DonorRank: Donor Language Selection for Low-Resource Cross-Lingual Speech RecognitionarXiv:2608.11441v1 Announce Type: new Abstract: Low-resource automatic speech recognition (ASR) commonly relies on cross-lingual transfer, where models are adapted from higher-resource donor languages. However, selectin·Research/深度学习, LLM·arxiv.org ↗arxiv.org ↗
Stigma and Support in Online Sexual Violence Narratives on RedditarXiv:2608.11433v1 Announce Type: new Abstract: Online communities increasingly provide spaces where survivors of sexual violence can share their experiences and seek support. Although prior research has examined stigma·Research/综合科学·arxiv.org ↗arxiv.org ↗
Is Convergence Inevitable? Tracing Output Homogeneity Back to Base ModelsarXiv:2608.11426v1 Announce Type: new Abstract: The lack of diversity in LM content is widely attributed to the alignment process, but how and where exactly in the pipeline this collapse begins is unknown. We argue that·Research/LLM, 深度学习·arxiv.org ↗arxiv.org ↗
Self-Evolving Embodied Agents via Skill-Harness EvolutionarXiv:2608.11350v1 Announce Type: new Abstract: Embodied agents are increasingly built as systems around foundation models, where performance depends not only on model weights but also on the skills, context, action int·Research/智能体·arxiv.org ↗arxiv.org ↗
Gloss-Free Representation Learning for Cross-Dataset Sign SpottingarXiv:2608.11332v1 Announce Type: new Abstract: Sign-language research for resource-constrained languages is often limited by the cost of dense linguistic labels such as glosses, temporal boundaries, and sign order. Bro·Research/综合科学·arxiv.org ↗arxiv.org ↗
TRACE Bench: Task-driven Roleplay Agentic Checklist EvaluationarXiv:2608.11236v1 Announce Type: new Abstract: Roleplay evaluation should do more than assign a single score: it should reveal which role requirements were tested, which failed, and which dialogue evidence supports the·Research/Agent·arxiv.org ↗arxiv.org ↗
Retrofitting Recurrent Depth into a Pretrained Language Model: Installation, Extrapolation, Transfer, and Retention at Two Parameter BudgetsarXiv:2608.11233v1 Announce Type: new Abstract: A dense, pretrained language model can be retrofitted with recurrent depth and learn an iterative latent transition that persists after outcome-only annealing. Qwen2.5-0.5·Research/LLM·arxiv.org ↗arxiv.org ↗
These mammal ancestors birthed live young far earlier than thoughtGrowth rings on a fossil bone suggest at least one mammal ancestor gave birth to live young rather than laying eggs 236 million years ago.·Research/生命科学, 综合科学·sciencenews.org ↗sciencenews.org ↗
Synchronizing Beliefs with Second-Order Theory-of-Mind in Human-Autonomy TeamsarXiv:2608.11229v1 Announce Type: new Abstract: Comparative feedback, asking people which of two behaviors they prefer, has become a standard way to align robot and agent behavior with human intent when the reward itsel·Research/Agent, 智能体·arxiv.org ↗arxiv.org ↗
Forecasting Side Effects of Activation SteeringarXiv:2608.11227v1 Announce Type: new Abstract: Activation steering modifies a language model by adding a learned direction to its hidden activations, enabling targeted behavioral changes without retraining. While effec·Research/LLM·arxiv.org ↗arxiv.org ↗
LLMs in Process Diagram Engineering: From Optimal PFDs to Validated P&IDsarXiv:2608.11220v1 Announce Type: new Abstract: Nowadays, the creation of a process flow diagram (PFD) and its subsequent transformation into a piping and instrumentation diagram (P&ID) is predominantly performed manual·Research/LLM, 深度学习·arxiv.org ↗arxiv.org ↗
From Monolithic to Modular: Segment-level Automatic Prompt OptimizationarXiv:2608.11219v1 Announce Type: new Abstract: Automatic Prompt Optimization (APO) often rewrites prompts monolithically, which can improve one behavior while degrading others. We present SAPO, a segment-level APO meth·Research/LLM, 深度学习·arxiv.org ↗arxiv.org ↗
MaSRead: Content-Addressed Reading of Replicated Latent StoresarXiv:2608.11218v1 Announce Type: new Abstract: Independent agents that reason in latent space can share computed state as key-value cache fragments rather than text. Merged by a conflict-free replicated data type, thes·Research/智能体, 航天·arxiv.org ↗arxiv.org ↗
AutoWorldModel-Bench: A State-Centric Benchmark for Automated World-Model ResearcharXiv:2608.11216v1 Announce Type: new Abstract: World modeling is an unsettled field: architectures, training objectives, and state representations interact in complex ways, and no single recipe dominates across environ·Research/训练优化·arxiv.org ↗arxiv.org ↗
Detecting a Route Flip Is Easier Than Knowing Whether to Fix It: Causal Route-Mediated Damage in Quantized Mixture-of-ExpertsarXiv:2608.11212v1 Announce Type: new Abstract: Top-k Mixture-of-Experts (MoE) routing is discontinuous, so a deployment-motivated numerical disturbance -- simulated 4-bit KV-cache quantization read by a protected BF16·Research/部署·arxiv.org ↗arxiv.org ↗
Distribird: Literature-Informed Prior Distribution Design for Bayesian Model CalibrationarXiv:2608.11210v1 Announce Type: new Abstract: Bayesian calibration of process-based models requires a prior distribution for each model parameter. Despite decades of methodological work, researchers almost always fall·Research/综合科学·arxiv.org ↗arxiv.org ↗
Toddler's tragic death from brain-destroying amoeba offers lessons for doctorsBoy's unusual symptoms could help identify the next case sooner.·Research/临床医学, 流行病学·arstechnica.com ↗arstechnica.com ↗
Have physicists finally discovered glueballs? New evidence points to yes.“It’s the strongest evidence yet that particles dominated by a glueball component can exist in nature.”·Research/综合科学·arstechnica.com ↗arstechnica.com ↗
MIT selected to lead new NSF materials research centerThe Materials Research Science and Engineering Center unites researchers across disciplines to develop technologies for medical imaging, sustainable metals production, and next-generation electronics.·Research/医学影像, 综合科学·news.mit.edu ↗news.mit.edu ↗
Estrogen therapy is linked to fewer signs of Alzheimer’s in the brainEstrogen-only menopause therapy in older women was linked to less Alzheimer’s brain pathology at autopsy but may not prevent the disease.·Research/综合科学·sciencenews.org ↗sciencenews.org ↗