TaoMedAI/RareSeek-R1
RareSeek-R1 is a 70 billion parameter, domain-specialized large language model developed by TaoMedAI, built upon the DeepSeek-R1-Distill-Llama-70B base. It is specifically tailored for rare disease diagnostic reasoning and clinical applications, utilizing a 32768 token context length. The model was instruction-tuned on the multi-source RareMed-Corpus and fine-tuned on RareMed-CoT to instill explicit, stepwise clinical reasoning. It excels at processing complex medical information for diagnostic support in rare diseases.
Loading preview...
RareSeek-R1: Specialized for Rare Disease Diagnosis
RareSeek-R1, developed by TaoMedAI, is a 70 billion parameter large language model built on the DeepSeek-R1-Distill-Llama-70B architecture. It is uniquely designed for rare disease diagnostic reasoning and clinical applications, supporting both English and Chinese languages.
Key Capabilities & Training
- Domain Specialization: Tailored specifically for the complexities of rare disease diagnosis.
- Progressive Parameter-Efficient Transfer Learning: Utilizes an advanced framework for efficient and targeted learning.
- Instruction-Tuned on RareMed-Corpus: Initially trained on a comprehensive, multi-source dataset integrating medical textbooks, guidelines, biomedical literature, and real-world Electronic Health Record (EHR) narratives.
- Fine-Tuned on RareMed-CoT: Further refined using a high-fidelity corpus to embed explicit, stepwise clinical reasoning processes, mirroring actual diagnostic workflows.
- Extensive Context Window: Features a 32768 token context length, enabling the processing of lengthy and detailed clinical information.
Datasets & Resources
The model's training leveraged the extensive RareMedData dataset, which is publicly accessible. This dataset is crucial for understanding the model's foundation in clinical knowledge.
Ideal Use Cases
RareSeek-R1 is particularly well-suited for scenarios requiring deep clinical reasoning in the context of rare diseases, offering support for diagnostic processes and information retrieval from complex medical texts.