quantizers`
#44412

Merged

chore(typing): Add type checking to `src/transformers/quantizers` #44412

tarekziade merged 43 commits into main from tarekziade-typing-quantizers

tarekziade force pushed from 7f3d268c to 80499115 15 days ago

tarekziade changed the title ~~chore(typing): Add type checking to `src/transformers/generation`~~ chore(typing): Add type checking to `src/transformers/quantizers` 15 days ago

tarekziade assigned

tarekziade 15 days ago

tarekziade force pushed from 80499115 to 87ac38b0 14 days ago

tarekziade force pushed from 87ac38b0 to ea212202 13 days ago

tarekziade commented on 2026-03-05

tarekziade requested a review from

vasqu 13 days ago

vasqu commented on 2026-03-06

tarekziade force pushed from 087b3ccd to 2139b1e6 12 days ago

tarekziade force pushed from 5714a089 to b202c773 12 days ago

tarekziade requested a review from

vasqu 12 days ago

tarekziade force pushed from dbe048c3 to ee5781d6 10 days ago

vasqu approved these changes on 2026-03-09

tarekziade force pushed from 9cbadfa0 to 7e6c53f4 9 days ago

vasqu commented on 2026-03-10

tarekziade force pushed from dacbfde5 to e8d63cc3 8 days ago

SunMarc approved these changes on 2026-03-10

chore(typing): add `ty` type checking for src/transformers/generation

6b3d450c

extend make file

78eb16e8

chore(typing): further reduce ty diagnostics (36→18), replace assert …

c0f9ac9c

more tweaking

f80e4079

fix get word size

52b10676

date

ab2bd374

cast value

afd50484

@zucchini-nlp reviews

32aa9e6c

model_kwargs cannot be None, fix type annotations

1cf87929

removed redundancy

ae423792

bypass type checking to reduce code complexity

46ba6171

de-vectorize

c501fb3f

unified _typing

07e44ed4

explain why we pollute inheritance with a protocol

0d44bffd

do not import torch for type checking, it breaks on CI logging usage

f9f6e280

Add check_types.py to wrap ty type checking

3414063b

improved ty usage

0fc155f2

chore(typing): Add type checking to src/transformers/quantizers

ad00556a

fix merge error

10361eda

reworked the typing using previous PR best practices

c86b5c64

format and torch.neuron fix

18592a01

revert bad upstream merge

32d18468

narrow isinstance

4f9f2c7e

cleaner const, also 1.23.99 vs 1.24.0

19cdf6dc

clean up old cast

ae569bcc

removing leftover

78b0b015

lets just use setattr

3288c138

tweaked one more optimum version

f75067c2

setattr again

c8f41de0

more setattr/getattr patyterns

d40101d6

more cleanups

9d5dd5ea

fmt

756794f1

cleanups from rebase abd ty 0.20.0

7e3cb429

fix ignores

d044b383

const for torchao version

1c9bd9fc

simplify call

d7e87d57

use is_torch_greater_or_equal

a0213d83

use type guard

c4dd7b41

remove the getattr() pattenr

f5f9cc2b

reworked the approach

a0b2785f

Update src/transformers/quantizers/quantizer_sinq.py

30c3100c

Update src/transformers/quantizers/quantizer_quanto.py

d29ec940

tarekziade force pushed from ce0ca963 to d29ec940 8 days ago

class is there we can use it

331b3687

tarekziade merged 8464284d into main 8 days ago

tarekziade deleted the tarekziade-typing-quantizers branch 8 days ago

Reviewers

SunMarc

vasqu

Cyrilvallez

Assignees

tarekziade

Labels

None yet

Milestone

No milestone

transformers chore(typing): Add type checking to `src/transformers/quantizers` #44412 Merged

chore(typing): Add type checking to `src/transformers/quantizers` #44412

transformers
chore(typing): Add type checking to `src/transformers/quantizers`
#44412

Merged