.. |
__init__.py
|
b112c99ea8
Fix loading a universal checkpoint (#5263)
|
7 月之前 |
bwc.py
|
08e0733e4a
Support MoE for pipeline models (#5338)
|
6 月之前 |
comms_logging.py
|
0b507253e5
fix comm logging for inference (#4043)
|
1 年之前 |
config.py
|
995ba11928
Add throughput timer configuration (#5363)
|
5 月之前 |
debug.py
|
40342055ce
Remove hooks on gradient accumulation on engine/optimizer destroy (#4858)
|
9 月之前 |
exceptions.py
|
b361c72761
Update DeepSpeed copyright license to Apache 2.0 (#3111)
|
1 年之前 |
groups.py
|
08e0733e4a
Support MoE for pipeline models (#5338)
|
6 月之前 |
init_on_device.py
|
b361c72761
Update DeepSpeed copyright license to Apache 2.0 (#3111)
|
1 年之前 |
logging.py
|
b361c72761
Update DeepSpeed copyright license to Apache 2.0 (#3111)
|
1 年之前 |
mixed_precision_linkage.py
|
18179807f5
Remove optimizer step on initialization (#5104)
|
8 月之前 |
numa.py
|
5dadf68771
support HBM in utils/numa.py (#3918)
|
1 年之前 |
nvtx.py
|
b421e8c8f3
Disable nvtx decorator to avoid graph break (#5697)
|
3 月之前 |
tensor_fragment.py
|
b112c99ea8
Fix loading a universal checkpoint (#5263)
|
7 月之前 |
timer.py
|
b3767d01d4
Fixed Windows inference build. (#5609)
|
4 月之前 |
torch.py
|
6dcced1d5c
Cleanup required_torch_version code and references. (#5370)
|
6 月之前 |
types.py
|
0a61d5d664
Hybrid Engine Refactor and Llama Inference Support (#3425)
|
1 年之前 |
z3_leaf_module.py
|
19e0dc39ba
Delay reduce-scatter for ZeRO3 leaf modules (#5008)
|
8 月之前 |
zero_to_fp32.py
|
2518cc429d
Support `exclude_frozen_parameters` for `zero_to_fp32.py` script (#4979)
|
8 月之前 |