-
First and foremost, I'd like to thank for your great work. It's truly impressive!
By the way, I have a question about the MRD module. When I looked into the implementation, it appears that the larg…
-
These might be useful augmentations in the audio domain: (see "An Ensemble of Convolutional Neural Networks for Audio Classification" https://www.mdpi.com/2076-3417/11/13/5796/htm section 4.2)
- [ …
-
### Problem Description
Many problems; not just one. But I'm not talking about "dysfunctions", exactly; just at stuff that makes no sense. When stuff does not make any sense, it kills immersive…
-
I am trying to save a quantized ternary model to a `.tflite` file, but larq doesn't seem to save the weights using datatypes with a reduced precision and thus compress the file size.
However, after c…
-
`torchaudio` is an extension library for PyTorch, designed to facilitate audio processing using the same PyTorch paradigms familiar to users of its tensor library. It provides powerful tools for audio…
-
## Quantization Method for conv, deconv and fc Layers.
Here I want to implement the quanzization on operation in conv, deconv and fc layers. Much quantization method are included in this paper: Ristr…
-
因为 task-specific data 有限，难蒸馏出好 student model，所以用 mixup 增强数据，teacher 对插值的输入预测，student 学习该预测结果(loss_3，带调节系数超参)，直接学插值输入对应的插值标签(loss_2，带调节系数超参)，学习原始数据(loss_1，无系数)。提供了一些理论证明，在GLUE上做了实验有效。
 deformers (many from audio degradation toolbox):
- [ ] Low-, band-, high-pass filter
- [ ] bit-crushing
- [ ] companding+quantization noise
- should wait until https://github.com/li…