{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":6,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":6,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"e87bc5552faa","filters":{"venue":"CUHK Research Data Repository"}},"results":[{"id":"W6967189273","doi":"10.48668/ottxne/vi7kp6","title":"Amatrice_CAT5.v20210504.tab","year":2021,"lang":"en","type":"dataset","venue":"CUHK Research Data Repository","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Process (computing); Identification (biology); Product (mathematics)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.395295085703934,"gpt":0.5052243669251737,"spread":0.1099292812212397,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001343112,0.003671909,0.002602817,0.004520009,0.001463348,0.003933512,0.00505871,0.003671356,0.249691],"category_scores_gemma":[0.006542171,0.00127154,0.001790822,0.008172868,0.0008897074,0.002489873,0.003032961,0.002121784,0.3090571],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002463581,"about_ca_system_score_gemma":0.003218105,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02584269,"about_ca_topic_score_gemma":0.03340759,"domain_scores_codex":[0.9983662,0.0002922851,0.0001768332,0.0005435209,0.0002783541,0.0003428329],"domain_scores_gemma":[0.9965192,0.0009126369,0.000233735,0.0009914651,0.0008690853,0.0004737801],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004356038,0.00001455156,0.0001422173,0.0004307631,0.00001428335,0.000007254401,0.00001140979,0.00009656636,0.00007564516,0.0002062722,0.9982022,0.0007551755],"study_design_scores_gemma":[0.0004422008,0.00004048449,0.001829865,0.0002534509,0.00003558088,0.00005283588,0.000088933,0.0004992262,0.0006151589,0.001208966,0.9948843,0.00004900539],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.00006176462,0.00005074571,0.00004973521,0.000059286,0.00003758033,0.00001253325,0.9984885,0.0006473556,0.000592472],"genre_scores_gemma":[0.0002513381,0.00004349764,0.0001754293,0.00004969197,0.00001156163,0.00007738387,0.9985524,0.0001755275,0.0006630369],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.750309,"threshold_uncertainty_score":0.8352997,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6929673858","doi":"10.48668/vt8ump/3z9a5c","title":"edible.PacBio.fastq.MD5","year":2023,"lang":"en","type":"dataset","venue":"CUHK Research Data Repository","topic":"Invertebrate Immune Response Mechanisms","field":"Immunology and Microbiology","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Queen's University","funders":"","keywords":"Process (computing); Identification (biology); Product (mathematics)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.2289916205990932,"gpt":0.4191394015655509,"spread":0.1901477809664576,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001298305,0.003795143,0.002565346,0.00404245,0.001338239,0.003681353,0.005368422,0.003582038,0.1067144],"category_scores_gemma":[0.00520899,0.001217246,0.002012954,0.006188191,0.0008361664,0.001971821,0.00272544,0.002137925,0.1784699],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002190956,"about_ca_system_score_gemma":0.002752585,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02364304,"about_ca_topic_score_gemma":0.03176385,"domain_scores_codex":[0.9987626,0.0001951242,0.0001143607,0.0004502234,0.0002666063,0.0002110738],"domain_scores_gemma":[0.9981407,0.0004984006,0.0001702205,0.0005563282,0.0003708022,0.0002635474],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001117896,0.00002338511,0.0005336202,0.001024745,0.00004960093,0.00002422928,0.00001981172,0.0003614738,0.0001900644,0.0005001939,0.9954994,0.001661485],"study_design_scores_gemma":[0.0004042144,0.00003042449,0.002408082,0.0002268043,0.0000573357,0.00007381348,0.00005855158,0.0007750769,0.0007761387,0.001507833,0.9936339,0.00004791102],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.00008621401,0.00008598001,0.00007496742,0.00005070141,0.00001821978,0.000007428233,0.9982898,0.0008967807,0.000489875],"genre_scores_gemma":[0.0003334298,0.00006803665,0.0002014088,0.00004046395,0.000005609074,0.00004161187,0.9987522,0.0001439848,0.000413181],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.8932856,"threshold_uncertainty_score":0.3569951,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6910853651","doi":"10.48668/vt8ump","title":"Jellyfish genomes","year":2023,"lang":"en","type":"dataset","venue":"CUHK Research Data Repository","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Queen's University","funders":"","keywords":"Jellyfish; Genome; Gene; Homeobox; Genomics","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.4482967197969963,"gpt":0.502938805463396,"spread":0.05464208566639972,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000727783,0.001872128,0.001786593,0.003965015,0.001770524,0.002002402,0.002274653,0.001805851,0.0429362],"category_scores_gemma":[0.003337564,0.0009202329,0.001597887,0.007428872,0.0005320203,0.001185278,0.002778328,0.002014862,0.0387946],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001879443,"about_ca_system_score_gemma":0.003392608,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05613162,"about_ca_topic_score_gemma":0.0803755,"domain_scores_codex":[0.9990495,0.00007656223,0.00009745317,0.000326655,0.0002291868,0.000220695],"domain_scores_gemma":[0.9985595,0.0002724376,0.0001497866,0.0002796208,0.0004753508,0.0002633049],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0006780851,0.00006351838,0.004923673,0.003399485,0.000203245,0.0001749874,0.0002609922,0.0008680735,0.004017692,0.001470622,0.975813,0.008126476],"study_design_scores_gemma":[0.0004240474,0.00004086217,0.0297562,0.0003359665,0.0001227585,0.0001686937,0.0002519377,0.0004879179,0.001317438,0.001064557,0.9659656,0.0000640098],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.00109031,0.0001435548,0.00006170748,0.00004033194,0.00001276555,0.00000986447,0.9977807,0.0002232135,0.0006373976],"genre_scores_gemma":[0.0006969533,0.00006835118,0.0001576895,0.00001310323,0.000002087369,0.00002895248,0.9986973,0.00003679563,0.0002987092],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.05613162,"threshold_uncertainty_score":0.1436359,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6966470483","doi":"10.48668/vt8ump/1elgza","title":"sanderia.PacBio.fastq","year":2023,"lang":"en","type":"dataset","venue":"CUHK Research Data Repository","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Queen's University","funders":"","keywords":"Process (computing); Identification (biology); Product (mathematics)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.5571518294041781,"gpt":0.5380763801755616,"spread":0.01907544922861648,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001849942,0.005080441,0.002975535,0.004397496,0.002230019,0.003867063,0.006334874,0.00399324,0.07999779],"category_scores_gemma":[0.006050076,0.001254639,0.002384999,0.006766827,0.001368711,0.00205733,0.002942649,0.002569172,0.1603776],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003319271,"about_ca_system_score_gemma":0.00402222,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05927213,"about_ca_topic_score_gemma":0.08283284,"domain_scores_codex":[0.9979512,0.0004012095,0.0001578249,0.0006657573,0.0004192937,0.0004046409],"domain_scores_gemma":[0.9974568,0.0005074025,0.000149237,0.0008974393,0.0006558149,0.0003333113],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007190575,0.00002554597,0.0002156532,0.000309321,0.00002015048,0.00001300401,0.00001062837,0.0002038981,0.0000844535,0.0002270649,0.9973687,0.001449678],"study_design_scores_gemma":[0.0005704417,0.00005560492,0.003002016,0.0002358054,0.00006119242,0.0001288848,0.0001414046,0.002034097,0.0012589,0.001738426,0.9906998,0.000073437],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0003054265,0.000198576,0.0001565952,0.0001568834,0.000107975,0.00002687083,0.9958888,0.002000209,0.001158721],"genre_scores_gemma":[0.0005070237,0.00006731552,0.0003725784,0.00004582462,0.00001509724,0.00007056678,0.9979512,0.0001620332,0.0008083789],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.9200022,"threshold_uncertainty_score":0.2676193,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6967200374","doi":"10.48668/vt8ump/c7v1mk","title":"sanderia.PacBio.fastq.gz.MD5","year":2023,"lang":"en","type":"dataset","venue":"CUHK Research Data Repository","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Queen's University","funders":"","keywords":"Process (computing); Identification (biology); Product (mathematics)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.5571518294041781,"gpt":0.5380763801755616,"spread":0.01907544922861648,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001762992,0.004175587,0.002559382,0.00405964,0.002047575,0.003579389,0.0056533,0.003369852,0.0880791],"category_scores_gemma":[0.006201304,0.001091543,0.00199083,0.006720627,0.00122899,0.001929051,0.002831442,0.002152755,0.1724776],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003349226,"about_ca_system_score_gemma":0.003709672,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05859774,"about_ca_topic_score_gemma":0.07813758,"domain_scores_codex":[0.9980922,0.0003775488,0.000158307,0.0006143349,0.0003842613,0.0003732277],"domain_scores_gemma":[0.9973956,0.0004853846,0.0001722574,0.0008979967,0.000719495,0.0003292731],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007004994,0.00002216553,0.0002559292,0.0003225439,0.00001883645,0.00001092297,0.00001117038,0.000201453,0.0000678566,0.0002638919,0.9974114,0.001343812],"study_design_scores_gemma":[0.0004653751,0.0000413369,0.003039751,0.0002234368,0.00005140091,0.00008980127,0.0001286525,0.001471895,0.0009816529,0.001552576,0.9918942,0.00005994481],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0002166726,0.0001344416,0.0001119865,0.0001188771,0.00006753561,0.0000194294,0.9970264,0.001338121,0.0009665777],"genre_scores_gemma":[0.0004614149,0.00005924594,0.0003072741,0.00003966084,0.00001248334,0.00006163595,0.9981669,0.0001358593,0.000755491],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.9119209,"threshold_uncertainty_score":0.294654,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W6910866301","doi":"10.48668/vt8ump/rk7s2t","title":"edible.PacBio.fastq","year":2023,"lang":"en","type":"dataset","venue":"CUHK Research Data Repository","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Queen's University","funders":"","keywords":"Process (computing); Identification (biology); Product (mathematics)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.5530324821035331,"gpt":0.5411763995475676,"spread":0.01185608255596549,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001757753,0.007333559,0.003265471,0.004972026,0.002325451,0.003792587,0.00740355,0.005209717,0.08626911],"category_scores_gemma":[0.005846187,0.001541576,0.003117033,0.006353791,0.001521762,0.002339647,0.003309359,0.003331263,0.1724052],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003214788,"about_ca_system_score_gemma":0.003675748,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04585858,"about_ca_topic_score_gemma":0.06588873,"domain_scores_codex":[0.9979112,0.0003584726,0.0001587038,0.0007235326,0.0004547382,0.0003933767],"domain_scores_gemma":[0.9976761,0.0005168847,0.0001390963,0.0008355692,0.0005388465,0.0002934027],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00009394668,0.00003572489,0.0002109608,0.0004148244,0.00002565082,0.00001805816,0.00001169495,0.0002397121,0.0001594824,0.0001968761,0.9966441,0.001948957],"study_design_scores_gemma":[0.0007882923,0.00009769025,0.002862092,0.0002928773,0.00008717227,0.0002236587,0.0001417762,0.002581764,0.00230908,0.001839923,0.9886822,0.00009347744],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0004031933,0.0003358402,0.0002337365,0.0001836721,0.0001561826,0.00004169457,0.9937621,0.003535647,0.001347863],"genre_scores_gemma":[0.0005159987,0.00008792317,0.0004962401,0.00006559918,0.0000180227,0.00008022026,0.9976846,0.0002103715,0.0008410556],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.9137309,"threshold_uncertainty_score":0.288599,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}