Open-weight model outputs CAN be published in full (unlike closed models). Add an open-weight scribe baseline (e.g. MedGemma or Llama via an OpenAI-compatible endpoint) and, because the license permits it, publish the generated notes alongside the scores so readers can see exactly what was fabricated. Great first contribution.
Open-weight model outputs CAN be published in full (unlike closed models). Add an open-weight scribe baseline (e.g. MedGemma or Llama via an OpenAI-compatible endpoint) and, because the license permits it, publish the generated notes alongside the scores so readers can see exactly what was fabricated. Great first contribution.