detail argument GPU in action-da & bingdingdb

examples/action_dann_lightn/test.py

@@ -19,7 +19,11 @@ def arg_parse():
     """Parsing arguments"""
     parser = argparse.ArgumentParser(description="Domain Adversarial Networks on Action Datasets")
     parser.add_argument("--cfg", required=True, help="path to config file", type=str)
-    parser.add_argument("--gpus", default="0", help="gpu id(s) to use", type=str)
+    parser.add_argument(
+        "--gpus",
+        default="0",
+        help="gpu id(s) to use. None/int(0) for cpu. list[x,y] for xth, yth GPU. str(x) for the first x GPUs. str(-1)/int(-1) for all available GPUs",
+    )
     parser.add_argument("--resume", default="", type=str)
     parser.add_argument("--ckpt", default="", help="pre-trained parameters for the model (ckpt files)", type=str)
     args = parser.parse_args()
@@ -64,11 +68,7 @@ def main():
 
     # ---- setup model and logger ----
     model, train_params = get_model(cfg, dataset, num_classes)
-    trainer = pl.Trainer(
-        logger=False,
-        resume_from_checkpoint=args.ckpt,
-        gpus=args.gpus,
-    )
+    trainer = pl.Trainer(logger=False, resume_from_checkpoint=args.ckpt, gpus=args.gpus,)
 
     model_test = weights_update(model=model, checkpoint=torch.load(args.ckpt))
 

examples/bindingdb_deepdta/main.py

@@ -15,7 +15,11 @@ def arg_parse():
     """Parsing arguments"""
     parser = argparse.ArgumentParser(description="DeepDTA on BindingDB dataset")
     parser.add_argument("--cfg", required=True, help="path to config file", type=str)
-    parser.add_argument("--gpus", default="0", help="gpu id(s) to use", type=str)
+    parser.add_argument(
+        "--gpus",
+        default="0",
+        help="gpu id(s) to use. None/int(0) for cpu. list[x,y] for xth, yth GPU. str(x) for the first x GPUs. str(-1)/int(-1) for all available GPUs",
+    )
     parser.add_argument("--resume", default="", type=str)
     args = parser.parse_args()
     return args