3 papers
cs.AI2026
The Decomposition Tax: LLM Pipelines Lose Up to 40 Accuracy Points at Their Own Interfaces
Tianqi Bu, YuXuan Peng, Junteng Tu +1
A four-stage LLM pipeline gives up as much as 40.5 accuracy points at its own interfaces (gemma-3-12B on MATH-500, Holm-corrected p = 1.66e-19; the largest tax in the primary famil…
cs.AI2026
Rank Collapse Is Recoverable, Growing Is Not: Out-of-Sample Early Warning for Value Divergence in High-UTD Soft Actor-Critic
Tianqi Bu, YuXuan Peng, Junteng Tu +1
Raising the update-to-data (UTD) ratio breaks off-policy critics in two ways grouped as "plasticity loss": collapsing representations and growing value magnitude . We separate…
cs.AI2026
Re-derivability Decides What a Staged Agent Pipeline Recovers After an Upstream Fault
Tianqi Bu, YuXuan Peng, Junteng Tu +1
One variable sets what an upstream fault costs a staged pipeline of language-model agents: re-derivability, how much of what a stage needs it can rebuild from the original problem.…