787 Commits

Author SHA1 Message Date
Zaida Zhou
5ef75fd7a7
[Docs] Introduce how to customize distributed training settings (#1279) 2023-07-31 15:40:45 +08:00
Xinyu Yang
4fdab5e9cb
[Fix] Fix Visualizer that built vis_backends will not be used when save_dir is None (#1275) 2023-07-31 15:29:44 +08:00
Zaida Zhou
2df93eb51f
Add the loop stage in message_hub (#1277) 2023-07-31 14:22:49 +08:00
Mashiro
237aee3866
[Enhance] Enhance error information in build function (#1088) 2023-07-28 11:21:09 +08:00
Mashiro
e56d6edf19
[Enhance] Ehance error message thrown by Config (#1270) 2023-07-28 10:09:47 +08:00
Mashiro
42fdbc2ddb
[Fix] Fix ConfigDict.items will be called during dump (#1272) 2023-07-28 10:08:10 +08:00
youkaichao
ee742da254
[Enhance] Use graph transform to deal with more general cases for efficient_conv_bn_eval (#1259) 2023-07-26 17:52:57 +08:00
vugia truong
c8a1264568
[Feature] Compare the difference of two configs (#1260) 2023-07-26 15:48:59 +08:00
黄启元
78205c3254
Support multi-node distributed training with MLU backend (#1266) 2023-07-26 10:32:53 +08:00
KerwinKai
68360e7ce8
[Feature] Add parameter save_begin for CheckpointHook (#1271) 2023-07-25 19:21:21 +08:00
Mashiro
3871881ef6
[Enhance] Support skipping initialization in BaseModule (#1263) 2023-07-25 12:56:42 +08:00
Mashiro
6187595677
Move data_preprocessor to target device in FSDPStrategy (#1261) 2023-07-24 10:42:53 +08:00
Mashiro
f4f2555324
Llama2 example (#1264) 2023-07-24 10:20:21 +08:00
Qingyun
86387da4a5
[Enhancement] Register DeepSpeedStrategy even if deepspeed is not installed (#1240) 2023-07-18 11:02:55 +08:00
Zaida Zhou
5bc841c09c
[Docs] Add the data flow of Runner in README (#1257) 2023-07-18 10:09:42 +08:00
Miguel Méndez
9f21d38907
[Docs] Add short explanation about registry scope (#1114) 2023-07-17 10:21:44 +08:00
youkaichao
66d828d8d3
[Enhancement] Rename fast_conv_bn_eval to efficient_conv_bn_eval (#1251) 2023-07-15 22:13:17 +08:00
i-aki-y
276c614bf5
[Fix] Fix scalar check in RuntimeInfoHook (#1250) 2023-07-15 22:03:43 +08:00
youkaichao
40e49ff747
[Feature] Enable fast conv bn eval (#1202) 2023-07-14 18:21:55 +08:00
Zaida Zhou
278f7f5666
[Docs] Add ecosystem in README (#1247) 2023-07-12 14:06:28 +08:00
Mashiro
0549e1ce01
Bump version to v0.8.2 (#1246) v0.8.2 2023-07-12 10:57:12 +08:00
Qingyun
de81d29abd
[Fix] Fix the logic of setting lazy_import (#1239) 2023-07-11 22:32:20 +08:00
Mashiro
955b5712c4
[Fix] Fix pickle the Python style config (#1241) 2023-07-11 20:37:48 +08:00
Mashiro
b2295a258c
bump version to v0.8.1 (#1237) v0.8.1 2023-07-05 23:27:51 +08:00
Mashiro
8d4bac2616
[Enhance] Enhance config (#1232) 2023-07-05 23:27:08 +08:00
Zaida Zhou
33e30b7cb6
[Docs] how to train a large model (#1228) 2023-07-05 18:20:07 +08:00
Mashiro
62f9504d70
[Fix] FSDP should call _get_ignored_modules (#1235) 2023-07-04 22:39:05 +08:00
Mashiro
529bab815f
[Fix] Fix docs (#1233) 2023-07-04 22:20:30 +08:00
Mashiro
317d8f3178
[Fix] Remove auto find_ignored_modules (#1231) v0.8.0 2023-07-03 19:54:54 +08:00
Mashiro
d5a46d4144
[Fix] Implement copy and __copy__ for ConfigDict (#1230) 2023-07-03 15:11:10 +08:00
Mashiro
20d477dae1
[Enhancement] Only recognize modules mixed gradients as ignored_modules (#1226) 2023-07-01 22:50:36 +08:00
Mashiro
b638d3b1fe
[Fix] Fix new config (#1227) 2023-07-01 22:35:11 +08:00
Mashiro
eea8a7135c
Bump version to v0.8.0 (#1221) 2023-06-30 17:35:56 +08:00
Mashiro
1480261e8f
[Enhance] Config adds copy method (#1224) 2023-06-30 17:11:16 +08:00
Mashiro
f930b9fe53
Fix docstring format (#1223) 2023-06-30 10:39:19 +08:00
Mashiro
399f76ffa8 [Experimental] Add support for FSDP (#1213) 2023-06-29 15:19:33 +08:00
Zaida Zhou
ccd5dc8b18 [Experimental] Add FlexibleRunner and Strategies (#1183) 2023-06-29 15:19:33 +08:00
Mashiro
22aa46bf56
[Docs] Fix config doc (#1218) 2023-06-28 19:01:08 +08:00
Mashiro
6c5366649e
[Fix] Fix undefined variable error in Runner (#1219) 2023-06-28 18:59:22 +08:00
Zaida Zhou
d03a1da9a9
[Docs] Add a document to introduce how to debug with vscode (#1212) 2023-06-27 16:46:47 +08:00
Maxime G
186e17dbe7
[Docs] Fix a comment in runner tutorial (#1210) 2023-06-27 11:34:25 +08:00
Mashiro
478c952a6d
[Refacor] Replace 'if base' with 'with read_base' context manager (#1207) 2023-06-25 13:53:19 +08:00
edkair
42dd97d029
[Fix] gpu collection during evaluation (#1208) 2023-06-24 16:42:25 +08:00
Mashiro
04b0ffee76
[Fix] Fix ut error in docker (#1204) 2023-06-16 22:05:08 +08:00
Zaida Zhou
1c3332c96e
Fix the syntax in github action (#1203) 2023-06-16 14:38:43 +08:00
Mashiro
6ece63ed35
[Feature] Support Pure Python style Configuration File (#1071) 2023-06-16 12:52:07 +08:00
Zaida Zhou
1ab646de17
Skip mmcv in CI when torch < 1.8 (#1199) 2023-06-15 14:15:00 +08:00
syo093c
0052873b41
[Docs] Fix typo in document (#1201) 2023-06-14 22:15:20 +08:00
Zaida Zhou
2c35c164d2
[Fix] CheckpointHook should check whether file exists before removing it (#1198) 2023-06-12 22:35:07 +08:00
Akide Liu
94e7a3bb57
[Enhance] Learning rate in log can show the base learning rate of optimizer (#1019) 2023-06-08 19:51:15 +08:00