save_to_ckpt(ckptfile, start_epoch + additional_epoch, net, optimizer,
scheduler)
# save the final model
torch.save(net.state_dict(), ckptfile)
ndcg_result = eval_ndcg_at_k(net, device, df_valid, valid_loader, 100000,
[10, 30], start_epoch + additional_epoch,
writer)
print(
get_time(), "finish training " + ", ".join(
["NDCG@{}: {:.5f}".format(k, ndcg_result[k])
for k in ndcg_result]), '\n\n')
if __name__ == "__main__":
parser = get_args_parser()
parser.add_argument("--sigma", dest="sigma", type=float, default=1.0)
args = parser.parse_args()
train(
args.start_epoch,
args.additional_epoch,
args.lr,
args.optim,
args.leaky_relu,
ndcg_gain_in_train=args.ndcg_gain_in_train,
sigma=args.sigma,
double_precision=args.double_precision,
standardize=args.standardize,
small_dataset=args.small_dataset,
debug=args.debug,
output_dir=args.output_dir,
from datetime import datetime
from flask import request
from flask_restful import Resource
from config import db
from models.user import User
from schemas.user import UserSchema
from utils import get_args_parser
user_schema = UserSchema(many=False)
parser = get_args_parser([
{'name': "num_mark_tasks", 'type': int, 'required': True,
'help': 'number of mark tasks the user has'},
{'name': "password", 'type': str, 'required': False,
'help': 'unprocessed password of the user'},
{'name': "name", 'type': str, 'required': True,
'help': 'name of the user'}
])
class UserResource(Resource):
"""Resource to handle CRUD operations for users table"""
@staticmethod
def get(user_id: int):
"""
Returns single user
"""
if not user_id:
import numpy as np
import os
import sys
import torch
from torch.utils.data import DataLoader
from augmentation.augmentation import RootAugmentation, RootBaseTransform
from dataloader.Eco2018Loader import DeviceLoader, Eco2018
from loss.loss import loss_fn, loss_fn2
from model.Textnet import Textnet
from model.functions import fit
from utils import get_device, get_args_parser, make_output_dir_name, print_config_file
if __name__ == '__main__':
args = get_args_parser().parse_args()
# Create output directory if it doesn't exist
output_dir = make_output_dir_name(args)
try:
os.makedirs(output_dir, exist_ok=True)
except IOError as e:
sys.exit(f'[ERROR] Could not create output directory: {e}')
# Write configuration to log file
try:
print_config_file(output_dir, args)
except IOError as e:
sys.exit(f'[ERROR] Could not write to output directory: {e}')
print('Running on device:', get_device())
import os
import sys
import torch
import torch.nn as nn
import my_model
import model
import utils
import data
import time
start = time.time()
# prepare
eval_batch_size = 10
args = utils.get_args_parser()
print('\n')
print('V ' * 80)
print('\n')
print(args)
print('\n')
print('#' * 80)
print("start!====")
# Set the random seed manually for reproducibility.
torch.manual_seed(args.seed)
device = utils.check_device(args)
if args.use_MyGRU:
print("train by MyGRU!")
###############################################################################
# Load data