[Tuner] Allow passing extra TD specs during tuning #925
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Outdated suggestions cannot be applied.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.
Adds a new flag to libtuner
--extra-spec-file
, which allows specifying a path for an additional tuning spec to use during compilation. This allows the tuner to account for already tuned shapes when compiling models, which can have an affect on what the optimal candidate is.Another flag is added to the simple tuner (
--simple-best-spec-output-path
), which specifies a path to dump the final best spec after tuning. If stop-after is set to stop after dispatch benchmarking, then it will dump the best dispatch tuning spec, and otherwise the top model tuning spec will be saved.These 2 flags together now allows tuning loops to be set up where a full tuning spec can be generated automatically/directly from tuning several dispatches in a loop, by dumping the spec from the previous dispatch, and passing it to the tuning for the next dispatch. This has 2 benefits: