Guangyu LLM: An AI Model Combining LLMs and Symbolic Reasoning 130

Guangyu Large Model is an AI large model released by Infinite Light Years, combining large language models with symbolic reasoning, and integrating visual and language processing technologies. Guangyu Large Model consists of a visual encoder, a visual feature adapter, and a large language model, with 1 billion visual model parameters and 13 billion language model parameters. Guangyu Large Model performs excellently in vertical domains such as finance and healthcare, effectively solving the hallucination problem and enhancing the model's credibility and reasoning accuracy.

Core Features of Guangyu Large Model:

• Gray-box Trustworthiness Technology: Guangyu Large Model ensures the stability and reliability of model outputs through gray-box trustworthiness technology, which is particularly important in fields such as finance and healthcare that demand extremely high accuracy.

• Hallucination Problem Solution: Through innovative technologies such as neuro-symbolic computation, it effectively reduces the inauthenticity of generated content and enhances user trust in model outputs.

• Vertical Domain Empowerment: Specifically optimized for professional scenarios such as financial services and medical diagnosis, improving reasoning accuracy, reducing service costs, and achieving productivity transformation of AI technology.

• Parameter Scale and Performance: Although the specific parameter scale has not been detailed, the trustworthy Guangyu Large Model is described as a "specialized" ten-billion-scale model, indicating that its performance in specific domains may surpass certain ultra-large-scale models, such as GPT-4 Turbo, demonstrating advantages in specific evaluations.

• Medical Domain Application: Guangyu Medical Large Model is a branch of Guangyu Large Model, developed in collaboration with Zhongshan Hospital affiliated with Fudan University, customized for medical scenarios. Its performance in knowledge, logical reasoning, etc. is leading, comparable to GPT-4's capabilities in medical industry assessments.

Guangyu Industry Large Models

• Guangyu Jinfan: A financial large model based on fully self-developed large model technology, with industry-leading capabilities in financial report parsing, table analysis, and information interpretation, actively empowering scenarios such as investment research and knowledge base search, helping customers build intelligent financial services in the era of large models.

• Guangyu Huiyi: Guangyu Medical Large Model, based on self-developed large model technology, extending from health management scenarios such as physical examination report intelligent generation systems and physical examination report interpretation medical scenario applications to full medical scenario applications.

• Guangyu Huiyan: An advanced AI Chatbot tool based on Guangyu Large Model, which can understand human language and provide smooth natural language interaction experiences for individuals and enterprises.
  

Application Scenarios of Guangyu Large Model

• Financial Services: Guangyu Financial Large Model achieves higher accuracy than GPT-4 Turbo and Llama3 on the CFA exam dataset, capable of providing precise search, efficient information summarization, paradigm writing, and Agent management comprehensive functions, and has served many first-tier financial institutions.

• Medical Diagnosis: Guangyu Medical Large Model ranks first on the Chinese medical domain MedBench leaderboard, capable of simulating doctors in physical examination report interpretation, disease diagnosis, etc., and has been deployed in many first-tier hospitals.

Advantages of Guangyu Large Model:

Improved Reasoning Accuracy: In evaluations within the vertical domains of finance and healthcare, Guangyu Large Model has outperformed many ultra-large-scale models, such as GPT-4 Turbo, effectively improving reasoning accuracy.

Reduced Service Costs: By enhancing model credibility and accuracy, Guangyu Large Model helps reduce service costs for enterprises in AI applications.

Infinite Light Years is an innovative company focused on large model technology and applications, dedicated to developing trustworthy large models in vertical domains and creating AI-native products and solutions that deeply address industry needs. The Infinite Light Years team consists of technology and industry experts from renowned tech companies such as Alibaba, Ant Group, Microsoft, ByteDance, Tencent, and Oracle, all graduated from prestigious universities both domestically and internationally, including MIT, Cornell, Berkeley, UCSD, UMICH, UCL, NUS, Tsinghua, Peking University, Fudan, Shanghai Jiao Tong University, and Zhejiang University, with over 70% holding master's or doctoral degrees.