Update hooks pseudocode #7713
Update hooks pseudocode #7713
Conversation
carmocca
requested review from
awaelchli,
Borda,
edenlightning,
justusschock,
kaushikb11,
SeanNaren,
tchaton and
williamFalcon
as code owners
Codecov Report
@@ Coverage Diff @@
## master #7713 +/- ##
======================================
- Coverage 93% 92% -0%
======================================
Files 199 199
Lines 12971 12960 -11
======================================
- Hits 12001 11952 -49
- Misses 970 1008 +38 |
| @@ -1064,7 +1064,9 @@ override :meth:`pytorch_lightning.core.LightningModule.tbptt_split_batch`: | |||
|
|
|||
| Hooks | |||
| ^^^^^ | |||
| This is the pseudocode to describe how all the hooks are called during a call to ``.fit()``. | |||
| This is the pseudocode to describe the structure of :meth:`~pytorch_lightning.trainer.Trainer.fit`. | |||
| The inputs and outputs of each function are not represented for simplicity. Please check each function's API reference | |||
There was a problem hiding this comment.
I prefer keeping outputs here since its very helpful to distinguish some hooks (on_train_epoch_end vs training_epoch_end) and since inputs/outputs across hooks are closely related to the order of execution
There was a problem hiding this comment.
I removed it because I find it weird having the argument for training_step but not all the other functions that take a batch argument or an outputs argument.
So either I would add the batch argument for all functions or for none of them.
But adding it for all decreases the simplicity
There was a problem hiding this comment.
that's fair. should we have a separate section to discuss how data is passed across hooks (not for this PR)?
it can clarify #7723
There was a problem hiding this comment.
Sure. If you have any ideas on how to clear it do not hesitate to send a patch or open an issue
| @@ -1064,7 +1064,9 @@ override :meth:`pytorch_lightning.core.LightningModule.tbptt_split_batch`: | |||
|
|
|||
| Hooks | |||
| ^^^^^ | |||
| This is the pseudocode to describe how all the hooks are called during a call to ``.fit()``. | |||
| This is the pseudocode to describe the structure of :meth:`~pytorch_lightning.trainer.Trainer.fit`. | |||
| The inputs and outputs of each function are not represented for simplicity. Please check each function's API reference | |||
There was a problem hiding this comment.
that's fair. should we have a separate section to discuss how data is passed across hooks (not for this PR)?
it can clarify #7723
What does this PR do?
See title.
I removed the inputs/outputs because IMO they decrease the readability if we want to do them correctly. Also because the value in this snippet is the hook order, not the inputs/arguments - we have the API reference for those.
Also included:
Before submitting
PR review