HF RL Explorer

Add support for computing DIC

Add support for computing DIC: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). When comparing Bayesian models fit with pymc3, one of the standard summary statistics people reach for is the Deviance Information Criterion (DIC). As far as I can tell there's no built-in way to…

The task

When comparing Bayesian models fit with pymc3, one of the standard summary statistics people reach for is the Deviance Information Criterion (DIC). As far as I can tell there's no built-in way to compute it from a fitted model + trace right now — I end up writing my own helper…

Part of XiaomiMiMo/MiMo-V2.6-RL-oss.