active-qa icon indicating copy to clipboard operation
active-qa copied to clipboard
verified publisher icon google

Model pretrained on UN and Paralex datasets

Open kaustumbh7 opened this issue 6 years ago • 0 comments

Reformulator Training We first train reformulator from a model pretrained on UN and Paralex datasets. It should take a week on a single P100 GPU to reach ~42 F1 score on SearchQA's dev set.

@rodrigonogueira4 How to make the model (pretrained on UN and Paralex datasets) from scratch on a different dataset ?

kaustumbh7 avatar Mar 15 '19 09:03 kaustumbh7