strisunshinewentingwang/pytorch

Pull Requests Commits

[Dynamo+FSDP] Update benchmarks with use_orig_params=True (#90100)

davidberard98 committed 1 year ago

8f079b89

[Dynamo][Easy] capture more exceptions when import skip modules (#90338)

yanboliang committed 1 year ago

898b46d6

[Inductor] More robust stride and offset extraction from index expressions (#90184)

peterbell10 committed 1 year ago

71f27f76

[Inductor] Add test for Scheduler fusions (#90014)

peterbell10 committed 1 year ago

4f448779

[Quant][fx][bc-breaking] Remove unused functions in fx/utils.py (#90025)

andrewor14 committed 1 year ago

13fcc412

Revert "[MPS] Fix median_out_mps caching (#90326)"

pytorchmergebot committed 1 year ago

f28927e9

[quant] Add fused "q - qlinear - dq" operator with skipped quant op for output of linear (#89882)

jerryzh168 committed 1 year ago

887249b2

[Vulkan] Partially fix and then disable copying of vulkan quantized tensors to cpu (#90275)

manuelcandales committed 1 year ago

22e36334

[MPS] Fix median_out_mps caching (#90326)

malfet committed 1 year ago

23c192c3

[fx][passes] Implement annotate getitem node FX passes (#90237)

lequytra committed 1 year ago

b7690059

[quant][fx] Add support for matching constant in the custom matcher code in quantization (#90092)

jerryzh168 committed 1 year ago

0e182c94

as_strided: Fix default storage_offset for reference implementation (#89513)

peterbell10 committed 1 year ago

5caa27a3

Ensure that we fakeify tensor subclasses when they are initially tracked (#90009)

ezyang committed 1 year ago

3d4b92b1

Replace assertEqualIgnoreType in test_nn.py (#90242)

kit1980 committed 1 year ago

f09e7b5c

[CI] Relax CMake requirements (#90307)

malfet committed 1 year ago

6c195881

Add `TORCH_FAKE_TENSOR_DEBUG` use it to enable storage of traces on fake tensors at init time (#90215)

voznesenskym committed 1 year ago

3b9a386d

Remove logging.CODE (#90234)

williamwen42 committed 1 year ago

d224ac7f

Remove non-existing parameter from docstring (#90163)

kit1980 committed 1 year ago

14894a73

Disable dynamo tracing torchrec.distributed (#90087)

yanboliang committed 1 year ago

7e9a8a13

Hotfix to unblock TRT unit tests internally (#90313)

seemethere committed 1 year ago

27ad2605

[CUDA Graphs] Add option to dump a captured graph for debugging (#85519)

eqy committed 1 year ago

62e450d5

[Upstream _NamedOptimzer] Reland PR (89480) (#90293)

fduwjj committed 1 year ago

1abe264e

[FSDP] Clarify loss dtype check in `_test_fsdp_parity` (#90251)

awgu committed 1 year ago

7436b19e

[FSDP][BE] Clean up dead code from `clip_grad_norm_()` testing (#90250)

awgu committed 1 year ago

919e09f2

[FSDP] Test `use_orig_params=True` in `test_fsdp_ignored_modules.py` (#90290)

awgu committed 1 year ago

3b578edd

Fix uniform ref implementation (#90094)

yanboliang committed 1 year ago

25f39c1b

ShapeEnv.create_symbolic_sizes_strides_storage_offset (#89962)

ezyang committed 1 year ago

a1ab06ab

reland #89222: [Composable API] replicate: change to per module call, remove mark_root_module() (#90254)

Charlie Yan committed 1 year ago

e818c366

[FSDP] Fix accidental change in `_test_fsdp_parity` (#90252)

awgu committed 1 year ago

bd9ad89a

Log1p for complex in CPU (#89691)

mfkasim1 committed 1 year ago

ce212628

Newer Older