arXiv:2510.25126v2 Announce Type: replace-cross Abstract: Many real-world prediction tasks, particularly those involving entities such as customers or patients, involve both {sequential} and {relational} data. Each entity maintains its own sequence of events while simultaneously engaging in relationships with others. Existing methods in sequence and graph modeling often overlook one modality in favor of the other. We argue that these two facets should instead be integrated and learned jointly. We introduce BRIDGE, a unified end-to-end architecture that couples a sequence model with a graph module under a single objective, allowing gradients to flow across both components to learn task-aligned representations. To enable fine-grained interaction, we propose TOKENXATTN, a token-level cross-attention layer that facilitates message passing between specific events in neighboring sequences. Across two settings, relationship prediction and fraud detection, BRIDGE consistently outperforms static graph models, temporal graph methods, as well as sequence-only baselines on both ranking and classification metrics.

arXiv:2603.08706v1 Announce Type: new Abstract: Training large language models (LLMs) as autonomous agents often begins with imitation learning, but it only teaches agents what to do without understanding why: agents never contrast successful actions against suboptimal alternatives and thus lack awareness of action quality. Recent approaches attempt to address this by introducing self-reflection supervision derived from contrasts between expert and alternative actions. However, the training paradigm fundamentally remains imitation learning: the model imitates pre-constructed reflection text rather than learning to reason autonomously. We propose Agentic Critical Training (ACT), a reinforcement learning paradigm that trains agents to identify the better action among alternatives. By rewarding whether the model's judgment is correct, ACT drives the model to autonomously develop reasoning about action quality, producing genuine self-reflection rather than imitating it. Across three challenging agent benchmarks, ACT consistently improves agent performance when combined with different post-training methods. It achieves an average improvement of 5.07 points over imitation learning and 4.62 points over reinforcement learning. Compared to approaches that inject reflection capability through knowledge distillation, ACT also demonstrates clear advantages, yielding an average improvement of 2.42 points. Moreover, ACT enables strong out-of-distribution generalization on agentic benchmarks and improves performance on general reasoning benchmarks without any reasoning-specific training data, highlighting the value of our method. These results suggest that ACT is a promising path toward developing more reflective and capable LLM agents.