Hi,
I got something wrong with the 5 labels of training and validation, after i submitted it, i got 0/20 for this part. It wasn’t like expected output. Could anybody tell me why…?
The number of units in the output dense layer is equal to the number of classes. This means that index 0 is a class id as well.
A Tokenizer uses 0 for internal purposes. The first valid integer representation of a token starts from 1.
It’s important to account for this offset when creating class ids from tokenizer output.