From fce4e1d1b48f23cd8332e60afce3df8d6209a6a7 Mon Sep 17 00:00:00 2001
From: gaochangfeng <54253717+gaochangfeng@users.noreply.github.com>
Date: 星期四, 11 四月 2024 14:59:22 +0800
Subject: [PATCH] SenseVoice对富文本解码的参数 (#1608)

---
 funasr/datasets/large_datasets/datapipes/filter.py |   24 ++++++++++++++++++++++++
 1 files changed, 24 insertions(+), 0 deletions(-)

diff --git a/funasr/datasets/large_datasets/datapipes/filter.py b/funasr/datasets/large_datasets/datapipes/filter.py
new file mode 100644
index 0000000..6fe7153
--- /dev/null
+++ b/funasr/datasets/large_datasets/datapipes/filter.py
@@ -0,0 +1,24 @@
+from torch.utils.data import IterableDataset
+
+def default_fn(data):
+    return data
+
+
+class FilterIterDataPipe(IterableDataset):
+
+    def __init__(self,
+                 datapipe,
+                 fn=default_fn):
+        self.datapipe = datapipe
+        self.fn = fn
+
+    def set_epoch(self, epoch):
+        self.datapipe.set_epoch(epoch)
+
+    def __iter__(self):
+        assert callable(self.fn)
+        for data in self.datapipe:
+            if self.fn(data):
+                yield data
+            else:
+                continue

--
Gitblit v1.9.1