ISSUE 458 · October 24, 2023TrendsSealed in glassThis may be Microsoft's most important contribution to tech so far. Project Silica’s coaster-size glass plates can store terrabytes of data for thousands of years. The plates can be burned, boiled, dropped, scoured, and microwaved and still, the data will persist, unscathed. Cool project! What will an AI revolution in medicine look like?Five years ago, the renowned AI expert, Geoffrey Hinton, proclaimed that machine learning tools would soon displace radiologists. So far, that hasn't happened and probably won't anytime soon. This article takes a look at the landscape of AI tooling, the challenges, and what's next now. Sponsored LinkLearning to learn learning-rate schedulesAmazon researchers are developing algorithms that can learn to schedule by harnessing data from past experiments. In a sequence of recent papers, they describe three phases of their research. Posts & TutorialsEmbeddings: What they are and why they matterGreat introduction to embeddings, which are important in LLMs and, of course, vector databases. This is easy to follow and manages to go deep without getting into a lot of jargon. Includes a recent presentation at PyBay, code snippets, and lots of resources along the way. Joining CSV Data Without SQLYou've got csv data. You want to run sophisticated queries against it. Does that mean you need to turn it into SQL tables first? This post shows how to approach the problem using the open-source tool, Zed. Interactive Linear Model Hypeparameter TuningThis set of interactive dashboards lets you explore the use of a hyperparameter to tune linear machine learning models. The hyperparameter controls the degree of fit to the data, known as model regularization. This notebook visualizes both ridge and LASSO regression. There's a lot of great work in this repo. Look around! 30-Day Map Challenge 🌎Whether you're a GIS expert or have never made a map before, if you're interested in making maps with data, you're sure to learn something useful during this annual challenge. If you just want to follow along, maps are published using the hashtag #30DayMapChallenge. Tools & CodeRTutorThis project has evolved a lot since being mentioned in Data Elixir in January. RTutor uses ChatGPT to translate natural language into R code, which is then executed. You can upload a data file and then request an analysis, just like asking a real person. It can also be run locally and the latest version uses GPT-4 and ggplot2 as defaults. ResourcesDeep Learning CourseAwesome deep learning course by François Fleuret at the University of Geneva, Switzerland. The course is a thorough introduction to deep-learning, with examples in the PyTorch framework. Everything is freely available, including handouts, slides, and recordings. 2024 Internship Opportunities in Engineering & DataGreat collection of internship announcements for 2024. This list covers a wide variety of organizations, such as Dropbox, NASA, Adobe, Nike, NVDIA, Pinecone, Tesla, and many more. Includes hourly pay and links to application info, typical interview questions, and Glassdoor ratings Was this email forwarded to you? Sign up here >> |
Поиск по этому блогу
Search1
123
вторник, 24 октября 2023 г.
Data Elixir - Issue 458
Подписаться на:
Комментарии к сообщению (Atom)
Комментариев нет:
Отправить комментарий
Примечание. Отправлять комментарии могут только участники этого блога.