LambdaPO: A Lambda Style Policy Optimization for Reasoning Language Models
Authors: Zhe Yuan, Yipeng Zhou, Jinghan Li et al.
Summary
arXiv:2605. 19416v1 Announce Type: new Abstract: Group Relative Policy Optimization(GRPO) has become a cornerstone of modern reinforcement learning alignment, prized for its efficacy in foregoing an explicit value-critic by leveraging reward normalization across sampled trajectory cohorts.
Relevance
Read next because LambdaPO: A Lambda Style Policy Optimization for Reasoning Language Models overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: alignment, line, rate, compare, language, model. Source: arxiv cs.CL (NLP).
Abstract
arXiv:2605.19416v1 Announce Type: new Abstract: Group Relative Policy Optimization(GRPO) has become a cornerstone of modern reinforcement learning alignment, prized for its efficacy in foregoing an explicit value-critic by leveraging reward normalization across sampled trajectory cohorts. However, the method's reliance on a monolithic statistical baseline, such as the group mean, collapses the relational topology of the trajectory space into a single scalar, thereby erasing the fine-grained preference information essential for navigating complex, rank-sensitive reward landscapes. To address this issue, we introduce a novel framework, Lambda Policy Optimization (LambdaPO), that addresses this information-theoretic bottleneck by re-conceptualizing advantage estimation from a scalar value to a decomposed, pairwise preference structure. Specifically, the advantage for any given trajectory is formulated as the integrated sum of reward differentials against all peers in its cohort, where each pairwise comparison is dynamically attenuated by the policy's own probabilistic confidence in the established preference. To further mitigate the sparsity of binary outcome supervision, we augment the objective with a semantic density reward, derived from the precision-recall alignment between generated reasoning traces and ground-truth solutions. As a result, our method can mine more fine-grained optimization signals from a group of rollouts, guiding the LLM to a better optima. Experimental results across challenging math reasoning and question-answering tasks demonstrates that LambdaPO improves performance compared to the baseline methods.