OpenAI Releases 1.5 Billion Parameter GPT-2 Model
OpenAI announced the final staged release of its 1.5 billion parameter language model GPT-2, along with all associated code and model weights. GPT-2 is a large language model that can generate realistic paragraphs of text.
(Synced) / (OpenAI) / GitHub
Baidu Announces Third Quarter 2019 Results
Baidu, Inc., the leading Chinese-language Internet search provider, announced its unaudited financial results for the third quarter ended September 30, 2019. “Baidu App traffic continues to grow robustly with DAUs reaching 189 million, up 25% year over year, in September and Baidu’s in-app search continues to gain market share.” said Robin Li, Chairman and CEO of Baidu.
EMNLP-IJCNLP 2019 Announces Best Papers
EMNLP 2019 | Best Paper: Specializing Word Embeddings (for Parsing) by Information Bottleneck
EMNLP 2019 | Best Paper Runner Up: Designing and Interpreting Probes with Control Tasks
EMNLP 2019 | Best Resource Paper: The FLORES Evaluation Datasets for Low-Resource Machine Translation: Nepali–English and Sinhala–English / GitHub
EMNLP 2019 | Best Demo: AllenNLP Interpret: A Framework for Explaining Predictions of NLP Models / Click here to visit the project
The Measure of Intelligence
Researchers propose a set of guidelines for what a general AI benchmark should look like. A new benchmark closely following these guidelines, the Abstraction and Reasoning Corpus (ARC), is built upon an explicit set of priors designed to be as close as possible to innate human priors.
High Fidelity Video Prediction with Large Stochastic Recurrent Neural Networks
In this work, researchers question whether handcrafted architectures are necessary and propose a different approach: finding minimal inductive bias for video prediction while maximizing network capacity. They investigate this question by performing the first large-scale empirical study and demonstrate state-of-the-art performance
(University of Michigan & Google Research & Google & Adobe Research)
Dancing to Music
In this paper researchers propose a synthesis-by-analysis learning framework to generate dance from music. In the analysis phase, they decompose a dance into a series of basic dance units, through which the model learns how to move.
(University of California, Merced & NVIDIA)
You May Also Like
Do Deep Neural Networks ‘See’ Faces Like Brains Do?
A new paper from the Weizmann Institute of Science proposes that the face-space geometries that form the foundation for facial recognition in SOTA deep convolutional neural networks (DCNNs) have already evolved to the point that they share pattern similarities with the face-space geometries of cortical face-selective neuronal groups in humans.
Google T5 Explores the Limits of Transfer Learning
Google researchers recently introduced a novel “Text-to-Text Transfer Transformer” (T5) neural network model which can convert any language problem into a text-to-text format. Synced invited Samuel R. Bowman, an Assistant Professor at New York University who works on artificial neural network models for natural language understanding, to share his thoughts on the T5 framework.
Global AI Events
November 13-14: AI & Big Data Expo in Santa Clara, United States
December 2-6: AWS re:Invent 2019 in Las Vegas, United States
December 8-14: 2019 Conference on Neural Information Processing Systems (NeurIPS 2019) in Vancouver, Canada
January 7-10: CES 2020 in Las Vegas, United States