RI

#1MetaAgent-X : Breaking the Ceiling of Automatic Multi-Agent Systems via End-to-End Reinforcement Learning

2026-05-14 | Amazon AGI, UCSD, Oregon State University, Pennsylvania State University, AG2AI, Inc. |2605.14212

MetaAgent-X is an end-to-end reinforcement learning framework that jointly optimizes a meta-agent (Designer) that generates multi-agent workflows and the downstream agents (Executors) that run them. The results are backed by strong empirical evidence, showing up to 21.7% improvement over baselines on math and code benchmarks using Qwen3-8B. The key insight is the Executor-Designer Hierarchical Rollout combined with stagewise co-evolution: by sampling M designs and executing each N times, the system isolates the designer's reward from the executor's reward, while alternating training phases stabilizes the non-stationary RL environment.

#2Harnessing Agentic Evolution

2026-05-13 | Tsinghua University, Université de Montréal & Mila, The Hong Kong University of Science and Technology (Guangzhou), Nanyang Technological University, Shanghai Jiao Tong University, Singapore University of Technology and Design, DeepWisdom |2605.13821

Zhang et al. introduce AEVO, a framework that formulates LLM evolutionary search as an interactive environment where a meta-agent edits the underlying search procedure or agent context rather than directly generating candidate solutions. The results are backed by strong empirical evidence, achieving state-of-the-art performance on Anthropic's Kernel optimization task (1138 cycles) and outperforming baselines like OpenEvolve and HyperAgents across multiple open-ended discovery and reasoning benchmarks. The key insight is separating the candidate generation loop from a meta-editing phase, allowing the system to dynamically revise its own selection rules, feedback summaries and memory structures when search plateaus.

#3Budget-Efficient Automatic Algorithm Design via Code Graph

2026-05-11 | Massachusetts Institute of Technology, University of California, Berkeley |2605.10598

This paper proposes a graph-based search framework for LLM-driven automatic algorithm design where algorithms are represented as paths in a directed acyclic graph (DAG) of code blocks, allowing the LLM to generate composable corrections rather than full rewrites. The approach is backed by strong empirical results on TSP, LRP and BRP, demonstrating up to 22% smaller optimality gaps than full-algorithm baselines at equal token budgets. The key insight is the use of a Random Forest surrogate model combined with Shapley values to perform correction-level credit assignment, effectively isolating the impact of specific code edits from the overall algorithm fitness. This is highly relevant for LLM evolutionary search; the DAG representation and Shapley-based credit assignment provide a concrete architectural blueprint for drastically improving sample efficiency, search signal quality and observability in automated heuristic discovery.

#4Teacher-Aware Evolution of Heuristic Programs from Learned Optimization Policies

2026-05-11 | Shanghai Jiao Tong University, Shenzhen Technology University, Shanghai Polytechnic University |2605.10634

This paper introduces a teacher-aware evolutionary framework that uses pre-trained neural optimization policies to provide step-by-step behavioral feedback to guide the LLM-based evolution of symbolic heuristics. The results are backed by strong empirical numbers, showing consistent improvements over standard performance-driven baselines (EoH, ReEvo) on scheduling, routing and graph optimization benchmarks, while maintaining the fast inference of static code. The key insight is that instead of relying solely on delayed endpoint performance, one can query a black-box neural policy on the specific states visited by a candidate heuristic and use the behavioral discrepancies (top-1 action agreement) to prompt targeted LLM revisions. This is highly relevant for research in LLM evolutionary search, as it provides a concrete, actionable implementation of a process reward model that directly improves search guidance and sample efficiency.

#5RubricEM: Meta-RL with Rubric-guided Policy Decomposition beyond Verifiable Rewards

2026-05-11 | Google Cloud AI Research, University of Illinois Urbana-Champaign |2605.10899

Li et al. introduce RubricEM, an RL framework for long-horizon LLM agents that uses dynamically generated rubrics to structure trajectories, assign stage-wise process rewards via Stage-Structured GRPO and train a reflection memory bank for cross-episode transfer. The results are rigorously backed by numbers, showing their 8B model outperforming 30B+ open models and approaching proprietary systems on four long-form research benchmarks. The key insight is that rubrics can serve as a shared interface across the entire RL loop, acting simultaneously as structural priors for planning, dynamic process reward models for intermediate credit assignment and structured memory for experience reuse.

#6LEVI: Stronger Search Architectures Can Substitute for Larger LLMs in Evolutionary Search

2026-05-10 | Independent Researcher |2605.09764

LEVI introduces a cost-efficient framework for LLM-guided evolutionary search that replaces expensive frontier-model reliance with a stronger search architecture. The authors demonstrate that by combining a CVT-MAP-Elites archive, role-aware LLM routing (using small models for local edits and large models only for structural paradigm shifts) and rank-preserving proxy benchmarks, they can outperform existing frameworks like ShinkaEvolve and GEPA on systems-research and prompt-optimization tasks at 3.3 to 6.7x lower cost. The results are rigorously backed by empirical cost and performance metrics across 11 benchmarks.

#7Adapting AlphaEvolve to Optimize Fully Homomorphic Encryption on TPUs

2026-05-14 | Google DeepMind, Google, Georgia Institute of Technology |2605.14718

This paper applies the AlphaEvolve framework to automatically optimize Fully Homomorphic Encryption (FHE) kernels for TPU execution, using LLM-driven code generation guided by hardware-in-the-loop latency feedback and strict correctness checks. The results are real and backed by physical hardware deployments, achieving up to a 2.5x speedup over human-engineered state-of-the-art baselines on TPUv5e chips. The key insight is that optimizing isolated operations often fails to yield end-to-end speedups; instead, co-evolving multiple components and scoring them against complete system execution, while feeding back low-level execution traces (e.g., Xprof) to the LLM, is necessary to discover non-obvious compiler and scheduling optimizations.

#8EvolveMem:Self-Evolving Memory Architecture via AutoResearch for LLM Agents

2026-05-13 | UNC-Chapel Hill, UC Berkeley, UCSC |2605.13941

EVOLVEMEM introduces a self-evolving memory architecture for LLM agents that autonomously optimizes its retrieval infrastructure (e.g., fusion weights, context budgets) using an LLM-powered diagnosis module that analyzes failure logs. The results are backed by strong empirical evidence, showing a 25.7% relative improvement over the state-of-the-art on the LoCoMo benchmark and positive transfer to MemBench. The key insight is treating the retrieval configuration as a structured action space and using an LLM to propose targeted adjustments based on per-question failure logs, safeguarded by a revert-on-regression mechanism.

#9CVEvolve: Autonomous Algorithm Discovery for Unstructured Scientific Data Processing

2026-05-12 | Argonne National Laboratory |2605.11359

CVEvolve is an LLM-based evolutionary search framework that autonomously discovers data-processing algorithms using a generate-tune-evolve loop, lineage-aware sampling and SQL-backed persistent memory. The results are backed by strong empirical evidence, showing significant improvements over baselines, such as reducing X-ray image registration error from 0.98 to 0.12. The key insight is the use of a relational SQL database to manage search history and lineage (avoiding context bloat) combined with a dedicated holdout-test agent to prevent over-optimization on small development sets.

#10TacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent Systems

2026-05-10 | Carnegie Mellon University, University of Science and Technology of China, National University of Singapore, Shanghai AI Lab |2605.09539

TacoMAS introduces a test-time co-evolution framework for LLM multi-agent systems that updates agent capabilities rapidly (via memory refinement) and communication topology slowly (via birth-death operations and edge edits). The results are backed by strong empirical evidence, achieving an average 13.3% accuracy improvement over 20 baselines across four diverse benchmarks. The key insight is the separation of evolution into two time scales—fast capability updates to handle immediate subtasks and slow topology updates to preserve coordination stability—which prevents the system from diverging during test-time adaptation. Furthermore, using a meta-judge to assign contribution scores to individual agents effectively acts as a process reward model to guide the replicator dynamics.

#11Graph-Grounded Optimization: Rao-Family Metaheuristics, Classical OR and SLM-Driven Formulation over Knowledge Graphs

2026-05-13 | Samyama.ai |2605.12204

This paper introduces a graph-grounded optimization paradigm where operations research problem components are sourced directly from property knowledge graphs via Cypher queries, rather than from lossy natural language descriptions. The results are backed by strong empirical numbers across 7 real-world KGs, demonstrating that while classical solvers and metaheuristics succeed, frontier SLMs (including the MILP-tuned OptiMind) fail almost completely at zero-shot formulation. The key insight is the identification of a 'data-binding pathology': LLMs generate syntactically valid optimization code that silently fails or returns trivial solutions because it binds to hallucinated data structures instead of the actual operational database.

#1Attention Once Is All You Need: Efficient Streaming Inference with Stateful Transformers

2026-05-13 | LayerScale, Inc. |2605.13784

Norgren introduces a stateful transformer inference architecture that decouples data ingestion from query processing, maintaining a persistent KV cache to achieve constant-time query latency for streaming workloads. The approach is backed by strong empirical results, demonstrating a 2.4x to 5.9x speedup over state-of-the-art engines like vLLM and SGLang on streaming benchmarks while maintaining approximately 43ms latency regardless of context size. The key insight is the use of Flash Queries, which utilize idle GPU cycles between data arrivals to pre-compute answers to registered queries against the evolving context, effectively pushing user-visible latency toward zero.

#2PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding

2026-05-13 | Zhejiang University, Northeastern University, University of Surrey, Zhongguancun Institute of Artificial Intelligence |2605.13319

PipeSD accelerates cloud-edge collaborative LLM inference by using dynamic programming to optimally batch and pipeline draft tokens, alongside a Bayesian optimization-tuned dual-threshold mechanism for triggering speculative verification. The results are backed by empirical hardware measurements, demonstrating 1.16x–2.16x speedups and up to 25% energy reduction over baselines like EdgeLLM on standard benchmarks. The key insight is formulating the token-batching decision—balancing communication startup overhead against immediate transmission—as a dynamic programming problem to perfectly overlap edge-side autoregressive generation with network transmission.

#3DisagMoE: Computation-Communication overlapped MoE Training via Disaggregated AF-Pipe Parallelism

2026-05-10 | ByteDance Seed, University of Washington, Cornell University |2605.11005

DisagMoE optimizes large-scale Mixture-of-Experts (MoE) model training by disaggregating attention and feed-forward network (FFN) layers onto separate GPU groups and using a multi-stage pipeline to overlap communication and computation. The results are backed by strong empirical evidence, demonstrating up to 1.81x throughput speedups over Megatron-LM and 1.34x over state-of-the-art overlap methods on a 128-GPU H800 cluster. The key insight is the use of a Compute-Communication roofline model, solved via MILP, to asymmetrically allocate GPU and NIC resources based on the distinct arithmetic intensities of different model components (compute-bound attention vs. communication-bound FFN).