torch+bf linear models.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "import batchflow as bf\n",
    "from batchflow import Dataset, V, F, B, action, Batch\n",
    "from batchflow.models.torch.layers import ConvBlock\n",
    "import torch\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Lets generate some linear data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def data_generator(size, dim):\n",
    "    step = 5 / size\n",
    "    x = np.array([np.arange(0, 5, step) for i in range(dim)]).T\n",
    "    coef = np.random.randint(-3, 3, size=dim)\n",
    "    y = x @ coef + np.random.uniform(-0.5, 0.5, size=size)\n",
    "    return x, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x4ef2fb10f0>]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "size, dim = 100, 1\n",
    "x, y = data_generator(100, 1) # generation of 100 points\n",
    "plt.plot(x, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define MyBatch class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MyBatch(Batch):\n",
    "    components = 'features', 'targets'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = Dataset(index=size, batch_class=MyBatch)\n",
    "dataset.split()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define MyModel class to specify our nn with 1 hidden layer without activation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MyModel(TorchModel):\n",
    "    @classmethod\n",
    "    def default_config(cls):\n",
    "        config = TorchModel.default_config()\n",
    "        config['body'].update(dict(layout='df', units=1))\n",
    "        return config\n",
    "    \n",
    "    def build_config(self, names=None):\n",
    "        config = super().build_config(names)\n",
    "        return config\n",
    "     \n",
    "    @classmethod\n",
    "    def body(cls, **kwargs):\n",
    "        kwargs = cls.get_defaults('body', kwargs)\n",
    "        x = ConvBlock(**kwargs)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Lets define  loss and optimizer. NN architechture already specified in MyModel class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "config =  {\n",
    "    'loss' : 'mse',\n",
    "    'optimizer': {'name': 'SGD',\n",
    "                 'lr': 0.1},\n",
    "    'inputs': dict(features={'shape': dim},\n",
    "                   targets={'shape' : 1}),\n",
    "    'initial_block/inputs': 'features',\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "ConvBlock: dropout_rate is zero or undefined, so dropout layer is skipped\n"
     ]
    },
    {
     "ename": "TypeError",
     "evalue": "train() got an unexpected keyword argument 'feed_dict'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-34-fa451fc03b59>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      5\u001b[0m             .train_model('linear', feed_dict=dict(features=B('features'), \n\u001b[0;32m      6\u001b[0m                                                   targets=B('targets')))\n\u001b[1;32m----> 7\u001b[1;33m             .run(batch_size, shuffle=True, n_epochs=1))\n\u001b[0m",
      "\u001b[1;32m~\\PROJECTS\\work\\batchflow\\batchflow\\pipeline.py\u001b[0m in \u001b[0;36mrun\u001b[1;34m(self, init_vars, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1280\u001b[0m                 \u001b[0mwarnings\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mwarn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'Pipeline will never stop as n_epochs=None'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1281\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1282\u001b[1;33m             \u001b[1;32mfor\u001b[0m \u001b[0m_\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mgen_batch\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1283\u001b[0m                 \u001b[1;32mpass\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1284\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\PROJECTS\\work\\batchflow\\batchflow\\pipeline.py\u001b[0m in \u001b[0;36m_gen_batch\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1212\u001b[0m             \u001b[1;32mfor\u001b[0m \u001b[0mbatch\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mbatch_generator\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1213\u001b[0m                 \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1214\u001b[1;33m                     \u001b[0mbatch_res\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexecute_for\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1215\u001b[0m                 \u001b[1;32mexcept\u001b[0m \u001b[0mSkipBatchException\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1216\u001b[0m                     \u001b[1;32mpass\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\PROJECTS\\work\\batchflow\\batchflow\\pipeline.py\u001b[0m in \u001b[0;36mexecute_for\u001b[1;34m(self, batch, new_loop)\u001b[0m\n\u001b[0;32m    607\u001b[0m             \u001b[0masyncio\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mset_event_loop\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0masyncio\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mnew_event_loop\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    608\u001b[0m         \u001b[0mbatch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpipeline\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 609\u001b[1;33m         \u001b[0mbatch_res\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_exec_all_actions\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    610\u001b[0m         \u001b[0mbatch_res\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpipeline\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    611\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mbatch_res\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\PROJECTS\\work\\batchflow\\batchflow\\pipeline.py\u001b[0m in \u001b[0;36m_exec_all_actions\u001b[1;34m(self, batch, action_list)\u001b[0m\n\u001b[0;32m    571\u001b[0m             \u001b[1;32melif\u001b[0m \u001b[0m_action\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'name'\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;32min\u001b[0m \u001b[0m_ACTIONS\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    572\u001b[0m                 \u001b[0maction_fn\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mgetattr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_ACTIONS\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0m_action\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'name'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 573\u001b[1;33m                 \u001b[0maction_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_action\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    574\u001b[0m             \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    575\u001b[0m                 \u001b[1;32mif\u001b[0m \u001b[0mjoin_batches\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\PROJECTS\\work\\batchflow\\batchflow\\pipeline.py\u001b[0m in \u001b[0;36m_exec_train_model\u001b[1;34m(self, batch, action)\u001b[0m\n\u001b[0;32m    885\u001b[0m         \u001b[0mmodel\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_model_by_name\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0maction\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'model_name'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mbatch\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    886\u001b[0m         \u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkwargs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_make_model_args\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maction\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 887\u001b[1;33m         \u001b[0moutput\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    888\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_save_output\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0moutput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maction\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'save_to'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maction\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'mode'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    889\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mTypeError\u001b[0m: train() got an unexpected keyword argument 'feed_dict'"
     ]
    }
   ],
   "source": [
    "pipeline = (dataset.train.p\n",
    "            .load(src=(x, y))\n",
    "             #.init_variable('loss')\n",
    "            .init_model('dynamic', MyModel, 'linear', config=config)\n",
    "            .train_model('linear', feed_dict=dict(features=B('features'), \n",
    "                                                  targets=B('targets')))\n",
    "            .run(batch_size, shuffle=True, n_epochs=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Another way to specify MyModel class pushing the single neuron of out net to HEAD instead of BODY block"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MyModel_1(TorchModel):\n",
    "    @classmethod\n",
    "    def default_config(cls):\n",
    "        config = TorchModel.default_config()\n",
    "        config['head'].update(dict(layout='df', units=1))\n",
    "        return config\n",
    "    \n",
    "    def build_config(self, names=None):\n",
    "        config = super().build_config(names)\n",
    "        config['head/units'] = self.num_classes('targets')\n",
    "        return config\n",
    "     \n",
    "    @classmethod\n",
    "    def head(cls, **kwargs):\n",
    "        kwargs = cls.get_defaults('head', kwargs)\n",
    "        x = ConvBlock(**kwargs)\n",
    "        return x"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}