Back to papers
October 27, 2026cs.LGcs.CV

When Visual Evidence Is Insufficient: How Instructions, Post-Training, and Internal Activations Shape Refusal in Multimodal Language Models

Ananth K Suresh, Aveen Dayal, Gouthaman KV, Aditya Pola, Vineeth N Balasubramanian
Read on arXivDownload PDF
Categories

cs.LG, cs.CV

When Visual Evidence Is Insufficient: How Instructions, Post-Training, and Internal | One9Founders