# -------------------------------------------------------- # The YiTrans End-to-End Speech Translation System for IWSLT 2022 Offline Shared Task (https://arxiv.org/abs/2206.05777) # Github source: https://github.com/microsoft/SpeechT5/tree/main/YiTrans # Copyright (c) 2022 Microsoft # Licensed under The MIT License [see LICENSE for details] # Based on fairseq code bases # https://github.com/facebookresearch/fairseq # -------------------------------------------------------- from .multihead_attention import MultiheadAttention from .relative_pos_enc import RelativePositionalEncoding from .transformer_layer import TransformerEncoderLayerBase, TransformerDecoderLayerBase from .w2v_encoder import TransformerEncoder, TransformerSentenceEncoderLayer from .learned_positional_embedding import LearnedPositionalEmbedding __all__ = [ "MultiheadAttention", "RelativePositionalEncoding", "TransformerEncoderLayerBase", "TransformerDecoderLayerBase", "TransformerEncoder", "TransformerSentenceEncoderLayer" ]