{"id":"W4383605036","doi":"10.48550/arxiv.2307.03042","title":"Parameter-Efficient Fine-Tuning of LLaMA for the Clinical Domain","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Topic Modeling","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"European Commission; Engineering and Physical Sciences Research Council; National Institute for Health and Care Research; Canadian Institute of Steel Construction; UK Research and Innovation; Nvidia; Accenture; Cisco Systems","keywords":"Adapter (computing); Computer science; Downstream (manufacturing); Domain adaptation; Language model; Domain (mathematical analysis); Set (abstract data type); Artificial intelligence; Machine learning; Programming language; Engineering; Mathematics","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":[0.003561668,0.001857534,0.001213592,0.001142258,0.000622758,0.001601373,0.001765149,0.001798042,0.003751968],"category_scores_gemma":[0.01862317,0.0006538245,0.001317811,0.0007256438,0.000877922,0.002502841,0.002295317,0.003463634,0.00411289],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008681249,"about_ca_system_score_gemma":0.002002642,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005578917,"about_ca_topic_score_gemma":0.008309737,"domain_scores_codex":[0.9984603,0.0006567852,0.000118801,0.0004404719,0.0001793475,0.0001442559],"domain_scores_gemma":[0.9953701,0.002868022,0.0002468706,0.0008084844,0.0005461034,0.0001604252],"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":[0.0007543393,0.0004888591,0.006851014,0.0003810392,0.0003156091,0.0002159791,0.0002994173,0.2106571,0.03438975,0.00234618,0.01292524,0.7303754],"study_design_scores_gemma":[0.0001071672,0.0002030888,0.002502081,0.00007196501,0.00009666264,0.0002889537,0.0001446178,0.9652982,0.01704401,0.007912976,0.006260819,0.00006949758],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.08523366,0.004411294,0.8755906,0.001336357,0.0003819554,0.0003505234,0.0007069287,0.02703436,0.004954235],"genre_scores_gemma":[0.7407476,0.001123507,0.2469871,0.001424597,0.0002516706,0.000749789,0.002246134,0.001587118,0.004882468],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005578917,"threshold_uncertainty_score":0.01883608,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2546703697093897,"score_gpt":0.2633227949557442,"score_spread":0.008652425246354567,"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."}}