Static Benchmarks Are Broken: The Case for Dynamic Evaluation of LLMs
- Farhan Ahmed
- Chad Deluca
- 2026
- ICML 2026
Workshop paper
Publications
1677 results for Machine Learning
DropKV: Decoupling Residual-Output Perturbation for Near-Optimal KV-Cache Eviction
- Aozhong Zhang
- Selcuk Gurses
- et al.
- 2026
- ICML 2026
Workshop paperHGR-TabE: Universal Tabular Embeddings via Maximal Correlation Alignment
- Niharika DSouza
- Liane Vogel
- et al.
- 2026
- ICML 2026
Workshop paperLearning to Mediate Equilibrium Selection in LLM Games
- Miao Liu
- Matthew Riemer
- et al.
- 2026
- ICML 2026
Workshop paperAgainst the Monolithic Wireless World Model: Why NextG Needs Composable and Agentic Intelligence
- Aladin Djuhera
- Farhan Ahmed
- et al.
- 2026
- ICML 2026
Workshop paperModel Merging via Averaged Representational Similarity
- Christopher Wang
- Vighnesh Subramaniam
- et al.
- 2026
- ICML 2026
WorkshopAraVQA: Building a New Arabic Factoid Visual Question Answering Dataset from Wikipedia
- Sultan Alrowili
- Younes Samih
- et al.
- 2026
- ACL 2026
Conference paperStop Guessing When to Stop Testing: Efficient Model Evaluation with Just Enough Data
- Ofir Arviv
- Kristjan Greenewald
- et al.
- 2026
- ACL 2026
Conference paperBeyond Outcome Verification: Verifiable Process Reward Models for Structured Reasoning
- Massimiliano Pronesti
- Anya Belz
- et al.
- 2026
- ACL 2026
PaperSTaD: Scaffolded Task Design for Identifying Compositional Skill Gaps in LLMs
- 2026
- ACL 2026
Paper