search

mlcommons / training

Reference implementations of MLPerf™ training benchmarks
https://mlcommons.org/en/groups/training
Apache License 2.0
1.59k stars 553 forks source link

[BERT] Unable to download Tensorflow checkpoint from GCS. #444

Closed ddl-kevin closed 3 years ago

ddl-kevin commented 3 years ago

I do not appear to be able to list or get objects from the indicated bucket pkanwar-bert. Have these files moved?

I have tried several different Google accounts, and using the CLI as well.

cc @pkanwar23

alphaRGB commented 3 years ago

Me too, the given url： https://console.cloud.google.com/storage/browser/pkanwar-bert ， it is empty.

dagrayvid commented 3 years ago

@ddl-kevin @alphaRGB,

Thanks to @TheKanter and @sgpyc these files have been added to the Google Drive with the wikipedia dataset (mentioned in issue 377).

I can see them as a member of mlcommons and the training WG on. I think at least one (maybe both) of these memberships is necessary to access the files.

ddl-kevin commented 3 years ago

Hey @dagrayvid, I'm referring to the other files which are needed: Tensorflow checkpoint, Vocab, and Config, mentioned in the README here: https://github.com/mlcommons/training/tree/master/language_model/tensorflow/bert#location-of-the-input-files

I was able to download the Wikipedia dump itself from the Google Drive that you mentioned.

TheKanter commented 3 years ago

This is a permissions problem. I'm fixing.

On Fri, Feb 5, 2021 at 8:10 AM Kevin notifications@github.com wrote:

Hey @dagrayvid https://github.com/dagrayvid, I'm referring to the other files which are needed: Tensorflow checkpoint, Vocab, and Config, mentioned in the README here: https://github.com/mlcommons/training/tree/master/language_model/tensorflow/bert#location-of-the-input-files

I was able to download the Wikipedia dump itself from the Google Drive that you mentioned.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/mlcommons/training/issues/444#issuecomment-774128188, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJXLOK2INKWEDLDTHGLUCCLS5QJ7TANCNFSM4WRZRFJA .

dagrayvid commented 3 years ago

@ddl-kevin, these input files are in the drive and visible to me. Thanks @TheKanter for working on making them public!

TheKanter commented 3 years ago

https://drive.google.com/drive/u/4/folders/1oQF4diVHNPCclykwdvQJw8n_VIWwV0PT contains all the files needed for BERT and is now publicly visible. Please let me know if there are any further problems.

ddl-kevin commented 3 years ago

Awesome, I can see them now! Thanks!