config.yaml

# Model configuration
model:
  hidden_dims: [256, 128]
  embed_dim: 64
  dropout: 0.1

# Training / runtime settings
training:
  batch_size: 512
  epochs: 20
  learning_rate: 0.001
  weight_decay: 0.000001
  # fraction of training data to sample (use 1.0 for full dataset)
  sample_frac: 0.2
  # number of dataloader workers (0 recommended for macOS MPS)
  num_workers: 0
  log_every: 100

# Paths used by scripts (relative to repo root)
paths:
  processed_dir: data/processed
  checkpoint_dir: data/checkpoints
  vectors_dir: data/vectors
  index_dir: data/index

# FAISS index settings
faiss:
  hnsw_m: 32
  ef_construction: 200
  ef_search: 100

# Server settings (used by Docker / deployment)
server:
  host: 0.0.0.0
  port: 8000

# Misc
seed: 42