bodo-tokenizers

Pre tokenized models for Bodo. This project includes all the tokenized models to be used in the Neural Machine Translation. The models include pre tokenized modelstrained using ByteLevelBPETokenizer, BPETokenizer, SentencePieceBPETokenizer, BertWordPieceTokenizer

Models file

There are two models in this repository

Models build with Google Sentencepiece

Models build with Huggingface Tokenizer

In order to use pip install tokenizers

from tokenizers import BPETokenizer

# Load a BPE Model
vocab = "./hf-models/brx-bpe-tokenizer-vocab.json"
merges = "./hf-models/brx-bpe-tokenizer-merges.txt"
tokenizer = BPETokenizer(vocab, merges)

# And then encode:
encoded = tokenizer.encode("साबद्राय बावबायसो। नों रेदि खुरनानैलʼ दोनखादों आं थांगासिनो दं ।")
print(encoded.ids)
[13173, 11108, 1309, 241, 9268, 7661, 272, 25759, 219, 6576, 323, 192]

print(encoded.tokens)
['▁साबद्राय', '▁बावबाय', 'सो।', '▁नों', '▁रेदि', '▁खुरनानै', 'लʼ', '▁दोनखादों', '▁आं', '▁थांगासिनो', '▁दं', '▁।']

# For new tokenizers

from tokenizers import Tokenizer
from tokenizers.models import BPE

vocab = "./hf-models/brx-bpe-tokenizer-vocab.json"
merges = "./hf-models/brx-bpe-tokenizer-merges.txt"
tokenizer = Tokenizer(BPE(vocab=vocab, merges=merges))
encoded = tokenizer.encode("साबद्राय बावबायसो। नों रेदि खुरनानैलʼ दोनखादों आं थांगासिनो दं ।")
print(encoded.ids)
[17205, 481, 367, 300, 127, 360, 436, 315, 2078, 760, 429, 620, 271, 375, 350, 353, 584, 355, 607, 127]
print(encoded.tokens)
['साबद्राय', 'बाव', 'बाय', 'सो', '।', 'नों', 'रे', 'दि', 'खुर', 'नानै', 'लʼ', 'दोन', 'खा', 'दों', 'आं', 'थां', 'गासि', 'नो', 'दं', '।']