https://github.com/JamesSand/UsefulCommands/blob/master/08-Research-utils/qiang-ideas/deepconf.md
I came across this from DeepConf issue section. Specifically, you've mentioned that we actually wanted the q to be sharper. This means the first item in the decomposed equation would have lower values (rather than higher values). So the problem lingers: does the author actually confuses certainty and uncertainty or the interpretation leads somewhere else.
Thanks in advance. This is a great repo!
https://github.com/JamesSand/UsefulCommands/blob/master/08-Research-utils/qiang-ideas/deepconf.md
I came across this from DeepConf issue section. Specifically, you've mentioned that we actually wanted the q to be sharper. This means the first item in the decomposed equation would have lower values (rather than higher values). So the problem lingers: does the author actually confuses certainty and uncertainty or the interpretation leads somewhere else.
Thanks in advance. This is a great repo!