Model-guided directed evolution seeks to identify high-fitness protein variants under limited oracle budgets. Protein language models (PLMs) provide rich representations for this task, but existing methods either rely on task-agnostic zero-shot scores or perform search in high-dimensional embedding spaces where surrogate modeling and uncertainty estimation can be sample-inefficient. We propose the Linear Fitness Subspace (LFS) hypothesis: mutation-induced fitness variation encoded by PLMs is concentrated in a low-dimensional linear subspace of residue-level representation changes. This view explains why zero-shot PLM scores are often useful but incomplete: they behave as fixed, task-agnostic projections, whereas few-shot supervision can recover assay-specific fitness directions. Building on this observation, we introduce Subspace-Guided Evolutionary Search (SGES), which learns the LFS from a small initial sample and performs surrogate modeling, uncertainty estimation, and acquisition directly in the learned subspace. Across 10 core ProteinGym assays and an extended set of 87 assays, SGES improves fitness prediction, budgeted search efficiency, computational cost, and biological interpretability over zero-shot PLMs and recent ML-guided protein optimization baselines. These results suggest that PLMs encode protein fitness in a geometry that is more linear, lower-dimensional, and more directly exploitable than previously assumed.

Vision–language models (VLMs) have demonstrated impressive capabilities in general medical visual question answering, yet it remains unclear whether their predictions reflect evidence-grounded clinical reasoning or reliance on spurious priors. We introduce Med-R2 Bench, a hierarchical benchmark aligned with the clinical workflow to evaluate adversarial robustness with visual grounding. We design stepwise QA tasks to assess whether reasoning chains are strictly grounded in visual evidence across the four clinical stages, and employ adversarial perturbations to test robustness against misleading cues. Med-R2 comprises 42,432 images, 31 task categories, and 110,406 QA pairs. Evaluation across 14 VLMs reveals a sequential performance degradation along the four-stage clinical workflow. Adversarial experiments show that models rely heavily on correct prompts to guess answers. Even when provided with explicit visual cues, the models struggle to accurately align textual descriptions. Finally, we demonstrate that stepwise fine-tuning using our hierarchical data significantly improves reasoning robustness, highlighting its potential to drive future improvements in evidence-based medical AI.

Generating long, informative, and factual outputs remains a major challenge for Large Language Models (LLMs). Existing benchmarks for long-form generation typically assess real-world queries with hard-to-verify metrics or use synthetic setups that ease evaluation but overlook real-world intricacies. In this paper, we introduce LongWeave, which balance real-world and verifiable assessment with Target-Anchored Evaluation (TAE). TAE constructs tasks by first defining verifiable targets within real-world scenarios, then systematically generating corresponding queries, textual materials, and anchors based on these targets. This ensures that tasks are both realistic and objectively assessable, enabling rigorous assessment of model capabilities in meeting complex real-world constraints. LongWeave supports customizable input/output lengths (up to 64K/8K tokens) across seven distinct tasks. Evaluation on 23 LLMs show that even state-of-the-art models encounter significant challenges in long-form generation as real-world complexity and output length increase. Dataset will be publicly available.

While Large Language Models (LLMs) support long contexts, they struggle with performance degradation within the context window. Current solutions incur prohibitive training costs, leaving statistical behaviors and cost-effective approaches underexplored. From the decoding perspective, we identify the Posterior Salience Attenuation (PSA) phenomenon, where the salience ratio correlates with long-text performance degradation. Notably, despite the attenuation, gold tokens still occupy high-ranking positions in the decoding space. Motivated by it, we propose the training-free Positional Contrastive Decoding (PCD) that contrasts the logits derived from long-aware attention with those from designed local-aware attention, enabling the model to focus on the gains introduced by large-scale short-to-long training. Through the analysis of long-term decay simulation, we demonstrate that PCD effectively alleviates attention score degradation. Experimental results show that PCD achieves state-of-the-art performance on long-context benchmarks.