[Doc] Add recurrent state lifecycle guide - #3792
Merged
Merged
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3792
Note: Links to docs will display an error until the docs builds have been completed. This comment was automatically generated by Dr. CI and updates every 15 minutes. |
theap06
force-pushed
the
feat/hidden-understandable
branch
from
May 24, 2026 06:15
7d5e1ca to
dc6cd0c
Compare
…andable # Conflicts: # torchrl/modules/tensordict_module/rnn.py
… to follow-up PR)
Collaborator
Author
|
@vmoens feel free to leave some feedback when ur free. also lmk if there is anything I can do to help with general maintainance of the library |
Collaborator
Author
|
@vmoens thanks for the changes! I had the InitTracker-missing failure mode wrong (the KeyError path vs the silent-content-wrong path), and the "final hidden" caveat is much more accurate now. |
vmoens
reviewed
Jun 1, 2026
| plumbing for a PPO-style update. It omits optimization, logging, and | ||
| multi-epoch training so the data path stays visible. | ||
|
|
||
| .. code-block:: python |
Collaborator
There was a problem hiding this comment.
@theap06 added an e2e example -- worth showing what we're really talking about
vmoens
approved these changes
Jun 1, 2026
theap06
added a commit
to theap06/rl
that referenced
this pull request
Jun 12, 2026
Adds tutorials/sphinx-tutorials/recurrent_sequence_training.py — the multi-step / sequence-training complement to dqn_with_rnn.py (which covers single-step recurrent DQN at collection time). Walks through the post-pytorch#3695 recurrent contract end-to-end: - Collector auto-wiring of InitTracker + the recurrent-state primer via auto_register_policy_transforms=True - Trajectory-aware sampling with SliceSampler - Multi-step LSTM forward under set_recurrent_mode(True) - Boundary safety: hand-built two-trajectory packed batch + isolation check that proves hidden state does not leak across is_init markers - A tiny end-to-end training loop closing the BC-style sequence path Runs in ~3s on CPU. Cross-references the recurrent state lifecycle guide (pytorch#3792), collector internals page (pytorch#3796), and the glossary. Toctree entry added to docs/source/index.rst next to dqn_with_rnn.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #3746
Summary
Adds a focused recurrent-state lifecycle guide and an integration test for the
full policy. Closes the "recurrent debugging requires jumping across LSTMModule, InitTracker, and
loss-side
is_initmasking with no unifying doc" gap, and the parallel gapof having no integration test that exercises a multi-trajectory batch with
mid-batch
done.