[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

2022-05-12: Ghost posting is now globally disabled. 2022: Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/sci/ - Science & Math

View post   
View page     

[ Toggle deleted replies ]
File: 1.64 MB, 1316x1152, craiyon.png [View same] [iqdb] [saucenao] [google]
14732710 No.14732710 [Reply] [Original] [archived.moe]

why is the 0.4 billion parameter craiyon able to do a fairly decent job? Note the parti-images were built up using some sort of in-paint feature whereas craiyon was all in one prompt.

>> No.14732771

Google engies are probably retarded or are doing a purely scaling approach.
Craiyon might be using heuristics on top of scaling.

>> No.14732788

Now ask it to do human faces, or use negation statements.

>> No.14732915
File: 688 KB, 598x608, craiyon2.png [View same] [iqdb] [saucenao] [google]

not bad actually 5/10 for Craiyon.

>> No.14732919

That banana melting into the board is so dalliesque

>> No.14733012

I mean 5/9
Yeah those are interesting to me. I recall doing something similar when I was a 3. When I would get anything wrong I'd draw over it.
If you look the bowl is yellow so somewhere in the steps to create it the banana ended up there and not there at the same time.

>> No.14733030

Parameter count is not a useful heuristic for comparing models with different architecture. Given a particular network design, scaling parameter count seems to pretty reliably improve performance.

Delete posts
Password [?]Password used for file deletion.