AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Normalization Helps Training of Quantized LSTM

Lu Hou, Jinhua Zhu, James Kwok, Fei Gao, Tao Qin, Tie-Yan Liu

NeurIPS 2019 · Architectures and efficiency › Quantization · approach: Post-training weight quantization · 29 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add layer norm (ablation)BinaryConnect LSTM · Penn Treebank char-leveltest BPC: 2.51 -> 1.49Improved
add layer norm (ablation)TerConnect LSTM · Penn Treebank char-leveltest BPC: 5.84 -> 1.43Improved
add layer norm (ablation)BinaryConnect LSTM · War and Peace char-leveltest BPC: 4.24 -> 1.69Improved
add weight norm (ablation)BinaryConnect LSTM · Penn Treebank word-level d=300test perplexity: 8247.4 -> 87.6Improved
add weight norm (ablation)TerConnect LSTM · Penn Treebank word-level d=300test perplexity: 113.8 -> 86.5Improved
add layer norm (ablation)BinaryConnect LSTM · Penn Treebank word-level d=650test perplexity: 1244.2 -> 82.3Improved
add layer norm (ablation)BinaryConnect LSTM · pMNISTtest accuracy (%): N/A -> 91.2Improved
add layer norm (ablation)LAT 2-bit LSTM · Penn Treebank char-leveltest BPC: 1.48 -> 1.40Improved
use shared instead of separate batch norm statistics (ablation)BinaryConnect LSTM · Penn Treebank char-leveltest BPC: 1.50 -> 1.51No clear effect
use shared instead of separate batch norm statistics (ablation)BinaryConnect LSTM · pMNISTtest accuracy (%): 91.2 -> 36.5Hurt
add normalization to full-precision LSTM (layer norm)Full-precision LSTM · Penn Treebank char-leveltest BPC: 1.45 -> 1.43Improved
apply normalization directly on quantization instead of SBN stochastic binarization with batch normBinaryConnect LSTM · Penn Treebank char-leveltest BPC: 1.60 -> 1.49Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy