PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
January 14, 20260 citationsOpen Access

Contrastive Geometric Transfer: Efficient Sentence Embeddings via Hyperbolic Projection with 24× Compression

View Full Paper
ÉRÉric Reis

Key Points

  • This research aims to evaluate the effectiveness of hyperbolic projection for compressing sentence embeddings.
  • Used Contrastive Geometric Transfer method for compression.
  • Evaluated performance on STS-B benchmark with MPNet and MiniLM models.
  • Measured quality retention using Spearman correlation coefficient.
  • Achieved 24× compression from 768d to 32d with 97.4% quality retention.
  • Outperformed PCA by 5.9% at 32d and 16.7% at 16d.
  • Identified semantic degradation threshold at 5 dimensions.

Abstract

Experimental results for Contrastive Geometric Transfer (CGT), a method for compressing sentence embeddings via hyperbolic projection. Achieves 24× compression (768d→32d) with 97.4% quality retention (Spearman ρ=0.812 vs baseline 0.834). Outperforms PCA by 5.9% at 32d and 16.7% at 16d. Identifies semantic degradation threshold at 5 dimensions. Evaluated on STS-B benchmark using MPNet and MiniLM teacher models. Includes LaTeX manuscript, JSON experimental results, and Jupyter notebooks for reproducibility

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Éric Reis (2026) studied this question.

synapsesocial.com/papers/6966f33b13bf7a6f02c0130dhttps://doi.org/10.5281/zenodo.18225189
Ask AI
Helpful
Bookmark
Share
View Full Paper