MemoryRewardBench: Benchmarking Reward Models for Long-Term Memory Management in Large Language Models

Existing works increasingly adopt memory-centric mechanisms to process long contexts in a segment manner, and effective memory management is one of the key capabilities that enables large language models to effectively propagate information across the entire sequence. Therefore, leveraging reward models (RMs) to automatically and reliably evaluate memory quality is critical. In this work, we introduce MemoryRewardBench, the first benchmark to systematically study the ability of RMs to evaluate long-term memory management processes. MemoryRewardBench covers both long-context comprehension and long-form generation tasks, featuring 10 distinct settings with different memory management patterns, with context length ranging from 8K to 128K tokens. Evaluations on 13 cutting-edge RMs indicate a diminishing performance gap between open-source and proprietary models, with newer-generation models consistently outperforming their predecessors regardless of parameter count. We further expose the capabilities and fundamental limitations of current RMs in evaluating LLM memory management across diverse settings.

$\texttt{MemoryRewardBench}$: Benchmarking Reward Models for Long-Term Memory Management in Large Language Models

Existing works increasingly adopt memory-centric mechanisms to process long contexts in a segment manner, and effective memory management is one of the key capabilities that enables large language models to effectively propagate information across the entire sequence. Therefore, leveraging reward models (RMs) to automatically and reliably evaluate memory quality is critical. In this work, we introduce $\texttt{MemoryRewardBench}$, the first benchmark to systematically study the ability of RMs to evaluate long-term memory management processes. $\texttt{MemoryRewardBench}$ covers both long-context comprehension and long-form generation tasks, featuring 10 distinct settings with different memory management patterns, with context length ranging from 8K to 128K tokens. Evaluations on 13 cutting-edge RMs indicate a diminishing performance gap between open-source and proprietary models, with newer-generation models consistently outperforming their predecessors regardless of parameter count. We further expose the capabilities and fundamental limitations of current RMs in evaluating LLM memory management across diverse settings.

LongRM: Revealing and Unlocking the Context Boundary of Reward Modeling

Reward model (RM) plays a pivotal role in aligning large language model (LLM) with human preferences. As real-world applications increasingly involve long history trajectories, e.g., LLM agent, it becomes indispensable to evaluate whether a model's responses are not only high-quality but also grounded in and consistent with the provided context. Yet, current RMs remain confined to short-context settings and primarily focus on response-level attributes (e.g., safety or helpfulness), while largely neglecting the critical dimension of long context-response consistency. In this work, we introduce Long-RewardBench, a benchmark specifically designed for long-context RM evaluation, featuring both Pairwise Comparison and Best-of-N tasks. Our preliminary study reveals that even state-of-the-art generative RMs exhibit significant fragility in long-context scenarios, failing to maintain context-aware preference judgments. Motivated by the analysis of failure patterns observed in model outputs, we propose a general multi-stage training strategy that effectively scales arbitrary models into robust Long-context RMs (LongRMs). Experiments show that our approach not only substantially improves performance on long-context evaluation but also preserves strong short-context capability. Notably, our 8B LongRM outperforms much larger 70B-scale baselines and matches the performance of the proprietary Gemini 2.5 Pro model.

High-Performance Parallel Optimization of the Fish School Behaviour on the Setonix Platform Using OpenMP

This paper presents an in-depth investigation into the high-performance parallel optimization of the Fish School Behaviour (FSB) algorithm on the Setonix supercomputing platform using the OpenMP framework. Given the increasing demand for enhanced computational capabilities for complex, large-scale calculations across diverse domains, there's an imperative need for optimized parallel algorithms and computing structures. The FSB algorithm, inspired by nature's social behavior patterns, provides an ideal platform for parallelization due to its iterative and computationally intensive nature. This study leverages the capabilities of the Setonix platform and the OpenMP framework to analyze various aspects of multi-threading, such as thread counts, scheduling strategies, and OpenMP constructs, aiming to discern patterns and strategies that can elevate program performance. Experiments were designed to rigorously test different configurations, and our results not only offer insights for parallel optimization of FSB on Setonix but also provide valuable references for other parallel computational research using OpenMP. Looking forward, other factors, such as cache behavior and thread scheduling strategies at micro and macro levels, hold potential for further exploration and optimization.

Multispectral Remote Sensing for Weed Detection in West Australian Agricultural Lands

The Kondinin region in Western Australia faces significant agricultural challenges due to pervasive weed infestations, causing economic losses and ecological impacts. This study constructs a tailored multispectral remote sensing dataset and an end-to-end framework for weed detection to advance precision agriculture practices. Unmanned aerial vehicles were used to collect raw multispectral data from two experimental areas (E2 and E8) over four years, covering 0.6046 km^{2} and ground truth annotations were created with GPS-enabled vehicles to manually label weeds and crops. The dataset is specifically designed for agricultural applications in Western Australia. We propose an end-to-end framework for weed detection that includes extensive preprocessing steps, such as denoising, radiometric calibration, image alignment, orthorectification, and stitching. The proposed method combines vegetation indices (NDVI, GNDVI, EVI, SAVI, MSAVI) with multispectral channels to form classification features, and employs several deep learning models to identify weeds based on the input features. Among these models, ResNet achieves the highest performance, with a weed detection accuracy of 0.9213, an F1-Score of 0.8735, an mIOU of 0.7888, and an mDC of 0.8865, validating the efficacy of the dataset and the proposed weed detection method.