mcqa:
  resources_servers:
    mcqa:
      entrypoint: app.py
      grading_mode: null
      domain: knowledge
      verified: false
      description: Multi-choice question answering problems
      value: Improve benchmarks like MMLU / GPQA / HLE
mcqa_simple_agent:
  responses_api_agents:
    simple_agent:
      entrypoint: app.py
      resources_server:
        type: resources_servers
        name: mcqa
      model_server:
        type: responses_api_models
        name: policy_model
      datasets:
      - name: train
        type: train
        jsonl_fpath: resources_servers/mcqa/data/train.jsonl
        gitlab_identifier:
          dataset_name: syn_gpqa_v1.1
          version: 1.1.2
          artifact_fpath: filtered_decontaminated.jsonl
        huggingface_identifier:
          repo_id: nvidia/Nemotron-RL-knowledge-mcqa
        license: Apache 2.0
      - name: validation
        type: validation
        jsonl_fpath: resources_servers/mcqa/data/validation.jsonl
        # Unified dataset source. `type` selects the backend; here the split
        # is pulled from the HuggingFace Hub. `artifact_fpath` is optional for huggingface sources.
        source:
          type: huggingface
          repo_id: nvidia/Nemotron-RL-knowledge-mcqa
        license: Apache 2.0
      - name: example
        type: example
        jsonl_fpath: resources_servers/mcqa/data/example.jsonl
      - name: example_with_template_metadata
        type: example
        jsonl_fpath: resources_servers/mcqa/data/example_with_template_metadata.jsonl