Opinion to text classification code I've found

Lady_Hangaku · April 8, 2018, 9:01pm

Heyho!

I’m currently researching for work that has been done in the field of text classification, because I’m pretty new to this topic… I stumbled upon this project:

github.com

jiangqy/LSTM-Classification-Pytorch/blob/master/main.py

import os
import torch
import copy
from torch.utils.data import DataLoader
import utils.DataProcessing as DP
import utils.LSTMClassifier as LSTMC
import torch.optim as optim
import torch.nn as nn
from torch.autograd import Variable

use_plot = True
use_save = True
if use_save:
    import pickle
    from datetime import datetime

DATA_DIR = 'data'
TRAIN_DIR = 'train_txt'
TEST_DIR = 'test_txt'
TRAIN_FILE = 'train_txt.txt'

This file has been truncated. show original

Here in the main.py, it seems like the maker of the project trains the model with his testing data for every epoch. But this shouldn’t be the case normally, right? What are you thinking about that? Just want to learn more about it, thank you!

SimonW · April 8, 2018, 9:19pm

I just glanced over, but it seems that the owner just test every epoch.