.. |
autotuning
|
0c979d6779
Update BUFSIZE to come from autotuner's constants.py, not numpy (#5686)
|
4 月之前 |
checkpoint
|
f0e3f01d7c
Add an argument to enable the injection of missing state during the conversion of universal checkpoints (#5608)
|
3 月之前 |
comm
|
d254d75ef0
[XPU] support op builder from intel_extension_for_pytorch kernel path (#5425)
|
4 月之前 |
compression
|
389bf69319
fix: Remove duplicate word the (#4051)
|
1 年之前 |
elasticity
|
a4cd5504ca
Fix latest pytorch '_get_socket_with_port' import error (#5654)
|
4 月之前 |
inference
|
b3767d01d4
Fixed Windows inference build. (#5609)
|
3 月之前 |
launcher
|
3f875d9519
add device config env for the accelerator (#5396)
|
6 月之前 |
linear
|
0fc19b6a32
Fix crash when creating Torch tensor on NPU with device=get_accelerator().current_device() (#5464)
|
5 月之前 |
model_implementations
|
d5a7c1e0b4
Capture short kernel sequences to graph (#4318)
|
10 月之前 |
module_inject
|
8ea995ee1f
enable yuan autotp & add conv tp (#5428)
|
4 月之前 |
moe
|
695d79ea06
Fix RuntimeError for moe on XPU: tensors found at least two devices (#5519)
|
5 月之前 |
monitor
|
54f98fd913
Monitor was always enabled causing performance degradation (#5633)
|
4 月之前 |
nebula
|
cd4e473ee6
fix typo with deepspeed/ (#3547)
|
1 年之前 |
ops
|
f4efef21b8
[INF] DSAttention allow input_mask to have false as value (#5546)
|
5 月之前 |
pipe
|
b361c72761
Update DeepSpeed copyright license to Apache 2.0 (#3111)
|
1 年之前 |
profiling
|
98c96e790b
Update flops profiler to handle attn and __matmul__ (#4724)
|
8 月之前 |
runtime
|
b421e8c8f3
Disable nvtx decorator to avoid graph break (#5697)
|
3 月之前 |
sequence
|
3bdd187e71
Fixing the reshape bug in sequence parallel alltoall, which corrupted all QKV data (#5664)
|
4 月之前 |
utils
|
b421e8c8f3
Disable nvtx decorator to avoid graph break (#5697)
|
3 月之前 |
__init__.py
|
dc3554f832
Add `distributed_port` for `deepspeed.initialize` (#5260)
|
6 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
|
c08e69f212
Make op builder detection adapt to accelerator change (#5206)
|
7 月之前 |
git_version_info.py
|
c08e69f212
Make op builder detection adapt to accelerator change (#5206)
|
7 月之前 |
pydantic_v1.py
|
604d701e35
Introduce pydantic_v1 compatibility module for pydantic>=2.0.0 support (#4407)
|
1 年之前 |