{"id":"W6966676709","doi":"10.48448/b7gs-gp03","title":"Sorted LLaMA: Unlocking the Potential of Intermediate Layers of Large Language Models for Dynamic Inference","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"York University","funders":"","keywords":"Inference; Language model; Modularity (biology); Transformer; Generative grammar; Sorting; Dynamic random-access memory; Natural language","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":[0.001479212,0.00163219,0.0007908352,0.0008952667,0.000633359,0.002138872,0.002508622,0.001378161,0.007708964],"category_scores_gemma":[0.005743462,0.0009552871,0.001440129,0.0005817575,0.001071957,0.005702746,0.002439413,0.002588095,0.00321493],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001369455,"about_ca_system_score_gemma":0.001863985,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01132571,"about_ca_topic_score_gemma":0.03371003,"domain_scores_codex":[0.9993358,0.000209248,0.0000391335,0.0002337716,0.0001168038,0.00006531852],"domain_scores_gemma":[0.9983848,0.0007541081,0.0000684066,0.0005619958,0.000161065,0.00006962248],"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":[0.0009294408,0.000332556,0.005371699,0.0004476288,0.0003004087,0.000457275,0.0004354482,0.4420363,0.01877839,0.03255438,0.02708318,0.4712732],"study_design_scores_gemma":[0.00003085021,0.00004272777,0.0001822392,0.00001909691,0.0000226256,0.00004183326,0.00002522944,0.9662338,0.004159995,0.02536614,0.003861026,0.00001446827],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04369603,0.001118984,0.8981569,0.0008142935,0.0001660085,0.0001450327,0.001694169,0.04752135,0.006687188],"genre_scores_gemma":[0.5567561,0.0004174794,0.4229516,0.001020578,0.00009577223,0.0003020012,0.006109417,0.003398643,0.008948503],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01132571,"threshold_uncertainty_score":0.02578908,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01711460245891395,"score_gpt":0.330973591178419,"score_spread":0.313858988719505,"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."}}