W. Fitzgerald

[Google Labs @GoogleLabs]

We're releasing a few more #NotebookLM features today – we've been so inspired by all the ways this community has been using the tool, and we're excited to see where you take it next ✨

[Google AI @GoogleAI]

Development of therapeutic drugs is often difficult and time consuming. A new model, Tx-LLM, is able to predict the properties of many entities of potential interest for therapeutic development with accuracy comparable state-of-the-art specialty models.→ https://t.co/58QE81P5gx

[Google Cloud @googlecloud]

We're building on our partnership with @Ginkgo with a launch that's one of the first-of-its-kind in the industry—a protein LLM designed to help enterprises accelerate drug development coming to Vertex AI Model Garden soon! Learn more → https://t.co/CJDDK6iZ9f

[Brad Little @bradlittletx]

Very proud of the role #googlecloudconsulting played in working with @Ginkgo to build and train this custom LLM. The potential for this technology cannot be overstated. Thank you Ginkgo for the partnership on this program with @googlecloud .

[Thomas Kurian @ThomasOrTK]

We’re excited to partner with @Ginkgo to bring their protein LLM and open API to both enterprises and individual scientists, which will help drive innovation in life sciences. https://t.co/m6jEjD5bIm

[Jason Kelly @jrkelly]

Excited to launch our newest offering -- Ginkgo Model API and the first publicly available model trained on Ginkgo's in-house data assets, AA-0! 🧬🎉

Over the past 10 years @Ginkgo has built a ton of amazing technologies from automation to long DNA construction to complex -omics to proprietary data and AI assets.

These tools had been kept in-house and only accessible to Ginkgo scientists -- in the remainder of this year you will see us announce the opening of these tools -- recently opening our automation to be placed in your lab and now opening up our AI tools to you over the cloud. AA-0 was trained on the 2 billion+ protein sequences in Ginkgo's codebase which have low overlap with the sequences available in public data sets.

So today if you are a developer or bioinformatician and you want to build on top of AA0-650M or ESM2-650M give it a try here!

https://t.co/WrHtamgoPR

We’ve structured our costs to make it easy for you to jump in, experiment, and get predictions without worrying about high fees or royalties. Our initial models will have a free tier, and our introductory pricing will be approximately $0.18 per million tokens. This means for a protein with around 500 amino acids, users should be able to get predictions on 2,000 sequences for roughly 20 cents.

Our first protein language model release supports two use-cases initially:
1. Generation via Masked Language Modeling: given a sequence of amino acids with one or more <mask> tokens, the model will complete the sequence.
2. Embedding calculation: Calculate the final hidden layer of the trained model to extract valuable representations for downstream tasks. To begin, our model returns the mean-pooled representation across the length axis.

This is just the start -- expect more open-source as well as models trained on proprietary Ginkgo data to be added to our API regularly by @agupta and Ginkgo's AI team. And let us know what else would be helpful to you -- both new models or biological data sets that you'd like to see available for training -- we are building these tools for you, help us make them better!

I'd like Ginkgo's platform to be easier (and cheaper) than downloading models and running on your own cloud instance. We're lucky to be partnered with @googlecloud to make this fast and easy for the biotech community! 🧬🤝💻

Ultra large lab data assets coupled to LLMs is the most exciting new tool I've seen in the last 20 years of bioengineering -- can't wait to see what you all grow with it! 🌱🌱

[Ginkgo Bioworks @Ginkgo]

Today, we’re thrilled to announce two exciting launches, powered by our partnership with @googlecloud: our first protein LLM, AA-0, and our new model API, a powerful tool designed to bring biological AI models directly to machine learning scientists.

https://t.co/Ku9Cx1gh0M

[W. Fitzgerald @wfitzgerald22]

@LifeSciVC Great idea. I am in Boston but out of town this time!

[Acyn @Acyn]

Trump: And they are keeping me, in a courtroom that’s freezing all day long. Biden is out there campaigning, that’s probably an advantage because he can't campaign nobody knows what he is doing he can't put two sentences together.

[W. Fitzgerald @wfitzgerald22]

@kfjonespallnp @JessicaMerlinMD @LisaWoodPhDRN @AmyEDelaneyPNP @andOneFineDay @KathleenBroglio @BriannaMorganNP @ErinKGeorge @Nickie_Nickie @hwbulls @ZacharySagerMD @jjhomd @rbernack @releiter @devon_check @SteveWechslerPT Awesome! So proud.

[W. Fitzgerald @wfitzgerald22]

@kfjonespallnp @BC_CSON @HPNAinfo Congratulations Katie!!!

[W. Fitzgerald @wfitzgerald22]

@BChighathletics @HokiesFB @CoachPryVT @BrianDohn247 @RivalsFriedman Awesome!

[Eric Feigl-Ding @DrEricDing]

💡BEST. VIDEO. ALL. YEAR. Please share with friends how the mRNA vaccine works to fight the coronavirus.

📌NOTA BENE—The mRNA never interacts with your DNA 🧬. #vaccinate

(Special thanks to the Vaccine Makers Project @vaccinemakers of @ChildrensPhila). #COVID19

[W. Fitzgerald @wfitzgerald22]

@ButchStearns thanks for stopping on your way to booth last night to capture this moment between me and my son! #GoSox

[W. Fitzgerald @wfitzgerald22]

@TonyMassarotti 100%

[W. Fitzgerald @wfitzgerald22]

@BCHighBaseball Giddy Up!

[W. Fitzgerald @wfitzgerald22]

@mindykaling @BakingSteel the absolute best!

https://t.co/NwoiqBttq2

[W. Fitzgerald @wfitzgerald22]

@FareedZakaria I have read Post American world and been a fan of yours for years. I would like to read and further understand the Israeli-Palestine conflict. Can you suggest a book or two?

[W. Fitzgerald @wfitzgerald22]

@cmclymer This reporter is aces. Well done

[W. Fitzgerald @wfitzgerald22]

@HackswithHaggs Haha this is awesome.