In the continuous pre-training phase, we introduce a hybrid instruction pre-training mechanism to mitigate the internal-injected knowledge inconsistency of LLMs for medical domain adaptation.

In this work, first, we construct and evaluate a straightforward strategy, "impute-then-detect", via combining state-of-the-art imputation methods with unsupervised anomaly detection methods, where the training data are composed of normal samples only.

Algorithmic indistinguishability yields a natural test for assessing whether experts incorporate this kind of "side

As such, we prevent policies from merely exploiting heuristic rewards without improving the task reward.