mirror of
https://github.com/hpcaitech/ColossalAI.git
synced 2025-09-09 04:50:17 +00:00
[devops] fix extention building (#5427)
This commit is contained in:
@@ -76,7 +76,7 @@ def check_installation():
|
||||
click.echo("")
|
||||
click.echo(f"Note:")
|
||||
click.echo(
|
||||
f"1. AOT (ahead-of-time) compilation of the CUDA kernels occurs during installation when the environment variable CUDA_EXT=1 is set"
|
||||
f"1. AOT (ahead-of-time) compilation of the CUDA kernels occurs during installation when the environment variable BUILD_EXT=1 is set"
|
||||
)
|
||||
click.echo(f"2. If AOT compilation is not enabled, stay calm as the CUDA kernels can still be built during runtime")
|
||||
|
||||
|
@@ -25,7 +25,7 @@ conda install -c conda-forge cupy cudnn cutensor nccl cuda-version=11.6
|
||||
|
||||
# install colossalai with PyTorch extensions
|
||||
cd <path_to_ColossalAI_repo>
|
||||
CUDA_EXT=1 pip install -e .
|
||||
BUILD_EXT=1 pip install -e .
|
||||
|
||||
# install other dependencies
|
||||
pip install triton==2.0.0.dev20221202
|
||||
|
@@ -25,7 +25,7 @@ conda install -c "nvidia/label/cuda-11.6.2" cuda-toolkit
|
||||
cd <path_to_ColossalAI_repo>
|
||||
pip install -r requirements/requirements.txt
|
||||
pip install -r requirements/requirements-test.txt
|
||||
CUDA_EXT=1 pip install -e .
|
||||
BUILD_EXT=1 pip install -e .
|
||||
|
||||
# install torchserve
|
||||
cd <path_to_torch_serve_repo>
|
||||
|
@@ -38,7 +38,7 @@ ARG VERSION=main
|
||||
RUN git clone -b ${VERSION} https://github.com/hpcaitech/ColossalAI.git && \
|
||||
cd ./ColossalAI && \
|
||||
git checkout 3e05c07bb8921f2a8f9736b6f6673d4e9f1697d0 && \
|
||||
CUDA_EXT=1 pip install -v --no-cache-dir .
|
||||
BUILD_EXT=1 pip install -v --no-cache-dir .
|
||||
|
||||
# install titans
|
||||
RUN pip install --no-cache-dir titans
|
||||
|
@@ -78,7 +78,7 @@ class CPUAdam(NVMeOptimizer):
|
||||
super(CPUAdam, self).__init__(model_params, default_args, nvme_offload_fraction, nvme_offload_dir)
|
||||
self.adamw_mode = adamw_mode
|
||||
cpu_adam = CPUAdamLoader().load()
|
||||
# if you find yourself stuck here, make sure that you install colossalai with CUDA_EXT=1 specification
|
||||
# if you find yourself stuck here, make sure that you install colossalai with BUILD_EXT=1 specification
|
||||
self.cpu_adam_op = cpu_adam.CPUAdamOptimizer(lr, betas[0], betas[1], eps, weight_decay, adamw_mode)
|
||||
|
||||
def torch_adam_update(
|
||||
|
Reference in New Issue
Block a user