training-phases-plots/training-phases-float16-vs-float32-tf.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "RJqZjpyIx7Qb",
    "outputId": "cdb783e0-54c5-4de2-d781-78b8ff88eea2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: deepxde in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (1.9.3)\n",
      "Requirement already satisfied: matplotlib in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from deepxde) (3.8.0)\n",
      "Requirement already satisfied: numpy in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from deepxde) (1.26.0)\n",
      "Requirement already satisfied: scikit-learn in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from deepxde) (1.3.1)\n",
      "Requirement already satisfied: scikit-optimize>=0.9.0 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from deepxde) (0.9.0)\n",
      "Requirement already satisfied: scipy in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from deepxde) (1.11.3)\n",
      "Requirement already satisfied: joblib>=0.11 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from scikit-optimize>=0.9.0->deepxde) (1.3.2)\n",
      "Requirement already satisfied: pyaml>=16.9 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from scikit-optimize>=0.9.0->deepxde) (23.9.7)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from scikit-learn->deepxde) (3.2.0)\n",
      "Requirement already satisfied: contourpy>=1.0.1 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (1.1.1)\n",
      "Requirement already satisfied: cycler>=0.10 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (0.12.0)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (4.43.0)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (1.4.5)\n",
      "Requirement already satisfied: packaging>=20.0 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (23.2)\n",
      "Requirement already satisfied: pillow>=6.2.0 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (10.0.1)\n",
      "Requirement already satisfied: pyparsing>=2.3.1 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (3.1.1)\n",
      "Requirement already satisfied: python-dateutil>=2.7 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from matplotlib->deepxde) (2.8.2)\n",
      "Requirement already satisfied: PyYAML in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from pyaml>=16.9->scikit-optimize>=0.9.0->deepxde) (6.0.1)\n",
      "Requirement already satisfied: six>=1.5 in /Users/Jeniffer/Documents/mixed-precision-sciml/venv/lib/python3.11/site-packages (from python-dateutil>=2.7->matplotlib->deepxde) (1.16.0)\n",
      "\n",
      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m23.2.1\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m23.3.2\u001b[0m\n",
      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip install --upgrade pip\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "!pip install deepxde"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "rd7uxjWKydXN",
    "outputId": "d080dd3c-4586-4669-f362-3008e6f3a6a7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "env: DDE_BACKEND=tensorflow\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-04 19:23:47.791818: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "Using backend: tensorflow\n",
      "Other supported backends: tensorflow.compat.v1, pytorch, jax, paddle.\n",
      "paddle supports more examples now and is recommended.\n"
     ]
    }
   ],
   "source": [
    "%env DDE_BACKEND=tensorflow\n",
    "import deepxde as dde\n",
    "import numpy as np\n",
    "from deepxde.backend import tf\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def func(x):\n",
    "    return x * np.sin(5 * x)\n",
    "\n",
    "def train(model, iterations, display_every, loss_test, loss_train):\n",
    "    for i in range(iterations):\n",
    "        model.train_state.set_data_train(\n",
    "            *model.data.train_next_batch(model.batch_size)\n",
    "        )\n",
    "        model.train_step(\n",
    "            model.train_state.X_train,\n",
    "            model.train_state.y_train,\n",
    "            model.train_state.train_aux_vars,\n",
    "        )\n",
    "\n",
    "        X_test, y_test = model.data.test()\n",
    "        y_pred = model.predict(X_test)\n",
    "        l2_loss_test = np.linalg.norm(y_pred - y_test)*np.linalg.norm(y_pred - y_test)\n",
    "        # l2r = np.linalg.norm(y_pred - y_test) / np.linalg.norm(y_test)\n",
    "        loss_test.append(l2_loss_test)\n",
    "\n",
    "        X_train, y_train = model.data.train_x, model.data.train_y\n",
    "        y_pred = model.predict(X_train)\n",
    "        l2_loss_train = np.linalg.norm(y_pred - y_train)*np.linalg.norm(y_pred - y_train)\n",
    "        # l2l = np.linalg.norm(y_pred - y_train)/np.linalg.norm(y_train)\n",
    "        loss_train.append(l2_loss_train)\n",
    "\n",
    "        model.train_state.epoch += 1\n",
    "        model.train_state.step += 1\n",
    "        if model.train_state.step % display_every == 0 or i + 1 == iterations:\n",
    "            print(str(model.train_state.step) + \" \" + str(loss_test[-1]))\n",
    "\n",
    "def test(seed):\n",
    "    dde.config.set_default_float(\"float32\")\n",
    "    dde.config.set_random_seed(seed)\n",
    "    print(\"Training Float32:\")\n",
    "    geom = dde.geometry.Interval(-1, 1)\n",
    "    num_train = 16\n",
    "    num_test = 100\n",
    "    data = dde.data.Function(geom, func, num_train, num_test)\n",
    "\n",
    "    activation = \"tanh\"\n",
    "    initializer = \"Glorot uniform\"\n",
    "    net = dde.nn.FNN([1] + [10] * 2 + [1], activation, kernel_initializer =  tf.keras.initializers.glorot_uniform(seed=seed))\n",
    "\n",
    "    model = dde.Model(data, net)\n",
    "    model.compile(\"adam\", lr=0.001, metrics=[\"l2 relative error\"])\n",
    "    loss_test_32 = []\n",
    "    loss_train_32 = []\n",
    "    train(model, 10000, 1000, loss_test_32, loss_train_32) # 10000\n",
    "\n",
    "    dde.config.set_default_float(\"float16\")\n",
    "    dde.config.set_random_seed(seed)\n",
    "    print(\"Training Float16:\")\n",
    "    geom = dde.geometry.Interval(-1, 1)\n",
    "    num_train = 16\n",
    "    num_test = 100\n",
    "    data = dde.data.Function(geom, func, num_train, num_test)\n",
    "\n",
    "    activation = \"tanh\"\n",
    "    initializer = \"Glorot uniform\"\n",
    "    net = dde.nn.FNN([1] + [10] * 2 + [1], activation, kernel_initializer =  tf.keras.initializers.glorot_uniform(seed=seed))\n",
    "\n",
    "    model = dde.Model(data, net)\n",
    "    model.compile(\"adam\", lr=0.001, metrics=[\"l2 relative error\"])\n",
    "    loss_test_16 = []\n",
    "    loss_train_16 = []\n",
    "    train(model, 10000, 1000, loss_test_16, loss_train_16) # 10000\n",
    "\n",
    "    return loss_test_32, loss_test_16, loss_train_32, loss_train_16"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 880
    },
    "id": "OZrzJTo-y9dk",
    "outputId": "24cf0cfe-f737-4fc0-87db-151250161ac3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Set the default float type to float32\n",
      "Training Float32:\n",
      "Compiling model...\n",
      "'compile' took 0.010250 s\n",
      "\n",
      "1000 3.652507\n",
      "2000 0.14346156\n",
      "3000 0.09265575\n",
      "4000 0.058981482\n",
      "5000 0.021714209\n",
      "6000 0.007991245\n",
      "7000 0.006644982\n",
      "8000 0.006146135\n",
      "9000 0.0054103755\n",
      "10000 0.005404102\n",
      "Set the default float type to float16\n",
      "Training Float16:\n",
      "Compiling model...\n",
      "'compile' took 0.003048 s\n",
      "\n",
      "1000 0.2372\n",
      "2000 0.1863\n",
      "3000 0.1628\n",
      "4000 0.1489\n",
      "5000 0.1304\n",
      "6000 0.11786\n",
      "7000 0.1064\n",
      "8000 0.1009\n",
      "9000 0.096\n",
      "10000 0.09357\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 360x270 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "seed = 3\n",
    "loss_test_32, loss_test_16, loss_train_32, loss_train_16 = test(seed)\n",
    "x = [i for i in range(len(loss_test_32))]\n",
    "#plt.figure(figsize=(4.8,3.6))\n",
    "plt.figure(figsize=(3.6,2.7))\n",
    "plt.xlabel('No. of iterations')\n",
    "plt.ylabel('L² training loss')\n",
    "plt.yscale('log')\n",
    "# plt.plot(x, loss_test_32, label = \"Testing float32\", color = 'orange', linestyle='dashed')\n",
    "# plt.plot(x, loss_test_16, label = \"Testing float16\", color = 'cornflowerblue', linestyle='dashed')\n",
    "# plt.plot(x, loss_train_32, label = \"Training float32\", color = 'red')\n",
    "# plt.plot(x, loss_train_16, label = \"Training float16\", color = 'midnightblue')\n",
    "plt.plot(x, loss_train_32, label = \"Float32\", color = 'red')\n",
    "plt.plot(x, loss_train_16, label = \"Float16\", color = 'midnightblue')\n",
    "plt.legend(loc='upper right', frameon = False)\n",
    "plt.savefig(\"phases-\" + str(seed) + \".pdf\", bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 357
    },
    "id": "Zln7o5NE8Glm",
    "outputId": "9652e51d-af3f-489c-8e29-096fd3e3bf4b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 360x270 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = [i for i in range(len(loss_test_32))]\n",
    "#plt.figure(figsize=(4.8,3.6))\n",
    "plt.figure(figsize=(3.6,2.7))\n",
    "plt.xlabel('No. of iterations')\n",
    "plt.ylabel('L² loss')\n",
    "plt.yscale('log')\n",
    "plt.plot(x, loss_test_32, label = \"Testing float32\", color = 'orange', linestyle='dashed')\n",
    "plt.plot(x, loss_test_16, label = \"Testing float16\", color = 'cornflowerblue', linestyle='dashed')\n",
    "plt.plot(x, loss_train_32, label = \"Training float32\", color = 'red')\n",
    "plt.plot(x, loss_train_16, label = \"Training float16\", color = 'midnightblue')\n",
    "# plt.plot(x, loss_train_32, label = \"Float32\", color = 'red')\n",
    "# plt.plot(x, loss_train_16, label = \"Float16\", color = 'midnightblue')\n",
    "plt.legend(loc='upper right', frameon = False)\n",
    "plt.savefig(\"phases-\" + str(seed) + \"-full.pdf\", bbox_inches='tight')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "gpuType": "T4",
   "provenance": []
  },
  "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": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}