strisunshinewentingwang/pytorch

Pull Requests Commits

[dynamo] Fix list contains check (#95092)

angelayi committed 1 year ago

ec10d23c

Fix a bug in nesting check_sparse_tensor_invariants context managers (#95372)

pearu committed 1 year ago

0c069449

Revert "Implement sparse semantics support in gradcheck (#94714)" (#95386)

ZainRizvi committed 1 year ago

808879ec

[mergebot] Fix for pagination error (#95333)

clee2000 committed 1 year ago

fb3ff774

Revert "During export, generate Python TENSOR_MATCH guards (#94970)"

pytorchmergebot committed 1 year ago

254b161d

Revert "Update docs that Parameters are immune to no_grad mode (#95232)"

pytorchmergebot committed 1 year ago

cb6e38d8

[MPS] Fix LSTM backward and forward pass (#95137)

alexdremov committed 1 year ago

b9e95158

[MPS] Fix view op slicing for 2nd dim in case of 0 offset (#95381)

DenisVieriu97 committed 1 year ago

86efa104

Update docs that Parameters are immune to no_grad mode (#95232)

janeyx99 committed 1 year ago

5783cee2

Add knobs for globally turning off 0/1 specialization and duck shaping (#95352)

ezyang committed 1 year ago

af202aea

Stop subclassing sympy Symbol (#95313)

ezyang committed 1 year ago

94fd063f

Add warn-once deprecation warning to legacy sparse constructors (#94850)

pearu committed 1 year ago

cece63f1

[autograd] disable backward/grad for complex scalar output (#92753)

kshitij12345 committed 1 year ago

3b966a6c

[Dynamo] No graph break on calling dict & collections.OrderedDict() (#95250)

yanboliang committed 1 year ago

b5ff41a4

std/var: support floating point correction value (#94073)

peterbell10 committed 1 year ago

bc438af6

SymFloat: Expose comparison operators in C++ API (#94812)

peterbell10 committed 1 year ago

56aed2a6

using float type to do the computation of norm reduce for cpu half and bfloat16 dtype (#95166)

XiaobingSuper committed 1 year ago

5730cabd

[DCP] Update DCP to use the updated FSDP optim state_dict APIs (#95303)

wz337 committed 1 year ago

6912cf40

Fix OOMing periodic shards (#95246)

ZainRizvi committed 1 year ago

c97275ac

[PTD][DCP] Add fsdp checkpoint example (#95258)

wz337 committed 1 year ago

bdb78e52

[vision hash update] update the pinned vision hash (#95340)

pytorchmergebot committed 1 year ago

c594a32f

[SDPA] Fix bug in parsing scaled_dot_product_attention arguments (#95311)

drisspg committed 1 year ago

29c235e5

use 4 warps for small block config in mm (#95339)

Natalia Gimelshein committed 1 year ago

8e391c73

[inductor] enable `test_nll_loss_forward_dynamic_shapes` (#95176)

nkaretnikov committed 1 year ago

ba8ff4be

Fix disbale typos (#95322)

kit1980 committed 1 year ago

f98733e9

Bugfix nested mem_efficient path in SDPA when E_qk != E_v (#95330)

mikaylagawarecki committed 1 year ago

586ac98c

[FSDP][Docs] Re-add why reg. post-bwd hook on 1st forward (#95326)

awgu committed 1 year ago

78175cee

Avoid FPE when running batch norm with zero batch size. (#95324)

ezyang committed 1 year ago

f247129f

coreml_delegate - Add input shape in error when throwing from predicting (#95249)

kylesyoon committed 1 year ago

a257486b

[pt2][inductor] switch dinfo representation (#95302)

nmacchioni committed 1 year ago

3ebab9ae

Newer Older