Rebase DBraun/sequence-layers mlx.001 onto main; bump to 0.3.dev1#15

Closed

JulianSlzr wants to merge 17 commits into

google:mainfrom

JulianSlzr:mlx.001

Commits on Jun 2, 2026

add MLX layers

DBraun
authored and
JulianSlzr
committed
Update attention.py

DBraun
authored and
JulianSlzr
committed
Create depthformer.py

DBraun
authored and
JulianSlzr
committed
Squashed commit of the following:

DBraun
authored and
JulianSlzr
committed
more efficient attention

DBraun
authored and
JulianSlzr
committed
Update attention.py

DBraun
authored and
JulianSlzr
committed
Update position.py

DBraun
authored and
JulianSlzr
committed
add to_quantized

DBraun
authored and
JulianSlzr
committed
optimizations

DBraun
authored and
JulianSlzr
committed
attention optimizations

DBraun
authored and
JulianSlzr
committed
Update convolution.py

DBraun
authored and
JulianSlzr
committed
Updates pyproject w/ recurrentgemma fix

chrisdonahue-goog
authored and
JulianSlzr
committed
test(mlx): Add ring buffer wrap-around tests (expected to fail)
JulianSlzr
committed
fix(mlx): Implement write-after-read for attention ring buffer
JulianSlzr
committed
Adds Parallel.Config on mlx side

KEHANG
authored and
JulianSlzr
committed
Fixes cache poisoning for dtype in simple.py

KEHANG
authored and
JulianSlzr
committed
chore(release): bump version to 0.3.dev1
JulianSlzr
committed