TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Pokemon GAN

77 点作者 aliabd超过 3 年前

8 条评论

catillac超过 3 年前
I am a big fan of hugging face, but whenever these examples come up, they always seem to be half baked and don’t quite work well, and are not compelling. I think it would behoove them to polish these a little more, in my opinion.
评论 #30337983 未加载
评论 #30336661 未加载
评论 #30340189 未加载
评论 #30336938 未加载
aliabd超过 3 年前
Just rebuilt the app with more examples, so the queue should stay a lot shorter now. This was built using Gradio[0] and is hosted on Spaces[1]. Check out the paper[2] and github repo[3].<p>[0]: <a href="https:&#x2F;&#x2F;github.com&#x2F;gradio-app&#x2F;gradio" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;gradio-app&#x2F;gradio</a> [1]: <a href="https:&#x2F;&#x2F;huggingface.co&#x2F;spaces" rel="nofollow">https:&#x2F;&#x2F;huggingface.co&#x2F;spaces</a> [2]: <a href="http:&#x2F;&#x2F;www.cvlibs.net&#x2F;publications&#x2F;Sauer2021NEURIPS.pdf" rel="nofollow">http:&#x2F;&#x2F;www.cvlibs.net&#x2F;publications&#x2F;Sauer2021NEURIPS.pdf</a> [3]: <a href="https:&#x2F;&#x2F;github.com&#x2F;autonomousvision&#x2F;projected_gan" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;autonomousvision&#x2F;projected_gan</a>
评论 #30336526 未加载
minimaxir超过 3 年前
For context, this is not the same technique that was used to create the AI-generated Pokemon a couple months ago: <a href="https:&#x2F;&#x2F;www.reddit.com&#x2F;r&#x2F;pokemon&#x2F;comments&#x2F;rgmyxp&#x2F;i_trained_an_ai_on_all_the_official_pokemon&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.reddit.com&#x2F;r&#x2F;pokemon&#x2F;comments&#x2F;rgmyxp&#x2F;i_trained_a...</a><p>That used a finetuned ruDALL-E which is not based on a GAN architecture and is much, much slower than a GAN, albeit the generated Pokemon are more coherent. You can play with it (with no queue) on your own Colab Notebook here: <a href="https:&#x2F;&#x2F;colab.research.google.com&#x2F;drive&#x2F;1A3t2gQofQGeXo5z1BAr1zqYaqVg3czKd?usp=sharing" rel="nofollow">https:&#x2F;&#x2F;colab.research.google.com&#x2F;drive&#x2F;1A3t2gQofQGeXo5z1BAr...</a>
评论 #30337799 未加载
评论 #30339895 未加载
arcticbull超过 3 年前
These are wonderfully low-effort and I love each and every one of them.
_v7gu超过 3 年前
Some examples look like they have picked two pokemon and tried to merge them in a very weird way. I&#x27;ll list what I see so others can compare:<p>- 0 as Arceus + Shaymin S<p>- 1 as Lickilicky + Espurr<p>- 10 as Kirlia + Porygon 2, don&#x27;t know where the colors are from<p>- 20 as Shellos + Froslass<p>- 30 as Roggenrolla&#x2F;Oddish&#x2F;Poliwag<p>- 42 as Garchomp + Trevenant<p>- I see the Kyurem in 60 but no idea where the tail is from<p>- 102 as Dragonite + Hitmonlee
评论 #30336918 未加载
评论 #30340288 未加载
aliabd超过 3 年前
Just updated the link again - much faster now (inference down to 0.5s from 9s) Queue will move a lot quicker now.
chrisma0超过 3 年前
This makes me want to build a Twitter Bot &quot;Cursed Pocket Monsters&quot;...
MauroIksem超过 3 年前
How long before game freak uses something like this to create new Pokemon?