-
Notifications
You must be signed in to change notification settings - Fork 3.1k
Pull requests: jax-ml/jax
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix logic for checking supported cross-host device transfers, since the (unsupported) TPU PjRt C API client also contains the string "TFRT TPU".
#29290
opened Jun 6, 2025 by
copybara-service
bot
Loading…
Delete extraneous
block_until_ready
s on JAX dispatch benchmarks.
#29289
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Improve device assignment string format and add AbslStringify implementation.
#29287
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Implemented cross-host memory transfer on GPU.
#29286
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Add fast-path for non-concrete Tracers in is_constant_dim to lower trace-time overhead from Exception creation
#29285
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Expose local/global
ExchangeTopologies
timeouts for PJRT CPU client.
#29283
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Remove forward_compat as it is past the support date.
#29282
opened Jun 5, 2025 by
copybara-service
bot
Loading…
[Mosaic] Adds both direct (where hardware can) support for int8 Transpose, and canonicalization for where it cannot.
#29280
opened Jun 5, 2025 by
copybara-service
bot
Loading…
#sdy don't make JAX export use
SdyRoundTripExportPipeline
to stringify attributes and convert ops to StableHLO CustomCallOp
s and back.
#29272
opened Jun 5, 2025 by
copybara-service
bot
Loading…
[pallas:mgpu] Optionally skip allocating registers in warp specialized pipelining.
#29271
opened Jun 5, 2025 by
copybara-service
bot
Loading…
[Mosaic GPU] Extract the type-related logic out of
reinterpret_smem_ref
.
#29270
opened Jun 5, 2025 by
copybara-service
bot
Loading…
[Mosaic GPU] Handle
None
transforms in swizzle_and_transforms_from_transforms_attr
#29269
opened Jun 5, 2025 by
copybara-service
bot
Loading…
[Mosaic GPU] Add support for lowering
2xbf16 -> 2xf8e4m3fn
converts.
#29264
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Change
use_shardy_partitioner
default to None
instead of False
.
#29262
opened Jun 5, 2025 by
copybara-service
bot
Loading…
Fixed NaN propagation in optimal_step_size using nanmin/nanmax
#29261
opened Jun 5, 2025 by
SuriyaaMM
Loading…
Fix handling of empty arrays in ufunc.reduce/accumulate
#29248
opened Jun 4, 2025 by
DanisNone
Loading…
[Pallas][Mosaic GPU] Add collective (CTA-pair) MMAs to blackwell matmul kernel.
#29244
opened Jun 4, 2025 by
copybara-service
bot
Loading…
[Pallas][Mosaic GPU] Expose partitioned collective loads to copy_gmem_to_smem.
#29243
opened Jun 4, 2025 by
copybara-service
bot
Loading…
Move converting a mixed SDY+GSPMD module down to a pure GSPMD targeting module to
MlirToXlaComputation
.
#29234
opened Jun 4, 2025 by
copybara-service
bot
Loading…
Previous Next
ProTip!
Type g i on any issue or pull request to go back to the issue listing page.