Like this is sick, but it removed gemma weights
OPEN SOURCE
-
Google Releases Gemma 3: 27B Multimodal Open Source Model
By
–
Very excited to see the release of Gemma 3, the latest in our open source models. It is only 27B parameters, is multimodal, and has a delightful footprint that fits in a single H100 GPU, and runs really well on TPUs as well. https://t.co/SzndoMOQnX
— Jeff Dean (@JeffDean) 12 mars 2025Very excited to see the release of Gemma 3, the latest in our open source models. It is only 27B parameters, is multimodal, and has a delightful footprint that fits in a single H100 GPU, and runs really well on TPUs as well.
-
NotebookLM adds citation support
By
–
NotebookLM notes now support citations as well. Earlier citations were only available in chat responses.
-
Reka Open-Sources Flash 3: 21B Parameter Reasoning Model
By
–
7. Reka open-sourced Flash 3, a 21B parameter reasoning modelpic.twitter.com/bvkOL1YQ1I
— Linus ✦ Ekenstam (@LinusEkenstam) 12 mars 20257. Reka open-sourced Flash 3, a 21B parameter reasoning model
-
Gemma Models: Underestimated LLMs with New Technical Report Released
By
–
Great news! Gemma models are one of the most underestimated LLMs imho. The technical report is available here: https://
storage.googleapis.com/deepmind-media
/gemma/Gemma3Report.pdf
… -
Using Meta pretrained weights without training or image tokens
By
–
Hm, did you use the pretrained weights from Meta? Does the model work fine if you don't do any training and don't pass any image tokens?
-
Ego Toll: Credibility Loss Resistance in Improvement Steps
By
–
What is the "ego toll"? It's when people push back against even a small step toward improvement, if that step involves a perceived loss of credibility, authority, or control https://
bit.ly/3EWiSgr #opensource #productdev -
OlympicCoder Beats Claude 3.7 in Competitive Programming
By
–
Wow, we've kept pushing our Open-R1 project, an open initiative to replicate and extend the techniques behind DeepSeek-R1 And even we were mind-blown by the results we got with this latest model we're releasing: OlympicCoder It's beating Claude 3.7 on (competitive)
-
Llama.cpp Image-Compatible Playground Guide Released
By
–
BTW Son's image-compatible playground is super cool Guide on how to run it: https://
github.com/ggml-org/llama
.cpp/discussions/12348
… -
Gemma 3 27B now on AI Studio
By
–
Gemma 3 27B is also now available on AI Studio 👀 https://t.co/Ri94aTOfxE pic.twitter.com/39LqzE5LKq
— 🚨 AI News | TestingCatalog (@testingcatalog) 12 mars 2025Gemma 3 27B is also now available on AI Studio