args.max_examples_per_epoch = args.num_training
print("Executing " + args.checkpoint_key)
with open("args-{}".format(args.checkpoint_key), "w") as args_file:
args_file.write(" ".join(sys.argv + ["--seed ", str(args.seed)]))
use_cuda = torch.cuda.is_available() and not args.no_cuda
device = torch.device("cuda" if use_cuda else "cpu")
is_parallel = False
best_acc = 0 # best test accuracy
start_epoch = 0 # start from epoch 0 or last checkpoint epoch
problem = None
if args.problem.startswith("genotyping:"):
problem = SbiGenotypingProblem(args.mini_batch_size,
code=args.problem,
num_workers=args.num_workers)
elif args.problem.startswith("struct_genotyping:"):
# struct_genotyping does not support multiprocessing data loading:
problem = StructuredSbiGenotypingProblem(args.mini_batch_size,
code=args.problem,
num_workers=1)
elif args.problem.startswith("somatic:"):
problem = SbiSomaticProblem(args.mini_batch_size,
code=args.problem,
num_workers=args.num_workers)
else:
print("Unsupported problem: " + args.problem)
exit(1)
def get_metric_value(all_perfs, query_metric_name):
default=512)
parser.add_argument("--num-workers",
type=int,
default=0,
help='Number of workers to feed data to the GPUs.')
parser.add_argument("-n",
type=int,
default=sys.maxsize,
help='Maximum number of examples to sample from each '
'dataset.')
args = parser.parse_args()
problem = None
if args.problem.startswith("genotyping:"):
problem = SbiGenotypingProblem(args.mini_batch_size,
code=args.problem,
num_workers=args.num_workers)
elif args.problem.startswith("somatic:"):
problem = SbiSomaticProblem(args.mini_batch_size,
code=args.problem,
num_workers=args.num_workers)
else:
print("Unsupported problem: " + args.problem)
exit(1)
sum_n = None
n = 0
mean = None
sum_sdm_n = None
std = None
try:
checkpoint_filename = '{}/models/pytorch_{}_{}.t7'.format(
args.model_path, args.checkpoint_key, args.model_label)
checkpoint = torch.load(checkpoint_filename)
except FileNotFoundError:
print("Unable to load model {} from checkpoint".format(
args.checkpoint_key))
exit(1)
if checkpoint is not None:
model = checkpoint['model']
problem = None
if args.problem.startswith("genotyping:"):
problem = SbiGenotypingProblem(args.mini_batch_size,
code=args.problem,
drop_last_batch=False,
num_workers=args.num_workers)
elif args.problem.startswith("somatic:"):
problem = SbiSomaticProblem(args.mini_batch_size,
code=args.problem,
drop_last_batch=False,
num_workers=args.num_workers)
elif args.problem.startswith("struct_genotyping:"):
problem = StructuredSbiGenotypingProblem(args.mini_batch_size,
code=args.problem,
drop_last_batch=False,
num_workers=args.num_workers)
else:
print("Unsupported problem: " + args.problem)
exit(1)
default=0)
parser.add_argument('--num-gpus', type=int, help='Number of GPUs to use for search.',
default=1)
parser.add_argument('--num-estimate-class-frequencies', type=int, help='Number of examples to look at to estimate '
'class frequencies.',
default=sys.maxsize)
parser.add_argument('--debug', action='store_true', help='Used to debug some code.')
parser.add_argument('--num-debug', type=int,
help='Maximum number of debug iterations.',
default=sys.maxsize)
args = parser.parse_args()
problem = None
if args.problem.startswith("genotyping:"):
problem = SbiGenotypingProblem(args.mini_batch_size, code=args.problem, num_workers=args.num_workers)
elif args.problem.startswith("somatic:"):
problem = SbiSomaticProblem(args.mini_batch_size, code=args.problem, num_workers=args.num_workers)
else:
print("Unsupported problem: " + args.problem)
exit(1)
args.num_training = min(args.num_training, len(problem.train_set()))
args.num_estimate_class_frequencies = min(args.num_estimate_class_frequencies, len(problem.train_set()))
args.num_validation = min(args.num_validation, len(problem.validation_set()))
trainers = []
count = 0
with open(args.commands, "r") as command_file:
trainer_arguments = command_file.readlines()
count = len(trainer_arguments)
i = 0