ProMORNA is a zero‑shot, end‑to‑end generative framework for designing full‑length therapeutic mRNA that balances intracellular stability and translation efficiency. It employs a BART‑style architecture pre‑trained on over 6 million protein‑mRNA pairs, followed by supervised fine‑tuning and multi‑objective reinforcement learning to align the generative policy with conflicting biological constraints. In silico and in‑vitro tests show that ProMORNA outperforms standard baselines and the state‑of‑the‑art GEMORNA, achieving higher predicted stability, translation efficiency, and reporter‑activity AUC in Fluc2P experiments.
By Zixi Shao, Tao Wang, Ye Shen, Tianyi Huang, Yibei Xiao
LocalAgent is a collaborative framework that combines a general multimodal large language model (MLLM) with lightweight specialist verifiers to verify fine‑grained instance‑level consistency between images. The MLLM first identifies suspicious local regions, which are then examined by the Local Evidence Verifier (LEV) toolkit—comprising Local Appearance Consistency (LAC) and Local Geometric Correspondence (LGC)—to provide structured scalar evidence for the final decision. Experiments demonstrate a 4.9% accuracy gain over the strongest baseline while invoking the verifier on only 12.13% of samples, preserving the MLLM’s general multimodal capabilities.
By Jiachen Guo, Xinshan Zhu, Boyun Wang, Yanyan Liang, Jun Wan, Feng Lin, Ajian Liu
CausalAlign is a new framework for diffusion‑based video generation models that tackles the problem of drifting camera ego‑motion over long horizons. It first diagnoses how ego‑motion is represented and changes during denoising, revealing a rise‑and‑fall pattern. Then it strengthens ego‑motion by aligning internal representations to a 3D foundation model in a delta‑token space that captures inter‑frame changes, improving motion accuracy, geometric consistency, and temporal smoothness while preserving visual quality.
By Kai Zhu, Yintianrun Zhang, Shuwei Jiang, Ruolin Zuo, Yixin Chen, Duomin Wang, Miao Liu, Yin Li, Daquan Zhou, Yiwu Zhong
arXiv:2610. 28871v1 Announce Type: new Abstract: Looped transformers reuse one block of layers many times and are often trained by backpropagating through only the last k loops.
By Aur Shalev Merin
arXiv:2610. 28872v1 Announce Type: new Abstract: The weights of a trained network do not have uniform sensitivity to model accuracy: the 1% most sensitive weights carry about a fifth of the damage a bit flip can do.
By Habibur Rahaman, Swarup Bhunia
GaP-MoE introduces a method for explicit expert specialization in Sparse Mixture-of-Experts models by manipulating expert output gradients instead of adding auxiliary loss terms. The technique selects a leader expert from the top‑K routed experts and projects follower experts’ gradients onto the orthogonal complement of the leader’s recent gradients, reducing directional overlap. Experiments on fine‑tuning BERT for GLUE and pretraining GPT‑2 on various corpora show that GaP‑MoE matches or surpasses standard SMoE performance while remaining a drop‑in module.
By Donghyun Jeong, Yoon-Yeong Kim
The paper introduces a transfer learning framework for VLC-based indoor localization that improves cross-area robustness and reduces data needs. Using real measurements from a Bosch facility, fine‑tuning a deep neural network with only 40% of target data recovers 97% of full‑data performance. The authors also propose a Transfer Efficiency Score (TES) to select the best source model, achieving localization errors as low as 40.9 cm and a 3.28 cm mean error on an unseen area.
By Masood Jan, Wafa Njima, Xun Zhang, Alexander Artemenko
The paper introduces HDC‑FL, a personalized federated learning method that clusters clients based on their classifier heads, constructs encoder prototypes for each cluster, and returns a soft mixture of these prototypes to clients while keeping their heads private. The approach requires no additional optimization steps or fine‑tuning and preserves the shared‑encoder trajectory. Experiments on CIFAR‑10 and CIFAR‑100 with ResNet‑18 and MobileNetV2 show that HDC‑FL matches or surpasses state‑of‑the‑art personalized and clustered baselines in macro‑averaged accuracy, worst‑client performance, and bottom‑20% accuracy across various heterogeneity settings.
By Amirardalan Dehghanpour, Arian Noori, Ehsan Tanghatari, Naimeh Omidvar, Hamid R. Rabiee
The paper presents a certified rule that determines whether unseen data should be incorporated into a pretrained model’s context or its weights, based on an exact law for their information value. This law quantifies the code length saved on held‑out queries as the expected bits stored about the query, adjusted by a prior repair and decoding loss, and applies to any model, architecture, dataset, or task. The authors demonstrate that the rule’s ceilings correctly certify choices across diverse models—including transformers, TabPFN, occupancy networks, and stereo networks—and that an anytime‑valid test can reliably select the better approach (e.g., TabPFN over XGBoost or GPT‑2’s cache over fine‑tuning) with controlled error rates.
By Leon Chlon, Julie Huang
Large language models (LLMs) frequently recommend AI-related options, a tendency that is already present in pretrained models (27%) and increases after instruction tuning (59%). Fine‑tuning experiments show that this rise is not inevitable: removing AI‑focused instruction examples reduces the rate, while data rewarding helpfulness or user preferences can raise or lower it. Additionally, how a question is phrased—such as asking for the "best" option or a current recommendation—significantly influences the likelihood of AI suggestions.
By Benaya Trabelsi, Jonathan Shaki, Sarit Kraus
The paper investigates how preference optimization can inadvertently create a shared directional shift—termed a common‑mode offset—in a model’s hidden states. This offset, which affects both chosen and rejected responses, accumulates without being penalized by the ranking loss, eventually projecting onto the output layer to form a vocabulary well that dominates decoding and leads to generation collapse. Experiments across multiple base‑model lineages and fine‑tuning regimes confirm that masking this offset or using per‑token ranking losses can prevent the collapse, while standard ranking metrics fail to detect the risk.
By Yun Li, Ehsan Javanmardi, Simon Thompson, Jiachen Gong, Qunli Zhang, Zifan Zeng, Shiming Liu, Peng Wang, Zixuan Guo, Manabu Tsukada
arXiv:2610.29545v1 Announce Type: new
Abstract: Linear-attention language models keep their context in a recurrent state of fixed size, where an inference system stores one copy of this state for eve...
By Patrick Haller, Filipe Laitenberger, Alan Akbik
arXiv:2505.23323v3 Announce Type: replace
Abstract: This position paper argues that many behaviours framed as either "stochastic parroting" or "advanced reasoning" are better explained as context-dir...
By Harish Tayyar Madabushi, Melissa Torgbi, Claire Bonial
arXiv:2610.28776v1 Announce Type: new
Abstract: Conventional surgical segmentation relies on closed-set categories or explicit text queries, which cannot resolve implicit queries that require reasoni...
By Yiqing Shen, Mathias Unberath
arXiv:2610.28899v1 Announce Type: new
Abstract: Synthesised data is a common strategy to address the lack of annotated handwritten music scores in Optical Music Recognition (OMR). Smashcima is a rece...
By Elona Shatri, Ji\v{r}\'i Mayer, Jan Haji\v{c} jr
arXiv:2610.29442v1 Announce Type: new
Abstract: Recent advances in generative models have enabled highly realistic face forgeries, yet generalizing deepfake detectors to unseen manipulation pipelines...
By Chao Shuai, Shaojing Fan, Kun Pan, Jinwu Hu, Weichen Lian, Ziqi Yang, Zhongjie Ba, Zhenguang Liu
arXiv:2610.29651v1 Announce Type: new
Abstract: Text-to-Image models reproduce societal stereotypes, and existing mitigation applies one demographic correction to the whole scene. In multi-person sce...
By Venkatesh Thirugnana Sambandham, Torsten Schon
The paper investigates how diffusion transformers (DiTs) denoise images by framing them as aggregator–predictor models that use context aggregation and patch prediction. It shows that for Gaussian Markov random field priors, a global Bayes denoiser can be approximated with a limited context, and that patchwise linear denoisers recover empirical local denoisers when aggregation is informative. Experiments confirm that DiTs trained on real images exhibit noise‑dependent denoising contexts and that replacing DiT MLPs with shared‑memory patch predictors yields competitive performance, while cross‑dataset component transplantation supports the proposed functional roles of attention and MLPs.
By Gyeonghun Kang, William Xingxu Chen, Xiang Cheng
The paper introduces LAW (Loopholing All the Way), a continuous generative model that relies entirely on Self‑Conditioning (SC) to propagate expected embeddings across diffusion steps, decoding tokens only once at the end. LAW outperforms both discrete and continuous diffusion models on Sudoku, Maze, and TinyGSM benchmarks, achieving higher accuracy and solving a larger proportion of GSM8K problems. Fine‑tuning a pre‑trained Uniform Diffusion Model with LAW significantly boosts its performance, and on OpenWebText LAW approaches autoregressive models while maintaining a favorable Pareto frontier.
By Adrien Deschenaux, Nadezhda Ilieva, Justin Deschenaux, Manuel Madeira, Dorina Thanou, Aur\'elien Lucchi, Antonio Orvieto, Caglar Gulcehre, Pascal Frossard
arXiv:2610.29110v1 Announce Type: cross
Abstract: Merging different models into a single model is highly desirable for multi-task deployment, particularly in regulated domains like healthcare. In suc...
By Di Wu, Shuaidong Pan