[
  {
    "source": "OpenJev-Decision-pilot",
    "rows": 260,
    "categories": [
      "constraint_planning",
      "evidence_validation",
      "policy_exception",
      "quantitative_temporal_reasoning",
      "score_rubric",
      "tool_skill_selection"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "adaptive_rag",
    "rows": 267,
    "categories": [
      "intent_routing"
    ],
    "urls": [
      "https://github.com/starsuzi/Adaptive-RAG"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "ag_news",
    "rows": 18,
    "categories": [
      "intent_routing"
    ],
    "urls": [
      "https://huggingface.co/datasets/fancyzhx/ag_news"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "arc",
    "rows": 3,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/ai2_arc"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "banking77",
    "rows": 2,
    "categories": [
      "intent_routing"
    ],
    "urls": [
      "https://huggingface.co/datasets/PolyAI/banking77"
    ],
    "recorded_terms": [
      "CC-BY-4.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "bfcl_v4",
    "rows": 4,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [
      "https://github.com/ShishirPatil/gorilla/tree/main/berkeley-function-call-leaderboard"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "bird_mini_dev",
    "rows": 80,
    "categories": [
      "code_database"
    ],
    "urls": [
      "https://huggingface.co/datasets/birdsql/bird_mini_dev"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "boolq",
    "rows": 52,
    "categories": [
      "reading_comprehension"
    ],
    "urls": [
      "https://github.com/google-research-datasets/boolean-questions"
    ],
    "recorded_terms": [
      "CC-BY-SA-3.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "codecontests",
    "rows": 45,
    "categories": [
      "code_database"
    ],
    "urls": [
      "https://github.com/google-deepmind/code_contests"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "complexfuncbench_agent_context",
    "rows": 167,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [
      "https://huggingface.co/datasets/zai-org/ComplexFuncBench"
    ],
    "recorded_terms": [
      "Apache-2.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "contractnli",
    "rows": 128,
    "categories": [
      "policy_exception"
    ],
    "urls": [
      "https://stanfordnlp.github.io/contract-nli/"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "cosmosqa",
    "rows": 75,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://github.com/wilburOne/cosmosqa"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "cruxeval",
    "rows": 60,
    "categories": [
      "code_database"
    ],
    "urls": [
      "https://github.com/facebookresearch/cruxeval"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "dair_emotion",
    "rows": 77,
    "categories": [
      "emotion_recognition"
    ],
    "urls": [
      "https://huggingface.co/datasets/dair-ai/emotion"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "enron_spam",
    "rows": 15,
    "categories": [
      "evidence_validation"
    ],
    "urls": [
      "https://www2.aueb.gr/users/ion/data/enron-spam/"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "finqa",
    "rows": 6,
    "categories": [
      "quantitative_temporal_reasoning"
    ],
    "urls": [
      "https://github.com/czyssrs/FinQA"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "freshqa",
    "rows": 165,
    "categories": [
      "quantitative_temporal_reasoning"
    ],
    "urls": [
      "https://github.com/freshllms/freshqa"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "go_emotions",
    "rows": 101,
    "categories": [
      "emotion_recognition"
    ],
    "urls": [
      "https://github.com/google-research/google-research/tree/master/goemotions"
    ],
    "recorded_terms": [
      "Apache-2.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "hellaswag",
    "rows": 15,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://rowanzellers.com/hellaswag/"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "helpsteer2",
    "rows": 70,
    "categories": [
      "answer_preference"
    ],
    "urls": [
      "https://huggingface.co/datasets/nvidia/HelpSteer2"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "hover",
    "rows": 39,
    "categories": [
      "reading_comprehension"
    ],
    "urls": [
      "https://github.com/hover-nlp/hover"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "humaneval_plus",
    "rows": 1,
    "categories": [
      "code_database"
    ],
    "urls": [
      "https://github.com/evalplus/humanevalplus_release/releases/tag/v0.1.10"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "if_rlvr_adapted",
    "rows": 15,
    "categories": [
      "evidence_validation"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/IF_multi_constraints_upto5"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "judgebench",
    "rows": 36,
    "categories": [
      "answer_preference"
    ],
    "urls": [
      "https://github.com/ScalerLab/JudgeBench"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "logiqa2",
    "rows": 141,
    "categories": [
      "deductive_reasoning"
    ],
    "urls": [
      "https://github.com/csitfun/LogiQA2.0"
    ],
    "recorded_terms": [
      "CC-BY-NC-SA-4.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "longbench_pro",
    "rows": 3,
    "categories": [
      "reading_comprehension"
    ],
    "urls": [
      "https://huggingface.co/datasets/caskcsg/LongBench-Pro"
    ],
    "recorded_terms": [
      "Apache-2.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "longbench_v2",
    "rows": 164,
    "categories": [
      "agent_state_tracking",
      "code_database",
      "reading_comprehension"
    ],
    "urls": [
      "https://github.com/THUDM/LongBench"
    ],
    "recorded_terms": [
      "Apache-2.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "massive",
    "rows": 44,
    "categories": [
      "intent_routing"
    ],
    "urls": [
      "https://huggingface.co/datasets/AmazonScience/massive"
    ],
    "recorded_terms": [
      "CC-BY-4.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "mathqa",
    "rows": 50,
    "categories": [
      "quantitative_temporal_reasoning"
    ],
    "urls": [
      "https://math-qa.github.io/"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "mmdm_context_state_programs",
    "rows": 441,
    "categories": [
      "agent_state_tracking"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "mmdm_v2_mutual_reports",
    "rows": 359,
    "categories": [
      "option_logic"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "mmdm_v2_practical_plans",
    "rows": 240,
    "categories": [
      "constraint_planning"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "mmdm_v2_revision_workflow",
    "rows": 250,
    "categories": [
      "workflow_decision"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "mmdm_v2_sampled_rubric",
    "rows": 598,
    "categories": [
      "score_rubric"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "mmdm_v2_transactional_state",
    "rows": 80,
    "categories": [
      "agent_state_tracking"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "mmlu_auxiliary",
    "rows": 6,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/cais/mmlu"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "mmlu_pro",
    "rows": 3,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/TIGER-Lab/MMLU-Pro"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "more_skillbank_review_v2_routing",
    "rows": 31,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "musique",
    "rows": 18,
    "categories": [
      "reading_comprehension"
    ],
    "urls": [
      "https://github.com/StonyBrookNLP/musique"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "new_program_composition",
    "rows": 152,
    "categories": [
      "option_logic"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "openbookqa",
    "rows": 10,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/openbookqa"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "piqa",
    "rows": 28,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/ybisk/piqa"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "public_style_executable",
    "rows": 157,
    "categories": [
      "evidence_validation",
      "policy_exception",
      "quantitative_temporal_reasoning",
      "reading_comprehension",
      "score_rubric"
    ],
    "urls": [],
    "recorded_terms": [
      "MIT (original MMDM contributions)"
    ],
    "type": "constructed mechanism"
  },
  {
    "source": "qasc",
    "rows": 23,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/qasc"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "quality",
    "rows": 156,
    "categories": [
      "reading_comprehension"
    ],
    "urls": [
      "https://github.com/nyu-mll/quality"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "retrievalqa",
    "rows": 154,
    "categories": [
      "evidence_validation"
    ],
    "urls": [
      "https://github.com/hyintell/RetrievalQA"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "reward_bench_2",
    "rows": 68,
    "categories": [
      "answer_preference"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/reward-bench-2"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "rm_bench",
    "rows": 59,
    "categories": [
      "answer_preference"
    ],
    "urls": [
      "https://github.com/THU-KEG/RM-Bench"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "ruletaker",
    "rows": 73,
    "categories": [
      "deductive_reasoning"
    ],
    "urls": [
      "https://github.com/allenai/ruletaker"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "sciq",
    "rows": 12,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/sciq"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "skilldag",
    "rows": 47,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "skillsbench",
    "rows": 18,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [
      "https://github.com/benchflow-ai/skillsbench"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "snli",
    "rows": 57,
    "categories": [
      "deductive_reasoning"
    ],
    "urls": [
      "https://nlp.stanford.edu/projects/snli/"
    ],
    "recorded_terms": [
      "CC-BY-SA-4.0"
    ],
    "type": "source adaptation"
  },
  {
    "source": "social_iqa",
    "rows": 76,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/social_i_qa"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "spider",
    "rows": 12,
    "categories": [
      "code_database"
    ],
    "urls": [
      "https://yale-lily.github.io/spider"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "tatqa",
    "rows": 3,
    "categories": [
      "quantitative_temporal_reasoning"
    ],
    "urls": [
      "https://github.com/NExTplusplus/TAT-QA"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "toolcall_risk",
    "rows": 3,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "toolret",
    "rows": 14,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [
      "https://huggingface.co/datasets/hugFu/ToolRet-Queries"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "tooltalk_agent_context",
    "rows": 72,
    "categories": [
      "tool_skill_selection"
    ],
    "urls": [
      "https://github.com/microsoft/ToolTalk"
    ],
    "recorded_terms": [
      "MIT"
    ],
    "type": "source adaptation"
  },
  {
    "source": "toxic_chat",
    "rows": 65,
    "categories": [
      "policy_exception"
    ],
    "urls": [
      "https://huggingface.co/datasets/lmsys/toxic-chat"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "winogrande",
    "rows": 22,
    "categories": [
      "world_knowledge"
    ],
    "urls": [
      "https://huggingface.co/datasets/allenai/winogrande"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  },
  {
    "source": "xnli",
    "rows": 38,
    "categories": [
      "evidence_validation"
    ],
    "urls": [
      "https://github.com/facebookresearch/XNLI"
    ],
    "recorded_terms": [],
    "type": "source adaptation"
  }
]