rolemrc: resources_servers: rolemrc: entrypoint: app.py domain: instruction_following verified: false description: RoleMRC role-play MRC scored with reference metrics (ROUGE/BLEU/METEOR/BERTScore); ROUGE-L is the reward. value: Improve role-play instruction following and knowledge-grounded responses. mode: reference # BERTScore is part of the reported reference metrics; it downloads a # roberta-large checkpoint on first use. Set false (and drop # bert-score from requirements.txt) for a lightweight reward signal. include_bertscore: true rolemrc_simple_agent: responses_api_agents: simple_agent: entrypoint: app.py resources_server: type: resources_servers name: rolemrc model_server: type: responses_api_models name: policy_model datasets: - name: example type: example jsonl_fpath: resources_servers/rolemrc/data/example.jsonl license: MIT # Build with `python resources_servers/rolemrc/prepare_rolemrc.py`. - name: test type: validation jsonl_fpath: resources_servers/rolemrc/data/test.jsonl huggingface_identifier: repo_id: Junrulu/RoleMRC artifact_fpath: roleMRC_test.jsonl license: MIT rolemrc_environment_server: environment_servers: single_agent_turn_legacy: entrypoint: app.py resources_server: type: resources_servers name: rolemrc agent_server: type: responses_api_agents name: rolemrc_simple_agent resources_tool_transports: - direct_http