1 paper · 1 filter
Oindrila Saha, Vojtech Krs, Radomir Mech +3
Recent progress in large language models (LLMs) has shown that reasoning improves when intermediate thoughts are externalized into explicit workspaces, such as chain-of-thought tra…