Merge pull request #157 from invoke-ai/maryhipp/flux-transformer-learning-rate

update default for transformer_learning_rate

Author: maryhipp

src/invoke_training/pipelines/flux/lora/config.py

@@ -55,7 +55,7 @@ class FluxLoraConfig(BasePipelineConfig):
     rate.
     """
 
-    transformer_learning_rate: float | None = 1e-4
+    transformer_learning_rate: float | None = 4e-4
     """The learning rate to use for the transformer model. If set, this overrides the optimizer's default learning
     rate.
     """