AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Unbounded cache model for online language modeling with open vocabulary

Edouard Grave, Moustapha M Cisse, Armand Joulin

NeurIPS 2017 · Architectures and efficiency › Attention and transformers · approach: Language model pretraining and induction · 11 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 the unbounded cache to the static language modelStatic model trained on News 2007 · News 2007–2011perplexity: 220.9 -> 166.5; 237.6 -> 191.4; 256.2 -> 202.6; 259.7 -> 204.8; 268.8 -> 214.3Improved
Use a unigram cache instead of only the static modelStatic model trained on News 2007 · News 2007–2011perplexity: 220.9 -> 220.3; 237.6 -> 235.9; 256.2 -> 252.6; 259.7 -> 256.1; 268.8 -> 264.3No clear effect
Use a local cache instead of only the static modelStatic model trained on News 2007 · News 2007–2011perplexity: 220.9 -> 218.9; 237.6 -> 234.5; 256.2 -> 250.5; 259.7 -> 256.2; 268.8 -> 265.2No clear effect
Add the unbounded cache instead of the strongest cache baselineStatic model trained on News · News → Commentaryperplexity: 288.5 -> 191.1Improved
Add the unbounded cache instead of the strongest cache baselineStatic model trained on News · News → Webperplexity: 581.1 -> 383.4Improved
Add the unbounded cache instead of the strongest cache baselineStatic model trained on News · News → Wikiperplexity: 316.5 -> 337.4Hurt
Add the unbounded cache instead of the strongest cache baselineStatic model trained on News · News → Booksperplexity: 240.3 -> 237.2Improved
Add the unbounded cache instead of the strongest cache baselineStatic model trained on Web · Web → Newsperplexity: 519.2 -> 306.3Improved
Add the unbounded cache instead of the strongest cache baselineStatic model trained on Web · Web → Commentaryperplexity: 391.3 -> 234.9Improved
Add the unbounded cache instead of the strongest cache baselineStatic model trained on Web · Web → Wikiperplexity: 321.5 -> 340.2Hurt
Add the unbounded cache instead of the strongest cache baselineStatic model trained on Web · Web → Booksperplexity: 235.8 -> 223.6Improved
Add the unbounded cache instead of the strongest cache baselineStatic model trained on Wiki · Wiki → Newsperplexity: 532.8 -> 318.7Improved

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