[KongchangAI]
Unverified50% confidenceFactExact time

When using MRL, 128-dimensional vectors can reduce vector database storage by approximately 6 times compared to 768-dimensional vectors.

1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen

Sources

Google EmbeddingGemma 2: A Lightweight, Open Multimodal Embedding Model

Related Entities

Matryoshka Representation Learningconcept

Related Claims

Unverified经过SVDQuant的4-bit量化后，FLUX模型显存需求可从约24GB降至约6GB67% similarVerifiedFP8 quantization can compress the weights of a 7B model from approximately 14GB (FP16) to approximately 7GB.65% similarVerifiedPreviously, mainstream models typically had context lengths ranging from 8K to 128K.64% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/985573
API
curl https://kongchang.com/api/v1/knowledge/claims/985573
MCP
get_claim(id=985573)