Fairseq: Reproduce Billion Word benchmark for paper by Baevski and Auli, 2018.

Created on 22 Jul 2019  路  3Comments  路  Source: pytorch/fairseq

There is only example on how to run for WikiText-103, but not Billion Word, for the adaptive input paper by Baevski and Auli, 2018.

There is a difference between the two, as Billion Word is of short sentences. Can anyone help out?

question
Source
picture yilegu

Most helpful comment

Would be interested in this too!

picture jeremyasapp on 5 Aug 2019
👍2

All 3 comments

@alexeib

huihuifan picture huihuifan on 26 Jul 2019

Would be interested in this too!

jeremyasapp picture jeremyasapp on 5 Aug 2019
👍2

Also interested in the correct setting to reproduce the results.

apeterswu picture apeterswu on 26 Aug 2019
Was this page helpful?
0 / 5 - 0 ratings

Related issues

ajesujoba picture ajesujoba  路  3Comments
Ir1d picture Ir1d  路  3Comments
neel04 picture neel04  路  3Comments
jordiae picture jordiae  路  3Comments
galphag picture galphag  路  3Comments