strisunshinewentingwang/pytorch

Pull Requests Commits

fix the formatting of runtime error msg in prims _cat_meta (#92124)

dshi7 committed 1 year ago

896b6d87

Shard mac to 3 (#91277)

clee2000 committed 1 year ago

703265e5

Don't allow recomputing a node that *must* be materialized in the backwards pass (#90896)

Chillee committed 1 year ago

d6c3468f

Fix GroupNorm backward prop on CUDA (#92671)

malfet committed 1 year ago

97b7e4cd

[CUDA][CUBLAS][BFloat16] Tenatively disable reduced precision reductions for some matmul tests (#92599)

eqy committed 1 year ago

8c0289a6

[inductor] Lower torch.exp2 and use it for torch.pow(2, x) (#92632)

peterbell10 committed 1 year ago

56440594

Add GHA side support for ciflow/inductor-perf-test-nightly (#92693)

ezyang committed 1 year ago

5a134440

Create autograd Function for aot_autograd backward only when needed (#92688)

soulitzer committed 1 year ago

a3efa9d7

[PT-D][checkpoint] Resolve no such file or directory issue when checkpointing on multi hosts (#92553)

wz337 committed 1 year ago

eee2869e

Foreach gradient clipping (#91846)

milesial committed 1 year ago

e4d83d54

Clean up argparser help (benchmarks/dynamo/distributed.py) (#92687)

wconstab committed 1 year ago

44b7a0b7

Deprecate capture hooks except distributed use case (#92653)

soulitzer committed 1 year ago

9db4323e

Delete get_pyobj() entirely (#92638)

ezyang committed 1 year ago

c4501593

Mark test monitoring as an optional process (#92658)

huydhn committed 1 year ago

56107660

Revert "Run dynamo/test_dynamic_shapes serially (#92215)"

pytorchmergebot committed 1 year ago

8b3e35ea

update vmap to accept nones (#91644)

srossross committed 1 year ago

fb3d9f39

[Dynamo] Preserve source_fn in node.meta (#92399)

SherlockNoMad committed 1 year ago

2fb328eb

[decomp] Use new squeeze.dims overload in decompositions (#91602)

peterbell10 committed 1 year ago

dd760c98

logaddexp2: Use log1p and exp2 (#92116)

peterbell10 committed 1 year ago

2af2952c

lint fix (#92685)

ZainRizvi committed 1 year ago

67bb5236

Revert "Add meta kernel coverage for aten.unsafe_split, aten.unsafe_chunk (#92608)"

pytorchmergebot committed 1 year ago

2891cecd

Update android/README.md, how to build pytorch android from source (#92356)

vfdev-5 committed 1 year ago

215f4fc3

[optim][adagrad] group tensors in foreach to maximize perf (#92362)

janeyx99 committed 1 year ago

b2ca2c86

Revert "Add `--timing` flag, phase timing to @dynamo_timed (#92637)"

pytorchmergebot committed 1 year ago

44132cc4

Optimized vertical flip using memcpy (#89414)

vfdev-5 committed 1 year ago

5ac22782

Log accuracy failure in more cases (#92645)

ezyang committed 1 year ago

38735753

Logcumsumexp for complex in CPU and CUDA (#90847)

mfkasim1 committed 1 year ago

64985123

Add meta kernel coverage for aten.unsafe_split, aten.unsafe_chunk (#92608)

tugsbayasgalan committed 1 year ago

4386f317

[vmap] unsafe_split : batching rule and OpInfo (#92291)

kshitij12345 committed 1 year ago

274958ef

Fix performance smoke test script bug (#92660)

weiwangmeta committed 1 year ago

f6acd95a

Newer Older