Modelpedia
Work in progress
About
Findings
Models
Concepts
Methods
Datasets
Sources
Light
Dark
Context-Parametric Inversion: Why Instruction Finetuning May Not Actually Improve Context Reliance
2025-01-22
· ICLR 2025 Oral ·
anchor
Findings
IC-248
Instruction fine-tuning causes context reliance under knowledge conflicts to initially increase then decrease (context-parametric inversion) in Llama2-7B, Pythia-6.9B, and Mistral-7B