Commit Graph

17 Commits (05bb28aacf71cde64bfc94e5ad7555d5607da77f)

Author SHA1 Message Date
Jiarui Fang 9214d1fe28
[Gemini] chunk init using runtime visited param order (#2115) 2022-12-12 18:06:16 +08:00
HELSON 63fbba3c19
[zero] add L2 gradient clipping for ZeRO (#2112)
* [zero] add L2 gradient clipping

* [testing] add MlpModel

* [zero] add unit test for grad clipping

* fix atol
2022-12-09 18:09:17 +08:00
Jiarui Fang 40b7d55bf3
[Gemini] add albert in test models. (#2075) 2022-12-05 14:09:34 +08:00
HELSON f6178728a0
[gemini] fix init bugs for modules (#2047)
* [gemini] fix init bugs for modules

* fix bugs
2022-11-30 17:06:10 +08:00
Jiarui Fang 1e885329f4
[test] align model name with the file name. (#2045) 2022-11-30 15:45:26 +08:00
Jiarui Fang 31c644027b
[hotfix] hotfix Gemini for no leaf modules bug (#2043) 2022-11-30 14:53:41 +08:00
HELSON 384cd26314
[zero] fix testing parameters (#2042) 2022-11-30 12:09:32 +08:00
HELSON 17a3c685b0
[zero] fix unit-tests (#2039) 2022-11-30 10:40:31 +08:00
Jiarui Fang eb7742a4bb
[Gemini] more tests for Gemini (#2038)
* [Gemini] more tests for Gemini

* polish code
2022-11-29 17:13:10 +08:00
Jiarui Fang 2e9cbfca12
[Gemini] add unitests to check gemini correctness (#2015) 2022-11-24 16:51:45 +08:00
Jiarui Fang f7e276fa71
[Gemini] add GeminiAdamOptimizer (#1960) 2022-11-16 14:44:28 +08:00
HELSON c6a1a62636
[hotfix] fix zero's incompatibility with checkpoint in torch-1.12 (#1786)
* [hotfix] fix zero's incompatibility with checkpoint in torch-1.12

* [zero] add cpu shard init

* [zero] add tiny example test

* [colo_tensor] fix bugs for torch-1.11
2022-11-02 16:11:34 +08:00
HELSON f69f9bf223
[zero] add chunk init function for users (#1729)
* add chunk manager init function

* fix unit tests

* add comment

* add flush=True
2022-10-18 16:31:22 +08:00
HELSON 1468e4bcfc
[zero] add constant placement policy (#1705)
* fixes memory leak when paramter is in fp16 in ZeroDDP init.
* bans chunk releasement in CUDA. Only when a chunk is about to offload, it is allowed to release.
* adds a constant placement policy. With it, users can allocate a reserved caching memory space for parameters.
2022-10-14 17:53:16 +08:00
HELSON b28991dd0a
[feature] A new ZeRO implementation (#1644) 2022-10-09 09:18:51 +08:00
Jiarui Fang c5d39215f6
Revert "[feature] new zero implementation (#1623)" (#1643)
This reverts commit 5be118f405.
2022-09-26 10:06:03 +08:00
HELSON 5be118f405
[feature] new zero implementation (#1623) 2022-09-24 19:58:18 +08:00