Skip to content

🏃 Migrate CI to self-hosted runners#3174

Merged
qgallouedec merged 71 commits into
mainfrom
migrate-ci-1
Mar 29, 2025
Merged

🏃 Migrate CI to self-hosted runners#3174
qgallouedec merged 71 commits into
mainfrom
migrate-ci-1

Conversation

@qgallouedec

@qgallouedec qgallouedec commented Mar 28, 2025

Copy link
Copy Markdown
Member

This PR:

  1. Migration of our CI to a self-hosted runner. This allows us to test on more relevant hardware, since it has 1 GPU. (instead of CPU only)

  2. Give up support for Windows. Reasons:

  • Windows is the recurring source of problems in our CI and takes up a lot of our time.
  • According to pypistats it's 3% of TRL usage.
  • Most optional dependencies don't support Windows anyway.
    Note that this doesn't mean that TRL no longer works with Windows, it's just that we no longer test it and no longer officially support it.

🏎️💨 2.5x faster

@HuggingFaceDocBuilderDev

Copy link
Copy Markdown

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.

Comment on lines -27 to -29
with:
fetch-depth: 0
submodules: recursive

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we don't need it

Comment on lines +45 to +47
defaults:
run:
shell: bash

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bash is needed otherwise it uses sh which doesn't have source

Comment on lines -53 to -56
cache: "pip"
cache-dependency-path: |
setup.py
requirements.txt

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

not compatible with the runner (I don't think it's very useful anyway)

run: |
apt-get update && apt-get install -y make git curl

- name: Install uv

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Much faster installation with uv

self.assertFalse(torch.equal(param, new_param), f"Parameter {n} has not changed.")

@unittest.skipIf(sys.platform.startswith("win"), "Skipping on Windows") # compiling seems to be broken on Windows
def test_training_torch_compile(self):

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this test was veeeery slow

@rtrompier rtrompier requested a review from glegendre01 March 29, 2025 08:54
@qgallouedec

Copy link
Copy Markdown
Member Author

I'll merge because for now the CI is broken, but I'm still interested in your remarks if any

@qgallouedec qgallouedec merged commit 2fe2337 into main Mar 29, 2025
@qgallouedec qgallouedec deleted the migrate-ci-1 branch March 29, 2025 18:56
kashif pushed a commit to kashif/trl that referenced this pull request Mar 31, 2025
BjarniHaukur pushed a commit to ASSERT-KTH/trl that referenced this pull request Apr 15, 2025
…ggingface#3131)

Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
Co-authored-by: Quentin Gallouédec <gallouedec.quentin@gmail.com>

log answer key to wandb

all Table

HTML logging

table

bump patch

hmm

formatting

html esacape

reward isnt string

[Liger] Liger KTO support (huggingface#2812)

Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Quentin Gallouédec <gallouedec.quentin@gmail.com>

🏃 Migrate CI to self-hosted runners (huggingface#3174)

❤️‍🩹 [CI] fix transformers dev CI failure (huggingface#3176)

Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>

⏯️ Fix: handle None inputs when resuming GRPO Trainer from checkpoint (huggingface#3148)

Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>

📎 Fix is_clipped to compute the effective clip_ratio (huggingface#3175)

Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
Co-authored-by: Quentin Gallouédec <gallouedec.quentin@gmail.com>

Fix breaking typo for flash_attention reducing_memory_usage.md (huggingface#3190)

Show unique prompts in GRPO WandB tables (huggingface#3191)

🐗 [CI] Fix trufflehog false positives (huggingface#3192)

[GRPO] Improve completion length logging (huggingface#3188)

preliminary openai compatible endpoint

early concept, needs refining

dedupe

debug print

some slop to work on

unslop, missing hist

almost valid pseudocode

middle-ware monkey patch in mp.Pool()...

remove unused

More accurate .md

need gpu

renting lambda again

much nicer

small

aider-chat and datasets conflict

risky reqs change

should work, but hacky

some insights, but monkeypatching probably wont suffice

refactor: Rewrite test script to use SWE-bench dataset with MultiProcessAider

refactor: Remove logging statements from test.py

one step closer

finally, the correct abstraction

doc

todo

unslop

unslop

undo accidental black

cleaner abstraction

new abstraction
yxliu-TAMU pushed a commit to mincheolseong/ECEN743-GRPO-Project-Proposal that referenced this pull request Apr 20, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants