config.yml

cache:
  name: redis
  port: 6379

root_path: "/viberary"

# Data Config
data:
  books: "/viberary/src/data/goodreads_books.json"
  authors: "/viberary/src/data/goodreads_book_authors.json"
  reviews: "/viberary/src/data/goodreads_reviews_dedup.json"
training_data:
  path: "src/training_data/"
  file: "2023071112_training.parquet"
embeddings_data:
  path: "src/training_data/"
  file: "20230711_learned_embeddings.snappy"

# Logging
logging:
  path: "logging.conf"

# Redis Search Config
search:
  index_name: "viberary"
  top_k: 50
  nvecs: 800000
  dim: 768 # embedding size, needs to be same for both towers
  max_edges: 40 # maximum allowed outgoing edges for each node in the graph in each layer.
  ef: 200 # maximum allowed potential outgoing edges candidates for each node in the graph
  distance_metric: "COSINE"
  float_type: "FLOAT64"
  index_type: "HNSW"

# Model Config
model:
  name: "sentence-transformers/msmarco-distilbert-base-v3"
  onnx_path: "sentence-transformers/msmarco-distilbert-base-v3_onnx/"

# Spaces Config

spaces:
  endpoint_url: "https://nyc3.digitaloceanspaces.com"
  region_name: "nyc3"
  bucket_name: "viberary-onnx"