EdgeML

EdgeML copied to clipboard

Phoneme Detection and Classifier Model Codes

2

AnirudhBHarish

Conv1d and batchnorm1d layers

5

AnirudhBHarish

port code to tensorflow2.0

6

Compile and run code with tf2.0. Eager execution is disabled. Tested code with python 3.5 Signed-off-by: Punit Vara

punitvara

Added Heirarchical prior.

3

Added path smoothing to prevent overfitting for deeper trees whilst improving accuracy. Added mse loss for regression on weighted layers. References: 1. [Path smoothing](https://ojs.uma.ac.id/index.php/jite/article/view/3856) 2. [IHT Gradient descent](https://arxiv.org/pdf/1910.13389.pdf)

meetdoshi90

KWS-demo throwing errors on running FastGRNNCUDA inference on google speech commands dataset

1

Error message:- `Exception in thread Thread-1: Traceback (most recent call last): File "/anaconda/envs/spr/lib/python3.6/threading.py", line 916, in _bootstrap_inner self.run() File "kws-demo.py", line 151, in run logits = fastgrnn(torch.FloatTensor(features)) File "/anaconda/envs/spr/lib/python3.6/site-packages/torch/nn/modules/module.py", line...

IannoIITR

FastCell Example Fixes, Generalized trainer for both batch_first args

4

Checked the changes with Aditya

SachinG007

SeeDot: Arduino Accuracy or Prediction Mode & Documentation

2

Is there any documentation on using the SeeDot generated Arduino sketch? The [README](https://github.com/microsoft/EdgeML/blob/master/tools/SeeDot/seedot/arduino/README.md) is empty. I'm having issues specifically using Accuracy mode so I can test a whole data set...

BrianEubanks

harsha-simhadri

This solves the issue which the visitor sghanta had when @harsha you were here in India in Jan. Files are under Apache license so shouldn't be an issue.

SachinG007

Xilinx HLS tool

4

In the paper, there is a comparison between SeeDot-generated c code and Vivado HLS c code. SeeDot can automatically generate hints(e.g. pragma HLS unroll, pipeline...) for HLS tools. How to...

Anjiang-Wei