schema_version: facet_pilot_config_v2

experiment:
  id: facet-deepseek-b40-v3

topics:
  path: trec-rag-data/trec-rag-2026/test-data/trec_rag_2026_queries.tsv

execution:
  topic_workers: 2

retrieval:
  index: climbmix-400b
  cache_dir: cache/retrieval/pyserini_remote
  query_sources: [original, subnarrative]
  documents_per_query: 1000
  corpus_epoch: climbmix-400b-operator-archive-facet-2025-v2-sha256-2ccad901eb908ac14747bafd2069f7c8aa96ecaaa8cc30457ff47c73ce3bf81f

passage:
  model: mixedbread-ai/mxbai-rerank-base-v2
  score_cache_dir: cache/reranker
  # Pinned rather than `auto` so the sealed passage-search identity is the same
  # string on every host. `auto` resolves per hardware, and a checkpoint scored
  # where it resolved to `cpu` cannot be resumed or exported where it resolves
  # to `cuda`. ROCm reports itself as `cuda`, so this covers the ROCm host and a
  # rented NVIDIA box alike.
  device: cuda
  passages_per_query: 100
  chunk_max_characters: 3500
  chunk_overlap_characters: 350

nuggets:
  evidence_budget_per_subnarrative: 40
  maximum_claims_per_subnarrative: 20
  maximum_supporting_documents_per_claim: 3