From 4ba1011b42e041ee1d71448eefd7ef2e7bd61bb6 Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期五, 31 三月 2023 15:31:26 +0800
Subject: [PATCH] export
---
funasr/export/models/encoder/sanm_encoder.py | 26 ++++----------------------
1 files changed, 4 insertions(+), 22 deletions(-)
diff --git a/funasr/export/models/encoder/sanm_encoder.py b/funasr/export/models/encoder/sanm_encoder.py
index 5437440..44a48ff 100644
--- a/funasr/export/models/encoder/sanm_encoder.py
+++ b/funasr/export/models/encoder/sanm_encoder.py
@@ -9,20 +9,6 @@
from funasr.modules.positionwise_feed_forward import PositionwiseFeedForward
from funasr.export.models.modules.feedforward import PositionwiseFeedForward as PositionwiseFeedForward_export
-def subsequent_mask(size, device="cpu", dtype=torch.bool):
- """Create mask for subsequent steps (size, size).
-
- :param int size: size of mask
- :param str device: "cpu" or "cuda" or torch.Tensor.device
- :param torch.dtype dtype: result dtype
- :rtype: torch.Tensor
- >>> subsequent_mask(3)
- [[1, 0, 0],
- [1, 1, 0],
- [1, 1, 1]]
- """
- ret = torch.ones(size, size, device=device, dtype=dtype)
- return torch.tril(ret, out=ret)
class SANMEncoder(nn.Module):
def __init__(
@@ -165,12 +151,7 @@
def prepare_mask(self, mask, sub_masks):
mask_3d_btd = mask[:, :, None]
- # sub_masks = subsequent_mask(mask.size(-1)).type(torch.float32)
- if len(mask.shape) == 2:
- mask_4d_bhlt = 1 - sub_masks[:, None, None, :]
- elif len(mask.shape) == 3:
- mask_4d_bhlt = 1 - sub_masks[:, None, :]
- mask_4d_bhlt = mask_4d_bhlt * -10000.0
+ mask_4d_bhlt = (1 - sub_masks) * -10000.0
return mask_3d_btd, mask_4d_bhlt
@@ -182,7 +163,8 @@
):
speech = speech * self._output_size ** 0.5
mask = self.make_pad_mask(speech_lengths)
- mask = self.prepare_mask(mask)
+ mask = self.prepare_mask(mask, sub_masks)
+ vad_mask = self.prepare_mask(mask, vad_mask)
if self.embed is None:
xs_pad = speech
else:
@@ -194,7 +176,7 @@
# encoder_outs = self.model.encoders(xs_pad, mask)
for layer_idx, encoder_layer in enumerate(self.model.encoders):
if layer_idx == len(self.model.encoders) - 1:
- mask = (mask[0], vad_mask)
+ mask = vad_mask
encoder_outs = encoder_layer(xs_pad, mask)
xs_pad, masks = encoder_outs[0], encoder_outs[1]
--
Gitblit v1.9.1