Reinforcement Learning with Action Chunking
arXiv:2507.07969v4 Announce Type: replace-cross Abstract: We present Q-chunking, a simple yet effective recipe for improving reinforcement learning (RL) algorithms for long-horizon, sparse-reward task
Knowledge catalogue
arXiv:2507.07969v4 Announce Type: replace-cross Abstract: We present Q-chunking, a simple yet effective recipe for improving reinforcement learning (RL) algorithms for long-horizon, sparse-reward task
arXiv:2605.09262v1 Announce Type: cross Abstract: Reinforcement Learning has significantly advanced the reasoning capabilities of Multimodal Large Language Models (MLLMs), yet the resulting policies r
arXiv:2506.04289v3 Announce Type: replace Abstract: Transformer-based models have demonstrated remarkable reasoning abilities, but the mechanisms underlying relational reasoning remain poorly understo
arXiv:2605.09420v1 Announce Type: cross Abstract: In this study, we tackle Generalized Category Discovery (GCD) via a Relational Retrieval perspective, explicitly coupling labeled and unlabeled data t
arXiv:2605.10317v1 Announce Type: cross Abstract: Knowledge graph embedding (KGE) models typically represent each relation as an operator on entity embeddings. In this work, we identify three structur
arXiv:2605.10218v1 Announce Type: new Abstract: Diffusion large language models (dLLMs) offer a promising route to parallel and efficient text generation, but improving their reasoning ability require
arXiv:2605.10414v1 Announce Type: new Abstract: Rotary Positional Encoding (RoPE) is widely used in modern large language models. However, when sequences are extended beyond the range seen during trai
arXiv:2605.09410v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models remain brittle in long-horizon, contact-rich manipulation because success-only imitation provides little supervisi
arXiv:2605.10615v1 Announce Type: new Abstract: Many studies have shown automatic speech processing (ASR) systems have unequal performance across speakergroups (SG's). However, the manner in which suc
arXiv:2605.08186v1 Announce Type: cross Abstract: Test-Time Adaptation (TTA) via entropy minimization (EM) has proven effective for classification tasks, yet its application to generative autoregressi
arXiv:2605.10047v1 Announce Type: cross Abstract: Loss reweighting is a widely used strategy for long-tailed classification, but existing reweighting strategies often rely on heuristics and rarely def
arXiv:2605.09212v1 Announce Type: new Abstract: Centralized training with decentralized execution (CTDE) is a standard framework for cooperative multi-agent policy-gradient reinforcement learning, all
arXiv:2605.06241v2 Announce Type: replace Abstract: Reinforcement learning has become the standard for improving reasoning in large language models, yet evidence increasingly suggests that RL does not
arXiv:2605.10909v1 Announce Type: new Abstract: This work revisits standard policy gradient methods used on restricted policy classes, which are known to get stuck in suboptimal critical points. We id
arXiv:2605.08311v1 Announce Type: cross Abstract: Model merging provides a compelling paradigm for integrating specialized expertise into a unified multi-task model, a goal that aligns naturally with
arXiv:2512.00920v4 Announce Type: replace Abstract: Reliable reward models (RMs) are critical for ensuring the safe alignment of large language models (LLMs). However, current RM evaluation methods fo
arXiv:2603.05066v2 Announce Type: replace Abstract: Single-task RL agents are typically trained under a fixed reward function, which limits their robustness to reward misspecification and their abilit
arXiv:2605.09196v1 Announce Type: cross Abstract: Learning-based simulation of multi-object rigid-body dynamics remains difficult because contact is discontinuous and errors compound over long horizon
arXiv:2605.10057v1 Announce Type: new Abstract: Compositional spatiotemporal reasoning often requires a system to invoke multiple heterogeneous specialists, such as geometric, temporal, topological, a
arXiv:2605.10899v1 Announce Type: new Abstract: Training deep research agents, namely systems that plan, search, evaluate evidence, and synthesize long-form reports, pushes reinforcement learning beyo
arXiv:2605.09346v1 Announce Type: cross Abstract: The Chain-of-Thought (CoT) paradigm, while enhancing the interpretability of Large Language Models (LLMs), is constrained by the inefficiencies and ex
arXiv:2605.08334v1 Announce Type: new Abstract: We present SalesSim, a framework and testbed for evaluating the ability of Multimodal Large Language Models (MLLMs) to simulate realistic, persona-drive
arXiv:2605.10289v1 Announce Type: new Abstract: Offline-to-online learning aims to improve online decision-making by leveraging offline logged data. A central challenge in this setting is the distribu
arXiv:2603.27977v2 Announce Type: replace Abstract: Reinforcement learning is critical to improving large reasoning models, but its success relies heavily on verifiable rewards (RLVR), making it hard
arXiv:2605.08528v1 Announce Type: cross Abstract: Autonomous-driving simulators typically trade physical fidelity for scalable parallelism. Physics-based platforms such as CARLA and MetaDrive provide
arXiv:2604.07383v2 Announce Type: replace Abstract: Cross-city transfer improves prediction in label-scarce cities by leveraging labeled data from other cities, but it becomes challenging when cities
arXiv:2605.05736v2 Announce Type: replace Abstract: Vector quantization (VQ) with autoregressive (AR) token modeling is a widely adopted and highly competitive paradigm for time-series generation. How
arXiv:2605.10603v1 Announce Type: new Abstract: Despite strong zero-shot performance, SAM is unreliable under domain shift due to Mask-level Confidence Confusion (MCC), where a single IoU-based mask s
arXiv:2605.09945v1 Announce Type: new Abstract: Many high-stakes decisions in health care, public policy, and clinical development require committing to a single policy that will be applied uniformly
arXiv:2605.09345v1 Announce Type: new Abstract: We identify a Selection Plateau phenomenon in one-shot neural network pruning: all rank-monotone weight scorers converge to identical accuracy at fixed
arXiv:2605.08874v1 Announce Type: new Abstract: Open-vocabulary semantic segmentation requires adapting image-level vision-language models such as CLIP to dense pixel-level prediction, which is challe
arXiv:2605.08302v1 Announce Type: cross Abstract: Real-world digital Parkinson's disease assessment faces challenges such as heterogeneous modalities, cross-device bias, and incomplete labeling. Exist
arXiv:2605.08453v1 Announce Type: cross Abstract: This paper studies the role of sinks and diagonal patterns as attention switch and anti-oversmoothing mechanisms. We analyze geometric conditions unde
arXiv:2605.09343v1 Announce Type: new Abstract: Decision making in large-scale complaint handling systems increasingly relies on heterogeneous evidence, including complaint narratives, screenshots, or
arXiv:2605.09359v1 Announce Type: cross Abstract: Agentic large language models often rely on skills, reusable natural language procedures that guide planning, action, and tool use. In practice, skill
arXiv:2605.10503v1 Announce Type: new Abstract: Large Language Models (LLMs) show remarkable semantic understanding but often struggle with structural understanding when processing graph topologies in
So basically Anthropic’s estimated valuation went up half a trillion dollars in a couple weeks (then back down a bit) on hype. Small wonder that companies like OpenAI, Anthropic and Tesla keep hyping
arXiv:2605.08183v1 Announce Type: new Abstract: Generalized Category Discovery (GCD) seeks to identify novel categories from unlabeled data while retaining the classification ability of seen categorie
speaks for itself Musk's lawyer: 'Are you completely trustworthy?' Altman: 'I believe so.' Musk's lawyer: 'But, you know, you don't know whether you're completely trustworthy.' Altman: 'I'll just amen
arXiv:2605.09498v1 Announce Type: cross Abstract: Time series, spatial data, and images are natural applications of Neural Processes. However, when such data exhibit strong periodicity and quasi-perio
arXiv:2605.10154v1 Announce Type: new Abstract: Long-horizon forecasting of time-dependent partial differential equations (PDEs) is critical for characterizing the sustained evolution of physical syst
arXiv:2605.09989v1 Announce Type: cross Abstract: Recent advances in robot imitation learning have yielded powerful visuomotor policies capable of manipulating a wide variety of objects directly from
arXiv:2604.01824v2 Announce Type: replace Abstract: We introduce STRIVE (SpatioTemporal Reinforcement with Importance-aware Variant Exploration), a structured reinforcement learning framework for vide
arXiv:2502.18334v5 Announce Type: replace Abstract: Graph-based learning excels at capturing interaction patterns in diverse domains like recommendation, fraud detection, and particle physics. However
arXiv:2605.09241v1 Announce Type: cross Abstract: Joint-Embedding Predictive Architectures (JEPAs) provide a simpleframework for learning world models by predicting future latent representations.Howev
If anyone, anywhere builds a superhuman artificial intelligence using present methods, the most likely outcome is catastrophe. There have accordingly been widespread calls for an international agreeme
arXiv:2605.09834v1 Announce Type: cross Abstract: Modern predictive systems encode beliefs that can act as useful prior information for statistical inference in data-limited settings. Using them for p
arXiv:2605.08963v1 Announce Type: cross Abstract: Machine Learning (ML) models trained on complex health surveys such as the National Health and Nutrition Examination Survey (NHANES) often ignore prim
arXiv:2605.10811v1 Announce Type: cross Abstract: This paper develops a joint spectral radius (JSR) framework for analyzing rank-one deflated Q-value iteration (Q-VI) in discounted Markov decision pro
arXiv:2605.08724v1 Announce Type: new Abstract: Unifying multimodal understanding and generation is a compelling frontier that is beginning to emerge in the medical field. However, the limited existin
arXiv:2602.04611v2 Announce Type: replace-cross Abstract: The synthetic control method (SCM) estimates causal effects in panel data with a single-treated unit by constructing a counterfactual outcome
arXiv:2605.09810v1 Announce Type: cross Abstract: Protein function is often controlled by ligands that bias the direction of state transitions, such as agonists and antagonists, rather than stabilizin
arXiv:2605.09922v1 Announce Type: cross Abstract: While recent self-training approaches have reduced reliance on human-labeled data for aligning LLMs, they still face critical limitations: (i) sensiti
arXiv:2505.11577v4 Announce Type: replace-cross Abstract: Recent application programming interface (API) restrictions on major social media platforms challenge compliance with the EU Digital Services
arXiv:2605.10698v1 Announce Type: cross Abstract: Multi-agent systems (MAS) assume that collaborating inherently improves Large Language Model (LLM) reasoning. We challenge this by demonstrating that
The FT says that Amazon employees are doing random unnecessary task automations to consume tokens and to show their bosses that they're using AI more https://www.ft.com/content/8ee0d3ef-9548-422d-8ff1
arXiv:2605.08464v1 Announce Type: new Abstract: The manifold hypothesis (MH) is often used to explain how machine learning can overcome the curse of dimensionality. However, the MH is only applicable
arXiv:2605.08837v1 Announce Type: cross Abstract: Abstract concepts - justice, theory, availability - have no single perceivable referent; in the human brain, their meaning emerges from a web of exper
arXiv:2605.09221v1 Announce Type: cross Abstract: Fairness impossibility results often look like distinct scalar incompatibility statements. We show that several share one RKHS geometry: fairness crit
arXiv:2512.18610v3 Announce Type: replace Abstract: Intuitively, a more deterministic time series should be easier to forecast. However, point-wise loss functions (e.g., MSE and MAE), serving as diffe