Model twice in README_Segmentation.md

sacmehta / EdgeNets

This repository contains the source code of our work on designing efficient CNNs for computer vision

MIT License

411 stars 82 forks source link

Model twice in README_Segmentation.md #36

Closed mitalbert closed 4 years ago

mitalbert commented 4 years ago

FYI, the model and s arguments are passed twice, so the arguments are getting overwritten.

Cityscapes dataset

CUDA_VISIBLE_DEVICES=0 python train_segmentation.py --model espnetv2 --s 2.0 --dataset city --data-path ./vision_datasets/cityscapes/ --batch-size 25 --crop-size 512 256 --model dicenet --s 1.75 --lr 0.009 --scheduler hybrid --clr-max 61 --epochs 100

sacmehta commented 4 years ago

Thanks for noting it. I have fixed it.