update

嘉渊

2023-04-24 80aeac6edc7ea1bc0bb0c2303e561931696e749b

update

 egs/aishell/paraformer/run.sh

@@ -3,8 +3,8 @@
. ./path.sh || exit 1;

# machines configuration
CUDA_VISIBLE_DEVICES="0,1"
gpu_num=2
CUDA_VISIBLE_DEVICES="0"
gpu_num=1
count=1
gpu_inference=true  # Whether to perform gpu decoding, set false for cpu decoding
# for gpu decoding, inference_nj=ngpu*njob; for cpu decoding, inference_nj=njob

 funasr/layers/abs_normalize.py

New file
@@ -0,0 +1,14 @@
from abc import ABC
from abc import abstractmethod
from typing import Tuple

import torch


class AbsNormalize(torch.nn.Module, ABC):
    @abstractmethod
    def forward(
        self, input: torch.Tensor, input_lengths: torch.Tensor = None
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        # return output, output_lengths
        raise NotImplementedError