pytorch/xla

Pull Requests Commits

Update to ubuntu latest (#8979)

qihqi committed 312 days ago

Verified a37a5852

@assume_pure (#8962)

tengyifei committed 312 days ago

Verified cf764ffa

Avoid re-computing computation hashes (#8976)

rpsilva-aws committed 312 days ago

Verified bcca5eba

Disable one splash attention test (#8970)

zpcore committed 313 days ago

Verified ba5cd4cb

Add tuned parameters for Qwen/Qwen2.5-32B (#8966)

yarongmu-google committed 316 days ago

Verified 4eea5e13

Adapt Splash Attention from TorchPrime (#8911)

zpcore committed 317 days ago

Verified 45830519

Make scan-based GRU support `batch_first` parameter. (#8964)

iwknow committed 317 days ago

Verified aad87e61

[call_jax] support returning PyTree from the JAX function (#8957)

tengyifei committed 317 days ago

Verified 297f5de6

Remove deprecated typing module in ansible flow (#8955)

tengyifei committed 317 days ago

Verified e77bd2f6

Add test to ensure scan-based and the standard GRU are interchangeable. (#8949)

iwknow committed 318 days ago

Verified d4d997dc

Update to pytorch 2.6 (#8944)

qihqi committed 319 days ago

Verified 085f8bc1

Pin update to 20250406 (#8945)

tengyifei committed 319 days ago

Verified 0837fac3

Use pages_per_seq * page_size instead of directly passing max_model_len (#8950)

yaochengji committed 320 days ago

Verified 9ec626e6

Fix ragged_paged_attention op signature (#8943)

yaochengji committed 320 days ago

Verified 80929a91

add rc4 trigger (#8941)

zpcore committed 322 days ago

Verified ac9a39f4

Add block size table for ragged_paged_attention (#8942)

yaochengji committed 323 days ago

Verified cbff76f5

Add alternative dynamo backend (#8893)

qihqi committed 323 days ago

Verified fb6038de

Trigger build for 2.6.1 patch (#8938)

zpcore committed 323 days ago

Verified 379ebd5f

Update and rename torch_xla2.yml to torchax.yml (#8936)

tengyifei committed 323 days ago

Verified 9e23ca85

Rewrite scan-based GRU based on nn.GRU (#8914)

iwknow committed 323 days ago

Verified 145c33ae

Fix up pin update (#8935)

tengyifei committed 323 days ago

Verified 980ead54

Include torchax in torch_xla (#8895)

tengyifei committed 324 days ago

Verified bb5c6a0b

pin update (#8908)

lsy323 committed 324 days ago

Verified 6400e169

[ragged-paged-attn] Unify kv strided load to one. (#8929)

bythew3i committed 325 days ago

Verified c044c691

Add heuristic default block sizes for different cases in ragged attention kernel (#8922)

yaochengji committed 325 days ago

Verified 6c3f2313

Clean up deprecated APIs (#8927)

zpcore committed 325 days ago

Verified f1fe8bc0

Use f32 scratch for output so we only need to transfer output with desired dtype back to HBM. (#8924)

vanbasten23 committed 325 days ago

Verified f0881b5a

[torchax] Support linalg.det and linalg.lu_solve (#8872)

dvhg committed 326 days ago

Verified 2d6f57a3

Fix cuda dependency not found (#8903)

zpcore committed 326 days ago

Verified dbb03aa1

Lower `isneginf()`. (#8912)

ysiraichi committed 326 days ago

Verified cba9ff9f

Newer Older