A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 5, 12:15 UTC
only patch runtime_env for torch >= 2.10 (#38763)
Signed-off-by: Rohan138 <rohanpotdar138@gmail.com>
Rohan Potdar committed
Apr 7, 2026 at 04:29 UTC
dd9342e6bc92a52a4674a3e472318c241cb18fe1
1 file changed
+1
-1
vllm/env_override.py
+1
-1
@@ -500,7 +500,7 @@ if is_torch_equal("2.9.0"):
500
# This mirrors the fix in https://github.com/pytorch/pytorch/pull/177558
501
# and can be removed once torch >=2.12 is the minimum supported version.
502
503
-if not is_torch_equal_or_newer("2.12.0"):
503
+if is_torch_equal_or_newer("2.10.0") and not is_torch_equal_or_newer("2.12.0"):
504
import builtins as _builtins
505
import pickle
506