ATLAS: Agentic Test-time Learning-to-Allocate Scaling

2026-06-01 · Peijia Qin, Qi Cao, Pengtao Xie arxiv

Test-time scaling has become a major way to improve large language model reasoning, but its orchestration has remained designer-engineered: a fixed sample budget, a fixed refinement loop, a fixed scoring rule, or a fixed search policy decides how compute is spent, leaving the model in charge of solving but not of orchestration. We introduce ATLAS, an agentic test-time scaling framework in which an LLM orchestrator owns the control loop end-to-end. Through a single action, explore, which dispatches a fresh independent solver on the original problem, the orchestrator decides whether to gather more evidence, when to stop, and how to synthesize the final answer; the action space is extensible, with each explore call optionally specifying solver, reasoning effort, or prompting strategy. We evaluate ATLAS on four benchmarks covering scientific question answering, code generation, and multimodal reasoning under a Claude Sonnet 4.6 backbone, where it reaches 56.00% on HLE-Verified, 82.29% on LiveCodeBench, 85.75% on GPQA-Diamond, and 23.71% on BabyVision while using far fewer API calls than fixed-workflow baselines. A multi-model extension, ATLAS-MM, that exposes solver choice as an additional action dimension further improves HLE-Verified to 60.00% and LiveCodeBench to 85.63%, with consistent gains on GPQA-Diamond and BabyVision. Ablations replacing the orchestrator's direct synthesis with a separate integrator degrade or fail to improve accuracy on three of four benchmarks, consistent with the role of stateful evidence management in producing the gains.

📄 PDF Abstract BibTeX arXiv:2606.01667

Code (0)

등록된 구현이 없습니다.

Tasks

Multimodal Reasoning Question Answering Code Generation

Results from the Paper

Rank	Task	Dataset	Model	Metrics
		GPQA	ATLAS	Accuracy: 85.63

Agentic Test-Time Scaling for WebAgents

2026-02-12 · Nicholas Lee, Lutfi Eren Erdogan, Chris Joseph John, Surya Krishnapillai 외 arxiv

Test-time scaling has become a standard way to improve performance and boost reliability of neural network models. However, its behavior on agentic, multi-step tasks remains less well-understood: small per-step errors ca…

Hive: A Multi-Agent Infrastructure for Algorithm- and Task-Level Scaling

2026-04-19 · Zizhang Luo, Yuhao Luo, Youwei Xiao, Yansong Xu 외 arxiv

Large language models are increasingly deployed as complex agentic systems that scale with task complexity. While prior work has extensively explored model- and system-level scaling, algorithm- and task-level scaling rem…

DeepSeq: High-Throughput Single-Cell RNA Sequencing Data Labeling via Web Search-Augmented Agentic Generative AI Foundation Models

2025-06-14 · Saleem A. Al Dajani, Abel Sanchez, John R. Williams

Generative AI foundation models offer transformative potential for processing structured biological data, particularly in single-cell RNA sequencing, where datasets are rapidly scaling toward billions of cells. We propos…

Latency and Token-Aware Test-Time Compute

2025-09-11 · Jenny Y. Huang, Mehul Damani, Yousef El-Kurdi, Ramon Astudillo 외 arxiv

Inference-time scaling has emerged as a powerful way to improve large language model (LLM) performance by generating multiple candidate responses and selecting among them. However, existing work on dynamic allocation for…

Position: Agentic Evolution is the Path to Evolving LLMs

2026-01-30 · Minhua Lin, Hanqing Lu, Zhan Shi, Bing He 외 arxiv

As Large Language Models (LLMs) move from curated training sets into open-ended real-world environments, a fundamental limitation emerges: static training cannot keep pace with continual deployment environment change. Sc…