mirror of https://github.com/hpcaitech/ColossalAI
1468e4bcfc
* 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. |
||
---|---|---|
.. | ||
process_group_initializer | ||
random | ||
__init__.py | ||
config.py | ||
moe_context.py | ||
parallel_context.py | ||
parallel_mode.py | ||
singleton_meta.py |