Fix MatrixFactorization trainer's warning #5071
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem to fix
MatrixFactorization's default options will throw a warning msg
Warning: insufficient blocks may slow down the trainingprocess (4*nr_threads^2+1 blocks is suggested)
when running on PC with more than 10 threads.Cause
The warning msg is from libmf, which will check nr_bin and nr_thread, if nr_bin <= 2* nr_thread, it will show that warning.
MatrixFactorization uses
max(20, 2*nr_thread)
to initialize nr_bin, when nr_thread < 10, it's fine because nr_bin will be 20, which is greater than 2*nr_thread, but if nr_thread is >= 10, nr_bin's value will be 2*nr_thread, which is less than 2*nr_thread + 1 and trigger warning.Fix
initialize nr_bin with 2 * nr_threads + 1.
Related issue
#5067