4 ms·
$6,000 for the Gigaword dataset they used to train... https://catalog.ldc.upenn.edu/LDC2012T21 https://catalog.ldc.upenn.edu/LDC2012T21
by trengrj 10y ago
$6,000 for the Gigaword dataset they used to train...
https://catalog.ldc.upenn.edu/LDC2012T21 https://catalog.ldc.upenn.edu/LDC2012T21
- emmab 10y agoThere's not even a way to buy it as an individual who's not part of an organization! They don't even state this as a possibility.
- schmit 10y agoFunny how Google is paying for data
- peterjliu 10y agoThe code can be used to train on other data. All you really need is a collection of news articles. I think there are some free ones available. This dataset was only used to benchmark against other published results. It was first proposed in https://arxiv.org/abs/1509.00685 https://arxiv.org/abs/1509.00685.