lineflow 0.2.0

Framework-Agnostic NLP Data Loader in Python

lineflow: Framework-Agnostic NLP Data Loader in Python

lineflow is a simple text dataset loader for NLP deep learning tasks.

• lineflow was designed to use in all deep learning frameworks.
• lineflow enables you to build pipelines and it's lazy evaluation.
• lineflow supports some functional API.

Installation

To install lineflow, simply:

$ pip install lineflow

If you'd like to use lineflow with AllenNLP:

$ pip install "lineflow[allennlp]"

Also, if you'd like to use lineflow with torchtext:

$ pip install "lineflow[torchtext]"

Usage

lineflow.TextDataset expects line-oriented text files:

import lineflow as lf

'''/path/to/text will looks like below:
i 'm a line 1 .
i 'm a line 2 .
i 'm a line 3 .
'''


def preprocess(x):
    return x.split()

ds = lf.TextDataset('/path/to/text')
ds.first()  # "i 'm a line 1 ."
ds[1]  # "i 'm a line 2 ."

ds = ds.map(preprocess)
ds.first()  # ["i", "'m", "a", "line", "1", "."]

ds = lf.TextDataset(['/path/to/text', '/path/to/text'])
ds.first()  # ("i 'm a line 1 .", "i 'm a line 1 .")

ds = ds.map(lambda x: (x[0].split(), x[1].split()))
ds.first()  # (["i", "'m", "a", "line", "1", "."], ["i", "'m", "a", "line", "1", "."])

Use lineflow with AllenNLP

Use lineflow with AllenNLP:

import math

from allennlp.common.tqdm import Tqdm
from allennlp.data.vocabulary import Vocabulary
from allennlp.data.iterators import BucketIterator

from lineflow.datasets import Seq2SeqDataset


ds = Seq2SeqDataset(
    source_file_path='/path/to/source',
    target_file_path='/path/to/target'
).to_allennlp()

vocab = Vocabulary.from_instances(ds)

iterator = BucketIterator(sorting_keys=[('source_tokens', 'num_tokens')])
iterator.index_with(vocab)

num_batches = math.ceil(len(ds) / iterator._batch_size)

for batch in Tqdm.tqdm(iterator(train, num_epochs=1), total=num_batches):
    ...  # Your training code here

See more in examples