Added RffBenchmark #1855
Added RffBenchmark #1855
Conversation
Anipik
requested review from
justinormont,
danmoseley,
eerhardt,
tannergooding and
artidoro
|
@justinormont @Ivanidzo4ka can you review this one ? |
|
|
||
| var pipeline = mlContext.Transforms.Projection.CreateRandomFourierFeatures("Features", "FeaturesRFF") | ||
| .Append(mlContext.Transforms.Concatenate("Features", "FeaturesRFF")) | ||
| .AppendCacheCheckpoint(mlContext) |
There was a problem hiding this comment.
AppendCacheCheckpoint [](start = 13, length = 21)
Did you try to play with swapping lines 45 and 46? I believe this may be how our auto-caching used to work.
Also, this might expose the need to specify 'prefetchColumns' to avoid overly lazy caching. See the Cache method in DataOperations.cs line 35
There was a problem hiding this comment.
Did you try to play with swapping lines 45 and 46? I believe this may be how our auto-caching used to work.
I tried this but still getting the same result.
There was a problem hiding this comment.
Also, this might expose the need to specify 'prefetchColumns' to avoid overly lazy caching. See the Cache method in DataOperations.cs line 35
how can we use this in this benchmark ?
There was a problem hiding this comment.
@Zruty0: Do we have documentation on how/where/when to do caching?
ghost
locked as resolved and limited conversation to collaborators
Adding an end to end to Benchmark for rffTransform and cpumathutils functions like matmul and matmiltran