deeplake_bug.ipynb

deeplake_bug.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyOEgStnfvYmMzgs2r86pnpi",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/daniel-falk/2fae689f61345ab109c9810c67ed7ecd/deeplake_bug.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "TOKEN=\"REPLACE_ME\""
      ],
      "metadata": {
        "id": "BOfKhcX1xlVL"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "MzyKTo-vxbbc"
      },
      "outputs": [],
      "source": [
        "!pip install deeplake"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import deeplake\n",
        "import sys\n",
        "\n",
        "print(\"Deeplake: \", deeplake.__version__)\n",
        "print(\"Python: \", sys.version)"
      ],
      "metadata": {
        "id": "RWLOYpkAyZwV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Fast\n",
        "Approx 70-130 images per second in notebook.\n",
        "Approx 110 images per socond on my desktop."
      ],
      "metadata": {
        "id": "LckqjGyWxxav"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "%%time\n",
        "\n",
        "ds = deeplake.load('hub://activeloop/imagenet-train', token=TOKEN)\n",
        "\n",
        "@deeplake.compute\n",
        "def calculate_mean(sample_in, sample_out):\n",
        "  sample_in.images.numpy()\n",
        "  return None\n",
        "\n",
        "tmp_ds = deeplake.empty(\"mem://tmp_ds\")\n",
        "tmp_ds.create_tensor(\"dummy\", htype=\"generic\")\n",
        "calculate_mean().eval(ds, tmp_ds, num_workers=8)"
      ],
      "metadata": {
        "id": "rombtM_Hx2HM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Slow\n",
        "The only difference is that the output dataset does not have any tensors specified.\n",
        "\n",
        "Approx 30 images per second in notebook. Approx 1.5 images per second on my desktop."
      ],
      "metadata": {
        "id": "s0ScH2pkxehZ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "%%time\n",
        "ds = deeplake.load('hub://activeloop/imagenet-train', token=TOKEN)\n",
        "\n",
        "@deeplake.compute\n",
        "def calculate_mean(sample_in, sample_out):\n",
        "  sample_in.images.numpy()\n",
        "  return None\n",
        "\n",
        "tmp_ds = deeplake.empty(\"mem://tmp_ds\")\n",
        "#tmp_ds.create_tensor(\"dummy\", htype=\"generic\")\n",
        "calculate_mean().eval(ds, tmp_ds, num_workers=8)"
      ],
      "metadata": {
        "id": "cwqmMDHXxiJI"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}