Word Embeddings Model - 1830s Text
This model provides word embeddings trained on historical text from the 1830s, specifically focusing on the Book of Mormon text. The embeddings capture semantic relationships between words in the context of early 19th-century religious literature.
Model Details
- Embedding dimension: 384
- Training corpus: Book of Mormon (1830 edition)
- Architecture: Transformer-based embedding model
- Context window size: 256
- Vocabulary size: 5999
Usage
from huggingface_hub import from_pretrained
model = WordEmbeddingModel.from_pretrained('tx-smitht/word-embeddings-1830')
Training Data
The model was trained on the 1830 edition of the Book of Mormon, representing historical English language patterns from the early 19th century.
Limitations
- Domain-specific vocabulary focused on religious and historical text
- Limited to early 19th-century American English language patterns
- Downloads last month
- 41
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API:
The model has no library tag.