arXiv:2411.06568v4 Announce Type: replace-cross Abstract: Evaluating preference optimization (PO) algorithms on LLM alignment is a challenging task that presents prohibitive costs, noise, and several variables like model size and hyper-parameters. In this work, we show that it is possible to gain insights on the efficacy of PO algorithm on simpler benchmarks. We design a diagnostic suite of MuJoCo tasks and datasets, which we use to systematically evaluate PO algorithms, establishing a more controlled and cheaper benchmark. We then […]

Ver mais

arXiv:2602.02512v1 Announce Type: new Abstract: We study the notion of unfairness in social networks, where a group such as females in a male-dominated industry are disadvantaged in access to important information, e.g. job posts, due to their less favorable positions in the network. We investigate a well-established network-based formulation of fairness called PageRank fairness, which refers to a fair allocation of the PageRank weights among distinct groups. Our goal is to enhance the PageRank fairness by modifying the […]

Ver mais

Safety Is Not a Switch, It Is a Process For a long time, AI safety seemed like something you could install. You specified constraints, tested for edge cases, built guardrails, and deployed with confidence. This paradigm was fine when systems were largely static. It starts to break when systems change. Today, AI systems do not merely follow pre-programmed logic. They learn. They adapt. They interact with other agents. In multi-agent settings, systems negotiate, compete, coordinate, and sometimes, game […]

Ver mais

arXiv:2602.03823v1 Announce Type: new Abstract: We introduce a new preference-based framework for conditional treatment effect estimation and policy learning, built on the Conditional Preference-based Treatment Effect (CPTE). CPTE requires only that outcomes be ranked under a preference rule, unlocking flexible modeling of heterogeneous effects with multivariate, ordinal, or preference-driven outcomes. This unifies applications such as conditional probability of necessity and sufficiency, conditional Win Ratio, and Generalized Pairwise Comparisons. Despite the intrinsic non-identifiability of comparison-based estimands, CPTE provides interpretable […]

Ver mais

Group Relative Policy Optimization (GRPO) has become a key technique for improving reasoning abilities in large language models, yet its behavior under different domain sequencing strategies is poorly understood. In particular, the impact of sequential (one domain at a time) versus mixed-domain (multiple domain at a time) training in GRPO has not been systematically studied. We provide the first systematic analysis of training-order effects across math, science, logic, and puzzle reasoning tasks. We found (1) single-domain generalization is […]

Ver mais

arXiv:2602.10262v1 Announce Type: new Abstract: The AMD MI300A APU integrates CDNA3 GPUs with high-bandwidth memory and advanced accelerator features: FP8 matrix cores, asynchronous compute engines (ACE), and 2:4 structured sparsity. These capabilities are increasingly relied upon by modern HPC and HPC-AI workloads, yet their execution characteristics and system-level implications remain insufficiently understood. In this paper, we present an execution-centric characterization of FP8 matrix execution, ACE concurrency, and structured sparsity on MI300A using targeted microbenchmarks. We quantify occupancy thresholds, […]

Ver mais

Table of Links Abstract and 1. Introduction Related Works 2.1 Traditional Index Selection Approaches 2.2 RL-based Index Selection Approaches Index Selection Problem Methodology 4.1 Formulation of the DRL Problem 4.2 Instance-Aware Deep Reinforcement Learning for Efficient Index Selection System Framework of IA2 5.1 Preprocessing Phase 5.2 RL Training and Application Phase Experiments 6.1 Experimental Setting 6.2 Experimental Results 6.3 End-to-End Performance Comparison 6.4 Key Insights Conclusion and Future Work, and References 6.3 End-to-End Performance Comparison reveals IA2’s distinctive […]

Ver mais