I try the tensorflow example, and i have some issues / or things that i don't understand.
First i use your csv files (i m talking about robmarkcole files); it generates very beautiful graphics for errors, etc... but when I change the seed, graphics (so errors evaluation...) are not so nice, and very different. i m ok for different, but ...
So after i try to generate my own csv files for punch and flex gestures; the results are very bad (loss).
For the classification part; the training model with your files is not working very well, with my files it's worse.
Maybe there's something i miss somewhere.
It don't where to exchange about this, i try with you
I already push that issue here ==> https://github.com/robmarkcole/arduino-tensorflow-example/issues/1
Hello,
I try the tensorflow example, and i have some issues / or things that i don't understand.
First i use your csv files (i m talking about robmarkcole files); it generates very beautiful graphics for errors, etc... but when I change the seed, graphics (so errors evaluation...) are not so nice, and very different. i m ok for different, but ...
So after i try to generate my own csv files for punch and flex gestures; the results are very bad (loss).
For the classification part; the training model with your files is not working very well, with my files it's worse.
Maybe there's something i miss somewhere.
It don't where to exchange about this, i try with you
Thx