Victor Geislinger (VictorsOtherVector)

Best to reach me not on Twitter (see website link)

Combination of machine learning engineer / data scientist and teacher. Focusing on AI/ML @Google

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

ML/AI community on Threads seems to already have a lot of life! I think it's going to be a long-term place to be for ML/AI

Here's me: https://t.co/voTrA8l1DG
Don't feel obligated to follow but you should follow all the bird ML folks who are now there like Jeremy 😄

[Jeremy Howard @jeremyphoward]

Seems like there's already more interesting ML/AI stuff in the new place than here - you can find me as: https://t.co/JfoqyGKOW8

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

@chrisalbon New idea: self-driving bikes

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

@chrisalbon Ah it's invite only right now?
It looks interesting though (is it exclusively 'toilet chats'?)

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

And my 'new' used books came in the mail today!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Today's haul from the library

(Shout out to https://t.co/1iXikOBDEu for recommending almost half of these books)

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Thanks Gemini for helping put my twins to bed!

We had a lot of energy & asked if "my phone can read stories". We did this before where they explain what they want in the story

The stories are simple but perfect for a pair of 3yr-olds
Here's one chat: https://t.co/zJT9SLqXna

[meg.ai 🇨🇦 @MeganRisdal]

Proud of the thoughtful, important clarity the team provides re: Gemma as an open model relative to definitions of open source that are often misused. Highly recommend checking out this post!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Been meaning to share this since I haven't seen anyone mention it, but seems more relevant now w/ the discussion on Gemini image gen

You can also check the conversation page directly here: https://t.co/fBGK9N6bXi

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Apparently you can hear what appears to be prompts for Gemini image generation, at least on Android

Wasn't sure the best way to share an example (upload kept saying 'media was invalid', so just linking the Threads post I made w/ the videos: https://t.co/gFcTC7gwYc

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Can't wait to see what the community does, and how other companies are going to respond to this! It's a little over 50 days into 2024 so a lot can change before the end of the year!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Been super excited for the Gemma announcement! Google releasing an open model based on Gemini with a generally open license is pretty awesome for the AI/ML field!

Announcement: https://t.co/Aj9NmzgUxJ
Model: https://t.co/dDn6UtOMUN
Tech report: https://t.co/ShBO28bql6

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

@andrew_n_carr Ooo that could be fun :D

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

@yacineMTB Wow! I never knew they had merch!!
https://t.co/wMUqsqXv1k

Thank you for pointing this out! Like, definitely donate too but also arxiv merch just tickles me in that particular nerdy way

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

@dward4 @vboykis @normconf Man do I miss @normconf...

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

@JackK Just gotta say that I think you & the team are doing an awesome job with the feedback!

Honestly, I'm pleasantly surprised how many people I've seen say they generally prefer Gemini over other AIs, but I know it also comes from all the improvements that have been over the week+!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Finally bit the bullet and got Manning's online subscription https://t.co/9DCcmH8TKu

Who am I kidding– I totally wanted to do this! Access to the full library and 50% off all future purchases? About to get so many more physical books!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

This interaction is just the best!

PyTorch founder at Meta (@soumithchintala) with a legit use case asking to test out Google's Gemini 1M context model

Just warms the AI heart ♥️♥️🤖

[Jeff Dean @JeffDean]

@soumithchintala Working on getting you access. Happy to hear any feedback you might have!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

It's been awesome being on a team involved with so much of the AI Google is developing, even if I joined only about a month ago!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Super excited for the Gemini 1.5 Pro model! Excited to see how multimodal models w/ a huge context are used!

I'm biased of course, but in my opinion Google's really flexing its AI 😄 And we're not even 50 days into 2024!

[Jeff Dean @JeffDean]

Gemini 1.5 Pro - A highly capable multimodal model with a 10M token context length

Today we are releasing the first demonstrations of the capabilities of the Gemini 1.5 series, with the Gemini 1.5 Pro model. One of the key differentiators of this model is its incredibly long context capabilities, supporting millions of tokens of multimodal input. The multimodal capabilities of the model means you can interact in sophisticated ways with entire books, very long document collections, codebases of hundreds of thousands of lines across hundreds of files, full movies, entire podcast series, and more.

Gemini 1.5 was built by an amazing team of people from @GoogleDeepMind, @GoogleResearch, and elsewhere at @Google. @OriolVinyals (my co-technical lead for the project) and I are incredibly proud of the whole team, and we’re so excited to be sharing this work and what long context and in-context learning can mean for you today!

There’s lots of material about this, some of which are linked to below.

Main blog post:
https://t.co/QAsDKXBdao

Technical report:
“Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context”
https://t.co/CTzTHNDCdo

Videos of interactions with the model that highlight its long context abilities:
Understanding the three.js codebase: https://t.co/yq7d6OSD6c
Analyzing a 45 minute Buster Keaton movie: https://t.co/adyMgDYHoK
Apollo 11 transcript interaction: https://t.co/Pqvq3Eac1R

Starting today, we’re offering a limited preview of 1.5 Pro to developers and enterprise customers via AI Studio and Vertex AI. Read more about this on these blogs:
Google for Developers blog:
https://t.co/x73Vun0kVS
Google Cloud blog:
https://t.co/OlaTW6PYGn

We’ll also introduce 1.5 Pro with a standard 128,000 token context window when the model is ready for a wider release. Coming soon, we plan to introduce pricing tiers that start at the standard 128,000 context window and scale up to 1 million tokens, as we improve the model.

Early testers can try the 1 million token context window at no cost during the testing period. We’re excited to see what developer’s creativity unlocks with a very long context window.

Let me walk you through the capabilities of the model and what I’m excited about!

[Victor Geislinger (VictorsOtherVector) @MrGeislinger]

Best thing I've done was show my 3yr old kids how to make the computer 'talk'

They've learned to type 'say' in the terminal and then whatever they want :D

Usually ask me how to spell a word and then they find & type the letters. They also do their names & gibberish on their own