Agentic evolution of physically constrained foundation models

• URL: http://arxiv.org/abs/2606.25532v1
• Date: Wed, 24 Jun 2026 08:07:59 GMT
• Title: Agentic evolution of physically constrained foundation models
• Abstract summary: Artificial intelligence increasingly drives automated scientific discovery, yet contemporary generalist agents lack physical grounding.<n>We present a physically grounded, multi-agent discovery engine that autonomously architects hardware-compliant computing systems.
• Score: 26.896066629768494
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Artificial intelligence increasingly drives automated scientific discovery, yet contemporary generalist agents lack physical grounding, frequently hallucinating hardware-incompatible designs. Here, we present a physically grounded, multi-agent discovery engine that autonomously architects hardware-compliant computing systems. Anchored by an Evolutionary Knowledge Graph structuring past scientific innovations, the framework extracts an "algorithmic Chain-of-Thought" to transform blind stochastic search into directed structural evolution. Applied to the extreme testbed of foundation model deployment, the engine evolved two hardware-aware compression methodologies surpassing human-engineered heuristics: Q-Enhance mitigates long-context accuracy loss in dense models, and MoE-Salient-AQ outperforms state-of-the-art manual sparse Mixture-of-Experts designs by 3.7% at sub-3-bit regimes. Utilizing a bandwidth-efficient Sensitivity Profile, we successfully deployed a massive 235-billion-parameter model onto a constrained dual-A100 server, reducing memory requirements by 75% with a marginal 0.64% accuracy degradation. By transforming unconstrained combinatorial search into knowledge-driven autonomy, this establishes a scalable hardware-software co-design paradigm for machine-driven discovery within strict physical boundaries.

Related papers

• Position: Weight Space Should Be a First-Class Generative AI Modality [67.10702320971893]
  This paper argues that model checkpoints should be treated as a first-class data modality.<n>It argues that generative modeling in weight space should be standardized as a core machine learning primitive.<n>Our goal is to shift the community's default mindset from optimizing models per task to sampling models from learned weight distributions.
  arXiv  Detail & Related papers  (2026-05-18T16:38:26Z)
• Generative AI for material design: A mechanics perspective from burgers to matter [0.0]
  We show that diffusion-based generative AI and computational mechanics are rooted in the same principles.<n>We illustrate this connection using a three-ingredient burger as a minimal benchmark for material design in a low-dimensional space.<n>We generate five new burgers and validate them in a restaurant-based sensory study with 100 participants.
  arXiv  Detail & Related papers  (2026-04-03T19:12:01Z)
• Shape Control of a Planar Hyper-Redundant Robot via Hybrid Kinematics-Informed and Learning-based Approach [42.55947704321053]
  Hyper-redundant robots offer high dexterity, making them good at operating in confined and unstructured environments.<n>To extend the reachable workspace, we built a multi-segment flexible rack actuated planar robot.<n>We propose a hybrid kinematics-informed and learning-based shape control method, named SpatioCoupledNet.
  arXiv  Detail & Related papers  (2026-03-11T04:35:06Z)
• Transolver-3: Scaling Up Transformer Solvers to Industrial-Scale Geometries [51.028432812178266]
  Transolver-3 is a new member of the Transolver family designed for high-fidelity physics simulations.<n>We show that Transolver-3 is capable of handling meshes with over 160 million cells, achieving impressive performance across three challenging simulation benchmarks.
  arXiv  Detail & Related papers  (2026-02-04T16:52:44Z)
• Cheap2Rich: A Multi-Fidelity Framework for Data Assimilation and System Identification of Multiscale Physics -- Rotating Detonation Engines [1.8796659304823702]
  Cheap2Rich is a multi-scale data assimilation framework that reconstructs high-fidelity state spaces from sparse sensor histories.<n>We demonstrate the performance on rotating detonation engines (RDEs)<n>Results highlight a general multi-fidelity framework for data assimilation and system identification in complex multi-scale systems.
  arXiv  Detail & Related papers  (2026-01-28T06:35:22Z)
• GENIUS: An Agentic AI Framework for Autonomous Design and Execution of Simulation Protocols [32.505127447635864]
  GENIUS is an AI-agentic workflow that fuses a smart Quantum ESPRESSO knowledge graph with a tiered hierarchy of large language models supervised by a finite-state error-recovery machine.<n> GENIUS translates free-form human-generated prompts into validated input files that run to completion on $approx$80% of 295 diverse benchmarks, where 76% are autonomously repaired, with success decaying exponentially to a 7% baseline.<n>The framework democratizes electronic-structure DFT simulations by intelligently automating protocol generation, validation, and repair, opening large-scale screening and accelerating ICME design loops across academia and industry worldwide.
  arXiv  Detail & Related papers  (2025-12-06T11:28:35Z)
• End-to-End Analysis of Charge Stability Diagrams with Transformers [0.7570270968488804]
  Transformer models and end-to-end learning frameworks are rapidly revolutionizing the field of artificial intelligence.<n>In this work, we apply object detection transformers to analyze charge stability diagrams in semiconductor quantum dot arrays.<n>We show that it surpasses convolutional neural networks in performance on three different spin qubit architectures.
  arXiv  Detail & Related papers  (2025-08-21T16:54:22Z)
• Mechanistic Design and Scaling of Hybrid Architectures [114.3129802943915]
  We identify and test new hybrid architectures constructed from a variety of computational primitives. We experimentally validate the resulting architectures via an extensive compute-optimal and a new state-optimal scaling law analysis. We find MAD synthetics to correlate with compute-optimal perplexity, enabling accurate evaluation of new architectures.
  arXiv  Detail & Related papers  (2024-03-26T16:33:12Z)
• Random resistive memory-based deep extreme point learning machine for unified visual processing [67.51600474104171]
  We propose a novel hardware-software co-design, random resistive memory-based deep extreme point learning machine (DEPLM) Our co-design system achieves huge energy efficiency improvements and training cost reduction when compared to conventional systems.
  arXiv  Detail & Related papers  (2023-12-14T09:46:16Z)
• OutlierNets: Highly Compact Deep Autoencoder Network Architectures for On-Device Acoustic Anomaly Detection [77.23388080452987]
  Human operators often diagnose industrial machinery via anomalous sounds. Deep learning-driven anomaly detection methods often require an extensive amount of computational resources which prohibits their deployment in factories. Here we explore a machine-driven design exploration strategy to create OutlierNets, a family of highly compact deep convolutional autoencoder network architectures.
  arXiv  Detail & Related papers  (2021-03-31T04:09:30Z)
• TELESTO: A Graph Neural Network Model for Anomaly Classification in Cloud Services [77.454688257702]
  Machine learning (ML) and artificial intelligence (AI) are applied on IT system operation and maintenance. One direction aims at the recognition of re-occurring anomaly types to enable remediation automation. We propose a method that is invariant to dimensionality changes of given data.
  arXiv  Detail & Related papers  (2021-02-25T14:24:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.