Search papers, labs, and topics across Lattice.
Affiliation:
2
0
4
4
OnPanda is presented, an interactive tool for efficiently annotating LLM alignment data and agent trajectories that adopts token-level correction as its core interaction and releases Panda-CVL, a dataset annotated with onPanda, together with a benchmark for token-level correction.
A brief domain refresh can recover performance in earlier tasks while preserving overall model integrity, challenging the notion of catastrophic forgetting in multi-domain RL.
