DeepMind’s New AIs: The Future is Here!

Guide for using DeepSeek on Lambda:

📝 The Gemma 3 paper and the rest are available here:

Sources:

📝 My paper on simulations that look almost like reality is available for free here:

Or this is the orig. Nature Physics link with clickable citations:

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:

[kylek]

I suspect one reason they're trying to reduce the footprint of their model is for the purpose of putting it on the Pixel devices, which hold smaller versions of models for specific tasks.

[WinonaNagy]

Gemma 3 optimizing with just one GPU? Now that's impressive! From creative writing to high-dex robotics, this is a game-changer. Hats off to Google DeepMind!

[TheAkdzyn]

This is very impressive. It feels like another leap with the coherent text in images, image editing on par with photoshop all in an incredible form factor. Sounds like a creative box of knowledge.

[bluehorizon]

Yess!! Local and Open models for the Win!

[June-]

With how fast things are moving, We will either be the last generation or the eternal generation. What a time be alive..

[JorgetePanete]

I remember watching here some years ago a technique for recoloring a photo of Abraham Lincoln, but the hair was slightly out of pose, it was revealed to be a GAN at work.
I can't believe the progress we've seen until today and what's left to come with video generation.

[mighty]

But can it fill my wine glass to the brim ?

[sirrobinofloxley]

That's definitely working almost perfectly, just a few little tweaks needed and I made a fantastic thing, which Gemma 3 helped a lot to fix. I just made the relevant fixes, put them back in and reasoned with the model about the aims of the work. One thing though, I did have to change the settings, in aistudio, and it was a after the full gamut of Gemini models before it shows u, but it gives the 2T option FOR FREE!!! Thanks for sharing!

[torarinvik]

I am fine-tuning gemma3 1b as we speak 😎

[colef]

The amount of competition in the AI space is crazy. I can see things accelerating even faster with how competitive this market is currently

[patrickdegenaar]

Holy moly... I uploaded a Glastonbury photo, and it recognised the pyramid stage!!! I'm getting 8 tokens/s on the 12B param version and 1 token/s on the full 27B, using my 5-year-old computer with a (8Gb VRAM) 3060TI. That is really impressive!

[fusn]

To add image generation and editing is only available in gemini 2.0 flash experimental (not the normal gemini 2 flash) its not available in gemma, and image generation is native from the model not an external model.

[xeleader]

"This is not amazing, this is beyond amazing!"
Another Two Minute Papers mood

[mrrfyW]

I LOVE that new AI image editing thing with Google’s AI Studio. I even made a video (shared somewhere else) using it! What a wonderful time to be alive!

[rando]

Generative edits are one thing, but I want to see the ability to create consistent characters and styles. Not similar, the same.

[npc-drew]

What you mean Gemma 2 was okay, it was literally one the best in the small llm models, the mode concise.

[Anders]

The iterative image editing with Google Gemini (experimental) I find interesting that the previous image remain unchanged and only some parts changes.

[Victorious]

Just subbed. Top-notch, no-nonsense approach. Love it!

[Sap-bk]

Tried to replicate the table with flowers but I wasn't able to.

I have ollama and Open WebUI, I updated to the latest versions of both, when I ask the table question I don't get an image back. I also asked the other image questions and used a screenshot from this video to have the exact same images and questions, I was never able to get an image back, instead I get something like this:


Here's the image with flowers added to the table:
<start_of_image>

I added a simple vase with a few stems of white flowers to the center of the table. I hope you like it!

Despite the answer having the "<start_of_image>", it's just that, just the text, the image is not there.

Has anyone else tried it?
Were you able to get this to work locally?

[RD-wmfo]

I like running these models locally on my computer.