About this document
SSRN 4367666 by dorjisither112 is a document available to read on EtoBox.
The document discusses the development of a next word prediction model for the Bodhi language using a Long Short-Term Memory (LSTM) approach, addressing the lack of linguistic research in this rare language. The authors trained the model on a dataset of approximately 1000 Bodhi words and achieved a loss of around 0.00040 and an accuracy of 50% after 500 epochs. The study highlights the challenges of processing Bodhi language due to its complexity and the absence of existing translation tools.
- Author
- dorjisither112
- Language
- EN