splendor.agents.our_agents package

Subpackages

• splendor.agents.our_agents.genetic_algorithm package
  • Submodules
  • splendor.agents.our_agents.genetic_algorithm.argument_parsing module
    • parse_args()
  • splendor.agents.our_agents.genetic_algorithm.constants module
  • splendor.agents.our_agents.genetic_algorithm.evolve module
    • crossover()
    • evaluate()
    • evolve()
    • generate_initial_population()
    • main()
    • mate()
    • mutate()
    • mutate_population()
    • single_game()
    • sort_by_fitness()
  • splendor.agents.our_agents.genetic_algorithm.genes module
    • Gene
      • Gene.LOWER_BOUND
      • Gene.SHAPE
      • Gene.UPPER_BOUND
      • Gene.dna
      • Gene.load()
      • Gene.mutate()
      • Gene.random()
      • Gene.raw_dna
      • Gene.save()
    • ManagerGene
      • ManagerGene.SHAPE
      • ManagerGene.select_strategy()
    • StrategyGene
      • StrategyGene.SHAPE
      • StrategyGene.evaluate_state()
  • splendor.agents.our_agents.genetic_algorithm.genetic_algorithm_agent module
    • GeneAlgoAgent
      • GeneAlgoAgent.INVALID_POPULATION_ID
      • GeneAlgoAgent.MANAGER_PATH
      • GeneAlgoAgent.STRATEGY_1_PATH
      • GeneAlgoAgent.STRATEGY_2_PATH
      • GeneAlgoAgent.STRATEGY_3_PATH
      • GeneAlgoAgent.SelectAction()
      • GeneAlgoAgent.evaluate_action()
      • GeneAlgoAgent.save()
    • myAgent
  • Module contents
• splendor.agents.our_agents.ppo package
  • Subpackages
    • splendor.agents.our_agents.ppo.ppo_rnn package
      • Subpackages
      • Submodules
      • splendor.agents.our_agents.ppo.ppo_rnn.recurrent_ppo module
      • Module contents
    • splendor.agents.our_agents.ppo.self_attn package
      • Submodules
      • splendor.agents.our_agents.ppo.self_attn.constants module
      • splendor.agents.our_agents.ppo.self_attn.network module
      • splendor.agents.our_agents.ppo.self_attn.ppo_agent module
      • Module contents
  • Submodules
  • splendor.agents.our_agents.ppo.arguments_parsing module
    • Arguments
      • Arguments.architecture
      • Arguments.device_name
      • Arguments.learning_rate
      • Arguments.opponent
      • Arguments.saved_weights
      • Arguments.seed
      • Arguments.test_opponent
      • Arguments.weight_decay
      • Arguments.working_dir
    • NeuralNetArch
      • NeuralNetArch.agent_relative_import_path
      • NeuralNetArch.default_saved_weights
      • NeuralNetArch.hidden_state_dim
      • NeuralNetArch.is_recurrent
      • NeuralNetArch.name
      • NeuralNetArch.ppo_factory
    • parse_args()
  • splendor.agents.our_agents.ppo.common module
    • calculate_advantages()
    • calculate_loss()
    • calculate_policy_loss()
    • calculate_returns()
  • splendor.agents.our_agents.ppo.constants module
  • splendor.agents.our_agents.ppo.input_norm module
    • InputNormalization
      • InputNormalization.forward()
  • splendor.agents.our_agents.ppo.network module
    • PPO
      • PPO.forward()
  • splendor.agents.our_agents.ppo.ppo module
    • extract_game_stats()
    • main()
    • save_model()
    • train()
  • splendor.agents.our_agents.ppo.ppo_agent module
    • PPOAgent
      • PPOAgent.SelectAction()
      • PPOAgent.load()
    • myAgent
  • splendor.agents.our_agents.ppo.ppo_agent_base module
    • PPOAgentBase
      • PPOAgentBase.SelectAction()
      • PPOAgentBase.load()
      • PPOAgentBase.load_policy()
  • splendor.agents.our_agents.ppo.ppo_base module
    • PPOBase
      • PPOBase.create_hidden_layers()
      • PPOBase.forward()
      • PPOBase.init_hidden_state()
    • PPOBaseFactory
  • splendor.agents.our_agents.ppo.rollout module
    • RolloutBuffer
      • RolloutBuffer.action_dim
      • RolloutBuffer.action_mask_history
      • RolloutBuffer.actions
      • RolloutBuffer.calculate_gae()
      • RolloutBuffer.cell_states
      • RolloutBuffer.clear()
      • RolloutBuffer.device
      • RolloutBuffer.dones
      • RolloutBuffer.full
      • RolloutBuffer.hidden_states
      • RolloutBuffer.hidden_states_shape
      • RolloutBuffer.index
      • RolloutBuffer.input_dim
      • RolloutBuffer.is_recurrent
      • RolloutBuffer.log_prob_actions
      • RolloutBuffer.remember()
      • RolloutBuffer.rewards
      • RolloutBuffer.size
      • RolloutBuffer.states
      • RolloutBuffer.unpack()
      • RolloutBuffer.values
  • splendor.agents.our_agents.ppo.training module
    • LearningParams
      • LearningParams.device
      • LearningParams.discount_factor
      • LearningParams.hidden_states_shape
      • LearningParams.is_recurrent
      • LearningParams.loss_fn
      • LearningParams.optimizer
      • LearningParams.ppo_clip
      • LearningParams.ppo_steps
      • LearningParams.seed
    • evaluate()
    • train_single_episode()
    • update_policy()
  • splendor.agents.our_agents.ppo.utils module
    • load_saved_model()
    • load_saved_ppo()
  • Module contents

Submodules

splendor.agents.our_agents.minmax module

Implementation of an agent that selects the first legal action.

class splendor.agents.our_agents.minmax.MiniMaxAgent(_id)

Bases: Agent

A Minimax agent, utilizing the zero-sum property of the game, there is only a single winner in each game, for determining which action to play.

SelectAction(actions, game_state, game_rule)

splendor.agents.our_agents.minmax.myAgent

alias of MiniMaxAgent

Module contents