Initialize SenseVoiceEncoderSmall. Args: input_size: Size/dimension parameter. output_size: Size/dimension parameter. attention_heads: TODO. linear_units: TODO. num_blocks: TODO. tp_b
(
self,
input_size: int,
output_size: int = 256,
attention_heads: int = 4,
linear_units: int = 2048,
num_blocks: int = 6,
tp_blocks: int = 0,
dropout_rate: float = 0.1,
positional_dropout_rate: float = 0.1,
attention_dropout_rate: float = 0.0,
stochastic_depth_rate: float = 0.0,
input_layer: Optional[str] = "conv2d",
pos_enc_class=SinusoidalPositionEncoder,
normalize_before: bool = True,
concat_after: bool = False,
positionwise_layer_type: str = "linear",
positionwise_conv_kernel_size: int = 1,
padding_idx: int = -1,
kernel_size: int = 11,
sanm_shfit: int = 0,
selfattention_layer_type: str = "sanm",
**kwargs,
)
| 494 | """ |
| 495 | |
| 496 | def __init__( |
| 497 | self, |
| 498 | input_size: int, |
| 499 | output_size: int = 256, |
| 500 | attention_heads: int = 4, |
| 501 | linear_units: int = 2048, |
| 502 | num_blocks: int = 6, |
| 503 | tp_blocks: int = 0, |
| 504 | dropout_rate: float = 0.1, |
| 505 | positional_dropout_rate: float = 0.1, |
| 506 | attention_dropout_rate: float = 0.0, |
| 507 | stochastic_depth_rate: float = 0.0, |
| 508 | input_layer: Optional[str] = "conv2d", |
| 509 | pos_enc_class=SinusoidalPositionEncoder, |
| 510 | normalize_before: bool = True, |
| 511 | concat_after: bool = False, |
| 512 | positionwise_layer_type: str = "linear", |
| 513 | positionwise_conv_kernel_size: int = 1, |
| 514 | padding_idx: int = -1, |
| 515 | kernel_size: int = 11, |
| 516 | sanm_shfit: int = 0, |
| 517 | selfattention_layer_type: str = "sanm", |
| 518 | **kwargs, |
| 519 | ): |
| 520 | """Initialize SenseVoiceEncoderSmall. |
| 521 | |
| 522 | Args: |
| 523 | input_size: Size/dimension parameter. |
| 524 | output_size: Size/dimension parameter. |
| 525 | attention_heads: TODO. |
| 526 | linear_units: TODO. |
| 527 | num_blocks: TODO. |
| 528 | tp_blocks: TODO. |
| 529 | dropout_rate: TODO. |
| 530 | positional_dropout_rate: TODO. |
| 531 | attention_dropout_rate: TODO. |
| 532 | stochastic_depth_rate: TODO. |
| 533 | input_layer: TODO. |
| 534 | pos_enc_class: TODO. |
| 535 | normalize_before: TODO. |
| 536 | concat_after: TODO. |
| 537 | positionwise_layer_type: TODO. |
| 538 | positionwise_conv_kernel_size: Size/dimension parameter. |
| 539 | padding_idx: TODO. |
| 540 | kernel_size: Size/dimension parameter. |
| 541 | sanm_shfit: TODO. |
| 542 | selfattention_layer_type: TODO. |
| 543 | **kwargs: Additional keyword arguments. |
| 544 | """ |
| 545 | super().__init__() |
| 546 | self._output_size = output_size |
| 547 | |
| 548 | self.embed = SinusoidalPositionEncoder() |
| 549 | |
| 550 | self.normalize_before = normalize_before |
| 551 | |
| 552 | positionwise_layer = PositionwiseFeedForward |
| 553 | positionwise_layer_args = ( |
nothing calls this directly
no test coverage detected