Comments (2)
The answer is a bit tricky.
some initial assumptions:
- transformed are harder to train than conv nets, and are more sensitive to hyperparameters (including lr)
- there is a major difference in the optimal learning rate when you train a network from scratch, compared to when you fine-tune it from a previous pretraining.
- if you don't have a lots (!) of GPU resources, hard to do hyperparameters tuning and searching on imagenet 21K.
given 1-3, the scheme I suggested in the article is to initially train a net on imagenet1K, and then transfer it to imagenet21K for 80 epochs. Since this is transfer learning, I used regular adam with lr=3e-4. i believe that this scheme is quite robust, on TResNet-m i trained with a large batch (4168). with ViT the batch was smaller (648).
you can try other learning rates, I think a reasonable range for adam optimizer is 2e-4 to 5e-4.
DeiT paper talks about training from scratch, which is different. notice that they suggest 1000 epochs !
from imagenet21k.
Thank you for the fast response! I'll try with the lr you suggested.
from imagenet21k.
Related Issues (20)
- why no data normlization in data pre-processing?when I use data normlization in data pre-processing, rate of convergence of the network is slow HOT 1
- Can you share mean, std of imagenet21k? HOT 2
- Is the 1k validation set included in the 21k data? HOT 2
- About hierarchy balancing HOT 2
- No parent for n09450163 (sun)
- Any label map for ImageNet-1K?
- could you please provide image-label map directly?
- When using your ImageNet21K pretrained ResNet50 model in Detectron2, performance degrades HOT 1
- I see no normalization of images.
- Hyperparameters to finetune ResNet50 from IN21k to IN1k
- Missing details on Dropout and momentum value used for SGD when fine tuning on ImageNet1k
- What is the teacher model when using semantic softmax with KD?
- Where can I find the data?
- Does there any decriptions abount classes in "imagenet21k_small_classes"
- Anyone here have trouble reaching the mentioned accuracy for ViT-B?
- How to test Imagenet1K with pretrained backbone MobilenetV3_large_100? Could you release the testing script? Thanks a lot.
- ask for the pretrained model on ImageNet-21k-P from single label
- ask for the Transfer Learning Code of train.py on cifar100
- Dependencies to run code
- CIFAR-100 pretrained ViT-B-16 weight HOT 2
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 imagenet21k.