human annotations
Labels or descriptions provided by human experts to datasets, used in supervised learning tasks. These annotations serve as ground truth for training AI models and evaluating their performance.
- Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond)
- Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond)
- Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning
- Contextual Integrity in LLMs via Reasoning and Reinforcement Learning
- Handling Missing Responses under Cluster Dependence with Applications to Language Model Evaluation
- Is This Tracker On? A Benchmark Protocol for Dynamic Tracking
- OS-Harm: A Benchmark for Measuring Safety of Computer Use Agents
- Single-pass Adaptive Image Tokenization for Minimum Program Search