We have 3 of 18 papers
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
This work shows that conventional hyperparameter scaling laws are insufficient for ultra-sparse MoEs: the optimal learning rate and batch size vary with activation ratio, and these shifts cannot be explained by either total or activated parameter count alone.
This work proposes SuperValid, a framework that synthesizes OOD, capability-aligned validation data by distilling core concepts from benchmarks within a capability domain and expanding them into diverse, knowledge-rich texts, which enables effective model selection, early stopping, and scaling decisions.
Results establish ELR as a common coordinate linking LR scheduling, norm control, and loss dynamics, and Controlled interventions further show that weight decay and Hyperball shape loss dynamics primarily through the ELR schedules they induce.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
