(
self,
hidden_size,
)
| 18 | """ |
| 19 | |
| 20 | def __init__( |
| 21 | self, |
| 22 | hidden_size, |
| 23 | ): |
| 24 | super(MLP, self).__init__() |
| 25 | self.hidden_size = hidden_size |
| 26 | # Project to 4h. |
| 27 | self.dense_h_to_4h = paddle.nn.Linear( |
| 28 | self.hidden_size, |
| 29 | 4 * self.hidden_size, |
| 30 | ) |
| 31 | |
| 32 | self.activation_func = fast_gelu |
| 33 | |
| 34 | # Project back to h. |
| 35 | self.dense_4h_to_h = paddle.nn.Linear( |
| 36 | 4 * self.hidden_size, |
| 37 | self.hidden_size, |
| 38 | ) |
| 39 | |
| 40 | def forward(self, hidden_states): |
| 41 | # [s, b, 4hp] |