Now that we're able to download and unpack the MultiNLI and SICK datasets, we want to pipeline these datasets to the model for training.
This so far has step 1, which is reconfiguring some functions a bit. The MultiNLI dataset is structured (for our purposes) just the same as NLI, but doesn't have {train, valid, test} files.
Description
Now that we're able to download and unpack the MultiNLI and SICK datasets, we want to pipeline these datasets to the model for training.
This so far has step 1, which is reconfiguring some functions a bit. The MultiNLI dataset is structured (for our purposes) just the same as NLI, but doesn't have {train, valid, test} files.