AI Paper Atlas

Research map / Learning theory and algorithms / Matrix and tensor methods

SVD-Softmax: Fast Softmax Approximation on Large Vocabulary Neural Networks

Kyuhong Shim, Minjae Lee, Iksoo Choi, Yoonho Boo, Wonyong Sung

NeurIPS 2017 · Learning theory and algorithms › Matrix and tensor methods · approach: Structured low-rank learning · 24 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
widen preview window W from 16 to 32 (D=256) (ablation)SVD-softmax · WikiText-2 LM D=256NLL (SVD): 4.518 -> 4.441Improved
widen preview window W from 64 to 128 (D=1024) (ablation)SVD-softmax · WikiText-2 LM D=1024NLL (SVD): 3.789 -> 3.761Improved
increase full-view vectors N from 32768 to 65536 (800K vocab) (ablation)SVD-softmax · OBW 800K LMNLL (SVD): 4.718 -> 4.690Improved
increase full-view vectors N from 1024 to 2048 (8K vocab) (ablation)SVD-softmax · OBW 8K LMNLL (SVD): 2.698 -> 2.687Improved
reduce W and N to 50 and 1000 (ablation)SVD-softmax · De-En NMTperplexity: 10.57 -> 11.04Hurt
reduce W and N to 50 and 1000 (ablation)SVD-softmax · De-En NMTBLEU: 21.98 -> 22.00No clear effect
replace full-softmax with SVD-softmax (W=256, N=16K)softmax layer · GPU timing V=262k D=2ktime (ms): 14.12 -> 4.43Improved
replace full-softmax with SVD-softmax (W=256, N=16K)softmax layer · CPU timing V=262k D=2ktime (ms): 1541.43 -> 303.57Improved

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