19.5. Asynchronous Successive Halving¶ Open the notebook in SageMaker Studio Lab
As we have seen in Section 19.3, we can accelerate HPO by distributing the evaluation of hyperparameter configurations across either multiple instances or multiples CPUs / GPUs on a single instance. However, compared to random search, it is not straightforward to run successive halving (SH) asynchronously in a distributed setting. Before we can decide which configuration to run next, we first have to collect all observations at the current rung level. This requires to synchronize workers at each rung level. For example, for the lowest rung level \(r_{\text{min}}\), we first have to evaluate all \(N = \eta^K\) configurations, before we can promote the \(\frac{1}{\eta}\) of them to the next rung level.
In any distributed system, synchronization typically implies idle time for workers. First, we often observe high variations in training time across hyperparameter configurations. For example, assuming the number of filters per layer is a hyperparameter, then networks with less filters finish training faster than networks with more filters, which implies idle worker time due to stragglers. Moreover, the number of slots in a rung level is not always a multiple of the number of workers, in which case some workers may even sit idle for a full batch.
Figure Fig. 19.5.1 shows the scheduling of synchronous SH with \(\eta=2\) for four different trials with two workers. We start with evaluating Trial-0 and Trial-1 for one epoch and immediately continue with the next two trials once they are finished. We first have to wait until Trial-2 finishes, which takes substantially more time than the other trials, before we can promote the best two trials, i.e., Trial-0 and Trial-3 to the next rung level. This causes idle time for Worker-1. Then, we continue with Rung 1. Also, here Trial-3 takes longer than Trial-0, which leads to an additional ideling time of Worker-0. Once, we reach Rung-2, only the best trial, Trial-0, remains which occupies only one worker. To avoid that Worker-1 idles during that time, most implementaitons of SH continue already with the next round, and start evaluating new trials (e.g Trial-4) on the first rung.
Fig. 19.5.1 Synchronous successive halving with two workers.¶
Asynchronous successive halving (ASHA) (Li et al., 2018) adapts SH to the asynchronous parallel scenario. The main idea of ASHA is to promote configurations to the next rung level as soon as we collected at least \(\eta\) observations on the current rung level. This decision rule may lead to suboptimal promotions: configurations can be promoted to the next rung level, which in hindsight do not compare favourably against most others at the same rung level. On the other hand, we get rid of all synchronization points this way. In practice, such suboptimal initial promotions have only a modest impact on performance, not only because the ranking of hyperparameter configurations is often fairly consistent across rung levels, but also because rungs grow over time and reflect the distribution of metric values at this level better and better. If a worker is free, but no configuration can be promoted, we start a new configuration with \(r = r_{\text{min}}\), i.e the first rung level.
Fig. 19.5.2 shows the scheduling of the same configurations for ASHA. Once Trial-1 finishes, we collect the results of two trials (i.e Trial-0 and Trial-1) and immediately promote the better of them (Trial-0) to the next rung level. After Trial-0 finishes on rung 1, there are too few trials there in order to support a further promotion. Hence, we continue with rung 0 and evaluate Trial-3. Once Trial-3 finishes, Trial-2 is still pending. At this point we have 3 trials evaluated on rung 0 and one trial evaluated already on rung 1. Since Trial-3 performs worse than Trial-0 at rung 0, and \(\eta=2\), we cannot promote any new trial yet, and Worker-1 starts Trial-4 from scratch instead. However, once Trial-2 finishes and scores worse than Trial-3, the latter is promoted towards rung 1. Afterwards, we collected 2 evaluations on rung 1, which means we can now promote Trial-0 towards rung 2. At the same time, Worker-1 continues with evaluating new trials (i.e., Trial-5) on rung 0.
Fig. 19.5.2 Asynchronous successive halving (ASHA) with two workers.¶
import logging
from d2l import torch as d2l
logging.basicConfig(level=logging.INFO)
import matplotlib.pyplot as plt
from syne_tune import StoppingCriterion, Tuner
from syne_tune.backend.python_backend import PythonBackend
from syne_tune.config_space import loguniform, randint
from syne_tune.experiments import load_experiment
from syne_tune.optimizer.baselines import ASHA
INFO:root:SageMakerBackend is not imported since dependencies are missing. You can install them with
pip install 'syne-tune[extra]'
AWS dependencies are not imported since dependencies are missing. You can install them with
pip install 'syne-tune[aws]'
or (for everything)
pip install 'syne-tune[extra]'
AWS dependencies are not imported since dependencies are missing. You can install them with
pip install 'syne-tune[aws]'
or (for everything)
pip install 'syne-tune[extra]'
INFO:root:Ray Tune schedulers and searchers are not imported since dependencies are missing. You can install them with
pip install 'syne-tune[raytune]'
or (for everything)
pip install 'syne-tune[extra]'
19.5.1. Objective Function¶
We will use Syne Tune with the same objective function as in Section 19.3.
def hpo_objective_lenet_synetune(learning_rate, batch_size, max_epochs):
from syne_tune import Reporter
from d2l import torch as d2l
model = d2l.LeNet(lr=learning_rate, num_classes=10)
trainer = d2l.HPOTrainer(max_epochs=1, num_gpus=1)
data = d2l.FashionMNIST(batch_size=batch_size)
model.apply_init([next(iter(data.get_dataloader(True)))[0]], d2l.init_cnn)
report = Reporter()
for epoch in range(1, max_epochs + 1):
if epoch == 1:
# Initialize the state of Trainer
trainer.fit(model=model, data=data)
else:
trainer.fit_epoch()
validation_error = trainer.validation_error().cpu().detach().numpy()
report(epoch=epoch, validation_error=float(validation_error))
We will also use the same configuration space as before:
min_number_of_epochs = 2
max_number_of_epochs = 10
eta = 2
config_space = {
"learning_rate": loguniform(1e-2, 1),
"batch_size": randint(32, 256),
"max_epochs": max_number_of_epochs,
}
initial_config = {
"learning_rate": 0.1,
"batch_size": 128,
}
19.5.2. Asynchronous Scheduler¶
First, we define the number of workers that evaluate trials concurrently. We also need to specify how long we want to run random search, by defining an upper limit on the total wall-clock time.
n_workers = 2 # Needs to be <= the number of available GPUs
max_wallclock_time = 12 * 60 # 12 minutes
The code for running ASHA is a simple variation of what we did for asynchronous random search.
mode = "min"
metric = "validation_error"
resource_attr = "epoch"
scheduler = ASHA(
config_space,
metric=metric,
mode=mode,
points_to_evaluate=[initial_config],
max_resource_attr="max_epochs",
resource_attr=resource_attr,
grace_period=min_number_of_epochs,
reduction_factor=eta,
)
INFO:syne_tune.optimizer.schedulers.fifo:max_resource_level = 10, as inferred from config_space
INFO:syne_tune.optimizer.schedulers.fifo:Master random_seed = 1420030479
Here, metric and resource_attr specify the key names used with
the report callback, and max_resource_attr denotes which input
to the objective function corresponds to \(r_{\text{max}}\).
Moreover, grace_period provides \(r_{\text{min}}\), and
reduction_factor is \(\eta\). We can run Syne Tune as before
(this will take about 12 minutes):
trial_backend = PythonBackend(
tune_function=hpo_objective_lenet_synetune,
config_space=config_space,
)
stop_criterion = StoppingCriterion(max_wallclock_time=max_wallclock_time)
tuner = Tuner(
trial_backend=trial_backend,
scheduler=scheduler,
stop_criterion=stop_criterion,
n_workers=n_workers,
print_update_interval=int(max_wallclock_time * 0.6),
)
tuner.run()
INFO:syne_tune.tuner:results of trials will be saved on /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470
INFO:root:Detected 3 GPUs
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.1 --batch_size 128 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/0/checkpoints
INFO:syne_tune.tuner:(trial 0) - scheduled config {'learning_rate': 0.1, 'batch_size': 128, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.5985177262140595 --batch_size 249 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/1/checkpoints
INFO:syne_tune.tuner:(trial 1) - scheduled config {'learning_rate': 0.5985177262140595, 'batch_size': 249, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.7930685343937359 --batch_size 167 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/2/checkpoints
INFO:syne_tune.tuner:(trial 2) - scheduled config {'learning_rate': 0.7930685343937359, 'batch_size': 167, 'max_epochs': 10}
INFO:syne_tune.tuner:Trial trial_id 1 completed.
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.01332236841633562 --batch_size 83 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/3/checkpoints
INFO:syne_tune.tuner:(trial 3) - scheduled config {'learning_rate': 0.01332236841633562, 'batch_size': 83, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.06395616061075522 --batch_size 157 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/4/checkpoints
INFO:syne_tune.tuner:(trial 4) - scheduled config {'learning_rate': 0.06395616061075522, 'batch_size': 157, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.03894552388726989 --batch_size 246 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/5/checkpoints
INFO:syne_tune.tuner:(trial 5) - scheduled config {'learning_rate': 0.03894552388726989, 'batch_size': 246, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.21589884448181618 --batch_size 171 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/6/checkpoints
INFO:syne_tune.tuner:(trial 6) - scheduled config {'learning_rate': 0.21589884448181618, 'batch_size': 171, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.02134388373929251 --batch_size 112 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/7/checkpoints
INFO:syne_tune.tuner:(trial 7) - scheduled config {'learning_rate': 0.02134388373929251, 'batch_size': 112, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.4281530822285822 --batch_size 236 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/8/checkpoints
INFO:syne_tune.tuner:(trial 8) - scheduled config {'learning_rate': 0.4281530822285822, 'batch_size': 236, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.14597685354921533 --batch_size 46 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/9/checkpoints
INFO:syne_tune.tuner:(trial 9) - scheduled config {'learning_rate': 0.14597685354921533, 'batch_size': 46, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.023606359106159654 --batch_size 141 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/10/checkpoints
INFO:syne_tune.tuner:(trial 10) - scheduled config {'learning_rate': 0.023606359106159654, 'batch_size': 141, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.020475406507570872 --batch_size 137 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/11/checkpoints
INFO:syne_tune.tuner:(trial 11) - scheduled config {'learning_rate': 0.020475406507570872, 'batch_size': 137, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.03072386439846424 --batch_size 52 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/12/checkpoints
INFO:syne_tune.tuner:(trial 12) - scheduled config {'learning_rate': 0.03072386439846424, 'batch_size': 52, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.2884807064939987 --batch_size 52 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/13/checkpoints
INFO:syne_tune.tuner:(trial 13) - scheduled config {'learning_rate': 0.2884807064939987, 'batch_size': 52, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.49517571939878474 --batch_size 47 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/14/checkpoints
INFO:syne_tune.tuner:(trial 14) - scheduled config {'learning_rate': 0.49517571939878474, 'batch_size': 47, 'max_epochs': 10}
INFO:syne_tune.tuner:Trial trial_id 13 completed.
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.07069910046322526 --batch_size 79 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/15/checkpoints
INFO:syne_tune.tuner:(trial 15) - scheduled config {'learning_rate': 0.07069910046322526, 'batch_size': 79, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.1393298222152186 --batch_size 92 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/16/checkpoints
INFO:syne_tune.tuner:(trial 16) - scheduled config {'learning_rate': 0.1393298222152186, 'batch_size': 92, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.01102850321799237 --batch_size 160 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/17/checkpoints
INFO:syne_tune.tuner:(trial 17) - scheduled config {'learning_rate': 0.01102850321799237, 'batch_size': 160, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.045553351306542623 --batch_size 163 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/18/checkpoints
INFO:syne_tune.tuner:(trial 18) - scheduled config {'learning_rate': 0.045553351306542623, 'batch_size': 163, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.18738829796624235 --batch_size 36 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/19/checkpoints
INFO:syne_tune.tuner:(trial 19) - scheduled config {'learning_rate': 0.18738829796624235, 'batch_size': 36, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.05077467757237889 --batch_size 228 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/20/checkpoints
INFO:syne_tune.tuner:(trial 20) - scheduled config {'learning_rate': 0.05077467757237889, 'batch_size': 228, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.2818254023025138 --batch_size 66 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/21/checkpoints
INFO:syne_tune.tuner:(trial 21) - scheduled config {'learning_rate': 0.2818254023025138, 'batch_size': 66, 'max_epochs': 10}
INFO:syne_tune.tuner:tuning status (last metric is reported)
trial_id status iter learning_rate batch_size max_epochs epoch validation_error worker-time
0 Stopped 2 0.100000 128 10 2.0 0.899031 16.655625
1 Completed 10 0.598518 249 10 10.0 0.201665 63.855728
2 Stopped 10 0.793069 167 10 10.0 0.178710 66.577238
3 Stopped 2 0.013322 83 10 2.0 0.900000 17.155993
4 Stopped 2 0.063956 157 10 2.0 0.900171 15.577330
5 Stopped 2 0.038946 246 10 2.0 0.900107 14.921423
6 Stopped 4 0.215899 171 10 4.0 0.332322 27.402298
7 Stopped 2 0.021344 112 10 2.0 0.900546 16.316442
8 Stopped 4 0.428153 236 10 4.0 0.323575 26.391071
9 Stopped 8 0.145977 46 10 8.0 0.227629 79.750229
10 Stopped 2 0.023606 141 10 2.0 0.900008 15.916068
11 Stopped 2 0.020475 137 10 2.0 0.899999 16.027407
12 Stopped 4 0.030724 52 10 4.0 0.899686 37.019265
13 Completed 10 0.288481 52 10 10.0 0.149586 92.460085
14 Stopped 10 0.495176 47 10 10.0 0.141497 97.445830
15 Stopped 2 0.070699 79 10 2.0 0.899757 16.952509
16 Stopped 4 0.139330 92 10 4.0 0.341133 30.270227
17 Stopped 4 0.011029 160 10 4.0 0.899802 28.178677
18 Stopped 2 0.045553 163 10 2.0 0.899761 14.328157
19 InProgress 3 0.187388 36 10 3.0 0.229159 35.283644
20 Stopped 2 0.050775 228 10 2.0 0.900010 14.251756
21 InProgress 0 0.281825 66 10 - - -
2 trials running, 20 finished (2 until the end), 437.04s wallclock-time
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.13450694334348853 --batch_size 226 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/22/checkpoints
INFO:syne_tune.tuner:(trial 22) - scheduled config {'learning_rate': 0.13450694334348853, 'batch_size': 226, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.1383206315022528 --batch_size 76 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/23/checkpoints
INFO:syne_tune.tuner:(trial 23) - scheduled config {'learning_rate': 0.1383206315022528, 'batch_size': 76, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.13182291830737192 --batch_size 137 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/24/checkpoints
INFO:syne_tune.tuner:(trial 24) - scheduled config {'learning_rate': 0.13182291830737192, 'batch_size': 137, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.20451149999816387 --batch_size 178 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/25/checkpoints
INFO:syne_tune.tuner:(trial 25) - scheduled config {'learning_rate': 0.20451149999816387, 'batch_size': 178, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.044544495073044366 --batch_size 214 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/26/checkpoints
INFO:syne_tune.tuner:(trial 26) - scheduled config {'learning_rate': 0.044544495073044366, 'batch_size': 214, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.026195132773807073 --batch_size 106 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/27/checkpoints
INFO:syne_tune.tuner:(trial 27) - scheduled config {'learning_rate': 0.026195132773807073, 'batch_size': 106, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.15508038457501633 --batch_size 120 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/28/checkpoints
INFO:syne_tune.tuner:(trial 28) - scheduled config {'learning_rate': 0.15508038457501633, 'batch_size': 120, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.7306582456725493 --batch_size 79 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/29/checkpoints
INFO:syne_tune.tuner:(trial 29) - scheduled config {'learning_rate': 0.7306582456725493, 'batch_size': 79, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.08473828437431256 --batch_size 130 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/30/checkpoints
INFO:syne_tune.tuner:(trial 30) - scheduled config {'learning_rate': 0.08473828437431256, 'batch_size': 130, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.10784996779411136 --batch_size 40 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/31/checkpoints
INFO:syne_tune.tuner:(trial 31) - scheduled config {'learning_rate': 0.10784996779411136, 'batch_size': 40, 'max_epochs': 10}
INFO:syne_tune.tuner:Trial trial_id 29 completed.
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.37677242671098565 --batch_size 205 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/32/checkpoints
INFO:syne_tune.tuner:(trial 32) - scheduled config {'learning_rate': 0.37677242671098565, 'batch_size': 205, 'max_epochs': 10}
INFO:root:running subprocess with command: /home/d2l-worker/miniconda3/envs/d2l-en-release-1/bin/python /home/d2l-worker/miniconda3/envs/d2l-en-release-1/lib/python3.9/site-packages/syne_tune/backend/python_backend/python_entrypoint.py --learning_rate 0.12320387549715911 --batch_size 41 --max_epochs 10 --tune_function_root /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/tune_function --tune_function_hash 40412c9643f2f0fda4865de73fad34e4 --st_checkpoint_dir /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470/33/checkpoints
INFO:syne_tune.tuner:(trial 33) - scheduled config {'learning_rate': 0.12320387549715911, 'batch_size': 41, 'max_epochs': 10}
INFO:syne_tune.stopping_criterion:reaching max wallclock time (720), stopping there.
INFO:syne_tune.tuner:Stopping trials that may still be running.
INFO:syne_tune.tuner:Tuning finished, results of trials can be found on /home/d2l-worker/syne-tune/python-entrypoint-2022-12-14-06-32-15-470
--------------------
Resource summary (last result is reported):
trial_id status iter learning_rate batch_size max_epochs epoch validation_error worker-time
0 Stopped 2 0.100000 128 10 2.0 0.899031 16.655625
1 Completed 10 0.598518 249 10 10.0 0.201665 63.855728
2 Stopped 10 0.793069 167 10 10.0 0.178710 66.577238
3 Stopped 2 0.013322 83 10 2.0 0.900000 17.155993
4 Stopped 2 0.063956 157 10 2.0 0.900171 15.577330
5 Stopped 2 0.038946 246 10 2.0 0.900107 14.921423
6 Stopped 4 0.215899 171 10 4.0 0.332322 27.402298
7 Stopped 2 0.021344 112 10 2.0 0.900546 16.316442
8 Stopped 4 0.428153 236 10 4.0 0.323575 26.391071
9 Stopped 8 0.145977 46 10 8.0 0.227629 79.750229
10 Stopped 2 0.023606 141 10 2.0 0.900008 15.916068
11 Stopped 2 0.020475 137 10 2.0 0.899999 16.027407
12 Stopped 4 0.030724 52 10 4.0 0.899686 37.019265
13 Completed 10 0.288481 52 10 10.0 0.149586 92.460085
14 Stopped 10 0.495176 47 10 10.0 0.141497 97.445830
15 Stopped 2 0.070699 79 10 2.0 0.899757 16.952509
16 Stopped 4 0.139330 92 10 4.0 0.341133 30.270227
17 Stopped 4 0.011029 160 10 4.0 0.899802 28.178677
18 Stopped 2 0.045553 163 10 2.0 0.899761 14.328157
19 Stopped 10 0.187388 36 10 10.0 0.170378 112.051995
20 Stopped 2 0.050775 228 10 2.0 0.900010 14.251756
21 Stopped 4 0.281825 66 10 4.0 0.416098 43.777161
22 Stopped 2 0.134507 226 10 2.0 0.900179 14.330927
23 Stopped 8 0.138321 76 10 8.0 0.226629 82.387080
24 Stopped 2 0.131823 137 10 2.0 0.900001 15.938905
25 Stopped 2 0.204511 178 10 2.0 0.899640 15.654077
26 Stopped 2 0.044544 214 10 2.0 0.899911 15.411022
27 Stopped 2 0.026195 106 10 2.0 0.899923 16.008510
28 Stopped 4 0.155080 120 10 4.0 0.339484 30.281955
29 Completed 10 0.730658 79 10 10.0 0.147388 75.855369
30 Stopped 2 0.084738 130 10 2.0 0.899983 16.240244
31 InProgress 7 0.107850 40 10 7.0 0.203700 74.223180
32 Stopped 8 0.376772 205 10 8.0 0.227063 50.852603
33 InProgress 0 0.123204 41 10 - - -
2 trials running, 32 finished (3 until the end), 723.68s wallclock-time
validation_error: best 0.1414971947669983 for trial-id 14
--------------------
Note that we are running a variant of ASHA where underperforming trials
are stopped early. This is different to our implementation in
Section 19.4.1, where each training job is started with a
fixed max_epochs. In the latter case, a well-performing trial which
reaches the full 10 epochs, first needs to train 1, then 2, then 4, then
8 epochs, each time starting from scratch. This type of pause-and-resume
scheduling can be implemented efficiently by checkpointing the training
state after each epoch, but we avoid this extra complexity here. After
the experiment has finished, we can retrieve and plot results.
d2l.set_figsize()
e = load_experiment(tuner.name)
e.plot()
WARNING:matplotlib.legend:No artists with labels found to put in legend. Note that artists whose label start with an underscore are ignored when legend() is called with no argument.
19.5.3. Visualize the Optimization Process¶
Once more, we visualize the learning curves of every trial (each color in the plot represents a trial). Compare this to asynchronous random search in Section 19.3. As we have seen for successive halving in Section 19.4, most of the trials are stopped at 1 or 2 epochs (\(r_{\text{min}}\) or \(\eta * r_{\text{min}}\)). However, trials do not stop at the same point, because they require different amount of time per epoch. If we ran standard successive halving instead of ASHA, we would need to synchronize our workers, before we can promote configurations to the next rung level.
d2l.set_figsize([6, 2.5])
results = e.results
for trial_id in results.trial_id.unique():
df = results[results["trial_id"] == trial_id]
d2l.plt.plot(
df["st_tuner_time"],
df["validation_error"],
marker="o"
)
d2l.plt.xlabel("wall-clock time")
d2l.plt.ylabel("objective function")
Text(0, 0.5, 'objective function')
19.5.4. Summary¶
Compared to random search, successive halving is not quite as trivial to run in an asynchronous distributed setting. To avoid synchronisation points, we promote configurations as quickly as possible to the next rung level, even if this means promoting some wrong ones. In practice, this usually does not hurt much, and the gains of asynchronous versus synchronous scheduling are usually much higher than the loss of the suboptimal decision making.