n_out=10)\n",
    "        pybuiltin(:super)(MLP, self)[:__init__]()\n",
    "        @pywith self.init_scope() begin\n",
    "            self.l1 = L.Linear(py\"None\", n_mid_units)\n",
    "            self.l2 = L.Linear(py\"None\", n_mid_units)\n",
    "            self.l3 = L.Linear(py\"None\", n_out)\n",
    "        end\n",
    "    end\n",
    "\n",
    "    function forward(self, x)\n",
    "        h1 = F.relu(self.l1(x))\n",
    "        h2 = F.relu(self.l2(h1))\n",
    "        return self.l3(h2)\n",
    "    end\n",
    "end"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "┌ Warning: `getindex(o::PyObject, s::Symbol)` is deprecated in favor of dot overloading (`getproperty`) so elements should now be accessed as e.g. `o.s` instead of `o[:s]`.\n",
      "│   caller = macro expansion at In[7]:3 [inlined]\n",
      "└ @ Core ./In[7]:3\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "PyObject <chainer.links.model.classifier.Classifier object at 0x7efdaa1551d0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max_epoch = 10\n",
    "model = MLP()\n",
    "model = L.Classifier(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PyObject <chainer.optimizers.momentum_sgd.MomentumSGD object at 0x7efdaa155470>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "optimizer = optimizers.MomentumSGD()\n",
    "optimizer.setup(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PyObject <chainer.training.updaters.standard_updater.StandardUpdater object at 0x7efdaa155518>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "updater = training.updaters.StandardUpdater(train_iter, optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer = training.Trainer(updater, (max_epoch, \"epoch\"), out=\"mnist_result\")\n",
    "trainer.extend(extensions.LogReport())\n",
    "trainer.extend(extensions.Evaluator(test_iter, model))\n",
    "trainer.extend(extensions.PrintReport([\"epoch\", \"main/loss\", \"main/accuracy\", \"validation/main/loss\", \"validation/main/accuracy\", \"elapsed_time\"]))\n",
    "trainer.extend(extensions.PlotReport([\"main/loss\", \"validation/main/loss\"], x_key=\"epoch\", file_name=\"loss.png\"))\n",
    "trainer.extend(extensions.PlotReport([\"main/accuracy\", \"validation/main/accuracy\"], x_key=\"epoch\", file_name=\"accuracy.png\"))\n",
    "trainer.extend(extensions.dump_graph(\"main/loss\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch       main/loss   main/accuracy  validation/main/loss  validation/main/accuracy  elapsed_time\n",
      "\u001b[J1           0.53921     0.850197       0.270433              0.919502                  151.316       \n",
      "\u001b[J2           0.236785    0.93122        0.199803              0.943335                  302.314       \n",
      "\u001b[J3           0.179879    0.947795       0.157251              0.953224                  451.812       \n",
      "\u001b[J4           0.143571    0.958417       0.134044              0.959652                  602.499       \n",
      "\u001b[J5           0.120292    0.965685       0.118129              0.964399                  751.069       \n",
      "\u001b[J6           0.10287     0.969716       0.101918              0.96875                   893.483       \n",
      "\u001b[J7           0.0892034   0.974264       0.0912552             0.972013                  1041.57       \n",
      "\u001b[J8           0.077978    0.977664       0.090307              0.971816                  1192.46       \n",
      "\u001b[J9           0.0701926   0.979278       0.087615              0.972903                  1362.7        \n",
      "\u001b[J10          0.0617166   0.98136        0.0808613             0.97498                   1504.98       \n"
     ]
    }
   ],
   "source": [
    "trainer.run()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Julia 1.1.0",
   "language": "julia",
   "name": "julia-1.1"
  },
  "language_info": {
   "file_extension": ".jl",
   "mimetype": "application/julia",
   "name": "julia",
   "version": "1.1.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}