{"id":"W7102431007","doi":"","title":"Agent Data Protocol: Unifying Datasets for Diverse, Effective Fine-tuning of LLM Agents","year":2025,"lang":"","type":"article","venue":"arXiv (Cornell University)","topic":"Multi-Agent Systems and Negotiation","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institute for Catastrophic Loss Reduction","keywords":"Bottleneck; Variety (cybernetics); Protocol (science); Parsing; Representation (politics); Simple (philosophy); Code (set theory); Pipeline (software); Data access","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01140473,0.001627514,0.0009456744,0.002292404,0.001369863,0.003758521,0.00498681,0.001827695,0.00424115],"category_scores_gemma":[0.04328548,0.001159793,0.001766655,0.002416184,0.001641785,0.008433696,0.007782484,0.006208708,0.004946275],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001862452,"about_ca_system_score_gemma":0.004418215,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005695102,"about_ca_topic_score_gemma":0.007012364,"domain_scores_codex":[0.9923801,0.002745605,0.001050329,0.001762153,0.001666418,0.0003954447],"domain_scores_gemma":[0.9804047,0.004129057,0.000968533,0.01161729,0.002225561,0.0006549042],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002043332,0.001787922,0.02266821,0.001548621,0.0004467242,0.0004296601,0.00199175,0.1336816,0.01657045,0.07745785,0.3163356,0.4250384],"study_design_scores_gemma":[0.0006951832,0.0003521253,0.005017086,0.0003489311,0.00009449069,0.0002038875,0.0006836625,0.6060951,0.03701568,0.09997497,0.2492819,0.0002370128],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04364667,0.00077932,0.7051379,0.003130887,0.0009061539,0.001951172,0.05074213,0.1822461,0.01145962],"genre_scores_gemma":[0.1792806,0.0003415531,0.65198,0.001323771,0.000115303,0.004770315,0.1495723,0.008695433,0.003920606],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01140473,"threshold_uncertainty_score":0.06031466,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1825657341906636,"score_gpt":0.2814527220478224,"score_spread":0.09888698785715877,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}