AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Representing Hyperbolic Space Accurately using Multi-Component Floats

Tao Yu, Christopher De Sa

NeurIPS 2021 · Architectures and efficiency › RNNs and normalization · approach: Dataset distillation and data-efficient training · 0 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
represent upper-half space coordinates with 2-component floats (2-xMC-Halfspace) instead of plain floatsHalfspace · Mammals 2DMAP: 92.07% -> 93.85%Improved
use Poincare upper-half space instead of Lorentz modelhyperbolic embedding · Mammals 2DMAP: 85.75% -> 92.07%Improved
2-component floats vs plain upper-half spaceHalfspace · WordNet Nouns 2DMAP (%): 23.31 -> 24.42Improved
2-component floats vs L-tiling2-xMC-Halfspace · WordNet Nouns 2DMAP (%): 23.39 -> 24.42Improved
2-component floats vs plain halfspaceHalfspace · WordNet Verbs 2DMAP (%): 63.65 -> 64.88Improved
2-component floats vs plain halfspaceHalfspace · Gr-QC 2DMAP (%): 55.60 -> 56.35Improved
increase embedding dimension from 2 to 5 and 102-xMC-Halfspace · WordNet NounsMAP (%): 24.42 -> 75.05Improved
2-component floats vs Lorentz at 10D2-xMC-Halfspace · WordNet Nouns 10DMAP (%): 78.43 -> 78.46No clear effect
2-component floats vs Lorentz at 10D2-xMC-Halfspace · WordNet Verbs 10DMAP (%): 90.71 -> 90.71No clear effect
use MCF on GPU instead of tiling models on CPU2-xMC-Halfspace · WordNet Nouns batch 128training time: saves 30.27 hours over 1000 epochsImproved
MCF (high precision) vs ordinary float halfspaceHalfspace · WordNet Nouns batch 128 2Dseconds per epoch: 17.4 -> 54Hurt
MCF with points near origin2-xMC-Halfspace · embeddings close to originMAPNo clear effect

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