this post was submitted on 29 Nov 2023
1 points (100.0% liked)

Machine Learning

1 readers
1 users here now

Community Rules:

founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] Zondartul@alien.top 1 points 11 months ago (12 children)

The point of the paper is that LLMs memorize an insane amount of training data and, with some massaging, can be made to output it verbatim. If that training data has PII (personally identifiable information), you're in trouble.

Another big takeaway is that training for more epochs leads to more memorization.

[–] Mandelmus100@alien.top 1 points 11 months ago (7 children)

Another big takeaway is that training for more epochs leads to more memorization.

Should be expected. It's overfitting.

[–] we_are_mammals@alien.top 1 points 11 months ago (1 children)

It's overfitting.

Overfitting, by definition, happens when your generalization error goes up.

[–] DigThatData@alien.top 1 points 11 months ago (2 children)

it's possible to "overfit" to a subset of the data. generalization error going up is a symptom of "overfitting" to the entire dataset. memorization is functionally equivalent to locally overfitting, i.e. generalization error going up in a specific neighborhood of the data. you can have a global reduction in generalization error while also having neighborhoods where generalization gets worse.

[–] seraphius@alien.top 1 points 11 months ago

On most tasks, memorization would be overfitting, but I think one would see that “overfitting” is task/generalization dependent. As long as accurate predictions are being made for new data, it doesn’t matter that it can cough up the old.

[–] Hostilis_@alien.top 1 points 11 months ago

Memorization is functionally equivalent to locally overfitting.

Uh, no it is not. Memorization and overfitting are not the same thing. You are certainly capable of memorizing things without degrading your generalization performance (I hope).

load more comments (5 replies)
load more comments (9 replies)