update betas

mindee · Dec 20, 2024 · a034ed8 · a034ed8
1 parent 2799d92
commit a034ed8
Show file tree

Hide file tree

Showing 10 changed files with 14 additions and 14 deletions.
diff --git a/references/classification/train_pytorch_character.py b/references/classification/train_pytorch_character.py
@@ -291,7 +291,7 @@ def main(args):
         optimizer = torch.optim.AdamW(
             [p for p in model.parameters() if p.requires_grad],
             args.lr,
-            betas=(0.9, 0.99),
+            betas=(0.9, 0.999),
             eps=1e-6,
             weight_decay=args.weight_decay or 1e-4,
         )

diff --git a/references/classification/train_pytorch_orientation.py b/references/classification/train_pytorch_orientation.py
@@ -297,7 +297,7 @@ def main(args):
         optimizer = torch.optim.AdamW(
             [p for p in model.parameters() if p.requires_grad],
             args.lr,
-            betas=(0.9, 0.99),
+            betas=(0.9, 0.999),
             eps=1e-6,
             weight_decay=args.weight_decay or 1e-4,
         )

diff --git a/references/classification/train_tensorflow_character.py b/references/classification/train_tensorflow_character.py
@@ -258,7 +258,7 @@ def main(args):
         optimizer = optimizers.Adam(
             learning_rate=scheduler,
             beta_1=0.95,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=None if args.weight_decay == 0 else args.weight_decay,
@@ -267,7 +267,7 @@ def main(args):
         optimizer = optimizers.AdamW(
             learning_rate=scheduler,
             beta_1=0.9,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=args.weight_decay or 1e-4,

diff --git a/references/classification/train_tensorflow_orientation.py b/references/classification/train_tensorflow_orientation.py
@@ -268,7 +268,7 @@ def main(args):
         optimizer = optimizers.Adam(
             learning_rate=scheduler,
             beta_1=0.95,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=None if args.weight_decay == 0 else args.weight_decay,
@@ -277,7 +277,7 @@ def main(args):
         optimizer = optimizers.AdamW(
             learning_rate=scheduler,
             beta_1=0.9,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=args.weight_decay or 1e-4,

diff --git a/references/detection/train_pytorch.py b/references/detection/train_pytorch.py
@@ -343,7 +343,7 @@ def main(args):
         optimizer = torch.optim.AdamW(
             [p for p in model.parameters() if p.requires_grad],
             args.lr,
-            betas=(0.9, 0.99),
+            betas=(0.9, 0.999),
             eps=1e-6,
             weight_decay=args.weight_decay or 1e-4,
         )

diff --git a/references/detection/train_pytorch_ddp.py b/references/detection/train_pytorch_ddp.py
@@ -359,7 +359,7 @@ def main(rank: int, world_size: int, args):
         optimizer = torch.optim.AdamW(
             [p for p in model.parameters() if p.requires_grad],
             args.lr,
-            betas=(0.9, 0.99),
+            betas=(0.9, 0.999),
             eps=1e-6,
             weight_decay=args.weight_decay or 1e-4,
         )

diff --git a/references/detection/train_tensorflow.py b/references/detection/train_tensorflow.py
@@ -303,7 +303,7 @@ def main(args):
         optimizer = optimizers.Adam(
             learning_rate=scheduler,
             beta_1=0.95,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=None if args.weight_decay == 0 else args.weight_decay,
@@ -312,7 +312,7 @@ def main(args):
         optimizer = optimizers.AdamW(
             learning_rate=scheduler,
             beta_1=0.9,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=args.weight_decay or 1e-4,

diff --git a/references/recognition/train_pytorch.py b/references/recognition/train_pytorch.py
@@ -349,7 +349,7 @@ def main(args):
         optimizer = torch.optim.AdamW(
             [p for p in model.parameters() if p.requires_grad],
             args.lr,
-            betas=(0.9, 0.99),
+            betas=(0.9, 0.999),
             eps=1e-6,
             weight_decay=args.weight_decay or 1e-4,
         )

diff --git a/references/recognition/train_pytorch_ddp.py b/references/recognition/train_pytorch_ddp.py
@@ -283,7 +283,7 @@ def main(rank: int, world_size: int, args):
         optimizer = torch.optim.AdamW(
             [p for p in model.parameters() if p.requires_grad],
             args.lr,
-            betas=(0.9, 0.99),
+            betas=(0.9, 0.999),
             eps=1e-6,
             weight_decay=args.weight_decay or 1e-4,
         )

diff --git a/references/recognition/train_tensorflow.py b/references/recognition/train_tensorflow.py
@@ -305,7 +305,7 @@ def main(args):
         optimizer = optimizers.Adam(
             learning_rate=scheduler,
             beta_1=0.95,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=None if args.weight_decay == 0 else args.weight_decay,
@@ -314,7 +314,7 @@ def main(args):
         optimizer = optimizers.AdamW(
             learning_rate=scheduler,
             beta_1=0.9,
-            beta_2=0.99,
+            beta_2=0.999,
             epsilon=1e-6,
             clipnorm=5,
             weight_decay=args.weight_decay or 1e-4,