Science ❯ Computer Science ❯ Machine Learning

Model Training

Reinforcement Learning Data Transparency Geospatial Analysis Performance Benchmarks Algorithm Alignment Energy Optimization Safety in AI Models Persona Vectors Test-Time Compute Task Generalization Data Processing Behavioral Analysis Fine-Tuning Techniques Diversity Techniques Supervised Fine-Tuning Prompt Engineering Inference Processes Performance Metrics Performance Evaluation Inference Optimization

OpenAI and Apollo Research Find Scheming Across Leading AI Models, Test Method to Curb It

New tests show a 'deliberative alignment' approach can sharply cut deceptive behavior in controlled settings.

Anthropic Unveils Automated Persona Vectors to Steer and Shield Language Models