Bump torchao >=0.15 and fix quantization CI by SunMarc · Pull Request #44604 · huggingface/transformers

SunMarc · 2026-03-11T16:44:38Z

What does this PR do?

This PR fix the quantization CI :

One major point in this PR is that I bump the min version of torchao to 0.15.0, thus we can clean a bunch of things.

SunMarc · 2026-03-11T16:45:19Z

run-slow: metal, mxfp4, fp_quant_integration, autoround

github-actions · 2026-03-11T16:46:41Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/autoround", "quantization/fp_quant_integration", "quantization/metal", "quantization/mxfp4"]

HuggingFaceDocBuilderDev · 2026-03-11T16:54:49Z

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.

github-actions · 2026-03-11T17:06:35Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	53bfea2b	workflow commit (merge commit)
PR	5c21405c	branch commit (from PR)
main	e45078f2	base commit (on `main`)

Quantization CI Report

❌ 3 new failed tests from this PR 😭

metal:
tests/quantization/metal/test_metal.py::MetalSlowIntegrationTest::test_quantized_model (❌ ⟹ ❌)
mxfp4:
tests/quantization/mxfp4/test_mxfp4.py::Mxfp4QuantizerTest::test_quantizer_validation_no_accelerator (❌ ⟹ ❌)
tests/quantization/mxfp4/test_mxfp4.py::Mxfp4QuantizerTest::test_quantizer_validation_order_dequantize_before_accelerator_check (❌ ⟹ ❌)

SunMarc · 2026-03-11T17:35:16Z

run-slow: bnb, fp_quant_integration, metal, mxfp4

github-actions · 2026-03-11T17:36:29Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/bnb", "quantization/fp_quant_integration", "quantization/metal", "quantization/mxfp4"]

github-actions · 2026-03-11T18:01:58Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	3767b4e5	workflow commit (merge commit)
PR	e99da79a	branch commit (from PR)
main	0a9fdeaa	base commit (on `main`)

✅ No failing test specific to this PR 🎉 👏 !

SunMarc · 2026-03-12T15:30:23Z

run-slow: bnb, fp_quant_integration, quark_integration

github-actions · 2026-03-12T15:31:49Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/bnb", "quantization/fp_quant_integration", "quantization/quark_integration"]

github-actions · 2026-03-12T15:52:23Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	7fb024b1	workflow commit (merge commit)
PR	6c9c89d9	branch commit (from PR)
main	e5a861d3	base commit (on `main`)

✅ No failing test specific to this PR 🎉 👏 !

SunMarc · 2026-03-12T17:08:30Z

run-slow: bnb

github-actions · 2026-03-12T17:09:53Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/bnb"]

github-actions · 2026-03-12T17:31:10Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	7fb024b1	workflow commit (merge commit)
PR	6c9c89d9	branch commit (from PR)
main	e5a861d3	base commit (on `main`)

✅ No failing test specific to this PR 🎉 👏 !

SunMarc · 2026-03-13T16:47:23Z

run-slow: torchao

SunMarc · 2026-03-13T17:11:26Z

run-slow: torchao_integration

github-actions · 2026-03-13T17:12:51Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/torchao_integration"]

github-actions · 2026-03-13T17:29:02Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	464c2078	workflow commit (merge commit)
PR	234cbb11	branch commit (from PR)
main	f2f7c894	base commit (on `main`)

Quantization CI Report

❌ 1 new failed tests from this PR 😭

torchao_integration:
tests/quantization/torchao_integration/test_torchao.py::TorchAoSerializationTest::test_serialization_accelerator_5 (✅ ⟹ ❌)

github-actions · 2026-03-13T18:45:08Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	82d7a58c	workflow commit (merge commit)
PR	4031b7ae	branch commit (from PR)
main	064f0e97	base commit (on `main`)

Quantization CI Report

❌ 1 new failed tests from this PR 😭

torchao_integration:
tests/quantization/torchao_integration/test_torchao.py::TorchAoAcceleratorTest::test_int4wo_offload (❌ ⟹ ❌)

ydshieh

👍

SunMarc · 2026-03-16T10:31:43Z

run-slow: bnb, fp_quant_integration, metal, mxfp4, quark_integration, torchao_integration

github-actions · 2026-03-16T10:32:59Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/bnb", "quantization/fp_quant_integration", "quantization/metal", "quantization/mxfp4", "quantization/quark_integration", "quantization/torchao_integration"]

Torchao made some API changes, which have to be reflected in the tests. Moreover, for this to pass, we also need transformers to make the corresponding adjustments: huggingface/transformers#44604 While working on this, I migrated the tests from unittest to pytest style.

github-actions · 2026-03-16T11:00:24Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	81e5d278	workflow commit (merge commit)
PR	f2753d47	branch commit (from PR)
main	d64a6d67	base commit (on `main`)

Quantization CI Report

❌ 1 new failed tests from this PR 😭

torchao_integration:
tests/quantization/torchao_integration/test_torchao.py::TorchAoAcceleratorTest::test_int4wo_offload (❌ ⟹ ❌)

…s-tests

github-actions · 2026-03-16T12:46:46Z

View the CircleCI Test Summary for this PR:

https://huggingface.co/spaces/transformers-community/circle-ci-viz?pr=44604&sha=d63de4

SunMarc · 2026-03-16T12:55:56Z

run-slow: bnb, fp_quant_integration, metal, mxfp4, quark_integration, torchao_integration, fouroversix

github-actions · 2026-03-16T12:57:11Z

Workflow Run ⚙️

This comment contains run-slow, running the specified jobs:

models: []
quantizations: ["quantization/bnb", "quantization/fp_quant_integration", "quantization/metal", "quantization/mxfp4", "quantization/quark_integration", "quantization/torchao_integration"]

github-actions · 2026-03-16T13:21:00Z

CI Results

Workflow Run ⚙️

Commit Info

Context	Commit	Description
RUN	9b030796	workflow commit (merge commit)
PR	d63de446	branch commit (from PR)
main	30adbf3f	base commit (on `main`)

✅ No failing test specific to this PR 🎉 👏 !

SunMarc · 2026-03-16T15:32:04Z

Omg everything is green

github-actions · 2026-03-16T15:33:17Z

[For maintainers] Suggested jobs to run (before merge)

run-slow: bnb, fp_quant_integration, metal, mxfp4, quark_integration, torchao_integration

…s-tests

github-actions · 2026-03-16T15:52:58Z

[For maintainers] Suggested jobs to run (before merge)

run-slow: bnb, fp_quant_integration, metal, mxfp4, quark_integration, torchao_integration

Fix

5c21405

update expected

b88d0ed

SunMarc added 3 commits March 11, 2026 17:29

fix fp_quant

d395127

fix metal

6409757

fix mxfp4

e99da79

fix

6c9c89d

fix bnb

41b9866

SunMarc added 5 commits March 12, 2026 18:12

style

10cc83c

Merge remote-tracking branch 'origin/main' into fix-quants-tests

ef941cc

bump torchao to 0.15.0 min to have safetensor support by default

f80a102

style

abc30c6

fix torchao tests

234cbb1

Fix !

03c24a8

Merge branch 'main' into fix-quants-tests

f2753d4

SunMarc changed the title ~~Fix quantization CI~~ Bump torchao >=0.15 and fix quantization CI Mar 16, 2026

ydshieh approved these changes Mar 16, 2026

View reviewed changes

BenjaminBossan mentioned this pull request Mar 16, 2026

FIX Broken tests with torchao >= 0.15 huggingface/peft#3101

Open

SunMarc added 4 commits March 16, 2026 12:17

update dockerfile

54ac4b1

build quant docker

c0c2663

Merge remote-tracking branch 'origin/fix-quants-tests' into fix-quant…

d1718a0

…s-tests

maybe this ?

d63de44

Merge branch 'main' into fix-quants-tests

6c78089

SunMarc enabled auto-merge March 16, 2026 15:32

SunMarc disabled auto-merge March 16, 2026 15:32

SunMarc added 2 commits March 16, 2026 15:51

uncomment

941689e

Merge remote-tracking branch 'origin/fix-quants-tests' into fix-quant…

7a42c83

…s-tests

SunMarc enabled auto-merge March 16, 2026 15:53

SunMarc added this pull request to the merge queue Mar 16, 2026

Merged via the queue into main with commit bfe00b4 Mar 16, 2026
16 of 17 checks passed

SunMarc deleted the fix-quants-tests branch March 16, 2026 16:07

Conversation

SunMarc commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What does this PR do?

Uh oh!

SunMarc commented Mar 11, 2026

Uh oh!

github-actions bot commented Mar 11, 2026

Uh oh!

HuggingFaceDocBuilderDev commented Mar 11, 2026

Uh oh!

github-actions bot commented Mar 11, 2026

CI Results

Commit Info

Quantization CI Report

Uh oh!

SunMarc commented Mar 11, 2026

Uh oh!

github-actions bot commented Mar 11, 2026

Uh oh!

github-actions bot commented Mar 11, 2026

CI Results

Commit Info

Uh oh!

SunMarc commented Mar 12, 2026

Uh oh!

github-actions bot commented Mar 12, 2026

Uh oh!

github-actions bot commented Mar 12, 2026

CI Results

Commit Info

Uh oh!

SunMarc commented Mar 12, 2026

Uh oh!

github-actions bot commented Mar 12, 2026

Uh oh!

github-actions bot commented Mar 12, 2026

CI Results

Commit Info

Uh oh!

SunMarc commented Mar 13, 2026

Uh oh!

SunMarc commented Mar 13, 2026

Uh oh!

github-actions bot commented Mar 13, 2026

Uh oh!

github-actions bot commented Mar 13, 2026

CI Results

Commit Info

Quantization CI Report

Uh oh!

github-actions bot commented Mar 13, 2026

CI Results

Commit Info

Quantization CI Report

Uh oh!

ydshieh left a comment

Choose a reason for hiding this comment

Uh oh!

SunMarc commented Mar 16, 2026

Uh oh!

github-actions bot commented Mar 16, 2026

Uh oh!

github-actions bot commented Mar 16, 2026

CI Results

Commit Info

Quantization CI Report

Uh oh!

github-actions bot commented Mar 16, 2026

Uh oh!

SunMarc commented Mar 16, 2026

Uh oh!

github-actions bot commented Mar 16, 2026

Uh oh!

github-actions bot commented Mar 16, 2026

CI Results

Commit Info

Uh oh!

SunMarc commented Mar 16, 2026

Uh oh!

SunMarc commented Mar 11, 2026 •

edited

Loading