Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

modify setup.py (#206)

n1ck-guo8610b888e72f
diff --git a/README.md b/README.md
index b9f614aba..ab03a9ee1 100644
--- a/README.md
+++ b/README.md
@@ -137,13 +137,13 @@ Please run the quantization code first
### Setup env for AutoRound format
-2 Set up the environment
-
**cuda**: git clone https://github.com/intel/auto-round.git && cd auto-round && pip install -vvv --no-build-isolation -e .
-cpu: pip install intel-extension-for-transformers
+**cpu**:
+* option 1: pip install auto-round && pip install intel-extension-for-transformers
+* option 2: git clone https://github.com/intel/auto-round.git && cd auto-round && pip install -vvv --no-build-isolation -e .
-hpu: docker image with Gaudi Software Stack is recommended, please refer to following script for environment setup. More details can be found in Gaudi Guide.
+**hpu**: docker image with Gaudi Software Stack is recommended. More details can be found in [Gaudi Guide](https://docs.habana.ai/en/latest/).
### Gaudi2/ CPU/ GPU
@@ -206,7 +206,7 @@ average accuracies of 11 zero-shot tasks.
## Reference
-If you find SignRound useful for your research, please cite our paper:
+If you find AutoRound useful for your research, please cite our paper:
```bash
@article{cheng2023optimize,
diff --git a/setup.py b/setup.py
index 55cf381ab..04a0284f4 100644
--- a/setup.py
+++ b/setup.py
@@ -19,13 +19,28 @@ version = __version__
BUILD_CUDA_EXT = int(os.environ.get('BUILD_CUDA_EXT', '1')) == 1
PYPI_RELEASE = os.environ.get('PYPI_RELEASE', None)
+def is_cpu_env():
+ try:
+ import torch
+ except Exception as e:
+ print(
+ f"Building extension requires PyTorch being installed, please install PyTorch first: {e}.\n NOTE: This issue may be raised due to pip build isolation system (ignoring local packages). Please use `--no-build-isolation` when installing with pip, and refer to https://github.com/intel/auto-round for more details.")
+ sys.exit(1)
+ if torch.cuda.is_available():
+ return False
+ try:
+ import habana_frameworks.torch.core as htcore
+ return False
+ except:
+ return True
+
def fetch_requirements(path):
requirements = []
with open(path, "r") as fd:
requirements = [r.strip() for r in fd.readlines()]
- if not BUILD_CUDA_EXT:
- requirements.append("intel-extension-for-transformers") # for leverage QBits woq_linear capability.
+ if is_cpu_env():
+ requirements.append("intel-extension-for-transformers")
return requirements
def detect_local_sm_architectures():
@@ -62,7 +77,7 @@ if BUILD_CUDA_EXT:
import torch
except Exception as e:
print(
- f"Building PyTorch CUDA extension requires PyTorch being installed, please install PyTorch first: {e}.\n NOTE: This issue may be raised due to pip build isolation system (ignoring local packages). Please use `--no-build-isolation` when installing with pip, and refer to https://github.com/AutoRound/AutoRound/pull/620 for more details.")
+ f"Building PyTorch CUDA extension requires PyTorch being installed, please install PyTorch first: {e}.\n NOTE: This issue may be raised due to pip build isolation system (ignoring local packages). Please use `--no-build-isolation` when installing with pip, and refer to https://github.com/intel/auto-round for more details.")
sys.exit(1)
if not torch.cuda.is_available():
print(