{"id":"W7123361559","doi":"10.1109/icm66518.2025.11322486","title":"Efficient Pruning and Acceleration of Encoder-Based LLM Transformers on eFPGAs","year":2025,"lang":"","type":"article","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Encoder; Pruning; Transformer; Computation; Edge device; High memory; Offset (computer science); Decoding methods","routes":{"ca_aff":true,"ca_fund":true,"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":[0.000528793,0.000879791,0.000572596,0.0006652133,0.0003897861,0.0009059643,0.001688988,0.0004444448,0.005623736],"category_scores_gemma":[0.00247793,0.0003221848,0.0004672071,0.0007108973,0.0004870071,0.001845685,0.001050081,0.0008322027,0.002088186],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008896031,"about_ca_system_score_gemma":0.001513699,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00527502,"about_ca_topic_score_gemma":0.01161089,"domain_scores_codex":[0.9994843,0.0001174636,0.00003877803,0.00008302672,0.000201352,0.00007514651],"domain_scores_gemma":[0.9990882,0.0004383842,0.00006730262,0.0002016247,0.0001663471,0.00003817721],"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":[0.0008902189,0.0001963299,0.002500475,0.0003682882,0.00007251567,0.000525135,0.0003501601,0.0947833,0.06114236,0.04028532,0.01510487,0.7837811],"study_design_scores_gemma":[0.00006568468,0.0001653661,0.0003854723,0.00003120416,0.00003569573,0.0001854147,0.00009443127,0.9076571,0.06976755,0.01162843,0.009962612,0.00002107567],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05218329,0.0004419865,0.9223227,0.0001820443,0.00008234753,0.0001196937,0.0003956364,0.01873918,0.00553316],"genre_scores_gemma":[0.4472689,0.0002357458,0.5429097,0.0001946994,0.00003651098,0.0001639587,0.00120121,0.0007040297,0.007285168],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005623736,"threshold_uncertainty_score":0.01881325,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01988389092937311,"score_gpt":0.2851108188255527,"score_spread":0.2652269278961796,"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."}}