fim-ode / base_model /model_architecture.txt
jrhuebers's picture
Upload FIM-ODE base model
b8f4b1f verified
==============================================================================================================
Layer (type:depth-idx) Output Shape Param #
==============================================================================================================
TrainingWrapper -- --
├─FimOdeon: 1-1 -- --
│ └─TrajectoryEncoder: 2-1 -- 896
│ │ └─TransformerEncoder: 3-1 [1, 1194, 256] 1,579,520
│ └─Sequential: 2-2 -- --
│ │ └─Linear: 3-2 [1, 2400, 256] 1,024
│ │ └─ReLU: 3-3 [1, 2400, 256] --
│ │ └─Linear: 3-4 [1, 2400, 256] 65,792
│ └─AttentionOperator: 2-3 -- --
│ │ └─ModuleList: 3-5 -- 6,318,080
│ │ └─MLP: 3-6 [1, 2400, 3] 132,355
├─UncertaintyEstimator: 1-2 -- --
│ └─AttentionOperator: 2-4 -- --
│ │ └─ModuleList: 3-7 -- 4,738,560
│ │ └─MLP: 3-8 [1, 2400, 1] 131,841
==============================================================================================================
Total params: 12,968,068
Trainable params: 12,968,068
Non-trainable params: 0
Total mult-adds (Units.MEGABYTES): 12.97
==============================================================================================================
Input size (MB): 0.09
Forward/backward pass size (MB): 771.15
Params size (MB): 51.87
Estimated Total Size (MB): 823.11
==============================================================================================================