CAST reduces object hallucination in LVLMs by 6.03% on average across five models and five benchmarks by identifying caption-sensitive attention heads and applying optimized steering directions to their outputs, with negligible added inference cost.
Towards a unified multi-dimensional evaluator for text generation.arXiv preprint arXiv:2210.07197
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
representative citing papers
A proposed pipeline shows LLMs introduce detectable race and gender biases when summarizing life narratives, creating potential for representational harm in research.
Multi-agent debate among LLMs yields more reliable text evaluations than single-agent prompting by simulating collaborative human judgment.
G-Eval uses GPT-4 with chain-of-thought and form-filling to reach 0.514 Spearman correlation with humans on summarization, beating prior NLG metrics while noting a bias toward LLM outputs.
A reference-free proxy scoring framework combined with GIRB calibration produces better-aligned evaluation metrics for summarization and outperforms baselines across seven datasets.
citing papers explorer
-
CAST: Mitigating Object Hallucination in Large Vision-Language Models via Caption-Guided Visual Attention Steering
CAST reduces object hallucination in LVLMs by 6.03% on average across five models and five benchmarks by identifying caption-sensitive attention heads and applying optimized steering directions to their outputs, with negligible added inference cost.
-
Whose Story Gets Told? Positionality and Bias in LLM Summaries of Life Narratives
A proposed pipeline shows LLMs introduce detectable race and gender biases when summarizing life narratives, creating potential for representational harm in research.
-
ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate
Multi-agent debate among LLMs yields more reliable text evaluations than single-agent prompting by simulating collaborative human judgment.
-
G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment
G-Eval uses GPT-4 with chain-of-thought and form-filling to reach 0.514 Spearman correlation with humans on summarization, beating prior NLG metrics while noting a bias toward LLM outputs.
-
Calibrating Model-Based Evaluation Metrics for Summarization
A reference-free proxy scoring framework combined with GIRB calibration produces better-aligned evaluation metrics for summarization and outperforms baselines across seven datasets.