Comments (5)
Updated results on VisDA17 classification (70.11%):
Difference is that previous result was obtained running the code under PyTorch 0.4, this one is using PyTorch 0.3. And running evaluation in either 0.3 or 0.4 gives the same accuracy. Suspect that behavior of some function(s) changed dramastically from 0.3 to 0.4.
Thanks to the author for pointing out that:)
from mcd_da.
Hi, @Ellenisawake
do you have any idea what makes this performance difference betewwn pytorch 0.3 and 0.4?
from mcd_da.
Hi, @Ellenisawake
do you use the code the author applied with VisDa classification directly? I couldn't understand the variable of 'val' in the test function, can you tell me your understand?
from mcd_da.
Hi, @Ellenisawake
do you have any idea what makes this performance difference betewwn pytorch 0.3 and 0.4?
I didn't check in detail. Suggest that you just use 0.3 for running it or adding anything on top. As PyTorch has already evolved to v1.0, personally don't think there's much significance in figuring out version conflicts between 0.3 & 0.4.
from mcd_da.
Hi, @Ellenisawake
do you use the code the author applied with VisDa classification directly? I couldn't understand the variable of 'val' in the test function, can you tell me your understand?
I think 'val' is just a flag used to distinguish a test run and a validation run within training.
'''
if not val and value > 60:
torch.save(F1.state_dict(), save_path+''+args.resnet+''+str(value)+'_'+'F1.pth')
'''
Here the author only wants the model saved when in validation and accuracy >60.
from mcd_da.
Related Issues (20)
- Loss is becoming negative HOT 3
- About datasets on classification HOT 3
- About the classifiers. HOT 1
- When do segmentation task, you also need labels for target domain.
- Code for " Strong-Weak Distribution Alignment for Adaptive Object Detection " HOT 1
- Question about the handwritten digit experiment HOT 7
- cityscapes/info.json
- Some problems with building my own datasets on classificaion HOT 1
- question for location HOT 2
- Asking for the visualization code
- Reproduce the segmentation result HOT 3
- sharing svhn2mnist result
- Model Selection
- USPS->Mnist source-only result can't reach 0.634 HOT 2
- The misnt_data.mat dataset
- drn_d_105-12b40979.pth HOT 3
- The link (http://crcv.ucf.edu/data/adaptationseg/ICCV_dataset.zip) is lost.
- The difference between trainer and trainer_one_step
- synth traffic dataset
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 mcd_da.