From 9d75e7f330f8032ef3599c4411dbbed0556a7d26 Mon Sep 17 00:00:00 2001 From: Yanam24 Date: Tue, 17 Dec 2024 19:01:55 +0100 Subject: [PATCH] Custom rnn layers for TFT --- nbs/models.tft.ipynb | 1590 +++++++++++++++++++++++++++------- neuralforecast/models/tft.py | 144 ++- 2 files changed, 1421 insertions(+), 313 deletions(-) diff --git a/nbs/models.tft.ipynb b/nbs/models.tft.ipynb index bae287acf..cc47e3dfe 100644 --- a/nbs/models.tft.ipynb +++ b/nbs/models.tft.ipynb @@ -4,7 +4,15 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "env: PYTORCH_ENABLE_MPS_FALLBACK=1\n" + ] + } + ], "source": [ "%set_env PYTORCH_ENABLE_MPS_FALLBACK=1" ] @@ -15,7 +23,7 @@ "metadata": {}, "outputs": [], "source": [ - "#| default_exp models.tft" + "# | default_exp models.tft" ] }, { @@ -52,17 +60,18 @@ "metadata": {}, "outputs": [], "source": [ - "#| export\n", - "from typing import Tuple, Optional, Callable\n", + "# | export\n", + "from typing import Callable, Optional, Tuple\n", "\n", + "import pandas as pd\n", "import torch\n", "import torch.nn as nn\n", "import torch.nn.functional as F\n", "from torch import Tensor\n", "from torch.nn import LayerNorm\n", - "import pandas as pd\n", - "from neuralforecast.losses.pytorch import MAE\n", - "from neuralforecast.common._base_windows import BaseWindows" + "\n", + "from neuralforecast.common._base_windows import BaseWindows\n", + "from neuralforecast.losses.pytorch import MAE" ] }, { @@ -71,11 +80,10 @@ "metadata": {}, "outputs": [], "source": [ - "#| hide\n", + "# | hide\n", "import logging\n", "import warnings\n", "\n", - "from fastcore.test import test_eq\n", "from nbdev.showdoc import show_doc" ] }, @@ -85,7 +93,7 @@ "metadata": {}, "outputs": [], "source": [ - "#| hide\n", + "# | hide\n", "logging.getLogger(\"pytorch_lightning\").setLevel(logging.ERROR)\n", "warnings.filterwarnings(\"ignore\")" ] @@ -132,32 +140,33 @@ "metadata": {}, "outputs": [], "source": [ - "#| exporti\n", + "# | exporti\n", "def get_activation_fn(activation_str: str) -> Callable:\n", " activation_map = {\n", - " 'ReLU': F.relu,\n", - " 'Softplus': F.softplus,\n", - " 'Tanh': F.tanh,\n", - " 'SELU': F.selu,\n", - " 'LeakyReLU': F.leaky_relu,\n", - " 'Sigmoid': F.sigmoid,\n", - " 'ELU': F.elu,\n", - " 'GLU': F.glu\n", - " }\n", + " \"ReLU\": F.relu,\n", + " \"Softplus\": F.softplus,\n", + " \"Tanh\": F.tanh,\n", + " \"SELU\": F.selu,\n", + " \"LeakyReLU\": F.leaky_relu,\n", + " \"Sigmoid\": F.sigmoid,\n", + " \"ELU\": F.elu,\n", + " \"GLU\": F.glu,\n", + " }\n", " return activation_map.get(activation_str, F.elu)\n", "\n", + "\n", "class MaybeLayerNorm(nn.Module):\n", " def __init__(self, output_size, hidden_size, eps):\n", " super().__init__()\n", " if output_size and output_size == 1:\n", " self.ln = nn.Identity()\n", " else:\n", - " self.ln = LayerNorm(output_size if output_size else hidden_size,\n", - " eps=eps)\n", + " self.ln = LayerNorm(output_size if output_size else hidden_size, eps=eps)\n", "\n", " def forward(self, x):\n", " return self.ln(x)\n", "\n", + "\n", "class GLU(nn.Module):\n", " def __init__(self, hidden_size, output_size):\n", " super().__init__()\n", @@ -168,14 +177,17 @@ " x = F.glu(x)\n", " return x\n", "\n", + "\n", "class GRN(nn.Module):\n", - " def __init__(self,\n", - " input_size,\n", - " hidden_size,\n", - " output_size=None,\n", - " context_hidden_size=None,\n", - " dropout=0,\n", - " activation='ELU',):\n", + " def __init__(\n", + " self,\n", + " input_size,\n", + " hidden_size,\n", + " output_size=None,\n", + " context_hidden_size=None,\n", + " dropout=0,\n", + " activation=\"ELU\",\n", + " ):\n", " super().__init__()\n", " self.layer_norm = MaybeLayerNorm(output_size, hidden_size, eps=1e-3)\n", " self.lin_a = nn.Linear(input_size, hidden_size)\n", @@ -186,7 +198,7 @@ " self.dropout = nn.Dropout(dropout)\n", " self.out_proj = nn.Linear(input_size, output_size) if output_size else None\n", " self.activation_fn = get_activation_fn(activation)\n", - " \n", + "\n", " def forward(self, a: Tensor, c: Optional[Tensor] = None):\n", " x = self.lin_a(a)\n", " if c is not None:\n", @@ -238,9 +250,11 @@ "metadata": {}, "outputs": [], "source": [ - "#| exporti\n", + "# | exporti\n", "class TFTEmbedding(nn.Module):\n", - " def __init__(self, hidden_size, stat_input_size, futr_input_size, hist_input_size, tgt_size):\n", + " def __init__(\n", + " self, hidden_size, stat_input_size, futr_input_size, hist_input_size, tgt_size\n", + " ):\n", " super().__init__()\n", " # There are 4 types of input:\n", " # 1. Static continuous\n", @@ -253,92 +267,111 @@ " self.stat_input_size = stat_input_size\n", " self.futr_input_size = futr_input_size\n", " self.hist_input_size = hist_input_size\n", - " self.tgt_size = tgt_size\n", + " self.tgt_size = tgt_size\n", "\n", " # Instantiate Continuous Embeddings if size is not None\n", - " for attr, size in [('stat_exog_embedding', stat_input_size), \n", - " ('futr_exog_embedding', futr_input_size),\n", - " ('hist_exog_embedding', hist_input_size),\n", - " ('tgt_embedding', tgt_size)]:\n", + " for attr, size in [\n", + " (\"stat_exog_embedding\", stat_input_size),\n", + " (\"futr_exog_embedding\", futr_input_size),\n", + " (\"hist_exog_embedding\", hist_input_size),\n", + " (\"tgt_embedding\", tgt_size),\n", + " ]:\n", " if size:\n", " vectors = nn.Parameter(torch.Tensor(size, hidden_size))\n", " bias = nn.Parameter(torch.zeros(size, hidden_size))\n", " torch.nn.init.xavier_normal_(vectors)\n", - " setattr(self, attr+'_vectors', vectors)\n", - " setattr(self, attr+'_bias', bias)\n", + " setattr(self, attr + \"_vectors\", vectors)\n", + " setattr(self, attr + \"_bias\", bias)\n", " else:\n", - " setattr(self, attr+'_vectors', None)\n", - " setattr(self, attr+'_bias', None)\n", - "\n", - " def _apply_embedding(self,\n", - " cont: Optional[Tensor],\n", - " cont_emb: Tensor,\n", - " cont_bias: Tensor,\n", - " ):\n", - "\n", - " if (cont is not None):\n", - " #the line below is equivalent to following einsums\n", - " #e_cont = torch.einsum('btf,fh->bthf', cont, cont_emb)\n", - " #e_cont = torch.einsum('bf,fh->bhf', cont, cont_emb) \n", + " setattr(self, attr + \"_vectors\", None)\n", + " setattr(self, attr + \"_bias\", None)\n", + "\n", + " def _apply_embedding(\n", + " self,\n", + " cont: Optional[Tensor],\n", + " cont_emb: Tensor,\n", + " cont_bias: Tensor,\n", + " ):\n", + " if cont is not None:\n", + " # the line below is equivalent to following einsums\n", + " # e_cont = torch.einsum('btf,fh->bthf', cont, cont_emb)\n", + " # e_cont = torch.einsum('bf,fh->bhf', cont, cont_emb)\n", " e_cont = torch.mul(cont.unsqueeze(-1), cont_emb)\n", " e_cont = e_cont + cont_bias\n", " return e_cont\n", - " \n", + "\n", " return None\n", "\n", - " def forward(self, target_inp, \n", - " stat_exog=None, futr_exog=None, hist_exog=None):\n", - " # temporal/static categorical/continuous known/observed input \n", + " def forward(self, target_inp, stat_exog=None, futr_exog=None, hist_exog=None):\n", + " # temporal/static categorical/continuous known/observed input\n", " # tries to get input, if fails returns None\n", "\n", " # Static inputs are expected to be equal for all timesteps\n", " # For memory efficiency there is no assert statement\n", - " stat_exog = stat_exog[:,:] if stat_exog is not None else None\n", - "\n", - " s_inp = self._apply_embedding(cont=stat_exog,\n", - " cont_emb=self.stat_exog_embedding_vectors,\n", - " cont_bias=self.stat_exog_embedding_bias)\n", - " k_inp = self._apply_embedding(cont=futr_exog,\n", - " cont_emb=self.futr_exog_embedding_vectors,\n", - " cont_bias=self.futr_exog_embedding_bias)\n", - " o_inp = self._apply_embedding(cont=hist_exog,\n", - " cont_emb=self.hist_exog_embedding_vectors,\n", - " cont_bias=self.hist_exog_embedding_bias)\n", + " stat_exog = stat_exog[:, :] if stat_exog is not None else None\n", + "\n", + " s_inp = self._apply_embedding(\n", + " cont=stat_exog,\n", + " cont_emb=self.stat_exog_embedding_vectors,\n", + " cont_bias=self.stat_exog_embedding_bias,\n", + " )\n", + " k_inp = self._apply_embedding(\n", + " cont=futr_exog,\n", + " cont_emb=self.futr_exog_embedding_vectors,\n", + " cont_bias=self.futr_exog_embedding_bias,\n", + " )\n", + " o_inp = self._apply_embedding(\n", + " cont=hist_exog,\n", + " cont_emb=self.hist_exog_embedding_vectors,\n", + " cont_bias=self.hist_exog_embedding_bias,\n", + " )\n", "\n", " # Temporal observed targets\n", - " # t_observed_tgt = torch.einsum('btf,fh->btfh', \n", - " # target_inp, self.tgt_embedding_vectors) \n", - " target_inp = torch.matmul(target_inp.unsqueeze(3).unsqueeze(4),\n", - " self.tgt_embedding_vectors.unsqueeze(1)).squeeze(3)\n", + " # t_observed_tgt = torch.einsum('btf,fh->btfh',\n", + " # target_inp, self.tgt_embedding_vectors)\n", + " target_inp = torch.matmul(\n", + " target_inp.unsqueeze(3).unsqueeze(4),\n", + " self.tgt_embedding_vectors.unsqueeze(1),\n", + " ).squeeze(3)\n", " target_inp = target_inp + self.tgt_embedding_bias\n", "\n", " return s_inp, k_inp, o_inp, target_inp\n", "\n", + "\n", "class VariableSelectionNetwork(nn.Module):\n", " def __init__(self, hidden_size, num_inputs, dropout, grn_activation):\n", " super().__init__()\n", - " self.joint_grn = GRN(input_size=hidden_size*num_inputs, \n", - " hidden_size=hidden_size, \n", - " output_size=num_inputs, \n", - " context_hidden_size=hidden_size,\n", - " activation=grn_activation)\n", + " self.joint_grn = GRN(\n", + " input_size=hidden_size * num_inputs,\n", + " hidden_size=hidden_size,\n", + " output_size=num_inputs,\n", + " context_hidden_size=hidden_size,\n", + " activation=grn_activation,\n", + " )\n", " self.var_grns = nn.ModuleList(\n", - " [GRN(input_size=hidden_size, \n", - " hidden_size=hidden_size, dropout=dropout, activation=grn_activation)\n", - " for _ in range(num_inputs)])\n", + " [\n", + " GRN(\n", + " input_size=hidden_size,\n", + " hidden_size=hidden_size,\n", + " dropout=dropout,\n", + " activation=grn_activation,\n", + " )\n", + " for _ in range(num_inputs)\n", + " ]\n", + " )\n", "\n", " def forward(self, x: Tensor, context: Optional[Tensor] = None):\n", " Xi = x.reshape(*x.shape[:-2], -1)\n", " grn_outputs = self.joint_grn(Xi, c=context)\n", " sparse_weights = F.softmax(grn_outputs, dim=-1)\n", - " transformed_embed_list = [m(x[...,i,:])\n", - " for i, m in enumerate(self.var_grns)]\n", + " transformed_embed_list = [m(x[..., i, :]) for i, m in enumerate(self.var_grns)]\n", " transformed_embed = torch.stack(transformed_embed_list, dim=-1)\n", - " #the line below performs batched matrix vector multiplication\n", - " #for temporal features it's bthf,btf->bth\n", - " #for static features it's bhf,bf->bh\n", - " variable_ctx = torch.matmul(transformed_embed, \n", - " sparse_weights.unsqueeze(-1)).squeeze(-1)\n", + " # the line below performs batched matrix vector multiplication\n", + " # for temporal features it's bthf,btf->bth\n", + " # for static features it's bhf,bf->bh\n", + " variable_ctx = torch.matmul(\n", + " transformed_embed, sparse_weights.unsqueeze(-1)\n", + " ).squeeze(-1)\n", "\n", " return variable_ctx, sparse_weights" ] @@ -375,7 +408,7 @@ "metadata": {}, "outputs": [], "source": [ - "#| exporti\n", + "# | exporti\n", "class InterpretableMultiHeadAttention(nn.Module):\n", " def __init__(self, n_head, hidden_size, example_length, attn_dropout, dropout):\n", " super().__init__()\n", @@ -467,17 +500,37 @@ "metadata": {}, "outputs": [], "source": [ - "#| exporti\n", + "# | exporti\n", "class StaticCovariateEncoder(nn.Module):\n", - " def __init__(self, hidden_size, num_static_vars, dropout, grn_activation):\n", + " def __init__(\n", + " self,\n", + " hidden_size,\n", + " num_static_vars,\n", + " dropout,\n", + " grn_activation,\n", + " rnn_type=\"lstm\",\n", + " n_rnn_layers=1,\n", + " one_rnn_initial_state=False,\n", + " ):\n", " super().__init__()\n", " self.vsn = VariableSelectionNetwork(\n", - " hidden_size=hidden_size, num_inputs=num_static_vars, dropout=dropout, grn_activation=grn_activation\n", + " hidden_size=hidden_size,\n", + " num_inputs=num_static_vars,\n", + " dropout=dropout,\n", + " grn_activation=grn_activation,\n", " )\n", + " self.rnn_type = rnn_type.lower()\n", + "\n", + " self.n_rnn_layers = n_rnn_layers\n", + "\n", + " self.n_states = 1 if one_rnn_initial_state else n_rnn_layers\n", + "\n", + " n_contexts = 2 + 2 * self.n_states if rnn_type == \"lstm\" else 2 + self.n_states\n", + "\n", " self.context_grns = nn.ModuleList(\n", " [\n", " GRN(input_size=hidden_size, hidden_size=hidden_size, dropout=dropout)\n", - " for _ in range(4)\n", + " for _ in range(n_contexts)\n", " ]\n", " )\n", "\n", @@ -489,9 +542,46 @@ " # enrichment context\n", " # state_c context\n", " # state_h context\n", - " cs, ce, ch, cc = tuple(m(variable_ctx) for m in self.context_grns) # type: ignore\n", "\n", - " return cs, ce, ch, cc, sparse_weights # type: ignore" + " cs, ce = list(m(variable_ctx) for m in self.context_grns[:2]) # type: ignore\n", + "\n", + " if self.n_states == 1:\n", + " ch = torch.cat(\n", + " self.n_rnn_layers\n", + " * list(\n", + " m(variable_ctx).unsqueeze(0)\n", + " for m in self.context_grns[2 : self.n_states + 2]\n", + " )\n", + " )\n", + "\n", + " if self.rnn_type == \"lstm\":\n", + " cc = torch.cat(\n", + " self.n_rnn_layers\n", + " * list(\n", + " m(variable_ctx).unsqueeze(0)\n", + " for m in self.context_grns[self.n_states + 2 :]\n", + " )\n", + " )\n", + "\n", + " else:\n", + " ch = torch.cat(\n", + " list(\n", + " m(variable_ctx).unsqueeze(0)\n", + " for m in self.context_grns[2 : self.n_states + 2]\n", + " )\n", + " )\n", + "\n", + " if self.rnn_type == \"lstm\":\n", + " cc = torch.cat(\n", + " list(\n", + " m(variable_ctx).unsqueeze(0)\n", + " for m in self.context_grns[self.n_states + 2 :]\n", + " )\n", + " )\n", + " if self.rnn_type != \"lstm\":\n", + " cc = ch\n", + "\n", + " return cs, ce, ch, cc, sparse_weights # type: ignore" ] }, { @@ -524,23 +614,64 @@ "metadata": {}, "outputs": [], "source": [ - "#| exporti\n", + "# | exporti\n", "class TemporalCovariateEncoder(nn.Module):\n", - " def __init__(self, hidden_size, num_historic_vars, num_future_vars, dropout, grn_activation):\n", + " def __init__(\n", + " self,\n", + " hidden_size,\n", + " num_historic_vars,\n", + " num_future_vars,\n", + " dropout,\n", + " grn_activation,\n", + " rnn_type=\"lstm\",\n", + " n_rnn_layers=1,\n", + " ):\n", " super(TemporalCovariateEncoder, self).__init__()\n", + " self.rnn_type = rnn_type.lower()\n", + " self.n_rnn_layers = n_rnn_layers\n", "\n", " self.history_vsn = VariableSelectionNetwork(\n", - " hidden_size=hidden_size, num_inputs=num_historic_vars, dropout=dropout, grn_activation=grn_activation\n", - " )\n", - " self.history_encoder = nn.LSTM(\n", - " input_size=hidden_size, hidden_size=hidden_size, batch_first=True\n", + " hidden_size=hidden_size,\n", + " num_inputs=num_historic_vars,\n", + " dropout=dropout,\n", + " grn_activation=grn_activation,\n", " )\n", + " if self.rnn_type == \"lstm\":\n", + " self.history_encoder = nn.LSTM(\n", + " input_size=hidden_size,\n", + " hidden_size=hidden_size,\n", + " batch_first=True,\n", + " num_layers=n_rnn_layers,\n", + " )\n", + "\n", + " self.future_encoder = nn.LSTM(\n", + " input_size=hidden_size,\n", + " hidden_size=hidden_size,\n", + " batch_first=True,\n", + " num_layers=n_rnn_layers,\n", + " )\n", + "\n", + " elif self.rnn_type == \"gru\":\n", + " self.history_encoder = nn.GRU(\n", + " input_size=hidden_size,\n", + " hidden_size=hidden_size,\n", + " batch_first=True,\n", + " num_layers=n_rnn_layers,\n", + " )\n", + " self.future_encoder = nn.GRU(\n", + " input_size=hidden_size,\n", + " hidden_size=hidden_size,\n", + " batch_first=True,\n", + " num_layers=n_rnn_layers,\n", + " )\n", + " else:\n", + " raise ValueError('RNN type should be in [\"lstm\",\"gru\"] !')\n", "\n", " self.future_vsn = VariableSelectionNetwork(\n", - " hidden_size=hidden_size, num_inputs=num_future_vars, dropout=dropout, grn_activation=grn_activation\n", - " )\n", - " self.future_encoder = nn.LSTM(\n", - " input_size=hidden_size, hidden_size=hidden_size, batch_first=True\n", + " hidden_size=hidden_size,\n", + " num_inputs=num_future_vars,\n", + " dropout=dropout,\n", + " grn_activation=grn_activation,\n", " )\n", "\n", " # Shared Gated-Skip Connection\n", @@ -552,7 +683,11 @@ " historical_features, history_vsn_sparse_weights = self.history_vsn(\n", " historical_inputs, cs\n", " )\n", - " history, state = self.history_encoder(historical_features, (ch, cc))\n", + " if self.rnn_type == \"lstm\":\n", + " history, state = self.history_encoder(historical_features, (ch, cc))\n", + "\n", + " elif self.rnn_type == \"gru\":\n", + " history, state = self.history_encoder(historical_features, ch)\n", "\n", " future_features, future_vsn_sparse_weights = self.future_vsn(future_inputs, cs)\n", " future, _ = self.future_encoder(future_features, state)\n", @@ -588,10 +723,17 @@ "metadata": {}, "outputs": [], "source": [ - "#| exporti\n", + "# | exporti\n", "class TemporalFusionDecoder(nn.Module):\n", " def __init__(\n", - " self, n_head, hidden_size, example_length, encoder_length, attn_dropout, dropout, grn_activation\n", + " self,\n", + " n_head,\n", + " hidden_size,\n", + " example_length,\n", + " encoder_length,\n", + " attn_dropout,\n", + " dropout,\n", + " grn_activation,\n", " ):\n", " super(TemporalFusionDecoder, self).__init__()\n", " self.encoder_length = encoder_length\n", @@ -602,7 +744,7 @@ " hidden_size=hidden_size,\n", " context_hidden_size=hidden_size,\n", " dropout=dropout,\n", - " activation=grn_activation\n", + " activation=grn_activation,\n", " )\n", " self.attention = InterpretableMultiHeadAttention(\n", " n_head=n_head,\n", @@ -615,7 +757,10 @@ " self.attention_ln = LayerNorm(normalized_shape=hidden_size, eps=1e-3)\n", "\n", " self.positionwise_grn = GRN(\n", - " input_size=hidden_size, hidden_size=hidden_size, dropout=dropout, activation=grn_activation\n", + " input_size=hidden_size,\n", + " hidden_size=hidden_size,\n", + " dropout=dropout,\n", + " activation=grn_activation,\n", " )\n", "\n", " # ---------------------- Decoder -----------------------#\n", @@ -657,7 +802,7 @@ "metadata": {}, "outputs": [], "source": [ - "#| export\n", + "# | export\n", "class TFT(BaseWindows):\n", " \"\"\"TFT\n", "\n", @@ -678,6 +823,9 @@ " `n_head`: int=4, number of attention heads in temporal fusion decoder.
\n", " `attn_dropout`: float (0, 1), dropout of fusion decoder's attention layer.
\n", " `grn_activation`: str, activation for the GRN module from ['ReLU', 'Softplus', 'Tanh', 'SELU', 'LeakyReLU', 'Sigmoid', 'ELU', 'GLU'].
\n", + " `rnn_type`: str=\"LSTM\", recurrent neural network (RNN) layer type from [\"LSTM\",\"GRU\"].
\n", + " `n_rnn_layers`: int=1, number of RNN layers.
\n", + " `one_rnn_initial_state`:str=False, Initialize all rnn layers with the same initial states computed from static covariates.
\n", " `loss`: PyTorch module, instantiated train loss class from [losses collection](https://nixtla.github.io/neuralforecast/losses.pytorch.html).
\n", " `valid_loss`: PyTorch module=`loss`, instantiated valid loss class from [losses collection](https://nixtla.github.io/neuralforecast/losses.pytorch.html).
\n", " `max_steps`: int=1000, maximum number of training steps.
\n", @@ -725,7 +873,10 @@ " hidden_size: int = 128,\n", " n_head: int = 4,\n", " attn_dropout: float = 0.0,\n", - " grn_activation: str = 'ELU',\n", + " grn_activation: str = \"ELU\",\n", + " n_rnn_layers: int = 1,\n", + " rnn_type: str = \"LSTM\",\n", + " one_rnn_initial_state: bool = False,\n", " dropout: float = 0.1,\n", " loss=MAE(),\n", " valid_loss=None,\n", @@ -748,10 +899,9 @@ " optimizer_kwargs=None,\n", " lr_scheduler=None,\n", " lr_scheduler_kwargs=None,\n", - " dataloader_kwargs = None,\n", + " dataloader_kwargs=None,\n", " **trainer_kwargs,\n", " ):\n", - "\n", " # Inherit BaseWindows class\n", " super(TFT, self).__init__(\n", " h=h,\n", @@ -784,32 +934,40 @@ " **trainer_kwargs,\n", " )\n", " self.example_length = input_size + h\n", - " self.interpretability_params = dict([]) # type: ignore\n", + " self.interpretability_params = dict([]) # type: ignore\n", " self.tgt_size = tgt_size\n", " self.grn_activation = grn_activation\n", " futr_exog_size = max(self.futr_exog_size, 1)\n", " num_historic_vars = futr_exog_size + self.hist_exog_size + tgt_size\n", + " self.n_rnn_layers = n_rnn_layers\n", + " # ------------------------------- Encoders -----------------------------#\n", + " self.embedding = TFTEmbedding(\n", + " hidden_size=hidden_size,\n", + " stat_input_size=self.stat_exog_size,\n", + " futr_input_size=futr_exog_size,\n", + " hist_input_size=self.hist_exog_size,\n", + " tgt_size=tgt_size,\n", + " )\n", "\n", - " #------------------------------- Encoders -----------------------------#\n", - " self.embedding = TFTEmbedding(hidden_size=hidden_size,\n", - " stat_input_size=self.stat_exog_size,\n", - " futr_input_size=futr_exog_size,\n", - " hist_input_size=self.hist_exog_size,\n", - " tgt_size=tgt_size)\n", - " \n", " if self.stat_exog_size > 0:\n", " self.static_encoder = StaticCovariateEncoder(\n", - " hidden_size=hidden_size,\n", - " num_static_vars=self.stat_exog_size,\n", - " dropout=dropout,\n", - " grn_activation=self.grn_activation)\n", + " hidden_size=hidden_size,\n", + " num_static_vars=self.stat_exog_size,\n", + " dropout=dropout,\n", + " grn_activation=self.grn_activation,\n", + " rnn_type=rnn_type,\n", + " n_rnn_layers=n_rnn_layers,\n", + " one_rnn_initial_state=one_rnn_initial_state,\n", + " )\n", "\n", " self.temporal_encoder = TemporalCovariateEncoder(\n", " hidden_size=hidden_size,\n", " num_historic_vars=num_historic_vars,\n", " num_future_vars=futr_exog_size,\n", " dropout=dropout,\n", - " grn_activation=self.grn_activation\n", + " grn_activation=self.grn_activation,\n", + " n_rnn_layers=n_rnn_layers,\n", + " rnn_type=rnn_type,\n", " )\n", "\n", " # ------------------------------ Decoders -----------------------------#\n", @@ -820,7 +978,7 @@ " encoder_length=self.input_size,\n", " attn_dropout=attn_dropout,\n", " dropout=dropout,\n", - " grn_activation=self.grn_activation\n", + " grn_activation=self.grn_activation,\n", " )\n", "\n", " # Adapter with Loss dependent dimensions\n", @@ -829,7 +987,6 @@ " )\n", "\n", " def forward(self, windows_batch):\n", - "\n", " # Parsiw windows_batch\n", " y_insample = windows_batch[\"insample_y\"][:, :, None] # <- [B,T,1]\n", " futr_exog = windows_batch[\"futr_exog\"]\n", @@ -851,17 +1008,19 @@ " # Static context\n", " if s_inp is not None:\n", " cs, ce, ch, cc, static_encoder_sparse_weights = self.static_encoder(s_inp)\n", - " ch, cc = ch.unsqueeze(0), cc.unsqueeze(0) # LSTM initial states\n", + " # ch, cc = ch.unsqueeze(0), cc.unsqueeze(0) # LSTM initial states\n", " else:\n", " # If None add zeros\n", " batch_size, example_length, target_size, hidden_size = t_observed_tgt.shape\n", " cs = torch.zeros(size=(batch_size, hidden_size), device=y_insample.device)\n", " ce = torch.zeros(size=(batch_size, hidden_size), device=y_insample.device)\n", " ch = torch.zeros(\n", - " size=(1, batch_size, hidden_size), device=y_insample.device\n", + " size=(self.n_rnn_layers, batch_size, hidden_size),\n", + " device=y_insample.device,\n", " )\n", " cc = torch.zeros(\n", - " size=(1, batch_size, hidden_size), device=y_insample.device\n", + " size=(self.n_rnn_layers, batch_size, hidden_size),\n", + " device=y_insample.device,\n", " )\n", " static_encoder_sparse_weights = []\n", "\n", @@ -942,16 +1101,17 @@ " self.mean_on_batch(hist_vsn_wgts).cpu().numpy(), columns=hist_exog_list\n", " )\n", " importances[\"Past variable importance over time\"] = hist_vsn_imp\n", - " # importances[\"Past variable importance\"] = hist_vsn_imp.mean(axis=0).sort_values()\n", + " # importances[\"Past variable importance\"] = hist_vsn_imp.mean(axis=0).sort_values()\n", "\n", " # Future feature importances\n", " if self.futr_exog_size > 0:\n", " future_vsn_wgts = self.interpretability_params.get(\"future_vsn_wgts\")\n", " future_vsn_imp = pd.DataFrame(\n", - " self.mean_on_batch(future_vsn_wgts).cpu().numpy(), columns=self.futr_exog_list\n", + " self.mean_on_batch(future_vsn_wgts).cpu().numpy(),\n", + " columns=self.futr_exog_list,\n", " )\n", " importances[\"Future variable importance over time\"] = future_vsn_imp\n", - " # importances[\"Future variable importance\"] = future_vsn_imp.mean(axis=0).sort_values()\n", + " # importances[\"Future variable importance\"] = future_vsn_imp.mean(axis=0).sort_values()\n", "\n", " # Static feature importances\n", " if self.stat_exog_size > 0:\n", @@ -969,16 +1129,16 @@ " )\n", "\n", " return importances\n", - " \n", + "\n", " def attention_weights(self):\n", - " \"\"\" \n", + " \"\"\"\n", " Batch average attention weights\n", - " \n", + "\n", " Returns:\n", " np.ndarray: A 1D array containing the attention weights for each time step.\n", - " \n", + "\n", " \"\"\"\n", - " \n", + "\n", " attention = (\n", " self.mean_on_batch(self.interpretability_params[\"attn_wts\"])\n", " .mean(dim=0)\n", @@ -987,11 +1147,11 @@ " )\n", "\n", " return attention\n", - " \n", - " def feature_importance_correlations(self)-> pd.DataFrame:\n", + "\n", + " def feature_importance_correlations(self) -> pd.DataFrame:\n", " \"\"\"\n", " Compute the correlation between the past and future feature importances and the mean attention weights.\n", - " \n", + "\n", " Returns:\n", " pd.DataFrame: A DataFrame containing the correlation coefficients between the past feature importances and the mean attention weights.\n", " \"\"\"\n", @@ -1013,54 +1173,318 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/markdown": [ + "---\n", + "\n", + "### TFT.fit\n", + "\n", + "> TFT.fit (dataset, val_size=0, test_size=0, random_seed=None,\n", + "> distributed_config=None)\n", + "\n", + "*Fit.\n", + "\n", + "The `fit` method, optimizes the neural network's weights using the\n", + "initialization parameters (`learning_rate`, `windows_batch_size`, ...)\n", + "and the `loss` function as defined during the initialization.\n", + "Within `fit` we use a PyTorch Lightning `Trainer` that\n", + "inherits the initialization's `self.trainer_kwargs`, to customize\n", + "its inputs, see [PL's trainer arguments](https://pytorch-lightning.readthedocs.io/en/stable/api/pytorch_lightning.trainer.trainer.Trainer.html?highlight=trainer).\n", + "\n", + "The method is designed to be compatible with SKLearn-like classes\n", + "and in particular to be compatible with the StatsForecast library.\n", + "\n", + "By default the `model` is not saving training checkpoints to protect\n", + "disk memory, to get them change `enable_checkpointing=True` in `__init__`.\n", + "\n", + "**Parameters:**
\n", + "`dataset`: NeuralForecast's `TimeSeriesDataset`, see [documentation](https://nixtla.github.io/neuralforecast/tsdataset.html).
\n", + "`val_size`: int, validation size for temporal cross-validation.
\n", + "`random_seed`: int=None, random_seed for pytorch initializer and numpy generators, overwrites model.__init__'s.
\n", + "`test_size`: int, test size for temporal cross-validation.
*" + ], + "text/plain": [ + "---\n", + "\n", + "### TFT.fit\n", + "\n", + "> TFT.fit (dataset, val_size=0, test_size=0, random_seed=None,\n", + "> distributed_config=None)\n", + "\n", + "*Fit.\n", + "\n", + "The `fit` method, optimizes the neural network's weights using the\n", + "initialization parameters (`learning_rate`, `windows_batch_size`, ...)\n", + "and the `loss` function as defined during the initialization.\n", + "Within `fit` we use a PyTorch Lightning `Trainer` that\n", + "inherits the initialization's `self.trainer_kwargs`, to customize\n", + "its inputs, see [PL's trainer arguments](https://pytorch-lightning.readthedocs.io/en/stable/api/pytorch_lightning.trainer.trainer.Trainer.html?highlight=trainer).\n", + "\n", + "The method is designed to be compatible with SKLearn-like classes\n", + "and in particular to be compatible with the StatsForecast library.\n", + "\n", + "By default the `model` is not saving training checkpoints to protect\n", + "disk memory, to get them change `enable_checkpointing=True` in `__init__`.\n", + "\n", + "**Parameters:**
\n", + "`dataset`: NeuralForecast's `TimeSeriesDataset`, see [documentation](https://nixtla.github.io/neuralforecast/tsdataset.html).
\n", + "`val_size`: int, validation size for temporal cross-validation.
\n", + "`random_seed`: int=None, random_seed for pytorch initializer and numpy generators, overwrites model.__init__'s.
\n", + "`test_size`: int, test size for temporal cross-validation.
*" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "show_doc(TFT.fit, name='TFT.fit', title_level=3)" + "show_doc(TFT.fit, name=\"TFT.fit\", title_level=3)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/markdown": [ + "---\n", + "\n", + "### TFT.predict\n", + "\n", + "> TFT.predict (dataset, test_size=None, step_size=1, random_seed=None,\n", + "> **data_module_kwargs)\n", + "\n", + "*Predict.\n", + "\n", + "Neural network prediction with PL's `Trainer` execution of `predict_step`.\n", + "\n", + "**Parameters:**
\n", + "`dataset`: NeuralForecast's `TimeSeriesDataset`, see [documentation](https://nixtla.github.io/neuralforecast/tsdataset.html).
\n", + "`test_size`: int=None, test size for temporal cross-validation.
\n", + "`step_size`: int=1, Step size between each window.
\n", + "`random_seed`: int=None, random_seed for pytorch initializer and numpy generators, overwrites model.__init__'s.
\n", + "`**data_module_kwargs`: PL's TimeSeriesDataModule args, see [documentation](https://pytorch-lightning.readthedocs.io/en/1.6.1/extensions/datamodules.html#using-a-datamodule).*" + ], + "text/plain": [ + "---\n", + "\n", + "### TFT.predict\n", + "\n", + "> TFT.predict (dataset, test_size=None, step_size=1, random_seed=None,\n", + "> **data_module_kwargs)\n", + "\n", + "*Predict.\n", + "\n", + "Neural network prediction with PL's `Trainer` execution of `predict_step`.\n", + "\n", + "**Parameters:**
\n", + "`dataset`: NeuralForecast's `TimeSeriesDataset`, see [documentation](https://nixtla.github.io/neuralforecast/tsdataset.html).
\n", + "`test_size`: int=None, test size for temporal cross-validation.
\n", + "`step_size`: int=1, Step size between each window.
\n", + "`random_seed`: int=None, random_seed for pytorch initializer and numpy generators, overwrites model.__init__'s.
\n", + "`**data_module_kwargs`: PL's TimeSeriesDataModule args, see [documentation](https://pytorch-lightning.readthedocs.io/en/1.6.1/extensions/datamodules.html#using-a-datamodule).*" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "show_doc(TFT.predict, name='TFT.predict', title_level=3)" + "show_doc(TFT.predict, name=\"TFT.predict\", title_level=3)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/markdown": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L679){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.feature_importances,\n", + "\n", + "> TFT.feature_importances, ()\n", + "\n", + "*Compute the feature importances for historical, future, and static features.\n", + "\n", + "Returns:\n", + " dict: A dictionary containing the feature importances for each feature type.\n", + " The keys are 'hist_vsn', 'future_vsn', and 'static_vsn', and the values\n", + " are pandas DataFrames with the corresponding feature importances.*" + ], + "text/plain": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L679){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.feature_importances,\n", + "\n", + "> TFT.feature_importances, ()\n", + "\n", + "*Compute the feature importances for historical, future, and static features.\n", + "\n", + "Returns:\n", + " dict: A dictionary containing the feature importances for each feature type.\n", + " The keys are 'hist_vsn', 'future_vsn', and 'static_vsn', and the values\n", + " are pandas DataFrames with the corresponding feature importances.*" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "show_doc(TFT.feature_importances, name='TFT.feature_importances,', title_level=3)" + "show_doc(TFT.feature_importances, name=\"TFT.feature_importances,\", title_level=3)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/markdown": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L738){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.attention_weights\n", + "\n", + "> TFT.attention_weights ()\n", + "\n", + "*Batch average attention weights\n", + "\n", + "Returns:\n", + "np.ndarray: A 1D array containing the attention weights for each time step.*" + ], + "text/plain": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L738){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.attention_weights\n", + "\n", + "> TFT.attention_weights ()\n", + "\n", + "*Batch average attention weights\n", + "\n", + "Returns:\n", + "np.ndarray: A 1D array containing the attention weights for each time step.*" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "show_doc(TFT.attention_weights , name='TFT.attention_weights', title_level=3)" + "show_doc(TFT.attention_weights, name=\"TFT.attention_weights\", title_level=3)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/markdown": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L738){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.attention_weights\n", + "\n", + "> TFT.attention_weights ()\n", + "\n", + "*Batch average attention weights\n", + "\n", + "Returns:\n", + "np.ndarray: A 1D array containing the attention weights for each time step.*" + ], + "text/plain": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L738){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.attention_weights\n", + "\n", + "> TFT.attention_weights ()\n", + "\n", + "*Batch average attention weights\n", + "\n", + "Returns:\n", + "np.ndarray: A 1D array containing the attention weights for each time step.*" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "show_doc(TFT.attention_weights , name='TFT.attention_weights', title_level=3)" + "show_doc(TFT.attention_weights, name=\"TFT.attention_weights\", title_level=3)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/markdown": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L756){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.feature_importance_correlations\n", + "\n", + "> TFT.feature_importance_correlations ()\n", + "\n", + "*Compute the correlation between the past and future feature importances and the mean attention weights.\n", + "\n", + "Returns:\n", + "pd.DataFrame: A DataFrame containing the correlation coefficients between the past feature importances and the mean attention weights.*" + ], + "text/plain": [ + "---\n", + "\n", + "[source](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/tft.py#L756){target=\"_blank\" style=\"float:right; font-size:smaller\"}\n", + "\n", + "### TFT.feature_importance_correlations\n", + "\n", + "> TFT.feature_importance_correlations ()\n", + "\n", + "*Compute the correlation between the past and future feature importances and the mean attention weights.\n", + "\n", + "Returns:\n", + "pd.DataFrame: A DataFrame containing the correlation coefficients between the past feature importances and the mean attention weights.*" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "show_doc(TFT.feature_importance_correlations , name='TFT.feature_importance_correlations', title_level=3)" + "show_doc(\n", + " TFT.feature_importance_correlations,\n", + " name=\"TFT.feature_importance_correlations\",\n", + " title_level=3,\n", + ")" ] }, { @@ -1075,55 +1499,365 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Seed set to 1\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "e36d46990ccd41b592b8b41272f824a8", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Sanity Checking: | | 0/? [00:00" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "import pandas as pd\n", + "# | eval: false\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", + "import pandas as pd\n", + "\n", "from neuralforecast import NeuralForecast\n", - "from neuralforecast.models import TFT\n", + "\n", + "# from neuralforecast.models import TFT\n", "from neuralforecast.losses.pytorch import DistributionLoss\n", "from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic\n", "\n", - "AirPassengersPanel['month']=AirPassengersPanel.ds.dt.month\n", - "Y_train_df = AirPassengersPanel[AirPassengersPanel.ds=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test\n", + "AirPassengersPanel[\"month\"] = AirPassengersPanel.ds.dt.month\n", + "Y_train_df = AirPassengersPanel[\n", + " AirPassengersPanel.ds < AirPassengersPanel[\"ds\"].values[-12]\n", + "] # 132 train\n", + "Y_test_df = AirPassengersPanel[\n", + " AirPassengersPanel.ds >= AirPassengersPanel[\"ds\"].values[-12]\n", + "].reset_index(drop=True) # 12 test\n", "\n", "nf = NeuralForecast(\n", - " models=[TFT(h=12, input_size=48,\n", - " hidden_size=20,\n", - " grn_activation='ELU',\n", - " loss=DistributionLoss(distribution='StudentT', level=[80, 90]),\n", - " learning_rate=0.005,\n", - " stat_exog_list=['airline1'],\n", - " futr_exog_list=['y_[lag12]','month'],\n", - " hist_exog_list=['trend'],\n", - " max_steps=300,\n", - " val_check_steps=10,\n", - " early_stop_patience_steps=10,\n", - " scaler_type='robust',\n", - " windows_batch_size=None,\n", - " enable_progress_bar=True),\n", + " models=[\n", + " TFT(\n", + " h=12,\n", + " input_size=48,\n", + " hidden_size=20,\n", + " grn_activation=\"ELU\",\n", + " rnn_type=\"lstm\",\n", + " n_rnn_layers=1,\n", + " one_rnn_initial_state=False,\n", + " loss=DistributionLoss(distribution=\"StudentT\", level=[80, 90]),\n", + " learning_rate=0.005,\n", + " stat_exog_list=[\"airline1\"],\n", + " futr_exog_list=[\"y_[lag12]\", \"month\"],\n", + " hist_exog_list=[\"trend\"],\n", + " max_steps=300,\n", + " val_check_steps=10,\n", + " early_stop_patience_steps=10,\n", + " scaler_type=\"robust\",\n", + " windows_batch_size=None,\n", + " enable_progress_bar=True,\n", + " ),\n", " ],\n", - " freq='M'\n", + " freq=\"M\",\n", ")\n", "nf.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)\n", "Y_hat_df = nf.predict(futr_df=Y_test_df)\n", "\n", "# Plot quantile predictions\n", - "Y_hat_df = Y_hat_df.reset_index(drop=False).drop(columns=['unique_id','ds'])\n", + "Y_hat_df = Y_hat_df.reset_index(drop=False).drop(columns=[\"unique_id\", \"ds\"])\n", "plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)\n", "plot_df = pd.concat([Y_train_df, plot_df])\n", "\n", - "plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)\n", - "plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')\n", - "plt.plot(plot_df['ds'], plot_df['TFT'], c='purple', label='mean')\n", - "plt.plot(plot_df['ds'], plot_df['TFT-median'], c='blue', label='median')\n", - "plt.fill_between(x=plot_df['ds'][-12:], \n", - " y1=plot_df['TFT-lo-90'][-12:].values, \n", - " y2=plot_df['TFT-hi-90'][-12:].values,\n", - " alpha=0.4, label='level 90')\n", + "plot_df = plot_df[plot_df.unique_id == \"Airline1\"].drop(\"unique_id\", axis=1)\n", + "plt.plot(plot_df[\"ds\"], plot_df[\"y\"], c=\"black\", label=\"True\")\n", + "plt.plot(plot_df[\"ds\"], plot_df[\"TFT\"], c=\"purple\", label=\"mean\")\n", + "plt.plot(plot_df[\"ds\"], plot_df[\"TFT-median\"], c=\"blue\", label=\"median\")\n", + "plt.fill_between(\n", + " x=plot_df[\"ds\"][-12:],\n", + " y1=plot_df[\"TFT-lo-90\"][-12:].values,\n", + " y2=plot_df[\"TFT-hi-90\"][-12:].values,\n", + " alpha=0.4,\n", + " label=\"level 90\",\n", + ")\n", "plt.legend()\n", "plt.grid()\n", "plt.plot()" @@ -1149,7 +1883,7 @@ "metadata": {}, "outputs": [], "source": [ - "#| eval: false\n", + "# | eval: false\n", "attention = nf.models[0].attention_weights()" ] }, @@ -1159,82 +1893,104 @@ "metadata": {}, "outputs": [], "source": [ - "#| eval: false\n", - "def plot_attention(self, plot:str=\"time\", output:str='plot', width:int=800, height:int=400):\n", - " \"\"\"\n", - " Plot the attention weights.\n", - "\n", - " Args:\n", - " plot (str, optional): The type of plot to generate. Can be one of the following:\n", - " - 'time': Display the mean attention weights over time.\n", - " - 'all': Display the attention weights for each horizon.\n", - " - 'heatmap': Display the attention weights as a heatmap.\n", - " - An integer in the range [1, model.h) to display the attention weights for a specific horizon.\n", - " output (str, optional): The type of output to generate. Can be one of the following:\n", - " - 'plot': Display the plot directly.\n", - " - 'figure': Return the plot as a figure object.\n", - " width (int, optional): Width of the plot in pixels. Default is 800.\n", - " height (int, optional): Height of the plot in pixels. Default is 400.\n", - "\n", - " Returns:\n", - " matplotlib.figure.Figure: If `output` is 'figure', the function returns the plot as a figure object.\n", - " \"\"\"\n", + "# | eval: false\n", + "def plot_attention(\n", + " self, plot: str = \"time\", output: str = \"plot\", width: int = 800, height: int = 400\n", + "):\n", + " \"\"\"\n", + " Plot the attention weights.\n", + "\n", + " Args:\n", + " plot (str, optional): The type of plot to generate. Can be one of the following:\n", + " - 'time': Display the mean attention weights over time.\n", + " - 'all': Display the attention weights for each horizon.\n", + " - 'heatmap': Display the attention weights as a heatmap.\n", + " - An integer in the range [1, model.h) to display the attention weights for a specific horizon.\n", + " output (str, optional): The type of output to generate. Can be one of the following:\n", + " - 'plot': Display the plot directly.\n", + " - 'figure': Return the plot as a figure object.\n", + " width (int, optional): Width of the plot in pixels. Default is 800.\n", + " height (int, optional): Height of the plot in pixels. Default is 400.\n", + "\n", + " Returns:\n", + " matplotlib.figure.Figure: If `output` is 'figure', the function returns the plot as a figure object.\n", + " \"\"\"\n", "\n", - " attention = (\n", - " self.mean_on_batch(self.interpretability_params[\"attn_wts\"])\n", - " .mean(dim=0)\n", - " .cpu()\n", - " .numpy()\n", - " )\n", + " attention = (\n", + " self.mean_on_batch(self.interpretability_params[\"attn_wts\"])\n", + " .mean(dim=0)\n", + " .cpu()\n", + " .numpy()\n", + " )\n", "\n", - " fig, ax = plt.subplots(figsize=(width / 100, height / 100))\n", - "\n", - " if plot == \"time\":\n", - " attention = attention[self.input_size:, :].mean(axis=0)\n", - " ax.plot(np.arange(-self.input_size, self.h), attention)\n", - " ax.axvline(x=0, color='black', linewidth=3, linestyle='--', label=\"prediction start\")\n", - " ax.set_title(\"Mean Attention\")\n", - " ax.set_xlabel(\"time\")\n", - " ax.set_ylabel(\"Attention\")\n", - " ax.legend()\n", - "\n", - " elif plot == \"all\":\n", - " for i in range(self.input_size, attention.shape[0]):\n", - " ax.plot(np.arange(-self.input_size, self.h), attention[i, :], label=f\"horizon {i-self.input_size+1}\")\n", - " ax.axvline(x=0, color='black', linewidth=3, linestyle='--', label=\"prediction start\")\n", - " ax.set_title(\"Attention per horizon\")\n", - " ax.set_xlabel(\"time\")\n", - " ax.set_ylabel(\"Attention\")\n", - " ax.legend()\n", - "\n", - " elif plot == \"heatmap\":\n", - " cax = ax.imshow(attention, aspect='auto', cmap='viridis',\n", - " extent=[-self.input_size, self.h, -self.input_size, self.h])\n", - " fig.colorbar(cax)\n", - " ax.set_title(\"Attention Heatmap\")\n", - " ax.set_xlabel(\"Attention (current time step)\")\n", - " ax.set_ylabel(\"Attention (previous time step)\")\n", - "\n", - " elif isinstance(plot, int) and (plot in np.arange(1, self.h + 1)):\n", - " i = self.input_size + plot - 1\n", - " ax.plot(np.arange(-self.input_size, self.h), attention[i, :], label=f\"horizon {plot}\")\n", - " ax.axvline(x=0, color='black', linewidth=3, linestyle='--', label=\"prediction start\")\n", - " ax.set_title(f\"Attention weight for horizon {plot}\")\n", - " ax.set_xlabel(\"time\")\n", - " ax.set_ylabel(\"Attention\")\n", - " ax.legend()\n", + " fig, ax = plt.subplots(figsize=(width / 100, height / 100))\n", "\n", - " else:\n", - " raise ValueError('plot has to be in [\"time\",\"all\",\"heatmap\"] or integer in range(1,model.h)')\n", + " if plot == \"time\":\n", + " attention = attention[self.input_size :, :].mean(axis=0)\n", + " ax.plot(np.arange(-self.input_size, self.h), attention)\n", + " ax.axvline(\n", + " x=0, color=\"black\", linewidth=3, linestyle=\"--\", label=\"prediction start\"\n", + " )\n", + " ax.set_title(\"Mean Attention\")\n", + " ax.set_xlabel(\"time\")\n", + " ax.set_ylabel(\"Attention\")\n", + " ax.legend()\n", + "\n", + " elif plot == \"all\":\n", + " for i in range(self.input_size, attention.shape[0]):\n", + " ax.plot(\n", + " np.arange(-self.input_size, self.h),\n", + " attention[i, :],\n", + " label=f\"horizon {i-self.input_size+1}\",\n", + " )\n", + " ax.axvline(\n", + " x=0, color=\"black\", linewidth=3, linestyle=\"--\", label=\"prediction start\"\n", + " )\n", + " ax.set_title(\"Attention per horizon\")\n", + " ax.set_xlabel(\"time\")\n", + " ax.set_ylabel(\"Attention\")\n", + " ax.legend()\n", + "\n", + " elif plot == \"heatmap\":\n", + " cax = ax.imshow(\n", + " attention,\n", + " aspect=\"auto\",\n", + " cmap=\"viridis\",\n", + " extent=[-self.input_size, self.h, -self.input_size, self.h],\n", + " )\n", + " fig.colorbar(cax)\n", + " ax.set_title(\"Attention Heatmap\")\n", + " ax.set_xlabel(\"Attention (current time step)\")\n", + " ax.set_ylabel(\"Attention (previous time step)\")\n", + "\n", + " elif isinstance(plot, int) and (plot in np.arange(1, self.h + 1)):\n", + " i = self.input_size + plot - 1\n", + " ax.plot(\n", + " np.arange(-self.input_size, self.h),\n", + " attention[i, :],\n", + " label=f\"horizon {plot}\",\n", + " )\n", + " ax.axvline(\n", + " x=0, color=\"black\", linewidth=3, linestyle=\"--\", label=\"prediction start\"\n", + " )\n", + " ax.set_title(f\"Attention weight for horizon {plot}\")\n", + " ax.set_xlabel(\"time\")\n", + " ax.set_ylabel(\"Attention\")\n", + " ax.legend()\n", + "\n", + " else:\n", + " raise ValueError(\n", + " 'plot has to be in [\"time\",\"all\",\"heatmap\"] or integer in range(1,model.h)'\n", + " )\n", "\n", - " plt.tight_layout()\n", + " plt.tight_layout()\n", "\n", - " if output == 'plot':\n", - " plt.show()\n", - " elif output == 'figure':\n", - " return fig\n", - " else:\n", - " raise ValueError(f\"Invalid output: {output}. Expected 'plot' or 'figure'.\")" + " if output == \"plot\":\n", + " plt.show()\n", + " elif output == \"figure\":\n", + " return fig\n", + " else:\n", + " raise ValueError(f\"Invalid output: {output}. Expected 'plot' or 'figure'.\")" ] }, { @@ -1248,9 +2004,20 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", + "# | eval: false\n", "plot_attention(nf.models[0], plot=\"time\")" ] }, @@ -1265,9 +2032,20 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", + "# | eval: false\n", "plot_attention(nf.models[0], plot=\"all\")" ] }, @@ -1282,9 +2060,20 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", + "# | eval: false\n", "plot_attention(nf.models[0], plot=8)" ] }, @@ -1300,9 +2089,20 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/plain": [ + "dict_keys(['Past variable importance over time', 'Future variable importance over time', 'Static covariates'])" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "#| eval: false\n", + "# | eval: false\n", "\n", "feature_importances = nf.models[0].feature_importances()\n", "feature_importances.keys()" @@ -1319,10 +2119,31 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/plain": [ + "" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + }, + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "feature_importances['Static covariates'].sort_values(by='importance').plot(kind='barh')" + "# | eval: false\n", + "feature_importances[\"Static covariates\"].sort_values(by=\"importance\").plot(kind=\"barh\")" ] }, { @@ -1336,10 +2157,33 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/plain": [ + "" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + }, + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "feature_importances['Past variable importance over time'].mean().sort_values().plot(kind='barh')" + "# | eval: false\n", + "feature_importances[\"Past variable importance over time\"].mean().sort_values().plot(\n", + " kind=\"barh\"\n", + ")" ] }, { @@ -1353,10 +2197,33 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/plain": [ + "" + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + }, + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "feature_importances['Future variable importance over time'].mean().sort_values().plot(kind='barh')" + "# | eval: false\n", + "feature_importances[\"Future variable importance over time\"].mean().sort_values().plot(\n", + " kind=\"barh\"\n", + ")" ] }, { @@ -1378,18 +2245,29 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "df=feature_importances['Future variable importance over time']\n", + "# | eval: false\n", + "df = feature_importances[\"Future variable importance over time\"]\n", "\n", "\n", "fig, ax = plt.subplots(figsize=(20, 10))\n", "bottom = np.zeros(len(df.index))\n", "for col in df.columns:\n", - " p = ax.bar(np.arange(-len(df),0), df[col].values, 0.6, label=col, bottom=bottom)\n", + " p = ax.bar(np.arange(-len(df), 0), df[col].values, 0.6, label=col, bottom=bottom)\n", " bottom += df[col]\n", - "ax.set_title('Future variable importance over time ponderated by attention')\n", + "ax.set_title(\"Future variable importance over time ponderated by attention\")\n", "ax.set_ylabel(\"Importance\")\n", "ax.set_xlabel(\"Time\")\n", "ax.grid(True)\n", @@ -1415,18 +2293,29 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "df= feature_importances['Past variable importance over time']\n", + "# | eval: false\n", + "df = feature_importances[\"Past variable importance over time\"]\n", "\n", "fig, ax = plt.subplots(figsize=(20, 10))\n", "bottom = np.zeros(len(df.index))\n", "\n", "for col in df.columns:\n", - " p = ax.bar(np.arange(-len(df),0), df[col].values, 0.6, label=col, bottom=bottom)\n", + " p = ax.bar(np.arange(-len(df), 0), df[col].values, 0.6, label=col, bottom=bottom)\n", " bottom += df[col]\n", - "ax.set_title('Past variable importance over time')\n", + "ax.set_title(\"Past variable importance over time\")\n", "ax.set_ylabel(\"Importance\")\n", "ax.set_xlabel(\"Time\")\n", "ax.legend()\n", @@ -1447,25 +2336,48 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ - "#| eval: false\n", - "df= feature_importances['Past variable importance over time']\n", - "mean_attention = nf.models[0].attention_weights()[nf.models[0].input_size:,:].mean(axis=0)[:nf.models[0].input_size]\n", + "# | eval: false\n", + "df = feature_importances[\"Past variable importance over time\"]\n", + "mean_attention = (\n", + " nf.models[0]\n", + " .attention_weights()[nf.models[0].input_size :, :]\n", + " .mean(axis=0)[: nf.models[0].input_size]\n", + ")\n", "df = df.multiply(mean_attention, axis=0)\n", "\n", "fig, ax = plt.subplots(figsize=(20, 10))\n", "bottom = np.zeros(len(df.index))\n", "\n", "for col in df.columns:\n", - " p = ax.bar(np.arange(-len(df),0), df[col].values, 0.6, label=col, bottom=bottom)\n", + " p = ax.bar(np.arange(-len(df), 0), df[col].values, 0.6, label=col, bottom=bottom)\n", " bottom += df[col]\n", - "ax.set_title('Past variable importance over time ponderated by attention')\n", + "ax.set_title(\"Past variable importance over time ponderated by attention\")\n", "ax.set_ylabel(\"Importance\")\n", "ax.set_xlabel(\"Time\")\n", "ax.legend()\n", "ax.grid(True)\n", - "plt.plot(np.arange(-len(df),0), mean_attention, color='black', marker='o', linestyle='-', linewidth=2, label='mean_attention')\n", + "plt.plot(\n", + " np.arange(-len(df), 0),\n", + " mean_attention,\n", + " color=\"black\",\n", + " marker=\"o\",\n", + " linestyle=\"-\",\n", + " linewidth=2,\n", + " label=\"mean_attention\",\n", + ")\n", "plt.legend()\n", "plt.show()" ] @@ -1482,9 +2394,103 @@ "cell_type": "code", "execution_count": null, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
trendy_[lag12]monthobserved_targetCorrelation with Mean Attention
trend1.00-0.45-0.29-0.41-0.43
y_[lag12]-0.451.00-0.56-0.180.68
month-0.29-0.561.000.18-0.38
observed_target-0.41-0.180.181.000.07
Correlation with Mean Attention-0.430.68-0.380.071.00
\n", + "
" + ], + "text/plain": [ + " trend y_[lag12] month observed_target \\\n", + "trend 1.00 -0.45 -0.29 -0.41 \n", + "y_[lag12] -0.45 1.00 -0.56 -0.18 \n", + "month -0.29 -0.56 1.00 0.18 \n", + "observed_target -0.41 -0.18 0.18 1.00 \n", + "Correlation with Mean Attention -0.43 0.68 -0.38 0.07 \n", + "\n", + " Correlation with Mean Attention \n", + "trend -0.43 \n", + "y_[lag12] 0.68 \n", + "month -0.38 \n", + "observed_target 0.07 \n", + "Correlation with Mean Attention 1.00 " + ] + }, + "execution_count": null, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ - "#| eval: false\n", + "# | eval: false\n", "nf.models[0].feature_importance_correlations()" ] } diff --git a/neuralforecast/models/tft.py b/neuralforecast/models/tft.py index f96d5646b..53b0c0cfc 100644 --- a/neuralforecast/models/tft.py +++ b/neuralforecast/models/tft.py @@ -4,16 +4,17 @@ __all__ = ['TFT'] # %% ../../nbs/models.tft.ipynb 5 -from typing import Tuple, Optional, Callable +from typing import Callable, Optional, Tuple +import pandas as pd import torch import torch.nn as nn import torch.nn.functional as F from torch import Tensor from torch.nn import LayerNorm -import pandas as pd -from ..losses.pytorch import MAE + from ..common._base_windows import BaseWindows +from ..losses.pytorch import MAE # %% ../../nbs/models.tft.ipynb 11 def get_activation_fn(activation_str: str) -> Callable: @@ -129,7 +130,6 @@ def _apply_embedding( cont_emb: Tensor, cont_bias: Tensor, ): - if cont is not None: # the line below is equivalent to following einsums # e_cont = torch.einsum('btf,fh->bthf', cont, cont_emb) @@ -270,7 +270,16 @@ def forward( # %% ../../nbs/models.tft.ipynb 19 class StaticCovariateEncoder(nn.Module): - def __init__(self, hidden_size, num_static_vars, dropout, grn_activation): + def __init__( + self, + hidden_size, + num_static_vars, + dropout, + grn_activation, + rnn_type="lstm", + n_rnn_layers=1, + one_rnn_initial_state=False, + ): super().__init__() self.vsn = VariableSelectionNetwork( hidden_size=hidden_size, @@ -278,10 +287,18 @@ def __init__(self, hidden_size, num_static_vars, dropout, grn_activation): dropout=dropout, grn_activation=grn_activation, ) + self.rnn_type = rnn_type.lower() + + self.n_rnn_layers = n_rnn_layers + + self.n_states = 1 if one_rnn_initial_state else n_rnn_layers + + n_contexts = 2 + 2 * self.n_states if rnn_type == "lstm" else 2 + self.n_states + self.context_grns = nn.ModuleList( [ GRN(input_size=hidden_size, hidden_size=hidden_size, dropout=dropout) - for _ in range(4) + for _ in range(n_contexts) ] ) @@ -293,16 +310,62 @@ def forward(self, x: Tensor) -> Tuple[Tensor, Tensor, Tensor, Tensor]: # enrichment context # state_c context # state_h context - cs, ce, ch, cc = tuple(m(variable_ctx) for m in self.context_grns) # type: ignore + + cs, ce = list(m(variable_ctx) for m in self.context_grns[:2]) # type: ignore + + if self.n_states == 1: + ch = torch.cat( + self.n_rnn_layers + * list( + m(variable_ctx).unsqueeze(0) + for m in self.context_grns[2 : self.n_states + 2] + ) + ) + + if self.rnn_type == "lstm": + cc = torch.cat( + self.n_rnn_layers + * list( + m(variable_ctx).unsqueeze(0) + for m in self.context_grns[self.n_states + 2 :] + ) + ) + + else: + ch = torch.cat( + list( + m(variable_ctx).unsqueeze(0) + for m in self.context_grns[2 : self.n_states + 2] + ) + ) + + if self.rnn_type == "lstm": + cc = torch.cat( + list( + m(variable_ctx).unsqueeze(0) + for m in self.context_grns[self.n_states + 2 :] + ) + ) + if self.rnn_type != "lstm": + cc = ch return cs, ce, ch, cc, sparse_weights # type: ignore # %% ../../nbs/models.tft.ipynb 21 class TemporalCovariateEncoder(nn.Module): def __init__( - self, hidden_size, num_historic_vars, num_future_vars, dropout, grn_activation + self, + hidden_size, + num_historic_vars, + num_future_vars, + dropout, + grn_activation, + rnn_type="lstm", + n_rnn_layers=1, ): super(TemporalCovariateEncoder, self).__init__() + self.rnn_type = rnn_type.lower() + self.n_rnn_layers = n_rnn_layers self.history_vsn = VariableSelectionNetwork( hidden_size=hidden_size, @@ -310,9 +373,36 @@ def __init__( dropout=dropout, grn_activation=grn_activation, ) - self.history_encoder = nn.LSTM( - input_size=hidden_size, hidden_size=hidden_size, batch_first=True - ) + if self.rnn_type == "lstm": + self.history_encoder = nn.LSTM( + input_size=hidden_size, + hidden_size=hidden_size, + batch_first=True, + num_layers=n_rnn_layers, + ) + + self.future_encoder = nn.LSTM( + input_size=hidden_size, + hidden_size=hidden_size, + batch_first=True, + num_layers=n_rnn_layers, + ) + + elif self.rnn_type == "gru": + self.history_encoder = nn.GRU( + input_size=hidden_size, + hidden_size=hidden_size, + batch_first=True, + num_layers=n_rnn_layers, + ) + self.future_encoder = nn.GRU( + input_size=hidden_size, + hidden_size=hidden_size, + batch_first=True, + num_layers=n_rnn_layers, + ) + else: + raise ValueError('RNN type should be in ["lstm","gru"] !') self.future_vsn = VariableSelectionNetwork( hidden_size=hidden_size, @@ -320,9 +410,6 @@ def __init__( dropout=dropout, grn_activation=grn_activation, ) - self.future_encoder = nn.LSTM( - input_size=hidden_size, hidden_size=hidden_size, batch_first=True - ) # Shared Gated-Skip Connection self.input_gate = GLU(hidden_size, hidden_size) @@ -333,7 +420,11 @@ def forward(self, historical_inputs, future_inputs, cs, ch, cc): historical_features, history_vsn_sparse_weights = self.history_vsn( historical_inputs, cs ) - history, state = self.history_encoder(historical_features, (ch, cc)) + if self.rnn_type == "lstm": + history, state = self.history_encoder(historical_features, (ch, cc)) + + elif self.rnn_type == "gru": + history, state = self.history_encoder(historical_features, ch) future_features, future_vsn_sparse_weights = self.future_vsn(future_inputs, cs) future, _ = self.future_encoder(future_features, state) @@ -439,6 +530,9 @@ class TFT(BaseWindows): `n_head`: int=4, number of attention heads in temporal fusion decoder.
`attn_dropout`: float (0, 1), dropout of fusion decoder's attention layer.
`grn_activation`: str, activation for the GRN module from ['ReLU', 'Softplus', 'Tanh', 'SELU', 'LeakyReLU', 'Sigmoid', 'ELU', 'GLU'].
+ `rnn_type`: str="LSTM", recurrent neural network (RNN) layer type from ["LSTM","GRU"].
+ `n_rnn_layers`: int=1, number of RNN layers.
+ `one_rnn_initial_state`:str=False, Initialize all rnn layers with the same initial states computed from static covariates.
`loss`: PyTorch module, instantiated train loss class from [losses collection](https://nixtla.github.io/neuralforecast/losses.pytorch.html).
`valid_loss`: PyTorch module=`loss`, instantiated valid loss class from [losses collection](https://nixtla.github.io/neuralforecast/losses.pytorch.html).
`max_steps`: int=1000, maximum number of training steps.
@@ -487,6 +581,9 @@ def __init__( n_head: int = 4, attn_dropout: float = 0.0, grn_activation: str = "ELU", + n_rnn_layers: int = 1, + rnn_type: str = "LSTM", + one_rnn_initial_state: bool = False, dropout: float = 0.1, loss=MAE(), valid_loss=None, @@ -512,7 +609,6 @@ def __init__( dataloader_kwargs=None, **trainer_kwargs, ): - # Inherit BaseWindows class super(TFT, self).__init__( h=h, @@ -550,7 +646,7 @@ def __init__( self.grn_activation = grn_activation futr_exog_size = max(self.futr_exog_size, 1) num_historic_vars = futr_exog_size + self.hist_exog_size + tgt_size - + self.n_rnn_layers = n_rnn_layers # ------------------------------- Encoders -----------------------------# self.embedding = TFTEmbedding( hidden_size=hidden_size, @@ -566,6 +662,9 @@ def __init__( num_static_vars=self.stat_exog_size, dropout=dropout, grn_activation=self.grn_activation, + rnn_type=rnn_type, + n_rnn_layers=n_rnn_layers, + one_rnn_initial_state=one_rnn_initial_state, ) self.temporal_encoder = TemporalCovariateEncoder( @@ -574,6 +673,8 @@ def __init__( num_future_vars=futr_exog_size, dropout=dropout, grn_activation=self.grn_activation, + n_rnn_layers=n_rnn_layers, + rnn_type=rnn_type, ) # ------------------------------ Decoders -----------------------------# @@ -593,7 +694,6 @@ def __init__( ) def forward(self, windows_batch): - # Parsiw windows_batch y_insample = windows_batch["insample_y"][:, :, None] # <- [B,T,1] futr_exog = windows_batch["futr_exog"] @@ -615,17 +715,19 @@ def forward(self, windows_batch): # Static context if s_inp is not None: cs, ce, ch, cc, static_encoder_sparse_weights = self.static_encoder(s_inp) - ch, cc = ch.unsqueeze(0), cc.unsqueeze(0) # LSTM initial states + # ch, cc = ch.unsqueeze(0), cc.unsqueeze(0) # LSTM initial states else: # If None add zeros batch_size, example_length, target_size, hidden_size = t_observed_tgt.shape cs = torch.zeros(size=(batch_size, hidden_size), device=y_insample.device) ce = torch.zeros(size=(batch_size, hidden_size), device=y_insample.device) ch = torch.zeros( - size=(1, batch_size, hidden_size), device=y_insample.device + size=(self.n_rnn_layers, batch_size, hidden_size), + device=y_insample.device, ) cc = torch.zeros( - size=(1, batch_size, hidden_size), device=y_insample.device + size=(self.n_rnn_layers, batch_size, hidden_size), + device=y_insample.device, ) static_encoder_sparse_weights = []