SpeechT5/SpeechLM/speechlm/modules/__init__.py

# --------------------------------------------------------
# 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"
]