Prepare for transformers v5 upgrade by winglian · Pull Request #3272 · axolotl-ai-cloud/axolotl

winglian · 2025-11-20T14:05:58Z

Summary by CodeRabbit

Chores
- Expanded build pipeline support for new dependency versions: CUDA 12.8.1, Python 3.11, PyTorch 2.9.0, and PyTorch 2.9.1. Updated both standard and cloud build configurations.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

coderabbitai · 2025-11-20T14:06:08Z

Important

Review skipped

Auto incremental reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

🔍 Trigger a full review

📝 Walkthrough

Walkthrough

Modifies .github/workflows/main.yml to add matrix entries for CUDA 12.8.1 configurations to both build-axolotl and build-axolotl-cloud jobs, introducing two PyTorch versions (2.9.0 and 2.9.1) with Python 3.11.

Changes

Cohort / File(s)	Summary
Workflow Matrix Configuration `\.github/workflows/main\.yml`	Adds two new matrix entries (PyTorch 2.9.0 and 2.9.1) with CUDA 12.8.1 and Python 3.11 to both `build-axolotl` and `build-axolotl-cloud` jobs

Estimated Code Review Effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Simple configuration matrix additions with no logic changes
Repetitive, homogeneous pattern across both jobs
No structural or behavioral modifications to workflow logic

Possibly Related PRs

#2784: Modifies the same GitHub Actions workflow matrices for CUDA 12.8.1 PyTorch configurations in build jobs
#3049: Updates GitHub Actions workflow matrix in .github/workflows/main.yml for the build-axolotl-cloud job
#3221: Adds matching CUDA 12.8.1 / Python 3.11 / PyTorch 2.9.0 matrix entries to CI workflow configurations

Suggested Reviewers

NanoCode012
SalmanMohammadi

Pre-merge checks and finishing touches

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Title check	⚠️ Warning	The PR title 'Prepare for transformers v5 upgrade' does not match the actual changes, which only add CUDA 12.8.1 and PyTorch 2.9.x matrix entries to CI workflows.	Update the title to accurately reflect the changes, such as 'Add CUDA 12.8.1 and PyTorch 2.9.x matrix entries to build workflows' or clarify how these CI changes relate to transformers v5 preparation.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

codecov · 2025-11-20T14:17:10Z

Codecov Report

❌ Patch coverage is 59.13978% with 38 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/axolotl/core/trainers/base.py	9.09%	10 Missing ⚠️
src/axolotl/train.py	12.50%	7 Missing ⚠️
...ypatch/models/mistral3/mistral_common_tokenizer.py	0.00%	6 Missing ⚠️
src/axolotl/cli/merge_sharded_fsdp_weights.py	0.00%	3 Missing ⚠️
src/axolotl/utils/schemas/model.py	50.00%	3 Missing ⚠️
src/axolotl/utils/callbacks/perplexity.py	50.00%	2 Missing ⚠️
src/axolotl/cli/quantize.py	0.00%	1 Missing ⚠️
src/axolotl/core/builders/base.py	66.66%	1 Missing ⚠️
src/axolotl/core/builders/causal.py	0.00%	1 Missing ⚠️
src/axolotl/loaders/model.py	66.66%	1 Missing ⚠️
... and 3 more

📢 Thoughts on this report? Let us know!

github-actions · 2025-11-20T16:43:12Z

📖 Documentation Preview: https://697908f5e23666577cd750fa--resonant-treacle-0fd729.netlify.app

Deployed on Netlify from commit 875ec43

salmanmohammadi · 2025-12-03T16:39:40Z

 tokenizers>=0.22.1
-transformers==4.57.1
+transformers @ git+https://github.com/huggingface/transformers.git@main
 accelerate==1.11.0


I think we need to bump the accelerate version @winglian

NanoCode012 · 2025-12-15T11:56:12Z

Require #3318

…ipped

more fdsp_Version fixes for v5 fix version in fsdp_config fix aliasing fix fsdp_version check check fsdp_version is 2 in both places

* bump dep * use latest fbgemm, grab model config as part of fixture, un-skip test * import AutoConfig * don't need more problematic autoconfig when specifying config.json manually * add fixtures for argilla ultrafeedback datasets * download phi4-reasoning * fix arg * update tests for phi fast tokenizer changes * use explicit model types for gemma3 --------- Co-authored-by: Wing Lian <wing@axolotl.ai>

winglian force-pushed the transformers-v5 branch from e3ba4e9 to 2192d1f Compare November 20, 2025 14:25

salmanmohammadi reviewed Nov 20, 2025

View reviewed changes

Comment thread .github/workflows/tests.yml

NanoCode012 reviewed Dec 1, 2025

View reviewed changes

Comment thread tests/hf_offline_utils.py

winglian force-pushed the transformers-v5 branch 4 times, most recently from 6ac8f13 to 9767e28 Compare December 3, 2025 13:29

salmanmohammadi reviewed Dec 3, 2025

View reviewed changes

winglian force-pushed the transformers-v5 branch from 1ba104b to 0946a9b Compare December 8, 2025 15:38

winglian force-pushed the transformers-v5 branch from adae23f to af690a6 Compare December 17, 2025 15:12

winglian force-pushed the transformers-v5 branch 2 times, most recently from 0341778 to c0d0fac Compare December 31, 2025 03:11

winglian force-pushed the transformers-v5 branch from e35b0fb to 924c76b Compare January 14, 2026 16:44

winglian added 13 commits January 22, 2026 10:55

Prepare for transformers v5 upgrade

5cae58e

fix hf cli

da2b737

update for hf hub changes

2df923c

fix tokenizer apply_chat_template args

4e65312

remap include_tokens_per_second

b8b9038

fix tps

e78360c

handle migration for warmup

94b0973

use latest hf hub

896b711

Fix scan -> ls

7ef0ebc

fix import

4940304

fix for renaming of mistral common tokenizer -> backend

ebcda8f

update for fixed tokenziation for llama

b9d6edb

Skip phi35 tests for now

8dede22

winglian and others added 20 commits January 22, 2026 10:56

update fsdp version tests for new v5 behavior

0c31460

Fail multigpu tests after 3 failures

407a722

skip known v5 broken tests for now and cleanup

58becb8

bump deps

4ae9627

unmark skipped test

45bb110

re-enable test_fsdp_qlora_prequant_packed test

9f79f3d

increase multigpu ci timeout

7a7152e

skip broken gemma3 test

9306a5e

reduce timout back to original 120min now that the hanging test is sk…

b099bbc

…ipped

fix for un-necessary collator for pretraining with bsz=1

d84ee54

fix: safe_serialization deprecated in transformers v5 rc01 (#3318)

9f8644a

torch_dtype deprecated

83099b5

load model in float32 for consistency with tests

8188cd5

revert some test fixtures back

7ae4a58

use hf cache ls instead of scan

a8e8fb6

don't strip fsdp_version

2a42a9a

more fdsp_Version fixes for v5 fix version in fsdp_config fix aliasing fix fsdp_version check check fsdp_version is 2 in both places

fix: AutoModelForVision2Seq -> AutoModelForImageTextToText

50a4a28

chore: remove duplicate

b33feed

fix: attempt fix gemma3 text mode

c62eb25

winglian force-pushed the transformers-v5 branch from 70476e0 to c62eb25 Compare January 22, 2026 15:57

NanoCode012 and others added 5 commits January 23, 2026 16:05

chore: lint

54fec5e

ga release of v5

0cab7dd

need property setter for name_or_path for mistral tokenizer

3bcbefc

vllm not compatible with transformers v5

cd34054

setter for chat_template w mistral too

875ec43

winglian merged commit fc4e379 into main Jan 27, 2026
23 of 24 checks passed

winglian deleted the transformers-v5 branch January 27, 2026 22:08

coderabbitai Bot mentioned this pull request Mar 3, 2026

Add torch 2.10.0 to unit tests and use python 3.14 #3450

Merged

coderabbitai Bot mentioned this pull request Jun 7, 2026

build docker uv images w torch 2.11 and 2.12 #3715

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Prepare for transformers v5 upgrade#3272

Prepare for transformers v5 upgrade#3272
winglian merged 47 commits into
mainfrom
transformers-v5

winglian commented Nov 20, 2025 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented Nov 20, 2025 •

edited

Loading

Review skipped

Walkthrough

Changes

Estimated Code Review Effort

Possibly Related PRs

Suggested Reviewers

Uh oh!

codecov Bot commented Nov 20, 2025 •

edited

Loading

Uh oh!

github-actions Bot commented Nov 20, 2025 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

salmanmohammadi Dec 3, 2025 •

edited

Loading

Uh oh!

NanoCode012 commented Dec 15, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

winglian commented Nov 20, 2025 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Nov 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review skipped

Walkthrough

Changes

Estimated Code Review Effort

Possibly Related PRs

Suggested Reviewers

Pre-merge checks and finishing touches

Uh oh!

codecov Bot commented Nov 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions Bot commented Nov 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

salmanmohammadi Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NanoCode012 commented Dec 15, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

winglian commented Nov 20, 2025 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Nov 20, 2025 •

edited

Loading

codecov Bot commented Nov 20, 2025 •

edited

Loading

github-actions Bot commented Nov 20, 2025 •

edited

Loading

salmanmohammadi Dec 3, 2025 •

edited

Loading