PaliGemma 2
Google DeepMind — Gemma
rich detail 27 method fields published
Evaluation
automatic chest X-ray report generation cast as a long captioning task; fine-tuned on MIMIC-CXR (377k images from 228k radiographic studies) with the same train/validation/test splits as [90]; an LLM (Gemini 1.5 Pro) removes mentions of prior X-rays since the model does not have access to them; measured with RadGraph F1 (entities and relationships vs reference), plus CIDEr, BLEU4 and Rouge-L
How it was run
Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.
rich detail 27 method fields published