Docs

C

In development…

Python

Requirements

• Linux platform

Installation

The library is available on the PyPI and can be installed with the command pip install ngram-lm.

Usage

Given an ARPA file, a trie can be built like this:

from ngram_lm.trie import build
lm_order = 3
arpa_path = "/path/to/arpa-file"
out_path = "/the/desired/output/path/where/trie/will/be/saved"
build(lm_order, arpa_path, out_path)

Then, the trie can be loaded into memory and queried for next word predictions:

from ngram_lm.trie import Trie
t = Trie(out_path)
context = ["ele", "foi"]
n_predictions = 5
predictions = t.next_word_predictions(context, n_predictions)
for p in predictions:
    print(p)