{"id":"W4385574232","doi":"10.18653/v1/2022.nllp-1.10","title":"Parameter-Efficient Legal Domain Adaptation","year":2022,"lang":"en","type":"article","venue":"","topic":"Artificial Intelligence in Law","field":"Social Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"Queen's University","funders":"","keywords":"Computer science; Domain adaptation; Adaptation (eye); Domain (mathematical analysis); Language model; Artificial intelligence; Legal case; Machine learning; Legal advice; Data science; Law; Mathematics","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":[0.002719735,0.001355526,0.000969158,0.001552817,0.0009372745,0.001628942,0.002719708,0.002262145,0.008612644],"category_scores_gemma":[0.01925526,0.0006398881,0.00102902,0.001109713,0.001250466,0.004244573,0.003119876,0.003348661,0.006416808],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001349443,"about_ca_system_score_gemma":0.002505235,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008449713,"about_ca_topic_score_gemma":0.01496693,"domain_scores_codex":[0.9974385,0.0009139816,0.0001277255,0.0007784287,0.0005129721,0.0002283974],"domain_scores_gemma":[0.9955873,0.001894416,0.0002315428,0.001281701,0.0008280237,0.0001769644],"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":[0.0002811259,0.0006153936,0.008140231,0.0002996567,0.0001386451,0.000299159,0.0005563569,0.3123799,0.01353664,0.0146964,0.06543054,0.583626],"study_design_scores_gemma":[0.00004208149,0.00003244777,0.0009308392,0.00004658064,0.00002123314,0.0001428743,0.0002062011,0.9638645,0.004776838,0.01691185,0.01298747,0.00003717963],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09540985,0.001873576,0.8420823,0.001912651,0.0006643004,0.0005671949,0.001870784,0.02890463,0.02671485],"genre_scores_gemma":[0.635613,0.0006868279,0.3368353,0.002237107,0.000231238,0.0006572874,0.006743636,0.002881836,0.01411378],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008612644,"threshold_uncertainty_score":0.02881211,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06623400985671193,"score_gpt":0.3431375893438574,"score_spread":0.2769035794871455,"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."}}