search

mlcommons / algorithmic-efficiency

MLCommons Algorithmic Efficiency is a benchmark and competition measuring neural network training speedups due to algorithmic improvements in both training algorithms and models.
https://mlcommons.org/en/groups/research-algorithms/
Apache License 2.0
335 stars 69 forks source link

Feat default dropout in doc #806

Open init-22 opened 3 weeks ago

init-22 commented 3 weeks ago

Adding default dropout values for each workloads mentioned in this issue: #786

github-actions[bot] commented 3 weeks ago

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅

init-22 commented 3 weeks ago

recheck

priyakasimbeg commented 1 week ago

Hi Isaac, thanks for this PR. Looks great. Can we remove cifar10 and mnist from the table? Those two are implemented more for testing purposes and aren't part of the 8 AlgoPerf benchmarking workloads.

priyakasimbeg commented 1 week ago

Also, can you sign the CLA with the email address you use for your github account so that the CLA check passes. If it still does not pass, could you email support@mlcommons.org and ask for help with the CLA check in this PR.

morphine00 commented 1 week ago

Hi folks, the CLA check is passing. The code needs to be reviewed presumably by @priyakasimbeg (or anyone else in wg-algorithms). Note the message after the previous comment, " init-22 dismissed priyakasimbeg’s stale review via 76b084b 2 hours ago".

priyakasimbeg commented 1 week ago

Hi Isaac can you run yapf -i -r -vv -p scoring/score_submissions.py to fix the yapf test and push the changes. Typically, we want to make PRs to the dev branch but I think this score_submissions.py file is out of sync with def and has a formatting error that is blocking this PR

priyakasimbeg commented 3 days ago

This branch should also be synced with dev, which will fix the yapf tests.