.. |
autotuning
|
0c979d6779
Update BUFSIZE to come from autotuner's constants.py, not numpy (#5686)
|
4 月之前 |
checkpoint
|
774b897736
fix the missing argument in test and typo (#5730)
|
3 月之前 |
comm
|
659f6be105
Avoid security issues of subprocess shell (#6498)
|
1 月之前 |
compression
|
389bf69319
fix: Remove duplicate word the (#4051)
|
1 年之前 |
elasticity
|
659f6be105
Avoid security issues of subprocess shell (#6498)
|
1 月之前 |
inference
|
645639bcf8
Rearrange inference OPS and stop using builder.load (#5490)
|
1 周之前 |
launcher
|
13c16c9562
Accept btl_tcp_if_include option through launcher_args (#6613)
|
1 周之前 |
linear
|
6e5d58d248
OptimizedLinear updates (#5791)
|
2 月之前 |
model_implementations
|
645639bcf8
Rearrange inference OPS and stop using builder.load (#5490)
|
1 周之前 |
module_inject
|
474a3288cd
Enabled Qwen2-MoE Tensor Parallelism (TP) inference (#6551)
|
1 周之前 |
moe
|
7260890452
reduce cpu host overhead when using moe (#5578)
|
2 月之前 |
monitor
|
0a4457cc48
Pydantic v2 migration (#5167)
|
2 月之前 |
nebula
|
cd4e473ee6
fix typo with deepspeed/ (#3547)
|
1 年之前 |
nvme
|
a5400974df
DeepNVMe perf tuning (#6560)
|
3 周之前 |
ops
|
645639bcf8
Rearrange inference OPS and stop using builder.load (#5490)
|
1 周之前 |
pipe
|
b361c72761
Update DeepSpeed copyright license to Apache 2.0 (#3111)
|
1 年之前 |
profiling
|
170b46e8b1
Add conditional on torch version for scaled_dot_product_attention (#6517)
|
1 月之前 |
runtime
|
85b7469ea0
Add first Step in LR Schedulers (#6597)
|
1 周之前 |
sequence
|
8b191d7ccf
Long sequence parallelism (Ulysses) integration with HuggingFace (#5774)
|
2 月之前 |
utils
|
ce468c3756
add option to disable logger while compiling to avoid graph breaks (#6496)
|
6 天之前 |
__init__.py
|
8b191d7ccf
Long sequence parallelism (Ulysses) integration with HuggingFace (#5774)
|
2 月之前 |
accelerator
|
9548d48f48
Abstract accelerator (step 2) (#2560)
|
1 年之前 |
constants.py
|
706a72562a
Allow env var for timeout (#4405)
|
1 年之前 |
env_report.py
|
74f3dcab62
Add Windows scripts (deepspeed, ds_report). (#5699)
|
3 月之前 |
git_version_info.py
|
c08e69f212
Make op builder detection adapt to accelerator change (#5206)
|
7 月之前 |