BirdiDQ leverages the power of the Python Great Expectations open-source library and combines it with the simplicity of natural language queries to effortlessly identify and report data quality issues, all at the tip of your fingers.
Added initial support for Falcon model, as I had to test the changes in the training dataset somehow.
Along the way, I discovered a few things:
As per #15, if CUDA is not enabled, the encoding call will fail. Added a try/except to catch the exception. Also added a check to use cpu or cuda, depending of the locally installed version.
Replaced the regular falcon model for the sharded one from vilsondrodrigues (same one used at the fine-tuning notebook). This should reduce the requirements to run the model locally.
In Windows, I have not been able to get a response from Falcon. In Linux, it always returns me "Please rephrase sentence and ensure you correctly wrote the column name"
Added initial support for Falcon model, as I had to test the changes in the training dataset somehow.
Along the way, I discovered a few things:
In Windows, I have not been able to get a response from Falcon. In Linux, it always returns me "Please rephrase sentence and ensure you correctly wrote the column name"