{"id":"W3043619075","doi":"10.1109/isca45697.2020.00092","title":"Echo: Compiler-based GPU Memory Footprint Reduction for LSTM RNN Training","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Memory footprint; Recurrent neural network; Bottleneck; Compiler; Parallel computing; Deep learning; Artificial intelligence; Artificial neural network; Embedded system; Programming 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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0002154713,0.0003893379,0.0004520515,0.0001236984,0.0002362957,0.0001910417,0.001805527,0.0002294263,0.0000131429],"category_scores_gemma":[0.00005200084,0.0004077256,0.0002855645,0.0003672713,0.00007792511,0.0001197408,0.001059829,0.0007003425,0.00003045763],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000158905,"about_ca_system_score_gemma":0.000301047,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009451347,"about_ca_topic_score_gemma":0.00000474573,"domain_scores_codex":[0.9972289,0.00006523039,0.0005348075,0.001426405,0.0003061028,0.0004385767],"domain_scores_gemma":[0.9975775,0.0002453871,0.0003570374,0.00142665,0.0001688723,0.0002245727],"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":[0.00006403694,0.0001969911,0.000005150467,0.000333341,0.0001034527,0.000007470189,0.001623379,0.3145252,0.01571171,0.1949074,0.008254612,0.4642673],"study_design_scores_gemma":[0.0006288076,0.0001245068,0.00007108415,0.0001011804,0.00003384319,0.00001328011,0.0001136434,0.8137529,0.02107988,0.1414627,0.02178639,0.0008317305],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0005578814,0.00005069284,0.9789234,0.01237035,0.001310705,0.001778626,0.00001734074,0.001294711,0.003696311],"genre_scores_gemma":[0.3561634,0.00001041194,0.6412086,0.0006387965,0.0006383287,0.001003093,0.00007707143,0.00004086899,0.0002193198],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.4992277,"threshold_uncertainty_score":0.9998375,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1184545215385083,"score_gpt":0.322659852216518,"score_spread":0.2042053306780097,"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."}}