5 papers
Same Beliefs, Different Policies: Learning Reusable Representations for Early Stopping in LLM Reasoning
Jingwei Ni, Dmitry Abulkhanov, Vladislav Smirnov +5
Reasoning LLMs often continue reasoning after reaching a correct answer, wasting computation and sometimes replacing it with an incorrect one. Deciding when to stop is a metareason…
Can LLMs Catch a Rigged Backtest? A Clean-Control Calibration Benchmark
Makar Ulesov, Vladislav Smirnov, Omar Ibrahim +1
Backtest auditing is a calibration problem: high flaw recall is not useful when the model falsely flags matched clean strategies. We build a 96-item paired benchmark in which every…
Bayesian control for coding agents
Theodore Papamarkou, Vladislav Smirnov, Viktor Mazanov +4
Modern coding agents pair LLM generators with various tools, including cheap diagnostics and expensive verifiers. The tool-use decisions are typically governed by orchestrators tha…
ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning
Vladislav Smirnov, Chieu Nguyen, Sergey Senichev +14
Test-time compute (TTC) scaling has emerged as a powerful paradigm for improving large language model (LLM) reasoning by allocating additional compute during inference, e.g., via m…
Uncertainty Quantification for Large Language Diffusion Models
Artem Vazhentsev, Vladislav Smirnov, David Li +3
Large Language Diffusion Models (LLDMs) are emerging as an alternative to autoregressive models, offering faster inference through higher parallelism. Similar to autoregressive LLM…