# K-FAC Memorization Suppression - Dependencies

# Core ML
torch>=2.0
transformers>=4.35
datasets>=2.14
accelerate>=0.24

# Numerical
numpy>=1.24
scipy>=1.11

# Utilities
tqdm>=4.65
python-Levenshtein>=0.21  # Fast Levenshtein distance

# Development
jupyter>=1.0
ipywidgets>=8.0

# Optional: for loading OLMo models
# ai2-olmo>=0.3.0