Palmyra Med 70B 32K
About
The Palmyra Med 70B 32K is a specialized large language model by Writer, tailored for the healthcare sector. It extends the capabilities of the original Palmyra Med 70B, offering an enhanced context window of 32,768 tokens, allowing it to handle lengthy medical documents and complex healthcare scenarios. Utilizing a Llama-based architecture, the model is fine-tuned with a bespoke medical instruction dataset and the DPO dataset, excel at summarizing clinical notes, recognizing medical entities, and enhancing data retention from various sources. It has surpassed prominent models like GPT-4 in biomedical benchmarks, achieving notable accuracy rates. However, it is not suited for direct patient care without extensive validation and is currently limited to English language applications 145.
