Jack Krawczyk

We’re bringing more to Google One AI Premium by adding Gemini in Gmail, Docs, Slides & more — so you can get more done without tab hopping. From drafting a note in Gmail to writing an itinerary in Docs, Gemini can help. Coming to 150 countries in English. goo.gle/3PepQ2X

Excited for Gemini to be available to Workspace customers! For people who use a Workspace domain, you can enable access to Ultra 1.0 for your accounts of all sizes. Looking forward to your feedback!

We are aware that Gemini is offering inaccuracies in some historical image generation depictions, and we are working to fix this immediately. As part of our AI principles ai.google/responsibility…, we design our image generation capabilities to reflect our global user base, and we take representation and bias seriously. We will continue to do this for open ended prompts (images of a person walking a dog are universal!) Historical contexts have more nuance to them and we will further tune to accommodate that. This is part of the alignment process - iteration on feedback. Thank you and keep it coming!

Wow. Google just released Gemma, the most powerful open LLM yet. Open for commercial use, it outperforms Mistral AI 7B and LLaMa 2 on Human Eval and MMLU. It's the first open LLM based on Gemini. Details: - Comes in two flavors: 2B and 7B. - Beats Mistral 7B, DeciLM 7B and Qwen1.5 7B - Instruction models in 2B and 7B variants. - 8192 Default context window. - MMLU score of 64.56, average leaderboard score 63.75 for 7B. -2B model compatible with mobile phones. Available on HuggingFace, Kaggle and Vertex AI.

We have a long history of supporting responsible open source & science, which can drive rapid research progress, so we’re proud to release Gemma: a set of lightweight open models, best-in-class for their size, inspired by the same tech used for Gemini blog.google/technology/dev…

@triswarkentin congrats on the huge launch!

Our newest launch -- check out my post on Google's main blog for the launch of Gemma! Excited to see what you all build!

👋 open models

@mckaywrigley credit goes to the @GoogleDeepMind team for pushing the research forward!

@JackK Amazing work!

I fed an *entire* biology textbook into Gemini 1.5 Pro. 491,002 tokens. I asked it 3 extremely specific questions, and it got each answer 100% correct. 1M token context windows are a gamechanger.

@devahaz @nikbonaddio and salads.

@JackK @nikbonaddio Weirdos y’all prob put fries in your sandwiches too

Mount Rushmore of apps: mozzarella sticks, crab rangoon, jalapeño poppers, and scotch eggs

@nikbonaddio @devahaz it’s definitely the fanciest app. bar none.

@devahaz man you’ve only ever had it in spears go to pittsburgh, get it sliced lengthwise on a mandolin and fried fresh with romano cheese on top @JackK let this man know

@adityaag 🫶

Gemini Advanced (nee Bard nee Ultra nee why doesn't Google get branding people) is actually pretty damn good. I like the little UI innovations like checking veracity etc.

@ai_andrew hmm can you confirm the workspace extension is turned on at gemini.google.com/extensions? /cc @asadovsky @AarushSelvan

@JackK It’s awesome. The Google Drive integration doesn’t quite seem to be working at the moment, regardless of the doc, the link, permissions. It hallucinates other files in the workspace. The code gen and test is awesome though.

As we continue to improve the coding experience with Gemini Advanced, one thing we’ve heard is many people want to try to make a few mods + run it first to ensure it works exactly as they want before merging into their workflow. Give it a shot and let us know what you think!

@itsPaulAi @dr_cintas appreciate it! adding our product lead for it @chrisgorgo

@dr_cintas @JackK I've been testing this for an hour. Even for a novice it's great! Maybe managing inputs and a few more libraries would be even better but honestly it's not a big deal.

Gemini has just received an update You can now execute and edit code in addition to generating it. Here's where to find these features and how to use them:

@dr_cintas @itsPaulAi looking forward to the feedback!

@itsPaulAi @JackK I need to try this!

@enzo_zuo follow up turn should address /cc @chrisgorgo

@JackK How can I give immediate feedback to the model so that it can quickly correct itself if the generated code runs with an error

@ryancarson credit belongs to @chrisgorgo and team!

Ok this is freaking awesome. You can run the code from Gemini right in the browser. Small thing but will make a huge impact in dev workflow. @JackK and team 👏🏻

@loganthorneloe yes! you can export any python responses to colab

@JackK This is really cool. Thanks for sharing! Does this connect to colab notebooks?