{"id":"W4416960172","doi":"10.48550/arxiv.2512.01405","title":"Fantastic Features and Where to Find Them: A Probing Method to combine Features from Multiple Foundation Models","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Engineering and Physical Sciences Research Council; Alliance de recherche numérique du Canada","keywords":"Scalability; Exploit; Hyperparameter; Focus (optics); Transformer; Feature (linguistics); Robustness (evolution); Artificial neural network; Joint (building)","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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0004568899,0.0003815953,0.0004422335,0.0002298968,0.0002641789,0.0005926923,0.0009475821,0.0002388975,0.0000251519],"category_scores_gemma":[0.0002855103,0.0003688751,0.00009434641,0.0003702693,0.00002463008,0.00030216,0.001891655,0.0007366923,0.00006630437],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000949411,"about_ca_system_score_gemma":0.0001587015,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007352956,"about_ca_topic_score_gemma":0.0002565487,"domain_scores_codex":[0.9975788,0.0002679154,0.0003396019,0.00112705,0.0003329833,0.0003536201],"domain_scores_gemma":[0.9980474,0.0006144785,0.000177604,0.000781065,0.0001613797,0.0002180393],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003764745,0.0003489629,0.04299631,0.0006879081,0.0006367256,0.00006451357,0.06690766,0.5055408,0.01883714,0.0360936,0.01079531,0.3167145],"study_design_scores_gemma":[0.002441236,0.0002085966,0.4072004,0.002435662,0.0001727765,0.00002860613,0.001018156,0.5332155,0.003548192,0.03781258,0.009792957,0.002125338],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1065256,0.0002715473,0.8871812,0.002997296,0.0006020742,0.0008172445,0.00002551062,0.0002354037,0.0013441],"genre_scores_gemma":[0.6021095,0.00002543324,0.3928824,0.001485629,0.0001035629,0.0001290839,0.00006595851,0.00002240169,0.003176005],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.4955839,"threshold_uncertainty_score":0.9998763,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06239051884779147,"score_gpt":0.3089751941152041,"score_spread":0.2465846752674126,"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."}}