Jan Jongboom
Jan Jongboom
I have an example program here: https://github.com/janjongboom/utensor-misclassifies The first classification is completely wrong... I added a second one to do sanity checking and that one seems to be OK. TF...
Running on DISCO-L475VG-IOT01A development board (Cortex-M4F) a quantized network is much slower than a non-quantized one. Simple MLP 33x20x10x5 with ReLu activation between the dense layers and SoftMax for output...
I have a simple neural network with two hidden layers. Every time I classify through uTensor I'm leaking ~250 bytes of memory. Here's how to log memory: 1. Set the...
Given the current bug around Softmax (https://github.com/uTensor/uTensor/issues/170) I figured I'd try Sigmoid as my activation function. Unfortunately it's not supported: ``` unsupported op type in uTensor: Sigmoid ``` Is there...
The first layer of my model is: 186 input nodes -> 150 hidden nodes. uTensor allocates 111,600 bytes for this. This should not be necessary. My weights are already distributed...
^ Should not be needed. @neil-tan @mbartling
uTensor always assumes that file system operations succeed, and will not throw helpful errors when they fail. E.g. fread on the header will just throw magic number invalid if fread...
To automatically set up all settings for pyOCD and GDB in projects exported from ARM mbed it'd be great if `gdb.client.executable` would be exposed as an Eclipse variable, in a...
@mvertes This upgrades node-parquet to Node.js v12 and Nan v2.14.0. Verified that all tests pass against node 12.13.0 on macOS. There are still deprecation warnings around the `toBoolean()` calls but...