Pull Requests pytorch/xla

Update tuned block table and add benchmarking/autotune scripts for quantized matmul pallas kernel.

#9322 opened 2025-06-10 02:24 by vanbasten23

Add DLRM DCN v2 with JAX Sparsecore

#9287 opened 2025-06-03 20:01 by chandrasekhard2

WIP - do not review; Support python 3.12

#9274 opened 2025-06-01 16:24 by zhanyong-wan

Migrate `torch_xla.device()` to `torch.device('xla')`

#9253 opened 2025-05-27 22:03 by ghpvnist

Switch bazel to default to Clang for C/C++ compilation

#9169 opened 2025-05-14 18:26 by zhanyong-wan

[Test] Add torch pin 152601

#9167 opened 2025-05-14 18:08 by bhavya01

(Do not submit) Set StableHLO Compilation to true for CI

#9141 opened 2025-05-12 09:33 by GleasonK

Lower clamp

#9123 opened 2025-05-08 23:25 by ghpvnist

Move requires_jax to inner flash_attention functions

#9098 opened 2025-05-05 23:47 by bhavya01

Add features needed for vllm

#9092 opened 2025-05-05 18:11 by wenxindongwork

Check the scan op input for requires_grad

#9083 opened 2025-05-02 20:33 by haifeng-jin

Assert on empty PJRT buffers

#9062 opened 2025-04-29 23:58 by rpsilva-aws

(Draft - PT27) Extend device data node binding to not override tensor IDs

#9052 opened 2025-04-28 23:47 by rpsilva-aws

reduce redundant graph in collective op distributed

#9044 opened 2025-04-25 22:24 by zpcore

[DRAFT] call_jax sets the JAX contextual mesh

#9043 opened 2025-04-25 19:03 by tengyifei

Update test_export_fx_passes.py

#8933 opened 2025-04-03 21:25 by avikchaudhuri

add CreateGlobalShardedData prototype

#8932 opened 2025-04-03 21:12 by pgmoka

test on debian12 DO_NOT_MERGE

#8928 opened 2025-04-02 22:59 by zpcore

Make GPU CUDA plugin require JAX

#8919 opened 2025-04-01 22:06 by tengyifei

Lower `upsample_bilinear2d_backward` operation.

#8916 opened 2025-04-01 19:45 by ysiraichi

Add Experimental limited sparse embedding bag

#8905 opened 2025-03-28 22:22 by amjames

[ragged-paged-attn] Use hidden states in kv cache and support any num…

#8891 opened 2025-03-26 23:23 by zpcore

[DRAFT/WIP] Add top-p masking

#8871 opened 2025-03-21 23:17 by hyeygit

Showcase jax.grad in torch_xla

#8800 opened 2025-03-05 22:02 by zpcore

Repro ragged paged attn kernel

#8752 opened 2025-02-26 20:50 by vanbasten23

Replace setup.py with pyproject.toml

#8744 opened 2025-02-26 09:15 by ManfeiBai

Follow up on ragged kernel wrapper

#8737 opened 2025-02-24 00:40 by vanbasten23

Document how to debug the dispatcher

#8712 opened 2025-02-15 01:03 by tengyifei

enable tensor.copy_ with torch native cpu tensor as source

#8682 opened 2025-02-06 01:51 by yaochengji

Lower cummin op lowering

#8565 opened 2025-01-14 03:40 by zyy-martin