Not able to change model to EfficientDet-D1 or D2

[drelvenkee]

Hi EJ

I am trying to change the model to D1 or D2. But it is not working. I received this error
ValueError: Cannot assign value to variable ' WeightSharedConvolutionalBoxPredictor/BoxPredictionTower/conv2d_0/BatchNorm/feature_0/beta:0': Shape mismatch.The variable shape (88,), and the assigned value shape (112,) are incompatible.

Possible to help? Here is my code.

chosen_model = 'efficientdet-d2'

MODELS_CONFIG = {
'efficientdet-d2': {
'model_name': 'efficientdet_d2_coco17_tpu-32',
'base_pipeline_file': 'ssd_efficientdet_d2_768x768_coco17_tpu-8.config',
'pretrained_checkpoint': 'efficientdet_d2_coco17_tpu-32.tar.gz',
}
}

model_name = MODELS_CONFIG[chosen_model]['model_name']
pretrained_checkpoint = MODELS_CONFIG[chosen_model]['pretrained_checkpoint']
base_pipeline_file = MODELS_CONFIG[chosen_model]['base_pipeline_file']

[sangyo1]

Hello did you solve the problem for this?
if so, how did you add the efficientdet-d2?