{ "cells": [ { "cell_type": "markdown", "source": [ "Self-Attention\r\n", "======================\r\n", "\r\n", "In this notebook, we look at how self-attention works. In other words, no RNN, but simply self attention of source sequence for the encoder, attention (both self and cross) of the **shifted target sequence** for the decoder. \r\n", "\r\n", "![Self Attention](../images/encdec_self_simplified.png)" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "## Download Untility Files for Plotting and Data Generation\r\n", "\r\n", "Download these utility functions first and place them in the same directory as this notebook. These files are the same as the ones in [Lab11 Sequence to Sequence Model](https://weiliu2k.github.io/CITS4012/LSTM/seq2seq.html)." ], "metadata": {} }, { "cell_type": "code", "execution_count": 2, "source": [ "from IPython.display import FileLink, FileLinks" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 3, "source": [ "FileLink('plots.py')" ], "outputs": [ { "output_type": "execute_result", "data": { "text/html": [ "plots.py
" ], "text/plain": [ "c:\\Users\\wei\\jupyter_book\\cits4012\\cits4012_natural_language_processing\\cits4012_natural_language_processing\\attention\\plots.py" ] }, "metadata": {}, "execution_count": 3 } ], "metadata": {} }, { "cell_type": "code", "execution_count": 4, "source": [ "FileLink('plots_seq2seq.py')" ], "outputs": [ { "output_type": "execute_result", "data": { "text/html": [ "plots_seq2seq.py
" ], "text/plain": [ "c:\\Users\\wei\\jupyter_book\\cits4012\\cits4012_natural_language_processing\\cits4012_natural_language_processing\\attention\\plots_seq2seq.py" ] }, "metadata": {}, "execution_count": 4 } ], "metadata": {} }, { "cell_type": "code", "execution_count": 5, "source": [ "FileLink('util.py')" ], "outputs": [ { "output_type": "execute_result", "data": { "text/html": [ "util.py
" ], "text/plain": [ "c:\\Users\\wei\\jupyter_book\\cits4012\\cits4012_natural_language_processing\\cits4012_natural_language_processing\\attention\\util.py" ] }, "metadata": {}, "execution_count": 5 } ], "metadata": {} }, { "cell_type": "code", "execution_count": 6, "source": [ "FileLink('replay.py')" ], "outputs": [ { "output_type": "execute_result", "data": { "text/html": [ "replay.py
" ], "text/plain": [ "c:\\Users\\wei\\jupyter_book\\cits4012\\cits4012_natural_language_processing\\cits4012_natural_language_processing\\attention\\replay.py" ] }, "metadata": {}, "execution_count": 6 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "## Imports" ], "metadata": {} }, { "cell_type": "code", "execution_count": 7, "source": [ "import copy\r\n", "import numpy as np\r\n", "\r\n", "import torch\r\n", "import torch.optim as optim\r\n", "import torch.nn as nn\r\n", "import torch.nn.functional as F\r\n", "from torch.utils.data import DataLoader, Dataset, random_split, TensorDataset\r\n", "from util import StepByStep\r\n", "from plots import *\r\n", "from plots_seq2seq import *" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "## Data Generation" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "We still make use of the Square Sequences, using the first two corners to predict the last two. Same method as before for generating noisy squares. " ], "metadata": {} }, { "cell_type": "code", "execution_count": 8, "source": [ "def generate_sequences(n=128, variable_len=False, seed=13):\r\n", " basic_corners = np.array([[-1, -1], [-1, 1], [1, 1], [1, -1]])\r\n", " np.random.seed(seed)\r\n", " bases = np.random.randint(4, size=n)\r\n", " if variable_len:\r\n", " lengths = np.random.randint(3, size=n) + 2\r\n", " else:\r\n", " lengths = [4] * n\r\n", " directions = np.random.randint(2, size=n)\r\n", " points = [basic_corners[[(b + i) % 4 for i in range(4)]][slice(None, None, d*2-1)][:l] + np.random.randn(l, 2) * 0.1 for b, d, l in zip(bases, directions, lengths)]\r\n", " return points, directions" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "## Multi-Headed Attention" ], "metadata": {} }, { "cell_type": "code", "execution_count": 9, "source": [ "class Attention(nn.Module):\r\n", " def __init__(self, hidden_dim, input_dim=None, proj_values=False):\r\n", " super().__init__()\r\n", " self.d_k = hidden_dim\r\n", " self.input_dim = hidden_dim if input_dim is None else input_dim\r\n", " self.proj_values = proj_values\r\n", " # Affine transformations for Q, K, and V\r\n", " self.linear_query = nn.Linear(self.input_dim, hidden_dim)\r\n", " self.linear_key = nn.Linear(self.input_dim, hidden_dim)\r\n", " self.linear_value = nn.Linear(self.input_dim, hidden_dim)\r\n", " self.alphas = None\r\n", " \r\n", " def init_keys(self, keys):\r\n", " self.keys = keys\r\n", " self.proj_keys = self.linear_key(self.keys)\r\n", " self.values = self.linear_value(self.keys) \\\r\n", " if self.proj_values else self.keys\r\n", " \r\n", " def score_function(self, query):\r\n", " proj_query = self.linear_query(query)\r\n", " # scaled dot product\r\n", " # N, 1, H x N, H, L -> N, 1, L\r\n", " dot_products = torch.bmm(proj_query, self.proj_keys.permute(0, 2, 1))\r\n", " scores = dot_products / np.sqrt(self.d_k)\r\n", " return scores\r\n", " \r\n", " def forward(self, query, mask=None):\r\n", " # Query is batch-first N, 1, H\r\n", " scores = self.score_function(query) # N, 1, L\r\n", " if mask is not None:\r\n", " scores = scores.masked_fill(mask == 0, -1e9)\r\n", " alphas = F.softmax(scores, dim=-1) # N, 1, L\r\n", " self.alphas = alphas.detach()\r\n", " \r\n", " # N, 1, L x N, L, H -> N, 1, H\r\n", " context = torch.bmm(alphas, self.values)\r\n", " return context" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 10, "source": [ "class MultiHeadAttention(nn.Module):\r\n", " def __init__(self, n_heads, d_model, input_dim=None, proj_values=True):\r\n", " super().__init__()\r\n", " self.linear_out = nn.Linear(n_heads * d_model, d_model)\r\n", " self.attn_heads = nn.ModuleList([Attention(d_model, \r\n", " input_dim=input_dim, \r\n", " proj_values=proj_values) \r\n", " for _ in range(n_heads)])\r\n", " \r\n", " def init_keys(self, key):\r\n", " for attn in self.attn_heads:\r\n", " attn.init_keys(key)\r\n", " \r\n", " @property\r\n", " def alphas(self):\r\n", " # Shape: n_heads, N, 1, L (source)\r\n", " return torch.stack([attn.alphas for attn in self.attn_heads], dim=0)\r\n", " \r\n", " def output_function(self, contexts):\r\n", " # N, 1, n_heads * D\r\n", " concatenated = torch.cat(contexts, axis=-1)\r\n", " # Linear transf. to go back to original dimension\r\n", " out = self.linear_out(concatenated) # N, 1, D\r\n", " return out\r\n", " \r\n", " def forward(self, query, mask=None):\r\n", " contexts = [attn(query, mask=mask) for attn in self.attn_heads]\r\n", " out = self.output_function(contexts)\r\n", " return out" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "## Self-Attention\r\n", "\r\n", "Instead of using RNN to generate the hidden states, we use attention mechanism to generate hidden states for inputs. This is called self-attention, i.e. attention on the input data points themselves. Each input will have its own context vector. All context vectors are concatenated and transformed through a linear layer, as an input to the feedforward network as hidden state for the corresponding input. " ], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Attention Scores from Input\r\n", "\r\n", "Inputs or their affine transformations are used as Keys, Values and Queries. " ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "\\alpha_{\\color{blue}{0}\\color{red}0}, \\alpha_{{\\color{blue}{0}\\color{red}1}} = softmax(\\frac{\\color{red}{Q_0}\\color{black}\\cdot K_0}{\\sqrt{2}}, \\frac{\\color{red}{Q_0}\\color{black}\\cdot K_1}{\\sqrt{2}})\r\n", "\\\\\r\n", "\\color{blue}{context\\ vector_0}\\color{black}= \\alpha_{\\color{blue}{0}\\color{red}0}V_0 + \\alpha_{{\\color{blue}{0}\\color{red}1}}V_1\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "\\alpha_{\\color{blue}{1}\\color{red}0}, \\alpha_{{\\color{blue}{1}\\color{red}1}} = softmax(\\frac{\\color{red}{Q_1}\\color{black}\\cdot K_0}{\\sqrt{2}}, \\frac{\\color{red}{Q_1}\\color{black}\\cdot K_1}{\\sqrt{2}})\r\n", "\\\\\r\n", "\\color{blue}{context\\ vector_1}\\color{black}= \\alpha_{\\color{blue}{1}\\color{red}0}V_0 + \\alpha_{{\\color{blue}{1}\\color{red}1}}V_1\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "\\begin{array}{c|cc}\r\n", "& source\\\\\r\n", "target& \\color{red}{x_0} & \\color{red}{x_1} \\\\\r\n", "\\hline\r\n", " \\color{blue}{h_0} & \\alpha_{\\color{blue}{0}\\color{red}0} & \\alpha_{{\\color{blue}{0}\\color{red}1}} \\\\\r\n", " \\color{blue}{h_1} & \\alpha_{\\color{blue}{1}\\color{red}0} & \\alpha_{{\\color{blue}{1}\\color{red}1}}\r\n", "\\end{array}\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "![](../images/transf_encself.png)" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Encoder with Self-Attention\r\n", "\r\n", "As you can see from the code below, each input has its own attention head, so it makes use of the multi-headed attention above. Check the code against the diagram for better understanding." ], "metadata": {} }, { "cell_type": "code", "execution_count": 11, "source": [ "class EncoderSelfAttn(nn.Module):\r\n", " def __init__(self, n_heads, d_model, ff_units, n_features=None):\r\n", " super().__init__()\r\n", " self.n_heads = n_heads\r\n", " self.d_model = d_model\r\n", " self.ff_units = ff_units\r\n", " self.n_features = n_features\r\n", " self.self_attn_heads = MultiHeadAttention(n_heads, d_model, input_dim=n_features)\r\n", " self.ffn = nn.Sequential(\r\n", " nn.Linear(d_model, ff_units),\r\n", " nn.ReLU(),\r\n", " nn.Linear(ff_units, d_model),\r\n", " )\r\n", " \r\n", " def forward(self, query, mask=None):\r\n", " self.self_attn_heads.init_keys(query)\r\n", " att = self.self_attn_heads(query, mask)\r\n", " out = self.ffn(att)\r\n", " return out" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Encoder the Square Sequence\r\n", "\r\n", "The perfect sequare with four corners, is split into a source sequence of two corners and a target sequence of two corners." ], "metadata": {} }, { "cell_type": "code", "execution_count": 12, "source": [ "full_seq = torch.tensor([[-1, -1], [-1, 1], [1, 1], [1, -1]]).float().view(1, 4, 2)\r\n", "source_seq = full_seq[:, :2]\r\n", "target_seq = full_seq[:, 2:]" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 13, "source": [ "torch.manual_seed(11)\r\n", "encself = EncoderSelfAttn(n_heads=3, d_model=2, ff_units=10, n_features=2)\r\n", "query = source_seq\r\n", "encoder_states = encself(query)\r\n", "encoder_states" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[-0.0498, 0.2193],\n", " [-0.0642, 0.2258]]], grad_fn=)" ] }, "metadata": {}, "execution_count": 13 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Decoder" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "This decoder has both self attention and cross-attention, which uses the decoder hidden state as query. " ], "metadata": {} }, { "cell_type": "code", "execution_count": 14, "source": [ "class DecoderSelfAttn(nn.Module):\r\n", " def __init__(self, n_heads, d_model, ff_units, n_features=None):\r\n", " super().__init__()\r\n", " self.n_heads = n_heads\r\n", " self.d_model = d_model\r\n", " self.ff_units = ff_units\r\n", " self.n_features = d_model if n_features is None else n_features\r\n", " self.self_attn_heads = MultiHeadAttention(n_heads, d_model, input_dim=self.n_features)\r\n", " self.cross_attn_heads = MultiHeadAttention(n_heads, d_model)\r\n", " self.ffn = nn.Sequential(\r\n", " nn.Linear(d_model, ff_units),\r\n", " nn.ReLU(),\r\n", " nn.Linear(ff_units, self.n_features),\r\n", " )\r\n", " \r\n", " def init_keys(self, states):\r\n", " self.cross_attn_heads.init_keys(states)\r\n", " \r\n", " def forward(self, query, source_mask=None, target_mask=None):\r\n", " self.self_attn_heads.init_keys(query)\r\n", " att1 = self.self_attn_heads(query, target_mask)\r\n", " att2 = self.cross_attn_heads(att1, source_mask)\r\n", " out = self.ffn(att2)\r\n", " return out" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "#### Subsequent Inputs and Teacher Forcing" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "The **shifted target sequence** containes the last element of the source sequence, and everything but the last element of the target sequence. " ], "metadata": {} }, { "cell_type": "code", "execution_count": 15, "source": [ "shifted_seq = torch.cat([source_seq[:, -1:], target_seq[:, :-1]], dim=1)" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "#### Attention Scores\r\n", "\r\n", "Because we feed the entire shifted target sequence to the decoder, we need to be careful not to allow the model to peek into the future. \r\n", "\r\n", "The attention score $\\alpha_{22}$ is problematic, because we are using the $K_2$ and $V_2$ which the model should not have seen yet for calculation. So we need **target mask** to ensure that it is not computed. " ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "\\alpha_{\\color{green}{2}\\color{red}1}, \\alpha_{{\\color{green}{2}\\color{red}2}} = softmax(\\frac{\\color{red}{Q_1}\\color{black}\\cdot K_1}{\\sqrt{2}}, \\frac{\\color{red}{Q_1}\\color{black}\\cdot K_2}{\\sqrt{2}})\r\n", "\\\\\r\n", "\\color{green}{context\\ vector_2}\\color{black}= \\alpha_{\\color{green}{2}\\color{red}1}V_1 + \\alpha_{{\\color{green}{2}\\color{red}2}}V_2\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "Below, it is fine for us to use $K_2$ and $V_2$ to caluate $\\alpha_3$." ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "\\alpha_{\\color{green}{3}\\color{red}1}, \\alpha_{{\\color{green}{3}\\color{red}2}} = softmax(\\frac{\\color{red}{Q_2}\\color{black}\\cdot K_1}{\\sqrt{2}}, \\frac{\\color{red}{Q_2}\\color{black}\\cdot K_2}{\\sqrt{2}})\r\n", "\\\\\r\n", "\\color{green}{context\\ vector_3}\\color{black}= \\alpha_{\\color{green}{3}\\color{red}1}V_1 + \\alpha_{{\\color{green}{3}\\color{red}2}}V_2\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "Any values in the attention score matrix that are above the diagnal line needs to be masked. Here we need to mask $\\alpha_{22}$ to 0.\r\n", "$$\r\n", "\\begin{array}{c|cc}\r\n", "& source\\\\\r\n", "target& \\color{red}{x_1} & \\color{red}{x_2} \\\\\r\n", "\\hline\r\n", " \\color{green}{h_2} & \\alpha_{\\color{green}{2}\\color{red}1} & \\alpha_{{\\color{green}{2}\\color{red}2}} \\\\\r\n", " \\color{green}{h_3} & \\alpha_{\\color{green}{3}\\color{red}1} & \\alpha_{{\\color{green}{3}\\color{red}2}}\r\n", "\\end{array}\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "#### Target Mask (Training)" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "The masked scoreing matrix should look like this:" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "\\begin{array}{c|cc}\r\n", "& source\\\\\r\n", "target& \\color{red}{x_1} & \\color{red}{x_2} \\\\\r\n", "\\hline\r\n", " \\color{green}{h_2} & \\alpha_{\\color{green}{2}\\color{red}1} & 0 \\\\\r\n", " \\color{green}{h_3} & \\alpha_{\\color{green}{3}\\color{red}1} & \\alpha_{{\\color{green}{3}\\color{red}2}}\r\n", "\\end{array}\r\n", "$$" ], "metadata": {} }, { "cell_type": "code", "execution_count": 16, "source": [ "def subsequent_mask(size):\r\n", " attn_shape = (1, size, size)\r\n", " subsequent_mask = (1 - torch.triu(torch.ones(attn_shape), diagonal=1)).bool()\r\n", " return subsequent_mask" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 17, "source": [ "subsequent_mask(2) # 1, L, L" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[ True, False],\n", " [ True, True]]])" ] }, "metadata": {}, "execution_count": 17 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ ":::{important} Subsequent Mask\r\n", "We must use this mask while querying the decoder to prevent it\r\n", "from cheating. You can choose to use an additional mask to\r\n", "\"hide\" more data from the decoder if you wish, but the\r\n", "subsequent mask is a strong requirement of the self-attention\r\n", "decoder.\r\n", ":::" ], "metadata": {} }, { "cell_type": "code", "execution_count": 18, "source": [ "torch.manual_seed(13)\r\n", "decself = DecoderSelfAttn(n_heads=3, d_model=2, ff_units=10, n_features=2)\r\n", "decself.init_keys(encoder_states)\r\n", "\r\n", "query = shifted_seq\r\n", "out = decself(query, target_mask=subsequent_mask(2))\r\n", "\r\n", "decself.self_attn_heads.alphas" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[[1.0000, 0.0000],\n", " [0.4011, 0.5989]]],\n", "\n", "\n", " [[[1.0000, 0.0000],\n", " [0.4264, 0.5736]]],\n", "\n", "\n", " [[[1.0000, 0.0000],\n", " [0.6304, 0.3696]]]])" ] }, "metadata": {}, "execution_count": 18 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "#### Target Mask (Evaluation/Prediction)" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "The only difference between training and evaluation, concerning the target mask,\r\n", "is that we'll be using larger masks as we go. The very first mask is actually trivial\r\n", "since there are no elements above the diagonal:" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "1^{st}\\ Step\r\n", "\\begin{cases}\r\n", "\\begin{array}{c|cc}\r\n", "target& source\\\\\r\n", "& \\color{red}{x_1} & \\\\\r\n", "\\hline\r\n", " \\color{green}{h_2} & \\alpha_{\\color{green}{2}\\color{red}1}\r\n", "\\end{array}\r\n", "\\end{cases}\r\n", "$$" ], "metadata": {} }, { "cell_type": "markdown", "source": [ "In evaluation/prediction time we only have the source sequence and, in our\r\n", "example, we use its last element as input for the decoder:" ], "metadata": {} }, { "cell_type": "code", "execution_count": 19, "source": [ "inputs = source_seq[:, -1:]\r\n", "trg_masks = subsequent_mask(1)\r\n", "out = decself(inputs, trg_masks)\r\n", "out" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[0.4132, 0.3728]]], grad_fn=)" ] }, "metadata": {}, "execution_count": 19 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "The mask is not actually masking anything in this case, and we get a prediction for\r\n", "the coordinates of $x_2$ as expected. In RNN based seq2seq models, this prediction would be used directly as the next input, but the self-attention decoder expects the full sequence as \"query\", so we concatenate the prediction to the previous \"query\"." ], "metadata": {} }, { "cell_type": "code", "execution_count": 20, "source": [ "inputs = torch.cat([inputs, out[:, -1:, :]], dim=-2)\r\n", "inputs" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[-1.0000, 1.0000],\n", " [ 0.4132, 0.3728]]], grad_fn=)" ] }, "metadata": {}, "execution_count": 20 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "Now there are two data points for querying the decoder, so we adjust the mask\r\n", "accordingly." ], "metadata": {} }, { "cell_type": "markdown", "source": [ "$$\r\n", "2^{nd}\\ Step\r\n", "\\begin{cases}\r\n", "\\begin{array}{c|cc}\r\n", "target& source\\\\\r\n", "& \\color{red}{x_1} & \\color{green}{x_2} \\\\\r\n", "\\hline\r\n", " \\color{green}{h_2} & \\alpha_{\\color{green}{2}\\color{red}1} & 0 \\\\\r\n", " \\color{green}{h_3} & \\alpha_{\\color{green}{3}\\color{red}1} & \\alpha_{{\\color{green}{3}\\color{green}2}}\r\n", "\\end{array}\r\n", "\\end{cases}\r\n", "$$" ], "metadata": {} }, { "cell_type": "code", "execution_count": 21, "source": [ "trg_masks = subsequent_mask(2)\r\n", "out = decself(inputs, trg_masks)\r\n", "out" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[0.4137, 0.3727],\n", " [0.4132, 0.3728]]], grad_fn=)" ] }, "metadata": {}, "execution_count": 21 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ ":::{important}\r\n", "The mask guarantees that the predicted $x_2$ (in the first step)\r\n", "won't change the predicted $x_2$ (in the second step) because\r\n", "predictions are made based on past data points only\r\n", ":::\r\n", "\r\n", "The last prediction is, once again, concatenated to the previous \"query\"." ], "metadata": {} }, { "cell_type": "code", "execution_count": 22, "source": [ "inputs = torch.cat([inputs, out[:, -1:, :]], dim=-2)\r\n", "inputs" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[-1.0000, 1.0000],\n", " [ 0.4132, 0.3728],\n", " [ 0.4132, 0.3728]]], grad_fn=)" ] }, "metadata": {}, "execution_count": 22 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "But, since we're actually done with the predictions (the desired target sequence has\r\n", "a length of two), we simply exclude the first data point in the query (the one coming\r\n", "from the source sequence) and that's the predicted target sequence" ], "metadata": {} }, { "cell_type": "code", "execution_count": 23, "source": [ "inputs[:, 1:]" ], "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "tensor([[[0.4132, 0.3728],\n", " [0.4132, 0.3728]]], grad_fn=)" ] }, "metadata": {}, "execution_count": 23 } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Encoder + Decoder + Self-Attention" ], "metadata": {} }, { "cell_type": "code", "execution_count": 24, "source": [ "class EncoderDecoderSelfAttn(nn.Module):\r\n", " def __init__(self, encoder, decoder, input_len, target_len):\r\n", " super().__init__()\r\n", " self.encoder = encoder\r\n", " self.decoder = decoder\r\n", " self.input_len = input_len\r\n", " self.target_len = target_len\r\n", " self.trg_masks = self.subsequent_mask(self.target_len)\r\n", " \r\n", " @staticmethod\r\n", " def subsequent_mask(size):\r\n", " attn_shape = (1, size, size)\r\n", " subsequent_mask = (1 - torch.triu(torch.ones(attn_shape), diagonal=1))\r\n", " return subsequent_mask\r\n", " \r\n", " def encode(self, source_seq, source_mask):\r\n", " # Encodes the source sequence and uses the result\r\n", " # to initialize the decoder\r\n", " encoder_states = self.encoder(source_seq, source_mask)\r\n", " self.decoder.init_keys(encoder_states)\r\n", " \r\n", " def decode(self, shifted_target_seq, source_mask=None, target_mask=None):\r\n", " # Decodes/generates a sequence using the shifted (masked)\r\n", " # target sequence - used in TRAIN mode\r\n", " outputs = self.decoder(shifted_target_seq, \r\n", " source_mask=source_mask,\r\n", " target_mask=target_mask)\r\n", " return outputs\r\n", " \r\n", " def predict(self, source_seq, source_mask):\r\n", " # Decodes/generates a sequence using one input\r\n", " # at a time - used in EVAL mode\r\n", " inputs = source_seq[:, -1:]\r\n", " for i in range(self.target_len):\r\n", " out = self.decode(inputs, source_mask, self.trg_masks[:, :i+1, :i+1])\r\n", " out = torch.cat([inputs, out[:, -1:, :]], dim=-2)\r\n", " inputs = out.detach()\r\n", " outputs = inputs[:, 1:, :]\r\n", " return outputs\r\n", " \r\n", " def forward(self, X, source_mask=None):\r\n", " # Sends the mask to the same device as the inputs\r\n", " self.trg_masks = self.trg_masks.type_as(X).bool()\r\n", " # Slices the input to get source sequence\r\n", " source_seq = X[:, :self.input_len, :]\r\n", " # Encodes source sequence AND initializes decoder\r\n", " self.encode(source_seq, source_mask)\r\n", " if self.training:\r\n", " # Slices the input to get the shifted target seq\r\n", " shifted_target_seq = X[:, self.input_len-1:-1, :]\r\n", " # Decodes using the mask to prevent cheating\r\n", " outputs = self.decode(shifted_target_seq, source_mask, self.trg_masks)\r\n", " else:\r\n", " # Decodes using its own predictions\r\n", " outputs = self.predict(source_seq, source_mask)\r\n", " \r\n", " return outputs" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Data Preparation" ], "metadata": {} }, { "cell_type": "code", "execution_count": 25, "source": [ "points, directions = generate_sequences()\r\n", "full_train = torch.as_tensor(points).float()\r\n", "target_train = full_train[:, 2:]" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 26, "source": [ "test_points, test_directions = generate_sequences(seed=19)\r\n", "full_test = torch.as_tensor(points).float()\r\n", "source_test = full_test[:, :2]\r\n", "target_test = full_test[:, 2:]" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 27, "source": [ "train_data = TensorDataset(full_train, target_train)\r\n", "test_data = TensorDataset(source_test, target_test)\r\n", "\r\n", "generator = torch.Generator()\r\n", "train_loader = DataLoader(train_data, batch_size=16, shuffle=True, generator=generator)\r\n", "test_loader = DataLoader(test_data, batch_size=16)" ], "outputs": [], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Model Configuration & Training" ], "metadata": {} }, { "cell_type": "code", "execution_count": 28, "source": [ "torch.manual_seed(23)\r\n", "encself = EncoderSelfAttn(n_heads=3, d_model=2, ff_units=10, n_features=2)\r\n", "decself = DecoderSelfAttn(n_heads=3, d_model=2, ff_units=10, n_features=2)\r\n", "model = EncoderDecoderSelfAttn(encself, decself, input_len=2, target_len=2)\r\n", "loss = nn.MSELoss()\r\n", "optimizer = optim.Adam(model.parameters(), lr=0.01)" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 29, "source": [ "sbs_seq_selfattn = StepByStep(model, loss, optimizer)\r\n", "sbs_seq_selfattn.set_loaders(train_loader, test_loader)\r\n", "sbs_seq_selfattn.train(100)" ], "outputs": [], "metadata": {} }, { "cell_type": "code", "execution_count": 30, "source": [ "fig = sbs_seq_selfattn.plot_losses()" ], "outputs": [ { "output_type": "display_data", "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {} } ], "metadata": {} }, { "cell_type": "markdown", "source": [ "### Visualizing Predictions" ], "metadata": {} }, { "cell_type": "code", "execution_count": 31, "source": [ "fig = sequence_pred(sbs_seq_selfattn, full_test, test_directions)" ], "outputs": [ { "output_type": "display_data", "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {} } ], "metadata": { "scrolled": false } } ], "metadata": { "orig_nbformat": 4, "language_info": { "name": "python", "version": "3.8.10", "mimetype": "text/x-python", "codemirror_mode": { "name": "ipython", "version": 3 }, "pygments_lexer": "ipython3", "nbconvert_exporter": "python", "file_extension": ".py" }, "kernelspec": { "name": "python3", "display_name": "Python 3.8.10 64-bit ('cits4012': conda)" }, "interpreter": { "hash": "d990147e05fc0cc60dd3871899a6233eb6a5324c1885ded43d013dc915f7e535" } }, "nbformat": 4, "nbformat_minor": 2 }