search

ViTAE-Transformer / ViTDet

Unofficial implementation for [ECCV'22] "Exploring Plain Vision Transformer Backbones for Object Detection"
Apache License 2.0
530 stars 46 forks source link

about lsj_100e config #2

Closed youngwanLEE closed 2 years ago

youngwanLEE commented 2 years ago

Hi,

I'm very impressed by your work!

I have a question the lsj_100e config.

I'm confused by the epoch number.

In this config, the epoch is set to 25.

https://github.com/ViTAE-Transformer/ViTDet/blob/0d3298c5f30ea63c2cbf89cc74ea8459526f5b0f/configs/ViTDet/lsj_100e_coco_instance_r1024.py#L90

ViTDet is trained for 100 epochs in the paper.

Could you fix me if I'm wrong?

Thanks in advance

Annbless commented 2 years ago

Hi,

Please note this line. The COCO dataset has been repeated 4 times to perform the 100 epochs training.

youngwanLEE commented 2 years ago

@Annbless Oh, I see.

Thanks for the quick reply :)