BoKenlm-syl - Tibetan KenLM Language Model

A KenLM n-gram language model trained on Tibetan text, tokenized with syllable tokenizer.

Model Details

Parameter	Value
Model Type	Modified Kneser-Ney 5-gram
Tokenizer	Tibetan syllable-based (botok-rs SimpleTokenizer)
Training Corpus	bo_corpus.txt
Pruning	0 0 1
Tokens	72,358,144
Vocabulary Size	83,721

N-gram Statistics

Order	Count	D1	D2	D3+
1	83,721	0.7208	1.0003	1.2406
2	1,513,875	0.6906	1.0405	1.3725
3	3,382,032	0.7399	1.0905	1.3764
4	6,201,199	0.8066	1.1508	1.3914
5	7,456,051	0.7466	1.2988	1.5043

Memory Estimates

Type	MB	Details
probing	384	assuming -p 1.5
probing	447	assuming -r models -p 1.5
trie	180	without quantization
trie	98	assuming -q 8 -b 8 quantization
trie	160	assuming -a 22 array pointer compression
trie	77	assuming -a 22 -q 8 -b 8 array pointer compression and quantization

Training Resources

Metric	Value
Peak Virtual Memory	12,333 MB
Peak RSS	3,641 MB
Wall Time	42.3s
User Time	41.4s
System Time	13.1s

Usage

import kenlm

model = kenlm.Model("BoKenlm-syl.arpa")

# Score a tokenized sentence
score = model.score("བོད་ སྐད་ ཀྱི་ ཚིག་ གྲུབ་ འདི་ ཡིན།")
print(score)

Files

• BoKenlm-syl.arpa — ARPA format language model
• README.md — This model card

License

Apache 2.0