Pablo de Castro
@pablodecm
Followers
468
Following
1K
Media
18
Statuses
244
Building awen (YC W25). Prev - Product and Tech @ reforestum. Data and ML @ treelogic. Physics PhD Statistical Learning and Inference @ LHC at CERN.
Spain, Europe
Joined April 2014
Very happy to launch and share what we have been building at @awen_ai and we are just getting started! Whatever you can think? You can see - you just have to say it.
🎨@Awen_ai is rebuilding Photoshop, with an AI-voice interface. Instead of navigating complex menus, creatives can simply describe their vision, and Awen will bring it to life. https://t.co/cnjKRpyNi3
1
2
15
Really powerful video to build (or refresh) some intuitions and scaling notions around Large Language Models (LLMs) like GPT. Slightly technical yet it is amenable for people with expertise in adjacent disciplines (e.g. physics, statistics, engineering, etc).
LLMs in Five Formulas: A somewhat idiosyncratic tutorial on LLMs. The goal is to highlight five independent areas in LLMs that we kinda understand, while being humble about how hard the rest is. https://t.co/JEWo3bgNOO
0
0
0
It is so easy to fall into a pure consumer/reader mode in this network if you are curiosity-driven, so much interesting stuff all the time... Thus, after a four-year long hiatus, I am back into creating and sharing stuff here often! 🚀 Anyone in a similar situation? Any tips?
1
0
1
📢 Open PhD positions👩‍🎓👨‍🎓 Fully-funded Ph.D. positions in the areas of Machine Learning, Computer Vision, and Physics-Based Simulation for 3D human modeling, human interaction, and understanding of crowds. All details: https://t.co/h4ewEsP7oE Likes and RTs highly appreciated!
8
181
404
An open-source server for hosting conda packages. A fast replacement to the conda command line utility. Check out our new blog post on "Open Software Packaging for Science". https://t.co/nCm4UKXUht
medium.com
Modern scientific applications typically depend on a very large number of libraries written in various programming languages, ranging from…
3
88
212
Assume training a model takes weeks, but testing is fast. Estimating the variance of the performance of this model is not realistically feasible using cross-validation. How would you proceed to evaluate the uncertainty around the performance estimate?
4
4
11
Been reading lots of stats and ML methods papers in high-energy physics for a review lately. Maybe is the clarity of almost one year doing something else, but it seems the barrier for much better data analyses is mainly a (software) tooling and integration problem. Any takes?
4
2
13
I have updated my course notes on automatic differentiation (last section of the PDF). Now includes dual numbers, adjoint state method, argmin layers, envelope theorem, reversible architectures. Thx to @PierreAblin for the constructive criticisms :) https://t.co/Bfc7FXpZpT
6
186
824
So #AcademicTwitter, a preprint takes on same problem with same conceptual solution but with a minor change from one of my PhD publications. Cite in related work incorrectly claiming different applicability and does not attribute ideas in intro or methodology. Any advice?
2
2
3
This is the model from Report 13 of the Imperial College COVID-19 response team. It fits the death data jointly from 11 European countries to estimate the reproduction number and the effect of lockdowns. Such a remarkable piece of @mcmc_stan
4
22
62
Equivariance is such a beautiful concept and so important for machine learning. @EdWagstaff and I spent many hours setting our wits to understanding it from both an intuitive and theoretical points of view. This is the first part of our write-up. 🦊 https://t.co/5iyYlSm0eV
1
13
75
Profoundly awesome thing for computational science: You can automatically get chi-squared contours with jax and inverting the Fisher matrix, with ONE function evaluation. Jax: https://t.co/ObN2wQ5nEv Here is a paper explaining Fisher matrices. https://t.co/JQd0LLpYvC
arxiv.org
Fisher matrices are used frequently in the analysis of combining cosmological constraints from various data sets. They encode the Gaussian uncertainties of multiple variables. They are simple to...
5
39
208
Quite comprehensive questionarie/check-list for data science projects, particularly relevant for consulting projects in external organizations.
I completed my 1st data science project ~30 years ago. Since then I've been continuously developing a questionnaire I use for all new data projects, to ensure the right info is available from the start. I'm sharing it publicly today for the first time. https://t.co/fwUobkPvvp
0
0
3
I found @artix41 meme-rich slides quite useful to understand a bit more about Quantum Machine Learning (QML) and where that field is going.
Thanks @FatrasKilian, @nicolas_courty and all the OBELIX team @irisa_lab (@CNRS) for inviting me to give a talk at their seminar. It was a blast! You can find the slides of my talk Quantum Machine Learning Beyond the Hype online: https://t.co/C8ASSO3iYG
1
0
5
Fantastic review of simulation-based inference (one of my main research interests during my PhD) by @KyleCranmer, J. Brehmer and @glouppe. Strong endorsement and a must read if you are into the intersection of machine learning, science and stats.
In our newest paper we discuss the frontier of simulation-based inference (aka likelihood-free inference) for a broad audience. We identify three main forces driving the frontier including: #ML, active learning, and integration of autodiff and probprog. https://t.co/R6vMUAnaul
2
6
20
Great blog post by Scott Aaronson explaining Google's quantum supremacy result, as well as the origin of the leak: https://t.co/9EkeTZbjZX
scottaaronson.blog
You’ve seen the stories—in the Financial Times, Technology Review, CNET, Facebook, Reddit, Twitter, or elsewhere—saying that a group at Google has now achieved quantum computation…
0
3
6
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: https://t.co/etxb3rqRZr 🙏
docs.google.com
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 https://t.co/EoqCIzOWMr
#ML4PS2019
3
52
105
We're hiring! Research Software Engineer for NYU’s Data Science and Software Services (DS3) Please share widely. @NYUDataScience @NYU_PRIISM @NYULibraries @MooreFound @SloanFoundation @UCBIDS @uwescience @digitalFlaneuse
https://t.co/z9ZHJ68h8k
0
35
52
Our article "INFERNO: Inference-Aware Neural Optimization" is now online in the Computer Physics Communications site - gold open access. Much improved from the preprint, check it out at https://t.co/tW3XZ8POrW
#MachineLearning @pablodecm @KyleCranmer @glouppe @DanielWhiteson
0
15
25