PICARD - Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Models. PICARD is a ServiceNow Research project that was started at Element AI.
first of all, thank you for the great work!
So, I have 2 tables from which i can create an sqlite db. As seen in your instructions, do i just need to place them in under database directory or there are additional steps to it? In general if you can comment how is the performance of t5-picard on unseen tables directly as I do not have dataset to fine tune the model. I haven't yet tried on how hard it is to create a fine tuning dataset. I would appreciate your comments on it.
Hi @tscholak ,
first of all, thank you for the great work! So, I have 2 tables from which i can create an sqlite db. As seen in your instructions, do i just need to place them in under database directory or there are additional steps to it? In general if you can comment how is the performance of t5-picard on unseen tables directly as I do not have dataset to fine tune the model. I haven't yet tried on how hard it is to create a fine tuning dataset. I would appreciate your comments on it.