Learn What's Left, Not What's Mastered: Saturation Aware Advantage Reweighting for Multi-Reward Policy Optimization
Research
Research News
New papers, breakthroughs and academic AI research
TRACE: Training-time Report-guided and Clinically Ordered Concept Editing
Research
Weight Oracles: Reading Neural Network Weights with Language Models
Research
Stepped MoE: Segment-Level Routing with Configurable Inference Complexity
Research
Dynamic Budget Allocation for LLM Evaluation under Hard Resource Constraints
Research
The Law of Stop: Interruptibility, Injunctions, and the Governance of Agentic AI
Research
Language-model ratings of depression reflect the rater more than the patient
Research
Capability-Gated Language Models: Security Composes, Utility Does Not
Research
TriCalRAG: A Three-Strategy, Retrieval-Augmented Benchmark for On-Premise LLM-Based Root Cause Analysis in AIOps
Research
Evaluating System One Models for Agent Security Decisions: Reliability, Calibration, and Selective Automation
Research
Inference and learning in sparse autoencoders as natural gradient flow
Research
Counterfactual Rollout Replay: Forkable Environments as Free Process Rewards for Software Engineering Agents
Research