Commit Graph

3196 Commits (5165bc854ac35c22c3d1d8c04629f5420972a23a)

Author SHA1 Message Date
liym27 ff25c5b36f
Fix bug: GetAttrValue should deal with attr with attrType vector<double> (#30536)
4 years ago
Leo Chen 81217a94d8
unify calling cudaSetDevice (#30470)
4 years ago
hutuxian 40ede12631
Ascend Framework Part1: OP & Wrapper (#30281)
4 years ago
liuyuhui 843dc3cdbd
[Kunlun]PR3: add xpu executor, multi xpu card train function optimization (#30317)
4 years ago
Adam Osewski c5ffad126c
[oneDNN] Refactor fuse pass helper functions to one place. (#30460)
4 years ago
pangyoki 13d757362c
Add Inplace strategy (Output reuse Input Varbase) in dygraph (#30103)
4 years ago
yaoxuefeng 6e0da01c61
Heter ps new (#30198)
4 years ago
cc 8e3a294045
skip quantizing ops in cpu inference (#30342)
4 years ago
alncat 7bbf3ac5ab
Added support for inference using quantization aware trained dygraph (#30288)
4 years ago
Zhang Jun 10a8f3e5c3
fix bug on compiling inference shared lib with crypto;test=develop (#30269)
4 years ago
JZ-LIANG 75936d838f
Recompute Offload (#30233)
4 years ago
tangwei12 5e839e4da5
add sparse embedding & load vars for 2.0 & gloo bug fix (#30306)
4 years ago
tangwei12 25f80fd304
Fix/distributed proto (#29981)
4 years ago
liym27 b4989fb744
Support vector<double> as type of op attribute and op set_value suppport vector<double> as value (#30126)
4 years ago
石晓伟 8ce2482b80
fix header file paths of gflags, commit 1, test=develop (#30271)
4 years ago
wangchaochaohu af80859dd6
reduce the occupied size of memory for the fused pattern of elementwise_add Op and activation Op(relu Op for example) (#29885)
4 years ago
Zhen Wang 7f7dfccf20
Support pure fp16 training for AMP API. (#29544)
4 years ago
Leo Chen 789743e190
use cuda generator in bernoulli cuda kernel (#30199)
4 years ago
Leo Chen 1f97d61c68
Add callback after TensorCopy (#30123)
4 years ago
Chengmo 528e03fc08
【Paddle.Fleet】Fix tensor table (#30075)
4 years ago
Huihuang Zheng 54bf3f5a56
Refine PADDLE_ENFORCE Error Messages. test=develop (#30149)
4 years ago
Chen Weihang d0fb06b27f
[Complex] Simplify prepared op impl to improve performance (#30153)
4 years ago
liuyuhui 15fac5e7fa
fix assign_op_xpu concat_op_xpu warining (#30120)
4 years ago
石晓伟 53bb126510
fix a bug in op_version_registry, test=develop, test=op_version (#29994)
4 years ago
liuyuhui 254ad61959
fix xpu pe sync, test=notest (#30095)
4 years ago
Thunderbrook 0b8e1fadc5
add topo-aware in heter-ps (#30087)
4 years ago
WangXi ee16006b5d
Optimization grad merge performance (#29784)
4 years ago
Shang Zhizhou 08dc5bc27e
fix op version checker of pass bug (#30028)
4 years ago
cc c3c064a8fc
Add mkldnn nearest_interp and bilinear_interp op (#30016)
5 years ago
wawltor cc2f94620c
add the support the op version check for matmul, test=op_version (#30011)
5 years ago
wawltor b33aaea86c
add the op version check for the elementwise ops, test=op_version (#30010)
5 years ago
Leo Chen 47d10c55d5
Enhance debugging (#30001)
5 years ago
wawltor 8f49f9d5c9
change the elementwise ops version check, test=op_version
5 years ago
Thunderbrook 0ca6de171f
add include (#29952)
5 years ago
cc 6a0102b038
map matmul/squeeze2+matmul/reshape2+matmul to mul (#29911)
5 years ago
Jack Zhou 5a4e42ca9a
add gru op_register_version; test=op_version; (#29931)
5 years ago
Wilber 2b1d796cd0
[Inference] Solve 2.0 trt performance reduce compare 1.8. (#29925)
5 years ago
石晓伟 181ea1870b
flush denormals to zero, test=develop (#29924)
5 years ago
liuyuhui 3d1741b794
[Kunlun] bug fix of PR2: Support MultiDevicePass and BKCL in parallel executor (#29926)
5 years ago
liym27 9602a182b2
[Dynamic Inplace] Support ShareInplaceVersionCounterWith for C++ Tensor (#29842)
5 years ago
liuyuhui 4427df37cf
[Kunlun] PR2: Support MultiDevicePass and BKCL in parallel executor (#29574)
5 years ago
YUNSHEN XIE 2a01756bf3
remove duplicate ut names (#29809)
5 years ago
Chen Weihang a6072055be
[Complex] Handle complex to real after type promotion (#29855)
5 years ago
Leo Chen 6b258317cb
fix TransferInplaceBack (#29830)
5 years ago
QingshuChen 59b47f3b32
feat: support check_nan_inf for kunlun/xpu device (#29694)
5 years ago
tangwei12 032414ca2a
[Feature] one ps (3/4) (#29604)
5 years ago
jakpiase edc06c6a1b
Added fc + activation fuse pass (currently only gelu, sigmoid and tanh are supported) (#29772)
5 years ago
YUNSHEN XIE 24ce051a84
remove duplicate ut reload (#29810)
5 years ago
Thunderbrook 09b6e71928
heter box (#29734)
5 years ago
Jacek Czaja 7b33720c90
[oneDNN] Tensor copy fix to oneDNN tensors (#29771)
5 years ago
Leo Chen 224f3bcbb1
format code (#29714)
5 years ago
石晓伟 8bd2879ef7
update the operator registration for incompatible upgrade, test=develop (#29720)
5 years ago
WangXi 9cbcc6cadc
fleet sync build strategy, test=develop (#29732)
5 years ago
Chen Weihang 6cfa59de1b
[Complex] Add real & imag op and api for complex tensor (#29672)
5 years ago
liuyuhui f13c3a9cd7
[Kunlun] PR1:Support one Kunlun card training in parallel executor (#29337)
5 years ago
lilong12 ff6a145011
update, test=develop (#29559)
5 years ago
Jacek Czaja f6cca62575
[oneDNN] Making ThreadID info in caching key optional (#29272)
5 years ago
JZ-LIANG d33d468f02
[Sharding] add hybrid-dp feature (#29518)
5 years ago
LoveAn b5d4a1f33d
Add the strategy of skipping cc/cu test compilation and execution in CI (#29499)
5 years ago
Aurelius84 2a42250699
Polish hash function of executor cache key (#29556)
5 years ago
jakpiase 57a4f16d9e
added internal and external reorders to profiler (#29443)
5 years ago
LoveAn 03b42d9fa7
fix unittest on windows, test=develop (#29365)
5 years ago
cc a623ce044f
Use different name_scope for different conv type, test=develop (#29355)
5 years ago
liym27 b10ecd9d3a
[inplace] Add ShareHolderWith for class Variable and SharePlaceholderWith in VarBase.detach() to share the same Tensor/SelectedRows (#29267)
5 years ago
Chen Weihang 9ad800ebb2
Support type promote for basic math ops (quantum required) (#29265)
5 years ago
Aurelius84 67c700b479
[Dy2Stat] Add cache for Executor and Context in run_program_op (#28421)
5 years ago
Chen Weihang 1de32f823d
Hot fix complle failed in gcc4.8 caused by complex impl (#29254)
5 years ago
GeminiCarrie 642abe2a48
Fix a bug when running on an operating system without "bash." (#29131)
5 years ago
ShenLiang 46b73e6cd9
Change the api of DataParallel and Fleet (#29224)
5 years ago
chentianyu03 8f45d14263
add complex64 and complex128 type; add +-*/@ and slice opreator for c… (#29199)
5 years ago
liym27 865a45984f
Check whether there is any inplace operation affecting gradient calculation. (#27901)
5 years ago
Chen Weihang 0b032faeee
Polish unittests details and execution conditions to adapt to MUSL (#29044)
5 years ago
Wojciech Uss 4fd4095d1b
Add quantization of multi_gru op and tests (#28615)
5 years ago
yaoxuefeng 545df287fc
add user_define_dump (#28596)
5 years ago
arlesniak bc902044a4
Fixes mkldnn dygraph learning rate scheduler crashes (#28988)
5 years ago
WangXi 173c22aec2
optimize fast graph executor (#28962)
5 years ago
Shibo Tao db41258501
add API serialize_program, serialize_persistables, save_to_file, deserialize_program, deserialize_persistables, load_from_file. (#29034)
5 years ago
joanna.wozna.intel b0d1ac161e
Add bf16 pool2d and unify bf16 unit tests (#29039)
5 years ago
joanna.wozna.intel fddea67445
Fix cpu_bfloat16_pass (#28730)
5 years ago
Chen Weihang fea0e294ee
Hide the C++ stack by default and add hints (#29042)
5 years ago
Wojciech Uss 7b5a8e46de
Add multi_gru_fuse_pass and tests (#28601)
5 years ago
Wojciech Uss 991345b368
Add multi_gru_seq_fuse_pass and tests (#28604)
5 years ago
lilong12 f77a78cdee
enable pipeline to run with Executor.run() (#28373)
5 years ago
Thunderbrook 0073f9bdb0
support ps-gpu (#28752)
5 years ago
Jacek Czaja bd1d6d3b30
extends oneDNN caching keys so caching objects are unique to executor/predictor (#28758)
5 years ago
gongweibao 1dad8ceaab
Fix gpu memory allocation bug. (#28703)
5 years ago
joanna.wozna.intel 8c0ea4bffe
Add bf16 matmul, fc, elementwise add and mul (#28729)
5 years ago
Wojciech Uss efc3b182f0
a fix for the fc_lstm_fuse_pass (#28709)
5 years ago
wanghuancoder 5aec7dbeb0
use forward declarations for framework.pb.h (#28494)
5 years ago
Jacek Czaja 6d8d3d4c22
[oneDNN] Layer norm bf16 kernel (#28619)
5 years ago
joanna.wozna.intel 2cb71c0cde
Add checkpoint to quantize (#28612)
5 years ago
lidanqing 804271cff9
Op version python mkldnn_inplace test (#28354)
5 years ago
Leo Chen 90805e2df7
Register op_version for new attribute use_addto (#28463)
5 years ago
Shang Zhizhou 8699f38d08
裁剪transformer模型trt支持;修复tensorRT不支持DeletePass的bug (#28517)
5 years ago
lidanqing 0fc181dbd0
[Fix bug] If the pass name is not found, IsCompatible should return false (#28475)
5 years ago
wangchaochaohu d7cfee9b31
Checkout point add (#28488)
5 years ago
Pei Yang 75196cda40
Paddle-TRT int8 support mul op channelwise quant (#28422)
5 years ago
YUNSHEN XIE 369605be1d
fix cmake error when execute build_inference_lib (#28503)
5 years ago
YUNSHEN XIE 1e698c600e
fix cmake error when setting ut timeout properity (#28492)
5 years ago
YUNSHEN XIE ba0756325a
exec ut no more than 15s 1 (#28439)
5 years ago