config.yml

create_index:
  batch_size: 10000
  fields:
    - doc_id
    - content
  indexer:
    name: "ElasticSearchIndexer"
    hparams:
      index_name: "elastic_index"
      hosts: "localhost:9200"
      algorithm: "bm25"
    other_kwargs:
      request_timeout: 60
      refresh: true

query_creator:
  size: 1000
  field: "content"
  query_pack_name: "query"

full_ranker:
  query_pack_name: "query"
  index_config:
    index_name: "elastic_index"
    hosts: "localhost:9200"
    algorithm: "bm25"
  field: "content"
  response_pack_name_prefix: "passage"


fullranker_evaluator:
  pack_name: "query"
  ground_truth_file: "data/collectionandqueries/qrels.dev.small.tsv"
  input_file: "data/collectionandqueries/queries.dev.small.tsv"
  output_file: "output/fullranker_results_dev.tsv"

reader:
  pack_name: "query"

reranker:
  size: 1000
  model_name: 'amberoad/bert-multilingual-passage-reranking-msmarco'
  max_seq_length: 512
  batch_size: 256

reranker_evaluator:
  pack_name: "query"
  ground_truth_file: "data/collectionandqueries/qrels.dev.small.tsv"
  input_file: "data/collectionandqueries/queries.dev.small.tsv"
  output_file: "output/reranker_results_dev.tsv"



qa_system:
  size: 1
  task_name: "question-answering"
  model_name: "deepset/roberta-base-squad2"

qa_evaluator:
  pack_name: "query"
  ground_truth_file: "data/dev_small_qa_references.json"
  filtered_gt_file: "data/dev_small_qa_references_filtered.json"
  predicted_output_file: "output/predicted_text_qa.txt"
  output_file: "output/qa_results.json"
  max_bleu_order: 4