sbrl-github/research-rainfallradar
1
0
Fork 0
mirror of https://github.com/sbrl/research-rainfallradar synced 2024-11-25 18:33:01 +00:00
Code Issues Projects Releases Packages Wiki Activity

actually use dice loss

Browse source
This commit is contained in:
Starbeamrainbowlabs 2022-12-09 18:35:17 +00:00
parent 649c262960
commit e22c0981e6
Signed by: sbrl
GPG key ID: 1BE5172E637709C2
1 changed files with 3 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
aimodel/src/lib/ai/model_rainfallwater_mono.py
View file

@ -7,6 +7,7 @@ from .components.convnext import make_convnext
from .components.convnext_inverse import do_convnext_inverse
from .components.LayerStack2Image import LayerStack2Image
from .components.LossCrossentropy import LossCrossentropy
from .components.LossDice import LossDice
def model_rainfallwater_mono(metadata, model_arch_enc="convnext_xtiny", model_arch_dec="convnext_i_xtiny", feature_dim=512, batch_size=64, water_bins=2, learning_rate=None, heightmap_input=False):
"""Makes a new rainfall / waterdepth mono model.
@ -87,7 +88,8 @@ def model_rainfallwater_mono(metadata, model_arch_enc="convnext_xtiny", model_ar
optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate)
model.compile(
optimizer=optimizer,
loss=LossCrossentropy(batch_size=batch_size),
# loss=LossCrossentropy(batch_size=batch_size),
loss=LossDice(),
# loss=tf.keras.losses.CategoricalCrossentropy(),
metrics=[tf.keras.metrics.CategoricalAccuracy()]
)