From ed1561ab9625d5e6dadcb77ec962b77a5c278d1a Mon Sep 17 00:00:00 2001 From: hyj Date: Tue, 28 Oct 2025 17:55:36 +0900 Subject: [PATCH] modi !!! --- preprocess.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/preprocess.py b/preprocess.py index 4c3b06e..5f07da2 100644 --- a/preprocess.py +++ b/preprocess.py @@ -11,7 +11,8 @@ MAX_INTERNAL_PATHS = 0 MAX_PATH_LENGTH = 0 MAX_RELATIVE_PATH_LENGTH = 0 -MAX_EXAMPLES_IN_SHARD = 500000 +#MAX_EXAMPLES_IN_SHARD = 500000 +MAX_EXAMPLES_IN_SHARD = 100000 # 수정함 def save_dictionaries(dataset_name, subtok_to_count, node_to_count, max_contexts, max_internal_paths, max_path_length, max_path_width, max_relative_path_length, max_child_id):