• Palmyra-Med (New Medical LLM) –
  • The writing team introduces a specialized healthcare model and achieves an average of 85.9% in all medical benchmarks.
• Can Medical LLMs Be Fooled? (Microsoft Research) –
  • Research into vulnerabilities in medical LLMs, highlighting risks such as patient data leakage.
• Improving RAG in medicine –
  • Researchers propose i-MedRAG, an iterative Retrieval-Augmented Generation method for medical LLMs.
• CollectiveSFT –
  • Chinese researchers improve the performance of medical LLMs using diverse, well-distributed datasets for fine-tuning.
• MedExpQA –
  • Researchers introduce first multilingual benchmark for medical question answering using LLMs.
• Medical LLM Preparatory Courses –
  • Study compares domain-specific and mixed-domain pre-degree programs for medical LLMs.
• …

View the full thread https://x.com/OpenlifesciAI/status/1822091419848237387

https://preview.redd.it/o3jvwixt8rhd1.png?width=1476&format=png&auto=webp&s=7f70f2876c7572ad13ef6cadd6202351555e617c