metadata
license: mit
!git clone https://github.com/ayulockin/llm_scratch
%cd llm_scratch
!uv pip install -r pyproject.toml
# place the tokenizers in a dir `data`
from llm import get_wmt_tokenizers
tokenizers = get_wmt_tokenizers("en-de", path="path/to/tokenizers/*.json")