Eric Hunsberger

My impression is that there's only two things in our models that really store state information over time: synapse values and neuron states. Assuming LIF neurons, we have two states...

Yes, I meant that making the synapse state a signal is the first step. On Jan 6, 2017 4:12 PM, "Jan Gosmann" wrote: Unfortunately, the synapse state is not stored...

We used to treat small `tau` in `Lowpass` and `Alpha` as zero, when we used to use a worse `expm` function. We removed this in 4ea0b0549553134d511e696faf83d5d763f55448. It would be nice...

I don't think we want to install them as part of the python package. Then they'd be in some out-of-the-way folder like /usr/lib/python2.7/dist-packages/nengo/examples or something. I'd install them as data...

My proposed solution is to simply not validate defaults. We will validate them when using them to create an object. This will still happen early (i.e. before the build process),...

The first part of this is fixed with the `coerce_defaults` attribute on Parameters. Setting a default for `Connection.function` still doesn't work because we use `function_info` internally, as stated above. I...

Somewhat related to #1026/#1028 (allowing neuron time constants to be heterogeneous).

One option would be to have a base tolerance (or pair of tolerances) that's passed into the test as a parameter. Then the test tolerances can be derived from that,...

You can also use the `keep_history` option in `nengo_dl.configure_settings` to have probes just store the last output (you can further configure this for individual probes; see the docs). If you...

Hi! The main point of this project is to provide a Python wrapper on the KITTI dataset, so that it's easy to load KITTI images, odometry data, velodyne points, etc....