pytorch/benchmark

Pull Requests Commits

Don't shard torchrec_dlrm by default

ezyang committed 2 years ago

978f8d30

add option to run group_fusion

chaekit committed 2 years ago

8a0f5e36

Move hf_MPT_7b_instruct to canary_models (#1810)

xuzhao9 committed 2 years ago

f8823121

checkout submodule one more time (#1808)

xuzhao9 committed 2 years ago

dfe152c5

Add MPT-7B-instruct HuggingFace Model (#1773)

apsonawane committed 2 years ago

0b7147fd

all the llamas in canary (#1803)

msaroufim committed 2 years ago

e7ca300f

Make `test.py` device agnostic (#1805)

hmellor committed 2 years ago

8dbc4ee0

Generalise BERT_pytorch's device option (#1801)

hmellor committed 2 years ago

00c01dfe

Update `detectron2` to fix Pillow errors (#1799)

hmellor committed 2 years ago

673279dd

Move jit from cpu userbenchmark supportted arg to an extra arg (#1797)

chuanqi129 committed 2 years ago

ecc91cb9

Disable llama_v2_7b_16h on optim benchmarks as they currently OOM (#1792)

janeyx99 committed 2 years ago

561abe38

Update workflow for new added models (#1790)

janeyx99 committed 2 years ago

648a32e1

Remove jit from readme and cleanup dependencies. (#1789)

xuzhao9 committed 2 years ago

b10fcd29

use default_factory and not a mutable dictionary (#1788)

janeyx99 committed 2 years ago

83f17f96

Remove jit from the framework code (#1787)

xuzhao9 committed 2 years ago

4ea9354a

Added Whisper from Huggingface. (#1769)

MaanavD committed 2 years ago

770d5cf7

torchmultimodal CLIP (#1780)

msaroufim committed 2 years ago

f968816b

fix regression detector after #1710 (#1786)

janeyx99 committed 2 years ago

8177771c

Update LLAMA v2 init (#1781)

msaroufim committed 2 years ago

28c45ec1

Exclude PT2 on most models, add more knobs to script (#1785)

janeyx99 committed 2 years ago

ed895d7e

Print out runnable repros with better time metadata (#1783)

janeyx99 committed 2 years ago

0d166d10

Skip CPU benchmarks since we don't observe their stats (#1784)

janeyx99 committed 2 years ago

725b72cc

Print torch git version (commit) in workflow (#1779)

janeyx99 committed 2 years ago

d9e6653a

fix uninitialized model_flop variable issue (#1778)

snadampal committed 2 years ago

c03597e8

llama v2 (#1775)

msaroufim committed 2 years ago

196b3b8a

Add functionality to TRT backend (#1753)

gs-olive committed 2 years ago

30a3c4a9

Get PT-D benchmark to support being run internally

H-Huang committed 2 years ago

36e1ed1c

Add torch.utils.flop_counter to TorchBench

monicatang committed 2 years ago

1dada69d

Fix fvcore flops counting for torchvision models

xuzhao9 committed 2 years ago

9d84f9e6

Add dependencies in the base docker build (#1776)

xuzhao9 committed 2 years ago

79bc754f

Older