Made it so that the hyperparameters are saved to a file for each run.

code/main.py

@@ -9,6 +9,9 @@ import argparse
 from datetime import datetime
 from torchvision import transforms
 
+# TODO: lr scheduler support.
+# TODO: save the gradients somehow (maybe interesting for only a small batch during training)
+
 parser = argparse.ArgumentParser()
 parser.add_argument('-d', '--data', default='fashion', type=str, help="The dataset to use")
 parser.add_argument('--lr', default=0.0002, type=float, help="Which learning rate to use")
@@ -51,6 +54,10 @@ criterion = nn.CrossEntropyLoss()
 basedir = "weights/"+str(int(datetime.now().timestamp()))
 os.makedirs(basedir, exist_ok=True)
 
+with open(f"{basedir}/settings", "w") as f:
+    for key, val in args.__dict__.items():
+        f.write(f"{key} : {val}\n")
+
 
 def iterate_over_dataset(dataloader, model, criterion, optimizer, device, train=True):
     model.train() if train else model.eval()