arXiv:2510.12044v2 Announce Type: replace-cross Abstract: Preference tuning typically distributes adapter updates across every Transformer layer. We study how the placement of a fixed LoRA budget affects Direct Preference Optimization (DPO). Hierarchical Alignment partitions deco…
arXiv cs.AI
TIER_1English(EN)·Leon Sick, Lukas Hoyer, Yannick Str\"umpler, Christian Reisswig, Timo Ropinski, Federico Tombari·
arXiv:2610.32770v1 Announce Type: cross Abstract: On-policy self-distillation enables post-training foundation models without relying on stronger external teachers. However, existing approaches rely on ground-truth question-answer pairs, human-curated problem sets, or other super…
arXiv:2601.09373v3 Announce Type: replace Abstract: Do Large Language Models (LLMs) genuinely grasp the compositional semantics of events, or do they rely on surface-level probabilistic heuristics? We investigate the Imperfective Paradox, a logical phenomenon where the past progr…
arXiv:2610.30435v1 Announce Type: new Abstract: Recent advances in search agents have enabled large language models to tackle complex tasks by retrieving and synthesizing external information. However, existing search agents often lack search-state awareness, i.e., the ability to…
arXiv cs.CL
TIER_1English(EN)·Ferdinand Kapl, Kaitlin Maile, Stefan Bauer, Johannes von Oswald·
arXiv:2610.31421v1 Announce Type: new Abstract: Looped and depth-grown language models reuse computations across depth and have attracted interest for their reasoning capabilities. Building on prior suggestive evidence of improved in-context-learning (ICL), we investigate how suc…
arXiv cs.AI
TIER_1English(EN)·Ha Minh Hieu, Tongyao Zhu, Do Xuan Long, Min-Yen Kan·
arXiv:2610.28706v1 Announce Type: new Abstract: LLM-driven evolutionary search is a promising paradigm for scientific discovery, but existing methods remain costly and slow, often applying expensive reasoning uniformly while underusing evidence from failed, redundant, and saturat…
arXiv cs.CL
TIER_1English(EN)·Shuo Cai, Yanggan Gu, Yuhang Liu, Su Lu, Yuanyi Wang, Yang Yu, Zhen Li, Zhitian Hou, Guanghao Zhu, Ming Li, Hongxia Yang·
arXiv:2610.30162v1 Announce Type: new Abstract: Model fusion aims to integrate the capabilities of multi-expert models into a single model. Existing evaluations offer limited coverage of reasoning with long chains of thought and tool use, while strong average performance can mask…
Foundation models can self-improve across text and vision tasks by generating synthetic questions and reasoning traces guided only by high-level topics, without external supervision.
arXiv:2507.17417v4 Announce Type: replace Abstract: For large language models (LLMs), post-training quantization (PTQ) significantly reduces memory and computational overhead. Despite rapid advancements, existing papers often lack experiments on the same ground. In addition, sinc…
arXiv cs.CL
TIER_1English(EN)·Muxuan He, Wang Bill Zhu, Yue Chen, Elsi Kaiser, Weizhi Du, Yao Du, Samuel Nastase, Yanjun Gao·
arXiv:2610.28992v1 Announce Type: new Abstract: Although large language models (LLMs) encode language in highly distributed spaces, they remain vulnerable to focal disruptions---a phenomenon strikingly similar to how localized brain damage causes clinical aphasia in humans. To in…
arXiv:2610.29907v1 Announce Type: new Abstract: Code-switching is a natural linguistic phenomenon, yet its impact on high-level reasoning tasks remains largely unexplored in under-resourced languages like Romanian. Natural language processing tasks, such as satire detection, rely…
arXiv:2610.28500v1 Announce Type: cross Abstract: For efficient deployment of large language models (LLMs), methods of ultra-low-bit backbones reduce weight storage, memory traffic, and linear-layer execution cost, while multi-token prediction (MTP) methods reduce sequential deco…
arXiv:2610.28559v1 Announce Type: cross Abstract: Continuous flow language models offer a path to parallel generation with tools for control and acceleration unavailable to discrete diffusion, yet effective techniques for building them at the billion-parameter scale remain less e…
arXiv:2510.03223v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) tend to generate a long reasoning chain when solving complex tasks. However, as the reasoning chain extends, critical intermediate steps and the original prompt will be buried in the context, r…
A large-scale continuous flow language model achieves efficient training, high-quality generation, and faster inference through blockwise flow design, self-speculative decoding, and multi-token prediction.
LM2LM uses a frozen research world model to provide scalable reinforcement learning rewards for generating language model improvements, maintaining reliability through periodic real execution and outperforming larger baselines at lower cost.
