Kite is a small, trained, 18 million parameter language model.
It was trained on a tokenized and truncated version of the "multi style" subset of
openbmb/Ultra-FineWeb-L3, using 2 epochs, 32 batch size, 1e-3 learning rate, and the
pika 4 tokenizer.
Also, evaluation on a tokenized and truncated
byunggill/gpt-2-output was done during training.
Due to its size, the model is not suitable for production workloads.
