RISED: Rubrics for Agentic Multi-Environment Selection and Self-Distillation

Researchers Jingtan Wang, Sirajul Salekin, Young mok Jung, Javier Movellan, Bryan Kian Hsiang Low and Manjot Bilkhu published RISED, a method that repurposes rubrics — textual descriptions of rollout behaviours tagged by an LLM judge — to guide both online data selection and policy supervision when training a single LLM agent across diverse interactive environments. RISED uses positive rubrics as privileged context for an on-policy self-distillation teacher and negative rubrics to steer subsequent rollout generation away from recurring failure modes, addressing the lack of group-relative reward signal when all-failure and all-success rollout groups coexist in a batch. Across model backbones, RISED achieved the highest mean pass rate across environments and ranked first or second in every individual environment, according to the paper published in October 2026.

content type paper https://machinelearning.apple.com/research/ published October 2026 RISED: Rubrics for Agentic Multi-Environment Selection and Self-Distillation AuthorsJingtan Wang† , Sirajul Salekin, Young mok Jung, Javier Movellan, Bryan Kian Hsiang Low†, Manjot Bilkhu Training a single LLM agent jointly across diverse interactive environments has attracted increasing attention as a route to generalist agents. Existing curriculum and data-selection strategies often allocate training at the environment level or prioritize local reward-based signals, without explicitly considering relationships between current rollouts across environments for prompt-group selection. Meanwhile, as environments are learned at different rates, all-failure and all-success rollout groups can coexist within a batch, leaving those data without group-relative reward signals. Both challenges highlight limitations of relying solely on scalar rewards in multi-environment RL: they provide limited information about cross-environment relationships and no within-group reward contrast when rewards are identical. This motivates richer textual feedback, such as rubrics describing rollout behaviours, to guide learning. Beyond rubrics’ usage as reward, we repurpose rubrics to guide both online data selection and policy supervision. An LLM judge tags each rollout using a predefined rubric vocabulary shared across environments. The resulting profiles guide the selection of data that aligns with the overall behavioural composition of the mixed-environment batch while limiting overlap with already-selected data. Available positive rubrics describing desired behaviours provide privileged context for an on-policy self-distillation teacher, supplying additional token-level supervision, while negative rubrics describing undesired behaviours guide subsequent rollout generation away from recurring failure modes. Together, these components form RISED. Across model backbones, RISED achieves the highest mean pass rate across environments and ranks first or second in every individual environment. Rubric-based analysis of RISED can further characterize the behavioural changes accompanying these gains. From Preferences to Principles: Rubric-Based Alignment for Grounded Knowledge Answers August 27, 2026 research area Methods and Algorithms https://machinelearning.apple.com/research/?domain=Methods%20and%20Algorithms , research area Speech and Natural Language Processing https://machinelearning.apple.com/research/?domain=Speech%20and%20Natural%20Language%20Processing Designing effective reward signals for open-domain question answering is challenging because high-quality responses must simultaneously satisfy multiple aspects of answer quality that are difficult to capture with a holistic scalar objective. We introduce a rubric-based reward framework that generates query-specific rubrics grounded in retrieved evidence and decomposed into multiple quality dimensions, providing fine-grained supervision during… RubiCap: Rubric-Guided Reinforcement Learning for Dense Image Captioning March 16, 2026 research area Computer Vision https://machinelearning.apple.com/research/?domain=Computer%20Vision , research area Data Science and Annotation https://machinelearning.apple.com/research/?domain=Data%20Science%20and%20Annotation Dense image captioning is critical for cross-modal alignment in vision-language pretraining and text-to-image generation, but scaling expert-quality annotations is prohibitively expensive. While synthetic captioning via strong vision-language models VLMs is a practical alternative, supervised distillation often yields limited output diversity and weak generalization. Reinforcement learning RL could overcome these limitations, but its…