Intern applications at
@DeepMind
for the 2023 calendar year open this Friday, September 16th! 📣
Feel free to contact me or others for info, be ready to apply, and please share with your network!
To make my research more reproducible, extensible and comparable to that of others & out of need to homogenize the language we use to express nn pruning methods, I contributed `nn.utils.prune` to
#PyTorch
1.4 (see highlights )
Try it out & build on it! 🔥
v1.4: customizable mobile builds, Distributed Model Parallelism via experimental RPC API, Java Bindings, Chaining LRSchedulers
Summary:
Release Notes:
Last release for Python 2 (bye bye!)
If haven’t replied to you, it’s not because I hate you or because I’m ignoring you. Life’s complicated at the moment and I’m struggling. Feels suffocating. Please be patient ❤️
I’m both thrilled and honored that the position paper “The Scientific Method in the Science of ML” Jessica Forde and I wrote was selected for an oral at the
#DebugML
workshop at
#ICLR2019
& I’m even more excited that Jessica will be joining me at
@facebookai
this summer!
My poster at
#WiML2017
! If you didn’t manage to attend our poster session today, get in touch to ask me about my work on
#GANs
for science simulation!
#NIPS2017
Originally motivated by reproducibility needs in the context of NN pruning research, where multi-stage experiment pipelines are common, I'm happy to finally open source *dagger*, the framework I've built for reproducible & reusable experiment orchestration
If you're in NOLA for
#ICLR2019
, come chat with us and check out our poster at the
#DebugML
workshop () tomorrow! Room R03. I look forward to conversations on science, ML, science of ML, and ML for science 👩🏼🔬👩🏻💻
@DeepMind
Role descriptions, eligibility criteria, preferred qualifications, and other information will be published on Friday, so stay tuned!
I'll make sure to drop the link here.
Due to the high volume of submissions, we are still recruiting *reviewers* for the
#NeurIPS
workshop on ML + Physical Sciences. If you or any of your students and colleagues are able to review a couple of 4-page short papers, PLEASE fill out: 🙏
Happy Monday! Don’t forget to work on your submissions for the
#NeurIPS2019
workshop on Machine Learning and the Physical Sciences.
Deadline: Sept 16, 2019
#ML4PS2019
Bob is training models with
#PyTorch
. Bob forgets to add volatile=True to the input Variables in eval mode. Bob’s laptop runs out of disk space and crashes for a week straight. Don’t be like Bob.
Today also marks the end of my experience at
@facebookai
. I’m thankful for this important stepping stone in my career. A special thanks goes out to my manager
@deviparikh
, to
@an_open_mind
, and to those who gave me this opportunity years ago. Cheers to a better 2021! 🥂
Upset at the latest
#NeurIPS2020
reviews? Interested in an alternative publication model for machine learning research? Looking for hypothesis-based science in AI? 👩🔬
Check out and submit to the workshop at NeurIPS.
#SCIENCE
!
Deadline: Oct 7. More info 👇
Calling all women in
@datascience
and
#ML
/
#AI
in
#NYC
!
If you're new to
@PyTorch
and would like to learn more about it, become proficient, and potentially contribute to it, sign up for the intro workshop I'll be teaching Nov 19 with
@wimlds
@WiMLDS_NYC
!
I’m very excited to share our work on Gemini today! Gemini is a family of multimodal models that demonstrate really strong capabilities across the image, audio, video, and text domains. Our most-capable model, Gemini Ultra, advances the state of the art in 30 of 32 benchmarks,
Don't forget to submit your papers at the intersection of
#Physics
and
#MachineLearning
to the
#NeuriPS2020
workshop on Machine Learning and the Physical Sciences by this *Friday Oct 2* ‼️
We’re introducing a new offering called Gemini Business, which lets organizations use generative AI in Workspace at a lower price point than Gemini Enterprise, which replaces Duet AI for Workspace Enterprise.
@xmpierinix
@tanvir_h
@CERNpress
@jesswade
@CERN
The issue is that women (or in general minorities, targeted individuals) are often expected, if not demanded, to take time off of their other obligations to provide thorough evidence to debunk clearly poorly researched inflammatory statements. It is frankly exhausting ☹️
Slides + video recording of today's
@CERN
seminar with
@DeepMindAI
's
@DeepSpiker
are now available
Topic: "Approximate Inference and Deep Generative Models"
@xmpierinix
@tanvir_h
@CERNpress
@jesswade
@CERN
In other words, it's not our job to do his research for him. He didn't take the time, why should we? (rhetorical)
Also related:
"The burden is on you to waste time disproving my wild unsupported claims" 💁♀️ no thanks.
#sealioning
New paper on
#LotteryTickets
in deep nets & transfer across datasets and optimizers now out on
@arxiv_org
! "One ticket to win them all: generalizing lottery ticket initializations across datasets and optimizers". Work led by
@arimorcos
at
@facebookai
➡️
Want to join one of our amazing teams at
@DeepMind
for an internship?
Apply for the 2023 positions at:
Lots of teams across all our offices look forward to welcoming the new cohort of interns 🤝
This was one of the most inclusive poster sessions I ever attended. Lots of engaging conversation, learning, and teaching. No flexing. These communities are special!
#WiML2019
#BAI2019
Our afternoon joint poster session with
@black_in_ai
was a success! Don’t miss the evening session 6:30pm-8:00pm in East Exhibition Hall B. All
#NeurIPS2019
attendees welcome!
#WiML2019
Do you want to get to know more people doing research in
#MachineLearning
for High Energy
#Physics
? Looking for speakers? Need advice on a project? Check out this directory we just started (modeled after the WiML list): And sign up if you too are in
#HEPML
📣 ATTENTION STUDENTS!!
Apply for a chance to work on an
#AI
project at
@CERN
this summer through the
@Google
Summer of Code program!
You’ll get to collab with some of my favorite people there 👩🏫🧑🏻💻👩🏾🏭👩🏿🎓👨🏼🔧🧑🔬👩🏽💻
RT + get the word out!
@WiMLworkshop
@WiMLDS_Milan
@debora_nozza
I will supervise a fun
#machinelearning
@gsoc
project with
@lukasheinrich_
, and I'm happy we got a number of applicants already - but I wonder how to increase the diversity of the candidate pool. Maybe
@WonderMicky
and
@wimlds
could help pointing out how to advertise this better?
✨ Applications for internships are now open✨
Interning at DeepMind offers a unique opportunity to closely collaborate with our team & work on various projects that aim to advance science and benefit humanity.
Interested? Learn more:
Members of the
@UCBerkeley
and
@BerkeleyLab
community: later today I'll be a guest at the seminar series organized by the Career Development Initiative for the Physical Sciences. I'll talk about my transition from Physics to
#AI
research. Come ask questions in 375 LeConte at 5pm.
Excited to announce this year's Deep Learning for Science Summer School to be held at Lawrence Berkeley National Lab (
@BerkeleyLab
) July 15-19.
No registration fee, candidates selected from application. Application deadline Mar 31. APPLY!
#DL4Sci2019
A reminder that some
@DeepMind
internship applications CLOSE on the 30th September 2022 at midnight UK time (GMT)!
Make sure to apply! Below are some of the role-specific links, but all open internship roles can be found at
🧵
I'm excited to head to
@NeurIPSConf
#NeurIPS2023
this week.
We'll be having a couple of "Chat with the Gemini Team" events in the
@GoogleDeepMind
/
@GoogleResearch
booth areas on Tuesday and Wednesday from 1:30 to 3:00 PM (New Orleans time). Quite a few Gemini team members will
Now in the 2nd half of my postdoc at FAIR, I’m starting to look for a suitable next step in my career. If your team is hiring a Research Scientist, get in touch!
#NeurIPS2019
#WiML2019
Curious to hear Nobel laureate and theoretical physicist Giorgio Parisi’s take on understanding neural networks. He is best known for his work on complex systems, disorder and fluctuations, spin glass theory, and more.
#ML4PS
#NeurIPS2022
On my way to
#NeurIPS2022
for my first in-person conference in three years. Looking forward to (re)connecting with everyone, talk about large scale models, efficiency and sparsity, interpretability, and more! Find me or DM me for chats
I went to shutterstock to find a bunch of laughable interpretations of my job, but instead I found the most accurate representation of what my days are actually like, and now I feel seen
#BadStockPhotosOfMyJob
Afternoon session at the
@MSFTResearch
“Physics meets ML” workshop now underway.
This is one of the first times I’ve felt like I belong to a community ever since transitioning to pure ML research 🤗
It's exciting to see that applications of
#ML
to the Physical Sciences will be once again showcased at a
#NeurIPS
workshop this year!
The field has grown very rapidly, and we're looking forward to sharing these advancements with the rest of the community. Join us in Vancouver!
I'm very excited that our
#NeurIPS2019
Workshop proposal on "Machine Learning and the Physical Sciences" was accepted! 🎉
It's an extension of our previous workshop ( ) with an extended organizing and program committee. More soon!
Happy Monday! Don’t forget to work on your submissions for the
#NeurIPS2019
workshop on Machine Learning and the Physical Sciences.
Deadline: Sept 16, 2019
#ML4PS2019
Impressive lineup of speakers at the
#DL4Sci
school today with
@GoogleAI
’s
@dumoulinv
’s talk on generative models, conditioning & featurewise transformations,
@_beenkim
’s talk on interpretability, and
@facebookai
’s Alex Kirillov’s talk on object detection & image segmentation
Coming back from this work trip I’m realizing that things will change eventually (or even soon!), but, professionally speaking, this is the happiest and motivated I’ve ever been. And it all comes down to people 🥰
PLEASE donate to the hospital in my home town, Busto Arsizio (Lombardy 🇮🇹) where my mom works as a nurse caring for the increasingly alarming number of
#COVID19
patients in one of the worst hit areas in the world right now.
Donate and share:
🙏😷👩⚕️👨⚕️🥼🚑☣️
Had another amazing experience at
#SOCML
this year. Thanks
@Google
for having us,
@goodfellow_ian
& org committee for creating this welcoming&empowering event. I learned so much, met INCREDIBLE people, opened up about painful topics, & shared my experience and insight w/others 🧠
If you're craving more science, here's where you can find scientists to talk to at an AI conference: the
#NeurIPS
workshop on Machine Learning and the Physical Sciences!
Meet ex-scientists now in AI as well as scientists using AI in the physical sciences
Upset at the latest
#NeurIPS2020
reviews? Interested in an alternative publication model for machine learning research? Looking for hypothesis-based science in AI? 👩🔬
Check out and submit to the workshop at NeurIPS.
#SCIENCE
!
Deadline: Oct 7. More info 👇
Talk “by”
@zacharylipton
at the “Integration of Deep Learning Theories” workshop at
#NeurIPS2018
argues that we should embrace a natural science view of machine learning
A personal favorite at
@WiMLworkshop
? The ability of this event to create a welcoming environment in which every participant can feel empowered to ask questions among a non-threatening audience 👩💻👩🏻💻👩🏼💻👩🏽💻👩🏾💻👩🏿💻💪Loving the Q&A sessions!
#WiML2018
In 2020, my life was turned upside down -- I was encountering failure after failure and in deep depression. But that also turned out to be the most enlightening period. In October I was invited to give a talk at Google (thanks to
@orf_bnw
). It's out today:
Neat thing I discovered in
#PyTorch
today: there is a convenient WeightedRandomSampler -- super useful for Physics (
#HEPML
) where all events come with physical weights
#WiML2021
is excited to host our first virtual Mentorship Roundtable session on December 10, 11:00-12:45 UTC. We will be hosting 22 roundtables on both research and career and life advice topics. Check out the full list here:
#NeurIPS2021
These incredible ladies worked days and nights to make
#WiML2019
happen! They handled review, sponsors, speakers, volunteers, mentors, logistics, travel funding, and so much more. I am so proud and thankful to have shared this experience with them. Thank you for everything! ❤️
Calling all women in
@datascience
and
#ML
/
#AI
in
#NYC
!
If you're new to
@PyTorch
and would like to learn more about it, become proficient, and potentially contribute to it, sign up for the intro workshop I'll be teaching Nov 19 with
@wimlds
@WiMLDS_NYC
!
Spread the word!
The WiML workshop page and Call for Participation are now live. Share it with your female colleagues, and invite them all to submit and attend. Great opportunity for exposure and networking! Fantastic community. Inspiring speakers. Only 1-page abstract needed👩🏼💻
Call for Participation! The 14th WiML Workshop will be co-located with
@NeurIPSConf
in Vancouver, Canada on Monday, December 9th, 2019.
Deadline for submission: August 15th, 2019
Read the full CFP:
#WiML
#WiML2019
#NeurIPS2019
📣 Join me,
@jefrankle
, and our coorganizers for this new event focused on *sparsity* in neural networks!
Call for papers at
Join us virtually to hear from experts on the AI theory, experiment, and hardware side.
NEW WORKSHOP: Sparsity in Neural Networks: Advancing Understanding and Practice (July 8-9, 2021). This workshop will bring together members of the many communities working on neural network sparsity to share their perspectives and the latest cutting-edge research (Deadline: 6/15)
This is what scientists from Berkeley, Princeton, Yale, JPL, CERN, and Facebook look like, y’all. Update your stereotypes 🤓
(okay, maybe we’re not the best representation of ethnic diversity, but it’s a start)
We're building a brand new team at
@DeepMind
dedicated to data, and we're looking for Data Research Engineers to join us.
Feel free to reach out, or apply online here:
Intense week ahead of me in Milan 🇮🇹 then Lausanne & Geneva 🇨🇭starting with a talk on model compression, pruning, the “lottery ticket hypothesis” & more at
@unimib
, hosted by
@debora_nozza
, then at
@PrometeiaGroup
on the 24th. Stop by if you’re in Milan!
Today’s breakfast reading: class notes in group theory from my Ph.D. days at Yale with prof. Iachello. Looking back, that was a great choice for an elective!
@YalePhysicsDept
🇮🇹
Just for fun, first gif for upcoming
#trackml
@kaggle
tracking challenge. Thanks
@edwardmoyse
for the viz). 100'000 3D points to be grouped into 10'000 tracks, arc of helices coming from the center. Your eyes can't do it. Your algorithm should.
Evaluating the
#Gemini
models and providing clear actionable direction for improving them across all capabilities day after day has been some of the most exciting and empowering work I've ever done. We need to aspire to deliver models that work well everywhere for everyone.