QAMPARI

anchor

Findings

• IC-1319Larger LLaMA and LLaMA2 models show better calibration on phrase-level tasks but not consistently on sentence- and paragraph-level tasks [eval]
• IC-170GPT-4, GPT-3.5, and Claude-3.5-Sonnet rely heavily on parametric knowledge in RAG settings, producing ungrounded responses with high answered ratios and low trust-scores [eval]
• IC-171ICL prompting produces binary response patterns in released LLMs, with answered ratios collapsing to near 0% or 100% rather than calibrated refusal, making prompting ineffective for RAG groundedness [eval]