documentation: move num_chunks doc comment in ConvertApplyOpPattern #3770
Conversation
| tensor sizes it handles. Higher values may increase compute overhead but reduce size of | ||
| communication buffers when lowered. | ||
| If there are several candidate prefetch ops, the one with the largest result buffer | ||
| size is selected. |
There was a problem hiding this comment.
Do newlines get included in the docs or is it more like latex where single newlines are no-ops?
There was a problem hiding this comment.
The latter, same rule with double newline, you can also add them explicitly, I think it's basically markdown for the online doc rendering (and I think for VSCode hover)
| """ | ||
|
|
||
| num_chunks: int = 1 | ||
| """ | ||
| number of chunks into which communication and computation should be split. |
There was a problem hiding this comment.
Should we start doc comments with capital letters?
There was a problem hiding this comment.
Hmm, I would say yes, and should end in .
There was a problem hiding this comment.
I wanted to minimise the changes here but actually I'll make this one.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #3770 +/- ##
=======================================
Coverage 91.25% 91.25%
=======================================
Files 461 461
Lines 57478 57479 +1
Branches 5543 5543
=======================================
+ Hits 52454 52455 +1
Misses 3602 3602
Partials 1422 1422 ☔ View full report in Codecov by Sentry. |
…dslproject#3770) Makes the doc comment more localized and is rendered better in the online doc (upcoming).
Makes the doc comment more localized and is rendered better in the online doc (upcoming).