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 15, 17:15 UTC
[Bugfix] Fix 'InductorAdaptor object has no attribute 'cache_dir' (#15674)
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Jee Jee Li committed
Mar 28, 2025 at 17:10 UTC
70f2c2a7094cc5fbc6788f2b0b9b9da6973290cd
1 file changed
+1
-1
vllm/compilation/compiler_interface.py
+1
-1
@@ -144,6 +144,7 @@ class InductorAdaptor(CompilerInterface):
144
return hash_str
145
146
def initialize_cache(self, cache_dir: str, disable_cache: bool = False):
147
+ self.cache_dir = cache_dir
148
if disable_cache:
149
return
150
# redirect the cache directory to a sub-directory
@@ -156,7 +157,6 @@ class InductorAdaptor(CompilerInterface):
157
triton_cache = os.path.join(cache_dir, "triton_cache")
158
os.makedirs(triton_cache, exist_ok=True)
159
os.environ["TRITON_CACHE_DIR"] = triton_cache
159
- self.cache_dir = cache_dir
160
161
def compile(
162
self,