pyproject.toml

[build-system]
requires = ["setuptools"]
build-backend = "setuptools.build_meta"

[project]
name = "lamoto"
version = "2025.01.01"  # Don't forget to change to change the lamoto.__version__ variable.
requires-python = ">= 3.10"
authors = [
  {name = "Thomas Bauwens", email = "thomas.bauwens@kuleuven.be"}
]
maintainers = [
  {name = "Thomas Bauwens", email = "thomas.bauwens@kuleuven.be"}
]
description = "Language Modelling Tasks as Objects (LaMoTO) provides a framework for language model training (masked and causal, pretraining and finetuning) where the tasks, not just the models, are classes themselves."
readme = "README.md"
license = {file = "LICENSE"}
keywords = ["language models", "LMs", "NLP", "PyTorch", "OOP", "object-oriented"]
classifiers = [
    "Programming Language :: Python :: 3",
    "Operating System :: OS Independent",
]
dependencies = [
    "transformers >= 4.45.1",  # Note: training is unstable between >4.45.2 and at least 4.48.0.
    "datasets >= 2.18.0",
    "evaluate >= 0.4.1",
    "pyfiglet >= 1.0.2",
    "wandb",
    "scikit-learn",  # Technically a dependency of evaluate, but they refuse to add it to their metadata.
    "conllu",
    "zstandard"  # Needed for SlimPajama dataset. HuggingFace won't tell you, of course. Instead, SentencePiece will complain that your corpus has 0 sentences.
]

# These GitHub dependencies aren't optional, but if you already have them installed as editable, pip will override
# that installation with an installation in site-packages. Hence, if this is the case for you, install these manually.
[project.optional-dependencies]
github = [
    "archit[github]   @ git+https://github.com/bauwenst/ArchIt",
    "tktkt[github]    @ git+https://github.com/bauwenst/TkTkT",
    "modest[github]   @ git+https://github.com/bauwenst/MoDeST",
    "fiject           @ git+https://github.com/bauwenst/fiject",

    "supar            @ git+https://github.com/bauwenst/supar"
]