BPulse · bioengineer.orgbioengineer.org · Nov 7Researchers Propose Decision Tree Framework to Help Developers Select AI Interpretability TechniquesRead more
arXivNov 4Multi-step SWE: Can Coding Agents Keep Building on Their Own Code Over Long Horizons?AI Long-horizon benchmark for coding agents building on their own code.Read more
arXivNov 5Scientist World: Grow a Research World for AI-ScientistsAI Executable research environments for AI-Scientists with evaluation.Read more
arXivNov 5XStack: A Benchmark for ML Repo MigrationAI Benchmark for ML repository migration between frameworks.Read more
arXivNov 5Benchmarks as Software: Terminal-Bench and the Case for Continuous BenchmarksAI Continuous benchmarks for coding agents.Read more
arXivNov 5T3R: Training-Free Two-Stage Token Refinement Towards Efficient and Robust VLA ModelsAI Training-free token refinement for efficient VLA models in robotics.Read more
arXivNov 2Must AI Spell It Out? Pragmatic Comprehension of Alerts in a Time-Critical TaskAI Pragmatic comprehension study for assistive system alerts.Read more
arXivNov 2One Mistake, Three Layers: Persistence Rules Under Depth CorrelationAI Probing and calibration method for depth curves in transformer models.Read more
arXivNov 2Stacking-Based Weighting for Large Language Bayes in M-Open SettingsAI Stacking-based weighting method for Large Language Bayes.Read more
arXivNov 2Ermakov--Based embeddings as a parameter-free layer for physics-informed neural networksAI Physics-informed neural network embeddings for evolution equations.Read more