Table 2 The specific parameters of the modality decomposition interaction module.

From: Offline reinforcement learning combining generalized advantage estimation and modality decomposition interaction

Architecture

Hyper-parameters

Value

CGM

Encoder

ConvFormer layers

1

Dual-TransFormer layers

2

Decoder layers

3

Attention head

4

Embedding dimension

128

Nonlinearity function

Relu

Back to article page