environment.yml

name: nlp-research-template
channels:
  - conda-forge
  - nvidia # nvidia channel for cuda packages
  # We want to have a reproducible setup, so we don't want default channels,
  # which may be different for different users. All required channels should
  # be listed explicitly here.
  - nodefaults
dependencies:
  - python=3.10 # or don't specify the version and use latest stable

  # ML libraries
  - pytorch::pytorch>=2.0 # or don't specify the version and use latest stable
  # CUDA only on linux and windows
  - pytorch::pytorch-cuda=11.8 # [linux]
  - pytorch::pytorch-cuda=11.8 # [win]
  
  - wandb
  - lightning
  - transformers
  - tokenizers
  - datasets
  - evaluate
  - accelerate
  - sentencepiece

  # Good to have
  - numpy
  - scipy
  - pandas
  - seqeval
  - scikit-learn 

  # Misc
  - jsonlines
  - zstandard
  - debugpy
  - tqdm

  # Dev dependencies
  - black
  - ruff
  - jupyterlab

  # pip dependencies
  - pip
  - pip:
      - pipx
      - simple_parsing
      - print-on-steroids[exceptions]

# Non-standard section listing target platforms for conda-lock:
platforms:
  - linux-64
  - osx-arm64 # This is for M1/M2 MacBooks # Note: needs mamba 1.5.3 to work properly
  # - linux-aarch64 # This is linux-arm64. We can build Docker images for M1/M2 Macs on top of this - won't support MPS though
  - win-64