Call for Papers
Foundation models are transforming AI, yet their resource demands are widening the gap between research breakthroughs and practical deployment. Closing this gap requires collaboration across traditionally separate communities of ML, systems, and hardware. AdaptFM offers a dedicated space for interdisciplinary exchange, bringing together diverse perspectives to tackle resource-efficient foundation model inference.
We welcome contributions in the following research areas:
- Flexible & Elastic Architectures: Once-for-all and supernet approaches, slimmable & any-width/any-depth networks, Matryoshka representation learning, MatFormer and elastic transformers, layer skipping and dynamic width, modular and reconfigurable architectures
- Adaptive Test-Time Compute: Input-adaptive inference, adaptive reasoning, universal transformer & recursive inference
- Dynamic Networks & Routing: Early-exit & adaptive-depth networks, MoEs & conditional computation, cascade & routing systems
- Efficient Decoding & Token-Level Adaptation: Speculative & parallel decoding, token pruning and merging, adaptive token computation, adaptive attention mechanisms, dynamic KV cache compression
- Model Compression & Optimization: Quantization, pruning, and sparsity, knowledge distillation, low-rank decomposition and factorization, hardware-aware model optimization
- Systems & Hardware: H/W–S/W co-design for adaptive inference, runtime systems for flexible computation, benchmarking and profiling across resource budgets
- Analysis & Trade-offs: Quality-resource tradeoff analysis, energy-efficient and sustainable inference, emerging applications of adaptive inference
Submissions
Solicited submissions include both full technical workshop papers and short position/experience papers. Maximum length is 6 pages (excluding references) using the official ICML'26 template. Authors may use as many pages of appendices as they wish, but reviewers are not required to read the appendix.
Submissions are non-archival, and we accept novel work that is under active submission to other venues, but not previously published. We actively discourage submissions that are simultaneously submitted to other ICML '26 workshops.
All submissions should be double-blind and will be peer-reviewed. For anonymity purposes, you must remove any author names and other uniquely identifying features in your submitted paper.
Authors may use generative AI tools (e.g. LLMs) to assist with writing, but bear full responsibility for all content. LLMs may not be listed as authors. Authors are encouraged to disclose any significant use of generative AI in their methodology. The usual standards of accuracy, originality, and scientific integrity apply.
Submission portal: openreview.net.
Any questions regarding submission issues should be directed to: Deepak Gupta or Stefanos Laskaridis.
Important Dates
| Paper Submission: | May 8, 2026 AoE |
| Notification of Acceptance: | May 30, 2026 AoE |
| Camera ready: | June 10, 2026 AoE |
| Workshop Event: | July 11, 2026 |