self_attention_analysis.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "id": "921e17cb",
      "metadata": {
        "id": "921e17cb"
      },
      "source": [
        "# Self-Attention Analysis"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "id": "203dfb55",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "203dfb55",
        "outputId": "05af1fa7-424e-4f96-a3d4-49e8fe49fc6d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Torch: 2.2.0+cu121 \n",
            "Availability: True\n",
            "Number: 1 \n",
            "Current device: 0 \n",
            "First device: <torch.cuda.device object at 0x7f30b5b58d50> \n",
            "Device name: NVIDIA GeForce RTX 3090\n"
          ]
        }
      ],
      "source": [
        "import torch\n",
        "import os\n",
        "\n",
        "#@markdown Execute this cell to check the Colab environment, including GPU availability.\n",
        "\n",
        "# check gpu env\n",
        "print(f\"Torch: {torch.__version__} \\n\" + \n",
        "      f\"Availability: {torch.cuda.is_available()}\")\n",
        "assert torch.cuda.is_available() == True, \"The GPU is turned off. To turn it on, navigate to: Runtime > Change Runtime Type.\"\n",
        "print(f\"Number: {torch.cuda.device_count()} \\n\" +\n",
        "      f\"Current device: {torch.cuda.current_device()} \\n\" +\n",
        "      f\"First device: {torch.cuda.device(0)} \\n\" +\n",
        "      f\"Device name: {torch.cuda.get_device_name(0)}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "id": "fff99e10",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 37
        },
        "id": "fff99e10",
        "outputId": "5b985121-aea7-4509-c5c1-f3d09235d659"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import math\n",
        "import numpy as np\n",
        "from torch.utils.data import DataLoader\n",
        "\n",
        "import torchvision.datasets as datasets\n",
        "import timm.data.transforms_factory as tff\n",
        "from fastai.data.external import untar_data, URLs\n",
        "from utils import subsample\n",
        "\n",
        "#@markdown Execute this cell to download and import the *Imagenette* dataset, which is a small subset of ImageNet, for simplicity.\n",
        "\n",
        "# use imagenette (a small subset of imagenet) as an example for simplicity\n",
        "imagenet_path = untar_data(URLs.IMAGENETTE_160)\n",
        "\n",
        "imagenet_mean = np.array([0.485, 0.456, 0.406])\n",
        "imagenet_std = np.array([0.229, 0.224, 0.225])\n",
        "\n",
        "transform_test = tff.transforms_imagenet_eval(\n",
        "    img_size=192, mean=imagenet_mean, std=imagenet_std,\n",
        ")\n",
        "\n",
        "test_dir = os.path.join(imagenet_path, 'val')\n",
        "dataset_test = datasets.ImageFolder(test_dir, transform_test)\n",
        "dataset_test = torch.utils.data.Subset(\n",
        "    dataset_test, \n",
        "    subsample(dataset_test, ratio=math.pow(2,-6))  # use a subsampled batch\n",
        ")\n",
        "\n",
        "dataset_test = DataLoader(\n",
        "    dataset_test, \n",
        "    num_workers=1, \n",
        "    batch_size=64,\n",
        ")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4532bc85",
      "metadata": {
        "id": "4532bc85"
      },
      "source": [
        "## Load the Models"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "id": "dba96775",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dba96775",
        "outputId": "05661634-cc0e-45d9-8c26-50258f61f243"
      },
      "outputs": [],
      "source": [
        "# wSimMIM\n",
        "import torch\n",
        "from configs.config import parameters_base as parameters\n",
        "import transformers\n",
        "\n",
        "T = 0.25\n",
        "w_ps = 8\n",
        "\n",
        "DIMENSION = parameters['dimension']                                 # dimension of image\n",
        "MODEL_PATCH_SIZE = parameters['model_patch_size']                   # patch size         \n",
        "MASKING_PATCH_SIZE = parameters['masking_patch_size']               # masking patch size\n",
        "MASK_RATIO = parameters['mask_ratio']                               # model masking ratio\n",
        "INTERMEDIATE_SIZE = parameters['intermediate_size']                      # model patch size to conform with VIT specs\n",
        "\n",
        "# model config\n",
        "HIDDEN_SIZE = parameters['hidden_size']\n",
        "HIDDEN_LAYERS = parameters['num_hidden_layer']\n",
        "ATTN_HEADS = parameters['num_attention_heads']\n",
        "\n",
        "config = transformers.ViTConfig(\n",
        "    hidden_size = HIDDEN_SIZE,\n",
        "    num_hidden_layers = HIDDEN_LAYERS,\n",
        "    num_attention_heads= ATTN_HEADS,\n",
        "    image_size = DIMENSION,\n",
        "    patch_size = MODEL_PATCH_SIZE,\n",
        "    intermediate_size= INTERMEDIATE_SIZE,\n",
        "    num_channels= 3,\n",
        ")\n",
        "\n",
        "model = transformers.ViTForMaskedImageModeling(config)\n",
        "model = model.cuda()\n",
        "model = model.eval()\n",
        "\n",
        "state_dict = torch.load(f'weights/ViT-B/wSimMIM [weight_ps = {w_ps}] [T = {T}]/checkpoint_final.pth')\n",
        "_ = model.load_state_dict(state_dict, strict=False)\n",
        "model_wsimmim = model.base_model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "id": "f959e0ff",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f959e0ff",
        "outputId": "b804d3eb-e7d5-4e3c-e4b9-0a6d3755b57d"
      },
      "outputs": [],
      "source": [
        "# SimMIM\n",
        "import torch\n",
        "from configs.config import parameters_base as parameters\n",
        "import transformers\n",
        "\n",
        "DIMENSION = parameters['dimension']                                 # dimension of image\n",
        "MODEL_PATCH_SIZE = parameters['model_patch_size']                   # patch size         \n",
        "MASKING_PATCH_SIZE = parameters['masking_patch_size']               # masking patch size\n",
        "MASK_RATIO = parameters['mask_ratio']                               # model masking ratio\n",
        "INTERMEDIATE_SIZE = parameters['intermediate_size']                      # model patch size to conform with VIT specs\n",
        "\n",
        "# model config\n",
        "HIDDEN_SIZE = parameters['hidden_size']\n",
        "HIDDEN_LAYERS = parameters['num_hidden_layer']\n",
        "ATTN_HEADS = parameters['num_attention_heads']\n",
        "\n",
        "config = transformers.ViTConfig(\n",
        "    hidden_size = HIDDEN_SIZE,\n",
        "    num_hidden_layers = HIDDEN_LAYERS,\n",
        "    num_attention_heads= ATTN_HEADS,\n",
        "    image_size = DIMENSION,\n",
        "    patch_size = MODEL_PATCH_SIZE,\n",
        "    intermediate_size= INTERMEDIATE_SIZE,\n",
        "    num_channels= 3,\n",
        ")\n",
        "\n",
        "model = transformers.ViTForMaskedImageModeling(config)\n",
        "model = model.cuda()\n",
        "model = model.eval()\n",
        "\n",
        "state_dict = torch.load('weights/ViT-B/SimMIM/checkpoint_final.pth')\n",
        "_ = model.load_state_dict(state_dict, strict=False)\n",
        "model_simmim = model.base_model"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "7e1b1d97",
      "metadata": {
        "id": "7e1b1d97"
      },
      "source": [
        "## Attention Map Visualization"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "id": "17b41b3c",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 336
        },
        "id": "17b41b3c",
        "outputId": "a1bdfb4a-f8d4-490a-c618-51c878518e07",
        "scrolled": false
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 200x200 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 150x150 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# visualize attns with samples\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "def mark_token(ax, xs, batch, i, j, color='tab:red'):\n",
        "    import matplotlib.patches as patches\n",
        "    from einops import rearrange, reduce, repeat\n",
        "    from utils import restore\n",
        "    \n",
        "    ax.imshow(rearrange(restore(xs.cpu())[batch], 'c h w -> h w c'))\n",
        "    ax.set_axis_off()\n",
        "\n",
        "    # create a rectangle patch to visualize the query token\n",
        "    rect = patches.Rectangle((12 * i, 12 * j), 12, 12, linewidth=3, edgecolor=color, facecolor='none')\n",
        "    ax.add_patch(rect)\n",
        "\n",
        "def visualize_attn(ax, attns, batch, i, j, depth, head):\n",
        "    \"\"\"\n",
        "    ax: matplotlib axes\n",
        "    xs: batch\n",
        "    batch, i, j: batch coords \n",
        "    attns, depth, head: \n",
        "    \"\"\"\n",
        "    \n",
        "    attn_list = [list(attn[batch].cpu()) if attn is not None else None for attn in attns]\n",
        "    attn = torch.stack(attn_list[depth])\n",
        "    attn = attn[head]  # take head-th head\n",
        "    attn = attn[1:,1:]  # drop cls token\n",
        "    attn = attn / attn.sum(dim=-1, keepdim=True)  # normalize\n",
        "\n",
        "    attn = attn[j * 12 + i]\n",
        "    attn = attn.reshape([12, 12])\n",
        "\n",
        "    ax.imshow(attn, cmap=\"plasma\", )\n",
        "    ax.set_xticks([])\n",
        "    ax.set_yticks([])\n",
        "    ax.set_axis_off()\n",
        "    \n",
        "    \n",
        "#@markdown Execute this cell to visualize the attention maps. The red square in the visualization indicates the location of the query.\n",
        "#@markdown \n",
        "#@markdown This visualization demonstrates that the self-attentions of CL (MoCo) generally captures global relationships and objects. However, they tend to produce homogeneous attention maps for all query tokens and heads, particularly when compared to MIM. In contrast, the self-attentions of MIM (SimMIM) primarily focus on localized areas.\n",
        "\n",
        "depth = 10  #@param {type:\"slider\", min:0, max:11, step:1}\n",
        "batch = 30  #@param {type:\"integer\"}\n",
        "coord_x = 3  #@param {type:\"integer\"}\n",
        "coord_y = 8  #@param {type:\"integer\"}\n",
        "head = 0  #@param {type:\"integer\"}\n",
        "\n",
        "model_option = \"wsimmim\"  #@param [\"wsimmim\", \"simmim\"]\n",
        "if model_option == \"wsimmim\":\n",
        "    model = model_wsimmim\n",
        "elif model_option == \"simmim\":\n",
        "    model = model_simmim\n",
        "else:\n",
        "    raise Exception(\"Invalid option: \", model_option)\n",
        "\n",
        "xs, ys = next(iter(dataset_test))\n",
        "with torch.no_grad():\n",
        "    xs = xs.cuda()\n",
        "    attns = model(xs, output_attentions = True)['attentions']\n",
        "\n",
        "\n",
        "\n",
        "fig, ax = plt.subplots(1, 1, figsize=(2, 2), dpi=100)\n",
        "mark_token(ax, xs, batch=batch, i=coord_x, j=coord_y, color='tab:red')\n",
        "\n",
        "fig, ax = plt.subplots(1, 1, figsize=(1, 1), dpi=150)\n",
        "visualize_attn(ax, attns, batch=batch, i=coord_x, j=coord_y, depth=depth, head=head)\n",
        "\n",
        "save_path = f'figures/{model_option}/'\n",
        "os.makedirs(save_path, exist_ok=True, mode = 511)\n",
        "plt.savefig(save_path + 'attention.png', facecolor = 'white')"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "b7f8f456",
      "metadata": {
        "id": "b7f8f456"
      },
      "source": [
        "## Attention Distance"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "id": "7dad3f9d",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 579
        },
        "id": "7dad3f9d",
        "outputId": "32c49d26-0d97-4ea6-f1b7-39a6dcf20c80"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 645x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# build off the implementation of https://github.com/sayakpaul/probing-vits\n",
        "import numpy as np\n",
        "\n",
        "def compute_distance_matrix(patch_size, num_patches, length):\n",
        "    \"\"\"Helper function to compute distance matrix.\"\"\"\n",
        "    distance_matrix = np.zeros((num_patches, num_patches))\n",
        "    for i in range(num_patches):\n",
        "        for j in range(num_patches):\n",
        "            if i == j: # zero distance\n",
        "                continue\n",
        "\n",
        "            xi, yi = (int(i/length)), (i % length)\n",
        "            xj, yj = (int(j/length)), (j % length)\n",
        "            distance_matrix[i, j] = patch_size * np.linalg.norm([xi - xj, yi - yj])\n",
        "\n",
        "    return distance_matrix\n",
        "\n",
        "def calculate_mean_attention_dist(patch_size, attention_weights):\n",
        "    \"\"\" The attention_weights shape = (batch, num_heads, num_patches, num_patches)\"\"\"\n",
        "    \n",
        "    num_patches = attention_weights.shape[-1]\n",
        "    length = int(np.sqrt(num_patches))\n",
        "    assert (length**2 == num_patches), (\"Num patches is not perfect square\")\n",
        "\n",
        "    distance_matrix = compute_distance_matrix(patch_size, num_patches, length)\n",
        "    h, w = distance_matrix.shape\n",
        "\n",
        "    distance_matrix = distance_matrix.reshape((1, 1, h, w))\n",
        "    # The attention_weights along the last axis adds to 1\n",
        "    # this is due to the fact that they are softmax of the raw logits\n",
        "    # summation of the (attention_weights * distance_matrix)\n",
        "    # should result in an average distance per token\n",
        "    mean_distances = attention_weights * distance_matrix\n",
        "    mean_distances = np.sum(mean_distances, axis=-1) # sum along last axis to get average distance per token\n",
        "    mean_distances = np.mean(mean_distances, axis=-1) # now average across all the tokens\n",
        "\n",
        "    return torch.tensor(mean_distances)\n",
        "\n",
        "\n",
        "#@markdown Execute this cell to calculate attention distance. Attention distance is defined as the average distance between the query tokens and key tokens considering their self-attention weights. Therefore, it conceptually corresponds to the size of the receptive fields in CNNs.\n",
        "#@markdown \n",
        "#@markdown These results shows that the attention distance of CL (MoCo) is significantly higher than that of MIM (SimMIM), especially in the later layers. As seen above, the qualitative visualization, this implies that the representations of CL contain global patterns and shape information, so CL can help ViTs distinguish between objects of images. Conversely, the self-attentions of MIM mainly capture local relationships; i.e., MIM may have difficulty recognizing whole objects and their shapes.\n",
        "\n",
        "from timm.utils import AverageMeter\n",
        "\n",
        "if model_option == \"wsimmim\":\n",
        "    model = model_wsimmim\n",
        "elif model_option == \"simmim\":\n",
        "    model = model_simmim\n",
        "else:\n",
        "    raise Exception(\"Invalid option: \", model_option)\n",
        "    \n",
        "encoder_length = 12     #len(model.blocks)  # for example 12 for ViT-B\n",
        "distances = [AverageMeter() for _ in range(encoder_length)]\n",
        "\n",
        "for idx, (xs, _) in enumerate(dataset_test):\n",
        "    xs = xs.cuda()\n",
        "    \n",
        "    with torch.no_grad():\n",
        "        attns = model(xs, output_attentions = True)['attentions']\n",
        "        \n",
        "    for i, attn in enumerate([attn for attn in attns if attn is not None]):    \n",
        "        attn = attn[:,:,1:,1:]\n",
        "        attn = attn + 1e-32\n",
        "        attn = attn / attn.sum(dim=-1, keepdim=True)  # normalize\n",
        "        attn = attn.cpu().float().detach().numpy()\n",
        "\n",
        "        distance = calculate_mean_attention_dist(patch_size=16, attention_weights=attn)\n",
        "        distances[i].update(torch.mean(distance, dim=0))  # average w.r.t. batch\n",
        "        \n",
        "    if idx > -1:\n",
        "        break\n",
        "    \n",
        "\n",
        "fig, ax = plt.subplots(1, 1, figsize=(4.3, 4), dpi=150)\n",
        "distances = [torch.mean(distance.avg) for distance in distances]\n",
        "ax.plot(range(1, 13), distances, marker=\"o\")\n",
        "ax.set_xlabel(\"Depth\")\n",
        "ax.set_ylabel(\"Attention distance (px)\")\n",
        "ax.set_ylim(top=120, bottom=0)\n",
        "plt.savefig(save_path + 'attention_distance.png', facecolor = 'white')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "34e8593d",
      "metadata": {
        "id": "34e8593d"
      },
      "source": [
        "## Normalized Mutual Information"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "id": "7cfc022f",
      "metadata": {
        "cellView": "form",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 579
        },
        "id": "7cfc022f",
        "outputId": "50c84d60-6e1c-4b61-ff90-8546d8ca038a"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 645x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import torch.nn.functional as F \n",
        "from einops import rearrange, reduce, repeat\n",
        "\n",
        "def calculate_nmi(attn): \n",
        "    \"\"\" Normalized mutual information with a return type of (batch, head) \"\"\"\n",
        "    b, h, q, k = attn.shape\n",
        "    pq = torch.ones([b, h, q]).to(attn.device)\n",
        "    pq = F.softmax(pq, dim=-1)\n",
        "    pq_ext = repeat(pq, \"b h q -> b h q k\", k=k)\n",
        "    pk = reduce(attn * pq_ext, \"b h q k -> b h k\", \"sum\")\n",
        "    pk_ext = repeat(pk, \"b h k -> b h q k\", q=q)\n",
        "    \n",
        "    mi = reduce(attn * pq_ext * torch.log(attn / pk_ext), \"b h q k -> b h\", \"sum\")\n",
        "    eq = - reduce(pq * torch.log(pq), \"b h q -> b h\", \"sum\")\n",
        "    ek = - reduce(pk * torch.log(pk), \"b h k -> b h\", \"sum\")\n",
        "    \n",
        "    nmiv = mi / torch.sqrt(eq * ek)\n",
        "    \n",
        "    return nmiv\n",
        "\n",
        "#@markdown Execute this cell to calculate the Normalized Mutual Information (NMI). NMI measures the inhomogeneity of self-attentions with respect to queries. In other words, *attention collapse* occurs when the NMI approaches zero.\n",
        "#@markdown \n",
        "#@markdown The results indicate that the mutual information of CL is significantly lower than that of MIM, especially in the later layers. This suggests that the self-attentions of CL tend to collapse into homogeneous distributions.\n",
        "\n",
        "\n",
        "from timm.utils import AverageMeter\n",
        "\n",
        "if model_option == \"wsimmim\":\n",
        "    model = model_wsimmim\n",
        "elif model_option == \"simmim\":\n",
        "    model = model_simmim\n",
        "else:\n",
        "    raise Exception(\"Invalid option: \", model_option)\n",
        "\n",
        "encoder_length = 12 # len(model.blocks)  # 12 for ViT-B\n",
        "nmis = [AverageMeter() for _ in range(encoder_length)]\n",
        "\n",
        "for idx, (xs, _) in enumerate(dataset_test):\n",
        "    xs = xs.cuda()\n",
        "    \n",
        "    with torch.no_grad():\n",
        "        attns = model(xs, output_attentions = True)['attentions']\n",
        "        \n",
        "    for i, attn in enumerate([attn for attn in attns if attn is not None]):\n",
        "        attn = attn[:,:,1:,1:]  # drop cls token\n",
        "        attn = attn + 1e-8\n",
        "        attn = attn / attn.sum(dim=-1, keepdim=True)  # normalize\n",
        "        attn = attn.cpu().float()\n",
        "        \n",
        "        nmi = calculate_nmi(attn)\n",
        "        nmis[i].update(torch.mean(nmi, dim=0))  # average w.r.t. batch\n",
        "        \n",
        "    if idx > -1:\n",
        "        break\n",
        "    \n",
        "    \n",
        "fig, ax = plt.subplots(1, 1, figsize=(4.3, 4), dpi=150)\n",
        "nmis = [torch.mean(nmi.avg) for nmi in nmis]\n",
        "ax.plot(range(1, 13), nmis, marker=\"o\")\n",
        "ax.set_xlabel(\"Depth\")\n",
        "ax.set_ylabel(\"Normalized MI\")\n",
        "# ax.set_ylim(top=0.01, bottom=0.0)\n",
        "plt.savefig(save_path + 'normalized_mi.png', facecolor = 'white')\n",
        "plt.show()"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "gpuType": "T4",
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "representation_analysis",
      "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": "3.11.0"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}