tensorflow / tensorflow/models
Mobilenetv3 quantitative model accuracy reproduction
Open
@marksandler2 is already working on this.
Since Jun 12, 2020.
models:research
type:docs
- Dominant language
- Python
- Stars
- 77.7k
- Forks
- 44.8k
- PR merge metrics
- No merged PRs in 30d
Description
I’m trying to reproduce the Mobilenetv3 quantization process, using mobilenetv3 large dm=1(top1: 75.2% in imagenet 2012). My int8 quantitative model top1 accuracy is only 71%.
How can I reproduce the int8 model accuracy in the document, which top1 accuracy is 73.9%.
Can you provide detailed reproduction steps, such as learning rate and batchsize parameters.
Do I need to skip some layers and not quantify them?
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.