Writer Introduces Specialized AI Models for Healthcare and Finance

Writer has announced the release of two new domain-specific AI models, Palmyra-Med 70B and Palmyra-Fin 70B, significantly expanding the capabilities of NVIDIA NIM, according to NVIDIA Technical Blog. These models bring unparalleled accuracy to medical and financial generative AI applications, outperforming other models like GPT-4, Med-PaLM 2, and Claude 3.5 Sonnet.

Enhanced Capabilities for Regulated Industries

While general-purpose large language models (LLMs) have been the focus of recent advancements, specialized models like Palmyra-Med 70B and Palmyra-Fin 70B offer improved accuracy and domain knowledge, making them ideal for complex, regulated industries such as finance and healthcare. These models are now available as NVIDIA NIM microservices, improving their composability with preconfigured containers that can be deployed across various NVIDIA-accelerated platforms, including cloud, data center, and local environments.

Optimized Performance with NVIDIA AI Software

Both Palmyra-Med 70B and Palmyra-Fin 70B have benefitted from NVIDIA AI software optimizations. Using NVIDIA TensorRT-LLM, the models have seen a reduction in inference latency (TTFT) by 23% and 30%, respectively, and a ~60% increase in the rate of token return (TPS), resulting in a more responsive prompting experience.

Improving Patient Outcomes with Medical Accuracy

Palmyra-Med 70B is the latest iteration of Writer’s healthcare model, achieving an average of 85.9% across all medical benchmarks, surpassing Med-PaLM 2 by close to 2 percentage points. This model excels in various medical disciplines, including clinical knowledge, anatomy, genetics, and biomedical research, making it a valuable tool for diagnostic accuracy, treatment planning, genetic counseling, and medical education.

	Palmyra-Med	Med-PaLM 2 (5-shot)	GPT-4	Gemini 1.0	GPT-3.5 Turbo
MMLU Clinical Knowledge	90.9	88.3	86	76.7	74.7
MMLU Medical Genetics	94	90	91	75.8	74
MMLU Anatomy	83.7	77.8	80	66.7	72.8
MMLU College Medicine	84.4	80.9	76.9	69.2	64.7
PubMedQA	79.6	79.2	75.2	70.7	72.7
Average*	85.9	84.1	82.8	70.8	66

Table 1. Comprehensive medical MMLU benchmark comparison of popular models
*Average performance is measured from all nine tests

Writer collaborates with leading healthcare companies to leverage Palmyra-Med 70B in various medical applications, including clinical decision support, clinical trial protocol development, and medical document generation.

A Powerful LLM for Finance

Palmyra-Fin 70B addresses the unique challenges of the financial sector, such as lengthy financial statements, complex terminology, and nuanced market analysis. This model supports financial trend analysis, investment evaluation, risk assessment, and asset allocation strategy. Notably, Palmyra-Fin scored 73% on the CFA Level III exam, marking it as the first model to pass this rigorous test, which typically has a passing score of around 60%.

The model also excelled in the long-fin-eval benchmark test, outperforming popular models like Claude 3.5 Sonnet and GPT-4, demonstrating its capability to analyze complex financial topics.

*Figure 2. Palmyra-Fin 70B outperformed other models in the long-fin-eval benchmark test*

Getting Started with Palmyra LLMs

Domain-specific LLMs like Palmyra-Med 70B and Palmyra-Fin 70B are set to lead AI innovation, transforming how industries build specialized AI applications. These models offer greater accuracy, efficiency, and improved data management and regulatory compliance. Developers in the medical and financial fields can explore Palmyra-Med 70B and Palmyra-Fin 70B through the NVIDIA API catalog.