chunked/content_aware_chunking/_generation_strategies/chunk_7.txt

The default decoding strategy is greedy search, which is the simplest decoding strategy that picks a token with the highest probability as the next token. For many tasks
and small output sizes this works well.