New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Great work,How to fintune #5

Open

zhangbaijin opened this issue Jan 4, 2024 · 1 comment

zhangbaijin commented Jan 4, 2024

Thanks for your contribution, it is a great work, how to fintune my dataset? like mural dataset, can you give the command or script?

Collaborator

PhyscalX commented Jan 4, 2024

Hi, @zhangbaijin

To train TAP models, you should implement the following components:

tokenize_anything/config/... (YAML Configuration System)
tokenize_anything/data/... (data loader, transforms)
tokenize_anything/engine/train_engine.py (train loop, optimizer, LR scheduler, mixed precision and DDP wrapper)
tokenize_anything/models/<my_model>.py (custom TAP builder, losses/targets getter and forward impl. for training)
scripts/train.py (argument parser, DDP initialization and dispatcher for your train engine)

You can follow detectron2 to complete the above components.
Currently, we have no plan to release the full code.

PhyscalX pinned this issue

PhyscalX mentioned this issue

question about Model D training #11

Closed

PhyscalX unpinned this issue

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment