min-dalle-test/min_dalle.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/kuprel/min-dalle/blob/main/min_dalle.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3WL-G_f2_ld8"
      },
      "source": [
        "# min(DALL·E)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Zl_ZFisFApeh"
      },
      "source": [
        "### Install"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "cellView": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ix_xt4X1_6F4",
        "outputId": "20acdc7e-e2e5-4d27-95cc-8b34f069fdcf"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting min-dalle\n",
            "  Downloading min-dalle-0.2.27.tar.gz (11 kB)\n",
            "Requirement already satisfied: torch>=1.10.0 in /usr/local/lib/python3.7/dist-packages (from min-dalle) (1.11.0+cu113)\n",
            "Requirement already satisfied: typing_extensions>=4.1.0 in /usr/local/lib/python3.7/dist-packages (from min-dalle) (4.1.1)\n",
            "Requirement already satisfied: numpy>=1.21 in /usr/local/lib/python3.7/dist-packages (from min-dalle) (1.21.6)\n",
            "Requirement already satisfied: pillow>=7.1 in /usr/local/lib/python3.7/dist-packages (from min-dalle) (7.1.2)\n",
            "Requirement already satisfied: requests>=2.23 in /usr/local/lib/python3.7/dist-packages (from min-dalle) (2.23.0)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests>=2.23->min-dalle) (2022.6.15)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests>=2.23->min-dalle) (2.10)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests>=2.23->min-dalle) (3.0.4)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests>=2.23->min-dalle) (1.24.3)\n",
            "Building wheels for collected packages: min-dalle\n",
            "  Building wheel for min-dalle (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for min-dalle: filename=min_dalle-0.2.27-py3-none-any.whl size=11247 sha256=788e2010694dd0085a6352e07de88d108d171e4f04457575a45e5b02c7914822\n",
            "  Stored in directory: /root/.cache/pip/wheels/da/00/12/9761a7506eb7e3e1f4ca383cd37e426dd146eea5bc5cb016f3\n",
            "Successfully built min-dalle\n",
            "Installing collected packages: min-dalle\n",
            "Successfully installed min-dalle-0.2.27\n",
            "Tue Jul  5 09:43:32 2022       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   34C    P0    27W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ]
        }
      ],
      "source": [
        "! pip install min-dalle\n",
        "! nvidia-smi"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kViq2dMbGDKt"
      },
      "source": [
        "### Load Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8W-L2ICFGFup",
        "outputId": "d652ed35-aed2-49c2-918e-5c77d580051a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "initializing MinDalle\n",
            "downloading tokenizer params\n",
            "intializing TextTokenizer\n",
            "downloading encoder params\n",
            "initializing DalleBartEncoder\n",
            "downloading decoder params\n",
            "initializing DalleBartDecoder\n",
            "downloading detokenizer params\n",
            "initializing VQGanDetokenizer\n"
          ]
        }
      ],
      "source": [
        "from PIL import Image\n",
        "from IPython.display import update_display\n",
        "import numpy\n",
        "from math import log2\n",
        "from min_dalle import MinDalle\n",
        "\n",
        "model = MinDalle(is_mega=True, is_reusable=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c52TV1GbBNgS"
      },
      "source": [
        "### Generate Images\n",
        "\n",
        "- `grid_size` Size of the image grid.  3x3 works best when displaying intermediate results.  4x4 has been tested to work on T4 and P100 with intermediate_image_count = 1\n",
        "\n",
        "- `seed` Set the seed to a positive number for reproducible results\n",
        "\n",
        "- `intermediate_image_count` Set the number of intermediate images to show while running\n",
        "\n",
        "- `supercondition_factor` Lower results in a wider variety of images but less agreement with the text"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 819
        },
        "id": "nQ0UG05dA4p2",
        "outputId": "6af365b0-2cf3-4cd9-f842-077712320fb9"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=768x768 at 0x7FCC11FB2F10>"
            ],
            "image/png": "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
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CPU times: user 32.6 s, sys: 263 ms, total: 32.8 s\n",
            "Wall time: 32.5 s\n"
          ]
        }
      ],
      "source": [
        "%%time\n",
        "\n",
        "text = \"Dali painting of WALL·E\" #@param {type:\"string\"}\n",
        "grid_size = 3 #@param {type:\"integer\"}\n",
        "seed = -1 #@param {type:\"integer\"}\n",
        "intermediate_image_count = 4 #@param [\"1\", \"2\", \"4\", \"8\", \"16\"] {type:\"raw\"}\n",
        "supercondition_factor = 8 #@param [\"2\", \"4\", \"8\", \"16\", \"32\", \"64\"] {type:\"raw\"}\n",
        "\n",
        "image_stream = model.generate_image_stream(\n",
        "    text=text,\n",
        "    seed=seed,\n",
        "    grid_size=grid_size,\n",
        "    log2_mid_count=log2(intermediate_image_count),\n",
        "    log2_supercondition_factor=log2(supercondition_factor)\n",
        ")\n",
        "\n",
        "is_first = True\n",
        "for image in image_stream:\n",
        "    display_image = display if is_first else update_display\n",
        "    display_image(image, display_id=1)\n",
        "    is_first = False"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [
        "Zl_ZFisFApeh"
      ],
      "name": "min-dalle",
      "provenance": [],
      "include_colab_link": true
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}