4 ms·
I welcome forward progress in the field, but something about this doesn't sit right with me. The authors have an unpublished/unreviewed set of results and they'
by nharada 6y ago
I welcome forward progress in the field, but something about this doesn't sit right with me. The authors have an unpublished/unreviewed set of results and they're already co-opting the YOLO name (without the original author) for it and all of this to promote a company? I guess this was inevitable when there's so much money in ML but it definitely feels against the spirit of the academic research community that they're building upon.
- syntaxing 6y agoTotally agreed, kinda seems dirty to call something "v5" when it this is a derivative work of the original.
- oehtXRwMkIs 6y agoI think derivative is a bit generous. This is just a reimplementation of v4 with a different framework.
- salty_biscuits 6y agoWell, very unlikely to get the original author. He doesn't do that kind of thing anymore https://twitter.com/pjreddie/status/1230524770350817280?s=19 https://twitter.com/pjreddie/status/1230524770350817280?s=19
- sicariusnoctis 6y ago> there's so much money in ML What do you mean? I thought the DL hypetrain was dying as companies failed to make returns on their investments.