From 796300bed045baf4f086c9937ed4f2354fdb1e10 Mon Sep 17 00:00:00 2001 From: Xiaohan Zhang Date: Thu, 23 May 2024 22:31:50 -0700 Subject: [PATCH] Remove prints --- streaming/base/dataset.py | 1 - streaming/base/stream.py | 1 - 2 files changed, 2 deletions(-) diff --git a/streaming/base/dataset.py b/streaming/base/dataset.py index a49099d52..43cd818e0 100644 --- a/streaming/base/dataset.py +++ b/streaming/base/dataset.py @@ -1236,7 +1236,6 @@ def get_item(self, sample_id: int, retry: int = 7) -> Any: raise RuntimeError('Background thread failed. Check other traceback.') # Locate the shard and sample offset within that shard where the sample lives. shard_id, shard_sample_id = self.spanner[sample_id] - #print('I am here 5.2', shard_id, shard_sample_id) shard = self.shards[shard_id] sample = None diff --git a/streaming/base/stream.py b/streaming/base/stream.py index 67cc37582..cad709177 100644 --- a/streaming/base/stream.py +++ b/streaming/base/stream.py @@ -543,7 +543,6 @@ def generate_unique_basename(self, url: str, index: int) -> str: """Generate a unique basename for the file path from the URL.""" hash_object = hashlib.md5(url.encode()) hex_dig = hash_object.hexdigest() - # basename = f"{hex_dig[:3]}/shard.{int(hex_dig, 16) % 100000:05d}.mds" basename = '.'.join(['shard', f'{index:05}', 'mds']) self.url_to_basename[url] = basename self.basename_to_url[basename] = url