Episode
CoRT: Counterfactual Replay for Token-Level Rubric-Guided Policy Optimization
- Podcast
- Daily Paper Cast
- Published
- Jul 31, 2026
- Duration seconds
- 1206
- Processing state
not_requested- Canonical source
- https://share.transistor.fm/s/7b7b9011
Actions
POST https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/cort-counterfactual-replay-for-token-level-rubric-guided-policy-optimization/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/daily-paper-cast-7079649/cort-counterfactual-replay-for-token-level-rubric-guided-policy-optimization.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
🤗 Upvotes: 78 | cs.AI Authors: Bo-Wen Zhang, Junwei He, Wen Wang, Song-Lin Lv, Wentao Ma, Rongyi Lin, Shuhan Zhong, Lan-Zhe Guo Title: CoRT: Counterfactual Replay for Token-Level Rubric-Guided Policy Optimization Arxiv: http://arxiv.org/abs/2607.25659v1 Abstract: Rubric-based reinforcement learning enriches language model training by evaluating model outputs against explicit criteria. Yet in GRPO-style pipelines, these structured judgments are reduced to a scalar response-level reward and converted into a response-level advantage, which is broadcast uniformly to all generated tokens. This leaves no explicit mechanism for allocating credit within a response, even when different criteria are grounded in different spans, formatting decisions, or semantic choices. We propose CoRT, a token-level credit weighting method for rubric-conditioned GRPO. Instead of training an auxiliary token scoring model, CoRT uses counterfactual replay to rescore the same sampled response under the original rubric-conditioned prompt and a matched criteria-free prompt. The resulting tokenwise log-likelihood contrasts serve as a proxy for dependence on the rubric context. CoRT maps these contrasts to bounded, response-normalized weights and uses them to redistribute the signed GRPO advantage across tokens, without introducing an auxiliary scorer or changing the response-level reward. Experiments across instruction-tuned models and reward granularities show that CoRT improves over matched response-level GRPO in the vast majority of comparisons, with an average gain of 4.4 percentage points. The method remains competitive with learned token-level credit baselines while avoiding a separate relevance-learning stage. These results suggest that policy-internal counterfactual likelihood contrasts provid…