0_Introducción_a_los_Transformers.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "ELIA_Transformers_1.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "kA1YMRaGbfdq"
      },
      "source": [
        "import torch\n",
        "import torch.nn.functional as F\n",
        "import torch.nn as nn\n",
        "import math"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gNXfETorg5lm"
      },
      "source": [
        "import numpy as np\n",
        "import random\n",
        "\n",
        "seed_val = 42\n",
        "\n",
        "random.seed(seed_val)\n",
        "np.random.seed(seed_val)\n",
        "torch.manual_seed(seed_val)\n",
        "torch.cuda.manual_seed_all(seed_val)"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f_VbI2UbdTj_"
      },
      "source": [
        "Creemos embeddings de prueba que reprenten dos palabras que insertaremos en el código. Hagamos que tengan las dimensiones de BERT: 768 features."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cXfuz8GIdS9U",
        "outputId": "c6e69bae-29fa-4410-8142-515a433a0183",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "embedding_palabras = torch.rand((3,768))*0.1\n",
        "embedding_palabras.shape"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 768])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DPHFe-LYfHjD"
      },
      "source": [
        "Creamos las matrices de pesos que nos darán nuestros vectores query, key y value."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OuyTiH4CfCrx"
      },
      "source": [
        "w_query = torch.rand((768, 768))*0.1\n",
        "w_key = torch.rand((768, 768))*0.1\n",
        "w_value = torch.rand((768, 768))*0.1"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fmw7ZguzgKRI"
      },
      "source": [
        "Creamos la constante de las dimensiones de nuestras matrices de proyección."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jXWdYepzhDZv"
      },
      "source": [
        "Multiplicamos nuestro embedding por las matrices de proyección para generar nuestros vectores query, key y value."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oL_CFvMOgvv8"
      },
      "source": [
        "query = torch.matmul(embedding_palabras, w_query)\n",
        "key = torch.matmul(embedding_palabras, w_key)\n",
        "value = torch.matmul(embedding_palabras, w_value)"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WncIIJWGgUwx",
        "outputId": "bd01e22d-779f-4543-9894-0347855e3f7d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "d_k = query.shape[-1]\n",
        "d_k"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "768"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3zF580THhSXz"
      },
      "source": [
        "Vemos que el tamaño de estos es (3,768)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hGpeoeD7hW3n",
        "outputId": "faca7d70-8177-48a2-b50a-464e10516215",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "query.shape"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 768])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fLdqhsea3vI3"
      },
      "source": [
        "## 1. Attention"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sJo3305K2WE_"
      },
      "source": [
        "Escribamos el código de una función attention paso a paso:\n",
        "![image.png](data:image/png;base64,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)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DzFnpe6H2lEi"
      },
      "source": [
        "La imagen arriba muestra que tenemos que ingresar a nuestra attention nuestras tres matrices de value, key, y query. Las capas Linear hacen referencia a nuestras matrices de proyección. \"h\" hace referencia al número de funciones attention que utilicemos, a esto le llamamos atención multi-headed. Lo veremos abajo."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I78ngEQhgeUg"
      },
      "source": [
        "Obtenemos los scores que resultan ser una matriz cuadrada. La podemos interpretar como una matriz que da pesos para cada palabra con respecto a las demás palabras donde en la diagonal tenemos la comparación entre mismas palabras."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xet1NO4FgfkN",
        "outputId": "d943635d-16c6-4f46-c19f-bd9c41ef3f76",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "scores = torch.matmul(query, key.transpose(-2, -1)) / math.sqrt(d_k)\n",
        "scores"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([[ 94.5383,  99.2352,  97.2765],\n",
              "        [ 99.2058, 104.1366, 102.0788],\n",
              "        [ 97.2091, 102.0398, 100.0245]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "R0MtW15ylgTw"
      },
      "source": [
        "Aplicamos la función softmax sobre la última dimensión (las columnas), en este caso hablamos de que a cada fila les vamos a aplicar la softmax. Vemos que al final los valores de todas las filas suman 1."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rzyz7srIghGo",
        "outputId": "3541ee61-cf37-4e00-d77f-ed2a9c484e66",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "p_attn = F.softmax(scores, dim = -1)\n",
        "p_attn"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([[0.0079, 0.8694, 0.1226],\n",
              "        [0.0064, 0.8811, 0.1125],\n",
              "        [0.0070, 0.8762, 0.1168]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vVgIeOhpvN59"
      },
      "source": [
        "Finalmente, multiplicamos nuestra matriz `p_attn` con nuetra matriz de `values` para crear nuestra matriz `Z` que pasará a la feed forward neural network. `Z` tiene forma (3,768), es decir, la misma que nuestro embedding que ingresamos originalmente."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZuB_6IhJvM9n",
        "outputId": "c507b67b-c343-4a34-846b-7d9f9cbdc55f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "z = torch.matmul(p_attn, value)\n",
        "z.shape"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 768])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LSXAXsfgydqw"
      },
      "source": [
        "Lo que estamos haciendo es es codificando la siguiente ecuación: \n",
        "$$\n",
        "\\text { Attention }(Q, K, V)=\\operatorname{softmax}\\left(\\frac{Q K^{T}}{\\sqrt{d_{k}}}\\right) V\n",
        "$$\n",
        "Concretemos todo en una única función llamada `attention`.\n",
        "Nota: la función de arriba es una función de atención de producto punto escalado. No sería escalado si no colocaramos el $\\sqrt{d_{k}}$. \n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UbNuvbsUbkbu"
      },
      "source": [
        "def attention(query, key, value):\n",
        "  \"Obtener 'Scaled Dot Product Attention'\"\n",
        "\n",
        "  d_k = query.shape[-1]\n",
        "\n",
        "  scores = torch.matmul(query, key.transpose(-2, -1)) / math.sqrt(d_k)\n",
        "  attn = F.softmax(scores, dim = -1)\n",
        "  z = torch.matmul(attn, value)\n",
        "\n",
        "  return z, attn"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nTVrpyeh0vT0",
        "outputId": "3e35d1ab-9068-4c25-83a3-f402f04e1c63",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "attention(query, key, value)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(tensor([[1.9599, 1.9514, 1.9623,  ..., 1.8916, 1.9233, 1.9816],\n",
              "         [1.9606, 1.9521, 1.9629,  ..., 1.8923, 1.9234, 1.9821],\n",
              "         [1.9603, 1.9518, 1.9627,  ..., 1.8920, 1.9234, 1.9819]]),\n",
              " tensor([[0.0079, 0.8694, 0.1226],\n",
              "         [0.0064, 0.8811, 0.1125],\n",
              "         [0.0070, 0.8762, 0.1168]]))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-QQeUFhT31QI"
      },
      "source": [
        "## 2. Atención multi-headed"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wHdc8jN0kM7W"
      },
      "source": [
        "\"La atención de múltiples cabezales permite que el modelo atienda de manera conjunta la información de diferentes subespacios de representación en diferentes posiciones. Con un solo cabezal de atención, el promedio lo inhibe.\" - [Harvard NLP](https://nlp.seas.harvard.edu/2018/04/03/attention.html)\n",
        "\n",
        "La traducción la hice yo pero aún así no parece español. Segunda traducción: si permitimos que diferentes cabezas (manera presumida de decir capas) de atención que comiencen su aprendizaje desde diferentes puntos (inicializamos los pesos aleatoriamente en todas las cabezas) entonces cada una llegará a diferentes conclusiones de las otras. Es como juntar ocho personas con experiencias diferentes de la vida piensen acerca de un mismo problema, cada quien desde su perspectiva, y al final juntamos las soluciones brindadas. Te aseguro que este metodo será mucho mejor a que si una sola persona piensa la solución al problema. Ese es el concepto de la atención multi-headed.\n",
        "\n",
        "Ahora, la ecuación es:\n",
        "\n",
        "\\begin{array}{l}\n",
        "\\text { MultiHead }(Q, K, V)=\\text { Concat }\\left(\\text { head }_{1}, \\ldots, \\text { head }_{\\mathrm{h}}\\right) W^{O} \\\\\n",
        "\\text { donde head }_{\\mathrm{i}}=\\text { Attention }\\left(Q W_{i}^{Q}, K W_{i}^{K}, V W_{i}^{V}\\right)\n",
        "\\end{array}\n",
        "\n",
        "La ecuación nos indica que para crear una atención multi-headed necesitamos ingresar nuestras matrices Query, Key y Value; estas entrarán a cada una de nuestas cabezas attention multiplicando las matrices de proyección $W$. Como vimos arriba, cada attention nos va a regresar un tensor que tendrá las dimensiones del embedding original que ingresamos (768 arriba). En este caso no será será así. Ahora, cada attention nos regresará un tensor de dimensión $768 / heads$. En bert $heads = 12$ entonces tenemos que cada attention regresará un tensor de dimensión 64. \n",
        "\n",
        "¿Por qué menos dimensiones? Porque vamos a concatener (pegar) los tensores lado a lado de forma que tendremos las $heads * 64 = 768$ dimensiones del embedding original. Teóricamente, como estamos calculando tensores más pequeños, el costo computacional debe ser similar al de calcular una sola cabeza más grande.\n",
        "\n",
        "Notar que estamos multiplicando nuestra concatenación por $W^{O}$. Estamos asignando pesos para que indicarán la mejor forma de acomodar la concatenación."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jaS6t8MhyYRN"
      },
      "source": [
        "Vayamos escribiendo paso a paso esta ecuación en código de pytorch.\n",
        "\n",
        "Primero notar que nuestras matrices Query, Key y Values tienen forma `(n_palabras, d_modelo)` donde `n_palabras` son el número de palabras que estamos ingresando y `d_modelo` es el tamaño de la dimensión de nuestro emebdding entrante (y, por lo tanto, saliente)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Pnv5lst-0ZOB"
      },
      "source": [
        "d_modelo =  768\n",
        "n_palabras = query.shape[0]"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mPYyaO1V2Plu"
      },
      "source": [
        "Ahora creamos una función `nn.Linear` que es equivalente a $y = xA^T + b$ donde X es una matriz de proyección. En otras palabras, equivale a lo que escribimos arriba manualmente pero ahora hecho con operaciones en la libreria `nn` de pytorch.\n",
        "\n",
        "Veremos que el resultado serán tensores de forma (3, 768) porque tenemos tres palabras y un embedding de 768 dimensiones."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JzeIFT9G09f3",
        "outputId": "8e6c5e77-e628-41d6-b2ce-a988954002aa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "linear_q = nn.Linear(in_features=d_modelo, out_features=d_modelo)\n",
        "linear_k = nn.Linear(in_features=d_modelo, out_features=d_modelo)\n",
        "linear_v = nn.Linear(in_features=d_modelo, out_features=d_modelo)\n",
        "\n",
        "linear_q(embedding_palabras).shape"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 768])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yxw427Iz3gPF"
      },
      "source": [
        "query = linear_q(embedding_palabras)\n",
        "key = linear_k(embedding_palabras)\n",
        "value = linear_v(embedding_palabras)"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "El8zcGlo5c8R"
      },
      "source": [
        "Hasta este punto todas las dimensones siguen iguales que con una sola cabeza. Ahora comienza el proceso de crear las múltiples cabezas. \n",
        "\n",
        "1. Vamos a cambiar la forma de nuestras matrices a `(n_palabras, cabezas, d_modelo / cabezas)`. En este caso el resultado es `(3, 12, 64)`. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qZuDgZqw4RlE"
      },
      "source": [
        "query = query.view(-1, 12, 64)\n",
        "key = key.view(-1, 12, 64)\n",
        "value = value.view(-1, 12, 64)"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JXsrJDDg6Si1",
        "outputId": "7badea1d-5c7e-4146-b482-9b2165b5eedb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "query.shape"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 12, 64])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4Iu_Vavs6ojG"
      },
      "source": [
        "Lo que hicimos fue crear para cada palabra una matriz de (12, 64). Esto equivale a tener por cada palabra 12 representaciones diferentes (las cabezass) para 64 valores de un embedding. En otras palabras, ¡tenemos 12 embeddings de 64 dimensiones por palabra! \n",
        "2. Ahora vamos a transponer estas matrices para cada una de nuestras palabras, es lo mismo que hicimos con una sola cabeza."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TQjhPdqX6iK9"
      },
      "source": [
        "query = query.transpose(1,2)\n",
        "key = key.transpose(1,2)\n",
        "value = value.transpose(1,2)"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6R346oUe7M2d",
        "outputId": "6b71d8aa-88fd-4dc2-f766-2e711d40b568",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "query.shape"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 64, 12])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YHcr1uLu7TwH"
      },
      "source": [
        "3. Apliquemos nuestra función de atención que creamos hace rato. Nos va a retornar nuestra `z` que tendrá las mismas dimensiones de lo que ingresamos y nuestro valor `attn`."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IqThxNio7Z_4"
      },
      "source": [
        "z, attn = attention(query, key, value)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fziDFT_L71ME",
        "outputId": "8bbee7da-300a-4410-cdbf-ea732c39f4ff",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "z.shape, attn.shape"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(torch.Size([3, 64, 12]), torch.Size([3, 64, 64]))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8wS1AevR-P1z"
      },
      "source": [
        "4. Vamos a concatenar nuestros resultados para obtener un embedding del tamaño original: (3, 768). Primero transponemos a la forma original. Por cuestiones de memoria, es buena práctica usar la función `contiguous()` después de transponer.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4bGOw3yQ9-WT"
      },
      "source": [
        "z = z.transpose(1,2).contiguous()"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HI8iryWK-luN",
        "outputId": "dfad908a-6cbb-456c-a89d-0d6d12f2d3fe",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "z.shape"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 12, 64])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2eMYog57_Elw"
      },
      "source": [
        "Ahora usamos view para regresar a la forma original y listo."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IQnYra6R-m7_"
      },
      "source": [
        "z = z.view(n_palabras, -1)"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uRLAdO2a_B5m",
        "outputId": "96da9807-39fa-4d92-b5eb-a89738ae7ee0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "z.shape"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 768])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L_kPWcG__OTG"
      },
      "source": [
        "Casi acabamos...\n",
        "5. Ahora solo multiplicamos el acodo resultante por la matriz $W^{O}$ usando de nuevo una función `nn.Linear`."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BZ1GE1so_lP_"
      },
      "source": [
        "linear_w0 = nn.Linear(768, 768)"
      ],
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4NL38zu6_w-v"
      },
      "source": [
        "z = linear_w0(z)"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FYijk3CV_01D",
        "outputId": "e074ed9b-1ae9-4458-ad0e-a997032ef8b1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "z.shape"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "torch.Size([3, 768])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e6ySN_LT3RvS"
      },
      "source": [
        "class MultiHeadedAttention(nn.Module):\n",
        "  def __init__(self, h, d_modelo):\n",
        "    \"Se reciben el número de cabezas (h) y las dimensiones de cada cabeza (d)\"\n",
        "    super(MultiHeadedAttention, self).__init__()\n",
        "    self.h = h\n",
        "    self.d_modelo = d_modelo\n",
        "    self.linear_w0 = nn.Linear(d_modelo, d_modelo)\n",
        "\n",
        "  def forward(self, query, key, value):\n",
        "    \"Definimos el grafo a seguir.\"\n",
        "    query = query.view(-1, self.h, int(self.d_modelo / self.h)).transpose(1,2)\n",
        "    key = key.view(-1, self.h, int(self.d_modelo / self.h)).transpose(1,2)\n",
        "    value = value.view(-1, self.h, int(self.d_modelo / self.h)).transpose(1,2)\n",
        "\n",
        "    z, _ = attention(query, key, value)\n",
        "\n",
        "    z = z.transpose(1,2).contiguous().view(n_palabras, -1)\n",
        "\n",
        "    z = linear_w0(z)\n",
        "\n",
        "    return z"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TkBkp8XbeVkt"
      },
      "source": [
        "Para probar nuestro `Module` creamos de nuevo nuestras matrices Query, Key y Value."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YJmBPAXseOUw"
      },
      "source": [
        "query = linear_q(embedding_palabras)\n",
        "key = linear_k(embedding_palabras)\n",
        "value = linear_v(embedding_palabras)"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7TGdgpfKekAN"
      },
      "source": [
        "multiHeadedAttention = MultiHeadedAttention(h = 12, d_modelo = d_modelo)"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ijV9rb9cfMnm",
        "outputId": "99a6d0fb-ce75-4a9f-d443-198637f8aaaa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "multiHeadedAttention"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "MultiHeadedAttention(\n",
              "  (linear_w0): Linear(in_features=768, out_features=768, bias=True)\n",
              ")"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ruOhbt5ngt5v"
      },
      "source": [
        "Obtendremos un tensor con el mismo resultado que obtuvimos creando la multi-headed attention paso a paso."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "suBMbPudhXix"
      },
      "source": [
        "z_con_modulo = multiHeadedAttention(query, key, value) "
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j4VhIOH9frLp"
      },
      "source": [
        "assert z_con_modulo.detach().numpy().all() == z.detach().numpy().all()"
      ],
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f7ZIOKunjsP3"
      },
      "source": [
        "Hay algo muy interesante y confuso. Si corremos nuestra instancia `MultiHeadedAttention` que creamos, vamos a ver que únicamene estamos agregando la matriz de pesos `W_0`, nada más eso. Los pesos de las matrices de proyección para Query, Key y Value siguen siendo las mismas. Mismo número de pesos. Por esta razón creo que la magia de los multi-headed attentions está en la matriz que decide la mejor forma de acomoda nuestro embedding resultante, `W_0`. \n",
        "\n",
        "Es también por esta misma razón que podemos esperar que una multi-headed attention no sea mucho más pesado computacionalmente que una attention de una sola cabeza."
      ]
    }
  ]
}