Simple example of TensorFlow.js in IJavaScript Notebook

iris_fit_dataset.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "(node:110831) Warning: N-API is an experimental feature and could change at any time.\n"
     ]
    }
   ],
   "source": [
    "const tf = require('@tensorflow/tfjs-node');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output shape              Param #   \n",
      "=================================================================\n",
      "dense_Dense1 (Dense)         [null,5]                  25        \n",
      "_________________________________________________________________\n",
      "dense_Dense2 (Dense)         [null,3]                  18        \n",
      "=================================================================\n",
      "Total params: 43\n",
      "Trainable params: 43\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = tf.sequential();\n",
    "model.add(tf.layers.dense({units: 5, activation: 'relu', inputShape: [4]}));\n",
    "model.add(tf.layers.dense({units: 3, activation: 'softmax'}));\n",
    "model.compile({loss: 'categoricalCrossentropy', optimizer: 'sgd'});\n",
    "model.summary();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class_1 { size: null }"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SPECIES = ['setosa', 'versicolor', 'virginica'];\n",
    "dataset = tf.data.csv(\n",
    "    'https://raw.githubusercontent.com/uiuc-cse/data-fa14/gh-pages/data/iris.csv')\n",
    "    .map(item => [\n",
    "        [item['sepal_length'], item['sepal_width'], item['petal_length'], item['petal_width']],\n",
    "        tf.oneHot(tf.scalar(SPECIES.indexOf(item['species']), 'int32'), 3)\n",
    "    ]).shuffle(100).batch(16);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "173ms 17325us/step - loss=1.29 \n",
      "Epoch 2 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "133ms 13324us/step - loss=1.10 \n",
      "Epoch 3 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "128ms 12849us/step - loss=1.07 \n",
      "Epoch 4 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "113ms 11344us/step - loss=1.06 \n",
      "Epoch 5 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "106ms 10620us/step - loss=1.05 \n",
      "Epoch 6 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "120ms 12047us/step - loss=1.04 \n",
      "Epoch 7 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "105ms 10519us/step - loss=1.03 \n",
      "Epoch 8 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "101ms 10133us/step - loss=1.02 \n",
      "Epoch 9 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "115ms 11505us/step - loss=1.01 \n",
      "Epoch 10 / 10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "98ms 9759us/step - loss=1.00 \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "History {\n",
       "  validationData: null,\n",
       "  params: \n",
       "   { epochs: 10,\n",
       "     initialEpoch: null,\n",
       "     samples: null,\n",
       "     steps: null,\n",
       "     batchSize: null,\n",
       "     verbose: 1,\n",
       "     doValidation: false,\n",
       "     metrics: [ 'loss' ] },\n",
       "  epoch: [ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 ],\n",
       "  history: \n",
       "   { loss: \n",
       "      [ 1.2919349670410156,\n",
       "        1.1000075340270996,\n",
       "        1.0673496723175049,\n",
       "        1.0560511350631714,\n",
       "        1.0499601364135742,\n",
       "        1.0428799390792847,\n",
       "        1.0348219871520996,\n",
       "        1.0247327089309692,\n",
       "        1.0146008729934692,\n",
       "        1.0047441720962524 ] } }"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fitDataset(dataset, {epochs: 10});"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Javascript (Node.js)",
   "language": "javascript",
   "name": "javascript"
  },
  "language_info": {
   "file_extension": ".js",
   "mimetype": "application/javascript",
   "name": "javascript",
   "version": "8.11.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}