Matrix Orthogonalization Improves Memory in Recurrent Modelsayushtambde.com85 ポイント·投稿者 at2005·13 日前·32 コメント
Tree Search Distillation for Language Models Using PPOayushtambde.com87 ポイント·投稿者 at2005·4 か月前·10 コメント
at2005·4 か月前·議論I didn't compare with the harness (focused on distillation) but the original ToT paper has a section on it: https://arxiv.org/abs/2305.10601
at2005·4 か月前·議論Ah, I meant that MCTS uses more inference-time compute (over GRPO) to produce a training sample