mt5-base-ko / README.md
seonjeongh's picture
commit files to HF hub
9cecef8
# Vocabulary Trimmed [google/mt5-base](https://huggingface.co/google/mt5-base): `seonjeongh/mt5-base-ko`
This model is a trimmed version of [google/mt5-base](https://huggingface.co/google/mt5-base) by [`vocabtrimmer`](https://github.com/asahi417/lm-vocab-trimmer), a tool for trimming vocabulary of language models to compress the model size.
Following table shows a summary of the trimming process.
| | google/mt5-base | seonjeongh/mt5-base-ko |
|:---------------------------|:------------------|:-------------------------|
| parameter_size_full | 582,401,280 | 310,904,064 |
| parameter_size_embedding | 384,172,032 | 112,674,816 |
| vocab_size | 250,112 | 73,356 |
| compression_rate_full | 100.0 | 53.38 |
| compression_rate_embedding | 100.0 | 29.33 |
Following table shows the parameter used to trim vocabulary.
| language | dataset | dataset_column | dataset_name | dataset_split | target_vocab_size | min_frequency |
|:-----------|:----------------------------|:-----------------|:---------------|:----------------|:--------------------|----------------:|
| ko | vocabtrimmer/mc4_validation | text | ko | validation | | 2 |