Update README.md

README.md

@@ -9,12 +9,12 @@ size_categories:
 - 10K<n<100K
 ---
 
-# Dataset Card for Mini Project Gutenberg Dataset
+# Dataset Card for Mini Project Gutenberg (Cleaned English Subset, Tokenized) Dataset
 
 
 This dataset is a mini subset of the dataset [nikolina-p/gutenberg_flat](nikolina-p/gutenberg_flat), created for **learning, testing streaming datasets, DDP training, and quick experimentation.**
 
-It is made from the first 24 books. Its structure is adapted for training of autoregressive models in distributed environment . 
+It is made from the first 24 books. The text is tokenized using OpenAI's tiktoken tokenizer. Its structure is adapted for training of autoregressive models in distributed environment: each split contains 8 shards, all shards within a split have the same number of tokens, and each row consists of 16×1,024 + 1 tokens. 
 
 
 # Usage