HelpSteer2

A new arXiv paper (2609.10996v1) reports that verbalized confidence has become the more robust soft-scoring mechanism for LLM-as-a-Judge on post-2025 proprietary models, reversing the standard advice …

A new arXiv paper introduces a scalable, inference-only data valuation pipeline that approximates the Shapley value to audit LLM alignment datasets, reducing manual audit search space by 99.1% on Help…

Anthropic's 2022 preference-training dataset (169K examples) and Nvidia's 2024 HelpSteer2 dataset, both public on Hugging Face, reveal that AI personalities are installed through specific, readable tr…