| 36 | |
| 37 | |
| 38 | class PositionalEmbedding(torch.nn.Module): |
| 39 | def __init__(self, hidden_size): |
| 40 | super(PositionalEmbedding, self).__init__() |
| 41 | |
| 42 | self.hidden_size = hidden_size |
| 43 | |
| 44 | inv_freq = 1 / (10000 ** (torch.arange(0.0, hidden_size, 2.0) / hidden_size)) |
| 45 | self.register_buffer('inv_freq', inv_freq) |
| 46 | |
| 47 | def forward(self, pos_seq, bsz=None): |
| 48 | sinusoid_inp = torch.ger(pos_seq, self.inv_freq) |
| 49 | pos_emb = torch.cat([sinusoid_inp.sin(), sinusoid_inp.cos()], dim=-1) |
| 50 | |
| 51 | if bsz is not None: |
| 52 | return pos_emb[None, :, :].expand(bsz, -1, -1) |
| 53 | else: |
| 54 | return pos_emb[None, :, :] |
| 55 | |
| 56 | |
| 57 | class ParallelCrossAttention(torch.nn.Module): |