Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

update seaco finetune

维石149063ced4d2
diff --git a/funasr/models/paraformer/model.py b/funasr/models/paraformer/model.py
index d47db1130..6c7957c5f 100644
--- a/funasr/models/paraformer/model.py
+++ b/funasr/models/paraformer/model.py
@@ -181,8 +181,6 @@ class Paraformer(torch.nn.Module):
text: (Batch, Length)
text_lengths: (Batch,)
"""
- # import pdb;
- # pdb.set_trace()
if len(text_lengths.size()) > 1:
text_lengths = text_lengths[:, 0]
if len(speech_lengths.size()) > 1:
@@ -190,7 +188,6 @@ class Paraformer(torch.nn.Module):
batch_size = speech.shape[0]
-
# Encoder
encoder_out, encoder_out_lens = self.encode(speech, speech_lengths)
diff --git a/funasr/models/seaco_paraformer/model.py b/funasr/models/seaco_paraformer/model.py
index 21b6abaec..8f8734025 100644
--- a/funasr/models/seaco_paraformer/model.py
+++ b/funasr/models/seaco_paraformer/model.py
@@ -97,7 +97,8 @@ class SeacoParaformer(BiCifParaformer, Paraformer):
smoothing=seaco_lsm_weight,
normalize_length=seaco_length_normalized_loss,
)
- self.train_decoder = kwargs.get("train_decoder", False)
+ self.train_decoder = kwargs.get("train_decoder", True)
+ self.seaco_weight = kwargs.get("seaco_weight", 0.01)
self.NO_BIAS = kwargs.get("NO_BIAS", 8377)
self.predictor_name = kwargs.get("predictor")
@@ -117,9 +118,10 @@ class SeacoParaformer(BiCifParaformer, Paraformer):
text: (Batch, Length)
text_lengths: (Batch,)
"""
- text_lengths = text_lengths.squeeze()
- speech_lengths = speech_lengths.squeeze()
- assert text_lengths.dim() == 1, text_lengths.shape
+ if len(text_lengths.size()) > 1:
+ text_lengths = text_lengths[:, 0]
+ if len(speech_lengths.size()) > 1:
+ speech_lengths = speech_lengths[:, 0]
# Check that batch_size is unified
assert (
speech.shape[0]
@@ -131,6 +133,8 @@ class SeacoParaformer(BiCifParaformer, Paraformer):
hotword_pad = kwargs.get("hotword_pad")
hotword_lengths = kwargs.get("hotword_lengths")
seaco_label_pad = kwargs.get("seaco_label_pad")
+ if len(hotword_lengths.size()) > 1:
+ hotword_lengths = hotword_lengths[:, 0]
batch_size = speech.shape[0]
# for data-parallel
@@ -156,11 +160,12 @@ class SeacoParaformer(BiCifParaformer, Paraformer):
loss_att, acc_att = self._calc_att_loss(
encoder_out, encoder_out_lens, text, text_lengths
)
- loss = loss_seaco + loss_att
+ loss = loss_seaco + loss_att * self.seaco_weight
stats["loss_att"] = torch.clone(loss_att.detach())
stats["acc_att"] = acc_att
else:
loss = loss_seaco
+
stats["loss_seaco"] = torch.clone(loss_seaco.detach())
stats["loss"] = torch.clone(loss.detach())