Unverified50% confidenceFactExact time
EmbeddingGemma 2 outputs 768-dimensional vectors by default and supports truncating the output dimension to as low as 128 dimensions via Matryoshka Representation Learning.
1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEmbeddingGemma 2's 740M parameters are divided into a text component of 270M (backbone 130M, embedding layer 140M), a vision encoder of 170M, and an audio encoder of 300M, with all three encoders loadable on demand.65% similarUnverifiedPPT核心坐标系统使用EMU（English Metric Unit）作为度量单位，精确到1/914400英寸62% similarUnverifiedAfter quantization, EmbeddingGemma 2 requires only approximately 191 MB of active memory for text-only weights on the Google Pixel 11 Pro, while the full multimodal model requires approximately 567 MB.61% similarUnverifiedU-Net 编码器通常通过最大池化逐步将特征图分辨率减半同时通道数加倍（如 64→128→256→512）60% similarUnverified自回归解码在批大小为1时算术强度约等于隐藏层维度（通常4096至16384），而现代GPU峰值算术强度往往超过200，导致解码处于内存带宽受限状态60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/985570API
curl https://kongchang.com/api/v1/knowledge/claims/985570MCP
get_claim(id=985570)