RI

#1CausalEvolve: Towards Open-Ended Discovery with Causal Scratchpad

2026-03-15 | Carnegie Mellon University, MBZUAI, Hong Kong Baptist University, The University of Sydney |2603.14575

CausalEvolve enhances LLM evolutionary search frameworks (like AlphaEvolve and ShinkaEvolve) by introducing a causal scratchpad that extracts outcome- and procedure-level factors to explicitly guide program mutations via a Multi-Armed Bandit. The results are backed by solid empirical numbers, showing it outperforms the state-of-the-art ShinkaEvolve across four algorithmic and mathematical tasks (Hadamard, Autocorrelation, Circle Packing, AIME) by up to 9.1% in best-found scores.

#2CreativeBench: Benchmarking and Enhancing Machine Creativity via Self-Evolving Challenges

2026-03-12 | Tsinghua University, Peking University, Southern University of Science and Technology, University of Bristol, The Hong Kong University of Science and Technology (Guangzhou), Xi’an Jiaotong University |2603.11863

Wang et al. introduce CreativeBench to evaluate LLM code generation creativity and propose EvoRePE, a representation engineering technique that extracts a 'creativity vector' from AlphaEvolve search trajectories to steer model activations at inference time. The results are backed by solid empirical evaluations, showing that injecting this vector improves novelty and correctness even without running the full evolutionary search.heuristic, evolvedheuristic) pairs, compute the PCA of their hidden state differences and inject this vector during standard inference to force the model into an exploratory mode.

#3KernelFoundry: Hardware-aware evolutionary GPU kernel optimization

2026-03-12 | Intel Corporation |2603.12440

KernelFoundry is an LLM-based evolutionary framework for GPU kernel optimization that combines MAP-Elites quality-diversity search with meta-prompt co-evolution and gradient-informed mutation hints. The results are rigorously backed by numbers, showing a 2.1x speedup over the AI CUDA Engineer baseline on KernelBench L2 and successful optimization of Llama 3 operations.

#4Procedural Generation of Algorithm Discovery Tasks in Machine Learning

2026-03-18 | University of Oxford, University College London, University of California, Santa Barbara, University of Wisconsin–Madison, Delft University of Technology |2603.17863

This paper introduces DiscoGen, a procedural generator that combinatorially creates millions of algorithm discovery tasks (varying domains, editable modules and datasets) with strict meta-train/meta-test splits to evaluate and train Algorithm Discovery Agents (ADAs). The results are backed by extensive empirical evaluation of open-source LLMs on a fixed subset (DiscoBench), demonstrating that current ADAs struggle with multi-module discovery and that prompt-tuning over a diverse set of procedurally generated tasks significantly improves generalization. The single most useful takeaway is the combinatorial task generation approach (toggling which modules are editable vs.

#1MetaClaw: Just Talk -- An Agent That Meta-Learns and Evolves in the Wild

2026-03-17 | Carnegie Mellon University, UC Berkeley, UNC-Chapel Hill, UC Santa Cruz |2603.17187

MetaClaw is a continual learning framework for LLM agents that combines gradient-free skill evolution (distilling failures into reusable prompt instructions) with asynchronous RL fine-tuning guided by a process reward model. The results are backed by solid empirical gains, showing an 8.25x improvement in end-to-end task completion on a 30-day simulated CLI benchmark.

#2Efficient LLM Serving for Agentic Workflows: A Data Systems Perspective

2026-03-17 | National University of Singapore |2603.16104

Helium optimizes LLM serving for batch agentic workflows by modeling them as query plans and using a Templated Radix Tree (TRT) to enable proactive KV caching and cache-aware scheduling. The results are rigorously backed by numbers, demonstrating up to 1.56x speedups over state-of-the-art systems (vLLM, Parrot, KVFlow) on complex multi-agent workflows and the authors even validate their greedy scheduler's optimality gap against an MILP solver. The most valuable takeaway is the TRT abstraction, which captures global prefix hierarchies across a DAG of LLM calls to maximize KV cache reuse, rather than relying on reactive, per-call caching.

#3IEMAS: An Incentive-Efficiency Routing Framework for Open Agentic Web Ecosystems

2026-03-18 | Shanghai Jiao Tong University |2603.17302

IEMAS routes client requests to distributed LLM agents by formulating the assignment as a Min-Cost Max-Flow bipartite matching problem, using VCG auctions to align economic incentives with KV-cache reuse. The results are backed by solid vLLM simulations, demonstrating an 80.2% KV-cache hit rate and a 35% cost reduction over baselines like GraphRouter.

#4inference-fleet-sim: A Queueing-Theory-Grounded Fleet Capacity Planner for LLM Inference

2026-03-17 | MBZUAI, McGill University, University of Chicago, Tensormesh Inc |2603.16054

This paper introduces a two-phase capacity planner (M/G/c analytical sweep followed by discrete-event simulation) to find minimum-cost GPU fleet configurations for LLM inference under strict latency SLOs. The results are rigorously backed by simulation across multiple GPU profiles and workload traces, demonstrating that intuitive sizing rules often fail (e.g., slower GPUs can be cheaper due to KV-slot multipliers and analytical models approve broken fleets for high-variance agent traffic).

#5Guaranteeing Semantic and Performance Determinism in Flexible GPU Sharing

2026-03-17 | Shanghai Jiao Tong University, Chinese Academy of Sciences, University of Chinese Academy of Sciences |2603.15042

DETSHARE introduces 'GPU coroutines' to decouple logical execution contexts from physical GPU resources, enabling fine-grained spatial sharing without modifying kernels to preserve semantic determinism. The results are highly credible and backed by strong empirical numbers on A800/Hopper GPUs, demonstrating up to 79% higher training throughput and 69% lower inference latency compared to temporal sharing baselines.

#6Cost-Efficient Multimodal LLM Inference via Cross-Tier GPU Heterogeneity

2026-03-13 | University of Illinois Urbana-Champaign |2603.12707

Yu et al. demonstrate that partitioning multimodal LLM inference at the modality boundary (vision encoder vs. language decoder) reduces cross-device transfer costs by O(L), dropping requirements from GB-scale NVLink to MB-scale PCIe. This enables heterogeneous serving architectures where cheap compute-dense GPUs (RTX 4090) handle vision and expensive bandwidth-dense GPUs (A100) handle language. Results are strongly backed by real hardware deployments, showing a 37% improvement in cost-efficiency over homogeneous vLLM baselines.