Created
August 9, 2020 23:02
-
-
Save isaacmg/32e13f0bd492c0d15669fa406044f292 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| import torch | |
| class TransformerXCBasic(torch.nn.Module): | |
| """ Transformer model """ | |
| def __init__(self, n_time_series, out_seq_len, device, d_model=128, dropout=.5, n_head=8): | |
| super(TransformerXCBasic, self).__init__() | |
| self.input_dim = n_time_series | |
| self.n_head = n_head | |
| self.seq_num = seq_num | |
| self.n_embd = n_embd | |
| self.win_len = win_len | |
| self.id_embed = torch.nn.Embedding(seq_num, n_embd) | |
| self.po_embed = torch.nn.Embedding(seq_len, n_embd) | |
| self.drop_em = torch.nn.Dropout(dropout) | |
| self.device = device | |
| self.xc_encoder = TransformerEncoderLayer(d_model, n_head) | |
| torch.nn.init.normal_(self.id_embed.weight, std=0.02) | |
| torch.nn.init.normal_(self.po_embed.weight, std=0.02) | |
| def forward(self, series_id, x): | |
| id_embedding = self.id_embed(series_id) | |
| length = x.size(1) # (Batch_size,length,input_dim) | |
| position = torch.tensor(torch.arange(length), dtype=torch.long).to(self.device) | |
| po_embedding = self.po_embed(position) | |
| batch_size = x.size(0) | |
| embedding_sum = torch.zeros(batch_size, length, self.n_embd).to(self.device) | |
| embedding_sum[:] = po_embedding | |
| embedding_sum = embedding_sum + id_embedding.unsqueeze(1) | |
| x = torch.cat((x, embedding_sum), dim=2) | |
| print(x.shape) | |
| return x |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment