Comments (2)
In Table 4, page 12 of the Dinov2 paper, the top 1% accuracy number for Dino(v1) using a linear probe is 79.2. On the other hand, in Table 2 of the I-JEPA paper the top 1% accuracy for Dino(v1) is 70. This discrepancy could possibly be due to I-IEPA reporting results at 300 epochs and Dinov2 reporting results at 500 epochs.
My question is the following. At 300 epochs I-JEPA produces a top 1% accuracy of 77.3 which is 7.3% greater than Dino(v1). If you run I-JEPA to 500 epochs, would it still be greater than Dino(v1) by 7.3%?
Another question is: What happens if you train I-JEPA with the LVD-142M dataset for 500 epochs? Would it be superior to Dinov2?
from ijepa.
@bumi001 Good question, I am commenting so this gets up on the list. Form the IJEPA paper it seems suprior conceptually to DINOV2. Since it does not rely on augmentations for training, making it more resilient. Imagine making augmentations that break stuff in various specialised images not just regular RGB stuff. Can some one from the research team address this here? Maybe I should try to email Yann Lacunn. At the end Facebook choose to put its weight behind Dinov2 and there is not large pretrained model for IJEPA as I can see?
from ijepa.
Related Issues (20)
- Struggling to replicate evaluation results HOT 14
- An error was reported while training on multiple Gpus HOT 1
- Difficulty continue self supervised pre training on custom dataset HOT 2
- How to load ijepa checkpoints? HOT 6
- Downstream task HOT 5
- Image resolution & folder structure for unsupervised pre-training HOT 1
- imagenet1k Huggingface extraction HOT 1
- Training from scratch HOT 13
- generative model decoder for target-encoder
- what is the difference between root_folder and image_folder? HOT 1
- Struggling to Train Downstream Classifier HOT 5
- Bias in Multiblock Mask Collator
- Why is there no class token? HOT 4
- Suggestions to train on low-end GPU HOT 5
- Did you ever train with the ViTC?
- Why there is no **unscale_** when you use amp?
- How to create the visualizations on your README and Figure 6 of your paper?
- Why normalize twice since the target encoder also normalize the feature at last๏ผ
- Valid mask not found when train on mini image1k dataset HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
๐ 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. ๐๐๐
-
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 โค๏ธ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from ijepa.