Trainer Loss #146
Merged
Trainer Loss #146
Conversation
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
meta-cla
bot
added
the
CLA Signed
joecummings
reviewed
Sep 15, 2025
Jack-Khuu
approved these changes
Sep 15, 2025
|
|
||
| def train_step( | ||
| self, inputs: list[dict[Tensor]], targets: list[dict[Tensor]] | ||
| ) -> None: |
| return tensor | ||
|
|
||
|
|
||
| def collate(batches: list[list[Episode]]): |
joecummings
approved these changes
Sep 15, 2025
allenwang28
reviewed
Sep 15, 2025
|
|
||
| print("Collecting Data...") | ||
| g = torch.manual_seed(0) | ||
| global_batch_size = cfg.replay_buffer.batch_size * cfg.replay_buffer.dp_size |
There was a problem hiding this comment.
no need to change right now, but wouldn't the global batch size technically be a trainer config and not a replay buffer config? Is it possible to tell omegaconf like "make this value the same as the one defined elsewhere"?
There was a problem hiding this comment.
Not a trainer config anymore since replay_buffer is it's own service and the trainer doesn't handle data or loading anymore
allenwang28
approved these changes
Sep 15, 2025
LucasLLC
approved these changes
Sep 15, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 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Outdated suggestions cannot be applied.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.
The goal of this PR is to enable titan training as well as keep the data definition and the loss easily accessible to the user. The loss, data type, and collation are left to the user so they can co-design all three to their specific data and are less likely to have to touch the trainer.