mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-09 01:52:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

remove outdated code in train.py (#1096)

Browse Source
This commit is contained in:
Fangjun Kuang 2023-05-25 07:47:38 +08:00 committed by GitHub
parent af8907e1ec
commit 1aeffa73bc
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 0 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
egs/librispeech/ASR/transducer/train.py
View File

@ -627,14 +627,6 @@ def run(rank, world_size, args):
train_cuts = train_cuts.filter(remove_short_and_long_utt) train_cuts = train_cuts.filter(remove_short_and_long_utt)
num_left = len(train_cuts)
num_removed = num_in_total - num_left
removed_percent = num_removed / num_in_total * 100
logging.info(f"Before removing short and long utterances: {num_in_total}")
logging.info(f"After removing short and long utterances: {num_left}")
logging.info(f"Removed {num_removed} utterances ({removed_percent:.5f}%)")
train_dl = librispeech.train_dataloaders(train_cuts) train_dl = librispeech.train_dataloaders(train_cuts)
valid_cuts = librispeech.dev_clean_cuts() valid_cuts = librispeech.dev_clean_cuts()

14
egs/librispeech/ASR/transducer_stateless/train.py
View File

@ -654,20 +654,6 @@ def run(rank, world_size, args):
train_cuts = train_cuts.filter(remove_short_and_long_utt) train_cuts = train_cuts.filter(remove_short_and_long_utt)
try:
num_left = len(train_cuts)
num_removed = num_in_total - num_left
removed_percent = num_removed / num_in_total * 100
logging.info(f"Before removing short and long utterances: {num_in_total}")
logging.info(f"After removing short and long utterances: {num_left}")
logging.info(f"Removed {num_removed} utterances ({removed_percent:.5f}%)")
except TypeError as e:
# You can ignore this error as previous versions of Lhotse work fine
# for the above code. In recent versions of Lhotse, it uses
# lazy filter, producing cutsets that don't have the __len__ method
logging.info(str(e))
train_dl = librispeech.train_dataloaders(train_cuts) train_dl = librispeech.train_dataloaders(train_cuts)
valid_cuts = librispeech.dev_clean_cuts() valid_cuts = librispeech.dev_clean_cuts()

14
egs/librispeech/ASR/transducer_stateless2/train.py
View File

@ -642,20 +642,6 @@ def run(rank, world_size, args):
train_cuts = train_cuts.filter(remove_short_and_long_utt) train_cuts = train_cuts.filter(remove_short_and_long_utt)
try:
num_left = len(train_cuts)
num_removed = num_in_total - num_left
removed_percent = num_removed / num_in_total * 100
logging.info(f"Before removing short and long utterances: {num_in_total}")
logging.info(f"After removing short and long utterances: {num_left}")
logging.info(f"Removed {num_removed} utterances ({removed_percent:.5f}%)")
except TypeError as e:
# You can ignore this error as previous versions of Lhotse work fine
# for the above code. In recent versions of Lhotse, it uses
# lazy filter, producing cutsets that don't have the __len__ method
logging.info(str(e))
train_dl = librispeech.train_dataloaders(train_cuts) train_dl = librispeech.train_dataloaders(train_cuts)
valid_cuts = librispeech.dev_clean_cuts() valid_cuts = librispeech.dev_clean_cuts()