ImageClassification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "authorship_tag": "ABX9TyPeLA4ZZFnQzzC0ZjmWisWH",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/serdarch/SerLab/blob/main/ImageClassification.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Image classification**\n",
        "\n",
        "Image classification tutorial based on the pretrained PyTorch baselines.\n",
        "Used model: ViT_h_14 with Weights IMAGENET1K_SWAG_E2E_V1\n",
        "\n",
        "Please save a copy of this tutorial into your own environment/drive folder."
      ],
      "metadata": {
        "id": "7kG9ZBiN238_"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from torchvision.io import read_image"
      ],
      "metadata": {
        "id": "UB3maNwt0roV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!wget http://images.cocodataset.org/val2017/000000005477.jpg -q -O input.jpg"
      ],
      "metadata": {
        "id": "5_vXR0dT0r9L"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "im1=read_image(\"input.jpg\")"
      ],
      "metadata": {
        "id": "-hGZui_z0w97"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from torchvision.models import vit_h_14, ViT_H_14_Weights\n",
        "\n",
        "# Step 1: Initialize model with the best available weights\n",
        "weights = ViT_H_14_Weights.DEFAULT\n",
        "model = vit_h_14(weights='DEFAULT')\n",
        "model.eval()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "h0JeHh_m5qmK",
        "outputId": "c9693af1-f20c-423b-a61f-1e13b42f1ea8"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Downloading: \"https://download.pytorch.org/models/vit_h_14_swag-80465313.pth\" to /root/.cache/torch/hub/checkpoints/vit_h_14_swag-80465313.pth\n",
            "100%|██████████| 2.36G/2.36G [00:26<00:00, 94.4MB/s]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "VisionTransformer(\n",
              "  (conv_proj): Conv2d(3, 1280, kernel_size=(14, 14), stride=(14, 14))\n",
              "  (encoder): Encoder(\n",
              "    (dropout): Dropout(p=0.0, inplace=False)\n",
              "    (layers): Sequential(\n",
              "      (encoder_layer_0): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_1): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_2): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_3): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_4): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_5): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_6): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_7): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_8): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_9): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_10): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_11): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_12): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_13): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_14): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_15): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_16): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_17): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_18): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_19): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_20): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_21): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_22): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_23): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_24): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_25): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_26): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_27): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_28): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_29): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_30): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "      (encoder_layer_31): EncoderBlock(\n",
              "        (ln_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (self_attention): MultiheadAttention(\n",
              "          (out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)\n",
              "        )\n",
              "        (dropout): Dropout(p=0.0, inplace=False)\n",
              "        (ln_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "        (mlp): MLPBlock(\n",
              "          (0): Linear(in_features=1280, out_features=5120, bias=True)\n",
              "          (1): GELU(approximate='none')\n",
              "          (2): Dropout(p=0.0, inplace=False)\n",
              "          (3): Linear(in_features=5120, out_features=1280, bias=True)\n",
              "          (4): Dropout(p=0.0, inplace=False)\n",
              "        )\n",
              "      )\n",
              "    )\n",
              "    (ln): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)\n",
              "  )\n",
              "  (heads): Sequential(\n",
              "    (head): Linear(in_features=1280, out_features=1000, bias=True)\n",
              "  )\n",
              ")"
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "th3lBuOHzu6e",
        "outputId": "e9778cb6-7e3f-48d1-cec1-bf4b4d2604fa"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "airliner: 91.4%\n"
          ]
        }
      ],
      "source": [
        "# Step 2: Initialize the inference transforms\n",
        "preprocess = weights.transforms()\n",
        "\n",
        "# Step 3: Apply inference preprocessing transforms\n",
        "batch = preprocess(im1).unsqueeze(0)\n",
        "\n",
        "# Step 4: Use the model and print the predicted category\n",
        "prediction = model(batch).squeeze(0).softmax(0)\n",
        "class_id = prediction.argmax().item()\n",
        "score = prediction[class_id].item()\n",
        "category_name = weights.meta[\"categories\"][class_id]\n",
        "print(f\"{category_name}: {100 * score:.1f}%\") #airliner: ~91.4%"
      ]
    }
  ]
}