Skip to content

Implement Beta-Dreamer in DreamerV3 architecture #6

@aarunsrinivas5

Description

@aarunsrinivas5

Replace vanilla actor critic with epistemic risk seeking actor critic (ERSAC) variant using risk seeking exponential as opposed variance reward bonus in the DreamerV3 architecture to determine the benefits of risk modulation using exponential function. ERSAC paper: https://arxiv.org/pdf/2302.09339; Exponential TD Learning paper: https://johnbaras.com/wp-content/uploads/2023/08/22-29-Exponential_TD_Learning_A_Risk-Sensitive_Actor-Critic_Reinforcement_Learning_Algorithm.pdf

Metadata

Metadata

Labels

enhancementNew feature or request

Type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions