At its recently concluded Think 2021 conference, IBM introduced Project CodeNet to develop machine learning models that can help in programming. The large dataset consists of 14 million code samples and 500 million lines of code in over 55 different languages, including C, Java, Go, Python, COBOL, Pascal, and Fortran. Modern computer programs have millions of lines of code and are hard to debug, maintain, update, and document. The use of artificial intelligence to write codes has been an important area of research for many years. However, it is easier said than done.

This is a basic Python shell (really, it's a fancy wrapper over the system shell) that takes a task and asks OpenAI for what Linux bash command to run based on your description. For safety reasons, you can look at the command and cancel before actually running it. To be clear, I'm not trying to convince you that having an AI model figure out what Linux command to run based on your written description is a good idea, but the commands that it generates are, well - watch the video if you want to see. There are several pre-canned ways of interacting with the models that OpenAI provides (the "GPT" models): completing a provided fragment, answering a question, generating "ideas" from a topic, summarizing a passage, etc. This shell uses the question-and-answer format and provides the model with an "example context" and examples of input and output.

MONTREAL, March 16, 2021 /CNW/ - Amgen announced today that it has entered into a multi-year partnership with Mila – Quebec Artificial Intelligence Institute. One of the world's leading biotechnology companies, Amgen uses cutting-edge science and technology to discover and develop innovative therapeutics, and AI is currently deployed in several of its R&D and manufacturing activities. This new partnership will permit Amgen to expand its knowledge of AI and deep learning by interacting and engaging with experts in Mila's unique ecosystem, both virtually and (eventually) on the Mila campus in Montreal. This announcement was made during Effervescence 2021, an international virtual gathering of scientists, entrepreneurs and industry professionals from the life sciences and health technology sectors in the presence of the Quebec Minister of Economy and Innovation, Pierre Fitzgibbon. "We are very happy to have signed this partnership with Mila," said Philip Tagari, Amgen's Vice President of Research, Therapeutic Discovery.

This code pattern explains how to classify an American Sign Language (ASL) alphabet using PyTorch and deep learning networks. It uses a pretrained model from the PyTorch models zoo and retrains the last part of the network. The code pattern uses PyTorch to build and train a deep learning model to classify images to 29 classes (26 ASL alphabet, space, Del, and nothing), which can be used later to help hard-of-hearing people communicate with others as well as with computers. The pattern uses a pretrained mobile network, defines a classifier, and connects it to network. It then trains this classifier along with some of the last blocks of the network on the data set.

The following paper, "Simba: Scaling Deep-Learning Inference with Chiplet-Based Architecture," by Shao et al. presents a scalable deep learning accelerator architecture that tackles issues ranging from chip integration technology to workload partitioning and non-uniform latency effects on deep neural network performance. Through a hardware prototype, they present a timely study of cross-layer issues that will inform next-generation deep learning hardware, software, and neural network architectures. Chip vendors face significant challenges with the continued slowing of Moore's Law causing the time between new technology nodes to increase, sky-rocketing manufacturing costs for silicon, and the end of Dennard scaling. In the absence of device scaling, domain specialization provides an opportunity for architects to deliver more performance and greater energy efficiency. However, domain specialization is an expensive proposition for chip manufacturers.

Over the last decade, deep learning systems have shown an astonishing ability to classify images, translate languages, and perform other tasks that once seemed uniquely human. However, these systems work opaquely and sometimes make elementary mistakes, and this fragility could be intentionally exploited to threaten security or safety. In 2018, for example, a group of undergraduates at the Massachusetts Institute of Technology (MIT) three-dimensionally (3D) printed a toy turtle that Google's Cloud Vision system consistently classified as a rifle, even when viewed from various directions. Other researchers have tweaked an ordinary-sounding speech segment to direct a smart speaker to a malicious website. These misclassifications sound amusing, but they could also represent a serious vulnerability as machine learning is widely deployed in medical, legal, and financial systems.

Many believe the company that enables real intelligence on end devices (such as mobile and IoT devices) will define the future of computing. Racing toward this goal, many companies, whether tech giants such as Google, Microsoft, Amazon, Apple and Facebook, or startups spent tens of billions of dollars each year on R&D. Assuming hardware is the major constraint for enabling real-time mobile intelligence, more companies dedicate their main efforts to developing specialized hardware accelerators for machine learning and inference. Billions of dollars have been spent to fuel this intelligent hardware race. This article challenges the view.

Package-level integration using multi-chip-modules (MCMs) is a promising approach for building large-scale systems. Compared to a large monolithic die, an MCM combines many smaller chiplets into a larger system, substantially reducing fabrication and design costs. Current MCMs typically only contain a handful of coarse-grained large chiplets due to the high area, performance, and energy overheads associated with inter-chiplet communication. This work investigates and quantifies the costs and benefits of using MCMs with finegrained chiplets for deep learning inference, an application domain with large compute and on-chip storage requirements. To evaluate the approach, we architected, implemented, fabricated, and tested Simba, a 36-chiplet prototype MCM system for deep-learning inference. Each chiplet achieves 4 TOPS peak performance, and the 36-chiplet MCM package achieves up to 128 TOPS and up to 6.1 TOPS/W. The MCM is configurable to support a flexible mapping of DNN layers to the distributed compute and storage units. To mitigate inter-chiplet communication overheads, we introduce three tiling optimizations that improve data locality. These optimizations achieve up to 16% speedup compared to the baseline layer mapping. Our evaluation shows that Simba can process 1988 images/s running ResNet-50 with a batch size of one, delivering an inference latency of 0.50 ms. Deep learning (DL) has become critical for addressing complex real-world problems.

Microsoft is using the power of GPT-3's natural language AI to help people who don't know how to code write their own software using Microsoft's PowerApps development platform. The announcement was made at Microsoft's Build developer conference today. Microsoft has hoped that PowerApps would become a powerful corollary to its Office suite, but the platform has languished a bit. Microsoft originally set up PowerApps in 2015 around a set of programming templates, pulling data from user-defined sources and then outputting results. Think of it like the next level of a traditional macro in Microsoft Office--it's a way for an average user to write a program to instruct Windows to perform a task, but with minimal or no knowledge of program coding.

The future is hurtling toward us, and with each passing day AI is getting smarter. We've seen AI that will play your games for you, and if you're using a 30-series Nvidia GPU, you've probably utilised the company's DLSS supersampling AI tech to upscale your game resolutions. But I bet you weren't aware AI can now not only detect sarcasm, it can also then rap about it. Even we humans have trouble with sarcasm in it's written form, especially in the case of Twitter where the majority of users refuse to use punctuation. Thankfully a study from DARPA researchers, at the University of Central Florida, has birthed an AI that can recognise the tone of an internet troll with over 80% accuracy on Reddit, and on Twitter it surpasses 90% accuracy (via Engadget).