Efficiency

A training-free approach that calibrates chain-of-thought reasoning in LLMs, improving accuracy while reducing computational overhead.

Efficient and federated learning for heterogeneous clients with different memory sizes