Why Cross-Modal Retrieval Works: Positive Samples Stay Close in Most Dimensions
gabriberton · x · 2026-08-19
Part of a thread on the "modality gap" in multimodal models: the author explains that along most dimensions, positive cross-modal pairs (e.g. text and its matching image) actually land close to each other — which is why cross-modal retrieval like text-to-image works so well. A previous note explains the hypersphere is high-dimensional (1000 dims), so the modality gap may exist in only one of them.
Related event: Modality Gap in VLMs: Flaw or Feature?(7 posts)→
More from Research
- LMR-BENCH at EMNLP 2025 tests whether LLM agents can reproduce real LLM research — ZhiyuChen4 · 2026-10-16
- Ex-OpenAI researcher claims most neurons are legible now; critics call it hype — anshulkundaje · 2026-10-16
- Physicist pushes back on ML community's acceptance that models may be beyond understanding — tszzl · 2026-10-16
- Survey: AI agent evaluation is getting realistic but still fails to measure safety and reliability — burkov · 2026-10-16
- COVA-SFT dataset trains models to solve text problems with visual chain-of-thought — yoavartzi · 2026-10-16
- USTC paper uses speculative sampling to cut latency in interactive video generation — ustc · 2026-10-16