# synthetic-data-kit default config
# Generated by llm-trainer bootstrap. Edit via the Config tab in the dashboard.

paths:
  input:     /opt/synthetic/synthetic-data-kit/data/input
  parsed:    /opt/synthetic/synthetic-data-kit/data/parsed
  generated: /opt/synthetic/synthetic-data-kit/data/generated
  curated:   /opt/synthetic/synthetic-data-kit/data/curated
  final:     /opt/synthetic/synthetic-data-kit/data/final

llm:
  provider: ollama
  api_base: http://localhost:11434
  model:    llama3.1:8b
  temperature: 0.3
  max_tokens: 2048

ingest:
  chunk_size:    2000
  chunk_overlap: 200

create:
  default_pair_type: qa
  default_num_pairs: 50

curate:
  threshold: 7.0
  use_llm_judge: true

save:
  default_format: jsonl