@valsai | prev gdm, stanford
San Francisco, CA
Joined December 2017
- Tweets331
- Following324
- Followers288
- Likes1.8K
Pinned Tweet
EmbeddingGemma 2 is our first open, natively multimodal embedding model.
With <1B params, the model is built to run on everyday hardware like your phone or laptop.
We’re super excited to see what people build!
More: youtu.be/anPsS6huQk0?is=zfAL…
Sebastian Russo retweeted
EmbeddingGemma 2: Unified and Efficient Multimodal Representations
EmbeddingGemma 2 natively unifies text, image, audio, and video into one embedding space, outperforming specialist baselines and models two to twelve times larger on tested benchmarks.
arxivb.org/abs/2610.37364
Sebastian Russo retweeted
📖 The @Google EmbeddingGemma2 team has published their technical report!
EmbeddingGemma2 was just 740M parameters, but its omni-modal nature and pareto-optimal performance makes it one of the most exciting releases of the last months for me.
huggingface.co/papers/2610.3…
Sebastian Russo retweeted
This is awesome. Just downloaded it and tried a few queries, solid so far
What if Spotlight actually understood the meaning inside your files?
DigUp is an open-source macOS menu bar app powered by EmbeddingGemma 2 that brings 100% local, multimodal semantic search to your Mac.
Instead of searching for file names, you can use natural language to:
📸 Find specific scenes in your photos
⏳ Jump to the exact timestamp in audio or video
💻 Search through your codebase by intent
Everything runs locally in a single vector space.
Sebastian Russo retweeted
Excited to launch Gemini 4 Argon, designed to help with hardest coding and knowledge work tasks with its frontier capabilities..
For developers and enterprises, it’s available in the paid tier of the Gemini API and the Gemini platform, for Ultra subscribers in Google Antigravity, and Ultra and Pro subscribers in the Gemini app and Search.
Read more: blog.google/innovation-and-a…
Sebastian Russo retweeted
📈@Google's EmbeddingGemma2 is looking to be a huge success!
After only a week, it has surpasses EG1's likes to become the 4th most liked Google model ever, as well as the 3rd most liked Sentence Transformers model ever.
Congratulations to the team for their work!
Links in 🧵
Sebastian Russo retweeted
Did you download Google Foresight yet?
(developers.google.com/edge/f… )
One of my fav demos yet cc @sahildua2305
1
1
174
Sebastian Russo retweeted
After 20M+ downloads of the original EmbeddingGemma, we just released EmbeddingGemma 2 (Apache 2.0)!
Built on the Gemma 4 architecture, it natively maps text, code, images, audio, and video into a unified 768D space—with modular encoders small enough to run fast, private, offline RAG and semantic search right on device.
Blog: blog.google/innovation-and-a…
Dev guide: developers.googleblog.com/en…
Introducing EmbeddingGemma 2! 🚀
Our lightweight, multimodal embedding model maps text, code, images, video, and audio into a single, unified embedding space. Optimized for on-device use cases, it features:
- 740M parameter form factor with modular encoders
- Flexible dimension sizes (768dim-128dim) via Matryoshka Representation Learning (MRL)
- 8K context window (4x larger than text-only EmbeddingGemma)
- A commercially permissive Apache 2.0 license
Sebastian Russo retweeted
EmbeddingGemma made it possible for so many AI for Science projects to get started:
🩺 Embedding Gemma 300m Medical
🌍 Geo-Gemma
🎗️ Oncology-Gemma
📚 EmbeddingGemma PubMed
EmbeddingGemma 2, which released and made available on Hugging Face yesterday, opens the door to so many new projects in science! 🚀 🧬 🌎
huggingface.co/google/embedd…
Check out this Jev-inspired demo, powered by EmbeddingGemma 2!!!
Kudus to @surils and his team for driving this
3/ (On-device Jev) MediaPipe Decision API (developers.google.com/edge/m…)
EG2 is a surprisingly good decision engine! With the MP Decision API, you can define strongly typed primitives (Choice, Boolean, Score) and run zero-shot classification and routing in the order of milliseconds!
3
214
Sebastian Russo retweeted
Imagine finding a photo of your cat by searching with their "Meow" 🙀
EmbeddingGemma 2 is a (very) small (740M), multimodal (text, code, images, video and audio) model you can use to search... pretty much anything. A photo, a sound, a line of code.
Btw: runs on phones / IoT 📱
Introducing EmbeddingGemma 2! 🚀
Our lightweight, multimodal embedding model maps text, code, images, video, and audio into a single, unified embedding space. Optimized for on-device use cases, it features:
- 740M parameter form factor with modular encoders
- Flexible dimension sizes (768dim-128dim) via Matryoshka Representation Learning (MRL)
- 8K context window (4x larger than text-only EmbeddingGemma)
- A commercially permissive Apache 2.0 license
We really took "pocket-sized" to heart .. thanks to @tomaarsen and the ST team for supporting :)
I'm so glad that Google Deepmind chose an accessible model size for their excellent multimodal EmbeddingGemma 2! Plus some clever work to make audio and video faster to process.
All nicely integrated in Sentence Transformers as well 👏
3
156
Wow! 77x less RAM for a ~5% drop in quality. Cool results @KShivendu_
We pushed 10M of Google's EmbeddingGemma 2 vectors from ~30GB of RAM down to ~0.4GB
float32, 768 dims: ~30GB
1-bit TurboQuant: ~1GB (99% nDCG@10 kept)
+ MRL to 256 dims: ~0.4GB (94.5% kept with rescoring)
77x less RAM for a ~5% drop in quality! TurboQuant + MRL is magic 🪄
2
114
Sebastian Russo retweeted
its preety strong overall! cool release and opensoured one :)
Sebastian Russo retweeted
EmbeddingGemma 2 is out! Our first native multimodal embeddings built on Gemma 4 under Apache 2.0. 🤗 Embeds 100+ languages, code, images, audio, and video into one vector.
🧮 8,192 context
3️⃣ 740M omni, 440M text+vision, 570M text+audio, or 270M text-only
📈 +14% jump on MTEB Code
🪆 Matryoshka: 768d, 512d, 256d, or 128d
🤗 Available in Sentence Transformers and LiteRT-LM
🔓 Apache 2.0
huggingface.co/google/embedd…