SavingLoading.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Saving / Loading models in MXNet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sample Network -> Saving"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "import mxnet as mx\n",
    "import numpy as np\n",
    "from mxnet import gluon\n",
    "ctx = mx.cpu()\n",
    "\n",
    "save_params = 'save_params.params'\n",
    "collect_params = 'collect_params.params'\n",
    "export_params = 'export-0000.params'\n",
    "sym_symbol = 'sym.json'\n",
    "export_symbol = 'export-symbol.json'\n",
    "\n",
    "# Create network\n",
    "def get_net(prefix=\"test_\"):\n",
    "    net = gluon.nn.HybridSequential(prefix=prefix)\n",
    "    with net.name_scope():\n",
    "        net.add(gluon.nn.Conv2D(10, (3, 3)))\n",
    "        net.add(gluon.nn.Dense(50))\n",
    "        net.add(gluon.nn.BatchNorm())\n",
    "    net.initialize()\n",
    "    return net\n",
    "\n",
    "net = get_net()\n",
    "# Save network    \n",
    "net.hybridize()\n",
    "data = mx.nd.ones((1,1,50,50))\n",
    "out = net(data).asnumpy()\n",
    "net.export('export', epoch=0)\n",
    "net.save_params(save_params)\n",
    "net.collect_params().save(collect_params)\n",
    "\n",
    "sym = net(mx.sym.Variable('data'))\n",
    "sym.save(sym_symbol)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.1.0'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mx.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading in Python / Gluon Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `save_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save_params / load_params\n",
    "net = get_net()\n",
    "net.load_params(save_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "test_ (\n",
       "  Parameter test_conv0_weight (shape=(10L, 0L, 3L, 3L), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_conv0_bias (shape=(10L,), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_dense0_weight (shape=(50, 0), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_dense0_bias (shape=(50,), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_batchnorm0_gamma (shape=(0,), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_batchnorm0_beta (shape=(0,), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_batchnorm0_running_mean (shape=(0,), dtype=<type 'numpy.float32'>)\n",
       "  Parameter test_batchnorm0_running_var (shape=(0,), dtype=<type 'numpy.float32'>)\n",
       ")"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "net.collect_params()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save_params / collect_params\n",
    "net = get_net(\"\")\n",
    "net.collect_params().load(save_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `collect_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# collect_params / load_params\n",
    "net = get_net()\n",
    "net._prefix = \"\"\n",
    "net.load_params(collect_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# collect_params / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(collect_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `export`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export / load_params\n",
    "net = get_net()\n",
    "net._prefix = \"\"\n",
    "net.load_params(export_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(export_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading in Python / Symbol Block - export json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "137c137\n",
      "<       \"inputs\": [[6, 0, 0], [7, 0, 0], [8, 0, 0], [9, 0, 2], [10, 0, 2]]\n",
      "---\n",
      ">       \"inputs\": [[6, 0, 0], [7, 0, 0], [8, 0, 0], [9, 0, 1], [10, 0, 1]]\n"
     ]
    }
   ],
   "source": [
    "!diff sym.json export-symbol.json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_net():\n",
    "    sym = mx.sym.load_json(open(export_symbol, 'r').read())\n",
    "    net = gluon.nn.SymbolBlock(outputs=sym, inputs=mx.sym.var('data'))\n",
    "    return net"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `save_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "ename": "AssertionError",
     "evalue": "Parameter test_conv0_weight is missing in file save_params.params",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mAssertionError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-30-bc3892642c64>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# save_params / load_params\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mnet\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_net\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mnet\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msave_params\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mctx\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcpu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;32massert\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray_equal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/gluon/block.pyc\u001b[0m in \u001b[0;36mload_params\u001b[0;34m(self, filename, ctx, allow_missing, ignore_extra)\u001b[0m\n\u001b[1;32m    315\u001b[0m         \"\"\"\n\u001b[1;32m    316\u001b[0m         self.collect_params().load(filename, ctx, allow_missing, ignore_extra,\n\u001b[0;32m--> 317\u001b[0;31m                                    self.prefix)\n\u001b[0m\u001b[1;32m    318\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    319\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mregister_child\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mblock\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/gluon/parameter.pyc\u001b[0m in \u001b[0;36mload\u001b[0;34m(self, filename, ctx, allow_missing, ignore_extra, restore_prefix)\u001b[0m\n\u001b[1;32m    667\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    668\u001b[0m                 \u001b[0;32massert\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 669\u001b[0;31m                     \u001b[0;34m\"Parameter %s is missing in file %s\"\u001b[0m\u001b[0;34m%\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mlprefix\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    670\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    671\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_params\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAssertionError\u001b[0m: Parameter test_conv0_weight is missing in file save_params.params"
     ]
    }
   ],
   "source": [
    "# save_params / load_params\n",
    "net = get_net()\n",
    "net.load_params(save_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "ename": "AssertionError",
     "evalue": "Parameter test_conv0_weight is missing in file save_params.params",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mAssertionError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-31-2378c5c25d8c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# save_params / collect_params\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mnet\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_net\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mnet\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcollect_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msave_params\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mctx\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcpu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;32massert\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray_equal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/gluon/parameter.pyc\u001b[0m in \u001b[0;36mload\u001b[0;34m(self, filename, ctx, allow_missing, ignore_extra, restore_prefix)\u001b[0m\n\u001b[1;32m    667\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    668\u001b[0m                 \u001b[0;32massert\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 669\u001b[0;31m                     \u001b[0;34m\"Parameter %s is missing in file %s\"\u001b[0m\u001b[0;34m%\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mlprefix\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    670\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    671\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_params\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAssertionError\u001b[0m: Parameter test_conv0_weight is missing in file save_params.params"
     ]
    }
   ],
   "source": [
    "# save_params / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(save_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `collect_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# collect_params / load_params\n",
    "net = get_net()\n",
    "net.load_params(collect_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# collect_params / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(collect_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `export`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export / load_params\n",
    "net = get_net()\n",
    "net.load_params(export_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(export_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading in Python / Symbol Block - mx.sym.var('data') json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_net():\n",
    "    sym = mx.sym.load_json(open(sym_symbol, 'r').read())\n",
    "    net = gluon.nn.SymbolBlock(outputs=sym, inputs=mx.sym.var('data'))\n",
    "    return net"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `save_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "ename": "AssertionError",
     "evalue": "Parameter test_conv0_weight is missing in file save_params.params",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mAssertionError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-47-bc3892642c64>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# save_params / load_params\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mnet\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_net\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mnet\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msave_params\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mctx\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcpu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;32massert\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray_equal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/gluon/block.pyc\u001b[0m in \u001b[0;36mload_params\u001b[0;34m(self, filename, ctx, allow_missing, ignore_extra)\u001b[0m\n\u001b[1;32m    315\u001b[0m         \"\"\"\n\u001b[1;32m    316\u001b[0m         self.collect_params().load(filename, ctx, allow_missing, ignore_extra,\n\u001b[0;32m--> 317\u001b[0;31m                                    self.prefix)\n\u001b[0m\u001b[1;32m    318\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    319\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mregister_child\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mblock\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/gluon/parameter.pyc\u001b[0m in \u001b[0;36mload\u001b[0;34m(self, filename, ctx, allow_missing, ignore_extra, restore_prefix)\u001b[0m\n\u001b[1;32m    667\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    668\u001b[0m                 \u001b[0;32massert\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 669\u001b[0;31m                     \u001b[0;34m\"Parameter %s is missing in file %s\"\u001b[0m\u001b[0;34m%\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mlprefix\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    670\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    671\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_params\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAssertionError\u001b[0m: Parameter test_conv0_weight is missing in file save_params.params"
     ]
    }
   ],
   "source": [
    "# save_params / load_params\n",
    "net = get_net()\n",
    "net.load_params(save_params, ctx=mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "ename": "AssertionError",
     "evalue": "Parameter test_conv0_weight is missing in file save_params.params",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mAssertionError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-48-dc38e98a2978>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# save_params / collect_params\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mnet\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_net\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mnet\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcollect_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msave_params\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcpu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;32massert\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray_equal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mout\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/gluon/parameter.pyc\u001b[0m in \u001b[0;36mload\u001b[0;34m(self, filename, ctx, allow_missing, ignore_extra, restore_prefix)\u001b[0m\n\u001b[1;32m    667\u001b[0m             \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    668\u001b[0m                 \u001b[0;32massert\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 669\u001b[0;31m                     \u001b[0;34m\"Parameter %s is missing in file %s\"\u001b[0m\u001b[0;34m%\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mlprefix\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfilename\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    670\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marg_dict\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    671\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_params\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAssertionError\u001b[0m: Parameter test_conv0_weight is missing in file save_params.params"
     ]
    }
   ],
   "source": [
    "# save_params / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(save_params, mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `collect_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "# collect_params / load_params\n",
    "net = get_net()\n",
    "net.load_params(collect_params, mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "# collect_params / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(collect_params, mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `export`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export / load_params\n",
    "net = get_net()\n",
    "net.load_params(export_params, mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export / collect_params\n",
    "net = get_net()\n",
    "net.collect_params().load(export_params, mx.cpu())\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading with module API"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_module():\n",
    "    sym = mx.sym.load_json(open(sym_symbol, 'r').read())\n",
    "    mod = mx.mod.Module(symbol=sym, context=ctx, label_names=None)\n",
    "    mod.bind(for_training=False, data_shapes=[('data', (1,1,50,50))], \n",
    "             label_shapes=mod._label_shapes)\n",
    "    return mod"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `save_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "need more than 1 value to unpack",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mValueError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-57-f179d8f58eab>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msystem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mu'cp $sym_symbol test-symbol.json'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msystem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mu'cp $save_params test-0000.params'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0msym\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0marg_params\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maux_params\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload_checkpoint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'test'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/model.pyc\u001b[0m in \u001b[0;36mload_checkpoint\u001b[0;34m(prefix, epoch)\u001b[0m\n\u001b[1;32m    423\u001b[0m     \u001b[0maux_params\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    424\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mk\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mv\u001b[0m \u001b[0;32min\u001b[0m \u001b[0msave_dict\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitems\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 425\u001b[0;31m         \u001b[0mtp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mk\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m':'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    426\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mtp\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'arg'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    427\u001b[0m             \u001b[0marg_params\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mv\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: need more than 1 value to unpack"
     ]
    }
   ],
   "source": [
    "!cp $sym_symbol test-symbol.json\n",
    "!cp $save_params test-0000.params\n",
    "sym, arg_params, aux_params = mx.model.load_checkpoint('test', 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `collect_params`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "need more than 1 value to unpack",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mValueError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-58-ae8bdf923005>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msystem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mu'cp $sym_symbol test-symbol.json'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msystem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mu'cp $collect_params test-0000.params'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0msym\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0marg_params\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maux_params\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload_checkpoint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'test'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/home/ubuntu/anaconda3/envs/mxnet_p27/lib/python2.7/site-packages/mxnet/model.pyc\u001b[0m in \u001b[0;36mload_checkpoint\u001b[0;34m(prefix, epoch)\u001b[0m\n\u001b[1;32m    423\u001b[0m     \u001b[0maux_params\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    424\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mk\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mv\u001b[0m \u001b[0;32min\u001b[0m \u001b[0msave_dict\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitems\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 425\u001b[0;31m         \u001b[0mtp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mk\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m':'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    426\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mtp\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'arg'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    427\u001b[0m             \u001b[0marg_params\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mv\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: need more than 1 value to unpack"
     ]
    }
   ],
   "source": [
    "!cp $sym_symbol test-symbol.json\n",
    "!cp $collect_params test-0000.params\n",
    "sym, arg_params, aux_params = mx.model.load_checkpoint('test', 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Parameters saved with `export`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "sym, arg_params, aux_params = mx.model.load_checkpoint('export', 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "mod = get_module()\n",
    "mod.set_params(arg_params, aux_params, allow_missing=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "mod.forward(mx.io.DataBatch([data]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "out = mod.get_outputs()[0].asnumpy()\n",
    "assert np.array_equal(out, net(data).asnumpy())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Environment (conda_mxnet_p27)",
   "language": "python",
   "name": "conda_mxnet_p27"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}