Hacker News new | past | comments | ask | show | jobs | submit login

I imagine fullwedgewhale was picturing them as being randomly distributed, not all on the same day every quarter.



And I imagine lallysingh meant that including April 1 headlines in the test corpus could help train the algorithms to correctly interpret false headlines, whether those headlines occur on a single day or not.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: