Ideally,languagemodelswould reflect the cultural norms of various regions around the world and generate culturally appropriate content when responding inlocallanguages oftheregions, unless otherwise specified.

LLMs excel at few-shot in-context learning (ICL) - learning from a few input-output examples ("shots") provided in context at inference, without any weight updates.

The wider application of end-to-end learning methods to embodied decision-making domains remains bottlenecked by their reliance on a superabundance of training data representative of the target domain.

Reinforcement learning (RL) is a sequential decision-making problem in which an agent tries to maximize its expected cumulative reward by interacting with an unknown environment over time.

We include all our annotations and extracted landmarks. This ensures that we uphold the highest standards of ethical data usage. In Table A1, we summarize the severity label distribution in Y ouTubePD. We also summarize the demographic distribution in Y ouTubePD, split between PD-positive and healthy control (HC), or PD-negative, subjects. This decision is based on the clinician's suggestion, since an accurate UPDRS facial expression rating would require more This strategy also allows for a finer classification.