MCPcopy Create free account

hub / github.com/brendanator/atari-rl / types & classes

Types & classes25 in github.com/brendanator/atari-rl

⨍ Functions 138◇ Types & classes 25
↓ 6 callersClassArraySyntax
networks/loss.py:193
↓ 3 callersClassRequiredFeeds
networks/inputs.py:101
↓ 2 callersClassMock
test/mock.py:1
↓ 2 callersClassProportionalPriorities
Track the priorities of each transition proportional to the TD-error Contains a sum tree and a max tree for tracking values needed Each tree is i
agents/replay_priorities.py:23
↓ 2 callersClassReplayMemory
agents/replay_memory.py:11
↓ 1 callersClassActionValueHead
networks/dqn.py:151
↓ 1 callersClassActorCriticHead
networks/dqn.py:182
↓ 1 callersClassAgent
agents/agent.py:7
↓ 1 callersClassAtari
atari/atari.py:9
↓ 1 callersClassExplorationBonus
agents/exploration_bonus.py:9
↓ 1 callersClassFastAtariEnv
atari/atari.py:107
↓ 1 callersClassInputs
networks/inputs.py:6
↓ 1 callersClassNetworkFactory
networks/factory.py:7
↓ 1 callersClassOffsetInput
networks/inputs.py:87
↓ 1 callersClassSampleBatch
agents/replay_memory.py:158
↓ 1 callersClassTrainer
agents/training.py:9
↓ 1 callersClassUniformPriorities
Each transition has equal priority
agents/replay_priorities.py:4
↓ 1 callersClassmemodict
util/util.py:102
ClassDisabledRewardScaling
An implementation that doesn't scale rewards
networks/reward_scaling.py:57
ClassLosses
networks/loss.py:5
ClassNetwork
networks/dqn.py:8
ClassReplayMemoryTest
test/test_replay_memory.py:9
ClassReplayPrioritiesTest
test/test_replay_priorities.py:9
ClassRewardScaling
Reward scaling is implemented using normalized SGD (algorithm 2 from paper)
networks/reward_scaling.py:4
ClassSummary
util/summary.py:4