Source code for

import json
import os
import signal

from deephyper.core.logs.logging import JsonMessage as jm
from deephyper.evaluator.evaluate import Encoder
from import util
from import NeuralArchitectureSearch
from import Optimizer

dhlogger = util.conf_logger("")

SERVICE_PERIOD = 2  # Delay (seconds) between main loop iterations
CHECKPOINT_INTERVAL = 1  # How many jobs to complete between optimizer checkpoints

[docs]def on_exit(signum, stack): global EXIT_FLAG EXIT_FLAG = True
[docs]class AMBNeuralArchitectureSearch(NeuralArchitectureSearch): """Asynchronous Model-Based Search. Args: problem (str): python attribute import of the ``NaProblem`` instance (e.g. ``mypackage.mymodule.myproblem``). run (str): python attribute import of the run function (e.g. ``mypackage.mymodule.myrunfunction``). evaluator (str): the name of the evaluator to use. surrogate_model (str, optional): Choices are ["RF", "ET", "GBRT", "DUMMY", "GP"]. ``RF`` is Random Forest, ``ET`` is Extra Trees, ``GBRT`` is Gradient Boosting Regression Trees, ``DUMMY`` is random, ``GP`` is Gaussian process. Defaults to "RF". liar_strategy (str, optional): ["cl_max", "cl_min", "cl_mean"]. Defaults to "cl_max". acq_func (str, optional): Acquisition function, choices are ["gp_hedge", "LCB", "EI", "PI"]. Defaults to "gp_hedge". n_jobs (int, optional): Number of parallel jobs to distribute the surrogate model (learner). Defaults to -1, means as many as the number of logical cores. """ def __init__( self, problem, run, evaluator, surrogate_model="RF", liar_strategy="cl_max", acq_func="gp_hedge", n_jobs=-1, **kwargs, ): super().__init__(problem=problem, run=run, evaluator=evaluator, **kwargs) self.free_workers = self.evaluator.num_workers jm( type="start_infos", alg="ambs-nas", nworkers=self.free_workers, encoded_space=json.dumps(, cls=Encoder), ) )"Initializing AMBS") self.optimizer = Optimizer( self.problem, self.num_workers, surrogate_model=surrogate_model, liar_strategy=liar_strategy, acq_func=acq_func, n_jobs=n_jobs, **kwargs, ) @staticmethod def _extend_parser(parser): NeuralArchitectureSearch._extend_parser(parser) parser.add_argument( "--surrogate-model", default="RF", choices=["RF", "ET", "GBRT", "DUMMY", "GP"], help="type of surrogate model (learner)", ) parser.add_argument( "--liar-strategy", default="cl_max", choices=["cl_min", "cl_mean", "cl_max"], help="Constant liar strategy", ) parser.add_argument( "--acq-func", default="gp_hedge", choices=["LCB", "EI", "PI", "gp_hedge"], help="Acquisition function type", ) parser.add_argument( "--acq-kappa", type=float, default=1.96, help='Controls how much of the variance in the predicted values should be taken into account. If set to be very high, then we are favouring exploration over exploitation and vice versa. Used when the acquisition is "LCB".', ) parser.add_argument( "--n-jobs", default=-1, type=int, help="Number of processes to use for surrogate model (learner).", ) return parser
[docs] def main(self): timer = util.DelayTimer(max_minutes=None, period=SERVICE_PERIOD) chkpoint_counter = 0 num_evals = 0"Generating {self.num_workers} initial points...") XX = self.optimizer.ask_initial(n_points=self.num_workers) self.evaluator.add_eval_batch(XX) # MAIN LOOP for elapsed_str in timer:"Elapsed time: {elapsed_str}") results = list(self.evaluator.get_finished_evals()) num_evals += len(results) chkpoint_counter += len(results) if EXIT_FLAG or num_evals >= self.max_evals: break if results:"Refitting model with batch of {len(results)} evals") self.optimizer.tell(results) f"Drawing {len(results)} points with strategy {self.optimizer.strategy}" ) # ! 'ask' is written as a generator because asking for a large batch is # ! slow. We get better performance when ask is batched. The RF is # ! constantly re-fitting during the call to ask. So it becomes slow # ! when there are a large number of workers. for batch in self.optimizer.ask(n_points=len(results)): self.evaluator.add_eval_batch(batch) if chkpoint_counter >= CHECKPOINT_INTERVAL: self.evaluator.dump_evals(saved_key="arch_seq") chkpoint_counter = 0"Hyperopt driver finishing") self.evaluator.dump_evals(saved_key="arch_seq")
if __name__ == "__main__": args = AMBNeuralArchitectureSearch.parse_args() search = AMBNeuralArchitectureSearch(**vars(args)) signal.signal(signal.SIGINT, on_exit) signal.signal(signal.SIGTERM, on_exit) search.main()