Synaptics/Google-EmbeddingGemma-2

EmbeddingGemma-2 (4-bit weights)

Model Overview

EmbeddingGemma-2 is a multimodal embedding model from Google. It maps text, images, video and audio into one 768-dimensional embedding space (Matryoshka: 768, 512, 256 or 128 dimensions), so a text query can be compared directly with photos, video frames and sound clips.

This build has been optimized for the Synaptics Astra™ SL2610-Series processors with Torq NPU: 4-bit block-quantized weights (block 32, from the onnx-community *_q4 export) with bf16 activations. The bf16 build (closest to the original model, more memory) is on the bf16 branch.

Model Features

• Model Type: Multimodal embedding model
• Input: Text, image, video, audio
• Output: 768-d embedding

Files

File	Runs on	Purpose
text_body_s128.vmfb, text_body_s512.vmfb	NPU	Text body for 128 / 512 tokens (queries, images; documents, video, audio)
vision_384x384.vmfb	NPU	Vision encoder: a 384x384 image or video frame -> 64 soft tokens
audio_140.vmfb … audio_1120.vmfb	NPU	Audio encoder for 1.4 / 2.8 / 5.6 / 11.2 s windows
token_embeddings.npy	CPU	Token-embedding table, memory-mapped
mel_filters.npy, mel_window.npy	CPU	Audio log-mel front end
tokenizer.json, config.json, embeddinggemma2_manifest.json	CPU	Tokenizer, model config, static shapes and NPU memory per network
onnx/	—	Source ONNX graphs (fp32 and 4-bit) that torq-tools builds these files from: a copy of onnx-community/embeddinggemma-2-ONNX (on main)

Performance (SL2619, one network per process)

Network	Latency	RAM	File
text S=128	549 ms	122 MB	87 MB
text S=512	2825 ms	140 MB	103 MB
vision 384x384	3310 ms	157 MB	122 MB
audio 1.4 s	279 ms	210 MB	180 MB
audio 2.8 s	600 ms	211 MB	181 MB
audio 5.6 s	1337 ms	217 MB	186 MB
audio 11.2 s	2670 ms	227 MB	196 MB

ONNX Runtime with the same 4-bit weights on the board CPU (2 threads) is 5.9–11.3x slower (text S=128 3.7 s, vision 21.6 s, audio 11.2 s 23.3 s).

Deployment

Usage and demos (text/image/video/audio similarity, and search over a media library) are in torq-examples under Google/Google-EmbeddingGemma-2:

python setup_demos.py Google-EmbeddingGemma-2

Learn More

• Synaptics AI Developer Zone: Get started with documentation, tutorials and resources for your Edge AI journey.
• Astra Support Portal: Connect with our engineering team and community.