Encoder-Decoder or Decoder-Only? Revisiting Encoder-Decoder Large Language Model

Recent large language model (LLM) research has undergone an architectural shift from encoder-decoder modeling to nowadays the dominant decoder-only modeling. This rapid transition, however, comes without a rigorous comparative analysis especially \textit{from the scaling perspective}, raising concerns that the potential of encoder-decoder models may have been overlooked. To fill this gap, we revisit encoder-decoder LLM (RedLLM), enhancing it with recent recipes from decoder-only LLM (DecLLM). We conduct a comprehensive comparison between RedLLM, pretrained with prefix language modeling (LM), and DecLLM, pretrained with causal LM, at different model scales, ranging from $\sim$150M to $\sim$8B. Using RedPajama V1 (1.6T tokens) for pretraining and FLAN for instruction tuning, our experiments show that RedLLM produces compelling scaling properties and surprisingly strong performance. While DecLLM is overall more compute-optimal during pretraining, RedLLM demonstrates comparable scaling and context length extrapolation capabilities. After instruction tuning, RedLLM achieves comparable and even better results on various downstream tasks while enjoying substantially better inference efficiency. We hope our findings could inspire more efforts on re-examining RedLLM, unlocking its potential for developing powerful and efficient LLMs.

Exploration of Summarization by Generative Language Models for Automated Scoring of Long Essays

BERT and its variants are extensively explored for automated scoring. However, a limit of 512 tokens for these encoder-based models showed the deficiency in automated scoring of long essays. Thus, this research explores generative language models for automated scoring of long essays via summarization and prompting. The results revealed great improvement of scoring accuracy with QWK increased from 0.822 to 0.8878 for the Learning Agency Lab Automated Essay Scoring 2.0 dataset.

HINT: Helping Ineffective Rollouts Navigate Towards Effectiveness

Reinforcement Learning (RL) has become a key driver for enhancing the long chain-of-thought (CoT) reasoning capabilities of Large Language Models (LLMs). However, prevalent methods like GRPO often fail when task difficulty exceeds the model's capacity, leading to reward sparsity and inefficient training. While prior work attempts to mitigate this using off-policy data, such as mixing RL with Supervised Fine-Tuning (SFT) or using hints, they often misguide policy updates In this work, we identify a core issue underlying these failures, which we term low training affinity. This condition arises from a large distributional mismatch between external guidance and the model's policy. To diagnose this, we introduce Affinity, the first quantitative metric for monitoring exploration efficiency and training stability. To improve Affinity, we propose HINT: Helping Ineffective rollouts Navigate Towards effectiveness, an adaptive hinting framework. Instead of providing direct answers, HINT supplies heuristic hints that guide the model to discover solutions on its own, preserving its autonomous reasoning capabilities. Extensive experiments on mathematical reasoning tasks show that HINT consistently outperforms existing methods, achieving state-of-the-art results with models of various scales, while also demonstrating significantly more stable learning and greater data efficiency.Code is available on Github.

AI Foundation Model for Time Series with Innovations Representation

This paper introduces an Artificial Intelligence (AI) foundation model for time series in engineering applications, where causal operations are required for real-time monitoring and control. Since engineering time series are governed by physical, rather than linguistic, laws, large-language-model-based AI foundation models may be ineffective or inefficient. Building on the classical innovations representation theory of Wiener, Kallianpur, and Rosenblatt, we propose Time Series GPT (TS-GPT) -- an innovations-representation-based Generative Pre-trained Transformer for engineering monitoring and control. As an example of foundation model adaptation, we consider Probabilistic Generative Forecasting, which produces future time series samples from conditional probability distributions given past realizations. We demonstrate the effectiveness of TS-GPT in forecasting real-time locational marginal prices using historical data from U.S. independent system operators.

A Novel Symbol Level Precoding based AFDM Transmission Framework: Offloading Equalization Burden to Transmitter Side

Affine Frequency Division Multiplexing (AFDM) has attracted considerable attention for its robustness to Doppler effects. However, its high receiver-side computational complexity remains a major barrier to practical deployment. To address this, we propose a novel symbol-level precoding (SLP)-based AFDM transmission framework, which shifts the signal processing burden in downlink communications from user side to the base station (BS), enabling direct symbol detection without requiring channel estimation or equalization at the receiver. Specifically, in the uplink phase, we propose a Sparse Bayesian Learning (SBL) based channel estimation algorithm by exploiting the inherent sparsity of affine frequency (AF) domain channels. In particular, the sparse prior is modeled via a hierarchical Laplace distribution, and parameters are iteratively updated using the Expectation-Maximization (EM) algorithm. We also derive the Bayesian Cramer-Rao Bound (BCRB) to characterize the theoretical performance limit. In the downlink phase, the BS employs the SLP technology to design the transmitted waveform based on the estimated uplink channel state information (CSI) and channel reciprocity. The resulting optimization problem is formulated as a second-order cone programming (SOCP) problem, and its dual problem is investigated by Lagrangian function and Karush-Kuhn-Tucker conditions. Simulation results demonstrate that the proposed SBL estimator outperforms traditional orthogonal matching pursuit (OMP) in accuracy and robustness to off-grid effects, while the SLP-based waveform design scheme achieves performance comparable to conventional AFDM receivers while significantly reducing the computational complexity at receiver, validating the practicality of our approach.

LEDOM: An Open and Fundamental Reverse Language Model

We introduce LEDOM, the first purely reverse language model, trained autoregressively on 435B tokens with 2B and 7B parameter variants, which processes sequences in reverse temporal order through previous token prediction. For the first time, we present the reverse language model as a potential foundational model across general tasks, accompanied by a set of intriguing examples and insights. Based on LEDOM, we further introduce a novel application: Reverse Reward, where LEDOM-guided reranking of forward language model outputs leads to substantial performance improvements on mathematical reasoning tasks. This approach leverages LEDOM's unique backward reasoning capability to refine generation quality through posterior evaluation. Our findings suggest that LEDOM exhibits unique characteristics with broad application potential. We will release all models, training code, and pre-training data to facilitate future research.

Think Twice, Act Once: Token-Aware Compression and Action Reuse for Efficient Inference in Vision-Language-Action Models

Vision-Language-Action (VLA) models have emerged as a powerful paradigm for general-purpose robot control through natural language instructions. However, their high inference cost-stemming from large-scale token computation and autoregressive decoding-poses significant challenges for real-time deployment and edge applications. While prior work has primarily focused on architectural optimization, we take a different perspective by identifying a dual form of redundancy in VLA models: (i) high similarity across consecutive action steps, and (ii) substantial redundancy in visual tokens. Motivated by these observations, we propose FlashVLA, the first training-free and plug-and-play acceleration framework that enables action reuse in VLA models. FlashVLA improves inference efficiency through a token-aware action reuse mechanism that avoids redundant decoding across stable action steps, and an information-guided visual token selection strategy that prunes low-contribution tokens. Extensive experiments on the LIBERO benchmark show that FlashVLA reduces FLOPs by 55.7% and latency by 36.0%, with only a 0.7% drop in task success rate. These results demonstrate the effectiveness of FlashVLA in enabling lightweight, low-latency VLA inference without retraining.

A system identification approach to clustering vector autoregressive time series

Clustering of time series based on their underlying dynamics is keeping attracting researchers due to its impacts on assisting complex system modelling. Most current time series clustering methods handle only scalar time series, treat them as white noise, or rely on domain knowledge for high-quality feature construction, where the autocorrelation pattern/feature is mostly ignored. Instead of relying on heuristic feature/metric construction, the system identification approach allows treating vector time series clustering by explicitly considering their underlying autoregressive dynamics. We first derive a clustering algorithm based on a mixture autoregressive model. Unfortunately it turns out to have significant computational problems. We then derive a `small-noise' limiting version of the algorithm, which we call k-LMVAR (Limiting Mixture Vector AutoRegression), that is computationally manageable. We develop an associated BIC criterion for choosing the number of clusters and model order. The algorithm performs very well in comparative simulations and also scales well computationally.

Multi-Constraint Safe Reinforcement Learning via Closed-form Solution for Log-Sum-Exp Approximation of Control Barrier Functions

The safety of training task policies and their subsequent application using reinforcement learning (RL) methods has become a focal point in the field of safe RL. A central challenge in this area remains the establishment of theoretical guarantees for safety during both the learning and deployment processes. Given the successful implementation of Control Barrier Function (CBF)-based safety strategies in a range of control-affine robotic systems, CBF-based safe RL demonstrates significant promise for practical applications in real-world scenarios. However, integrating these two approaches presents several challenges. First, embedding safety optimization within the RL training pipeline requires that the optimization outputs be differentiable with respect to the input parameters, a condition commonly referred to as differentiable optimization, which is non-trivial to solve. Second, the differentiable optimization framework confronts significant efficiency issues, especially when dealing with multi-constraint problems. To address these challenges, this paper presents a CBF-based safe RL architecture that effectively mitigates the issues outlined above. The proposed approach constructs a continuous AND logic approximation for the multiple constraints using a single composite CBF. By leveraging this approximation, a close-form solution of the quadratic programming is derived for the policy network in RL, thereby circumventing the need for differentiable optimization within the end-to-end safe RL pipeline. This strategy significantly reduces computational complexity because of the closed-form solution while maintaining safety guarantees. Simulation results demonstrate that, in comparison to existing approaches relying on differentiable optimization, the proposed method significantly reduces training computational costs while ensuring provable safety throughout the training process.

Towards Intelligent Edge Sensing for ISCC Network: Joint Multi-Tier DNN Partitioning and Beamforming Design

The combination of Integrated Sensing and Communication (ISAC) and Mobile Edge Computing (MEC) enables devices to simultaneously sense the environment and offload data to the base stations (BS) for intelligent processing, thereby reducing local computational burdens. However, transmitting raw sensing data from ISAC devices to the BS often incurs substantial fronthaul overhead and latency. This paper investigates a three-tier collaborative inference framework enabled by Integrated Sensing, Communication, and Computing (ISCC), where cloud servers, MEC servers, and ISAC devices cooperatively execute different segments of a pre-trained deep neural network (DNN) for intelligent sensing. By offloading intermediate DNN features, the proposed framework can significantly reduce fronthaul transmission load. Furthermore, multiple-input multiple-output (MIMO) technology is employed to enhance both sensing quality and offloading efficiency. To minimize the overall sensing task inference latency across all ISAC devices, we jointly optimize the DNN partitioning strategy, ISAC beamforming, and computational resource allocation at the MEC servers and devices, subject to sensing beampattern constraints. We also propose an efficient two-layer optimization algorithm. In the inner layer, we derive closed-form solutions for computational resource allocation using the Karush-Kuhn-Tucker conditions. Moreover, we design the ISAC beamforming vectors via an iterative method based on the majorization-minimization and weighted minimum mean square error techniques. In the outer layer, we develop a cross-entropy based probabilistic learning algorithm to determine an optimal DNN partitioning strategy. Simulation results demonstrate that the proposed framework substantially outperforms existing two-tier schemes in inference latency.