arXiv:2607.06648v3 cs.LG

Final Checkpoints Are Not Enough: Analyzing Latent Reasoning Faithfulness Along Training Trajectories

Hengyu Jin, Shu Yang, Di Wang

제출 2026-11-05

원문 초록

Latent reasoning performs multi-step inference in continuous hidden states, promising more compact and efficient reasoning. However, these opaque states raise a question of faithfulness: whether the latent reasoning steps drive the final answer. Prior work reports unfaithful behaviors at selected checkpoints, leaving their development during training unexamined. We track behavioral and activation-based evidence across training using verified counterfactual edits and interventions on the latent reasoning states. We find that accuracy can improve while models become less responsive to edits that change the required reasoning path and the correct answer. This counterfactual responsiveness follows distinct trajectories across latent reasoning approaches. On the graph reachability task ProsQA, output sensitivity to noise replacement declines alongside counterfactual responsiveness. Across separately trained binary-choice and open-ended versions of GSM8K math problems, interventions on all latent states show opposite sensitivity trajectories. Guided by these trajectories, we intervene on the training procedure and improve counterfactual responsiveness. These results show that evaluating only a final checkpoint can obscure both when counterfactual responsiveness changes and what the latent states contribute.

제출 이력

확인된 버전만 표시합니다. 전체 제출 이력은 arXiv 원문에서 확인하세요.

  1. arXiv:2607.06648v3제출 시각 미확인

함께 살펴볼 논문

공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.

공통 태그로 연결된 공개 논문이 아직 없습니다.

토론

이 논문과 연결된 포럼 글이 아직 없습니다.