strisunshinewentingwang/pytorch

Pull Requests Commits

add BFloat16 support for logcumsumexp on CPU (#72694)

jiayisunx committed 2 years ago

97deda4f

Lazy codegen change for xla (#76717)

JackCaoG committed 2 years ago

e36a8c1f

Enable BFloat16 ELU, SELU and CELU in CPU path (#62546)

yanbing-j committed 2 years ago

4f82f439

Use _process_group in ReplicatedTensor and ShardedTensor.

pritamdamania87 committed 2 years ago

bdbb7fe3

Fix pre-compiled header

peterbell10 committed 2 years ago

29e615f7

[chalf] getitem (#77339)

kshitij12345 committed 2 years ago

9e3eb329

add a new at::lift operator, fix torch.tensor for functionalization

bdhirsh committed 2 years ago

47dd092b

Fix docs "quantization" instead of "quantiztion" (#77300)

ftorres16 committed 2 years ago

e06400e7

Sparse BSR: Enable addmm, addmv, triangular_solve for BSR layout (#77255)

IvanYashchuk committed 2 years ago

09be44de

Sparse CSR CUDA: add batched support for torch.sparse.sampled_addmm

IvanYashchuk committed 2 years ago

d1beda53

[RecordFunction] Don't lazily construct the guts of RecordFunction. (#76016)

Taylor Robie committed 2 years ago

689df639

nvfuser opinfo test fixes masked_var/std (#77273)

jjsjann123 committed 2 years ago

b010c345

[complex32] fft support (cuda only) (#74857)

kshitij12345 committed 2 years ago

ada65fdd

Revert autoformat of tools/fast_nvcc/fast_nvcc.py

suo committed 2 years ago

b825e1d4

make clamp_min/max use minimum/maximum kernels, make clamp* correctly propagate nans (#77306)

ngimel committed 2 years ago

257c55f4

add mixed data type mode for LayerNorm forward path

mingfeima committed 2 years ago

3b56efd4

[quant][core][gpu][feature] Implemented quantized conv1d cudnn op

dzdang committed 2 years ago

af80329c

[black][codemod] formatting changes from black 22.3.0

amyreese committed 2 years ago

d973ece8

Added logsumexp decomposition (#77219)

Chillee committed 2 years ago

c25bdeea

Add RefineTypes JIT pass for Tuple (#76919)

henrytwo committed 2 years ago

f6eb8117

torch/deadlockdetection: add TORCH_DISABLE_DEADLOCK_DETECTION env for use with torch deploy (#77270)

d4l3k committed 2 years ago

2881e0ea

[FSDP] Do not check fwd order in eval mode

awgu committed 2 years ago

e912d243

docs: Add section for Hardware / Software support (#77304)

seemethere committed 2 years ago

e8f53ad1

lshift and rshift stop support floating types (#77146)

zasdfgbnm committed 2 years ago

cc9d0f30

improve LayerNorm bfloat16 performance on CPU

mingfeima committed 2 years ago

166a466e

Remove old grouped_accessor operator in static runtime (#77193)

Yixin Bao committed 2 years ago

355ca32b

[shard] Add deepcopy for ShardedTensor

wanchaol committed 2 years ago

03036470

ci: Move lint to be run on linux.20_04.16x runners

seemethere committed 2 years ago

440bc815

[static-runtime] plug in caching logic (#76210)

tenpercent committed 2 years ago

bcbc1a3e

Separate input moving to utils file

rohan-varma committed 2 years ago

bbb1f106

Newer Older