biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

use .dev for version comparison with pytorch nightly release (#20031)

Browse Source 
Signed-off-by: Boyuan Feng <boyuan@meta.com>
This commit is contained in:
Boyuan Feng
2025-06-24 14:52:16 -07:00
committed by GitHub
parent ead369845d
commit c01d1c5aba
4 changed files with 19 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/model_executor/layers/quantization/torchao.py
View File

@@ -44,14 +44,14 @@ class TorchAOConfig(QuantizationConfig):
"""
# TorchAO quantization relies on tensor subclasses. In order,
# to enable proper caching this needs standalone compile
if is_torch_equal_or_newer("2.8.0a"):
if is_torch_equal_or_newer("2.8.0.dev"):
os.environ["VLLM_TEST_STANDALONE_COMPILE"] = "1"
logger.info(
"Using TorchAO: Setting VLLM_TEST_STANDALONE_COMPILE=1")
# TODO: remove after the torch dependency is updated to 2.8
if is_torch_equal_or_newer(
"2.7.0") and not is_torch_equal_or_newer("2.8.0a"):
"2.7.0") and not is_torch_equal_or_newer("2.8.0.dev"):
os.environ["VLLM_DISABLE_COMPILE_CACHE"] = "1"
logger.info("Using TorchAO: Setting VLLM_DISABLE_COMPILE_CACHE=1")
"""