[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/ic/ - Artwork/Critique

Search:


View post   

>> No.6449681 [View]
File: 580 KB, 1014x539, 1671288049800900.png [View same] [iqdb] [saucenao] [google]
6449681

>>6449677
the database of images was thrown away like the peel of a banana but the essence is still there, and it's not a smooth mixture and can result in images that were put in spit back out almost exactly the same

>> No.6424162 [View]
File: 580 KB, 1014x539, 1671214127969466.png [View same] [iqdb] [saucenao] [google]
6424162

>> No.6424009 [View]
File: 580 KB, 1014x539, 1671214127969466.png [View same] [iqdb] [saucenao] [google]
6424009

>> No.6423984 [View]
File: 580 KB, 1014x539, 1671214127969466.png [View same] [iqdb] [saucenao] [google]
6423984

>>6423945

>> No.6422073 [View]
File: 580 KB, 1014x539, 391df1acca93eadda4347173ad5d11ea.png [View same] [iqdb] [saucenao] [google]
6422073

>"The goal of this study was to evaluate whether diffusion models are capable of reproducing high-fidelity content from their training data, and we find that they are. While typical images from large-scale models do not appear to contain copied content that was detectable using our feature extractors, copies do appear to occur often enough that their presence cannot be safely ignored;"

>"The dataset that we searched in our study is a small subset of this fine-tuning data, comprising less than 0.6% of the total training data. [...] Furthermore, it is highly likely that replication exists that our retrieval method is unable to identify. For both of these reasons, the results here systematically underestimate the amount of replication in Stable Diffusion and other models."

>https://arxiv.org/abs/2212.03860

AI sisters...I don't feel so good...

Navigation
View posts[+24][+48][+96]