[Retiarii] Grid search, random and evolution strategy #3377
Conversation
|
|
||
|
|
||
| class Random(BaseStrategy): | ||
| def __init__(self, variational=False, dedup=True): |
There was a problem hiding this comment.
what is the meaning of variational?
| else: | ||
| _logger.info('Random search running in fixed size mode. Dedup: %s.', 'on' if self.dedup else 'off') | ||
| search_space = dry_run_for_search_space(base_model, applied_mutators) | ||
| for sample in random_generator(search_space, dedup=self.dedup): |
There was a problem hiding this comment.
what will webui show if strategy exits but maxtrialnum and maxduration are not reached?
There was a problem hiding this comment.
I tested. The whole experiment directly exits and dispatcher (strategy) is terminated.
There was a problem hiding this comment.
dispatcher is not strategy, why dispatcher exits?...
There was a problem hiding this comment.
"Dispatcher terminated" is printed on the console. Afterwards, no more trials appear on the WebUI. I'm not sure about the details.
There was a problem hiding this comment.
got it, this is a normal behavior
nni/retiarii/strategy/evolution.py
Outdated
| population_size : int | ||
| The number of individuals to keep in the population. | ||
| cycles : int | ||
| The number of cycles (trials) the algorithm should run for. |
There was a problem hiding this comment.
cycle is a little misleading, cycle means trial in the paper?
There was a problem hiding this comment.
The terminology is used in paper and their open-source implementation.
This PR refines and adds new strategies to Retiarii.