From 7ea3836893bfdf1aac03952bb1ff2da2c6ef6e57 Mon Sep 17 00:00:00 2001
From: 嘉渊 <wangjiaming.wjm@alibaba-inc.com>
Date: 星期二, 01 八月 2023 14:18:32 +0800
Subject: [PATCH] update
---
funasr/datasets/small_datasets/sequence_iter_factory.py | 6 ++++--
1 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/funasr/datasets/small_datasets/sequence_iter_factory.py b/funasr/datasets/small_datasets/sequence_iter_factory.py
index 44f1db3..8ad72d7 100644
--- a/funasr/datasets/small_datasets/sequence_iter_factory.py
+++ b/funasr/datasets/small_datasets/sequence_iter_factory.py
@@ -57,6 +57,7 @@
data_path_and_name_and_type,
preprocess=preprocess_fn,
dest_sample_rate=dest_sample_rate,
+ speed_perturb=args.speed_perturb if mode=="train" else None,
)
# sampler
@@ -65,8 +66,9 @@
batch_bins=dataset_conf["batch_conf"]["batch_size"] * args.ngpu,
shape_files=shape_files,
sort_in_batch=dataset_conf["sort_in_batch"] if hasattr(dataset_conf, "sort_in_batch") else "descending",
- sort_batch=dataset_conf["sort_batch"] if hasattr(dataset_conf, "sort_batch") else "ascending",
+ sort_batch=dataset_conf["sort_batch"] if hasattr(dataset_conf, "sort_batch") else "descending",
drop_last=False,
+ min_batch_size=torch.distributed.get_world_size(),
padding=True,
)
@@ -83,7 +85,7 @@
args.max_update = len(bs_list) * args.max_epoch
logging.info("Max update: {}".format(args.max_update))
- if args.distributed:
+ if args.distributed and mode=="train":
world_size = torch.distributed.get_world_size()
rank = torch.distributed.get_rank()
for batch in batches:
--
Gitblit v1.9.1