zlacker

[return to "Imagen, a text-to-image diffusion model"]
1. jandre+M6[view] [source] 2022-05-23 21:29:22
>>kevema+(OP)
Is there a way to try this out? DALL-E2 also had amazing demos but the limitations became apparent once real people had a chance to run their own queries.
◧◩
2. wmfrov+D8[view] [source] 2022-05-23 21:38:59
>>jandre+M6
Looks like no, "The potential risks of misuse raise concerns regarding responsible open-sourcing of code and demos. At this time we have decided not to release code or a public demo. In future work we will explore a framework for responsible externalization that balances the value of external auditing with the risks of unrestricted open-access."
◧◩◪
3. nomel+j9[view] [source] 2022-05-23 21:43:09
>>wmfrov+D8
> the risks of unrestricted open-access

What exactly is the risk?

◧◩◪◨
4. varenc+cg[view] [source] 2022-05-23 22:22:17
>>nomel+j9
See section 6 titled “Conclusions, Limitations and Societal Impact” in the research paper: https://gweb-research-imagen.appspot.com/paper.pdf

One quote:

> “On the other hand, generative methods can be leveraged for malicious purposes, including harassment and misinformation spread [20], and raise many concerns regarding social and cultural exclusion and bias [67, 62, 68]”

◧◩◪◨⬒
5. userbi+di[view] [source] 2022-05-23 22:34:33
>>varenc+cg
But do we trust that those who do have access won't be using it for "malicious purposes" (which they might not think is malicious, but perhaps it is to those who don't have access)?
[go to top]