Prépublication
· 2026 arXiv
EN
Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding the capacity of large language models (LLMs) with limited additional computation. Beyond model scaling, this architecture has demonstrated the p...
Hongru Cai, Ran Wei et al.
#cs.CL
Prépublication
· 2026 arXiv
EN
Vision-language-action models (VLAs) are strikingly sensitive to instruction phrasing and do not inherit the language robustness of the vision-language models they are built on. A one-word edit can move success by tens of points: $π_{0.5}$ turns on a LIBERO st...
Mikey Watts, Yuchen Cui
#cs.RO
Prépublication
· 2026 arXiv
EN
Prompted embedding models have recently received increasing attention, particularly for retrieval, where detailed retrieval instructions are provided as part of the retrieval prompt. Several new datasets and studies have examined this setting, showing that the...
Amanda Myntti, Jenna Kanerva et al.
#cs.CL
Prépublication
· 2026 arXiv
EN
Large language models are increasingly applied to tasks grounded in long, heterogeneous information sources. Conventional Retrieval-Augmented Generation (RAG) relies on fixed similarity-based retrieval, while agentic variants adapt queries and tool use but rem...
Yilun Hao, Krishna Sayana et al.
#cs.AI
Prépublication
· 2026 arXiv
EN
Many of the questions now put to large language models have no correct answer to score against: what a policy is worth, which option a user should choose, how to weigh competing values. Stated-preference economics has faced this problem for decades. It judges...
Daniel Robert Kling Alexander, Catherine Louise Kling
#cs.AI
Prépublication
· 2026 arXiv
EN
Robot foundation models provide strong visuomotor control, yet their performance can degrade when object positions or task instructions change. Further improvements often require post-training on substantial robot data, which can be costly to collect through m...
Python Song, Zhixuan Liang et al.
#cs.AI
Prépublication
· 2026 arXiv
EN
How can we predict which base checkpoint is worth an expensive round of agentic post-training? End-to-end pass@$K$ tests whether successful behavior already appears in a base model's distribution, but it is a poor fit for agentic coding: many base checkpoints...
Tan Yu, Alexander Bukharin et al.
#cs.AI
Prépublication
· 2026 arXiv
EN
We present FoldBack, a self-correcting masked generative policy for long-horizon garment folding. Existing long-trajectory policies may continue after a missed or slipped grasp even when the garment has not reached the intended configuration. We structure Fold...
Lipeng Zhuang, Shiyu Fan et al.
#cs.RO
Prépublication
· 2026 arXiv
EN
Multi-teacher on-policy distillation (MOPD) is used in two settings. In common-domain composition, several teachers score each student rollout from one prompt domain and their signals form a single target; in routed-domain distillation, prompts from different...
Hejian Sang, Zhengze Zhou et al.
#cs.LG
Prépublication
· 2026 arXiv
EN
Hallucinated information can propagate through multi-stage LLM systems and become part of the context for subsequent reasoning. Existing studies of post-hallucination reasoning (PHR) mainly characterize changes in final outcomes and aggregate reasoning dynamic...
Linghao Meng, Feng He et al.
#cs.CL
Prépublication
· 2026 arXiv
EN
Reinforcement Learning (RL) from outcome rewards suffers from sparse supervision, particularly on difficult, long-horizon tasks where successful trajectories are rare and costly to generate. On-Policy Distillation (OPD) offers an attractive alternative by prov...
Randy Ardywibowo, Arnav Dalal et al.
#cs.LG
Prépublication
· 2026 arXiv
EN
Flow policies capture rich and diverse action distributions, and fine-tuning them with off-policy RL to improve beyond the demonstrations has drawn growing interest. However, fine-tuning a flow policy against a learned value function is not trivial, because th...
Yonghoon Dong, Minsung Yoon et al.
#cs.LG