federated-learning-based-RS.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "8f15be42",
   "metadata": {},
   "outputs": [],
   "source": [
    "import collections\n",
    "import functools\n",
    "import statistics\n",
    "from typing import Callable, Iterator, List, Optional, Tuple\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import nest_asyncio\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "import tensorflow_federated as tff\n",
    "from mizani.breaks import date_breaks\n",
    "from mizani.formatters import date_format\n",
    "from plotnine import *\n",
    "from surprise import SVD\n",
    "from surprise import Dataset, Reader\n",
    "from surprise.model_selection import cross_validate"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "59c513d5",
   "metadata": {},
   "source": [
    "# Table of Contents\n",
    "\n",
    "- [Step 1](#Step-1)\n",
    "- [Step 2](#Step-2-+-Step-3)\n",
    "- [Step 3](#Step-2-+-Step-3)\n",
    " * [Substep: EDA](#Substep:-EDA)\n",
    "- [Step 4](#Step-4)\n",
    "- [Step 5](#Step-5)\n",
    "- [Step 6](#Step-6)\n",
    "- [Step 7](#Step-7)\n",
    "- [Step 8](#Step-8)\n",
    "- [Step 9](#Step-9)\n",
    "- [Step 10](#Step-10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "d2c97ce8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def inspect_df(df: pd.DataFrame, n: int = 5) -> pd.DataFrame:\n",
    "    \"\"\"Helper method to easily inspect DataFrames.\"\"\"\n",
    "\n",
    "    print(f\"shape: {df.shape}\")\n",
    "\n",
    "    return df.head(n)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "734ab202",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2022-03-29 22:45:30--  https://files.grouplens.org/datasets/movielens/ml-latest-small.zip\n",
      "Resolving files.grouplens.org (files.grouplens.org)... 128.101.65.152\n",
      "Connecting to files.grouplens.org (files.grouplens.org)|128.101.65.152|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 978202 (955K) [application/zip]\n",
      "Saving to: ‘data/ml-latest-small.zip’\n",
      "\n",
      "ml-latest-small.zip 100%[===================>] 955.28K   315KB/s    in 3.0s    \n",
      "\n",
      "2022-03-29 22:45:33 (315 KB/s) - ‘data/ml-latest-small.zip’ saved [978202/978202]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# disclaimer: may not work as is in Windows OS\n",
    "# dataset source: https://grouplens.org/datasets/movielens/\n",
    "!wget --backups=1 https://files.grouplens.org/datasets/movielens/ml-latest-small.zip -P data/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "6bc846ed",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Archive:  data/ml-latest-small.zip\r\n",
      "  inflating: data/ml-latest-small/links.csv  \r\n",
      "  inflating: data/ml-latest-small/tags.csv  \r\n",
      "  inflating: data/ml-latest-small/ratings.csv  \r\n",
      "  inflating: data/ml-latest-small/README.txt  \r\n",
      "  inflating: data/ml-latest-small/movies.csv  \r\n"
     ]
    }
   ],
   "source": [
    "# disclaimer: may not work as is in Windows OS\n",
    "# if `unzip` does not exist in your system: apt-get install unzip\n",
    "!unzip -o data/ml-latest-small.zip -d data/"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56f05686",
   "metadata": {},
   "source": [
    "## Step 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "83c43429",
   "metadata": {},
   "source": [
    "The prediction $r_{ui}$ is set as:\n",
    "\n",
    "$$ r_{ui} = \\mu + b_u + b_i + q_{i}^T p_u$$\n",
    "\n",
    "If user $u$ is unknown, then the bias $b_u$ and the factors $p_u$ are assumed to be zero. \n",
    "\n",
    "The same applies for item $i$ with $b_i$ and $q_i$.\n",
    "\n",
    "To estimate all the unknown, we minimize the following regularized squared error:\n",
    "\n",
    "$$ \\sum_{r_{ui} \\in R_{train}} (r_{ui} - \\hat{r_{ui}})^2 + \\lambda (b_i^2 + b_u^2 + ||q_i||^2 + ||p_u||^2)$$\n",
    "\n",
    "The minimization is performed by a stochastic gradient descent:\n",
    "\n",
    "$$ b_u \\leftarrow b_u + \\gamma (e_{ui} - \\lambda b_u)$$\n",
    "$$ b_i \\leftarrow b_i + \\gamma (e_{ui} - \\lambda b_i)$$\n",
    "$$ p_u \\leftarrow p_u + \\gamma (e_{ui} q_i - \\lambda p_u)$$\n",
    "$$ q_i \\leftarrow q_i + \\gamma (e_{ui} p_u - \\lambda q_i)$$\n",
    "\n",
    "where: $e_{ui} = r_{ui} - \\hat{r_{ui}}$, $\\gamma$ is the learning rate and $\\lambda$ the regularization term. These steps are performed over all the ratings of the trainset and repeated $n$ times, where $n$ is the number of epochs.\n",
    "\n",
    "Baselines are initialized to zero. User and item factors are randomly initialized according to a normal distribution.\n",
    "\n",
    "[reference](https://surprise.readthedocs.io/en/stable/matrix_factorization.html#surprise.prediction_algorithms.matrix_factorization.SVD)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "15066828",
   "metadata": {},
   "source": [
    "## Step 2 + Step 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f9427823",
   "metadata": {},
   "outputs": [],
   "source": [
    "reader = Reader(\n",
    "    line_format=\"user item rating timestamp\", sep=\",\", skip_lines=1, rating_scale=(1, 5)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9e3da310",
   "metadata": {},
   "outputs": [],
   "source": [
    "data = Dataset.load_from_file(\"data/ml-latest-small/ratings.csv\", reader=reader)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "269efcb1",
   "metadata": {},
   "source": [
    "#### Substep: EDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ca14c48d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape: (100836, 4)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user</th>\n",
       "      <th>item</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964982703</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964981247</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964982224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>5.0</td>\n",
       "      <td>964983815</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>50</td>\n",
       "      <td>5.0</td>\n",
       "      <td>964982931</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user  item  rating  timestamp\n",
       "0     1     1     4.0  964982703\n",
       "1     1     3     4.0  964981247\n",
       "2     1     6     4.0  964982224\n",
       "3     1    47     5.0  964983815\n",
       "4     1    50     5.0  964982931"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "movielens = pd.read_csv(\n",
    "    \"data/ml-latest-small/ratings.csv\",\n",
    "    names=[\"user\", \"item\", \"rating\", \"timestamp\"],\n",
    "    skiprows=1,\n",
    ")\n",
    "\n",
    "inspect_df(movielens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c6c7ecc2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user</th>\n",
       "      <th>item</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>234</th>\n",
       "      <td>2</td>\n",
       "      <td>1704</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1445715228</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>237</th>\n",
       "      <td>2</td>\n",
       "      <td>8798</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1445714960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>240</th>\n",
       "      <td>2</td>\n",
       "      <td>58559</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1445715141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>242</th>\n",
       "      <td>2</td>\n",
       "      <td>68157</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1445715154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>247</th>\n",
       "      <td>2</td>\n",
       "      <td>80489</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1445715340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100820</th>\n",
       "      <td>610</td>\n",
       "      <td>160341</td>\n",
       "      <td>2.5</td>\n",
       "      <td>1479545749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100821</th>\n",
       "      <td>610</td>\n",
       "      <td>160527</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1479544998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100826</th>\n",
       "      <td>610</td>\n",
       "      <td>162350</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1493849971</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100827</th>\n",
       "      <td>610</td>\n",
       "      <td>163937</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1493848789</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100828</th>\n",
       "      <td>610</td>\n",
       "      <td>163981</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1493850155</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>30398 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        user    item  rating   timestamp\n",
       "234        2    1704     4.5  1445715228\n",
       "237        2    8798     3.5  1445714960\n",
       "240        2   58559     4.5  1445715141\n",
       "242        2   68157     4.5  1445715154\n",
       "247        2   80489     4.5  1445715340\n",
       "...      ...     ...     ...         ...\n",
       "100820   610  160341     2.5  1479545749\n",
       "100821   610  160527     4.5  1479544998\n",
       "100826   610  162350     3.5  1493849971\n",
       "100827   610  163937     3.5  1493848789\n",
       "100828   610  163981     3.5  1493850155\n",
       "\n",
       "[30398 rows x 4 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "movielens[movielens[\"rating\"].isin([0.5, 1.5, 2.5, 3.5, 4.5])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "cf75d7f9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of users: 610\n"
     ]
    }
   ],
   "source": [
    "print(f\"number of users: {len(movielens['user'].unique())}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "65708b1b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of items: 9724\n"
     ]
    }
   ],
   "source": [
    "print(f\"number of items: {len(movielens['item'].unique())}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3ddc16c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "movielens[\"created_at\"] = pd.to_datetime(movielens[\"timestamp\"], unit=\"s\")\n",
    "movielens[\"created_at\"] = pd.to_datetime(movielens[\"created_at\"], format=\"%Y-%m\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "342f7dea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (8773687233172)>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(\n",
    "    ggplot(movielens, aes(x=\"rating\"))\n",
    "    + geom_bar(color=\"black\")\n",
    "    + labs(x=\"rating\", title=\"distribution of ratings\")\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "2489f352",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1400x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (8773566562467)>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(\n",
    "    ggplot(\n",
    "        movielens.groupby(pd.Grouper(key=\"created_at\", freq=\"M\")).count().reset_index()\n",
    "    )\n",
    "    + geom_line(aes(x=\"created_at\", y=\"item\"), color=\"crimson\")\n",
    "    + labs(x=\"date\", y=\"count\", title=\"#ratings: evolution through the years\")\n",
    "    + scale_x_datetime(breaks=date_breaks(\"6 months\"), labels=date_format(\"%Y/%m\"))\n",
    "    + theme(axis_text_x=element_text(rotation=45, hjust=1), figure_size=(14, 7))\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3abbe16",
   "metadata": {},
   "source": [
    "## Step 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "17c416bf",
   "metadata": {},
   "outputs": [],
   "source": [
    "recsys_small = SVD(\n",
    "    n_factors=4,\n",
    "    n_epochs=50,\n",
    "    biased=True,  # use baselines\n",
    "    random_state=123456,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "eea20aef",
   "metadata": {},
   "outputs": [],
   "source": [
    "recsys_large = SVD(\n",
    "    n_factors=20,\n",
    "    n_epochs=50,\n",
    "    biased=True,  # use baselines\n",
    "    random_state=123456,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "917b2e90",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating MSE, RMSE, MAE of algorithm SVD on 5 split(s).\n",
      "\n",
      "                  Fold 1  Fold 2  Fold 3  Fold 4  Fold 5  Mean    Std     \n",
      "MSE (testset)     0.7580  0.7693  0.7609  0.7574  0.7652  0.7622  0.0045  \n",
      "RMSE (testset)    0.8707  0.8771  0.8723  0.8703  0.8747  0.8730  0.0026  \n",
      "MAE (testset)     0.6647  0.6702  0.6690  0.6644  0.6681  0.6673  0.0023  \n",
      "MSE (trainset)    0.5122  0.5030  0.5075  0.5085  0.5037  0.5070  0.0034  \n",
      "RMSE (trainset)   0.7157  0.7092  0.7124  0.7131  0.7097  0.7120  0.0024  \n",
      "MAE (trainset)    0.5514  0.5458  0.5469  0.5488  0.5460  0.5478  0.0021  \n",
      "Fit time          1.70    1.71    1.69    1.68    1.68    1.69    0.01    \n",
      "Test time         0.09    0.09    0.09    0.10    0.10    0.09    0.00    \n"
     ]
    }
   ],
   "source": [
    "cv_results = {}\n",
    "\n",
    "cv_results[\"SVD w/ 4 factors\"] = cross_validate(\n",
    "    recsys_small,\n",
    "    data=data,\n",
    "    measures=[\"mse\", \"rmse\", \"mae\"],\n",
    "    return_train_measures=True,\n",
    "    verbose=True,\n",
    "    cv=5,\n",
    "    n_jobs=-1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "1fcc3326",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating MSE, RMSE, MAE of algorithm SVD on 5 split(s).\n",
      "\n",
      "                  Fold 1  Fold 2  Fold 3  Fold 4  Fold 5  Mean    Std     \n",
      "MSE (testset)     0.7823  0.7784  0.7886  0.7773  0.7846  0.7822  0.0041  \n",
      "RMSE (testset)    0.8845  0.8823  0.8880  0.8817  0.8858  0.8844  0.0023  \n",
      "MAE (testset)     0.6741  0.6735  0.6790  0.6763  0.6786  0.6763  0.0022  \n",
      "MSE (trainset)    0.3015  0.3050  0.2999  0.3041  0.3005  0.3022  0.0020  \n",
      "RMSE (trainset)   0.5491  0.5523  0.5476  0.5515  0.5482  0.5497  0.0018  \n",
      "MAE (trainset)    0.4242  0.4262  0.4225  0.4257  0.4222  0.4241  0.0016  \n",
      "Fit time          3.37    3.51    3.32    3.29    3.26    3.35    0.09    \n",
      "Test time         0.11    0.10    0.12    0.10    0.11    0.11    0.01    \n"
     ]
    }
   ],
   "source": [
    "cv_results[\"SVD w/ 20 factors\"] = cross_validate(\n",
    "    recsys_large,\n",
    "    data=data,\n",
    "    measures=[\"mse\", \"rmse\", \"mae\"],\n",
    "    return_train_measures=True,\n",
    "    verbose=True,\n",
    "    cv=5,\n",
    "    n_jobs=-1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "28ceac08",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.DataFrame(cv_results).T.reset_index()\n",
    "\n",
    "df[\"mean RMSE (train)\"] = df[\"train_rmse\"].map(lambda x: statistics.mean(x))\n",
    "df[\"mean RMSE (test)\"] = df[\"test_rmse\"].map(lambda x: statistics.mean(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "596350da",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (8773566197998)>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(\n",
    "    ggplot(df)\n",
    "    + geom_col(aes(x=\"index\", y=\"mean RMSE (train)\"), stat=\"identity\")\n",
    "    + theme(axis_text_x=element_text(rotation=45, hjust=1), figure_size=(8, 6))\n",
    "    + labs(x=\"\", y=\"mean RMSE (train)\")\n",
    "    + coord_flip()\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "7a4a5af0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (8773678482682)>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(\n",
    "    ggplot(df)\n",
    "    + geom_col(aes(x=\"index\", y=\"mean RMSE (test)\"), stat=\"identity\")\n",
    "    + theme(axis_text_x=element_text(rotation=45, hjust=1), figure_size=(8, 6))\n",
    "    + labs(x=\"\", y=\"mean RMSE (test)\")\n",
    "    + coord_flip()\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "02bf0f4f",
   "metadata": {},
   "source": [
    "## Step 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "0bdaf204",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape: (100836, 3)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user</th>\n",
       "      <th>item</th>\n",
       "      <th>rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>50</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user  item  rating\n",
       "0     1     1     4.0\n",
       "1     1     3     4.0\n",
       "2     1     6     4.0\n",
       "3     1    47     5.0\n",
       "4     1    50     5.0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings = pd.read_csv(\n",
    "    \"data/ml-latest-small/ratings.csv\",\n",
    "    usecols=[\"user\", \"item\", \"rating\"],\n",
    "    names=[\"user\", \"item\", \"rating\"],\n",
    "    skiprows=1,\n",
    ")\n",
    "\n",
    "inspect_df(ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "b9cf6e82",
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings[\"user\"] = ratings.groupby(\"user\").ngroup()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "6db13a0b",
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings[\"item\"] = ratings.groupby(\"item\").ngroup()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "f8a883f8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def transform_batch(batch: tf.Tensor) -> collections.OrderedDict:\n",
    "    return collections.OrderedDict(\n",
    "        x=tf.cast(batch[:, 0:2], tf.int64), y=tf.cast(batch[:, 2:3], tf.float32)\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "6ecd6f4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_tf_datasets(\n",
    "    data: pd.DataFrame,\n",
    "    transform_func: Callable,\n",
    "    batch_size: int = 1,\n",
    "    max_clients: int = None,\n",
    "    max_items_per_client: int = None,\n",
    ") -> Iterator[tf.data.Dataset]:\n",
    "\n",
    "    n_users = len(data[\"user\"].unique())\n",
    "\n",
    "    if max_clients is not None:\n",
    "        n_users = min(n_users, max_clients)\n",
    "\n",
    "    for _, user_ratings in data.groupby(\"user\"):\n",
    "\n",
    "        if n_users < 0:\n",
    "            break\n",
    "\n",
    "        tf_dataset = tf.data.Dataset.from_tensor_slices(user_ratings)\n",
    "\n",
    "        yield (\n",
    "            tf_dataset.take(max_items_per_client)\n",
    "            .shuffle(buffer_size=max_items_per_client, seed=42)\n",
    "            .batch(batch_size)\n",
    "            .map(transform_func, num_parallel_calls=tf.data.experimental.AUTOTUNE)\n",
    "        )\n",
    "\n",
    "        n_users -= 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "ac77a7bf",
   "metadata": {},
   "outputs": [],
   "source": [
    "def split_tf_datasets(\n",
    "    tf_datasets: List[tf.data.Dataset],\n",
    "    train_fraction: float = 0.8,\n",
    ") -> Tuple[List[tf.data.Dataset], List[tf.data.Dataset]]:\n",
    "    \"\"\"Splits a list of user TF datasets into train/test by user.\"\"\"\n",
    "    np.random.seed(42)\n",
    "    np.random.shuffle(tf_datasets)\n",
    "\n",
    "    train_idx = int(len(tf_datasets) * train_fraction)\n",
    "\n",
    "    # Note that the test data contains completely different users, not\n",
    "    # just unseen ratings from train users.\n",
    "    return (\n",
    "        tf_datasets[:train_idx],\n",
    "        tf_datasets[train_idx:],\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "b575fa05",
   "metadata": {},
   "outputs": [],
   "source": [
    "tf_datasets = [\n",
    "    *create_tf_datasets(\n",
    "        data=ratings,\n",
    "        transform_func=transform_batch,\n",
    "        batch_size=10,\n",
    "        max_clients=100,\n",
    "        max_items_per_client=3000,\n",
    "    )\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "88d3e0e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "tf_trainsets, tf_testsets = split_tf_datasets(tf_datasets, train_fraction=0.8)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2601cc29",
   "metadata": {},
   "source": [
    "## Step 6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "744d0c9b",
   "metadata": {},
   "outputs": [],
   "source": [
    "nest_asyncio.apply()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "2ea601d5",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_users = len(ratings[\"user\"].unique())\n",
    "n_items = len(ratings[\"item\"].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "f6050674",
   "metadata": {},
   "outputs": [],
   "source": [
    "def matrix_factorization_model(\n",
    "    n_users: int,\n",
    "    n_items: int,\n",
    "    n_latent_factors: int,\n",
    ") -> tf.keras.Model:\n",
    "    \"\"\"A federation-friendly Matrix Factorization model.\"\"\"\n",
    "    combined_input = tf.keras.layers.Input(\n",
    "        shape=[2], dtype=\"int32\", name=\"combined_input\"\n",
    "    )\n",
    "    u = tf.keras.layers.Embedding(\n",
    "        input_dim=n_users + 1, output_dim=n_latent_factors, name=\"users_embedding\"\n",
    "    )(combined_input[:, 0])\n",
    "    u = tf.keras.layers.Flatten()(u)\n",
    "\n",
    "    i = tf.keras.layers.Embedding(\n",
    "        input_dim=n_items + 1, output_dim=n_latent_factors, name=\"items_embedding\"\n",
    "    )(combined_input[:, 1])\n",
    "    i = tf.keras.layers.Flatten()(i)\n",
    "\n",
    "    output = tf.output_layer = tf.keras.layers.Dot(1, normalize=False, name=\"Dot\")(\n",
    "        [i, u]\n",
    "    )\n",
    "\n",
    "    model = tf.keras.Model(inputs=combined_input, outputs=output)\n",
    "\n",
    "    input_spec = collections.OrderedDict(\n",
    "        x=tf.TensorSpec(shape=[None, 2], dtype=tf.int64),\n",
    "        y=tf.TensorSpec(shape=[None, 1], dtype=tf.float32),\n",
    "    )\n",
    "\n",
    "    return tff.learning.from_keras_model(\n",
    "        model,\n",
    "        input_spec=input_spec,\n",
    "        loss=tf.keras.losses.MeanSquaredError(),\n",
    "        metrics=[tf.keras.metrics.RootMeanSquaredError()],\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "2fb19c41",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_fn = functools.partial(\n",
    "    matrix_factorization_model,\n",
    "    n_users=n_users,\n",
    "    n_items=n_items,\n",
    "    n_latent_factors=4,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "5cec8717",
   "metadata": {},
   "outputs": [],
   "source": [
    "iterative_process = tff.learning.build_federated_averaging_process(\n",
    "    model_fn,\n",
    "    client_optimizer_fn=lambda: tf.keras.optimizers.SGD(learning_rate=0.02),\n",
    "    server_optimizer_fn=lambda: tf.keras.optimizers.SGD(learning_rate=1.0),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "c9244c9b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "( -> <\n",
      "  model=<\n",
      "    trainable=<\n",
      "      float32[9725,4],\n",
      "      float32[611,4]\n",
      "    >,\n",
      "    non_trainable=<>\n",
      "  >,\n",
      "  optimizer_state=<\n",
      "    int64\n",
      "  >,\n",
      "  delta_aggregate_state=<\n",
      "    value_sum_process=<>,\n",
      "    weight_sum_process=<>\n",
      "  >,\n",
      "  model_broadcast_state=<>\n",
      ">@SERVER)\n"
     ]
    }
   ],
   "source": [
    "print(iterative_process.initialize.type_signature.formatted_representation())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "eb31d2d2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ModelWeights(trainable=[array([[ 0.01468258,  0.00567931,  0.00495758,  0.00324589],\n",
      "       [ 0.03179849, -0.0492647 , -0.02005481, -0.04163704],\n",
      "       [-0.03475098, -0.01713616,  0.01399508,  0.04243547],\n",
      "       ...,\n",
      "       [ 0.00862901,  0.00322223,  0.01444075,  0.04583659],\n",
      "       [ 0.03038353,  0.02589783,  0.00930526,  0.02227395],\n",
      "       [ 0.01414761, -0.00780897,  0.04580608,  0.02988801]],\n",
      "      dtype=float32), array([[-0.02925836, -0.01624753,  0.03568989,  0.0261209 ],\n",
      "       [ 0.01137955,  0.03505001, -0.00913109, -0.0486768 ],\n",
      "       [ 0.02154956,  0.03719026, -0.04055684, -0.04193204],\n",
      "       ...,\n",
      "       [-0.04047085,  0.02363309,  0.01748785, -0.01520915],\n",
      "       [-0.02010256,  0.00147226,  0.0054672 , -0.00792523],\n",
      "       [ 0.03594568,  0.01957921,  0.04275305,  0.00249139]],\n",
      "      dtype=float32)], non_trainable=[])\n"
     ]
    }
   ],
   "source": [
    "state = iterative_process.initialize()\n",
    "\n",
    "print(state.model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "d2af493a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "User variables shape: (611, 4)\n"
     ]
    }
   ],
   "source": [
    "print(\"User variables shape:\", state.model.trainable[1].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "cdfb5694",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Item variables shape: (9725, 4)\n"
     ]
    }
   ],
   "source": [
    "print(\"Item variables shape:\", state.model.trainable[0].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "a8d5c961",
   "metadata": {},
   "outputs": [],
   "source": [
    "evaluation_computation = tff.learning.build_federated_evaluation(model_fn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "d6cb7370",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<\n",
      "  server_model_weights=<\n",
      "    trainable=<\n",
      "      float32[9725,4],\n",
      "      float32[611,4]\n",
      "    >,\n",
      "    non_trainable=<>\n",
      "  >@SERVER,\n",
      "  federated_dataset={<\n",
      "    x=int64[?,2],\n",
      "    y=float32[?,1]\n",
      "  >*}@CLIENTS\n",
      "> -> <\n",
      "  eval=<\n",
      "    root_mean_squared_error=float32,\n",
      "    loss=float32,\n",
      "    num_examples=int64,\n",
      "    num_batches=int64\n",
      "  >\n",
      ">@SERVER)\n"
     ]
    }
   ],
   "source": [
    "print(evaluation_computation.type_signature.formatted_representation())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "b8e7b6ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "NUM_ROUNDS = 200  # we will set up a training loop to train over multiple rounds."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "29cfe58b",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 0: OrderedDict([('root_mean_squared_error', 3.6591628), ('loss', 13.389473), ('num_examples', 8644), ('num_batches', 888)])\n",
      "Train round 1: OrderedDict([('root_mean_squared_error', 3.6203775), ('loss', 13.107134), ('num_examples', 8908), ('num_batches', 917)])\n",
      "Train round 2: OrderedDict([('root_mean_squared_error', 3.6475613), ('loss', 13.304705), ('num_examples', 9100), ('num_batches', 932)])\n",
      "Train round 3: OrderedDict([('root_mean_squared_error', 3.6928227), ('loss', 13.636939), ('num_examples', 7853), ('num_batches', 811)])\n",
      "Train round 4: OrderedDict([('root_mean_squared_error', 3.6223826), ('loss', 13.121655), ('num_examples', 9232), ('num_batches', 947)])\n",
      "Train round 5: OrderedDict([('root_mean_squared_error', 3.682296), ('loss', 13.559304), ('num_examples', 8628), ('num_batches', 887)])\n",
      "Train round 6: OrderedDict([('root_mean_squared_error', 3.7147858), ('loss', 13.799635), ('num_examples', 9198), ('num_batches', 943)])\n",
      "Train round 7: OrderedDict([('root_mean_squared_error', 3.6334302), ('loss', 13.201815), ('num_examples', 8964), ('num_batches', 920)])\n",
      "Train round 8: OrderedDict([('root_mean_squared_error', 3.739837), ('loss', 13.986381), ('num_examples', 8255), ('num_batches', 851)])\n",
      "Train round 9: OrderedDict([('root_mean_squared_error', 3.723842), ('loss', 13.866998), ('num_examples', 8955), ('num_batches', 918)])\n",
      "Train round 10: OrderedDict([('root_mean_squared_error', 3.6539574), ('loss', 13.351404), ('num_examples', 9013), ('num_batches', 925)])\n",
      "Train round 11: OrderedDict([('root_mean_squared_error', 3.6886353), ('loss', 13.60603), ('num_examples', 8700), ('num_batches', 890)])\n",
      "Train round 12: OrderedDict([('root_mean_squared_error', 3.7174873), ('loss', 13.819713), ('num_examples', 7784), ('num_batches', 801)])\n",
      "Train round 13: OrderedDict([('root_mean_squared_error', 3.811807), ('loss', 14.529873), ('num_examples', 6809), ('num_batches', 706)])\n",
      "Train round 14: OrderedDict([('root_mean_squared_error', 3.7484138), ('loss', 14.050607), ('num_examples', 6496), ('num_batches', 673)])\n",
      "Train round 15: OrderedDict([('root_mean_squared_error', 3.7378972), ('loss', 13.971875), ('num_examples', 7801), ('num_batches', 805)])\n",
      "Train round 16: OrderedDict([('root_mean_squared_error', 3.6660635), ('loss', 13.440021), ('num_examples', 8958), ('num_batches', 919)])\n",
      "Train round 17: OrderedDict([('root_mean_squared_error', 3.664927), ('loss', 13.43169), ('num_examples', 8425), ('num_batches', 865)])\n",
      "Train round 18: OrderedDict([('root_mean_squared_error', 3.6878982), ('loss', 13.600593), ('num_examples', 7861), ('num_batches', 808)])\n",
      "Train round 19: OrderedDict([('root_mean_squared_error', 3.64611), ('loss', 13.294119), ('num_examples', 8687), ('num_batches', 892)])\n",
      "Train round 20: OrderedDict([('root_mean_squared_error', 3.962864), ('loss', 15.704291), ('num_examples', 5689), ('num_batches', 593)])\n",
      "Train round 21: OrderedDict([('root_mean_squared_error', 3.7502804), ('loss', 14.064603), ('num_examples', 8108), ('num_batches', 834)])\n",
      "Train round 22: OrderedDict([('root_mean_squared_error', 3.7337413), ('loss', 13.9408245), ('num_examples', 6601), ('num_batches', 686)])\n",
      "Train round 23: OrderedDict([('root_mean_squared_error', 3.6602836), ('loss', 13.3976755), ('num_examples', 9762), ('num_batches', 1000)])\n",
      "Train round 24: OrderedDict([('root_mean_squared_error', 3.67574), ('loss', 13.511065), ('num_examples', 9136), ('num_batches', 939)])\n",
      "Train round 25: OrderedDict([('root_mean_squared_error', 3.8084397), ('loss', 14.504213), ('num_examples', 7066), ('num_batches', 730)])\n",
      "Train round 26: OrderedDict([('root_mean_squared_error', 3.6823215), ('loss', 13.559492), ('num_examples', 7223), ('num_batches', 747)])\n",
      "Train round 27: OrderedDict([('root_mean_squared_error', 3.8015914), ('loss', 14.452097), ('num_examples', 6879), ('num_batches', 711)])\n",
      "Train round 28: OrderedDict([('root_mean_squared_error', 3.6549137), ('loss', 13.358393), ('num_examples', 9590), ('num_batches', 983)])\n",
      "Train round 29: OrderedDict([('root_mean_squared_error', 3.702004), ('loss', 13.704833), ('num_examples', 8166), ('num_batches', 842)])\n",
      "Train round 30: OrderedDict([('root_mean_squared_error', 3.6087236), ('loss', 13.022887), ('num_examples', 8568), ('num_batches', 879)])\n",
      "Train round 31: OrderedDict([('root_mean_squared_error', 3.684204), ('loss', 13.57336), ('num_examples', 8438), ('num_batches', 868)])\n",
      "Train round 32: OrderedDict([('root_mean_squared_error', 3.6723416), ('loss', 13.486092), ('num_examples', 9541), ('num_batches', 979)])\n",
      "Train round 33: OrderedDict([('root_mean_squared_error', 3.6569965), ('loss', 13.373623), ('num_examples', 7322), ('num_batches', 756)])\n",
      "Train round 34: OrderedDict([('root_mean_squared_error', 3.7816293), ('loss', 14.30072), ('num_examples', 7761), ('num_batches', 798)])\n",
      "Train round 35: OrderedDict([('root_mean_squared_error', 3.656026), ('loss', 13.366525), ('num_examples', 9494), ('num_batches', 973)])\n",
      "Train round 36: OrderedDict([('root_mean_squared_error', 3.7702253), ('loss', 14.2146), ('num_examples', 8095), ('num_batches', 832)])\n",
      "Train round 37: OrderedDict([('root_mean_squared_error', 3.7252617), ('loss', 13.877576), ('num_examples', 7972), ('num_batches', 822)])\n",
      "Train round 38: OrderedDict([('root_mean_squared_error', 3.7053854), ('loss', 13.729881), ('num_examples', 7593), ('num_batches', 784)])\n",
      "Train round 39: OrderedDict([('root_mean_squared_error', 3.7589848), ('loss', 14.129967), ('num_examples', 6754), ('num_batches', 702)])\n",
      "Train round 40: OrderedDict([('root_mean_squared_error', 3.8193574), ('loss', 14.587491), ('num_examples', 7282), ('num_batches', 754)])\n",
      "Train round 41: OrderedDict([('root_mean_squared_error', 3.6422112), ('loss', 13.265703), ('num_examples', 8483), ('num_batches', 873)])\n",
      "Train round 42: OrderedDict([('root_mean_squared_error', 3.7374535), ('loss', 13.968559), ('num_examples', 7528), ('num_batches', 776)])\n",
      "Train round 43: OrderedDict([('root_mean_squared_error', 3.7570102), ('loss', 14.115126), ('num_examples', 7943), ('num_batches', 820)])\n",
      "Train round 44: OrderedDict([('root_mean_squared_error', 3.634956), ('loss', 13.212905), ('num_examples', 9859), ('num_batches', 1007)])\n",
      "Train round 45: OrderedDict([('root_mean_squared_error', 3.674501), ('loss', 13.501958), ('num_examples', 7269), ('num_batches', 751)])\n",
      "Train round 46: OrderedDict([('root_mean_squared_error', 3.7057426), ('loss', 13.732529), ('num_examples', 7746), ('num_batches', 799)])\n",
      "Train round 47: OrderedDict([('root_mean_squared_error', 3.7090235), ('loss', 13.756855), ('num_examples', 8504), ('num_batches', 875)])\n",
      "Train round 48: OrderedDict([('root_mean_squared_error', 3.7327347), ('loss', 13.933308), ('num_examples', 7268), ('num_batches', 750)])\n",
      "Train round 49: OrderedDict([('root_mean_squared_error', 3.7948282), ('loss', 14.400722), ('num_examples', 6793), ('num_batches', 702)])\n",
      "Train round 50: OrderedDict([('root_mean_squared_error', 3.7356174), ('loss', 13.954837), ('num_examples', 8457), ('num_batches', 869)])\n",
      "Train round 51: OrderedDict([('root_mean_squared_error', 3.7032526), ('loss', 13.71408), ('num_examples', 8277), ('num_batches', 853)])\n",
      "Train round 52: OrderedDict([('root_mean_squared_error', 3.6907763), ('loss', 13.62183), ('num_examples', 8520), ('num_batches', 873)])\n",
      "Train round 53: OrderedDict([('root_mean_squared_error', 3.6336803), ('loss', 13.203633), ('num_examples', 8247), ('num_batches', 848)])\n",
      "Train round 54: OrderedDict([('root_mean_squared_error', 3.6875064), ('loss', 13.597703), ('num_examples', 7565), ('num_batches', 782)])\n",
      "Train round 55: OrderedDict([('root_mean_squared_error', 3.6925838), ('loss', 13.635178), ('num_examples', 8913), ('num_batches', 915)])\n",
      "Train round 56: OrderedDict([('root_mean_squared_error', 3.7225816), ('loss', 13.8576145), ('num_examples', 8549), ('num_batches', 879)])\n",
      "Train round 57: OrderedDict([('root_mean_squared_error', 3.6325462), ('loss', 13.195392), ('num_examples', 9357), ('num_batches', 959)])\n",
      "Train round 58: OrderedDict([('root_mean_squared_error', 3.7623813), ('loss', 14.155513), ('num_examples', 6712), ('num_batches', 695)])\n",
      "Train round 59: OrderedDict([('root_mean_squared_error', 3.7048562), ('loss', 13.725959), ('num_examples', 7882), ('num_batches', 813)])\n",
      "Train round 60: OrderedDict([('root_mean_squared_error', 3.6488593), ('loss', 13.314174), ('num_examples', 8618), ('num_batches', 884)])\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 61: OrderedDict([('root_mean_squared_error', 3.762341), ('loss', 14.15521), ('num_examples', 8117), ('num_batches', 836)])\n",
      "Train round 62: OrderedDict([('root_mean_squared_error', 3.7075777), ('loss', 13.746133), ('num_examples', 8083), ('num_batches', 831)])\n",
      "Train round 63: OrderedDict([('root_mean_squared_error', 3.685967), ('loss', 13.586352), ('num_examples', 7246), ('num_batches', 750)])\n",
      "Train round 64: OrderedDict([('root_mean_squared_error', 3.6558838), ('loss', 13.365486), ('num_examples', 9553), ('num_batches', 979)])\n",
      "Train round 65: OrderedDict([('root_mean_squared_error', 3.6361732), ('loss', 13.221756), ('num_examples', 8580), ('num_batches', 883)])\n",
      "Train round 66: OrderedDict([('root_mean_squared_error', 3.6869874), ('loss', 13.593877), ('num_examples', 7509), ('num_batches', 773)])\n",
      "Train round 67: OrderedDict([('root_mean_squared_error', 3.6451082), ('loss', 13.286815), ('num_examples', 9124), ('num_batches', 933)])\n",
      "Train round 68: OrderedDict([('root_mean_squared_error', 3.6647522), ('loss', 13.430409), ('num_examples', 7913), ('num_batches', 818)])\n",
      "Train round 69: OrderedDict([('root_mean_squared_error', 3.6932847), ('loss', 13.640352), ('num_examples', 8945), ('num_batches', 919)])\n",
      "Train round 70: OrderedDict([('root_mean_squared_error', 3.6120346), ('loss', 13.046794), ('num_examples', 8295), ('num_batches', 852)])\n",
      "Train round 71: OrderedDict([('root_mean_squared_error', 3.6791992), ('loss', 13.536507), ('num_examples', 8651), ('num_batches', 888)])\n",
      "Train round 72: OrderedDict([('root_mean_squared_error', 3.6786215), ('loss', 13.532257), ('num_examples', 7878), ('num_batches', 812)])\n",
      "Train round 73: OrderedDict([('root_mean_squared_error', 3.642324), ('loss', 13.266525), ('num_examples', 8332), ('num_batches', 860)])\n",
      "Train round 74: OrderedDict([('root_mean_squared_error', 3.8625987), ('loss', 14.919669), ('num_examples', 6991), ('num_batches', 725)])\n",
      "Train round 75: OrderedDict([('root_mean_squared_error', 3.6561816), ('loss', 13.367662), ('num_examples', 7837), ('num_batches', 807)])\n",
      "Train round 76: OrderedDict([('root_mean_squared_error', 3.7721555), ('loss', 14.229157), ('num_examples', 7791), ('num_batches', 802)])\n",
      "Train round 77: OrderedDict([('root_mean_squared_error', 3.6417613), ('loss', 13.262425), ('num_examples', 9149), ('num_batches', 938)])\n",
      "Train round 78: OrderedDict([('root_mean_squared_error', 3.6236215), ('loss', 13.130632), ('num_examples', 8744), ('num_batches', 898)])\n",
      "Train round 79: OrderedDict([('root_mean_squared_error', 3.6204095), ('loss', 13.107365), ('num_examples', 9267), ('num_batches', 949)])\n",
      "Train round 80: OrderedDict([('root_mean_squared_error', 3.7247353), ('loss', 13.8736515), ('num_examples', 7336), ('num_batches', 756)])\n",
      "Train round 81: OrderedDict([('root_mean_squared_error', 3.741287), ('loss', 13.997229), ('num_examples', 7766), ('num_batches', 800)])\n",
      "Train round 82: OrderedDict([('root_mean_squared_error', 3.6019537), ('loss', 12.9740715), ('num_examples', 8255), ('num_batches', 848)])\n",
      "Train round 83: OrderedDict([('root_mean_squared_error', 3.7199862), ('loss', 13.838297), ('num_examples', 8224), ('num_batches', 846)])\n",
      "Train round 84: OrderedDict([('root_mean_squared_error', 3.6536655), ('loss', 13.349272), ('num_examples', 8729), ('num_batches', 896)])\n",
      "Train round 85: OrderedDict([('root_mean_squared_error', 3.574472), ('loss', 12.776852), ('num_examples', 8730), ('num_batches', 896)])\n",
      "Train round 86: OrderedDict([('root_mean_squared_error', 3.7521572), ('loss', 14.078685), ('num_examples', 7192), ('num_batches', 743)])\n",
      "Train round 87: OrderedDict([('root_mean_squared_error', 3.6122985), ('loss', 13.048701), ('num_examples', 8612), ('num_batches', 886)])\n",
      "Train round 88: OrderedDict([('root_mean_squared_error', 3.6806939), ('loss', 13.547508), ('num_examples', 9227), ('num_batches', 945)])\n",
      "Train round 89: OrderedDict([('root_mean_squared_error', 3.5980222), ('loss', 12.945764), ('num_examples', 8409), ('num_batches', 864)])\n",
      "Train round 90: OrderedDict([('root_mean_squared_error', 3.5925965), ('loss', 12.90675), ('num_examples', 8772), ('num_batches', 900)])\n",
      "Train round 91: OrderedDict([('root_mean_squared_error', 3.579993), ('loss', 12.81635), ('num_examples', 9457), ('num_batches', 971)])\n",
      "Train round 92: OrderedDict([('root_mean_squared_error', 3.7264152), ('loss', 13.88617), ('num_examples', 6466), ('num_batches', 671)])\n",
      "Train round 93: OrderedDict([('root_mean_squared_error', 3.5553367), ('loss', 12.640419), ('num_examples', 8942), ('num_batches', 918)])\n",
      "Train round 94: OrderedDict([('root_mean_squared_error', 3.5390487), ('loss', 12.524866), ('num_examples', 7670), ('num_batches', 790)])\n",
      "Train round 95: OrderedDict([('root_mean_squared_error', 3.7428987), ('loss', 14.009291), ('num_examples', 7123), ('num_batches', 738)])\n",
      "Train round 96: OrderedDict([('root_mean_squared_error', 3.620186), ('loss', 13.105747), ('num_examples', 7967), ('num_batches', 821)])\n",
      "Train round 97: OrderedDict([('root_mean_squared_error', 3.573356), ('loss', 12.768872), ('num_examples', 9208), ('num_batches', 943)])\n",
      "Train round 98: OrderedDict([('root_mean_squared_error', 3.768281), ('loss', 14.199942), ('num_examples', 8010), ('num_batches', 825)])\n",
      "Train round 99: OrderedDict([('root_mean_squared_error', 3.6350417), ('loss', 13.213528), ('num_examples', 7263), ('num_batches', 750)])\n",
      "Train round 100: OrderedDict([('root_mean_squared_error', 3.45036), ('loss', 11.904984), ('num_examples', 8769), ('num_batches', 902)])\n",
      "Train round 101: OrderedDict([('root_mean_squared_error', 3.6142645), ('loss', 13.062908), ('num_examples', 8336), ('num_batches', 857)])\n",
      "Train round 102: OrderedDict([('root_mean_squared_error', 3.584599), ('loss', 12.84935), ('num_examples', 8172), ('num_batches', 842)])\n",
      "Train round 103: OrderedDict([('root_mean_squared_error', 3.496488), ('loss', 12.225429), ('num_examples', 8976), ('num_batches', 921)])\n",
      "Train round 104: OrderedDict([('root_mean_squared_error', 3.469719), ('loss', 12.03895), ('num_examples', 8913), ('num_batches', 917)])\n",
      "Train round 105: OrderedDict([('root_mean_squared_error', 3.515269), ('loss', 12.357116), ('num_examples', 7792), ('num_batches', 803)])\n",
      "Train round 106: OrderedDict([('root_mean_squared_error', 3.5219202), ('loss', 12.403922), ('num_examples', 8990), ('num_batches', 922)])\n",
      "Train round 107: OrderedDict([('root_mean_squared_error', 3.584861), ('loss', 12.85123), ('num_examples', 8648), ('num_batches', 889)])\n",
      "Train round 108: OrderedDict([('root_mean_squared_error', 3.4127152), ('loss', 11.646626), ('num_examples', 8205), ('num_batches', 844)])\n",
      "Train round 109: OrderedDict([('root_mean_squared_error', 3.6398811), ('loss', 13.2487335), ('num_examples', 7737), ('num_batches', 799)])\n",
      "Train round 110: OrderedDict([('root_mean_squared_error', 3.6495922), ('loss', 13.319524), ('num_examples', 6955), ('num_batches', 719)])\n",
      "Train round 111: OrderedDict([('root_mean_squared_error', 3.3102596), ('loss', 10.957819), ('num_examples', 8459), ('num_batches', 869)])\n",
      "Train round 112: OrderedDict([('root_mean_squared_error', 3.6786213), ('loss', 13.532255), ('num_examples', 5860), ('num_batches', 612)])\n",
      "Train round 113: OrderedDict([('root_mean_squared_error', 3.3462229), ('loss', 11.197208), ('num_examples', 8797), ('num_batches', 902)])\n",
      "Train round 114: OrderedDict([('root_mean_squared_error', 3.4118736), ('loss', 11.6408825), ('num_examples', 9401), ('num_batches', 963)])\n",
      "Train round 115: OrderedDict([('root_mean_squared_error', 3.6723206), ('loss', 13.485938), ('num_examples', 6366), ('num_batches', 661)])\n",
      "Train round 116: OrderedDict([('root_mean_squared_error', 3.5400572), ('loss', 12.532005), ('num_examples', 8773), ('num_batches', 902)])\n",
      "Train round 117: OrderedDict([('root_mean_squared_error', 3.4065843), ('loss', 11.604816), ('num_examples', 7475), ('num_batches', 771)])\n",
      "Train round 118: OrderedDict([('root_mean_squared_error', 3.2958894), ('loss', 10.862886), ('num_examples', 8730), ('num_batches', 897)])\n",
      "Train round 119: OrderedDict([('root_mean_squared_error', 3.4399893), ('loss', 11.833528), ('num_examples', 7816), ('num_batches', 804)])\n",
      "Train round 120: OrderedDict([('root_mean_squared_error', 3.5696096), ('loss', 12.742113), ('num_examples', 6714), ('num_batches', 697)])\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 121: OrderedDict([('root_mean_squared_error', 3.3955605), ('loss', 11.529832), ('num_examples', 7966), ('num_batches', 819)])\n",
      "Train round 122: OrderedDict([('root_mean_squared_error', 3.4272811), ('loss', 11.746257), ('num_examples', 7438), ('num_batches', 768)])\n",
      "Train round 123: OrderedDict([('root_mean_squared_error', 3.2188706), ('loss', 10.361129), ('num_examples', 8850), ('num_batches', 910)])\n",
      "Train round 124: OrderedDict([('root_mean_squared_error', 3.2137053), ('loss', 10.327903), ('num_examples', 9814), ('num_batches', 1005)])\n",
      "Train round 125: OrderedDict([('root_mean_squared_error', 3.4567873), ('loss', 11.94938), ('num_examples', 8583), ('num_batches', 882)])\n",
      "Train round 126: OrderedDict([('root_mean_squared_error', 3.1835425), ('loss', 10.134944), ('num_examples', 9440), ('num_batches', 967)])\n",
      "Train round 127: OrderedDict([('root_mean_squared_error', 3.548442), ('loss', 12.59144), ('num_examples', 7796), ('num_batches', 805)])\n",
      "Train round 128: OrderedDict([('root_mean_squared_error', 3.4131255), ('loss', 11.649426), ('num_examples', 7780), ('num_batches', 803)])\n",
      "Train round 129: OrderedDict([('root_mean_squared_error', 3.2463307), ('loss', 10.538666), ('num_examples', 9625), ('num_batches', 986)])\n",
      "Train round 130: OrderedDict([('root_mean_squared_error', 3.4588075), ('loss', 11.963348), ('num_examples', 7413), ('num_batches', 767)])\n",
      "Train round 131: OrderedDict([('root_mean_squared_error', 3.1571395), ('loss', 9.96753), ('num_examples', 10098), ('num_batches', 1033)])\n",
      "Train round 132: OrderedDict([('root_mean_squared_error', 3.2340326), ('loss', 10.458967), ('num_examples', 8827), ('num_batches', 906)])\n",
      "Train round 133: OrderedDict([('root_mean_squared_error', 3.1868496), ('loss', 10.156011), ('num_examples', 9609), ('num_batches', 984)])\n",
      "Train round 134: OrderedDict([('root_mean_squared_error', 3.1482348), ('loss', 9.911383), ('num_examples', 9031), ('num_batches', 929)])\n",
      "Train round 135: OrderedDict([('root_mean_squared_error', 3.1441803), ('loss', 9.88587), ('num_examples', 8768), ('num_batches', 900)])\n",
      "Train round 136: OrderedDict([('root_mean_squared_error', 3.0754273), ('loss', 9.458253), ('num_examples', 9600), ('num_batches', 983)])\n",
      "Train round 137: OrderedDict([('root_mean_squared_error', 3.1840296), ('loss', 10.138044), ('num_examples', 9206), ('num_batches', 945)])\n",
      "Train round 138: OrderedDict([('root_mean_squared_error', 3.3546789), ('loss', 11.253872), ('num_examples', 7703), ('num_batches', 795)])\n",
      "Train round 139: OrderedDict([('root_mean_squared_error', 3.1134946), ('loss', 9.69385), ('num_examples', 8744), ('num_batches', 899)])\n",
      "Train round 140: OrderedDict([('root_mean_squared_error', 3.0671957), ('loss', 9.407689), ('num_examples', 9122), ('num_batches', 935)])\n",
      "Train round 141: OrderedDict([('root_mean_squared_error', 3.4147825), ('loss', 11.66074), ('num_examples', 7123), ('num_batches', 736)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 22:51:44.461731: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 142: OrderedDict([('root_mean_squared_error', 3.2199574), ('loss', 10.368126), ('num_examples', 7594), ('num_batches', 783)])\n",
      "Train round 143: OrderedDict([('root_mean_squared_error', 3.2875922), ('loss', 10.808262), ('num_examples', 7867), ('num_batches', 812)])\n",
      "Train round 144: OrderedDict([('root_mean_squared_error', 3.2483127), ('loss', 10.551537), ('num_examples', 6985), ('num_batches', 722)])\n",
      "Train round 145: OrderedDict([('root_mean_squared_error', 3.1832283), ('loss', 10.132943), ('num_examples', 9292), ('num_batches', 952)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 22:51:54.985730: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 146: OrderedDict([('root_mean_squared_error', 3.0594504), ('loss', 9.360237), ('num_examples', 8417), ('num_batches', 863)])\n",
      "Train round 147: OrderedDict([('root_mean_squared_error', 2.9767308), ('loss', 8.860927), ('num_examples', 9616), ('num_batches', 986)])\n",
      "Train round 148: OrderedDict([('root_mean_squared_error', 3.1844366), ('loss', 10.140637), ('num_examples', 6920), ('num_batches', 716)])\n",
      "Train round 149: OrderedDict([('root_mean_squared_error', 2.9482503), ('loss', 8.69218), ('num_examples', 9125), ('num_batches', 937)])\n",
      "Train round 150: OrderedDict([('root_mean_squared_error', 3.3859828), ('loss', 11.46488), ('num_examples', 6930), ('num_batches', 717)])\n",
      "Train round 151: OrderedDict([('root_mean_squared_error', 3.1647265), ('loss', 10.015494), ('num_examples', 8848), ('num_batches', 908)])\n",
      "Train round 152: OrderedDict([('root_mean_squared_error', 3.1561167), ('loss', 9.961073), ('num_examples', 8204), ('num_batches', 842)])\n",
      "Train round 153: OrderedDict([('root_mean_squared_error', 3.1790698), ('loss', 10.106484), ('num_examples', 7314), ('num_batches', 755)])\n",
      "Train round 154: OrderedDict([('root_mean_squared_error', 2.9483812), ('loss', 8.692952), ('num_examples', 8407), ('num_batches', 864)])\n",
      "Train round 155: OrderedDict([('root_mean_squared_error', 2.896803), ('loss', 8.391467), ('num_examples', 8324), ('num_batches', 857)])\n",
      "Train round 156: OrderedDict([('root_mean_squared_error', 3.0900385), ('loss', 9.548338), ('num_examples', 7458), ('num_batches', 771)])\n",
      "Train round 157: OrderedDict([('root_mean_squared_error', 2.9931605), ('loss', 8.95901), ('num_examples', 9386), ('num_batches', 961)])\n",
      "Train round 158: OrderedDict([('root_mean_squared_error', 2.9876506), ('loss', 8.926057), ('num_examples', 9041), ('num_batches', 927)])\n",
      "Train round 159: OrderedDict([('root_mean_squared_error', 2.846244), ('loss', 8.101105), ('num_examples', 10128), ('num_batches', 1036)])\n",
      "Train round 160: OrderedDict([('root_mean_squared_error', 3.075937), ('loss', 9.461389), ('num_examples', 7232), ('num_batches', 747)])\n",
      "Train round 161: OrderedDict([('root_mean_squared_error', 2.7798095), ('loss', 7.72734), ('num_examples', 9202), ('num_batches', 942)])\n",
      "Train round 162: OrderedDict([('root_mean_squared_error', 2.9853513), ('loss', 8.912323), ('num_examples', 8398), ('num_batches', 863)])\n",
      "Train round 163: OrderedDict([('root_mean_squared_error', 3.2151258), ('loss', 10.337034), ('num_examples', 5975), ('num_batches', 620)])\n",
      "Train round 164: OrderedDict([('root_mean_squared_error', 3.224904), ('loss', 10.400007), ('num_examples', 7409), ('num_batches', 764)])\n",
      "Train round 165: OrderedDict([('root_mean_squared_error', 3.078979), ('loss', 9.480111), ('num_examples', 8701), ('num_batches', 894)])\n",
      "Train round 166: OrderedDict([('root_mean_squared_error', 2.7670872), ('loss', 7.656772), ('num_examples', 9175), ('num_batches', 940)])\n",
      "Train round 167: OrderedDict([('root_mean_squared_error', 2.8135011), ('loss', 7.915789), ('num_examples', 9140), ('num_batches', 936)])\n",
      "Train round 168: OrderedDict([('root_mean_squared_error', 2.941725), ('loss', 8.653747), ('num_examples', 7953), ('num_batches', 818)])\n",
      "Train round 169: OrderedDict([('root_mean_squared_error', 2.8652463), ('loss', 8.209637), ('num_examples', 8818), ('num_batches', 904)])\n",
      "Train round 170: OrderedDict([('root_mean_squared_error', 2.9536517), ('loss', 8.724058), ('num_examples', 8443), ('num_batches', 869)])\n",
      "Train round 171: OrderedDict([('root_mean_squared_error', 3.0566216), ('loss', 9.342935), ('num_examples', 7531), ('num_batches', 777)])\n",
      "Train round 172: OrderedDict([('root_mean_squared_error', 2.6962912), ('loss', 7.2699866), ('num_examples', 9117), ('num_batches', 934)])\n",
      "Train round 173: OrderedDict([('root_mean_squared_error', 2.87809), ('loss', 8.283402), ('num_examples', 8793), ('num_batches', 903)])\n",
      "Train round 174: OrderedDict([('root_mean_squared_error', 2.7197857), ('loss', 7.3972344), ('num_examples', 9028), ('num_batches', 929)])\n",
      "Train round 175: OrderedDict([('root_mean_squared_error', 2.8709714), ('loss', 8.242476), ('num_examples', 8096), ('num_batches', 834)])\n",
      "Train round 176: OrderedDict([('root_mean_squared_error', 3.1626945), ('loss', 10.002637), ('num_examples', 6851), ('num_batches', 712)])\n",
      "Train round 177: OrderedDict([('root_mean_squared_error', 2.8850467), ('loss', 8.323495), ('num_examples', 7778), ('num_batches', 800)])\n",
      "Train round 178: OrderedDict([('root_mean_squared_error', 2.7345734), ('loss', 7.477892), ('num_examples', 8319), ('num_batches', 856)])\n",
      "Train round 179: OrderedDict([('root_mean_squared_error', 2.9187107), ('loss', 8.518872), ('num_examples', 7693), ('num_batches', 794)])\n",
      "Train round 180: OrderedDict([('root_mean_squared_error', 2.6605358), ('loss', 7.0784516), ('num_examples', 9309), ('num_batches', 953)])\n",
      "Train round 181: OrderedDict([('root_mean_squared_error', 2.7544532), ('loss', 7.587013), ('num_examples', 8820), ('num_batches', 904)])\n",
      "Train round 182: OrderedDict([('root_mean_squared_error', 3.0381753), ('loss', 9.230509), ('num_examples', 7861), ('num_batches', 810)])\n",
      "Train round 183: OrderedDict([('root_mean_squared_error', 2.6708336), ('loss', 7.1333523), ('num_examples', 8538), ('num_batches', 878)])\n",
      "Train round 184: OrderedDict([('root_mean_squared_error', 2.8169253), ('loss', 7.935068), ('num_examples', 8188), ('num_batches', 844)])\n",
      "Train round 185: OrderedDict([('root_mean_squared_error', 2.5318208), ('loss', 6.4101167), ('num_examples', 9510), ('num_batches', 977)])\n",
      "Train round 186: OrderedDict([('root_mean_squared_error', 2.9866447), ('loss', 8.920047), ('num_examples', 7427), ('num_batches', 767)])\n",
      "Train round 187: OrderedDict([('root_mean_squared_error', 2.578247), ('loss', 6.647358), ('num_examples', 9518), ('num_batches', 973)])\n",
      "Train round 188: OrderedDict([('root_mean_squared_error', 2.8511918), ('loss', 8.129294), ('num_examples', 7074), ('num_batches', 732)])\n",
      "Train round 189: OrderedDict([('root_mean_squared_error', 2.7233088), ('loss', 7.4164104), ('num_examples', 7993), ('num_batches', 825)])\n",
      "Train round 190: OrderedDict([('root_mean_squared_error', 2.7935274), ('loss', 7.803795), ('num_examples', 8500), ('num_batches', 873)])\n",
      "Train round 191: OrderedDict([('root_mean_squared_error', 2.5978014), ('loss', 6.7485723), ('num_examples', 9073), ('num_batches', 930)])\n",
      "Train round 192: OrderedDict([('root_mean_squared_error', 2.6493695), ('loss', 7.0191584), ('num_examples', 9521), ('num_batches', 973)])\n",
      "Train round 193: OrderedDict([('root_mean_squared_error', 2.7634234), ('loss', 7.63651), ('num_examples', 8809), ('num_batches', 904)])\n",
      "Train round 194: OrderedDict([('root_mean_squared_error', 2.6389666), ('loss', 6.964144), ('num_examples', 10071), ('num_batches', 1031)])\n",
      "Train round 195: OrderedDict([('root_mean_squared_error', 2.9117837), ('loss', 8.478484), ('num_examples', 7140), ('num_batches', 738)])\n",
      "Train round 196: OrderedDict([('root_mean_squared_error', 2.637187), ('loss', 6.9547553), ('num_examples', 9278), ('num_batches', 949)])\n",
      "Train round 197: OrderedDict([('root_mean_squared_error', 2.8527598), ('loss', 8.138239), ('num_examples', 7036), ('num_batches', 730)])\n",
      "Train round 198: OrderedDict([('root_mean_squared_error', 2.5579243), ('loss', 6.5429764), ('num_examples', 9112), ('num_batches', 935)])\n",
      "Train round 199: OrderedDict([('root_mean_squared_error', 2.9334795), ('loss', 8.605303), ('num_examples', 6887), ('num_batches', 713)])\n"
     ]
    }
   ],
   "source": [
    "train_losses, train_accs = [], []\n",
    "\n",
    "state = iterative_process.initialize()\n",
    "\n",
    "for i in range(NUM_ROUNDS):\n",
    "    federated_train_data = np.random.choice(\n",
    "        tf_trainsets, size=50, replace=False\n",
    "    ).tolist()\n",
    "    state, metrics = iterative_process.next(state, federated_train_data)\n",
    "    print(f\"Train round {i}:\", metrics[\"train\"])\n",
    "    train_losses.append(metrics[\"train\"][\"loss\"])\n",
    "    train_accs.append(metrics[\"train\"][\"root_mean_squared_error\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "12ef3191",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"dark_background\")\n",
    "plt.rcParams[\"figure.figsize\"] = [10, 8]\n",
    "plt.plot(range(NUM_ROUNDS), train_losses)\n",
    "plt.ylabel(\"Train Loss\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "3caacef2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"dark_background\")\n",
    "plt.rcParams[\"figure.figsize\"] = [10, 8]\n",
    "plt.plot(range(NUM_ROUNDS), train_accs)\n",
    "plt.ylabel(\"Train RSME\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train RMSE\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "87a60c71",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 22:54:11.661978: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE on Test Set: 3.7825847\n"
     ]
    }
   ],
   "source": [
    "eval_metrics = evaluation_computation(state.model, tf_testsets)\n",
    "print(\"RMSE on Test Set:\", eval_metrics[\"eval\"][\"root_mean_squared_error\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b4214ff3",
   "metadata": {},
   "source": [
    "## Step 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "0f061c36",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_fn = functools.partial(\n",
    "    matrix_factorization_model,\n",
    "    n_users=n_users,\n",
    "    n_items=n_items,\n",
    "    n_latent_factors=4,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "f970ff30",
   "metadata": {},
   "outputs": [],
   "source": [
    "fedprox_learning_process = tff.learning.algorithms.build_unweighted_fed_prox(\n",
    "    model_fn,\n",
    "    proximal_strength=10e-4,\n",
    "    server_optimizer_fn=lambda: tf.keras.optimizers.SGD(1.0),\n",
    "    client_optimizer_fn=lambda: tf.keras.optimizers.SGD(0.5),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "2fe7efd7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "( -> <\n",
      "  global_model_weights=<\n",
      "    trainable=<\n",
      "      float32[9725,4],\n",
      "      float32[611,4]\n",
      "    >,\n",
      "    non_trainable=<>\n",
      "  >,\n",
      "  distributor=<>,\n",
      "  client_work=<>,\n",
      "  aggregator=<\n",
      "    <>,\n",
      "    <>\n",
      "  >,\n",
      "  finalizer=<\n",
      "    int64\n",
      "  >\n",
      ">@SERVER)\n"
     ]
    }
   ],
   "source": [
    "print(fedprox_learning_process.initialize.type_signature.formatted_representation())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "f5bf191d",
   "metadata": {},
   "outputs": [],
   "source": [
    "evaluation_computation = tff.learning.build_federated_evaluation(\n",
    "    model_fn,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "1e5b98e6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<\n",
      "  server_model_weights=<\n",
      "    trainable=<\n",
      "      float32[9725,4],\n",
      "      float32[611,4]\n",
      "    >,\n",
      "    non_trainable=<>\n",
      "  >@SERVER,\n",
      "  federated_dataset={<\n",
      "    x=int64[?,2],\n",
      "    y=float32[?,1]\n",
      "  >*}@CLIENTS\n",
      "> -> <\n",
      "  eval=<\n",
      "    root_mean_squared_error=float32,\n",
      "    loss=float32,\n",
      "    num_examples=int64,\n",
      "    num_batches=int64\n",
      "  >\n",
      ">@SERVER)\n"
     ]
    }
   ],
   "source": [
    "print(evaluation_computation.type_signature.formatted_representation())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "9f807bfb",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 0 | train loss: 14.445063591003418 | train RMSE: 3.800666093826294\n",
      "Round 1 | train loss: 12.863300323486328 | train RMSE: 3.5865442752838135\n",
      "Round 2 | train loss: 12.981673240661621 | train RMSE: 3.603008985519409\n",
      "Round 3 | train loss: 14.06955337524414 | train RMSE: 3.7509403228759766\n",
      "Round 4 | train loss: 11.10766315460205 | train RMSE: 3.3328161239624023\n",
      "Round 5 | train loss: 6.883570671081543 | train RMSE: 2.6236557960510254\n",
      "Round 6 | train loss: 5.4634175300598145 | train RMSE: 2.337395429611206\n",
      "Round 7 | train loss: 4.481811046600342 | train RMSE: 2.1170287132263184\n",
      "Round 8 | train loss: 4.603305816650391 | train RMSE: 2.14553165435791\n",
      "Round 9 | train loss: 4.599572658538818 | train RMSE: 2.1446614265441895\n",
      "Round 10 | train loss: 3.901719093322754 | train RMSE: 1.9752769470214844\n",
      "Round 11 | train loss: 3.5184707641601562 | train RMSE: 1.8757586479187012\n",
      "Round 12 | train loss: 2.954676389694214 | train RMSE: 1.7189173698425293\n",
      "Round 13 | train loss: 2.610311508178711 | train RMSE: 1.6156457662582397\n",
      "Round 14 | train loss: 2.6270694732666016 | train RMSE: 1.6208237409591675\n",
      "Round 15 | train loss: 2.5494112968444824 | train RMSE: 1.5966875553131104\n",
      "Round 16 | train loss: 2.290870189666748 | train RMSE: 1.5135620832443237\n",
      "Round 17 | train loss: 2.321784734725952 | train RMSE: 1.523740291595459\n",
      "Round 18 | train loss: 2.4207236766815186 | train RMSE: 1.5558674335479736\n",
      "Round 19 | train loss: 1.9332191944122314 | train RMSE: 1.3904025554656982\n",
      "Round 20 | train loss: 2.1960678100585938 | train RMSE: 1.4819135665893555\n",
      "Round 21 | train loss: 1.8564629554748535 | train RMSE: 1.3625208139419556\n",
      "Round 22 | train loss: 1.8616533279418945 | train RMSE: 1.3644241094589233\n",
      "Round 23 | train loss: 1.7164902687072754 | train RMSE: 1.3101489543914795\n",
      "Round 24 | train loss: 1.6988073587417603 | train RMSE: 1.3033829927444458\n",
      "Round 25 | train loss: 1.5877660512924194 | train RMSE: 1.2600659132003784\n",
      "Round 26 | train loss: 1.532824993133545 | train RMSE: 1.2380731105804443\n",
      "Round 27 | train loss: 1.4131680727005005 | train RMSE: 1.188767433166504\n",
      "Round 28 | train loss: 1.4255163669586182 | train RMSE: 1.193949818611145\n",
      "Round 29 | train loss: 1.2780342102050781 | train RMSE: 1.1305017471313477\n",
      "Round 30 | train loss: 1.4367046356201172 | train RMSE: 1.198626160621643\n",
      "Round 31 | train loss: 1.4813635349273682 | train RMSE: 1.2171127796173096\n",
      "Round 32 | train loss: 1.245526671409607 | train RMSE: 1.1160316467285156\n",
      "Round 33 | train loss: 1.1422648429870605 | train RMSE: 1.0687679052352905\n",
      "Round 34 | train loss: 1.2416480779647827 | train RMSE: 1.1142926216125488\n",
      "Round 35 | train loss: 1.1806912422180176 | train RMSE: 1.086596131324768\n",
      "Round 36 | train loss: 1.0245171785354614 | train RMSE: 1.0121843814849854\n",
      "Round 37 | train loss: 1.144567847251892 | train RMSE: 1.0698447227478027\n",
      "Round 38 | train loss: 1.0855532884597778 | train RMSE: 1.0418988466262817\n",
      "Round 39 | train loss: 1.1117740869522095 | train RMSE: 1.054407000541687\n",
      "Round 40 | train loss: 0.9189454317092896 | train RMSE: 0.9586164355278015\n",
      "Round 41 | train loss: 0.9636000990867615 | train RMSE: 0.9816313982009888\n",
      "Round 42 | train loss: 0.9204195737838745 | train RMSE: 0.9593849778175354\n",
      "Round 43 | train loss: 0.9929705858230591 | train RMSE: 0.9964790940284729\n",
      "Round 44 | train loss: 0.9410591721534729 | train RMSE: 0.9700820446014404\n",
      "Round 45 | train loss: 0.917229950428009 | train RMSE: 0.9577212333679199\n",
      "Round 46 | train loss: 0.9164596199989319 | train RMSE: 0.9573189616203308\n",
      "Round 47 | train loss: 0.8723824620246887 | train RMSE: 0.9340141415596008\n",
      "Round 48 | train loss: 0.7847704887390137 | train RMSE: 0.8858727216720581\n",
      "Round 49 | train loss: 0.7793186902999878 | train RMSE: 0.8827902674674988\n",
      "Round 50 | train loss: 0.9441626071929932 | train RMSE: 0.9716802835464478\n",
      "Round 51 | train loss: 0.8154547810554504 | train RMSE: 0.9030253291130066\n",
      "Round 52 | train loss: 0.7761537432670593 | train RMSE: 0.8809958696365356\n",
      "Round 53 | train loss: 0.7592304944992065 | train RMSE: 0.8713383078575134\n",
      "Round 54 | train loss: 0.8836650252342224 | train RMSE: 0.9400346279144287\n",
      "Round 55 | train loss: 0.6823935508728027 | train RMSE: 0.8260711431503296\n",
      "Round 56 | train loss: 0.7740892171859741 | train RMSE: 0.8798233866691589\n",
      "Round 57 | train loss: 0.6623160243034363 | train RMSE: 0.8138279914855957\n",
      "Round 58 | train loss: 0.6804985404014587 | train RMSE: 0.8249233365058899\n",
      "Round 59 | train loss: 0.8460507392883301 | train RMSE: 0.9198101758956909\n",
      "Round 60 | train loss: 0.7450012564659119 | train RMSE: 0.8631345629692078\n",
      "Round 61 | train loss: 0.6683663725852966 | train RMSE: 0.8175367712974548\n",
      "Round 62 | train loss: 0.6548648476600647 | train RMSE: 0.8092371821403503\n",
      "Round 63 | train loss: 0.7338700890541077 | train RMSE: 0.8566621541976929\n",
      "Round 64 | train loss: 0.6863004565238953 | train RMSE: 0.828432559967041\n",
      "Round 65 | train loss: 0.6464200019836426 | train RMSE: 0.8040024638175964\n",
      "Round 66 | train loss: 0.6127615571022034 | train RMSE: 0.7827908396720886\n",
      "Round 67 | train loss: 0.6349047422409058 | train RMSE: 0.7968090772628784\n",
      "Round 68 | train loss: 0.6367912292480469 | train RMSE: 0.7979919910430908\n",
      "Round 69 | train loss: 0.5807299017906189 | train RMSE: 0.7620563507080078\n",
      "Round 70 | train loss: 0.6529632210731506 | train RMSE: 0.8080613613128662\n",
      "Round 71 | train loss: 0.6482197046279907 | train RMSE: 0.8051209449768066\n",
      "Round 72 | train loss: 0.6079626679420471 | train RMSE: 0.7797195911407471\n",
      "Round 73 | train loss: 0.6207424998283386 | train RMSE: 0.7878721356391907\n",
      "Round 74 | train loss: 0.5376670360565186 | train RMSE: 0.7332578301429749\n",
      "Round 75 | train loss: 0.6188115477561951 | train RMSE: 0.786645770072937\n",
      "Round 76 | train loss: 0.5587414503097534 | train RMSE: 0.7474901080131531\n",
      "Round 77 | train loss: 0.5372636318206787 | train RMSE: 0.7329826951026917\n",
      "Round 78 | train loss: 0.5888270139694214 | train RMSE: 0.7673506140708923\n",
      "Round 79 | train loss: 0.5927349925041199 | train RMSE: 0.7698928713798523\n",
      "Round 80 | train loss: 0.5931534767150879 | train RMSE: 0.7701645493507385\n",
      "Round 81 | train loss: 0.6705809235572815 | train RMSE: 0.8188900351524353\n",
      "Round 82 | train loss: 0.5347535610198975 | train RMSE: 0.7312684655189514\n",
      "Round 83 | train loss: 0.575218915939331 | train RMSE: 0.7584318518638611\n",
      "Round 84 | train loss: 0.48877498507499695 | train RMSE: 0.6991243958473206\n",
      "Round 85 | train loss: 0.6044665575027466 | train RMSE: 0.7774744629859924\n",
      "Round 86 | train loss: 0.4857383370399475 | train RMSE: 0.696949303150177\n",
      "Round 87 | train loss: 0.6017065644264221 | train RMSE: 0.7756974697113037\n",
      "Round 88 | train loss: 0.4842602610588074 | train RMSE: 0.695888102054596\n",
      "Round 89 | train loss: 0.5129145383834839 | train RMSE: 0.7161805033683777\n",
      "Round 90 | train loss: 0.506810188293457 | train RMSE: 0.7119060158729553\n",
      "Round 91 | train loss: 0.4633477032184601 | train RMSE: 0.6806964874267578\n",
      "Round 92 | train loss: 0.48492828011512756 | train RMSE: 0.6963679194450378\n",
      "Round 93 | train loss: 0.5184579491615295 | train RMSE: 0.7200402021408081\n",
      "Round 94 | train loss: 0.5113931894302368 | train RMSE: 0.7151175737380981\n",
      "Round 95 | train loss: 0.4808100461959839 | train RMSE: 0.6934046745300293\n",
      "Round 96 | train loss: 0.4521389901638031 | train RMSE: 0.6724128127098083\n",
      "Round 97 | train loss: 0.4835810959339142 | train RMSE: 0.6953999400138855\n",
      "Round 98 | train loss: 0.4694739580154419 | train RMSE: 0.6851816177368164\n",
      "Round 99 | train loss: 0.4035968780517578 | train RMSE: 0.6352927684783936\n",
      "Round 100 | train loss: 0.43525776267051697 | train RMSE: 0.659740686416626\n",
      "Round 101 | train loss: 0.4718698263168335 | train RMSE: 0.6869277954101562\n",
      "Round 102 | train loss: 0.460260808467865 | train RMSE: 0.6784252524375916\n",
      "Round 103 | train loss: 0.5119157433509827 | train RMSE: 0.7154828906059265\n",
      "Round 104 | train loss: 0.47176504135131836 | train RMSE: 0.6868515610694885\n",
      "Round 105 | train loss: 0.4301404058933258 | train RMSE: 0.655850887298584\n",
      "Round 106 | train loss: 0.41127339005470276 | train RMSE: 0.6413059830665588\n",
      "Round 107 | train loss: 0.4369323253631592 | train RMSE: 0.6610085964202881\n",
      "Round 108 | train loss: 0.509668231010437 | train RMSE: 0.7139105200767517\n",
      "Round 109 | train loss: 0.3813445270061493 | train RMSE: 0.6175310015678406\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 110 | train loss: 0.3956719934940338 | train RMSE: 0.6290246248245239\n",
      "Round 111 | train loss: 0.40241721272468567 | train RMSE: 0.63436359167099\n",
      "Round 112 | train loss: 0.4614361822605133 | train RMSE: 0.6792909502983093\n",
      "Round 113 | train loss: 0.4122992753982544 | train RMSE: 0.6421053409576416\n",
      "Round 114 | train loss: 0.39232948422431946 | train RMSE: 0.6263620853424072\n",
      "Round 115 | train loss: 0.42629292607307434 | train RMSE: 0.6529111266136169\n",
      "Round 116 | train loss: 0.38912293314933777 | train RMSE: 0.6237971186637878\n",
      "Round 117 | train loss: 0.38028720021247864 | train RMSE: 0.6166743040084839\n",
      "Round 118 | train loss: 0.4149329662322998 | train RMSE: 0.6441528797149658\n",
      "Round 119 | train loss: 0.44785764813423157 | train RMSE: 0.6692216396331787\n",
      "Round 120 | train loss: 0.487199991941452 | train RMSE: 0.6979970932006836\n",
      "Round 121 | train loss: 0.3961804509162903 | train RMSE: 0.6294286847114563\n",
      "Round 122 | train loss: 0.4422721564769745 | train RMSE: 0.6650354266166687\n",
      "Round 123 | train loss: 0.37625664472579956 | train RMSE: 0.6133975982666016\n",
      "Round 124 | train loss: 0.3584384322166443 | train RMSE: 0.598697304725647\n",
      "Round 125 | train loss: 0.3694812059402466 | train RMSE: 0.607849657535553\n",
      "Round 126 | train loss: 0.36921635270118713 | train RMSE: 0.6076317429542542\n",
      "Round 127 | train loss: 0.39895346760749817 | train RMSE: 0.63162761926651\n",
      "Round 128 | train loss: 0.39504095911979675 | train RMSE: 0.6285228133201599\n",
      "Round 129 | train loss: 0.37750503420829773 | train RMSE: 0.614414393901825\n",
      "Round 130 | train loss: 0.3891589343547821 | train RMSE: 0.6238260269165039\n",
      "Round 131 | train loss: 0.3241063952445984 | train RMSE: 0.5693033933639526\n",
      "Round 132 | train loss: 0.455028772354126 | train RMSE: 0.674558162689209\n",
      "Round 133 | train loss: 0.36401644349098206 | train RMSE: 0.6033377647399902\n",
      "Round 134 | train loss: 0.3889177739620209 | train RMSE: 0.6236327290534973\n",
      "Round 135 | train loss: 0.3675841987133026 | train RMSE: 0.6062871813774109\n",
      "Round 136 | train loss: 0.3566758334636688 | train RMSE: 0.5972234010696411\n",
      "Round 137 | train loss: 0.37028542160987854 | train RMSE: 0.6085108518600464\n",
      "Round 138 | train loss: 0.3480401337146759 | train RMSE: 0.5899492502212524\n",
      "Round 139 | train loss: 0.40121227502822876 | train RMSE: 0.6334131956100464\n",
      "Round 140 | train loss: 0.31356388330459595 | train RMSE: 0.5599677562713623\n",
      "Round 141 | train loss: 0.3448795676231384 | train RMSE: 0.5872644782066345\n",
      "Round 142 | train loss: 0.3701198399066925 | train RMSE: 0.6083747148513794\n",
      "Round 143 | train loss: 0.37312090396881104 | train RMSE: 0.6108362078666687\n",
      "Round 144 | train loss: 0.3423112630844116 | train RMSE: 0.585073709487915\n",
      "Round 145 | train loss: 0.325788289308548 | train RMSE: 0.5707786679267883\n",
      "Round 146 | train loss: 0.4004151523113251 | train RMSE: 0.6327836513519287\n",
      "Round 147 | train loss: 0.36125484108924866 | train RMSE: 0.601044774055481\n",
      "Round 148 | train loss: 0.3798156976699829 | train RMSE: 0.616291880607605\n",
      "Round 149 | train loss: 0.3046800494194031 | train RMSE: 0.5519782900810242\n",
      "Round 150 | train loss: 0.4239760637283325 | train RMSE: 0.6511344313621521\n",
      "Round 151 | train loss: 0.3321913480758667 | train RMSE: 0.5763604044914246\n",
      "Round 152 | train loss: 0.32785192131996155 | train RMSE: 0.5725835561752319\n",
      "Round 153 | train loss: 0.3429737091064453 | train RMSE: 0.5856395959854126\n",
      "Round 154 | train loss: 0.3453896939754486 | train RMSE: 0.5876986384391785\n",
      "Round 155 | train loss: 0.44329893589019775 | train RMSE: 0.6658069491386414\n",
      "Round 156 | train loss: 0.32765164971351624 | train RMSE: 0.5724086165428162\n",
      "Round 157 | train loss: 0.3415634036064148 | train RMSE: 0.5844342708587646\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:02:18.229724: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Round 158 | train loss: 0.40516671538352966 | train RMSE: 0.6365270614624023\n",
      "Round 159 | train loss: 0.35933825373649597 | train RMSE: 0.5994483232498169\n",
      "Round 160 | train loss: 0.2873004972934723 | train RMSE: 0.5360041856765747\n",
      "Round 161 | train loss: 0.35588911175727844 | train RMSE: 0.5965644121170044\n",
      "Round 162 | train loss: 0.37288394570350647 | train RMSE: 0.6106422543525696\n",
      "Round 163 | train loss: 0.35760197043418884 | train RMSE: 0.597998321056366\n",
      "Round 164 | train loss: 0.32704833149909973 | train RMSE: 0.5718814134597778\n",
      "Round 165 | train loss: 0.31858476996421814 | train RMSE: 0.5644331574440002\n",
      "Round 166 | train loss: 0.2841043174266815 | train RMSE: 0.5330143570899963\n",
      "Round 167 | train loss: 0.30964675545692444 | train RMSE: 0.5564591288566589\n",
      "Round 168 | train loss: 0.36923450231552124 | train RMSE: 0.6076467037200928\n",
      "Round 169 | train loss: 0.36678096652030945 | train RMSE: 0.6056244373321533\n",
      "Round 170 | train loss: 0.3755810260772705 | train RMSE: 0.6128466129302979\n",
      "Round 171 | train loss: 0.3092816472053528 | train RMSE: 0.5561309456825256\n",
      "Round 172 | train loss: 0.29065757989883423 | train RMSE: 0.5391266942024231\n",
      "Round 173 | train loss: 0.35890182852745056 | train RMSE: 0.5990841388702393\n",
      "Round 174 | train loss: 0.35533544421195984 | train RMSE: 0.5961002111434937\n",
      "Round 175 | train loss: 0.36818304657936096 | train RMSE: 0.6067808866500854\n",
      "Round 176 | train loss: 0.34901824593544006 | train RMSE: 0.5907776355743408\n",
      "Round 177 | train loss: 0.401565283536911 | train RMSE: 0.6336917877197266\n",
      "Round 178 | train loss: 0.3899414837360382 | train RMSE: 0.6244529485702515\n",
      "Round 179 | train loss: 0.3123263418674469 | train RMSE: 0.5588616728782654\n",
      "Round 180 | train loss: 0.3388032913208008 | train RMSE: 0.582068145275116\n",
      "Round 181 | train loss: 0.31718483567237854 | train RMSE: 0.5631916522979736\n",
      "Round 182 | train loss: 0.3143618106842041 | train RMSE: 0.5606797933578491\n",
      "Round 183 | train loss: 0.31817325949668884 | train RMSE: 0.5640684962272644\n",
      "Round 184 | train loss: 0.26761525869369507 | train RMSE: 0.5173154473304749\n",
      "Round 185 | train loss: 0.3049139976501465 | train RMSE: 0.5521901845932007\n",
      "Round 186 | train loss: 0.3055853843688965 | train RMSE: 0.5527977347373962\n",
      "Round 187 | train loss: 0.2984467148780823 | train RMSE: 0.5463027358055115\n",
      "Round 188 | train loss: 0.30782175064086914 | train RMSE: 0.5548168420791626\n",
      "Round 189 | train loss: 0.29937922954559326 | train RMSE: 0.5471555590629578\n",
      "Round 190 | train loss: 0.31262218952178955 | train RMSE: 0.5591261982917786\n",
      "Round 191 | train loss: 0.31573233008384705 | train RMSE: 0.5619006156921387\n",
      "Round 192 | train loss: 0.3401448130607605 | train RMSE: 0.5832193493843079\n",
      "Round 193 | train loss: 0.3051333427429199 | train RMSE: 0.5523887276649475\n",
      "Round 194 | train loss: 0.30515581369400024 | train RMSE: 0.5524090528488159\n",
      "Round 195 | train loss: 0.29512032866477966 | train RMSE: 0.543249785900116\n",
      "Round 196 | train loss: 0.33726808428764343 | train RMSE: 0.5807478427886963\n",
      "Round 197 | train loss: 0.33953267335891724 | train RMSE: 0.5826943516731262\n",
      "Round 198 | train loss: 0.28275373578071594 | train RMSE: 0.5317459106445312\n",
      "Round 199 | train loss: 0.2782978117465973 | train RMSE: 0.5275393724441528\n"
     ]
    }
   ],
   "source": [
    "train_losses, train_rmses = [], []\n",
    "\n",
    "state = fedprox_learning_process.initialize()\n",
    "\n",
    "# This may take a couple minutes to run.\n",
    "for i in range(NUM_ROUNDS):\n",
    "\n",
    "    federated_train_data = np.random.choice(\n",
    "        tf_trainsets, size=50, replace=False\n",
    "    ).tolist()\n",
    "    state = fedprox_learning_process.next(\n",
    "        state if i == 0 else state.state, federated_train_data\n",
    "    )\n",
    "\n",
    "    train_loss = state.metrics[\"client_work\"][\"train\"][\"loss\"]\n",
    "    train_rmse = state.metrics[\"client_work\"][\"train\"][\"root_mean_squared_error\"]\n",
    "\n",
    "    print(f\"Round {i} | train loss: {train_loss} | train RMSE: {train_rmse}\")\n",
    "    train_losses.append(train_loss)\n",
    "    train_rmses.append(train_rmse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "03f05d41",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_final_weights = state.state.global_model_weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "a98c988f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"dark_background\")\n",
    "plt.rcParams[\"figure.figsize\"] = [10, 8]\n",
    "plt.plot(range(NUM_ROUNDS), train_losses)\n",
    "plt.ylabel(\"Train Loss\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "ea95ad1a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"dark_background\")\n",
    "plt.rcParams[\"figure.figsize\"] = [10, 8]\n",
    "plt.plot(range(NUM_ROUNDS), train_rmses)\n",
    "plt.ylabel(\"Train RSME\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train RMSE\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "56a9bc84",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE on Test Set: 3.7838557\n"
     ]
    }
   ],
   "source": [
    "eval_metrics = evaluation_computation(model_final_weights, tf_testsets)\n",
    "print(\"RMSE on Test Set:\", eval_metrics[\"eval\"][\"root_mean_squared_error\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "103c86a9",
   "metadata": {},
   "source": [
    "## Step 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "a0124de0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def transform_batch_for_fed_recon(batch: tf.Tensor) -> collections.OrderedDict:\n",
    "    \"\"\"we do not need the user since each client will only look at their own data.\"\"\"\n",
    "    return collections.OrderedDict(\n",
    "        x=tf.cast(batch[:, 1:2], tf.int64), y=tf.cast(batch[:, 2:3], tf.float32)\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "d48ac99b",
   "metadata": {},
   "outputs": [],
   "source": [
    "tf_datasets = [\n",
    "    *create_tf_datasets(\n",
    "        data=ratings,\n",
    "        transform_func=transform_batch_for_fed_recon,\n",
    "        batch_size=10,\n",
    "        max_clients=100,\n",
    "        max_items_per_client=3000,\n",
    "    )\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "8dc00a10",
   "metadata": {},
   "outputs": [],
   "source": [
    "tf_trainsets, tf_testsets = split_tf_datasets(\n",
    "    tf_datasets,\n",
    "    train_fraction=0.8,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "67f10ea5",
   "metadata": {},
   "outputs": [],
   "source": [
    "class UserEmbedding(tf.keras.layers.Layer):\n",
    "    \"\"\"Keras layer representing an embedding for a single user.\"\"\"\n",
    "\n",
    "    def __init__(self, num_latent_factors, **kwargs):\n",
    "        super().__init__(**kwargs)\n",
    "        self.num_latent_factors = num_latent_factors\n",
    "\n",
    "    def build(self, input_shape):\n",
    "        self.embedding = self.add_weight(\n",
    "            shape=(1, self.num_latent_factors),\n",
    "            initializer=\"uniform\",\n",
    "            dtype=tf.float32,\n",
    "            name=\"UserEmbeddingKernel\",\n",
    "        )\n",
    "        super().build(input_shape)\n",
    "\n",
    "    def call(self, inputs):\n",
    "        return self.embedding\n",
    "\n",
    "    def compute_output_shape(self):\n",
    "        return (1, self.num_latent_factors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "3701e86e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_matrix_factorization_reconstruction_model(\n",
    "    n_items: int, n_latent_factors: int\n",
    ") -> tff.learning.reconstruction.Model:\n",
    "    \"\"\"Defines a Keras matrix factorization model.\"\"\"\n",
    "    # layers with variables will be partitioned into global and local layers.\n",
    "    global_layers = []\n",
    "    local_layers = []\n",
    "\n",
    "    item_input = tf.keras.layers.Input(shape=[1], name=\"Item\")\n",
    "    item_embedding_layer = tf.keras.layers.Embedding(\n",
    "        n_items, n_latent_factors, name=\"ItemEmbedding\"\n",
    "    )\n",
    "    global_layers.append(item_embedding_layer)\n",
    "    flat_item_vec = tf.keras.layers.Flatten(name=\"FlattenItems\")(\n",
    "        item_embedding_layer(item_input)\n",
    "    )\n",
    "\n",
    "    user_embedding_layer = UserEmbedding(n_latent_factors, name=\"UserEmbedding\")\n",
    "    local_layers.append(user_embedding_layer)\n",
    "\n",
    "    flat_user_vec = user_embedding_layer(item_input)\n",
    "\n",
    "    pred = tf.keras.layers.Dot(1, normalize=False, name=\"Dot\")(\n",
    "        [flat_user_vec, flat_item_vec]\n",
    "    )\n",
    "\n",
    "    input_spec = collections.OrderedDict(\n",
    "        x=tf.TensorSpec(shape=[None, 1], dtype=tf.int64),\n",
    "        y=tf.TensorSpec(shape=[None, 1], dtype=tf.float32),\n",
    "    )\n",
    "\n",
    "    model = tf.keras.Model(inputs=item_input, outputs=pred)\n",
    "\n",
    "    return tff.learning.reconstruction.from_keras_model(\n",
    "        keras_model=model,\n",
    "        global_layers=global_layers,\n",
    "        local_layers=local_layers,\n",
    "        input_spec=input_spec,\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "fdc73aa9",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_fn = functools.partial(\n",
    "    get_matrix_factorization_reconstruction_model, n_items=n_items, n_latent_factors=4\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "7f34dfa4",
   "metadata": {},
   "outputs": [],
   "source": [
    "class RatingAccuracy(tf.keras.metrics.Mean):\n",
    "    \"\"\"Keras metric computing accuracy of reconstructed ratings.\"\"\"\n",
    "\n",
    "    def __init__(self, name: str = \"rating_accuracy\", **kwargs):\n",
    "        super().__init__(name=name, **kwargs)\n",
    "\n",
    "    def update_state(\n",
    "        self,\n",
    "        y_true: tf.Tensor,\n",
    "        y_pred: tf.Tensor,\n",
    "        sample_weight: Optional[tf.Tensor] = None,\n",
    "    ):\n",
    "        absolute_diffs = tf.abs(y_true - y_pred)\n",
    "        # A [batch_size, 1] tf.bool tensor indicating correctness within the\n",
    "        # threshold for each example in a batch. A 0.5 threshold corresponds\n",
    "        # to correctness when predictions are rounded to the nearest whole\n",
    "        # number.\n",
    "        example_accuracies = tf.less_equal(absolute_diffs, 0.5)\n",
    "        super().update_state(example_accuracies, sample_weight=sample_weight)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "9a3dda33",
   "metadata": {},
   "outputs": [],
   "source": [
    "loss_fn = lambda: tf.keras.losses.MeanSquaredError()\n",
    "metrics_fn = lambda: [RatingAccuracy(), tf.keras.metrics.RootMeanSquaredError()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "694b574f",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_process = tff.learning.reconstruction.build_training_process(\n",
    "    model_fn=model_fn,\n",
    "    loss_fn=loss_fn,\n",
    "    metrics_fn=metrics_fn,\n",
    "    server_optimizer_fn=lambda: tf.keras.optimizers.SGD(1.0),\n",
    "    client_optimizer_fn=lambda: tf.keras.optimizers.SGD(0.5),\n",
    "    reconstruction_optimizer_fn=lambda: tf.keras.optimizers.SGD(0.1),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "8e822bb6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "( -> <\n",
      "  model=<\n",
      "    trainable=<\n",
      "      float32[9724,4]\n",
      "    >,\n",
      "    non_trainable=<>\n",
      "  >,\n",
      "  optimizer_state=<\n",
      "    int64\n",
      "  >,\n",
      "  delta_aggregate_state=<\n",
      "    value_sum_process=<>,\n",
      "    weight_sum_process=<>\n",
      "  >,\n",
      "  model_broadcast_state=<>\n",
      ">@SERVER)\n"
     ]
    }
   ],
   "source": [
    "print(training_process.initialize.type_signature.formatted_representation())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "45369923",
   "metadata": {},
   "outputs": [],
   "source": [
    "evaluation_computation = tff.learning.reconstruction.build_federated_evaluation(\n",
    "    model_fn,\n",
    "    loss_fn=loss_fn,\n",
    "    metrics_fn=metrics_fn,\n",
    "    reconstruction_optimizer_fn=functools.partial(tf.keras.optimizers.SGD, 0.1),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "3b3b3617",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<\n",
      "  server_model_weights=<\n",
      "    trainable=<\n",
      "      float32[9724,4]\n",
      "    >,\n",
      "    non_trainable=<>\n",
      "  >@SERVER,\n",
      "  federated_dataset={<\n",
      "    x=int64[?,1],\n",
      "    y=float32[?,1]\n",
      "  >*}@CLIENTS\n",
      "> -> <\n",
      "  broadcast=<>,\n",
      "  eval=<\n",
      "    loss=float32,\n",
      "    rating_accuracy=float32,\n",
      "    root_mean_squared_error=float32\n",
      "  >\n",
      ">@SERVER)\n"
     ]
    }
   ],
   "source": [
    "print(evaluation_computation.type_signature.formatted_representation())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "30ce7aff",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ModelWeights(trainable=[array([[ 0.032397  ,  0.00683475, -0.03634749,  0.01848885],\n",
      "       [-0.0056657 , -0.01406003, -0.03018404,  0.03773868],\n",
      "       [-0.03315934, -0.01758755,  0.0484832 , -0.04275289],\n",
      "       ...,\n",
      "       [ 0.00108583,  0.03362406, -0.01576094,  0.04999611],\n",
      "       [ 0.00850909, -0.00396698,  0.04703293, -0.02560939],\n",
      "       [ 0.03974214,  0.02007693,  0.04207966, -0.02986131]],\n",
      "      dtype=float32)], non_trainable=[])\n"
     ]
    }
   ],
   "source": [
    "state = training_process.initialize()\n",
    "\n",
    "print(state.model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "8bce0b02",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Item variables shape: (9724, 4)\n"
     ]
    }
   ],
   "source": [
    "print(\"Item variables shape:\", state.model.trainable[0].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "301748aa",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 0: OrderedDict([('rating_accuracy', 0.0037977688), ('root_mean_squared_error', 3.6690292), ('loss', 13.461775)])\n",
      "Train round 1: OrderedDict([('rating_accuracy', 0.0034514496), ('root_mean_squared_error', 3.608174), ('loss', 13.01892)])\n",
      "Train round 2: OrderedDict([('rating_accuracy', 0.008320216), ('root_mean_squared_error', 3.6286469), ('loss', 13.167078)])\n",
      "Train round 3: OrderedDict([('rating_accuracy', 0.003153745), ('root_mean_squared_error', 3.684771), ('loss', 13.577538)])\n",
      "Train round 4: OrderedDict([('rating_accuracy', 0.006672598), ('root_mean_squared_error', 3.5905397), ('loss', 12.891975)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:06:17.278523: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 5: OrderedDict([('rating_accuracy', 0.00831749), ('root_mean_squared_error', 3.545916), ('loss', 12.573521)])\n",
      "Train round 6: OrderedDict([('rating_accuracy', 0.015844677), ('root_mean_squared_error', 3.360343), ('loss', 11.291905)])\n",
      "Train round 7: OrderedDict([('rating_accuracy', 0.070332184), ('root_mean_squared_error', 2.9890742), ('loss', 8.934566)])\n",
      "Train round 8: OrderedDict([('rating_accuracy', 0.054640718), ('root_mean_squared_error', 2.895365), ('loss', 8.383139)])\n",
      "Train round 9: OrderedDict([('rating_accuracy', 0.11322484), ('root_mean_squared_error', 2.6239276), ('loss', 6.884996)])\n",
      "Train round 10: OrderedDict([('rating_accuracy', 0.09685506), ('root_mean_squared_error', 2.5549), ('loss', 6.5275145)])\n",
      "Train round 11: OrderedDict([('rating_accuracy', 0.15380986), ('root_mean_squared_error', 2.3354461), ('loss', 5.4543095)])\n",
      "Train round 12: OrderedDict([('rating_accuracy', 0.12344041), ('root_mean_squared_error', 2.4590793), ('loss', 6.047071)])\n",
      "Train round 13: OrderedDict([('rating_accuracy', 0.1652993), ('root_mean_squared_error', 2.4154356), ('loss', 5.834329)])\n",
      "Train round 14: OrderedDict([('rating_accuracy', 0.10596872), ('root_mean_squared_error', 2.5068395), ('loss', 6.284244)])\n",
      "Train round 15: OrderedDict([('rating_accuracy', 0.17521141), ('root_mean_squared_error', 2.1274543), ('loss', 4.526062)])\n",
      "Train round 16: OrderedDict([('rating_accuracy', 0.22980087), ('root_mean_squared_error', 1.9880575), ('loss', 3.952372)])\n",
      "Train round 17: OrderedDict([('rating_accuracy', 0.15938492), ('root_mean_squared_error', 2.1230154), ('loss', 4.507195)])\n",
      "Train round 18: OrderedDict([('rating_accuracy', 0.21258192), ('root_mean_squared_error', 1.9670119), ('loss', 3.8691359)])\n",
      "Train round 19: OrderedDict([('rating_accuracy', 0.24657534), ('root_mean_squared_error', 1.9075228), ('loss', 3.6386435)])\n",
      "Train round 20: OrderedDict([('rating_accuracy', 0.13986014), ('root_mean_squared_error', 2.281647), ('loss', 5.2059126)])\n",
      "Train round 21: OrderedDict([('rating_accuracy', 0.21909115), ('root_mean_squared_error', 1.9369777), ('loss', 3.7518828)])\n",
      "Train round 22: OrderedDict([('rating_accuracy', 0.19648463), ('root_mean_squared_error', 1.926318), ('loss', 3.7107015)])\n",
      "Train round 23: OrderedDict([('rating_accuracy', 0.26076004), ('root_mean_squared_error', 1.7701523), ('loss', 3.1334393)])\n",
      "Train round 24: OrderedDict([('rating_accuracy', 0.27578476), ('root_mean_squared_error', 1.6930884), ('loss', 2.8665485)])\n",
      "Train round 25: OrderedDict([('rating_accuracy', 0.17617793), ('root_mean_squared_error', 1.9912001), ('loss', 3.964878)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:07:14.985714: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 26: OrderedDict([('rating_accuracy', 0.23715188), ('root_mean_squared_error', 1.8264917), ('loss', 3.336072)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:07:17.154150: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 27: OrderedDict([('rating_accuracy', 0.20981337), ('root_mean_squared_error', 1.9048679), ('loss', 3.6285217)])\n",
      "Train round 28: OrderedDict([('rating_accuracy', 0.26602563), ('root_mean_squared_error', 1.6554478), ('loss', 2.7405076)])\n",
      "Train round 29: OrderedDict([('rating_accuracy', 0.24943253), ('root_mean_squared_error', 1.6764331), ('loss', 2.8104281)])\n",
      "Train round 30: OrderedDict([('rating_accuracy', 0.27859026), ('root_mean_squared_error', 1.6288921), ('loss', 2.6532893)])\n",
      "Train round 31: OrderedDict([('rating_accuracy', 0.28637582), ('root_mean_squared_error', 1.631946), ('loss', 2.6632478)])\n",
      "Train round 32: OrderedDict([('rating_accuracy', 0.30057928), ('root_mean_squared_error', 1.5276153), ('loss', 2.3336084)])\n",
      "Train round 33: OrderedDict([('rating_accuracy', 0.26235527), ('root_mean_squared_error', 1.6665064), ('loss', 2.7772434)])\n",
      "Train round 34: OrderedDict([('rating_accuracy', 0.24170868), ('root_mean_squared_error', 1.6950136), ('loss', 2.8730712)])\n",
      "Train round 35: OrderedDict([('rating_accuracy', 0.31726217), ('root_mean_squared_error', 1.4950947), ('loss', 2.2353082)])\n",
      "Train round 36: OrderedDict([('rating_accuracy', 0.2984506), ('root_mean_squared_error', 1.6241789), ('loss', 2.637957)])\n",
      "Train round 37: OrderedDict([('rating_accuracy', 0.32223943), ('root_mean_squared_error', 1.6227764), ('loss', 2.6334033)])\n",
      "Train round 38: OrderedDict([('rating_accuracy', 0.32637572), ('root_mean_squared_error', 1.5956553), ('loss', 2.546116)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:07:51.049702: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 39: OrderedDict([('rating_accuracy', 0.23006135), ('root_mean_squared_error', 1.7036579), ('loss', 2.90245)])\n",
      "Train round 40: OrderedDict([('rating_accuracy', 0.31336927), ('root_mean_squared_error', 1.6111524), ('loss', 2.5958123)])\n",
      "Train round 41: OrderedDict([('rating_accuracy', 0.34195888), ('root_mean_squared_error', 1.5037882), ('loss', 2.261379)])\n",
      "Train round 42: OrderedDict([('rating_accuracy', 0.29692647), ('root_mean_squared_error', 1.5507448), ('loss', 2.4048095)])\n",
      "Train round 43: OrderedDict([('rating_accuracy', 0.3442198), ('root_mean_squared_error', 1.5192074), ('loss', 2.307991)])\n",
      "Train round 44: OrderedDict([('rating_accuracy', 0.35087353), ('root_mean_squared_error', 1.4029318), ('loss', 1.9682178)])\n",
      "Train round 45: OrderedDict([('rating_accuracy', 0.2826765), ('root_mean_squared_error', 1.5821984), ('loss', 2.503352)])\n",
      "Train round 46: OrderedDict([('rating_accuracy', 0.3113358), ('root_mean_squared_error', 1.5610486), ('loss', 2.436873)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:08:12.204235: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 47: OrderedDict([('rating_accuracy', 0.35898674), ('root_mean_squared_error', 1.4609743), ('loss', 2.1344461)])\n",
      "Train round 48: OrderedDict([('rating_accuracy', 0.25603864), ('root_mean_squared_error', 1.5929017), ('loss', 2.5373359)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:08:18.421781: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 49: OrderedDict([('rating_accuracy', 0.2578173), ('root_mean_squared_error', 1.6124587), ('loss', 2.600023)])\n",
      "Train round 50: OrderedDict([('rating_accuracy', 0.34799513), ('root_mean_squared_error', 1.4185333), ('loss', 2.0122368)])\n",
      "Train round 51: OrderedDict([('rating_accuracy', 0.3699379), ('root_mean_squared_error', 1.3987942), ('loss', 1.9566255)])\n",
      "Train round 52: OrderedDict([('rating_accuracy', 0.3134978), ('root_mean_squared_error', 1.4356194), ('loss', 2.061003)])\n",
      "Train round 53: OrderedDict([('rating_accuracy', 0.3841723), ('root_mean_squared_error', 1.3653293), ('loss', 1.8641241)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:08:32.777431: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 54: OrderedDict([('rating_accuracy', 0.29124627), ('root_mean_squared_error', 1.481534), ('loss', 2.1949432)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:08:35.323712: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 55: OrderedDict([('rating_accuracy', 0.37853366), ('root_mean_squared_error', 1.3982567), ('loss', 1.9551219)])\n",
      "Train round 56: OrderedDict([('rating_accuracy', 0.35506374), ('root_mean_squared_error', 1.4102874), ('loss', 1.9889106)])\n",
      "Train round 57: OrderedDict([('rating_accuracy', 0.38226902), ('root_mean_squared_error', 1.3506715), ('loss', 1.8243136)])\n",
      "Train round 58: OrderedDict([('rating_accuracy', 0.28063118), ('root_mean_squared_error', 1.572263), ('loss', 2.472011)])\n",
      "Train round 59: OrderedDict([('rating_accuracy', 0.34593326), ('root_mean_squared_error', 1.3872275), ('loss', 1.9244004)])\n",
      "Train round 60: OrderedDict([('rating_accuracy', 0.33611444), ('root_mean_squared_error', 1.3852856), ('loss', 1.9190162)])\n",
      "Train round 61: OrderedDict([('rating_accuracy', 0.3674089), ('root_mean_squared_error', 1.3937169), ('loss', 1.942447)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:08:55.508347: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n",
      "2022-03-29 23:08:55.562285: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 62: OrderedDict([('rating_accuracy', 0.32542717), ('root_mean_squared_error', 1.4053484), ('loss', 1.9750043)])\n",
      "Train round 63: OrderedDict([('rating_accuracy', 0.3148413), ('root_mean_squared_error', 1.4788363), ('loss', 2.186957)])\n",
      "Train round 64: OrderedDict([('rating_accuracy', 0.3816156), ('root_mean_squared_error', 1.2848895), ('loss', 1.6509409)])\n",
      "Train round 65: OrderedDict([('rating_accuracy', 0.38803828), ('root_mean_squared_error', 1.3527839), ('loss', 1.8300245)])\n",
      "Train round 66: OrderedDict([('rating_accuracy', 0.34607977), ('root_mean_squared_error', 1.4514132), ('loss', 2.1066003)])\n",
      "Train round 67: OrderedDict([('rating_accuracy', 0.3900293), ('root_mean_squared_error', 1.2631702), ('loss', 1.5955992)])\n",
      "Train round 68: OrderedDict([('rating_accuracy', 0.33315912), ('root_mean_squared_error', 1.3628981), ('loss', 1.8574913)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:09:14.121058: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 69: OrderedDict([('rating_accuracy', 0.40298164), ('root_mean_squared_error', 1.2765367), ('loss', 1.629546)])\n",
      "Train round 70: OrderedDict([('rating_accuracy', 0.3806804), ('root_mean_squared_error', 1.2851179), ('loss', 1.6515282)])\n",
      "Train round 71: OrderedDict([('rating_accuracy', 0.3663578), ('root_mean_squared_error', 1.3466508), ('loss', 1.8134685)])\n",
      "Train round 72: OrderedDict([('rating_accuracy', 0.35744235), ('root_mean_squared_error', 1.3424605), ('loss', 1.8022003)])\n",
      "Train round 73: OrderedDict([('rating_accuracy', 0.4217284), ('root_mean_squared_error', 1.2814304), ('loss', 1.642064)])\n",
      "Train round 74: OrderedDict([('rating_accuracy', 0.3286052), ('root_mean_squared_error', 1.3965927), ('loss', 1.9504716)])\n",
      "Train round 75: OrderedDict([('rating_accuracy', 0.39684626), ('root_mean_squared_error', 1.3088683), ('loss', 1.7131361)])\n",
      "Train round 76: OrderedDict([('rating_accuracy', 0.36081654), ('root_mean_squared_error', 1.3016156), ('loss', 1.6942034)])\n",
      "Train round 77: OrderedDict([('rating_accuracy', 0.43924814), ('root_mean_squared_error', 1.201949), ('loss', 1.4446814)])\n",
      "Train round 78: OrderedDict([('rating_accuracy', 0.42143527), ('root_mean_squared_error', 1.28546), ('loss', 1.6524074)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:09:43.119381: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 79: OrderedDict([('rating_accuracy', 0.41858408), ('root_mean_squared_error', 1.2274554), ('loss', 1.5066469)])\n",
      "Train round 80: OrderedDict([('rating_accuracy', 0.34770358), ('root_mean_squared_error', 1.429111), ('loss', 2.0423584)])\n",
      "Train round 81: OrderedDict([('rating_accuracy', 0.38451305), ('root_mean_squared_error', 1.3289323), ('loss', 1.766061)])\n",
      "Train round 82: OrderedDict([('rating_accuracy', 0.4059925), ('root_mean_squared_error', 1.3027214), ('loss', 1.697083)])\n",
      "Train round 83: OrderedDict([('rating_accuracy', 0.3865462), ('root_mean_squared_error', 1.2658734), ('loss', 1.6024356)])\n",
      "Train round 84: OrderedDict([('rating_accuracy', 0.40122294), ('root_mean_squared_error', 1.2389178), ('loss', 1.5349176)])\n",
      "Train round 85: OrderedDict([('rating_accuracy', 0.40939122), ('root_mean_squared_error', 1.2453008), ('loss', 1.550774)])\n",
      "Train round 86: OrderedDict([('rating_accuracy', 0.34137237), ('root_mean_squared_error', 1.3506145), ('loss', 1.8241599)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:10:06.197717: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 87: OrderedDict([('rating_accuracy', 0.4402771), ('root_mean_squared_error', 1.2304751), ('loss', 1.5140691)])\n",
      "Train round 88: OrderedDict([('rating_accuracy', 0.44059515), ('root_mean_squared_error', 1.173345), ('loss', 1.3767384)])\n",
      "Train round 89: OrderedDict([('rating_accuracy', 0.4418093), ('root_mean_squared_error', 1.2515676), ('loss', 1.5664214)])\n",
      "Train round 90: OrderedDict([('rating_accuracy', 0.42451945), ('root_mean_squared_error', 1.2245748), ('loss', 1.4995836)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:10:16.962321: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 91: OrderedDict([('rating_accuracy', 0.44187054), ('root_mean_squared_error', 1.2156322), ('loss', 1.4777617)])\n",
      "Train round 92: OrderedDict([('rating_accuracy', 0.33354715), ('root_mean_squared_error', 1.4406828), ('loss', 2.0755668)])\n",
      "Train round 93: OrderedDict([('rating_accuracy', 0.43742824), ('root_mean_squared_error', 1.2246498), ('loss', 1.4997671)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:10:24.943281: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 94: OrderedDict([('rating_accuracy', 0.4209108), ('root_mean_squared_error', 1.29187), ('loss', 1.6689279)])\n",
      "Train round 95: OrderedDict([('rating_accuracy', 0.3871437), ('root_mean_squared_error', 1.3469673), ('loss', 1.814321)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:10:30.357720: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 96: OrderedDict([('rating_accuracy', 0.43012422), ('root_mean_squared_error', 1.2150958), ('loss', 1.4764576)])\n",
      "Train round 97: OrderedDict([('rating_accuracy', 0.45063516), ('root_mean_squared_error', 1.18161), ('loss', 1.3962023)])\n",
      "Train round 98: OrderedDict([('rating_accuracy', 0.4234536), ('root_mean_squared_error', 1.2324942), ('loss', 1.5190421)])\n",
      "Train round 99: OrderedDict([('rating_accuracy', 0.39254197), ('root_mean_squared_error', 1.3078433), ('loss', 1.7104542)])\n",
      "Train round 100: OrderedDict([('rating_accuracy', 0.46029514), ('root_mean_squared_error', 1.1555456), ('loss', 1.3352855)])\n",
      "Train round 101: OrderedDict([('rating_accuracy', 0.4265648), ('root_mean_squared_error', 1.217281), ('loss', 1.4817729)])\n",
      "Train round 102: OrderedDict([('rating_accuracy', 0.3918099), ('root_mean_squared_error', 1.2269413), ('loss', 1.505385)])\n",
      "Train round 103: OrderedDict([('rating_accuracy', 0.43582705), ('root_mean_squared_error', 1.1644258), ('loss', 1.3558875)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:10:53.358228: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 104: OrderedDict([('rating_accuracy', 0.45806453), ('root_mean_squared_error', 1.1437256), ('loss', 1.3081083)])\n",
      "Train round 105: OrderedDict([('rating_accuracy', 0.43430078), ('root_mean_squared_error', 1.2195084), ('loss', 1.4872009)])\n",
      "Train round 106: OrderedDict([('rating_accuracy', 0.44943053), ('root_mean_squared_error', 1.1717879), ('loss', 1.3730867)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:01.846816: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 107: OrderedDict([('rating_accuracy', 0.39371577), ('root_mean_squared_error', 1.1928055), ('loss', 1.4227852)])\n",
      "Train round 108: OrderedDict([('rating_accuracy', 0.4440542), ('root_mean_squared_error', 1.2111769), ('loss', 1.4669495)])\n",
      "Train round 109: OrderedDict([('rating_accuracy', 0.410795), ('root_mean_squared_error', 1.227109), ('loss', 1.5057966)])\n",
      "Train round 110: OrderedDict([('rating_accuracy', 0.37928978), ('root_mean_squared_error', 1.3034933), ('loss', 1.6990948)])\n",
      "Train round 111: OrderedDict([('rating_accuracy', 0.43804663), ('root_mean_squared_error', 1.164507), ('loss', 1.3560765)])\n",
      "Train round 112: OrderedDict([('rating_accuracy', 0.3075284), ('root_mean_squared_error', 1.4323655), ('loss', 2.051671)])\n",
      "Train round 113: OrderedDict([('rating_accuracy', 0.44213718), ('root_mean_squared_error', 1.1639066), ('loss', 1.3546786)])\n",
      "Train round 114: OrderedDict([('rating_accuracy', 0.47161573), ('root_mean_squared_error', 1.0956346), ('loss', 1.2004153)])\n",
      "Train round 115: OrderedDict([('rating_accuracy', 0.34793982), ('root_mean_squared_error', 1.3239864), ('loss', 1.7529402)])\n",
      "Train round 116: OrderedDict([('rating_accuracy', 0.4291735), ('root_mean_squared_error', 1.116509), ('loss', 1.2465923)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:29.153787: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 117: OrderedDict([('rating_accuracy', 0.43469894), ('root_mean_squared_error', 1.1873184), ('loss', 1.4097251)])\n",
      "Train round 118: OrderedDict([('rating_accuracy', 0.4473189), ('root_mean_squared_error', 1.1045893), ('loss', 1.2201178)])\n",
      "Train round 119: OrderedDict([('rating_accuracy', 0.3968254), ('root_mean_squared_error', 1.2140071), ('loss', 1.4738133)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:38.065689: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 120: OrderedDict([('rating_accuracy', 0.3917749), ('root_mean_squared_error', 1.2840548), ('loss', 1.6487966)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:40.962057: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 121: OrderedDict([('rating_accuracy', 0.4125387), ('root_mean_squared_error', 1.1997086), ('loss', 1.4393008)])\n",
      "Train round 122: OrderedDict([('rating_accuracy', 0.42095026), ('root_mean_squared_error', 1.2284247), ('loss', 1.5090271)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:45.385404: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 123: OrderedDict([('rating_accuracy', 0.4654933), ('root_mean_squared_error', 1.1092916), ('loss', 1.2305279)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:48.569719: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 124: OrderedDict([('rating_accuracy', 0.48622128), ('root_mean_squared_error', 1.0407519), ('loss', 1.0831647)])\n",
      "Train round 125: OrderedDict([('rating_accuracy', 0.45740342), ('root_mean_squared_error', 1.1188623), ('loss', 1.251853)])\n",
      "Train round 126: OrderedDict([('rating_accuracy', 0.4736614), ('root_mean_squared_error', 1.0665087), ('loss', 1.1374408)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:11:57.907851: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 127: OrderedDict([('rating_accuracy', 0.41646808), ('root_mean_squared_error', 1.1793358), ('loss', 1.3908329)])\n",
      "Train round 128: OrderedDict([('rating_accuracy', 0.44008484), ('root_mean_squared_error', 1.1717918), ('loss', 1.373096)])\n",
      "Train round 129: OrderedDict([('rating_accuracy', 0.45750105), ('root_mean_squared_error', 1.0615158), ('loss', 1.1268158)])\n",
      "Train round 130: OrderedDict([('rating_accuracy', 0.40549543), ('root_mean_squared_error', 1.192835), ('loss', 1.4228554)])\n",
      "Train round 131: OrderedDict([('rating_accuracy', 0.49929106), ('root_mean_squared_error', 1.0363958), ('loss', 1.0741162)])\n",
      "Train round 132: OrderedDict([('rating_accuracy', 0.45439777), ('root_mean_squared_error', 1.1099253), ('loss', 1.2319341)])\n",
      "Train round 133: OrderedDict([('rating_accuracy', 0.48646343), ('root_mean_squared_error', 1.026906), ('loss', 1.0545361)])\n",
      "Train round 134: OrderedDict([('rating_accuracy', 0.4844424), ('root_mean_squared_error', 1.0493115), ('loss', 1.1010548)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:12:20.133795: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 135: OrderedDict([('rating_accuracy', 0.47992486), ('root_mean_squared_error', 1.0784314), ('loss', 1.1630142)])\n",
      "Train round 136: OrderedDict([('rating_accuracy', 0.49049753), ('root_mean_squared_error', 1.103782), ('loss', 1.2183348)])\n",
      "Train round 137: OrderedDict([('rating_accuracy', 0.48623446), ('root_mean_squared_error', 1.0563391), ('loss', 1.1158525)])\n",
      "Train round 138: OrderedDict([('rating_accuracy', 0.42516732), ('root_mean_squared_error', 1.1845919), ('loss', 1.4032578)])\n",
      "Train round 139: OrderedDict([('rating_accuracy', 0.48778194), ('root_mean_squared_error', 1.0749232), ('loss', 1.1554598)])\n",
      "Train round 140: OrderedDict([('rating_accuracy', 0.5014597), ('root_mean_squared_error', 1.0772468), ('loss', 1.1604607)])\n",
      "Train round 141: OrderedDict([('rating_accuracy', 0.46022066), ('root_mean_squared_error', 1.1944859), ('loss', 1.4267967)])\n",
      "Train round 142: OrderedDict([('rating_accuracy', 0.4471611), ('root_mean_squared_error', 1.1429961), ('loss', 1.3064401)])\n",
      "Train round 143: OrderedDict([('rating_accuracy', 0.4440669), ('root_mean_squared_error', 1.1131922), ('loss', 1.239197)])\n",
      "Train round 144: OrderedDict([('rating_accuracy', 0.42358854), ('root_mean_squared_error', 1.2129742), ('loss', 1.4713063)])\n",
      "Train round 145: OrderedDict([('rating_accuracy', 0.47945204), ('root_mean_squared_error', 1.0400978), ('loss', 1.0818034)])\n",
      "Train round 146: OrderedDict([('rating_accuracy', 0.5026895), ('root_mean_squared_error', 1.1113168), ('loss', 1.2350252)])\n",
      "Train round 147: OrderedDict([('rating_accuracy', 0.520545), ('root_mean_squared_error', 1.0097462), ('loss', 1.0195874)])\n",
      "Train round 148: OrderedDict([('rating_accuracy', 0.4472347), ('root_mean_squared_error', 1.1982635), ('loss', 1.4358354)])\n",
      "Train round 149: OrderedDict([('rating_accuracy', 0.5030371), ('root_mean_squared_error', 1.049102), ('loss', 1.100615)])\n",
      "Train round 150: OrderedDict([('rating_accuracy', 0.42121032), ('root_mean_squared_error', 1.2162052), ('loss', 1.4791551)])\n",
      "Train round 151: OrderedDict([('rating_accuracy', 0.4623606), ('root_mean_squared_error', 1.0965458), ('loss', 1.2024128)])\n",
      "Train round 152: OrderedDict([('rating_accuracy', 0.4424668), ('root_mean_squared_error', 1.0982369), ('loss', 1.2061243)])\n",
      "Train round 153: OrderedDict([('rating_accuracy', 0.4551101), ('root_mean_squared_error', 1.1661192), ('loss', 1.3598341)])\n",
      "Train round 154: OrderedDict([('rating_accuracy', 0.48304465), ('root_mean_squared_error', 1.0756348), ('loss', 1.1569903)])\n",
      "Train round 155: OrderedDict([('rating_accuracy', 0.5003714), ('root_mean_squared_error', 1.0804435), ('loss', 1.1673584)])\n",
      "Train round 156: OrderedDict([('rating_accuracy', 0.46004978), ('root_mean_squared_error', 1.1485267), ('loss', 1.3191135)])\n",
      "Train round 157: OrderedDict([('rating_accuracy', 0.49520278), ('root_mean_squared_error', 1.0487182), ('loss', 1.0998099)])\n",
      "Train round 158: OrderedDict([('rating_accuracy', 0.4982974), ('root_mean_squared_error', 1.0452763), ('loss', 1.0926025)])\n",
      "Train round 159: OrderedDict([('rating_accuracy', 0.49485785), ('root_mean_squared_error', 0.9778833), ('loss', 0.95625573)])\n",
      "Train round 160: OrderedDict([('rating_accuracy', 0.45830953), ('root_mean_squared_error', 1.1528888), ('loss', 1.3291526)])\n",
      "Train round 161: OrderedDict([('rating_accuracy', 0.48392138), ('root_mean_squared_error', 1.0524476), ('loss', 1.1076458)])\n",
      "Train round 162: OrderedDict([('rating_accuracy', 0.49082458), ('root_mean_squared_error', 1.070343), ('loss', 1.1456343)])\n",
      "Train round 163: OrderedDict([('rating_accuracy', 0.3951049), ('root_mean_squared_error', 1.2848815), ('loss', 1.6509204)])\n",
      "Train round 164: OrderedDict([('rating_accuracy', 0.4247072), ('root_mean_squared_error', 1.1725985), ('loss', 1.3749872)])\n",
      "Train round 165: OrderedDict([('rating_accuracy', 0.48120123), ('root_mean_squared_error', 1.0385), ('loss', 1.0784823)])\n",
      "Train round 166: OrderedDict([('rating_accuracy', 0.49988833), ('root_mean_squared_error', 1.0211213), ('loss', 1.0426886)])\n",
      "Train round 167: OrderedDict([('rating_accuracy', 0.50504595), ('root_mean_squared_error', 1.0541995), ('loss', 1.1113366)])\n",
      "Train round 168: OrderedDict([('rating_accuracy', 0.4704969), ('root_mean_squared_error', 1.0888882), ('loss', 1.1856775)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:13:57.519159: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 169: OrderedDict([('rating_accuracy', 0.48253378), ('root_mean_squared_error', 1.0472294), ('loss', 1.0966895)])\n",
      "Train round 170: OrderedDict([('rating_accuracy', 0.47840938), ('root_mean_squared_error', 1.0741622), ('loss', 1.1538247)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:14:03.244773: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 171: OrderedDict([('rating_accuracy', 0.4578644), ('root_mean_squared_error', 1.1023153), ('loss', 1.2150991)])\n",
      "Train round 172: OrderedDict([('rating_accuracy', 0.4988759), ('root_mean_squared_error', 1.0235493), ('loss', 1.0476532)])\n",
      "Train round 173: OrderedDict([('rating_accuracy', 0.4995336), ('root_mean_squared_error', 1.0532734), ('loss', 1.109385)])\n",
      "Train round 174: OrderedDict([('rating_accuracy', 0.5188465), ('root_mean_squared_error', 1.0447065), ('loss', 1.0914116)])\n",
      "Train round 175: OrderedDict([('rating_accuracy', 0.51865953), ('root_mean_squared_error', 1.0967755), ('loss', 1.2029164)])\n",
      "Train round 176: OrderedDict([('rating_accuracy', 0.40532365), ('root_mean_squared_error', 1.151339), ('loss', 1.3255816)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:14:20.689504: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 177: OrderedDict([('rating_accuracy', 0.45676392), ('root_mean_squared_error', 1.1125801), ('loss', 1.2378343)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:14:22.126071: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 178: OrderedDict([('rating_accuracy', 0.4977778), ('root_mean_squared_error', 1.0680823), ('loss', 1.1408)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:14:26.316976: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 179: OrderedDict([('rating_accuracy', 0.46382636), ('root_mean_squared_error', 1.1139178), ('loss', 1.2408129)])\n",
      "Train round 180: OrderedDict([('rating_accuracy', 0.52126956), ('root_mean_squared_error', 1.0119665), ('loss', 1.0240762)])\n",
      "Train round 181: OrderedDict([('rating_accuracy', 0.49744305), ('root_mean_squared_error', 1.0094811), ('loss', 1.019052)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:14:36.570799: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 182: OrderedDict([('rating_accuracy', 0.44759652), ('root_mean_squared_error', 1.0885965), ('loss', 1.1850423)])\n",
      "Train round 183: OrderedDict([('rating_accuracy', 0.53471553), ('root_mean_squared_error', 1.0553436), ('loss', 1.1137502)])\n",
      "Train round 184: OrderedDict([('rating_accuracy', 0.5187139), ('root_mean_squared_error', 1.0608515), ('loss', 1.1254058)])\n",
      "Train round 185: OrderedDict([('rating_accuracy', 0.5228547), ('root_mean_squared_error', 0.9993747), ('loss', 0.99874973)])\n",
      "Train round 186: OrderedDict([('rating_accuracy', 0.46014994), ('root_mean_squared_error', 1.0849102), ('loss', 1.1770301)])\n",
      "Train round 187: OrderedDict([('rating_accuracy', 0.5371047), ('root_mean_squared_error', 0.94648385), ('loss', 0.89583164)])\n",
      "Train round 188: OrderedDict([('rating_accuracy', 0.4170076), ('root_mean_squared_error', 1.1426402), ('loss', 1.3056266)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:14:57.643617: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 189: OrderedDict([('rating_accuracy', 0.49276486), ('root_mean_squared_error', 1.0643469), ('loss', 1.1328344)])\n",
      "Train round 190: OrderedDict([('rating_accuracy', 0.47569528), ('root_mean_squared_error', 1.033097), ('loss', 1.0672896)])\n",
      "Train round 191: OrderedDict([('rating_accuracy', 0.5228403), ('root_mean_squared_error', 1.0039006), ('loss', 1.0078166)])\n",
      "Train round 192: OrderedDict([('rating_accuracy', 0.5161152), ('root_mean_squared_error', 1.0015631), ('loss', 1.0031286)])\n",
      "Train round 193: OrderedDict([('rating_accuracy', 0.48848033), ('root_mean_squared_error', 1.0237516), ('loss', 1.0480673)])\n",
      "Train round 194: OrderedDict([('rating_accuracy', 0.5494015), ('root_mean_squared_error', 0.93677104), ('loss', 0.87754005)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:15:15.494616: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n",
      "2022-03-29 23:15:15.621702: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 195: OrderedDict([('rating_accuracy', 0.45751446), ('root_mean_squared_error', 1.1158794), ('loss', 1.2451869)])\n",
      "Train round 196: OrderedDict([('rating_accuracy', 0.52707183), ('root_mean_squared_error', 0.98150325), ('loss', 0.9633488)])\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-03-29 23:15:21.369733: W tensorflow/core/data/root_dataset.cc:200] Optimization loop failed: CANCELLED: Operation was cancelled\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train round 197: OrderedDict([('rating_accuracy', 0.44408476), ('root_mean_squared_error', 1.0732454), ('loss', 1.1518557)])\n",
      "Train round 198: OrderedDict([('rating_accuracy', 0.53515095), ('root_mean_squared_error', 0.9740603), ('loss', 0.94879365)])\n",
      "Train round 199: OrderedDict([('rating_accuracy', 0.41984963), ('root_mean_squared_error', 1.1337314), ('loss', 1.2853467)])\n"
     ]
    }
   ],
   "source": [
    "train_losses, train_accs, train_rmses = [], [], []\n",
    "\n",
    "state = training_process.initialize()\n",
    "\n",
    "# This may take a couple minutes to run.\n",
    "for i in range(NUM_ROUNDS):\n",
    "    federated_train_data = np.random.choice(\n",
    "        tf_trainsets, size=50, replace=False\n",
    "    ).tolist()\n",
    "    state, metrics = training_process.next(state, federated_train_data)\n",
    "    print(f\"Train round {i}:\", metrics[\"train\"])\n",
    "    train_losses.append(metrics[\"train\"][\"loss\"])\n",
    "    train_accs.append(metrics[\"train\"][\"rating_accuracy\"])\n",
    "    train_rmses.append(metrics[\"train\"][\"root_mean_squared_error\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "5f7b7e88",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use(\"dark_background\")\n",
    "plt.rcParams[\"figure.figsize\"] = [10, 8]\n",
    "plt.plot(range(NUM_ROUNDS), train_losses)\n",
    "plt.ylabel(\"Train Loss\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "bae5a02a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(NUM_ROUNDS), train_accs)\n",
    "plt.ylabel(\"Train accuracy\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "fc7d63ee",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(NUM_ROUNDS), train_rmses)\n",
    "plt.ylabel(\"Train RMSE\")\n",
    "plt.xlabel(\"Round\")\n",
    "plt.title(\"Train RMSE\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "3b49fe7c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE on Test Set: 1.6366868\n"
     ]
    }
   ],
   "source": [
    "eval_metrics = evaluation_computation(state.model, tf_testsets)\n",
    "print(\"RMSE on Test Set:\", eval_metrics[\"eval\"][\"root_mean_squared_error\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1cdb703",
   "metadata": {},
   "source": [
    "## Step 9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "1e12276d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# already incorporated in previous steps"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "54047244",
   "metadata": {},
   "source": [
    "## Step 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "7f8dd92a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# report"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}