{"id":"W7104481854","doi":"10.71781/10885","title":"Towards efficient large language models : training low-bitwidth variants and low-rank decomposition of pretrained models","year":2024,"lang":"en","type":"dissertation","venue":"Open MIND","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Oak Ridge National Laboratory; Office of Science; Canadian Institute for Advanced Research; Canada Excellence Research Chairs, Government of Canada; U.S. Department of Energy","keywords":"Training set; Statistical analysis; Context (archaeology)","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":[0.0007429103,0.0003351045,0.0005274083,0.0002505561,0.00010111,0.0005112497,0.001189283,0.0002811546,0.00008552177],"category_scores_gemma":[0.00001925976,0.0003311548,0.0001075695,0.0002865268,0.00001803535,0.0006108688,0.0003701855,0.0003300293,0.000016886],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006373321,"about_ca_system_score_gemma":0.0004082809,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009995112,"about_ca_topic_score_gemma":0.0001056418,"domain_scores_codex":[0.9975373,0.00007840572,0.0005717648,0.0009427358,0.0004787041,0.000391084],"domain_scores_gemma":[0.9988549,0.00004937371,0.0002352721,0.0006210059,0.000113353,0.0001261474],"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":[0.0001258589,0.0003314546,8.762956e-7,0.0009125221,0.000230252,0.0001755241,0.3528855,0.1269378,0.008846338,0.01745962,0.00004511572,0.4920491],"study_design_scores_gemma":[0.0005437123,0.00004716386,0.0000205569,0.001147116,0.00005964874,0.00001125229,0.002147675,0.9886455,0.00271983,0.00430972,0.00001024904,0.0003375816],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5624469,0.0005898811,0.4144196,0.00004847208,0.0005650427,0.0006947756,0.0000967386,0.00001550511,0.02112312],"genre_scores_gemma":[0.9373785,0.0000158966,0.060961,0.00002411132,0.00005974702,0.00003332901,0.0002578578,0.00003627596,0.001233333],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8617077,"threshold_uncertainty_score":0.9999141,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03243441090778062,"score_gpt":0.3186131654673854,"score_spread":0.2861787545596048,"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."}}