tf_agents.agents.dqn.dqn_agent.DqnLossInfo
Stay organized with collections
Save and categorize content based on your preferences.
DqnLossInfo is stored in the extras field of the LossInfo instance.
tf_agents.agents.dqn.dqn_agent.DqnLossInfo(
td_loss, td_error
)
Both td_loss and td_error have a validity mask applied to ensure that
no loss or error is calculated for episode boundaries.
td_loss: The weighted TD loss (depends on choice of loss metric and
any weights passed to the DQN loss function.
td_error: The unweighted TD errors, which are just calculated as:
td_error = td_targets - q_values
These can be used to update Prioritized Replay Buffer priorities.
Note that, unlike td_loss, td_error may contain a time dimension when
training with RNN mode. For td_loss, this axis is averaged out.
Attributes |
|---|
td_loss
|
A namedtuple alias for field number 0
|
td_error
|
A namedtuple alias for field number 1
|
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2024-04-26 UTC.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2024-04-26 UTC."],[],[]]
View source on GitHub