ett_tmlr2022's People
ett_tmlr2022's Issues
training code
will the training code for this model be released?
About the the initialization of prefix
Hello! Thank you for releasing your outstanding work! While using your code, I have a small question: In your paper, you emphasized the importance of the initial prefix, and you adopted the initialization method of the attentive prototype. However, the implementation in your code seems to be simply averaging all patch embeddings from each category. I wonder if there is a comparable performance between the two approaches? It appears that you provide a simpler option in your code.
Incorrect parser argument for model checkpoint loading
There is a bug in the example code loading a pre-trained ViT model.
Code Line
The argument should be "model.ckpt" instead of "data.ckpt".
This bug got introduced in commit "update doc"/2f15355
About the reproduction
Hi, I am very interested in your work. When I attempted to reproduce the model provided in #1 using the default settings in your code, I achieved an accuracy of 75.27% on Omniglot and 76.77% on Aircraft. This shows a notable difference from the results mentioned in your paper (78.11% and 79.94%). Could you please help me identify if there are any details I might have overlooked?
some hyperparameter detail problem in the code
hello, it's a fantastic work which achieve a ViT-applied adapter. It can expand adapter method from Resnet to ViT-like network. Could you provide some suggestions on how to choose a good learning rate with adam/adamw? Also, I wanna know how the hyperparameters in code you get. Appreciate your efforts on it.
关于在ImageNet1K的train set的训练超参数
尊敬的作者,您好!感谢您开源这篇富有创造力的工作。最近想在您工作的基础上做一些尝试,请问是否方便明确一下您在使用DINO进行预训练时,ViT-Small和ViT-Tiny使用的训练超参数(特别是batch size 和 epoch)都是怎样的呢?是都采用您代码的默认设置吗?大概使用多少张多大显存的GPU能完成呢?此外,您论文中还提到,尝试过用DeiT进行过微调,微调时学习率和epoch也都是严格仿照DeiT原论文吗?
期待您能为我解惑,将不胜感激!
Request for support data finetuning code.
Hello.
Could you please upload the finetuning code that was used for the support data. Also, I would like to implement this with SWIN_VIT. Could you give a brief methodology on what changes have to be made at model level, ie. how to implement DRA in the layers.
Thanks.
Issues about Pretrained DINO
The DINO Pretrain used in this experiment is primary different from official dino by 1000 classes VS 712 classes?
And why do such a setting? It's clear that the main target of this paper is cross-domain application, so use full ImageNet dataset don't affect the result. Maybe this is for compare with other method like TSA and CTX?
Looking forward to your reply! Thanks!
pre-training dino
Training commands
Hello. Could you please provide the training commands for reproduction ? Thanks a lot
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.