Open JamesTV1996 opened 3 years ago
I haven't touched Tensorflow in a while. But wouldn't this be for loading a checkpoint?
saver.restore(sess, r'C:\Users\james\Desktop\Title Generation - SENG 6245\Dataset250K.csv')
It doesn't seem like normal Tensorflow checkpoint.
for i in range(0, len(val_batches_text)):
This is the first mention of val_batches_text that I found in the code. The issue seems to be that val_batches_text is completely empty that which your validation dataset is empty. Which would mean that the issue is probably somewhere outside this code snippet wherever you are preparing this val_batches_text.
Where are you loading the datasets btw?
I loaded a dataset that I cleaned from my own pc. I first ran my dataset through the preprocessing program you have posted as well. For my project, I was trying to adapt your implementation so that I can generate a title based off the description of the given problem. I assume this would work for this application. The dataset I used was originally downloaded from the StackOverflow dataset found on Google Cloud API.
It should work, but the main error seems to be that len(val_batches_text) is 0. That means the source of the bug is in the code snippet where val_batches_text is being created. For some reason no data is being loaded. If you are using my pre-processing, it has some heavy filters. Is it possible that no data is passing through the filters when you are you preprocessing your dataset?
I've tried what you said in the last comment but the data set is not loading and it return the error as the zero division. We are using your pre-processing, so please help me to reduce the heavy filter.
Probably you have to figure out what max length/min length are suitable for your dataset: https://github.com/JRC1995/Abstractive-Summarization/blob/master/Data_Pre-Processing.ipynb You have to probably change them: "text_max_len = 500 text_min_len = 25 summary_max_len = 30"
I have changed the max length/min length values as you said, but it gives the error, how can I rectify this error. I have used only the dataset reviews.csv with 50000 lines.
Trying your method for a different dataset and I am getting a ZeroDivisionError for the Training and Validation Section. I assume that something is not loading properly because there should be no zero values. Here is the code:
`import pickle import random
with tf.Session() as sess: # Start Tensorflow Session display_step = 100 patience = 5
I can get rid of the error with exception handling but I was wondering if you had and idea of why it's not working in the first place.