testID=None,
unique_fname=fname))
# All previous controllers control the agent during the epochs it goes through. However, we want to interleave a
# "validation epoch" between each training epoch ("one of two epochs", hence the periodicity=2). We do not want
# these validation epoch to interfere with the training of the agent, which is well established by the
# TrainerController, EpsilonController and alike. Therefore, we will disable these controllers for the whole
# duration of the validation epochs interleaved this way, using the controllersToDisable argument of the
# InterleavedTestEpochController. For each validation epoch, we want also to display the sum of all rewards
# obtained, hence the showScore=True. Finally, we want to call the summarizePerformance method of ALE_env every
# [parameters.period_btw_summary_perfs] *validation* epochs.
agent.attach(
bc.InterleavedTestEpochController(
id=ALE_env.VALIDATION_MODE,
epochLength=parameters.steps_per_test,
controllersToDisable=[0, 1, 2, 3, 4],
periodicity=2,
showScore=True,
summarizeEvery=1))
# --- Run the experiment ---
try:
os.mkdir("params")
except Exception:
pass
dump(vars(parameters), "params/" + fname + ".jldump")
agent.run(parameters.epochs, parameters.steps_per_epoch)
# --- Show results ---
basename = "scores/" + fname
scores = load(basename + "_scores.jldump")
initialLearningRate=parameters.learning_rate,
learningRateDecay=parameters.learning_rate_decay,
periodicity=1))
agent.attach(
bc.DiscountFactorController(
initialDiscountFactor=parameters.discount,
discountFactorGrowth=parameters.discount_inc,
discountFactorMax=parameters.discount_max,
periodicity=1))
agent.attach(
bc.EpsilonController(initialE=parameters.epsilon_start,
eDecays=parameters.epsilon_decay,
eMin=parameters.epsilon_min,
evaluateOn='action',
periodicity=1,
resetEvery='none'))
agent.attach(
bc.InterleavedTestEpochController(
id=0,
epochLength=parameters.steps_per_test,
controllersToDisable=[0, 1, 2, 3, 4],
periodicity=2,
showScore=True,
summarizeEvery=parameters.period_btw_summary_perfs))
# --- Run the experiment ---
agent.run(parameters.epochs, parameters.steps_per_epoch)
testID=MG_two_storages_env.TEST_MODE,
unique_fname=fname))
# All previous controllers control the agent during the epochs it goes through. However, we want to interleave a
# "validation epoch" between each training epoch ("one of two epochs", hence the periodicity=2). We do not want
# these validation epoch to interfere with the training of the agent, which is well established by the
# TrainerController, EpsilonController and alike, nor with its testing (see next controller). Therefore, we will
# disable these controllers for the whole duration of the validation epochs interleaved this way, using the
# controllersToDisable argument of the InterleavedTestEpochController. For each validation epoch, we want also to
# display the sum of all rewards obtained, hence the showScore=True. Finally, we never want this controller to call
# the summarizePerformance method of MG_two_storage_env.
agent.attach(
bc.InterleavedTestEpochController(
id=MG_two_storages_env.VALIDATION_MODE,
epochLength=parameters.steps_per_test,
controllersToDisable=[0, 1, 2, 3, 4, 7],
periodicity=2,
showScore=True,
summarizeEvery=-1))
# Besides inserting a validation epoch (required if one wants to find the best neural network over all training
# epochs), we also wish to interleave a "test epoch" between each training epoch ("one of two epochs", hence the
# periodicity=2). We do not want these test epoch to interfere with the training of the agent nor with its
# validation. Therefore, we will disable these controllers for the whole duration of the test epochs interleaved
# this way, using the controllersToDisable argument of the InterleavedTestEpochController. For each test epoch, we
# want also to display the sum of all rewards obtained, hence the showScore=True. Finally, we want to call the
# summarizePerformance method of MG_two_storage_env every [parameters.period_btw_summary_perfs] *test* epochs.
agent.attach(
bc.InterleavedTestEpochController(
id=MG_two_storages_env.TEST_MODE,
epochLength=parameters.steps_per_test,