Hi Srini,
Do you have more than 100k instances? I found, in your google drive, the data is already truncated to 100k. Looks like CONCODE is a well-established dataset for code gen task. I am wondering if I can explore more CONCODE data. Btw, how the research going?
Thanks,