Jiang-Ming Yang

Models trained with ACA-RL not only outperform on missing-premise tasks but also redefine how we evaluate reasoning under uncertainty in NLP.

Timing the entry of preference dimensions can lead to substantial performance gains in multi-preference alignment for LLMs.

Fair behavior comparisons in agent interactions can dramatically improve performance, reducing response times from nearly 5 seconds to just over 1 second.

Transforming agent failures into actionable recovery strategies, DARC enhances performance without bloating context, proving that less can be more in self-correction.

Explicitly modeling the asymmetry between information growth and action expiration can drastically improve decision-making efficiency in complex environments.

Timeflies reveals that accurately forecasting time series requires not just predicting values, but also understanding whether observations will occur at all.

Achieve state-of-the-art forecasting by mapping heterogeneous time series data into a unified latent space, enabling nuanced interaction modeling.