{
  "dev": [
    "FOMO-MRI_FOMO60K",
    "espnet_yodas_owsmv4",
    "hiyouga_geometry3k",
    "bezirganyan_LUMA",
    "callanwu_WebWalkerQA",
    "IGNF_PASTIS-HD",
    "hiyouga_math12k",
    "hails_mmlu_no_train",
    "CIFAR_30field",
    "osunlp_TravelPlanner",
    "opencompass_NeedleBench",
    "lukaemon_bbh",
    "bigcode_bigcodebench",
    "MMMU_30field"
  ],
  "test": [
    "AI4Math_MathVista",
    "BianYx_VAP-Data",
    "ChongyanChen_VQAonline",
    "Cnam-LMSSC_vibravox",
    "DigitalLearningGmbH_MATH-lighteval",
    "EleutherAI_hendrycks_math",
    "EleutherAI_lambada_openai",
    "EleutherAI_wikitext_document_level",
    "FLORES_30field",
    "Idavidrein_gpqa",
    "LeoFan01_RoboBench",
    "Lin-Chen_MMStar",
    "MINT-SJTU_RoboFAC-dataset",
    "MLS_30field",
    "MMLU_30field",
    "MMMU_MMMU",
    "MSCOCO_30field",
    "MathLLMs_VoiceAssistant-Eval",
    "MathVista_30field",
    "MohamedRashad_arabic-books",
    "NLPCoreTeam_mmlu_ru",
    "OpenGVLab_MVBench",
    "OpenGVLab_VideoChat-Flash-Training-Data",
    "PleIAs_common_corpus",
    "Rowan_hellaswag",
    "SWE-Gym_SWE-Gym",
    "SWE-bench_SWE-bench_Verified",
    "ScaleAI_SWE-bench_Pro",
    "TIGER-Lab_MMLU-Pro",
    "Visual_Genome_30field",
    "WINGNUS_ACL-OCL",
    "allenai_ai2_arc",
    "allenai_math_qa",
    "allenai_openbookqa",
    "allenai_sciq",
    "allenai_winogrande",
    "asahi417_seamless-align-enA-hiA.speaker-embedding.xlsr-2b",
    "baber_piqa",
    "ceval_ceval-exam",
    "chcorbi_helvipad",
    "cimec_lambada",
    "common-canvas_commoncatalog-cc-by",
    "deepmind_code_contests",
    "eriktks_conll2003",
    "fleaven_Retargeted_AMASS_for_robotics",
    "google-research-datasets_mbpp",
    "google-research-datasets_natural_questions",
    "google_IFEval",
    "google_boolq",
    "google_fleurs",
    "google_xtreme_s",
    "ibrahimhamamci_CT-RATE",
    "livecodebench_code_generation_lite",
    "lmms-lab_ChartQA",
    "lmms-lab_DocVQA",
    "lmms-lab_EgoLife",
    "lmms-lab_MMMU",
    "lmms-lab_POPE",
    "lmms-lab_ScienceQA",
    "lmms-lab_Video-MME",
    "lmms-lab_textvqa",
    "locuslab_TOFU",
    "mlfoundations_MINT-1T-PDF-CC-2024-18",
    "mrmrx_CADS-dataset",
    "mteb_sts22-crosslingual-sts",
    "ncoop57_mmmlu",
    "nebius_SWE-rebench",
    "nuprl_MultiPL-E",
    "openai_gsm8k",
    "openai_openai_humaneval",
    "princeton-nlp_SWE-bench",
    "princeton-nlp_SWE-bench_Lite",
    "princeton-nlp_SWE-bench_Verified",
    "rajpurkar_squad",
    "russellyq_VQA",
    "ryanmarten_OpenThoughts-1k-sample",
    "sci-benchmark_self-contradictory",
    "tanganke_dtd",
    "tanganke_eurosat",
    "tanganke_resisc45",
    "tau_commonsense_qa",
    "tencent_ArtifactsBenchmark",
    "truthfulqa_truthful_qa",
    "ucla-contextual_contextual_test",
    "uoft-cs_cifar100",
    "ybisk_piqa",
    "zai-org_LongBench",
    "ziyjiang_MMEB_Test_Instruct"
  ],
  "selection_criteria": "stratified by length (5 short/medium + 5 medium + 5 long), diversity maximized on RAI keyword density and Claude null count",
  "_meta": {
    "excluded_outliers": [
      "m-a-p_SuperGPQA"
    ],
    "excluded_reason": "Out-of-distribution token length (344K tokens; next largest is 102K). Excluded 2026-04-20 per decisions.md.",
    "total_papers_after_exclusion": 102,
    "updated": "2026-04-20T22:24:46.358516"
  }
}