zbian 61e687831d fixed using zero with tp cannot access weight correctly

2023-02-28 10:52:30 +08:00

..

__init__.py

moved env variables to global variables; (#215)

2022-02-15 11:31:13 +08:00

_utils.py

fixed using zero with tp cannot access weight correctly

2023-02-28 10:52:30 +08:00

dropout.py

fixed using zero with tp cannot access weight correctly

2023-02-28 10:52:30 +08:00

embedding.py

[model checkpoint] reworked unified layers for ease of save/load states (#593)

2022-04-01 16:49:56 +08:00

linear.py

added skip_bias_add for non-tp linear

2022-11-09 15:41:08 +08:00

normalization.py

[TP] allow layernorm without bias (#750)

2022-04-14 11:43:56 +08:00