{"id":"W4401042062","doi":"10.18653/v1/2024.semeval-1.265","title":"Edinburgh Clinical NLP at SemEval-2024 Task 2: Fine-tune your model unless you have access to GPT-4","year":2024,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"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":"SemEval; Task (project management); Computer science; Natural language processing; Artificial intelligence; Speech recognition; Engineering","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":[0.02707124,0.002359291,0.001958898,0.0017516,0.0008174782,0.002859881,0.003160504,0.004077689,0.01986729],"category_scores_gemma":[0.1004883,0.001043276,0.003019432,0.001524623,0.0008296362,0.003209095,0.00264755,0.00397353,0.01030003],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002229877,"about_ca_system_score_gemma":0.004509461,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01145861,"about_ca_topic_score_gemma":0.01679722,"domain_scores_codex":[0.988134,0.007637199,0.000770614,0.002136829,0.001059062,0.0002623249],"domain_scores_gemma":[0.9510677,0.04019938,0.001162229,0.004750521,0.001827022,0.0009931951],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00878991,0.0009682124,0.02761113,0.005901439,0.002812391,0.001378387,0.0005828979,0.1307939,0.008511049,0.005653057,0.4365502,0.3704473],"study_design_scores_gemma":[0.007662748,0.002537271,0.01979873,0.0009311268,0.001589664,0.002530341,0.0004420163,0.723986,0.02235359,0.03224441,0.1854538,0.0004704257],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2435621,0.01713825,0.3037397,0.03264062,0.003807662,0.003861931,0.2536138,0.1078952,0.03374076],"genre_scores_gemma":[0.5689449,0.001553265,0.1986541,0.005535336,0.0007054276,0.00233454,0.2048334,0.00537669,0.01206227],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.02707124,"threshold_uncertainty_score":0.1431681,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1398657191291666,"score_gpt":0.4029602116996834,"score_spread":0.2630944925705168,"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."}}