EGG is a toolkit that allows researchers to create game simulations in which two agents devise their own communication system to solve a task together. | Continue reading
Hermes improves app performance, focusing on React Native apps, even on mass-market devices with limited memory, storage, and computing power. | Continue reading
Hermes improves app performance, focusing on React Native apps, even on mass-market devices with limited memory, storage, and computing power. | Continue reading
OIL+VCache is a new take on generic storage abstractions with distributed virtual memory that provides significant efficiency and performance gains. | Continue reading
At Performance@Scale, we shared open source browser contributions to improve web app startup time, including isInputPending, Self-profiling API, and BinAST. | Continue reading
Aria is a set of initiatives to dramatically increase PrestoDB efficiency. Our goal is a 2-3x decrease in CPU time for Hive queries against ORC tables. | Continue reading
Delos, a fundamentally new architecture for replicated storage systems, provides flexibility and simplicity without sacrificing performance or reliability. | Continue reading
Tupperware advances the state of the art for cluster management with seamless support for stateful service and a single control plane across data centers. | Continue reading
How we built our service encryption infrastructure to optimize for operability and performance, while satisfying the right security model for each service. | Continue reading
The latest incarnation of our DHCP infrastructure extends DHCPLB to operate as a server. We can now handle the same volume with 10 times fewer servers. | Continue reading
Pythia is a new open source deep learning framework that enables researchers to quickly build, reproduce, and benchmark AI models. | Continue reading
The idea of intelligent machines has fascinated people for centuries. And while it may seem like something out of science fiction, people today use artificial intelligence every day in their smartphones, houses, cars and more. It’s becoming more and more prevalent in our world. A … | Continue reading
We are announcing our participation as a founding member of the Urban Computing Foundation (UCF), a continuation of Facebook’s commitment to open standards. | Continue reading
[...]Read More... | Continue reading
At F8, we shared open source tools that simplify ML experimentation, speed up test execution times, and help solve memory-related performance issues. | Continue reading
Topics at F8 2019 day 1 included making Messenger and WhatsApp fast, secure, and private, and the latest innovations in AR and VR. | Continue reading
To simplify the process of selecting the right hash table, we built F14, part of Folly, our open source library of C++ components. | Continue reading
To simplify the process of selecting the right hash table, we built F14, part of Folly, our open source library of C++ components. | Continue reading
To address queueing time delays on the web, we worked with engineers at Chrome to develop isInputPending. The origin trial for this API is now available. | Continue reading
We created Aroma, a code-to-code search and recommendation tool that uses ML to make the process of gaining insights from big codebases much easier. | Continue reading
At our first-ever London Facebook Engineering Fair, attendees got a look at the products and technologies being worked on by our U.K. engineers. | Continue reading
We are releasing the first experimental back end for our Glow compiler and runtime project, designed to target Habana's existing hardware accelerator. | Continue reading
Facebook Reality Labs is using groundbreaking 3D capture technology and AI systems to create lifelike virtual avatars that can be animated in real time. | Continue reading
A new tool from FAIR, PyTorch-BigGraph enables training of multi-relation graph embeddings for graphs with billions of nodes and trillions of edges. | Continue reading
Python Test Runner (ptr) allows developers to run Python unit test suites in parallel by crawling a repository to find projects with defined unit tests. | Continue reading
Today we celebrated the grand opening of our Los Lunas data center. This center will be one of our first to be certified Gold under LEED version 4. | Continue reading
A new approach to pretraining cross-lingual models for natural language processing tasks improves both supervised and unsupervised machine translation. | Continue reading
[...]Read More... | Continue reading
Facebook AI Research is working with the AI research community on the ZeroSpeech 2019 challenge to teach AI systems to learn speech more like children do. | Continue reading
We set out to learn how much modern sentence embeddings gain from training (surprisingly little) and to provide the field with more appropriate baselines. | Continue reading
We're sharing a new predictive model for electrical grid mapping. We believe this is the most accurate, globally scalable product of its kind. | Continue reading
Our natural language processing toolkit, LASER, performs zero-shot cross-lingual transfer with more than 90 languages and is now open source. | Continue reading
To make the mobile image upload process more efficient, we are releasing Spectrum 1.0.0, an image processing library for Android and iOS. | Continue reading
The Singapore data center is our first custom-built hyperscale center in Asia, designed with features that minimize the use of land, energy, and water. | Continue reading
In 2018, our AI strategy included investing in long-term research, building cutting-edge tools and platforms, and using AI to benefit the world. | Continue reading
Facebook believes in the value of open source technology. In 2018, we released 153 new open source projects. Our portfolio contains a total of 474 projects. | Continue reading
Wav2letter++ is the fastest state-of-the-art end-to-end speech recognition system available. We're also releasing flashlight, a fast, flexible ML library. | Continue reading
Facebook and Cloudflare are piloting DNS over TLS to better understand how the protocol behaves in a production workload. | Continue reading
We are open-sourcing Nevergrad, a Python3 library that makes it easier to perform gradient-free optimizations used in many machine learning tasks. | Continue reading
Zstandard, our open source data compression solution, increased compression speed up to 10x. How to mplement new advanced capabilities for similar benefits. | Continue reading
Facebook Reality Labs has built DeepFocus, a new AI-powered framework for rendering natural, realistic focus effects in real time. We're now open-sourcing our networks and data set. | Continue reading
HyperLogLog (HLL) in Presto, a distributed SQL query engine, provides an approximate count of distinct elements using a function called APPROX_DISTINCT. | Continue reading
We support the MLPerf initiative to develop Industry-standard ML models and benchmarks. As part of this support, we are also open-sourcing Mask R-CNN2Go. | Continue reading
A framework for using CNNs on satellite imagery can produce more accurate information in far less time than current disaster mapping methods. | Continue reading
[...]Read More... | Continue reading
The stable release of PyTorch 1.0 is now available, providing researchers and engineers with new capabilities, such as production-oriented features and support from major cloud platforms, for accelerating the AI development workflow | Continue reading
Recently, we hosted Mobile @Scale, an invitation-only technical conference for engineers working on large-scale mobile applications. The first-ever @Scale event in Tel Aviv attracted more than 300 attendees to hear talks on topics such as the efficiency of tools and teams, and te … | Continue reading