-
Notifications
You must be signed in to change notification settings - Fork 40
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Results on different slices #6
Comments
Hello, I am a newbie who has just entered this field. I tried to directly clone the author's project for experiments, but encountered many problems. Can you share your experimental project with me? Thank you very much if you can. |
do u got the experimental results?Can you share with me please |
no, I didn't even get through the code,Can you give me some advice or help? pls,I will also share my progress with you later |
I run the purchase data and set shard=1, slice=1 and epoch=1, the output model is trained on 0.25 million data for 1 epoch and the accuracy is around 0.955.
Then I set shard=1, slice=16, and epoch=1. The train always take places on the 8th slice and epoch for it is 1 while for other slices is 0, according to the 'avg_epochs_per_slice' and 'slice_epochs' calculated. But the output model still uses 0.14 million data and the accuracy is around 0.945.
I didn't see significant difference between them as Fig5 in paper shows.
Is there any mistake using the above settings?
The text was updated successfully, but these errors were encountered: