DeepSpeed
40e6764e
- Merge branch 'master' into loadams/rocm6
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Hide Minimap (CTRL+M)
Commit
1 year ago
Merge branch 'master' into loadams/rocm6
References
loadams/rocm6
Author
loadams
Parents
53cdfa1c
c1e02052
Files
61
.github/workflows
cpu-inference.yml
nv-inference.yml
nv-sd.yml
accelerator
hpu_accelerator.py
real_accelerator.py
xpu_accelerator.py
csrc
adam
multi_tensor_adam.cu
aio/py_lib
deepspeed_aio_thread.cpp
cpu/comm
ccl.cpp
xpu
adagrad
cpu_adagrad.cpp
adam
cpu_adam.cpp
cpu_adam_impl.cpp
fused_adam_frontend.cpp
multi_tensor_adam.dp.cpp
multi_tensor_apply.dp.hpp
common
custom_cuda_kernel.dp.cpp
includes
compat.h
cpu_adagrad.h
cpu_adam.h
simd.h
type_shim.h
deepspeed
comm
ccl.py
inference/v2
model_implementations
falcon
model.py
inference_transformer_base.py
llama_v2
model.py
mistral
model.py
mixtral
model.py
opt
model.py
phi
model.py
modules/implementations/attention
dense_blocked_attention.py
module_inject
auto_tp.py
containers
vae.py
fusedqkv_utils.py
replace_module.py
tp_shard.py
runtime
engine.py
sparse_tensor.py
swap_tensor
optimizer_utils.py
partitioned_optimizer_swapper.py
pipelined_optimizer_swapper.py
zero
mics.py
stage3.py
docs
_tutorials
accelerator-abstraction-interface.md
code-docs/source
schedulers.rst
op_builder
builder.py
npu
__init__.py
async_io.py
xpu
__init__.py
async_io.py
builder.py
cpu_adagrad.py
cpu_adam.py
fused_adam.py
requirements
requirements-inf.txt
requirements-sd.txt
tests/unit
inference
test_inference.py
test_inference_config.py
runtime/zero
test_nvme_checkpointing.py
test_zero.py
test_zero_context.py
test_zero_offloadpp.py
Loading