protobuf<7,>=6.33.5
blake3>=1.0.10
packaging>=24
ipython<10,>=9.17.1
tokenizers<0.24,>=0.23.1
fasteners>=0.19
regex>=2025.10.22
tldextract<6,>=5.3
spacy<4,>=3.8
fsspec[http]>=2023.12.0
marin-dupekit==0.1.1
pyarrow<24,>=18
transformers<6,>=5.10
sentence-transformers<6,>=5.7
einops<1,>=0.8
datasets<6,>=5.0.1

[:sys_platform != "darwin" or platform_machine != "x86_64"]
datatrove==0.10.1
fasttext-numpy2-wheel>=0.9.2
torch<3,>=2.5

[:sys_platform == "darwin" and platform_machine == "x86_64"]
datatrove==0.3.0
numpy<2,>=1.26
fasttext-wheel==0.9.2
torch<2.3,>=2.2

[huggingface]
datasets<6,>=5.0.1
