AI Weekly Malaysia

We’ve scaled Kubernetes clusters to 7,500 nodes, producing a scalable infrastructure for large models like GPT-3, CLIP, and DALL·E, but also for rapid small-scale iterative research such as Scaling Laws for Neural Language Models.

We’ve trained a neural network called DALL·E that creates images from text captions for a wide range of concepts expressible in natural language.

We’re introducing a neural network called CLIP which efficiently learns visual concepts from natural language supervision. CLIP can be applied to any visual classification benchmark by simply providing the names of the visual categories to be recognized, similar to the “zero-shot” capabilities of GPT-2 and GPT-3.

It’s been a year of dramatic change and growth at OpenAI.

ECF, P2P financing funds raised in 2020 to date benefitting more than 2.5k MSMEs Demand for online brokerage services in 2020 increased 270% compared to 2019 [Caption above has been edited for clarity.] “Given it is a tech event, it is only fitting we use technology to connect,” says Securities Commission (SC) Malaysia chairman Syed Zaid Albar when giving his welcome remarks to the SC x SC Fintech Conference 2020. This marks the 7th iteration of the annual conference that, due to the Covid-19 pandemic, is held virtually for the first time. Held from 5 to 7 October 2020, the event covers discussions on a broad range of topics that include local fintech players navigating the ‘new normal’, investment behaviour in turbulent times, and the role of fintech in responsible fundraising and investing. Zaid Albar acknowledges the difficulty of the current times. Yet there are silver linings amidst the gloom. For one, Syed notes that despite the challenging environment, investors have continued to show confidence in the Malaysian capital markets. “We have seen a considerable increase in individual investor participation via digital investment managers, equity crowdfunding (ECF) and peer-to-peer (P2P) financing platforms, digital asset exchanges (DAX) and online brokers. So, the retail participation in this challenging environment has only increased,” he says. The silver lining certainly shines bright here, with some fascinating discoveries. For one, the total funds raised on ECF and P2P platforms have broken the US$240 million (RM1 billion) mark in 2020, benefitting more than 2,500 MSMEs. “Supporting this growth are investors under the age of 35, who account for 60% of individual investors. The presence of retail investors – constituting 84% of participating individual investors - further underscores the appeal of these platforms,” Zaid Albar elaborates. Other numbers of significance There are other positive things to note. This year, SC issued three more digital investment management licenses, totalling seven overall. “I am pleased to note that these digital investment managers have attracted many first-time investors… with close to 90,000 new accounts opened this year,” Zaid Albar continues. In addition to that, demand for online brokerage services have also increased – the number of new account openings through online-only brokers, grew by more than 270% this year compared 2019. “That is an almost four-fold increase. Their average trading volume has also tripled this year, so once again, lots of silver lining.” Zaid Albar goes on to say that SC approved three digital asset exchanges last year – and all three have now gone live with four digital assets permitted on trading on these platforms. Collectively, these DAXes have seen more than 400,000 accounts being opened, with the value of trades surpassing US$24 million (RM100 million) in the month of August. According to Zaid Albar, the country has seen rapid adoption of dig

OpenAI has agreed to license GPT-3 to Microsoft for their own products and services.

We’ve applied reinforcement learning from human feedback to train language models that are better at summarization.

Our third class of OpenAI Scholars presented their final projects at virtual Demo Day, showcasing their research results from over the past five months.

We’re excited to announce that OpenAI is co-organizing two NeurIPS 2020 competitions with AIcrowd, Carnegie Mellon University, and DeepMind, using Procgen Benchmark and MineRL.

We find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. By establishing a correlation between sample quality and image classification accuracy, we show that our best generative model also contains features competitive with top convolutional nets in the unsupervised setting.

We’re releasing an API for accessing new AI models developed by OpenAI.