def train_wrapper(config, ray_params):
train_ray(
path="/data/classification.parquet",
num_workers=4,
num_boost_rounds=100,
num_files=25,
regression=False,
use_gpu=False,
ray_params=ray_params,
xgboost_params=config,
)
def train_wrapper(config):
ray_params = RayParams(elastic_training=False,
max_actor_restarts=2,
num_actors=32,
cpus_per_actor=1,
gpus_per_actor=0)
train_ray(
path="/data/classification.parquet",
num_workers=32,
num_boost_rounds=100,
num_files=128,
regression=False,
use_gpu=False,
ray_params=ray_params,
xgboost_params=config,
)
Test owner: krfricke
Acceptance criteria: Should run through and report final results.
"""
import ray
from xgboost_ray import RayParams
from _train import train_ray
if __name__ == "__main__":
ray.init(address="auto")
ray_params = RayParams(elastic_training=False,
max_actor_restarts=2,
num_actors=32,
cpus_per_actor=4,
gpus_per_actor=0)
train_ray(
path="/data/classification.parquet",
num_workers=32,
num_boost_rounds=100,
num_files=128,
regression=False,
use_gpu=False,
ray_params=ray_params,
xgboost_params=None,
)
print("PASSED.")
ray_params = RayParams(elastic_training=False,
max_actor_restarts=2,
num_actors=4,
cpus_per_actor=4,
gpus_per_actor=0)
_, additional_results, _ = train_ray(
path="/data/classification.parquet",
num_workers=4,
num_boost_rounds=100,
num_files=200,
regression=False,
use_gpu=False,
ray_params=ray_params,
xgboost_params=None,
callbacks=[
TrackingCallback(),
FailureInjection(id="first_fail",
state=failure_state,
ranks=[2],
iteration=14),
FailureInjection(id="second_fail",
state=failure_state,
ranks=[0],
iteration=34)
])
actor_1_world_size = set(additional_results["callback_returns"][1])
assert len(actor_1_world_size) == 1 and 4 in actor_1_world_size, \
"Training with fewer than 4 actors observed, but this was " \
"non-elastic training. Please report to test owner."
