We took the Tinyllama formate and fine-tuned the model with a history focus.
Model Description
This model was fined tuned using a dataset based on the opensource textbooks of the American Yawp and the OpenStax US History. Questions and answer pairs from the dataset were
generated using Claude.ai and ChatGPT 3.5.
Updated:
v1.5 is an updated from v1.1 by training on the new dataset generated by ambrosfitz/mighty-history-merge.
I have included the python script to generate most of this data at the might-history-merge dataset. Take a look if you are needing
a way to generate text from ChatGPT or from books/material you need to generate Q&A from.
Developed by: ambrosfitz
Model type: llama
Language(s) (NLP): English
License: [More Information Needed]
Finetuned from model: Tinyllama
Uses
The purpose of this model is to facilitate a more fine-tuned and specific model for questions on history. Further version will focus on opensource history journals,
primarly from an American History persepective.
