Conversation
|
It is important to note that since we are using a big model (~67M params) and CPU based systems, the CI checks will take a long time. The example can be executed on a T4 GPU with a duration of (~10 mins). |
|
@kAIto47802 Could you review this PR? |
|
Could you reduce the computational time by using a much smaller model? |
|
I have updated the example to run on a much smaller model. Now the CI is being completed in a reasonable amount of time :) cc: @nabenabe0928 |
|
This pull request has not seen any recent activity. |
|
Not stale. |
|
Could you please review the changes? cc: @kAIto47802 |
|
This pull request has not seen any recent activity. |
|
Not stale. |
|
Could you please review? cc: @kAIto47802 |
nzw0301
left a comment
There was a problem hiding this comment.
I'm using transformers nowadays, so let me leave a few suggestions.
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
|
Let me change the reviewer since @nzw0301 started the review for this PR. |
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
nzw0301
left a comment
There was a problem hiding this comment.
Thank you for quickly updating the PR. I would appreciate it if you followed my suggestions carefully, so as not to repeat a similar suggestion or the same suggestion.
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
nzw0301
left a comment
There was a problem hiding this comment.
Thanks. let me leave a few more comments.
|
Please kindly stop commenting just Best regards, |
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
Co-authored-by: Kento Nozawa <k_nzw@klis.tsukuba.ac.jp>
Description of the changes
Refs the following issue.
This PR adds an example showing the integration of transformers with optuna. We use the imdb dataset and
prajjwal1/bert-tinyas the model. The Best trial is:BestRun(run_id='1', objective=0.622, hyperparameters={'learning_rate': 3.341916221533824e-05, 'per_device_train_batch_size': 16}, run_summary=None)
Could you please review?
cc: @c-bata