Internal change

PiperOrigin-RevId: 529495239
This commit is contained in:
MediaPipe Team
2023-05-04 14:51:19 -07:00
committed by jqtang
parent 64cad80543
commit 12b0b6fad1
11 changed files with 534 additions and 518 deletions
@@ -106,7 +106,7 @@ class Dataset(classification_dataset.ClassificationDataset):
...
Each <file0>.xml annotation file should have the following format:
<annotation>
<filename>file0.jpg<filename>
<filename>file0.jpg</filename>
<object>
<name>kangaroo</name>
<bndbox>
@@ -114,6 +114,7 @@ class Dataset(classification_dataset.ClassificationDataset):
<ymin>89</ymin>
<xmax>386</xmax>
<ymax>262</ymax>
</bndbox>
</object>
<object>...</object>
</annotation>
@@ -27,8 +27,6 @@ class HParams(hp.BaseHParams):
learning_rate: Learning rate to use for gradient descent training.
batch_size: Batch size for training.
epochs: Number of training iterations over the dataset.
do_fine_tuning: If true, the base module is trained together with the
classification layer on top.
cosine_decay_epochs: The number of epochs for cosine decay learning rate.
See
https://www.tensorflow.org/api_docs/python/tf/keras/optimizers/schedules/CosineDecay
@@ -39,13 +37,13 @@ class HParams(hp.BaseHParams):
"""
# Parameters from BaseHParams class.
learning_rate: float = 0.003
batch_size: int = 32
epochs: int = 10
learning_rate: float = 0.3
batch_size: int = 8
epochs: int = 30
# Parameters for cosine learning rate decay
cosine_decay_epochs: Optional[int] = None
cosine_decay_alpha: float = 0.0
cosine_decay_alpha: float = 1.0
@dataclasses.dataclass
@@ -67,8 +65,8 @@ class QATHParams:
for more information.
"""
learning_rate: float = 0.03
batch_size: int = 32
epochs: int = 10
decay_steps: int = 231
learning_rate: float = 0.3
batch_size: int = 8
epochs: int = 15
decay_steps: int = 8
decay_rate: float = 0.96