blester125 / word-vectors

Easy loading of word vectors and a faster format.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Benchmarking example

blester125 opened this issue · comments

Read Google News vectors and write as Dense and as Glove (conditionally if they don't exist)

Output stats on data (vocab size, vector dim), read speed, and file size

[Benchmark] Beginning Benchmark...
[Benchmark] Vocab Size: 3000000
[Benchmark] Vector dim: 300
[Benchmark] Running 5 trials.
[Benchmark] Glove size: 10.035GB
[Benchmark] Glove: 187.35 ± 3.83
[Benchmark] W2V size: 3.39GB
[Benchmark] W2V: 19.44 ± 0.24
[Benchmark] Dense size: 3.63GB
[Benchmark] Dense: 15.82 ± 0.30