Full Papers
Main submissions must be 4–9 pages (excluding references), with optional technical appendices (no page limit) and up to 100MB of supplementary material. Accepted papers receive an additional page for camera-ready version.
Machine Learning/CORE Unranked
ICLR 2026
ICLR 2026 Workshop on Foundation Models for Science: Real-World Impact and Science-First Design
The FM4Science workshop at ICLR 2026 focuses on advancing foundation models for scientific discovery across physical, life, and Earth sciences. It aims to develop models that incorporate physical laws, causal structure, and multi-modal data while integrating with classical scientific tools to ensure reliability, interpretability, and real-world impact.
Key deadlines
Verified September 9, 2026Abstract registration
February 9, 2026
AoE
Full paper
February 11, 2026
AoE
Workshop timeline
Submission and decisions
Abstract registrationKey deadline
February 9, 2026 · AoE
Full paperKey deadline
February 11, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short Paper Track
Dedicated track for early-stage, high-potential ideas. Papers must be original and not AI-generated. Selected papers will be presented as posters or short talks.
Research areas in scope
01
Core Research Themes
Designing datasets, architectures, and training algorithms for better scaling laws in scientific foundation modelsData augmentation and multi-modal self-supervised pretraining for scientific problemsEfficient fine-tuning with scientific awarenessQuantifying and reducing uncertainty in scientific foundation modelsImproving out-of-distribution generalizationIntegrating foundation models with classical scientific tools (simulators, solvers)Leveraging LLM reasoning and in-context learning for scientific tasksCombining symbolic learning and data-driven learningUsing foundation models to facilitate scientific visualizationAccelerating scientific discovery and data collection/assimilationDiagnosing failure modes of scientific foundation modelsAligning foundation models with scientific facts to prevent hallucination
02
Scientific Domains
Astrophysics and Space ScienceBiomedicine (e.g., proteins, biosequences, virtual screening)Computational Science (e.g., PDEs, forecasting)Earth ScienceMaterials Science (e.g., batteries, chemical synthesis)Quantum Mechanics (e.g., nuclear fusion)Small Molecules
Policies worth checking twice
- Submissions must be anonymized for double-blind review; no identifying information in text, figures, or supplementary material.
- Authors must not use 'our previous work' in citations—must anonymize self-references (e.g., 'Smith et al. [1]').
- Abstract submission is mandatory and must be substantive; placeholder or substantially changed abstracts risk rejection.
- Author list cannot be changed after abstract deadline; additions/removals require program chair approval.
- AI-generated papers are prohibited in the short paper track.
- Authors must fully disclose use of LLMs if they played a role in methodology, data processing, or writing.
- Authors are fully responsible for all content, including correctness and originality of text and figures.
- Supplementary material (code, data) must be anonymized and may be up to 100MB.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026