Add option to load text pairs with CSVClassificationCorpus

[alanakbik]

You can now load a data pair object with the CSV classification corpus, in case you want to train a classifier for sentence pair.

Example:

import torch

from flair.data import Corpus
from flair.datasets import CSVClassificationCorpus
from flair.embeddings import TransformerDocumentEmbeddings

# 1. get you text pair corpus
corpus = CSVClassificationCorpus("path/to/your/dataset",
                                 train_file="what.csv",
                                 column_name_map={0: "text", 1: "pair", 2: "label_entailment"},
                                 skip_header=True,
                                 in_memory=False,
                                 max_chars_per_doc=50,
                                 )

# 2. make the tag dictionary from the corpus
label_dictionary = corpus.make_label_dictionary()

# 3. initialize text pair tagger
from flair.models import TextPairClassifier

tagger = TextPairClassifier(
    document_embeddings=TransformerDocumentEmbeddings(),
    label_dictionary=label_dictionary,
)

# 4. train trainer with AdamW
from flair.trainers import ModelTrainer

trainer = ModelTrainer(tagger, corpus, optimizer=torch.optim.AdamW)

# 5. run training
trainer.train('resources/taggers/text-pair-classifier',
              learning_rate=2e-5,
              mini_batch_size=4,
              mini_batch_chunk_size=1, # this can be removed if you hae a big GPU
              train_with_dev=True,
              monitor_test=True,
              max_epochs=10)