Change default batch_size for finetuning to max_batch_size for a model #189
Conversation
This file contains 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
thepowerfuldeez
approved these changes
Sep 23, 2024
azahed98
approved these changes
Sep 23, 2024
mryab
reviewed
Sep 24, 2024
src/together/resources/finetune.py
Outdated
| @@ -55,7 +58,7 @@ def create( | |||
| n_evals (int, optional): Number of evaluation loops to run. Defaults to 0. | |||
| n_checkpoints (int, optional): Number of checkpoints to save during fine-tuning. | |||
| Defaults to 1. | |||
| batch_size (int, optional): Batch size for fine-tuning. Defaults to 32. | |||
| batch_size (int, optional): Batch size for fine-tuning. Defaults to auto. | |||
There was a problem hiding this comment.
Suggested change
| batch_size (int, optional): Batch size for fine-tuning. Defaults to auto. | |
| batch_size (int, optional): Batch size for fine-tuning. Defaults to max. |
mryab
reviewed
Sep 24, 2024
Comment on lines
+109
to
+124
| else: | ||
| if model_limits.full_training is None: | ||
| raise ValueError( | ||
| "Full training is not supported for the selected model." | ||
| ) |
There was a problem hiding this comment.
It looks like you have duplicated validation logic here and in cli/finetune.py. Maybe it's best to extract it to a function, call it in cli/finetune.py and reraise the exception as click.BadParameter if necessary?
There was a problem hiding this comment.
This logic is mostly for mypy -- it will error out that full_training is None in the following lines without this check
artek0chumak
force-pushed
the
artem/auto-batch-size
branch
from
b4595a0 to
e3b2cea
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Issue #ENG-10437
I added a logic to request limits for hyperparameters for a requested model. If a user does not specify hyperparameters, togther-python will change it to recommended values.