Can captions teach medical VLMs to see artifacts? (SPIE) Collection Adapters (Qwen3-VL-8B, MedGemma-4B) and evaluation from our SPIE study on caption fine-tuning for medical figure captioning. • 2 items • Updated 18 days ago
Just Ask for a Table: A Thirty-Token User Prompt Defeats Sponsored Recommendations in Twelve LLMs Paper • 2605.12772 • Published May 12 • 1
Just Ask for a Table: A Thirty-Token User Prompt Defeats Sponsored Recommendations in Twelve LLMs Paper • 2605.12772 • Published May 12 • 1