def __init__(self,
config_file_path,
debug_mode=False,
import_dir='',
export_dir=''):
#
# parse config file and save off the information we need
#
config_dict = parse_config(config_file_path)
self.server_url = config_dict.get('server_url', 'https://127.0.0.1')
self.api_token = config_dict.get('api_token', '')
self.sites = config_dict.get('sites', [])
self.debug = config_dict.get('debug', False)
self.export_dir = export_dir
self.import_dir = import_dir
self.http_proxy_url = config_dict.get('http_proxy_url', None)
self.https_proxy_url = config_dict.get('https_proxy_url', None)
if self.export_dir and not os.path.exists(self.export_dir):
os.mkdir(self.export_dir)
#
# Test Cb Response connectivity
#
try:
self.cb = CbApi(server=self.server_url,
token=self.api_token,
ssl_verify=False)
self.cb.feed_enum()
except:
logger.error(traceback.format_exc())
sys.exit(-1)
def __init__(self, config_file_path, debug_mode=False, import_dir='', export_dir=''):
#
# parse config file and save off the information we need
#
config_dict = parse_config(config_file_path)
self.server_url = config_dict.get('server_url', 'https://127.0.0.1')
self.api_token = config_dict.get('api_token', '')
self.sites = config_dict.get('sites', [])
self.debug = config_dict.get('debug', False)
self.export_dir = export_dir
self.import_dir = import_dir
if self.export_dir and not os.path.exists(self.export_dir):
os.mkdir(self.export_dir)
#
# Test Cb Response connectivity
#
try:
self.cb = CbApi(server=self.server_url, token=self.api_token, ssl_verify=False)
self.cb.feed_enum()
except:
logger.error(traceback.format_exc())
sys.exit(-1)
pred = pred.view(batch_size, group_size, -1)
pred = nn.functional.softmax(pred, dim=-1).mean(1)
pred_classes = pred.argmax(dim=1)
num_each_class[0] += pred_classes.eq(0).sum().item()
num_each_class[1] += pred_classes.eq(1).sum().item()
total = num_each_class.sum()
print_and_log('Prediction set classification results:')
print_and_log('A: {0} / {2}\tB: {1} / {2}'.format(num_each_class[0], num_each_class[1], total))
return num_each_class
if __name__ == '__main__':
parser = config_util.make_parser()
args = parser.parse_args()
config = config_util.parse_config(args.config)
config_util.update_config(config, args)
# Insert the git hash into the config so that it gets logged
config['Logging']['githash'] = get_git_hash()
try:
os.mkdir(config['Logging']['log_root'])
except FileExistsError:
pass
model_config = config['Model']
if model_config['model'] not in nn_models.models.keys():
print('--model {} not in list of available models'.format(args.model))
sys.exit(1)
type=float,
default=None,
help='If set, does elastic net with the given l1_ratio.'
' (0 corresponds to purely L2, 1 corresponds to purely L1)')
parser.add_argument('--plot',
action='store_true',
help='If set, plots the paths at the end')
parser.add_argument(
'--val-precision',
action='store_true',
help='Use average precision as the "val accuracy" metric')
parser.add_argument('--order-rescale', nargs='+', type=float, default=None)
parser.add_argument('--doping-level', type=float, default=None)
args = parser.parse_args()
config = parse_config(args.config)
config['Model']['saved_model'] = args.model_file
config['Preprocessing']['oversample'] = False
if args.doping_level is not None:
config['Loader Kwargs']['doping_level'] = args.doping_level
if torch.cuda.is_available():
print('Loaded CUDA successfully!')
device = 'cuda'
else:
print('Could not load CUDA!')
device = 'cpu'
model = nn_models.from_config(config).to(device=device)