Tag: adversarial machine learning
Browse our exclusive articles!
AdaRubric: Dynamic Task-Adaptive Rubrics for LLM Evaluation
Discover AdaRubric, a dynamic rubric system that adapts to tasks for accurate evaluation and improved training of LLM agents across diverse benchmarks.
AgentHER: Boost LLM Performance with Trajectory Relabeling
AgentHER enhances LLM agents by relabeling failed trajectories, improving task success rates and data efficiency in AI training.
Value Alignment Tax: Quantifying Trade-offs in LLMs
Discover how the Value Alignment Tax framework measures value trade-offs in large language model alignment for ethical AI development.
Rethinking Ground Truth: Overcoming Bias in Data Annotation
Explore how embracing subjectivity and diverse perspectives improves data annotation and challenges the flawed 'ground truth' in machine learning.
DenoGrad: Enhance Data Quality for Tabular & Time-Series AI
Discover DenoGrad, a gradient-based framework improving data quality in tabular and time-series learning without clean references or strict assumptions.