Merged
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
qgallouedec
commented
Oct 31, 2024
| import numpy as np | ||
| from accelerate import Accelerator | ||
| from huggingface_hub import InferenceClient | ||
| from scipy.special import softmax |
qgallouedec
commented
Oct 31, 2024
| assert nb_trainable_params == 10273 | ||
|
|
||
| @require_bitsandbytes | ||
| @require_torch_gpu_if_bnb_not_multi_backend_enabled |
Member
Author
|
Failing test are expected and will be solved later (in #2288 for example) |
kashif
reviewed
Oct 31, 2024
5 tasks
lewtun
approved these changes
Oct 31, 2024
Member
lewtun
left a comment
lewtun
left a comment
There was a problem hiding this comment.
LGTM with a small tweak to the softmax
| cache: "pip" | ||
| cache-dependency-path: | | ||
| setup.py | ||
| requirements.txt |
Member
There was a problem hiding this comment.
If we find the dependencies are a hassle to maintain, we might want to switch to the transformers approach, which uses a look-up table to give fine-grained control: https://github.com/huggingface/transformers/blob/main/setup.py
yxliu-TAMU
pushed a commit
to mincheolseong/ECEN743-GRPO-Project-Proposal
that referenced
this pull request
Apr 20, 2025
* clean deps * new tests * tests * Add tests without optional dependencies workflow * Update dependencies in tests.yml * cpu version of torch * Update dependencies and installation commands * Disable fail-fast in test workflow * Update test matrix in workflows file * try fix windows * Remove "rich" from required packages in setup.py * Update dependency installation in tests.yml * Add torch and deepspeed installation for windows-latest * Fix conditional statement in workflow file * Add torch and deepspeed installation for Windows * Fix if statement * Update torch and deepspeed dependencies * Update liger package requirement for non-Windows platforms * remove scipy dep * Add torch GPU requirement for testing_utils * Update trl/trainer/judges.py
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
What does this PR do?
This PR cleans up TRL dependencies by better dividing and following best practice.
It updates the test workflow accordingly.
deepspeedandliger-kernelare only supported for linux. This PR ensure that they are not part of optional dependencies with windows.Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.