Tensorflow-Lets-Cook-in-ML.ipynb

tensorflow-lets-cook-in-ml.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Tensorflow-Lets-Cook-in-ML.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "swift",
      "display_name": "Swift"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/rakesh4real/33fcad0f4fab7b0ccafe04f1397b2279/tensorflow-lets-cook-in-ml.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kZRlD4utdPuX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import TensorFlow"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VoWRtV0ujokr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "struct Model: Layer {\n",
        "    var conv = Conv2D<Float>(filterShape: (5, 5, 3, 6))\n",
        "    var maxpool = MaxPool2D<Float>(poolSize: (2, 2), strides: (2, 2))\n",
        "    var flatten = Flatten<Float>()\n",
        "    var dense = Dense<Float>(inputSize: 36 * 6, outputSize: 10)\n",
        "\n",
        "    @differentiable\n",
        "    func callAsFunction(_ input: Tensor<Float>) -> Tensor<Float> {\n",
        "        return input.sequenced(through: conv, maxpool, flatten, dense)\n",
        "    }\n",
        "}\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5O-5kCEniSpJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "// Use random training data.\n",
        "let x = Tensor<Float>(randomNormal: [10, 16, 16, 3]) //dimensions\n",
        "let y = Tensor<Int32>(rangeFrom: 0, to: 10, stride: 1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x44kFjA0tRFB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "var model = Model()\n",
        "let opt = SGD(for: model)\n",
        "Context.local.learningPhase = .training"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mMxjNQIDtxgR",
        "colab_type": "code",
        "outputId": "6ced9d9a-3e4e-4431-b5c0-a7f6e93175a7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 697
        }
      },
      "source": [
        "for i in 1...20 {\n",
        "    print(\"Starting training step \\(i)\")\n",
        "    let (loss, grads) = valueWithGradient(at: model) { model -> Tensor<Float> in\n",
        "        let logits = model(x)\n",
        "        return softmaxCrossEntropy(logits: logits, labels: y)\n",
        "    }\n",
        "    print(\"Loss: \\(loss)\")\n",
        "    opt.update(&model.allDifferentiableVariables, along: grads)\n",
        "}"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Starting training step 1\n",
            "Loss: 2.5361626\n",
            "Starting training step 2\n",
            "Loss: 2.2827177\n",
            "Starting training step 3\n",
            "Loss: 2.059002\n",
            "Starting training step 4\n",
            "Loss: 1.8636631\n",
            "Starting training step 5\n",
            "Loss: 1.688523\n",
            "Starting training step 6\n",
            "Loss: 1.5316066\n",
            "Starting training step 7\n",
            "Loss: 1.389894\n",
            "Starting training step 8\n",
            "Loss: 1.2644584\n",
            "Starting training step 9\n",
            "Loss: 1.1527299\n",
            "Starting training step 10\n",
            "Loss: 1.0534394\n",
            "Starting training step 11\n",
            "Loss: 0.9644995\n",
            "Starting training step 12\n",
            "Loss: 0.8861335\n",
            "Starting training step 13\n",
            "Loss: 0.81693286\n",
            "Starting training step 14\n",
            "Loss: 0.75583524\n",
            "Starting training step 15\n",
            "Loss: 0.70112455\n",
            "Starting training step 16\n",
            "Loss: 0.6510164\n",
            "Starting training step 17\n",
            "Loss: 0.6062659\n",
            "Starting training step 18\n",
            "Loss: 0.5659787\n",
            "Starting training step 19\n",
            "Loss: 0.52945524\n",
            "Starting training step 20\n",
            "Loss: 0.4967416\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zCt9VnVuSLcS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}