Fix ASHA termination condition and branching with fidelity() #274
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
bouthilx
added
bug
bouthilx
force-pushed
the
hotfix/fidelity_branching
branch
from
fd4f94a to
41a3ebf
Compare
Codecov Report
@@ Coverage Diff @@
## develop #274 +/- ##
===========================================
+ Coverage 93.19% 93.26% +0.07%
===========================================
Files 119 120 +1
Lines 14392 14512 +120
Branches 999 1003 +4
===========================================
+ Hits 13412 13534 +122
+ Misses 879 877 -2
Partials 101 101
Continue to review full report at Codecov.
|
Why: There was no documentation on the fact that ASHA requires the SubParallelStrategy. There was also no documentation at all about the parallel strategies...
Why: If an algo is done, but only *realizes* during an update in `produce`, then the algo will opt out indefinitely and the process will crash. This should be handled and leave gracefully. How: Check for algo.is_done at each loop iteration in `produce` and stop if True.
Why: The EVC is using the default_value to build adapters. Also this default value can be used as a fall-back if the algorithms does not support multi-fidelity optimization. This is handy is we want to try different algorithms with the same experiment configuration.
Why: The termination condition was wrong and would stop when the penultimate rung was filled, making it impossible to ever run the final trial with max resources.
bouthilx
force-pushed
the
hotfix/fidelity_branching
branch
from
41a3ebf to
b850d93
Compare
Delaunay
approved these changes
Aug 29, 2019
Merged
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Improve documentation of ASHA
Why:
There was no documentation on the fact that ASHA requires the
SubParallelStrategy. There was also no documentation at all about the
parallel strategies...
Stop producer.produce when algo is done
Why:
If an algo is done, but only realizes during an update in
produce,then the algo will opt out indefinitely and the process will crash. This
should be handled and leave gracefully.
How:
Check for algo.is_done at each loop iteration in
produceand stop ifTrue.
Add default_value to fidelity
Why:
The EVC is using the default_value to build adapters. Also this default
value can be used as a fall-back if the algorithms does not support
multi-fidelity optimization. This is handy is we want to try different
algorithms with the same experiment configuration.
Fix ASHA termination condition
Why:
The termination condition was wrong and would stop when the penultimate
rung was filled, making it impossible to ever run the final trial with
max resources.