ifbench:
  resources_servers:
    ifbench:
      entrypoint: app.py
      domain: instruction_following
      verified: false
      description: IFBench instruction following evaluation using AllenAI's IFBench library (57 instruction types)
      value: Improve IFBench instruction following

ifbench_simple_agent:
  responses_api_agents:
    simple_agent:
      entrypoint: app.py
      resources_server:
        type: resources_servers
        name: ifbench
      model_server:
        type: responses_api_models
        name: policy_model
      datasets:
      - name: example
        type: example
        jsonl_fpath: resources_servers/ifbench/data/example.jsonl