Merge pull request #1 from yunjuanwang/fix-readme-download-requirements

Fix README, requirements, download files

Author: hazimehh

README.md

@@ -33,11 +33,31 @@ DomainRobust includes the following datasets:
 
 ## Quick start
 
+To download the datasets:
+```sh
+python3 -m domainrobust.download \
+       --data_dir=./domainrobust/data
+```
+
+We first generate a pretrained model using DANN (this pretrained model will be used by algorithms like SROUDA and DART):
+
+```sh
+python3 -m domainrobust.scripts.train\
+       --data_dir=/my/datasets/path\
+       --output_dir=/my/pretrained/model/path\
+       --algorithm DANN\
+       --dataset DIGIT\
+       --task domain_adaptation\
+       --source_envs 0\
+       --target_envs 2
+```
+
 To train a single model:
 
 ```sh
-python3 -m scripts.train\
+python3 -m domainrobust.scripts.train\
        --data_dir=/my/datasets/path\
+       --output_dir=/output/path\
        --algorithm AT\
        --dataset DIGIT\
        --task domain_adaptation\
@@ -55,7 +75,7 @@ python3 -m scripts.train\
 To launch a sweep (over a range of hyperparameters and possibly multiple algorithms and datasets):
 
 ```sh
-python -m scripts.sweep launch\
+python -m domainrobust.scripts.sweep launch\
        --data_dir=/my/datasets/path\
        --output_dir=/my/sweep/output/path\
        --command_launcher MyLauncher\
@@ -100,4 +120,4 @@ python -m domainrobust.scripts.test\
        --atk_lr 0.004 \
        --atk_iter 20 \
        --attack pgd 
-````
+````

domainrobust/download.py

@@ -54,10 +54,6 @@ def download_digit(data_dir):
     download_and_extract("https://drive.google.com/uc?id=1wnY5M74Zj-lPuOQzcmVO9KQVLVhgK7Vu",
                          os.path.join(data_dir, "DIGIT.zip"))
 
-    os.rename(os.path.join(data_dir, "kfold"),
-              full_path)
-
-
 
 # PACS ########################################################################
 
@@ -94,8 +90,6 @@ def download_visda(data_dir):
     download_and_extract("https://drive.google.com/uc?id=1VrIlU6yrm-XTcwfpRIWuALdTAYulgdYp",
                          os.path.join(data_dir, "VISDA.zip"))
 
-    os.rename(os.path.join(data_dir, "kfold"),
-              full_path)
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description='Download datasets')

requirements.txt

@@ -1,10 +1,10 @@
-numpy==1.20.3
+numpy==1.21.6
 wilds==1.2.2
-imageio==2.9.0
-gdown==3.13.0
-torchvision==0.8.2
-torch==1.7.1
-tqdm==4.62.2
+imageio==2.31.2
+gdown==4.7.3
+torchvision==0.14.1
+torch==1.13.1
+tqdm==4.66.2
 backpack==0.1
 parameterized==0.8.1
 Pillow==8.3.2