When you run a neural network script (for text classification) via cronTab, the network produces completely different results compared to manually launching the script. When you manually run the script, the classification is as necessary, but for some reason, when you run the script through cronTab, the network produces completely different results. Has anyone encountered this problem? How can I fix it?
Example problem:
When manually launching, the network determines that the test text belongs to the first category, which is correct, but when you run the script through cronTab, the network produces a completely different result, relating it for some reason to the 8th category, considering that it was trained only in 6 categories.
