Comments (3)
@learningman7777 Thanks for the report.
I just cross-checked with the original repository.
As I committed new code now, it is now producing same transformed images as author's code. Zero level should be replaced to 1.
But, as you said, transformation is less when level for posterize is bigger. I think randaugment's all augmentations should be harder when we increase level.
@BarretZoph , Can you check on this? Below images are generated using your code and levels are 0, 1, 2, 3, and 4, respectively. I expected orders of below images are reversed.
level=0
level=1
level=2
level=3
level=4
Reference
- https://github.com/tensorflow/tpu/blob/a0d7a57a782d76d3c3c1343c06dbb148cc2e1be1/models/official/efficientnet/autoaugment.py#L513
- https://github.com/tensorflow/tpu/blob/a0d7a57a782d76d3c3c1343c06dbb148cc2e1be1/models/official/efficientnet/autoaugment.py#L225
from pytorch-randaugment.
@learningman7777 I found the author's response on this. If you have any further questions and think this resolve the problem, please close this issue.
from pytorch-randaugment.
thanks!
from pytorch-randaugment.
Related Issues (20)
- Reproducing CIFAR-10 results HOT 4
- Reproducing CIFAR-100 results HOT 1
- Questions about pyramidNet
- Small remark on --cv-ratio (i.e. validation set)
- The order of augmentation
- Why do you use M scale [0, 30] instead of [0, 10]?
- CIFAR10(0) w/ batch size 128 HOT 1
- The prob of applying augmentation? HOT 1
- Question about the modified operation types
- Can not reproduce the result (both paper & yours) HOT 3
- Brightness, Sharpness, Constrast, Color at magnitude 0 should have v=1?
- view size is not compatible with input tensor's size and stride HOT 1
- TypeError: 'NoneType' object is not callable
- RandAugment hyperparameters for markdown tables
- How to change the architecture?
- SmoothCrossEntropyLoss
- Can I use this project to generate a strategy for my own dataset? HOT 2
- Unable to reproduce Cifar-10 results for WRN-28-10 HOT 1
- It seems that you dont use Linear Evaluation Protocol?
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from pytorch-randaugment.