Model or Harness? An Interaction-Centric Taxonomy for Localizing Agent Failures

MixtureVitae: Open Web-Scale Pretraining Dataset With High Quality Instruction and Reasoning Data Built from Permissive-First Text Sources

OpenThoughts-Agent: Data Recipes for Agentic Models

Consistency as a Testable Property: Statistical Methods to Evaluate AI Agent Reliability

Improving Consistency in Large Language Models through Chain of Guidance

Mitigating Unsafe Feedback with Learning Constraints

On transfer of adversarial robustness from pretraining to downstream tasks

Evaluating the robustness of biomedical concept normalization

Measuring reliability of large language models through semantic consistency

GANDALF: Gated Adaptive Network for Deep Automated Learning of Features

AskYourDB: An end-to-end system for querying and visualizing relational databases using natural language