Ya Ge
Ya Ge
Looks like the input image was not loaded successfully. Could you double check the image path?
@jkcarney did you have any findings? I'm asking because I have exactly the same observation. I was using RTX A4000 without xformers: * Single GPU speed is 2.62it/s, which is...
Probably you already know about this, but a potential get around is to specify the config file path in the command line, and specify the index folder there. Something like...
谢谢小明！我有个顾虑是这样全局的mask file不是太实用。一方面浪费手机的显示空间，一方面给所有的群用同一个mask也没有必要。我更倾向用一个dict来表示mask，每个群用不同的mask，比如知乎群用刘看山logo，温兆伦群用老赵近照等等。你看如何？ 一些低优先级的想法： * 可以加入随机增加趣味性，比如每10张tagcloud出一张有mask的。 * 对于`/mytag`可以用用户头像来做mask。
Sorry for the late reply. Somehow github unwatched myself... I just fixed the download link. It's still https://lab.grapeot.me/head.weights.
Sure. I just uploaded it to Cow Transfer: https://yage.cowtransfer.com/s/b4b9851c0ce949 click link to check [ faceimgs ] or access cowtransfer.com and input extract code: 6fe3e5 to check;
应该是这个原因：`ERROR : We need at least 1 word to plot a word cloud, got 0.` 在群里说几句话应该就好了~
如果是这样的话，感觉每次造tag cloud的函数都被调用了一次。。这个比较奇怪。。要不跟踪debug看一下？
关于斗图的那个问题，我开了个新issue哈。https://github.com/grapeot/WechatForwardBot/issues/5
如果还是一直抛的话，感觉要自己debug一下了。。我没有遇到过这样的情况。。