model and quantized model both with attention

replace fp32 model with one with attention layers

onnx model with attention

include output_attentions in tokenizer_config.json

include output_attentions config in config.json

rename q8 model

add quantized models

Imported model from jaxmef repo

initial commit