-
**System information**
- Have I written custom code (as opposed to using a stock example script provided in TensorFlow):
yes
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04): ubuntu 16…
ricvo updated
4 years ago
-
What is the architecture of the face_landmark_68_model. I wanted to train my own landmark detection model with lesser points. please let me know the model architecture.
-
ENV:
Tensorflow 1.14, CUDA10.0, cuDNN7.5.0,
After compiled the ucudnn with Tensorflow 1.14, I tested the performance, the test model is InceptionV3, and I noticed the performance is about 10%~20%…
-
As far as I can see, you use something in the lines of 3-layer LSTM. Did you try other networks architectures? I am asking, because we are planning to train Tacotron2 + WaveGlow for 30-40 Russian spea…
-
Is there any suggested way to perform convolutions having input and output as q15_t and weights as q7_t? (excluding of course converting weights to q15_t and then use `arm_convolve_HWC_q15_*`, which w…
-
Hi guys, could you make this library work in a browser environment? I think Node.js dependencies should be easy to get rid of.
Then, you could make an online demo, so we could compare your library …
-
I am seeing strange behavior running tensor flow with MKL-DNN on a KNL CPU. There are a number of strange results.
Initially, I was testing single node performance for a model that is relatively l…
-
Thank you for your work and making it public!
My question relates to the SparseTensor. In the provided example you wrote
`
sinput = ME.SparseTensor(features - 0.5, coords=coordinates).to(device…
MacOS updated
5 years ago
-
Your samples at epoch 3200 have strange noises at unvoiced segments, while there is no such phenomenon in samples at epoch 1600.
As the paper setting, the keypoint is to define variant kernel size a…