Large vision-language models (LVLMs) perform well on long-horizon GUI tasks, but their high inference costs make small VLMs (SVLMs) attractive for cost-effective local deployment despite a substantial capability gap. Group Relative Policy Optimization (GRPO) is a widely used RL algorithm for improving SVLMs, yet its binary outcome rewards provide no intermediate progress signal across trajectories and its uniform advantages lead to semantically-blind credit assignment within each trajectory. We propose PAGE, an RL training method built on automatic progress reward construction and hierarchical advantage computation. The former derives structured evaluation criteria from successful trajectories through iterative extraction and produces stable continuous progress rewards through multi-round aggregation, while the latter combines progress-based inter-rollout advantages with intra-rollout subtask-level advantages obtained through LVLM-driven segmentation and completion verification. Across three long-horizon GUI scenarios, PAGE achieves a 79.3% average success rate with an 8B SVLM, a 16.0 pp gain over the GRPO baseline, and competitive performance against the evaluated closed-source LVLMs.

In the time series domain, increasing efforts extend large language models (LLMs) to multimodal large language models (MLLMs) by incorporating temporal data for downstream temporal understanding tasks. In conventional MLLM training, models primarily rely on language signals, while temporal information provides substantially weaker supervision. This imbalance leads to insufficient learning of temporal patterns. To address this issue, we propose TimeSense, a framework with a disentangled reconstruction structure that introduces explicit temporal supervision and encourages the learning of both global trends and local temporal variations that are often underrepresented in text-dominated multimodal training. Existing QA benchmarks provide limited coverage of temporal variations and compositional temporal dependencies, making it difficult to systematically diagnose these MLLMs capabilities across diverse temporal understanding tasks. To facilitate a more fine-grained evaluation, we introduce EvalTS, a comprehensive benchmark spanning three difficulty levels (Atomic, Molecular, and Compositional) with diverse high-quality queries for evaluating temporal understanding abilities in large language and multimodal language models. Experiments on EvalTS and existing benchmarks demonstrate that TimeSense consistently outperforms prior methods and substantially improves temporal understanding performance.

Large language models (LLMs) possess extensive world knowledge, including geospatial knowledge, which has been successfully applied to various geospatial tasks such as mobility prediction and social indicator prediction. However, LLMs often generate inaccurate geospatial knowledge, leading to geospatial hallucinations—incorrect or inconsistent representations of geospatial information—that compromise their reliability. While the phenomenon of general knowledge hallucination in LLMs has been widely studied, the systematic evaluation and mitigation of geospatial hallucinations remain largely unexplored. To address this gap, we propose a comprehensive evaluation framework for geospatial hallucinations, leveraging structured geospatial knowledge graphs for controlled assessment. Through extensive evaluation across 20 advanced LLMs, we uncover the hallucinations in their geospatial knowledge. Building on these insights, we introduce a dynamic factuality aligning method based on Kahneman-Tversky Optimization (KTO) to mitigate geospatial hallucinations in LLMs, leading to a performance improvement of over 29.6% on the proposed benchmark. Extensive experimental results demonstrate the effectiveness of our benchmark and learning algorithm in enhancing the trustworthiness of LLMs in geospatial knowledge and reasoning tasks.