"enc3": torch.ones(128, 128),
"enc4": torch.ones(8, 128),
"dec1": torch.ones(128, 8),
"dec2": torch.ones(128, 128),
"dec3": torch.ones(128, 128),
"dec4": torch.ones(128, 128),
"dout": torch.ones(640, 128),
}
# Setup cuda
use_cuda = False # torch.cuda.is_available() #Don't really need CUDA here at the moment, so just set false
device = torch.device("cuda:0" if use_cuda else "cpu")
torch.backends.cudnn.benchmark = True
# Load datasets
#yrdy_dataset = jet_dataset.ParticleJetDataset('train_data/train/', yamlConfig)
test_dataset = jet_dataset.ParticleJetDataset(options.test, yamlConfig)
test_labels = test_dataset.labels_list
#train_loader = torch.utils.data.DataLoader(full_dataset, batch_size=10000,
# shuffle=True, num_workers=0)
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=25000,
shuffle=False,
num_workers=0)
dir = "model_files/"
dir = options.model_files
#
# try:
# if options.batnorm:
# loadmodel = models.three_layer_model_batnorm_masked(prune_mask_set, bn_affine=options.bn_affine,
import hls4ml
## Load yaml config
def parse_config(config_file):
print("Loading configuration from", config_file)
config = open(config_file, 'r')
return yaml.load(config, Loader=yaml.FullLoader)
yamlConfig = parse_config("yamlConfig.yml")
# Setup test data set
test_dataset = jet_dataset.ParticleJetDataset("./train_data/test/", yamlConfig)
test_size = len(test_dataset)
print("test dataset size: " + str(len(test_dataset)))
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=test_size,
shuffle=False,
num_workers=10,
pin_memory=True)
X_test = None
y_test = None
for i, data in enumerate(test_loader, 0):
X_test, y_test = data[0].numpy(), data[1].numpy()
# Setup cuda
use_cuda = torch.cuda.is_available()
device = torch.device("cuda:0" if use_cuda else "cpu")
print("Using Device: {}".format(device))
torch.backends.cudnn.benchmark = True
torch.backends.cudnn.fastest = True
if use_cuda:
print("cuda:0 device type: {}".format(torch.cuda.get_device_name(0)))
# Set Batch size and split value
batch_size = 1024
train_split = 0.75
# Setup and split dataset
full_dataset = jet_dataset.ParticleJetDataset(options.inputFile,
yamlConfig)
test_dataset = jet_dataset.ParticleJetDataset(options.test, yamlConfig)
train_size = int(
train_split *
len(full_dataset)) # 25% for Validation set, 75% for train set
val_size = len(full_dataset) - train_size
test_size = len(test_dataset)
num_val_batches = math.ceil(val_size / batch_size)
num_train_batches = math.ceil(train_size / batch_size)
print("train_batches " + str(num_train_batches))
print("val_batches " + str(num_val_batches))
train_dataset, val_dataset = torch.utils.data.random_split(
full_dataset, [train_size, val_size])
torch.backends.cudnn.enabled = True
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
else:
torch.backends.cudnn.enabled = True
torch.backends.cudnn.benchmark = True
torch.backends.cudnn.fastest = True
# Set Batch size and split value
batch_size = 1024
if options.fold is None: #No fold passed, just load a whole folder and randomly split train/test
train_split = 0.75
# Setup and split dataset
full_dataset = jet_dataset.ParticleJetDataset(options.inputFile,yamlConfig)
train_size = int(train_split * len(full_dataset)) # 25% for Validation set, 75% for train set
val_size = len(full_dataset) - train_size
num_val_batches = math.ceil(val_size/batch_size)
num_train_batches = math.ceil(train_size/batch_size)
print("train_batches " + str(num_train_batches))
print("val_batches " + str(num_val_batches))
train_dataset, val_dataset = torch.utils.data.random_split(full_dataset,[train_size,val_size])
else:
train_filenames = []
val_filename = ""
