paddlespeech.s2t.modules 包

子模块

• paddlespeech.s2t.modules.activation 模块
  • ConvGLUBlock
    • ConvGLUBlock.forward()
  • GLU
    • GLU.forward()
  • LinearGLUBlock
    • LinearGLUBlock.forward()
  • brelu()
  • get_activation()
• paddlespeech.s2t.modules.align 模块
  • BatchNorm1D
  • Conv1D
  • Conv2D
  • 嵌入
  • LayerNorm
  • 线性
• paddlespeech.s2t.modules.attention 模块
  • MultiHeadedAttention
    • MultiHeadedAttention.forward()
    • MultiHeadedAttention.forward_attention()
    • MultiHeadedAttention.forward_qkv()
  • RelPositionMultiHeadedAttention
    • RelPositionMultiHeadedAttention.forward()
    • RelPositionMultiHeadedAttention.init_weights()
    • RelPositionMultiHeadedAttention.rel_shift()
  • RoPERelPositionMultiHeadedAttention
    • RoPERelPositionMultiHeadedAttention.align()
    • RoPERelPositionMultiHeadedAttention.apply_rotary_position_embeddings()
    • RoPERelPositionMultiHeadedAttention.forward()
• paddlespeech.s2t.modules.cmvn 模块
  • GlobalCMVN
    • GlobalCMVN.forward()
• paddlespeech.s2t.modules.conformer_convolution 模块
  • ConvolutionModule
    • ConvolutionModule.forward()
    • ConvolutionModule.init_weights()
• paddlespeech.s2t.modules.crf 模块
  • CRF
    • CRF.decode()
    • CRF.forward()
    • CRF.init_weights()
    • CRF.log_likelihood()
• paddlespeech.s2t.modules.ctc 模块
  • CTCDecoder
    • CTCDecoder.decode()
    • CTCDecoder.decode_probs_offline()
    • CTCDecoder.del_decoder()
    • CTCDecoder.get_decoder()
    • CTCDecoder.init_decoder()
    • CTCDecoder.next()
    • CTCDecoder.reset_decoder()
• paddlespeech.s2t.modules.decoder 模块
  • TransformerDecoder
    • TransformerDecoder.batch_score()
    • TransformerDecoder.forward()
    • TransformerDecoder.forward_one_step()
    • TransformerDecoder.score()
• paddlespeech.s2t.modules.decoder_layer 模块
  • DecoderLayer
    • 解码器层.forward()
• paddlespeech.s2t.modules.embedding 模块
  • NoPositionalEncoding
    • NoPositionalEncoding.forward()
    • NoPositionalEncoding.position_encoding()
  • PositionalEncoding
    • 位置编码.forward()
    • PositionalEncoding.position_encoding()
  • PositionalEncodingInterface
    • PositionalEncodingInterface.forward()
    • PositionalEncodingInterface.position_encoding()
  • RelPositionalEncoding
    • RelPositionalEncoding.forward()
• paddlespeech.s2t.modules.encoder 模块
  • BaseEncoder
    • BaseEncoder.forward()
    • BaseEncoder.forward_chunk()
    • BaseEncoder.forward_chunk_by_chunk()
    • BaseEncoder.output_size()
  • ConformerEncoder
  • SqueezeformerEncoder
    • SqueezeformerEncoder.calculate_downsampling_factor()
    • SqueezeformerEncoder.check_ascending_list()
    • SqueezeformerEncoder.forward()
    • SqueezeformerEncoder.forward_chunk()
    • SqueezeformerEncoder.output_size()
  • TransformerEncoder
    • TransformerEncoder.forward_one_step()
• paddlespeech.s2t.modules.encoder_layer 模块
  • ConformerEncoderLayer
    • ConformerEncoderLayer.forward()
  • SqueezeformerEncoderLayer
    • SqueezeformerEncoderLayer.forward()
  • TransformerEncoderLayer
    • TransformerEncoderLayer.forward()
• paddlespeech.s2t.modules.initializer 模块
  • DefaultInitializerContext
• paddlespeech.s2t.modules.loss 模块
  • CTCLoss
    • CTCLoss.forward()
  • 标签平滑损失
    • 标签平滑损失.forward()
• paddlespeech.s2t.modules.mask 模块
  • add_optional_chunk_mask()
  • make_non_pad_mask()
  • make_pad_mask()
  • make_xs_mask()
  • mask_finished_preds()
  • mask_finished_scores()
  • subsequent_chunk_mask()
  • subsequent_mask()
• paddlespeech.s2t.modules.positionwise_feed_forward 模块
  • PositionwiseFeedForward
    • PositionwiseFeedForward.forward()
    • PositionwiseFeedForward.init_weights()
• paddlespeech.s2t.modules.subsampling 模块
  • Conv2dSubsampling4
    • Conv2dSubsampling4.forward()
  • Conv2dSubsampling6
    • Conv2dSubsampling6.forward()
  • Conv2dSubsampling8
    • Conv2dSubsampling8.forward()
  • DepthwiseConv2DSubsampling4
    • DepthwiseConv2DSubsampling4.forward()
  • LinearNoSubsampling
    • LinearNoSubsampling.forward()