undefined | Better HN

0 pointsqsera2mo ago0 comments

>using synthetic training data

yea, look up how it is done.

It is exactly how a perpetual motion machine scam would project an appearance of working like using a generator to drive a motor, and the motor driving the generator..something that would obscure the fact that there is energy loss happening along the way....

0 comments

3 comments · 1 top-level

WarmWash2mo ago· 2 in thread

I'm confused with the point you are trying to make, because they are using synthetic data, and the models are getting stronger.

There is no "conservation of fallacy" law (bad data must conserve it's level of bad), so I'm struggling to connect the dots on the analogy, unless I ignore the fact that training on synthetic data works, is being used, and the models are getting better.

qseraOP2mo ago

If the training that did not use synthetic data failed to capture some aspect of the information contained, then using data synthesized from the original data could help to capture it, thus it could result in the models getting better.

But that is because the synthetic data helped the model capture what was already there in the training data.

But after all such information has been extracted, then it would not be possible to use synthetic data or anything that is derived from the original data to create "new" information for training....

dgb232mo ago

Better by which metrics?

j / k navigate · click thread line to collapse