The paper introduces a unified conditional-flow framework that integrates text-driven motion generation, semantic editing, and intra-structural retargeting into a single rectified-flow model. By treating editing as a change in semantic condition and retargeting as a change in skeletal condition, the approach eliminates fragmented pipelines and allows a single model to perform generation, zero‑shot editing, and zero‑shot retargeting on articulated 3D motion data. Experiments on SnapMoGen and a Mixamo subset demonstrate that the model can handle all three tasks without task‑specific fine‑tuning, preserving both motion semantics and skeletal structure.
By Junlin Li, Xinhao Song, Siqi Wang, Haibin Huang, Yili Zhao
The paper introduces Planning Diffusion Policy Optimization (PDPO), an offline‑to‑online reinforcement‑learning framework that employs a diffusion policy to produce short‑horizon action chunks for robot crowd navigation. PDPO is pretrained on collision‑avoidance demonstrations and fine‑tuned online with PPO, generating five‑step action sequences applied in a receding‑horizon manner. The authors also identify a benchmark artifact where agents can leave the valid domain without explicit boundary constraints, and they mitigate this by treating boundary violations as collisions, leading to improved success rates over strong baselines.
By Wendong Li, Jochen Garcke
The paper introduces an unsupervised domain adaptation framework that aligns redundancy-reducing features to enable accurate 3D segmentation of cone-beam CT (CBCT) without target-domain annotations or inference-time adaptation. The method is architecture-agnostic, working with both CNN-based and ViT-based foundation models, and is evaluated on two liver segmentation benchmarks for interventional vascular procedures and radiation therapy. Results show that even large pretrained segmentation networks need explicit feature-space bridging to generalize across diagnostic CT and CBCT, and the proposed approach consistently outperforms existing pretrained foundation models and UDA strategies.
By Gauthier Miralles, Loic Le Folgoc, Vincent Jugnon, Pietro Gori
The paper introduces GRAS, a method that improves training‑free reward alignment for discrete diffusion models by reducing variance in guided proposals and adapting the resampling temperature during search. It achieves this without adding denoiser cost, using Rao‑Blackwellized estimates for differentiable rewards and a leave‑one‑out baseline for non‑differentiable ones. Experiments on regulatory DNA and protein design show GRAS outperforms existing training‑free techniques and rivals reward‑fine‑tuned models.
By Kwanyoung Kim
SimCast‑S2S is a generative latent‑diffusion model designed for probabilistic subseasonal‑to‑seasonal precipitation forecasting. It tackles three key challenges: it uses a diffusion pipeline to capture uncertainty, operates in a compact latent space to enable efficient large‑ensemble generation, and leverages transfer learning with low‑rank adaptation to train on limited reanalysis data after pretraining on climate simulations. The model outperforms deep‑learning baselines and competes with, or surpasses, operational systems such as the ECMWF‑S2S baseline without requiring extensive post‑processing.
By Hiep V. Dang, Antonios Mamalakis
SIMPLER is a pre‑fine‑tuning method that reduces inference and deployment costs for Earth Observation foundation models by pruning redundant layers. It uses layer‑wise representation similarity on unlabeled task data to identify and remove up to 79% of parameters without requiring gradients, magnitude heuristics, or hyperparameter tuning. Experiments on Prithvi‑EO‑2, TerraMind, and ImageNet‑pretrained ViT‑MAE show that SIMPLER retains 94% of baseline performance while achieving 2.1× faster training and 2.6× faster inference.
By V\'ictor Barreiro, Johannes Jakubik, Francisco Arg\"uello, Dora B. Heras
The paper proposes a scalable, automated method for auditing candidate‑job matching systems for demographic bias. It employs large‑language‑model agents to generate neutral resumes, injects controlled demographic variations, ranks candidates with a fine‑tuned embedding model, and evaluates nine fairness metrics across counterfactual, group‑fairness, and merit‑aware families, producing a composite risk report. Experiments on a small corpus show that single‑score audits miss nuanced issues, underscoring the need for multi‑metric evaluation and LLM‑generated audits as a low‑cost complement to human reviews.
By Sai Yashwant, Shruti Bansal, Anurag Dubey, Samaroha Chatterjee, Satyam Kumar, Shreyash Gupta, Gantala Thulsiram
The paper investigates whether language prompts selected by zero‑shot accuracy remain effective after visual adaptation in source‑free cross‑domain few‑shot learning. Using a paired protocol, the authors compare generic class‑name templates with detailed class descriptions before and after Low‑Rank Adaptation (LoRA) on datasets such as EuroSAT, CropDisease, ISIC, and ChestX. They identify two regimes: semantic saturation, where detailed prompts are already useful before adaptation, and semantic emergence, where detailed prompts become more useful only after visual representation updates, driven by changes in prediction patterns.
By Wei Liu, Xing Deng, Haijian Shao
The paper investigates how transformers acquire deep semantic dependencies, proposing a mechanistic framework that frames learning as a competition between surface statistics and deep semantics. It identifies a "Gradient Starvation" effect that suppresses error signals for sparse semantic dependencies early in training, delaying structural reasoning until a sudden phase transition. The study also explains the success of Chain-of-Thought strategies and introduces a topology‑aligned contrastive objective that improves variable binding performance by more than twice the gain of standard fine‑tuning.
By Jiangrui Zhao, Xiaoting Du
GRAIN is a single-agent reinforcement learning framework that improves large language models’ robustness to real‑world shifts in node identifiers and task formulations by treating reasoning as a semantic parsing and tool‑execution pipeline. It introduces a Structure Invariance Reward that validates intermediate graphs against ground‑truth topologies, encouraging the model to learn genuine text‑to‑structure mappings instead of overfitting to surface patterns. On the new GRIT benchmark, GRAIN surpasses multi‑agent baselines by 16.45% in accuracy, reduces latency by about 24%, and halves the out‑of‑distribution gap of fine‑tuned models while remaining robust on large‑scale graphs beyond the training distribution.
By Zike Yuan, Han Zhang, Jianzhi Yan, Le Liu, Cai Ke, Huozhi Zhou, Jian Xie, Jiran Yin, Yukun Cao, Yue Yu, Hui Wang, Ming Liu, Bing Qin
The paper investigates how scaling the amount of model-generated, off‑task distillation data affects the recoverability of teacher‑induced traits in student models. In a controlled subliminal‑learning setup, teachers are prompted to express a target trait, producing restricted data such as number‑only completions. Students trained on larger independent datasets show a clearer manifestation of the teacher’s trait in a separate evaluation domain, with the effect being strongest when the trait is already favored or when alternative traits are present. The authors find this trend holds across model families, trait types, multi‑trait settings, and cross‑model transfer, and suggest that scaling should be coupled with trait‑aware curation and evaluation.
By Zhichen Dong, Zhixuan Liu, Yuyu Fan, Xiangtian Li, Shuyang Zhang, Chao Yang
The paper evaluates large language models (LLMs) for annotating German parliamentary debates on migration, achieving macro‑F1 scores comparable to human agreement, particularly with GPT‑5 and gpt‑oss‑120B. It combines soft‑label outputs with Design‑based Supervised Learning to mitigate systematic bias and applies the method to a 150‑plus‑year corpus, revealing high solidarity post‑war and a sharp rise in anti‑solidarity since 2015. The study demonstrates that LLMs can enable large‑scale social‑scientific analysis while highlighting the need for rigorous validation and bias correction.
By Aida Kostikova, Ole P\"utz, Steffen Eger, Olga Sabelfeld, Benjamin Paassen
The paper proposes a method called Modular Expert Merging for Biomedical Retrieval, which combines independently trained domain‑specialized experts instead of large mixed‑domain training. Experiments across four decoder‑only LLM families (0.6B‑7B) and twelve retrieval tasks from MTEB show that merging experts consistently outperforms mixed‑domain training. The authors also introduce a Synthesize‑Train‑Merge (STM) framework that generates hard negatives with a top‑tier LLM, fine‑tunes experts via LoRA, and merges them, achieving strong biomedical retrieval performance while retaining competitive general‑domain results.
By Sameh Khattab, Jean-Philippe Corbeil, Osman Alperen \c{C}inar-Kora\c{s}, Amin Dada, Julian Friedrich, Jiawei He, Douglas Teodoro, Jens Kleesiek
RATIO (Retrieval Across Typed Ideation Operations) is a large-scale benchmark designed to evaluate how well retrieval systems can support scientific inspiration. It defines relevance through three ideation moves—Address, Broaden, and Specify—each targeting different levels of abstraction in literature retrieval. The benchmark is built from millions of full-text CS papers using a novel discourse-marker distant supervision method, and includes extensive LLM and human vetting to ensure quality.
By Maayan Sharon, Tom Hope
LLaVAFlow is an information‑theoretic distillation framework designed to preserve cross‑modal alignment in Multimodal Large Language Models during visual instruction tuning. It compresses the mutual information between extracted relations and MLLM embeddings to refine alignment flow, and then maximizes mutual information between pretrained and fine‑tuned alignment flows to transfer compact alignment information. Experiments demonstrate that LLaVAFlow effectively maintains alignment flow, improving downstream performance and generalization.
By Muyao Yuan, Muyan Jiao, Jiangyong Ying, Weizhan Zhang, Yuanhong Zhang, Lan Ma, Yuan Gao, Haipeng Du
The paper introduces FAN-LoRA, a Fourier‑adaptive nonlinear low‑rank adaptor designed to improve domain adaptation of vision foundation models like SAM for medical imaging. By decoupling frequency components into a low‑pass B‑spline branch for global structure and a high‑pass Fourier branch for local texture, FAN‑LoRA addresses performance drops caused by domain shifts. Experiments on three cross‑modality and cross‑center benchmarks show that FAN‑LoRA outperforms existing PEFT methods, achieving higher Dice scores and lower boundary errors while remaining computationally efficient.
By Ziquan Liu, Zhewei Zhu, Xuyang Shi
The authors present the Cross-Platform Fairness Evaluation (CPFE) framework, a five‑axis audit protocol that assesses discriminative performance, calibration, statistical significance, prediction equity, and attribution stability of transformer models. Applying CPFE to four models trained on a Kaggle mental‑health corpus and tested on Reddit and Twitter, they find substantial cross‑platform degradation in AUC (30–40%) and severe calibration failures (ECE rising to 0.5 on Twitter). The study demonstrates that platform‑specific temperature scaling can largely fix calibration without harming discrimination, while prediction equity and attribution stability analyses reveal significant disparities and vocabulary divergence across platforms. The results argue that cross‑platform validation across all CPFE axes should become a standard requirement for mental‑health NLP systems deployed in heterogeneous environments.
By Rajveer Singh Pall, Sameer Yadav
The paper introduces a low‑rank auditing method called LoRA as Oracle, which fits a small adapter to a hypothesis and analyzes the geometry, energy, and alignment of the resulting update relative to frozen weights. This approach directly measures what a model has internalized, independent of its output behavior, enabling detection of backdoors that behavioral audits miss. By identifying and erasing malicious internalizations within the same low‑rank subspace, the method consistently detects target classes across multiple datasets and architectures while preserving clean accuracy and operating at far lower parameter and memory cost than full‑model baselines.
By Marco Arazzi, Antonino Nocera
The paper introduces Echo-GRPO, a method that rewrites privileged reasoning traces into a model’s own idiolect to align off‑policy supervision with the student policy’s vocabulary. By preserving semantics through Dual‑Reference Decoding, Echo‑GRPO mitigates gradient clipping on critical reasoning tokens and improves reasoning distillation. The approach is instantiated as VideoEcho‑R1 for video reasoning, yielding consistent gains across multiple multimodal LLM backbones and benchmarks, and it can be applied as a plug‑in to both RL and supervised fine‑tuning frameworks.
By Ji Soo Lee, Jinyoung Park, Seohyun Lee, Jongha Kim, Joonmyung Choi, Jinsung Yoon, Hyunwoo J. Kim
SWE-Prime introduces a two‑stage, multi‑granularity supervised fine‑tuning (SFT) data selection process for large language models tackling real‑world software problems. The first stage filters entire trajectories based on process quality, result quality, and representativeness, while the second stage evaluates semantic segments for contribution, learnability, and risk, keeping all segments in context but only penalizing selected ones during training. Experiments on SWE‑Bench Pro and Verified demonstrate that training on just 10% of trajectories chosen by SWE‑Prime outperforms full‑dataset training, achieving up to 12.2% and 24.2% relative gains.