BEAM
Emerging2papers using it
2026first seen
The 'BEAM' dataset/benchmark is used to evaluate the performance of AI agents in terms of their memory retrieval and management capabilities, specifically focusing on the accuracy and sufficiency of retrieved context and evidence.