{"id":"W4414726848","doi":"10.1145/3737902.3768352","title":"WiP: Efficient Speculative Decoding for AI PCs via Hierarchical N-Gram Retrieval","year":2025,"lang":"en","type":"article","venue":"","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Advanced Micro Devices (Canada)","funders":"","keywords":"Decoding methods; Speedup; Inference; Look-ahead; Quality (philosophy); Face (sociological concept)","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002617092,0.0001234932,0.0001564041,0.0001145758,0.0002345235,0.0001737903,0.0006423477,0.00006449527,0.00001866519],"category_scores_gemma":[0.00008375719,0.00009551745,0.00008502416,0.0004337624,0.00004325261,0.0001802719,0.0005845347,0.0001451243,0.00001703071],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005255988,"about_ca_system_score_gemma":0.00007864225,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001155507,"about_ca_topic_score_gemma":0.000001956807,"domain_scores_codex":[0.9988031,0.0000356439,0.0002136938,0.0004445808,0.0002195047,0.0002835028],"domain_scores_gemma":[0.9990565,0.0002531118,0.00003941178,0.0004660738,0.00009966717,0.00008522123],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00009120698,0.0002196846,0.0001104421,0.00003156087,0.00003812281,0.00001073335,0.0003062096,0.001260375,0.001694257,0.7823303,0.01370936,0.2001977],"study_design_scores_gemma":[0.0004247598,0.00006651264,0.0004468364,0.0000320902,0.000005121263,0.000002704876,0.000009121642,0.9671682,0.005167347,0.01637109,0.01017877,0.0001274488],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.003149491,0.00004811767,0.9905467,0.001593449,0.0006187175,0.0002960433,0.00000613473,0.0001556461,0.003585695],"genre_scores_gemma":[0.3454007,0.000003742711,0.6506708,0.001776831,0.0001380295,0.00001553384,0.00001767517,0.000008947723,0.001967684],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9659078,"threshold_uncertainty_score":0.3895089,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01182716123524061,"score_gpt":0.2985599656614644,"score_spread":0.2867328044262238,"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."}}