.. |
autotuning
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
checkpoint
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
comm
|
db15ef578a
deepspeed.init_distributed() support for TCP protocols (#2905)
|
1 年之前 |
compression
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
elasticity
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
inference
|
87eaf8f99a
Check for local CUDA graphs when enable_cuda_graph=True (#2941)
|
1 年之前 |
launcher
|
8d53ac0cd3
Add MPICH Multinode Runner (#2839)
|
1 年之前 |
model_implementations
|
87eaf8f99a
Check for local CUDA graphs when enable_cuda_graph=True (#2941)
|
1 年之前 |
module_inject
|
0acf7e9c48
[RFC] add device abstraction to allow other device than CUDA be used (#2221)
|
1 年之前 |
moe
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
monitor
|
91d7090e47
Fixes `AttributeError` in #2853 (#2854)
|
1 年之前 |
nebula
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
ops
|
17fa0876ad
Always convert input mask to half (#2851)
|
1 年之前 |
pipe
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
profiling
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
runtime
|
80d8fcbdb3
Improve overflow handling (#2944)
|
1 年之前 |
utils
|
541e423ae6
Enable tensor fragments for zero 2 & 3 (#2727)
|
1 年之前 |
__init__.py
|
4abf637f96
Remove mutable default parameter in init_inference() (#2540)
|
1 年之前 |
accelerator
|
9548d48f48
Abstract accelerator (step 2) (#2560)
|
1 年之前 |
constants.py
|
1ed5aa96a8
Elastic Training support in DeepSpeed (#2153) (#2156)
|
2 年之前 |
env_report.py
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |
git_version_info.py
|
da84e60d98
add missing license info to top of all source code (#2889)
|
1 年之前 |