speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

auto frontend

Browse Source
This commit is contained in:
游雁 2024-06-07 03:21:53 +08:00
parent a0becf3cd8
commit 0ba1bdd476
1 changed files with 34 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

76
funasr/datasets/openai_datasets/datasets.py
View File

@ -180,51 +180,43 @@ class OpenAIDataset(torch.utils.data.Dataset):
return output return output
def collator(self, samples: list = None): def collator(self, samples: list = None):
outputs = {}
for sample in samples:
if sample is None:
continue
for key in sample.keys():
if key not in outputs:
outputs[key] = []
outputs[key].append(sample[key])
for key, data_list in outputs.items(): for idx in range(self.retry):
if isinstance(data_list[0], torch.Tensor): badcase_flag = False
if data_list[0].dtype == torch.int64 or data_list[0].dtype == torch.int32:
pad_value = self.int_pad_value outputs = {}
else: for sample in samples:
pad_value = self.float_pad_value if sample is None:
continue
for key in sample.keys():
if key not in outputs:
outputs[key] = []
outputs[key].append(sample[key])
outputs[key] = torch.nn.utils.rnn.pad_sequence(
data_list, batch_first=True, padding_value=pad_value
)
if self.batch_type != "example":
for i in range(10):
outputs = self._filter_badcase(outputs, i=i)
return outputs
def _filter_badcase(self, outputs, i=0):
b, t = outputs["input_ids"].shape
if b * t > self.batch_size * 2:
beg = torch.randint(0, 2, ()).item()
if b < 2:
beg = 0
logging.info(
f"Warning, b * t: {b * t} > {self.batch_size}, b: {b}, t: {t}, drop half data {i}th, beg:{beg}"
)
for key, data_list in outputs.items(): for key, data_list in outputs.items():
outputs[key] = outputs[key][beg : beg + b : 2] if isinstance(data_list[0], torch.Tensor):
# if data_list[0].dtype == torch.int64 or data_list[0].dtype == torch.int32:
# speech_lengths_max = outputs["speech_lengths"].max().item()
# outputs["speech"] = outputs["speech"][:, :speech_lengths_max, :] pad_value = self.int_pad_value
# text_lengths_max = outputs["text_lengths"].max().item() else:
# outputs["text"] = outputs["text"][:, :text_lengths_max] pad_value = self.float_pad_value
# target_mask_lengths_max = outputs["target_mask_lengths"].max().item()
# outputs["target_mask"] = outputs["target_mask"][:, :target_mask_lengths_max] outputs[key] = torch.nn.utils.rnn.pad_sequence(
data_list, batch_first=True, padding_value=pad_value
)
if self.batch_type != "example":
b, t = outputs["input_ids"].shape
if b * t > self.batch_size * 2:
beg = torch.randint(0, 2, ()).item()
if b < 2:
beg = 0
logging.info(
f"Warning, b * t: {b * t} > {self.batch_size}, b: {b}, t: {t}, drop half data {idx}th, beg:{beg}"
)
samples = samples[beg : beg + b : 2]
continue
break
return outputs return outputs