Add functions for input-masked loss calculation and batching #825
+156
−1
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
awni
reviewed
Nov 4, 2024
llms/mlx_lm/tuner/trainer.py
Outdated
| def iterate_input_masked_batches( | ||
| input_text, output_text, tokenizer, max_seq_length=2048 | ||
| ): | ||
| batch_size = len(input_text) |
There was a problem hiding this comment.
Why set the batch size to the length of the dataset?
awni
reviewed
Nov 4, 2024
llms/mlx_lm/tuner/trainer.py
Outdated
| input_lengths = mx.array(input_lengths) | ||
| lengths = mx.array(adjusted_lengths) | ||
|
|
||
| return batch[:, :-1], input_lengths, lengths |
There was a problem hiding this comment.
This only returns one example? Is that intentional? I assumed this is a drop-in replacement for iterate_batches but it's not clear that is the case based on how it's written..
… an updated attempt to better sync with iterate_batches logic
chimezie
changed the title
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Outdated suggestions cannot be applied.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.
Adds support for completion-only finetuning via functions for iterating over batching that also calculates input masks along with padding and a loss function using the masks
-- Updated 5 months later to keep up with mlx(_lm) changes, etc.