Traditional Chinese Medicine Multimodal Pre-training Corpus

Over 5 billion tokens of Traditional Chinese Medicine text form the largest existing TCM corpus, sourced from websites and books. FreedomIntelligence released this multimodal dataset for pre-training the ShizhenGPT model. It was last updated in September 2025.

Use Cases

• Pre-training a TCM-specific LLM using the 5B token text corpus.
• Training a multimodal model for TCM image-text alignment tasks.
• Fine-tuning models for TCM knowledge retrieval from the curated text data.
• Benchmarking model performance on specialized medical terminology and concepts.

Strengths

• Corpus contains over 5 billion tokens of TCM text.
• Described as the largest existing TCM corpus and image-text dataset.

Limitations

• Specific row counts, column names, and sample data are unavailable.
• Dataset size and file formats are unknown.
• Potential bias towards information available on specific TCM websites and books.

Provenance

Source

FreedomIntelligence.

Collection Method

Collected from TCM-related websites and books.

Freshness

Last updated September 2025.