-
Review major architectures and techniques that are hardware friendly:
- quantization
- binarization (full and partial)
- XNOR networks
**DOD**
- [x] chose network to implement on Zybo-Z7-20
-
### 🚀 The feature
Add quantization support for `BackboneWithFPN`.
### Motivation, pitch
Currently, it is possible to use `from torchvision.models.detection.backbone_utils.resnet_fpn_backbone/Backbo…
-
* [deep compression](https://arxiv.org/pdf/1605.07678.pdf): pruning and quantization , 35x reduction
* [squeezenet](https://arxiv.org/pdf/1602.07360.pdf): its [openreview](https://openreview.net/foru…
-
Dear @AlexeyAB ,
I design a network and train it on my custom dataset by using darknet framework.
My network weight is about 280MB. I want to use Model compression technology
(Model Pruning, K…
-
**Describe the bug**
Hello, I want to do full 8-bit quantization(input, weight all 8-bit) to the network with a bilinear upsampling layer. The fake QAT result in validation set is closed to FP32 re…
-
I tried out the MNIST classification [example](https://github.com/parimalp/xup-vitis-ai-tutorial/blob/main/docs/pt_mnist.md) using PyTorch in an AWS F1 instance. The quantization accuracy of the netwo…
-
Question for you guys: as best I can tell, there is no support at present for keeping activations in fp8 between the "output" matmul (of either an attention block or MLP block) and the next norm (laye…
-
Hi,
I had a question about the optimal weighting of the losses during training.
I notice that initially if you weigh the fc loss, vq_loss, and reconstruction loss equally, the quantizer is not t…
-
Problem: The current implementation of [GPTQ](https://arxiv.org/abs/2210.17323) (A technique used to improve quantization accuracy) relies on [model tracing](https://github.com/pytorch/ao/blob/main/to…
-
```
import paddle.fluid as fluid
from pyramidbox_test import PyramidBox
from paddle.fluid.framework import IrGraph
from paddle.fluid import core
from paddle.fluid.contrib.slim.quantization.quanti…