From 358e064fd202dd72e81447ec880dc220439d74c5 Mon Sep 17 00:00:00 2001 From: snarayan21 Date: Wed, 2 Aug 2023 19:19:43 -0700 Subject: [PATCH] Updated StreamingTextDataloader to pass take in shuffle_block_size (#503) --- llmfoundry/data/text_data.py | 1 + 1 file changed, 1 insertion(+) diff --git a/llmfoundry/data/text_data.py b/llmfoundry/data/text_data.py index b3765c25ef..9ed550df41 100644 --- a/llmfoundry/data/text_data.py +++ b/llmfoundry/data/text_data.py @@ -129,6 +129,7 @@ def __init__(self, shuffle=shuffle, shuffle_algo=shuffle_algo, shuffle_seed=shuffle_seed, + shuffle_block_size=shuffle_block_size, ) self.tokenizer = tokenizer self.max_seq_len = max_seq_len