Tootfinder

From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring
Yang Li, Qiang Sheng, Yehan Yang, Xueyao Zhang, Juan Cao
https://arxiv.org/abs/2506.09996

From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring
Though safety alignment has been applied to most large language models (LLMs), LLM service providers generally deploy a subsequent moderation as the external safety guardrail in real-world products. Existing moderators mainly practice a conventional full detection, which determines the harmfulness based on the complete LLM output, causing high service latency. Recent works pay more attention to partial detection where moderators oversee the generation midway and early stop the output if harmful…

Independent increments and group sequential tests
Anastasios A. Tsiatis, Marie Davidian
https://arxiv.org/abs/2506.15599 https://arxi…

Independent increments and group sequential tests
Widely used methods and software for group sequential tests of a null hypothesis of no treatment difference that allow for early stopping of a clinical trial depend primarily on the fact that sequentially-computed test statistics have the independent increments property. However, there are many practical situations where the sequentially-computed test statistics do not possess this property. Key examples are in trials where the primary outcome is a time to an event but where the a…

Fairness Dynamics During Training
Krishna Patel, Nivedha Sivakumar, Barry-John Theobald, Luca Zappella, Nicholas Apostoloff
https://arxiv.org/abs/2506.01709

Fairness Dynamics During Training
We investigate fairness dynamics during Large Language Model (LLM) training to enable the diagnoses of biases and mitigations through training interventions like early stopping; we find that biases can emerge suddenly and do not always follow common performance metrics. We introduce two new metrics to evaluate fairness dynamics holistically during model pre-training: Average Rank and Jensen-Shannon Divergence by Parts. These metrics provide insights into the Pythia models' progression of biases…