From 1bedfe305b8fedfc6cf7c372ef62535006e101ec Mon Sep 17 00:00:00 2001 From: Jerry Jiarui XU Date: Mon, 20 Jul 2020 15:17:59 +0800 Subject: [PATCH] Fixed training tricks (#26) --- docs/getting_started.md | 2 +- docs/tutorials/training_tricks.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/getting_started.md b/docs/getting_started.md index 16cb3b4..e0a1cae 100644 --- a/docs/getting_started.md +++ b/docs/getting_started.md @@ -132,7 +132,7 @@ Assume that you have already downloaded the checkpoints to the directory `checkp checkpoints/pspnet_r50-d8_512x1024_40k_cityscapes_20200605_003338-2966598c.pth \ 4 --out results.pkl --eval mIoU cityscapes ``` - Note: There is some inconsistency between cityscapes mIoU and our mIoU. The reason is that cityscapes average each class with class size by default. + Note: There is some gap (~0.1%) between cityscapes mIoU and our mIoU. The reason is that cityscapes average each class with class size by default. We use the simple version without average for all datasets. 5. Test PSPNet on cityscapes test split with 4 GPUs, and generate the png files to be submit to the official evaluation server. diff --git a/docs/tutorials/training_tricks.md b/docs/tutorials/training_tricks.md index 8555216..2a56daf 100644 --- a/docs/tutorials/training_tricks.md +++ b/docs/tutorials/training_tricks.md @@ -8,7 +8,7 @@ In semantic segmentation, some methods make the LR of heads larger than backbone In MMSegmentation, you may add following lines to config to make the LR of heads 10 times of backbone. ```python -optimizer_config=dict( +optimizer=dict( paramwise_cfg = dict( custom_keys={ 'head': dict(lr_mult=10.)}))