117 点作者 ludovicianul大约 1 年前

5 条评论

gmerc大约 1 年前

I like how they use stealing .... while maintaining steadfast that it's fair use when they use other people's data to train the model.

评论 #39678945 未加载

评论 #39681112 未加载

评论 #39679149 未加载

tellarin大约 1 年前

Discussion: <a href="https://news.ycombinator.com/item?id=39675735">https://news.ycombinator.com/item?id=39675735</a>

cuu508大约 1 年前

Better link: <a href="https://arxiv.org/abs/2403.06634" rel="nofollow">https://arxiv.org/abs/2403.06634</a>

nippoo大约 1 年前

I'm so curious what an example set of queries actually looks like here. Like - what were the actual prompts used!

fragmede大约 1 年前

[dupe], discussed previously at <a href="https://news.ycombinator.com/item?id=39675735">https://news.ycombinator.com/item?id=39675735</a><p>The twitter post adds no new information and poorly links to the same paper.

Google Announces Stealing Part of a Production Language Model

5 条评论

Google Announces Stealing Part of a Production Language Model

5 条评论