Minseong Sim

Interoperable Modular Language Models I: From Decomposition to Interchangeability

> Large language models are usually trained, distributed, and consumed as monolithic systems even though their internal computations are structurally modular. This paper asks a stronger question than model splitting, routing, merging, or adapter composition: can internal language...

Diagnosing Estimator Homomorphism Compatibility in Reinforcement Learning with Verifiable Rewards

Reinforcement learning with verifiable rewards (RLVR) evaluates decoded outcomes, whereas policy-gradient estimators operate on concrete trajectories and batch-dependent training state. These two levels need not respect the same equivalence relation. We formalize estimator homomo...

Diagnosing Estimator Homomorphism Compatibility in Reinforcement Learning with Verifiable Rewards

Reinforcement learning with verifiable rewards (RLVR) evaluates decoded outcomes, whereas policy-gradient estimators operate on concrete trajectories and batch-dependent training state. These two levels need not respect the same equivalence relation. We formalize estimator homomo...