Multimodal Large Language Models (MLLMs) excel in general tasks but struggle with specialized, structured cultural symbols. We introduce BoYaEval, the first comprehensive benchmark dedicated to deciphering diverse Ancient Chinese musical notations, including five types of ancient Chinese music notation systems. These systems utilize unique spatial layouts and specialized ideograms to encode pitch and intricate playing techniques. BoYaEval comprises 3,175 high-quality images across these notation styles and establishes a three-tier evaluation: Structural Parsing (symbol recognition), Instructional Translation (technique mapping), and Musical Reasoning (melody derivation). We evaluate 21 leading MLLMs. Results indicate that while models perform adequately in basic recognition, they fail in cross-system compositional logic, scoring only around 27% on reasoning tasks. BoYaEval highlights the limitations of current MLLMs in processing diverse spatial-symbolic dependencies, bridging the gap between ancient wisdom and modern AI for digitizing intangible cultural heritage. The BoYaEval benchmark is publicly available at https://huggingface.co/datasets/MYTH-Lab/BoYaEval.

Early Long-context Document Visual Question Answering (DocVQA) methods struggle with preserving visual semantics or handling finite context windows. Conversely, recent RAG-based approaches suffer from "semantic gaps" and "structural disconnections" due to passive retrieval mechanisms that ignore logical dependencies. To address these challenges, we introduce TRACE (Traversal Retrieval-Augmented Chain of Evidence). By navigating a Bi-Layered Graph that encodes both physical adjacency and semantic relevance, TRACE transforms retrieval from static matching into adaptive evidence chain construction. Furthermore, we propose M5BookVQA, a benchmark designed to assess deep, multi-hop reasoning in books, addressing the limitations of existing datasets. Extensive experiments show that TRACE achieves an average accuracy improvement of 14.07% on M5BookVQA and exhibits robust generalization with a 13.38% gain across four established benchmarks. Our source code is available at https://github.com/shimurenhlq/TRACE.

Autoregressive decoding in Large Language Models (LLMs) generates one token per step, causing high inference latency. Speculative decoding (SD) mitigates this through a guess-and-verify strategy, but existing training-free variants face trade-offs: retrieval-based drafts break when no exact match exists, while logits-based drafts lack structural guidance. We propose RACER (Retrieval-Augmented Contextual Rapid Speculative Decoding), a lightweight and training-free method that integrates retrieved exact patterns with logit-driven future cues. This unification supplies both reliable anchors and flexible extrapolation, yielding richer speculative drafts. Experiments on Spec-Bench, HumanEval, and MGSM-ZH demonstrate that RACER consistently accelerates inference, achieving more than 2× speedup over autoregressive decoding, and outperforms prior training-free methods, offering a scalable, plug-and-play solution for efficient LLM decoding. Our source code is available at https://github.com/hkr04/RACER.

In fine-grained sparse Mixture-of-Experts (MoE) models, a large pool of specialized experts replaces a small homogeneous set, shifting performance and throughput to be governed by inference-time expert activation. Yet most existing optimization recipes implicitly assume a fixed activation budget (e.g., a constant Top-k per layer), whose behavior in fine-grained MoEs is poorly understood. We first characterize runtime skipping strategies, quantifying the accuracy–efficiency trade-off of (i) uniform fixed activation and (ii) static layer-wise Top-k allocation found by search. Our analysis reveals that static skipping can already provide substantial throughput gains, but optimal static schedules vary significantly across models and routing mechanisms. We therefore introduce Adaptive Skipping with Entropy-Penalized Thresholding (ASET), a training-free policy that adapts token-level activation using router confidence and entropy while remaining within the model’s original budget. Across the fine-grained MoEs we study, static skipping policies yield 10–78% throughput gains with minimal performance degradation, including ≥10% improvement on DeepSeek-V3 without measurable loss. On the OLMoE testbed, ASET yields a Pareto frontier between average activation and task quality. Overall, these results identify expert skipping as a practical lever for faster fine-grained MoE inference, with adaptive activation helping when fixed budgets are too rigid.

Long-video understanding is bottlenecked by the high cost of processing massive visual tokens. Current reduction strategies often rely on static allocation or inefficient in-network selection that disrupts optimized attention kernels. In this paper, we introduce Vista-LLM, a decoupled framework for query-guided visual token pruning. By filtering redundancy prior to inference with minimal overhead, Vista-LLM ensures full compatibility with Flash Attention. Our method employs a coarse-to-fine pipeline: (1) Query-Guided Dynamic Budgeting for adaptive temporal allocation; (2) a lightweight Semantic Scout for fine-grained, query-specific selection; and (3) Structure-Aware Compensation to preserve global context. Extensive experiments on benchmarks like Video-MME and MLVU demonstrate a significantly improved Pareto frontier. Notably, on LLaVA-OneVision, Vista-LLM reduces visual tokens by 90% and accelerates inference while retaining over 98% of baseline performance on average, effectively filtering visual noise.