Make Mistral HF to NeMo checkpoint converter iterable to allow for lo… #9972
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
…w memory usage during conversion Signed-off-by: Alexandros Koumparoulis <[email protected]>
akoumpa
force-pushed
the
akoumparouli/iterable_ckpt_converter_mistral_2407
branch
from
7fa962d to
504e416
Compare
Signed-off-by: akoumpa <[email protected]>
yaoyu-33
approved these changes
Aug 2, 2024
adityavavre
pushed a commit
to adityavavre/NeMo
that referenced
this pull request
Sep 15, 2024
NVIDIA#9972) * Make Mistral HF to NeMo checkpoint converter iterable to allow for low memory usage during conversion Signed-off-by: Alexandros Koumparoulis <[email protected]> * Apply isort and black reformatting Signed-off-by: akoumpa <[email protected]> --------- Signed-off-by: Alexandros Koumparoulis <[email protected]> Signed-off-by: akoumpa <[email protected]> Co-authored-by: akoumpa <[email protected]> Signed-off-by: adityavavre <[email protected]>
monica-sekoyan
pushed a commit
that referenced
this pull request
Oct 14, 2024
#9972) * Make Mistral HF to NeMo checkpoint converter iterable to allow for low memory usage during conversion Signed-off-by: Alexandros Koumparoulis <[email protected]> * Apply isort and black reformatting Signed-off-by: akoumpa <[email protected]> --------- Signed-off-by: Alexandros Koumparoulis <[email protected]> Signed-off-by: akoumpa <[email protected]> Co-authored-by: akoumpa <[email protected]>
hainan-xv
pushed a commit
to hainan-xv/NeMo
that referenced
this pull request
Nov 5, 2024
NVIDIA#9972) * Make Mistral HF to NeMo checkpoint converter iterable to allow for low memory usage during conversion Signed-off-by: Alexandros Koumparoulis <[email protected]> * Apply isort and black reformatting Signed-off-by: akoumpa <[email protected]> --------- Signed-off-by: Alexandros Koumparoulis <[email protected]> Signed-off-by: akoumpa <[email protected]> Co-authored-by: akoumpa <[email protected]> Signed-off-by: Hainan Xu <[email protected]>
XuesongYang
pushed a commit
to paarthneekhara/NeMo
that referenced
this pull request
Jan 18, 2025
NVIDIA#9972) * Make Mistral HF to NeMo checkpoint converter iterable to allow for low memory usage during conversion Signed-off-by: Alexandros Koumparoulis <[email protected]> * Apply isort and black reformatting Signed-off-by: akoumpa <[email protected]> --------- Signed-off-by: Alexandros Koumparoulis <[email protected]> Signed-off-by: akoumpa <[email protected]> Co-authored-by: akoumpa <[email protected]>
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.
Makes the
convertfunction of the checkpoint converter an iterator. This allows for reduced memory usage during conversion, in fact less memory usage than the total RAM requirements of a full checkpoint.If the
--low-memCLI option is used, the converter will load only the requested layers from the HF checkpoint and upon conversion of a full layer (e.g. transformer) it will save it to disk. Next, it will load those partially converted checkpoints via mmap, thus only copying to ram layers that are being used.During the model initialization stage the converter may need the full checkpoint ram, but due to memory mapping it can do with less ram also, thus reducing the ram requirements significantly.
What does this PR do ?
Add a one line overview of what this PR aims to accomplish.
Collection: [Note which collection this PR will affect]
Changelog
Usage
# Add a code snippet demonstrating how to use thisGitHub Actions CI
The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.
The GitHub Actions CI will run automatically when the "Run CICD" label is added to the PR.
To re-run CI remove and add the label again.
To run CI on an untrusted fork, a NeMo user with write access must first click "Approve and run".
Before your PR is "Ready for review"
Pre checks:
PR Type:
If you haven't finished some of the above items you can still open "Draft" PR.
Who can review?
Anyone in the community is free to review the PR once the checks have passed.
Contributor guidelines contains specific people who can review PRs to various areas.
Additional Information