Commit Graph

681 Commits (6474e31556ae011410f29d8d0a2d80de67ed6956)

Author SHA1 Message Date
Sze-qq f8b9aaef47 [NFC] polish colossalai/kernel/cuda_native/csrc/type_shim.h code style (#1260)
2 years ago
superhao1995 f660152c73 [NFC] polish colossalai/nn/layer/parallel_3d/_operation.py code style (#1258)
2 years ago
Thunderbeee 9738fb0f78 [NFC] polish colossalai/nn/lr_scheduler/__init__.py (#1255)
2 years ago
Kai Wang (Victor Kai) 50f2ad213f [NFC] polish colossalai/engine/ophooks/utils.py code style (#1256)
2 years ago
Ofey Chan 2dd4d556fb
[NFC] polish colossalai/nn/init.py code style (#1292)
2 years ago
Jiarui Fang 556b9b7e1a
[hotfix] Dist Mgr gather torch version (#1284)
2 years ago
HELSON abba4d84e1
[hotfix] fix bert model test in unitests (#1272)
2 years ago
ver217 7aadcbd070
hotfix colotensor _scan_for_pg_from_args (#1276)
2 years ago
oahzxl 0cf8e8e91c
[NFC] polish <colossalai/nn/lr_scheduler/poly.py> code style (#1267)
2 years ago
Jiarui Fang c92f84fcdb
[tensor] distributed checkpointing for parameters (#1240)
2 years ago
Frank Lee fb35460595
[fx] added ndim property to proxy (#1253)
2 years ago
Frank Lee 4a09fc0947
[fx] fixed tracing with apex-based T5 model (#1252)
2 years ago
Frank Lee 7531c6271f
[fx] refactored the file structure of patched function and module (#1238)
2 years ago
YuliangLiu0306 17ed33350b
[hotfix] fix an assertion bug in base schedule. (#1250)
2 years ago
YuliangLiu0306 97d713855a
[fx] methods to get fx graph property. (#1246)
2 years ago
YuliangLiu0306 30b4fc0eb0
[fx]add split module pass and unit test from pipeline passes (#1242)
2 years ago
Jiarui Fang 1aad903c15
[tensor] redistribute among different process groups (#1247)
2 years ago
Jiarui Fang 9bcd2fd4af
[tensor] a shorter shard and replicate spec (#1245)
2 years ago
Jiarui Fang 2699dfbbfd
[rename] convert_to_dist -> redistribute (#1243)
2 years ago
HELSON f6add9b720
[tensor] redirect .data.__get__ to a tensor instance (#1239)
2 years ago
Jiarui Fang 20da6e48c8
[checkpoint] save sharded optimizer states (#1237)
2 years ago
Jiarui Fang 4a76084dc9
[tensor] add zero_like colo op, important for Optimizer (#1236)
2 years ago
Jiarui Fang 3b500984b1
[tensor] fix some unittests (#1234)
2 years ago
ver217 a45ddf2d5f
[hotfix] fix sharded optim step and clip_grad_norm (#1226)
2 years ago
HELSON f071b500b6
[polish] polish __repr__ for ColoTensor, DistSpec, ProcessGroup (#1235)
2 years ago
HELSON 0453776def
[tensor] fix a assertion in colo_tensor cross_entropy (#1232)
2 years ago
Jiarui Fang 0e199d71e8
[hotfix] fx get comm size bugs (#1233)
2 years ago
HELSON 42ab36b762
[tensor] add unitest for colo_tensor 1DTP cross_entropy (#1230)
2 years ago
Yi Zhao 04537bf83e
[checkpoint]support generalized scheduler (#1222)
2 years ago
Jiarui Fang a98319f023
[tensor] torch function return colotensor (#1229)
2 years ago
YuliangLiu0306 2b7dca44b5
[fx]get communication size between partitions (#1224)
2 years ago
Frank Lee 84f2298a96
[fx] added patches for tracing swin transformer (#1228)
2 years ago
Frank Lee b6cb5a47ad
[fx] added timm model tracing testing (#1221)
2 years ago
HELSON 280a81243d
[tensor] improve robustness of class 'ProcessGroup' (#1223)
2 years ago
Jiarui Fang 15d988f954
[tensor] sharded global process group (#1219)
2 years ago
Jiarui Fang db1bef9032
[hotfix] fx shard 1d pass bug fixing (#1220)
2 years ago
Frank Lee 11973d892d
[fx] added torchvision model tracing testing (#1216)
2 years ago
Jiarui Fang 52736205d9
[checkpoint] make unitest faster (#1217)
2 years ago
Jiarui Fang f38006ea83
[checkpoint] checkpoint for ColoTensor Model (#1196)
2 years ago
XYE 291e22aac6
[fx] temporarily used (#1215)
2 years ago
Jiarui Fang ae7d3f4927
[refactor] move process group from _DistSpec to ColoTensor. (#1203)
2 years ago
Frank Lee 5da87ce35d
[fx] added testing for all albert variants (#1211)
2 years ago
Frank Lee 2d13a45a3b
[fx] added testing for all gpt variants (#1210)
2 years ago
YuliangLiu0306 189946c5c4
[fx]add uniform policy (#1208)
2 years ago
Frank Lee 426a279ce7
[fx] added testing for all bert variants (#1207)
2 years ago
Jiarui Fang b5f25eb32a
[Tensor] add cpu group to ddp (#1200)
2 years ago
Frank Lee f7878f465c
[fx] supported model tracing for huggingface bert (#1201)
2 years ago
Jiarui Fang 060b917daf
[refactor] remove gpc dependency in colotensor's _ops (#1189)
2 years ago
Frank Lee abf6a262dc
[fx] added module patch for pooling layers (#1197)
2 years ago
YuliangLiu0306 63d2a93878
[context]support arbitary module materialization. (#1193)
2 years ago
Jiarui Fang a444633d13
warmup ratio configration (#1192)
2 years ago
ver217 dba7e0cfb4
make AutoPlacementPolicy configurable (#1191)
2 years ago
YuliangLiu0306 2053e138a2
[context]use meta tensor to init model lazily. (#1187)
2 years ago
Frank Lee 2c8c05675d
[fx] patched conv and normalization (#1188)
2 years ago
Frank Lee 6d86f1bc91
[fx] supported data-dependent control flow in model tracing (#1185)
2 years ago
Jiarui Fang c463f8adf9
[tensor] remove gpc in tensor tests (#1186)
2 years ago
Jiarui Fang 372f791444
[refactor] move chunk and chunkmgr to directory gemini (#1182)
2 years ago
ver217 6b2f2ab9bb
[ddp] ColoDDP uses bucket all-reduce (#1177)
2 years ago
Jiarui Fang 7487215b95
[ColoTensor] add independent process group (#1179)
2 years ago
YuliangLiu0306 26ba87272d
[hotfix]fixed p2p process send stuck (#1181)
2 years ago
Jiarui Fang 1b657f9ce1
[tensor] revert local view back (#1178)
2 years ago
Jiarui Fang 0dd4e2bbfb
[Tensor] rename some APIs in TensorSpec and Polish view unittest (#1176)
2 years ago
Ziyue Jiang dd0420909f
[Tensor] rename parallel_action (#1174)
2 years ago
YuliangLiu0306 e27645376d
[hotfix]different overflow status lead to communication stuck. (#1175)
2 years ago
Jiarui Fang aa7bef73d4
[Tensor] distributed view supports inter-process hybrid parallel (#1169)
2 years ago
ver217 9e1daa63d2
[zero] sharded optim supports loading local state dict (#1170)
2 years ago
ver217 561e90493f
[zero] zero optim supports loading local state dict (#1171)
2 years ago
Jiarui Fang 4b9bba8116
[ColoTensor] rename APIs and add output_replicate to ComputeSpec (#1168)
2 years ago
Jiarui Fang f4ef224358
[Tensor] remove ParallelAction, use ComputeSpec instread (#1166)
2 years ago
Jiarui Fang 177c374401
remove gather out in parallel action (#1163)
2 years ago
ver217 634eecb98e
mark sanity_check of dist_spec_mgr as staticmethod (#1161)
2 years ago
Ziyue Jiang 955ac912de
remove log (#1160)
2 years ago
ver217 4e67b2a890
fix chunk move device (#1158)
2 years ago
Jiarui Fang 07f9c781f9
[graph] improve the graph building. (#1157)
2 years ago
ver217 22717a856f
[tensor] add embedding bag op (#1156)
2 years ago
ver217 ae86151968
[tensor] add more element-wise ops (#1155)
2 years ago
ver217 54aabb8da4
[gemini] refactor gemini mgr (#1151)
2 years ago
Frank Lee f8eec98ff5
[tensor] fixed non-serializable colo parameter during model checkpointing (#1153)
2 years ago
ver217 ffa025e120
[tensor] dist spec s2s uses all-to-all (#1136)
2 years ago
YuliangLiu0306 f1f51990b9
[hotfix]fix some bugs caused by refactored schedule. (#1148)
2 years ago
Jiarui Fang 8cdce0399c
[ColoTensor] improves init functions. (#1150)
2 years ago
ver217 8106d7b8c7
[ddp] refactor ColoDDP and ZeroDDP (#1146)
2 years ago
Frank Lee 0e4e62d30d
[tensor] added __repr__ to spec (#1147)
2 years ago
YuliangLiu0306 70dd88e2ee
[pipeline]add customized policy (#1139)
2 years ago
YuliangLiu0306 18091581c0
[pipeline]support more flexible pipeline (#1138)
2 years ago
ver217 ccf3c58c89
embedding op use gather_out (#1143)
2 years ago
ver217 6690a61b4d
[hotfix] prevent nested ZeRO (#1140)
2 years ago
Frank Lee 15aab1476e
[zero] avoid zero hook spam by changing log to debug level (#1137)
2 years ago
Frank Lee 73ad05fc8c
[zero] added error message to handle on-the-fly import of torch Module class (#1135)
2 years ago
ver217 e4f555f29a
[optim] refactor fused sgd (#1134)
2 years ago
ver217 d26902645e
[ddp] add save/load state dict for ColoDDP (#1127)
2 years ago
YuliangLiu0306 946dbd629d
[hotfix]fix bugs caused by refactored pipeline (#1133)
2 years ago
ver217 789cad301b
[hotfix] fix param op hook (#1131)
2 years ago
ver217 a1a7899cae
[hotfix] fix zero init ctx numel (#1128)
2 years ago
ver217 f0a954f16d
[ddp] add set_params_to_ignore for ColoDDP (#1122)
2 years ago
YuliangLiu0306 3175bcb4d8
[pipeline]support List of Dict data (#1125)
2 years ago
Frank Lee 91a5999825
[ddp] supported customized torch ddp configuration (#1123)
2 years ago
YuliangLiu0306 fcf55777dd
[fx]add autoparallel passes (#1121)
2 years ago
ver217 e127b4375b
cast colo ddp v2 inputs/outputs (#1120)
2 years ago
Frank Lee 16302a5359
[fx] added unit test for coloproxy (#1119)
2 years ago
ver217 7d14b473f0
[gemini] gemini mgr supports "cpu" placement policy (#1118)
2 years ago
ver217 f99f56dff4
fix colo parameter torch function (#1117)
2 years ago
Frank Lee e1620ddac2
[fx] added coloproxy (#1115)
2 years ago
Frank Lee 6f82ac9bcb
[pipeline] supported more flexible dataflow control for pipeline parallel training (#1108)
2 years ago
ver217 895c1c5ee7
[tensor] refactor param op hook (#1097)
2 years ago
YuliangLiu0306 1e9f9c227f
[hotfix]change to fit latest p2p (#1100)
2 years ago
Frank Lee 72bd7c696b
[amp] included dict for type casting of model output (#1102)
2 years ago
Frank Lee 7f2d2b2b5b
[engine] fixed empty op hook check (#1096)
2 years ago
Frank Lee 14e5b11d7f
[zero] fixed api consistency (#1098)
2 years ago
Frank Lee cb18922c47
[doc] added documentation to chunk and chunk manager (#1094)
2 years ago
ver217 1f894e033f
[gemini] zero supports gemini (#1093)
2 years ago
Frank Lee 2b2dc1c86b
[pipeline] refactor the pipeline module (#1087)
2 years ago
Frank Lee bad5d4c0a1
[context] support lazy init of module (#1088)
2 years ago
ver217 be01db37c8
[tensor] refactor chunk mgr and impl MemStatsCollectorV2 (#1077)
2 years ago
Frank Lee 50ec3a7e06
[test] skip tests when not enough GPUs are detected (#1090)
2 years ago
Ziyue Jiang 0653c63eaa
[Tensor] 1d row embedding (#1075)
3 years ago
junxu d66ffb4df4
Remove duplication registry (#1078)
3 years ago
Jiarui Fang bcab249565
fix issue #1080 (#1071)
3 years ago
ver217 1b17859328
[tensor] chunk manager monitor mem usage (#1076)
3 years ago
ver217 98cdbf49c6
[hotfix] fix chunk comm src rank (#1072)
3 years ago
Frank Lee bfdc5ccb7b
[context] maintain the context object in with statement (#1073)
3 years ago
ver217 c5cd3b0f35
[zero] zero optim copy chunk rather than copy tensor (#1070)
3 years ago
Ziyue Jiang 4fc748f69b
[Tensor] fix optimizer for CPU parallel (#1069)
3 years ago
Jiarui Fang 49832b2344
[refactory] add nn.parallel module (#1068)
3 years ago
Ziyue Jiang 6754f1b77f
fix module utils bug (#1066)
3 years ago
Jiarui Fang a00644079e
reorgnize colotensor directory (#1062)
3 years ago
Frank Lee 3d10be33bd
[cudnn] set False to cudnn benchmark by default (#1063)
3 years ago
Ziyue Jiang df9dcbbff6
[Tensor] add hybrid device demo and fix bugs (#1059)
3 years ago
YuliangLiu0306 b167258b6a
[pipeline]refactor ppschedule to support tensor list (#1050)
3 years ago
ver217 e3fde4ee6b
fix import error in sharded model v2 (#1053)
3 years ago
ver217 e1922ea4f6
[zero] add chunk size search for chunk manager (#1052)
3 years ago
アマデウス 2c42b230f3
updated collective ops api (#1054)
3 years ago
ver217 51b9a49655
[zero] add zero optimizer for ColoTensor (#1046)
3 years ago
ver217 7faef93326
fix dist spec mgr (#1045)
3 years ago
ver217 9492a561c3
[tensor] ColoTensor supports ZeRo (#1015)
3 years ago
Ziyue Jiang 7c530b9de2
[Tensor] add Parameter inheritance for ColoParameter (#1041)
3 years ago
ver217 7cfd6c827e
[zero] add load_state_dict for sharded model (#894)
3 years ago
Ziyue Jiang 6c5996a56e
[Tensor] add module check and bert test (#1031)
3 years ago
YuliangLiu0306 7106bd671d
[p2p]add object list send/recv (#1024)
3 years ago
Frank Lee e4685832f8
[engine] fixed bug in gradient accumulation dataloader to keep the last step (#1030)
3 years ago
Ziyue Jiang 32291dd73f
[Tensor] add module handler for linear (#1021)
3 years ago
Ryan Russell 9b0c037027
fix typo in constants (#1027)
3 years ago
ver217 007ca0df92
fix colo init context (#1026)
3 years ago
YuliangLiu0306 d182b0bd47
[hotfix] fix some bugs caused by size mismatch. (#1011)
3 years ago
ver217 cefc29ff06
[tensor] impl ColoDDP for ColoTensor (#1009)
3 years ago
zhengzangw ae7c338105 [NFC] polish colossalai/kernel/cuda_native/csrc/colossal_C_frontend.cpp code style
3 years ago
ver217 a3b66f6def
[tensor] refactor parallel action (#1007)
3 years ago
ver217 ad536e308e
[tensor] refactor colo-tensor (#992)
3 years ago
Frank Lee 1467d83edf
[cli] remove unused imports (#1001)
3 years ago
Frank Lee 533d0c46d8
[kernel] fixed the include bug in dropout kernel (#999)
3 years ago