AI训练新招：用“反事实”给每个词打分
训练大模型时，我们常给整段回答一个总分，但好回答里可能只有某句话关键，其他是废话。这篇论文让模型自己对比“有评分标准”和“没评分标准”时，每个词出现的概率变化——如果某个词在没标准时几乎不会出现，那它就是关键，应获得更多训练权重。实验显示，这种“反事实”权重分配让模型在多数任务上平均提升4.4个百分点，且无需额外训练一个打分模型。它不是你明天能用上的，但揭示了更精细的AI训练方向：让模型自己学会区分哪些词真正重要。
📄 原文摘要(英文)
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 provide an effective training signal for within-response credit allocation while retaining the simplicity and stability of GRPO.