Comments (8)
I noticed similar pattern, but it only happens in the first few (<3~5) epochs and is not as obvious as you described. Below is an example of the loss, with x axis being the # of epochs:
from cmc.
Thanks for your kindly response.
from cmc.
Hi, thanks for sharing the code.
I've got the same issue, the loss is decay durning the epoch but reversed to the same point in the beginning of the next epoch. did you fin a solution?
from cmc.
I also got the same issue. I guess the reason leads to this situation might be that the features stored in the memory bank come from the previous epoch and leads to high loss when new features come in. But if you average the loss of each epoch, it still decays.
from cmc.
I also got the same issue, which has also been discussed in #27. Yes, the average loss still decays, but I find this may hurt the performance of CMC on small datasets.
from cmc.
I also got the same issue, which has also been discussed in #27. Yes, the average loss still decays, but I find this may hurt the performance of CMC on small datasets.
I have the same experiment.
Tuning the hyperparameters: nce_k, nce_m and learning rate improved the issue in my experiments
from cmc.
I also got the same issue, which has also been discussed in #27. Yes, the average loss still decays, but I find this may hurt the performance of CMC on small datasets.
I have the same experiment.
Tuning the hyperparameters: nce_k, nce_m and learning rate improved the issue in my experiments
Yes, I also find decreasing nce_k can improve the performance. Could you please share some experience on tuning nce_m and lr?
from cmc.
decreasing nce_m helped in some case to make the training more stable, meaning that the loss start follow the loss of the last epoch sooner.
For the lr, i played with it until i get the right shape of the loss curve. The same as:
from cmc.
Related Issues (20)
- Weight shape in Linear classifier HOT 1
- Train/Test split when training linear classifier (ImageNet) HOT 3
- Gpu memory heavy on gpu0 HOT 1
- mean std value for LAB/YCbCr normalization HOT 1
- How to prevent positive sample is sampled by AliasMethod? HOT 1
- Format of dataset HOT 2
- index drawn by AliasMethod is not on the same gpu as the model HOT 1
- About the training epochs for ImageNet-1K
- question about shuffle_ids
- Training Loss Doesn't Improve for CIFAR10
- pretrained model of ResNet50v1
- Feature Tensor Shape Question HOT 1
- how to use the pretrained MoCo
- Protential memory leak when training?
- Modified AlexNet architecture for STL-10 dataset
- question about input image
- Questions about NCEAverage.py
- How can it be used for more than two views datasets?
- Pre-trained model for Video task HOT 1
- CMC for small custom 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 cmc.