The SDDS (Self Describing Data Sets) Library is a C/C++ library designed for handling structured scientific data efficiently. It provides a flexible data format and a set of utilities for writing, ...
SDPG is the main contribution. It extends GRPO with an exact per-token forward KL between the actor (without privileged context) and itself conditioned on privileged context c: ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results