Hyperparameter setting for Table Retrieval Model

[Doreenruirui]

Dear developers,

Could you please share the hyper-parameters used for the table retrieval model to replicate the results here? In the paper, you mentioned that you used warmup ratio 0.2 to train the large retrieval model, but in the training script you provided the warmup ratio is 0.01. Also, is the hyper-parameter setting for the medium model the same with the large model? Thanks a lot!

Thanks!

[SyrineKrichene]

Hi, Can you give me the exact name of the model/ models that you want to replicate, I can look at our experiments and give you the exact hyper parameters to run.

Thanks, Syrine

On Tue, Aug 24, 2021 at 9:40 PM Doreen @.***> wrote:

Dear developers,

Could you please share the hyper-parameters used for the table retrieval model to replicate the results here https://github.com/google-research/tapas/blob/master/DENSE_TABLE_RETRIEVER.md? In the paper, you mentioned that you used warmup ratio 0.2 to train the large retrieval model, but in the training script you provided the warmup ratio is 0.01. Also, is the hyper-parameter setting for the medium model the same with the large model? Thanks a lot!

Thanks!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/google-research/tapas/issues/135, or unsubscribe https://github.com/notifications/unsubscribe-auth/APARZOMQDWZMKZK6UBWSZUTT6PYSPANCNFSM5CXQWUPA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email .