Difference Feedback: Generating Multimodal Process-Level Supervision for VLM Reinforcement Learning

Vision--language models (VLMs) are increasingly aligned via Group Relative Policy Optimization (GRPO)-style training. However, relying solely on terminal outcome rewards yields sparse credit assignment in multi-step reasoning, weakening the linkage between visual evidence and intermediate steps and often causing unstable optimization and visual hallucinations. We propose Differential Feedback, which automatically constructs token/step-level supervision masks by repairing erroneous reasoning trajectories, explicitly marking the key positions that require correction. Without costly large-scale step-by-step human annotations, our method enables process-level visual alignment and can be seamlessly integrated into existing GRPO-like frameworks. Experiments on multimodal reasoning benchmarks including MMMStar and MathVista show an average 3% improvement under matched compute budgets. Our approach offers an effective, low-cost solution for accurate vision--reasoning process alignment.

Multi-Layer GRPO: Enhancing Reasoning and Self-Correction in Large Language Models

The Group Relative Policy Optimization (GRPO) algorithm has demonstrated considerable success in enhancing the reasoning capabilities of large language models (LLMs), as evidenced by DeepSeek-R1. However, the absence of intermediate supervision in GRPO frequently leads to inefficient exploration dynamics. A single error in a complex reasoning chain can invalidate the entire solution, resulting in abrupt reward vanishing and compromising training stability.To address these challenges, we propose MGRPO (Multi-layer GRPO). MGRPO operates in two layers: the first layer employs standard GRPO to generate an initial response. This response, along with the original query, is then fed into a second-layer GRPO process. This second layer is specifically trained to identify and correct errors in the initial response, effectively creating a self-correction loop. This mechanism provides implicit process-level supervision by rewarding successful error correction, without requiring an explicit, densely-annotated reward model. Experimental results on several mathematical reasoning benchmarks demonstrate that MGRPO significantly outperforms standard GRPO, achieving superior performance by fostering both reasoning and self-correction abilities.

Deep Sparse Latent Feature Models for Knowledge Graph Completion

Recent progress in knowledge graph completion (KGC) has focused on text-based approaches to address the challenges of large-scale knowledge graphs (KGs). Despite their achievements, these methods often overlook the intricate interconnections between entities, a key aspect of the underlying topological structure of a KG. Stochastic blockmodels (SBMs), particularly the latent feature relational model (LFRM), offer robust probabilistic frameworks that can dynamically capture latent community structures and enhance link prediction. In this paper, we introduce a novel framework of sparse latent feature models for KGC, optimized through a deep variational autoencoder (VAE). Our approach not only effectively completes missing triples but also provides clear interpretability of the latent structures, leveraging textual information. Comprehensive experiments on the WN18RR, FB15k-237, and Wikidata5M datasets show that our method significantly improves performance by revealing latent communities and producing interpretable representations.