Better Language Models and Their Implications
“Our model, called GPT-2 (a successor to GPT), was trained simply to predict the next word in 40GB of Internet text. Due to our concerns about malicious applications of the technology, we are not releasing the trained model. As an experiment in responsible disclosure, we are instead releasing a much smaller model for researchers to experiment with, as well as a technical paper.”
AI Researchers Debate the Ethics of Sharing Potentially Harmful Programs
The debate has been wide-ranging and sometimes contentious. It even turned into a bit of a meme among AI researchers, who joked that they’ve had an amazing breakthrough in the lab, but the results were too dangerous to share at the moment.
Dear OpenAI: Please Open Source Your Language Model
“While OpenAI is correct to be concerned about potential misuse, I disagree with their decision not to open source the GPT-2. To justify this, I first argue that only certain types of dangerous technology should be controlled by suppressing access. Then, on the basis of this analysis, I argue that withholding the full GPT-2 model is both unnecessary for safety reasons and detrimental to future progress in AI.”
PyTorch Reimplementation of OpenAI GPT-2 Small Model Released
Github developer Hugging Face has updated its repository with a PyTorch reimplementation of the GPT-2 language model small version that OpenAI open-sourced last week, along with pretrained models and fine-tuning examples.
Optimizing Network Performance for Distributed DNN Training on GPU Clusters: ImageNet/AlexNet Training in 1.5 Minutes
“We propose a communication backend named GradientFlow for distributed DNN training, and employ a set of network optimization techniques. First, we integrate ring-based allreduce, mixed-precision training, and computation/communication overlap into GradientFlow. Second, we…”
(SenseTime Research & Nanyang Technological University)
SC-FEGAN: Face Editing Generative Adversarial Network with User’s Sketch and Color
“We present a novel image editing system that generates images as the user provides free-form mask, sketch and color as an input. Our system consist of a end-to-end trainable convolutional network. Contrary to the existing methods, our system wholly utilizes free-form user input with color and shape.”
(ETRI South Korea)
NAS-Generated Model Achieves SOTA In Super-Resolution
Single image super resolution (SISR) is a critical research challenge for smartphone image processing, but current state-of-the-art models in this domain are hand-crafted by human experts. Chinese smartphone giant Xiaomi is challenging this labour-intensive approach with a new machine-generated model that achieves impressive results in the super-resolution domain.
You May Also Like
Yann LeCun Cake Analogy 2.0
Yann LeCun updated his cake recipe last week at the 2019 International Solid-State Circuits Conference (ISSCC) in San Francisco, replacing “unsupervised learning” with “self-supervised learning,” a variant of unsupervised learning where the data provides the supervision.
Oxford University AI Policy Researcher Says Trump’s AI Initiative Falls Short on Immigration and Ethics Issues
Last Monday US President Donald Trump signed the “American AI Initiative,” an executive order designed to spur US investment in AI and boost the domestic AI industry. The initiative has five highlights: Investing in AI R&D, Unleashing AI Resources, Setting AI Governance Standards, Building the AI Workforce, International Engagement and Protecting our AI Advantage.
Global AI Events
March 8~10, World Symposium on Artificial Intelligence Xi’an, China
March 11~12, Conversational Interaction Conference San Jose, United States
March 12~13, AI Tech World London, United Kingdom
March 17~20, ACM IUI Los Angeles, United States
Global AI Opportunities