Policy optimization (PO) for Large Language Models faces a stability–exploration trade-off, currently mediated by an action-side Policy-KL regularizer. This puts practitioners in a double bind: keeping Policy-KL constrains response behavior and consumes the action-side exploration budget, while dropping it leaves the optimization without an explicit drift control. We argue for an alternative that breaks the dilemma by moving regularization to the input side. As training progresses, the distribution over training queries induced by the current policy drifts unchecked from its pre-RL reference distribution.

Concretely, Environment-Regularized Policy Optimization (ERPO) introduces a Query-KL (QKL) term that bounds this query distribution shift, together with a dataset-static reference-derived per-query weight that biases each per-query update toward queries typical under the reference. The QKL gradient flows strictly through the query likelihood; the response score function used by policy-gradient estimators does not appear in the QKL term, so QKL exerts no direct gradient pressure on the response distribution—exploration is preserved. ERPO plugs into GRPO/PPO/REINFORCE-style pipelines without additional forward passes. On six mathematical reasoning benchmarks, ERPO replaces the standard Policy-KL regularizer while achieving effective control over query distribution drift, delivering stronger accuracy and substantially more stable behavior under high-temperature decoding and long-horizon training.

Zero-resource cross-lingual named entity recognition (ZRCL-NER) aims to leverage rich labeled source language data to address the NER problem in the zero-resource target language. Existing methods are built either based on data transfer or representation transfer. However, the former usually leads to additional computation costs, and the latter lacks explicit optimization specific to the NER task. To overcome the above limitations, we propose a novel prototype-based representation alignment model (PRAM) for the challenging ZRCL-NER task. PRAM models the cross-lingual (CL) NER task and transfers knowledge from source languages to target languages in a unified neural network, and performs end-to-end training, avoiding additional computation costs. Moreover, PRAM borrows the CL inference ability of multilingual language models and enhances it with a novel training objective—attribution-prediction consistency (APC)—for explicitly enforcing the entity-level alignment between entity representations and predictions, as well as that across languages using prototypes as bridges. The experimental results show that PRAM significantly outperforms existing state-of-the-art methods, especially in some challenging scenarios.

Distance metric learning has become a popular solution for few-shot Named Entity Recognition (NER). The typical setup aims to learn a similarity metric for measuring the semantic similarity between test samples and referents, where each referent represents an entity class. The effect of this setup may, however, be compromised for two reasons. First, there is typically a limited optimization exerted on the representations of entity tokens after initing by pre-trained language models. Second, the referents may be far from representing corresponding entity classes due to the label scarcity in the few-shot setting. To address these challenges, we propose a novel approach named COntrastive learning with Prompt guiding for few-shot NER (COPNER). We introduce a novel prompt composed of class-specific words to COPNER to serve as 1) supervision signals for conducting contrastive learning to optimize token representations; 2) metric referents for distance-metric inference on test samples. Experimental results demonstrate that COPNER outperforms state-of-the-art models with a significant margin in most cases. Moreover, COPNER shows great potential in the zero-shot setting.