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

..

amp

[doc] update rst and docstring (#1351)

2022-07-21 15:54:53 +08:00

auto_parallel

[autoparallel] refactored the autoparallel module for organization (#1706)

2022-10-14 13:27:00 +08:00

builder

[NFC] polish colossalai/builder/__init__.py code style (#1560)

2022-09-08 22:11:04 +08:00

cli

[hotfix] fix some bugs caused by size mismatch. (#1011)

2022-05-23 14:02:28 +08:00

communication

[communication] add p2p_v2.py to support communication with List[Any] (#1407)

2022-08-09 11:40:04 +08:00

context

[zero] add constant placement policy (#1705)

2022-10-14 17:53:16 +08:00

device

[tensor]add 1D device mesh (#1492)

2022-08-25 16:48:12 +08:00

engine

[engin/schedule] use p2p_v2 to recontruct pipeline_schedule (#1408)

2022-08-12 11:33:26 +08:00

fx

[autoparallel] adapt runtime passes (#1703)

2022-10-14 10:14:07 +08:00

gemini

[zero] add constant placement policy (#1705)

2022-10-14 17:53:16 +08:00

kernel

[hotfix] fix CPUAdam kernel nullptr (#1410)

2022-08-05 19:45:45 +08:00

logging

[doc] improved docstring in the logging module (#861)

2022-04-25 13:42:00 +08:00

nn

[zero] add constant placement policy (#1705)

2022-10-14 17:53:16 +08:00

pipeline

[pipeline/fix-bug] num_microbatches support any integrate | stable chimera | launch tool for rpc pp framework (#1684)

2022-10-10 16:01:02 +08:00

registry

Remove duplication registry (#1078)

2022-06-08 07:47:24 +08:00

tensor

[autoparallel] added sharding spec conversion for linear handler (#1687)

2022-10-12 11:16:18 +08:00

testing

[unittest] added doc for the pytest wrapper (#1704)

2022-10-14 10:56:17 +08:00

trainer

[NFC] polish ./colossalai/trainer/hooks/_lr_scheduler_hook.py code style (#1576)

2022-09-08 22:11:04 +08:00

utils

[zero] add constant placement policy (#1705)

2022-10-14 17:53:16 +08:00

zero

[zero] add constant placement policy (#1705)

2022-10-14 17:53:16 +08:00

__init__.py

update version to 0.1.10 (#1676)

2022-10-09 10:43:29 +08:00

_meta_registrations.py

[fx/profiler] tuned the calculation of memory estimation (#1619)

2022-09-23 10:59:47 +08:00

constants.py

fix typo in constants (#1027)

2022-05-26 08:45:08 +08:00

core.py

[Tensor] distributed view supports inter-process hybrid parallel (#1169)

2022-06-27 09:45:26 +08:00

global_variables.py

[MOE] add unitest for MOE experts layout, gradient handler and kernel (#469)

2022-03-21 13:35:04 +08:00

initialize.py

[hotfix] remove potiential circle import (#1307)

2022-07-14 13:44:26 +08:00