towards_data_science_42.py

import padl
import torch
from padl import params

from my_codebase.pipelines import build_string_processor, build_preprocessor, build_model
from my_codebase.models import build_classifier
from my_codebase.transforms import Dictionary, build_dictionary

nn = padl.transform(torch.nn)

string_processor = build_string_processor()
words = padl.value(build_dictionary(string_processor, 'corpus.txt'))
dictionary = Dictionary(words)

preprocess = build_preprocessor(dictionary, string_processor)

rnn = torch.nn.GRU(
    num_layers=1,
    **params(
        'rnn',
        input_size=len(dictionary),
        hidden_size=512,
    ),
)

layer = build_classifier(
    rnn,
    len(dictionary),
)

m = build_model(layer, preprocess)