Carroll Vance

I had the same problem, here is the code snippet I used to get the position data back: ``` heading = MadgwickAHRS() while True: ninedofxyz = self._ninedof.read() heading.update(ninedofxyz[0], ninedofxyz[2], ninedofxyz[1])...

My problem was with how I was handling the gyroscope data from the IMU I was using. Everything was extremely unstable because (1) my unit conversion was not working as...

Hi, thank you very much for implementing this, it seems to work very well in full precision mode. However, I do get some issues with numerical stability when used automatic...

1. While the actual involution unfold / multiply does not work with mixed precision, the rest of the autocastable ops do. So if you cast to fp32 before calling the...

It explodes even in forward pass because the activation values tend to be much larger when using weight standardization. This is because the weights are normalized to std=1 instead of...

Just to summarize my thoughts overall using this method: 1. People are used to training nets with He initialization but weight standardization activation scales are quite higher than those initialized...

https://github.com/csvance/roboclaw Just got SLAM working using a node I wrote in C++. It doesn't have a ton of bells and whistles yet, but Odometry and differential drive works with move_base...

Question, in your discord.py configuration file have you set DISCORD_LEARN_FROM_ALL to True? Try that, then restart the bot, and it should start storing data from discord for training the next...

The discord connector does as well, the exception needs to be caught and the connections restarted.

What version of python are you using? I have tested armchair-expert with python 3.5 and 3.6. Several people have got it working on Windows as well. I recommend running armchair-expert...