{"id":"W4386273179","doi":"10.2139/ssrn.4733627","title":"Speech Self-Supervised Representations Benchmarking: A Case for Larger Probing Heads","year":2024,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"Concordia University","funders":"","keywords":"Benchmarking; Computer science; Ranking (information retrieval); Inference; Downstream (manufacturing); Task (project management); Generalization; Feature (linguistics); Set (abstract data type); Artificial intelligence; Architecture; Machine learning; Natural language processing; Engineering; Mathematics; Geography","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01439235,0.001224056,0.002048881,0.0009287573,0.001349329,0.002771685,0.004049137,0.003842878,0.01306165],"category_scores_gemma":[0.06787573,0.0004874058,0.0007974447,0.001405232,0.001890971,0.005323101,0.005474724,0.003115617,0.003593171],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009672928,"about_ca_system_score_gemma":0.001541297,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003217125,"about_ca_topic_score_gemma":0.004925088,"domain_scores_codex":[0.987357,0.00617517,0.000733373,0.002588615,0.002394449,0.0007514155],"domain_scores_gemma":[0.9584442,0.02095922,0.0007171858,0.01305259,0.005816917,0.001009849],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.003956913,0.001209541,0.009962489,0.001021923,0.0005061434,0.0007807676,0.001469412,0.1458718,0.05514231,0.0148796,0.03265017,0.732549],"study_design_scores_gemma":[0.0002601176,0.001555435,0.009200254,0.0001917067,0.0002264836,0.0009757577,0.001600691,0.8188267,0.08936004,0.04137878,0.03623619,0.0001878837],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.31758,0.002197563,0.6356847,0.002609626,0.001083264,0.000567033,0.003120383,0.02288067,0.01427671],"genre_scores_gemma":[0.8180303,0.0001971386,0.1671887,0.0008286976,0.0001705905,0.0003822324,0.005790458,0.0028221,0.00458983],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01439235,"threshold_uncertainty_score":0.07611489,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02543416354791751,"score_gpt":0.2923444770256426,"score_spread":0.2669103134777251,"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."}}