A PyTorch re-implementation of GPT training. minGPT tries to be small, clean, interpretable and educational, as most of the currently available ones are a bit sprawling. GPT is not a complicated model and this implementation is appropriately about 300 lines of code, including boilerplate and a totally unnecessary custom causal self-attention module...
An A.I. Training Tool Has Been Passing Its Bias to Algorithms for Almost Two Decades
CoNLL-2003 — [is] biased in an important way: The roughly 20,000 news wire sentences...annotated contain many more men’s names than women’s names, according to a recent experiment by [a] data annotation firm... this means that a model trained on CoNNL-2003 [cited more than 2,500 times in research literature in the past 17 years] wouldn’t just fall short when it comes to identifying the current names included in the dataset — it would fall short in the future, too, and likely perform worse over time. It would have more trouble with women’s names, but it would also likely be worse at recognizing names more common to minorities, immigrants, young people, and any other group that wasn’t regularly covered in the news two decades ago...
Challenges of Real-World Reinforcement Learning
Reinforcement learning (RL) has proven its worth in a series of artificial domains, and is beginning to show some successes in real-world scenarios. However, much of the research advances in RL are often hard to leverage in real-world systems due to a series of assumptions that are rarely satisfied in practice. We present a set of nine unique challenges that must be addressed to productionize RL to real world problems. For each of these challenges, we specify the exact meaning of the challenge, present some approaches from the literature, and specify some metrics for evaluating that challenge...
A Message From This Week's Sponsor
Data scientists are in demand on Vettery
Vettery is an online hiring marketplace that's changing the way people hire and get hired. Ready for a bold career move? Make a free profile, name your salary, and connect with hiring managers from top employers today.
Data Science Articles & Videos
A data science approach to 138 years of congressional speeches
Here we apply a quantitative analysis to a large dataset of USA congressional speeches made over a period of 138 years. The analysis reveals that the readability index of congressional speeches increased consistently until the 96th congress, and then started to decline. Congressional speeches have also become more positive over time, and in general express more sentiments compared to speeches made in the 19th century or early 20th century...
Photonics startup Lightmatter details its AI optical accelerator chip
Ahead of the Hot Chips conference this week, photonics chip startup Lightmatter revealed the first technical details about its upcoming test chip, which is on track for a fall 2021 release. Unlike conventional processors and graphics cards, the test chip uses light to send signals, promising orders of magnitude higher performance and efficiency...
FastMRI breakthrough shows AI-accelerated MRIs interchangeable with traditional MRIs
FastMRI, a joint research initiative from Facebook AI and NYU Langone Health, aims to develop new ways to use AI to accelerate the MRI scanning process. Unlike most AI medical imaging projects, which try to use AI to automatically review images to detect anomalies, fastMRI is using AI to create images in a new way that requires much less data...In a rigorous new clinical study, radiologists found fastMRI’s AI-generated images — created with about 4x less data from the scanning machine — were diagnostically interchangeable with traditional MRIs. This means fastMRI can make the scanning process much faster....
Vid2Player: Controllable Video Sprites that Behave and Appear like Professional Tennis Players
We present a system that converts annotated broadcast video of tennis matches into interactively controllable video sprites that behave and appear like professional tennis players. Our approach is based on controllable video textures, and utilizes domain knowledge of the cyclic structure of tennis rallies to place clip transitions and accept control inputs at key decision-making moments of point play. Most importantly, we use points from the video collection to model a player’s court positioning and shot selection decisions during points...
A Simulation Suite for Tackling Applied Reinforcement Learning Challenges
In “Challenges of Real-World Reinforcement Learning” [See Editor's Picks above], we identify and discuss nine different challenges that hinder the application of current RL algorithms to applied systems. We then follow up this work with an empirical investigation in which we simulated versions of these challenges on state-of-the-art RL algorithms, and benchmark the effects of each. We have open-sourced these simulated challenges in the Real-World RL (RWRL) task suite to help draw attention to these important issues, as well as accelerate research toward solving them...
The Computational Limits of Deep Learning
This article reports on the computational demands of Deep Learning applications in five prominent application areas and shows that progress in all five is strongly reliant on increases in computing power. Extrapolating forward this reliance reveals that progress along current lines is rapidly becoming economically, technically, and environmentally unsustainable. Thus, continued progress in these applications will require dramatically more computationally-efficient methods, which will either have to come from changes to deep learning or from moving to other machine learning methods...
Fast reinforcement learning with generalized policy updates
The combination of reinforcement learning with deep learning is a promising approach to tackle important sequential decision-making problems that are currently intractable. One obstacle to overcome is the amount of data needed by learning systems of this type. In this article, we propose to address this issue through a divide-and-conquer approach. We argue that complex decision problems can be naturally decomposed into multiple tasks that unfold in sequence or in parallel. By associating each task with a reward function, this problem decomposition can be seamlessly accommodated within the standard reinforcement-learning formalism...
Quick Question For You: Do you want a Data Science job?
After helping hundred of readers like you get Data Science jobs, we've distilled all the real-world-tested advice into a self-directed course
The course is broken down into three guides:
Click here to learn more
- Data Science Getting Started Guide. This guide shows you how to figure out the knowledge gaps that MUST be closed in order for you to become a data scientist quickly and effectively (as well as the ones you can ignore)
- Data Science Project Portfolio Guide. This guide teaches you how to start, structure, and develop your data science portfolio with the right goals and direction so that you are a hiring manager's dream candidate
- Data Science Resume Guide. This guide shows how to make your resume promote your best parts, what to leave out, how to tailor it to each job you want, as well as how to make your cover letter so good it can't be ignored!
*Sponsored post. If you want to be featured here, or as our main sponsor, contact us!
Want to post a job here? Email us for details >> email@example.com
Data Scientist (Entry Level) - Saturn Cloud - Remote
Saturn Cloud helps companies perform data science at a new level of scale, with one-click solutions, to solve the world’s hardest problems. Our product is a SaaS platform which equips data science teams with high-leverage automation tools, eliminating hours of traditional, manual work. The platform is user-friendly, scalable and secure.
You will be an entry-level Data Scientist for Saturn Cloud, an exciting new venture founded by the creators of Anaconda, NumPy, and SciPy. The role features drafting the first generation of Saturn resource materials, tutorials, and technical content...
Training & Resources
Hidden Gems and Underappreciated Resources [Reddit Discussion]
I'm currently working on a project to curate the currently massive number of ML resources, and I noticed that there are courses like CS231n or David Silver's that come up repeatedly (for a good reason). But there seems to be lots of other quality resources that don't receive as much widespread appreciation...So, here are a few hidden gems that, imo, deserve more love...
Hands-On Machine Learning with scikit-learn and Scientific Python Toolkits
Integrate scikit-learn with various tools such as NumPy, pandas, imbalanced-learn, and scikit-surprise and use it to solve real-world machine learning problems...
For a detailed list of books covering Data Science, Machine Learning, AI and associated programming languages check out our resources page
P.S., Enjoy the newsletter? Please forward it to your friends and colleagues - we'd love to have them onboard :) All the best, Hannah & Sebastian