Expose torch.compiler.config.force_disable_caches as a public API (pytorch#166699)

gmagogsfm · pytorchmergebot · commit 8b2365094dbb · 2025-11-06T05:59:05.000Z
Exposing this flag as some upstream frameworks (like vLLM) could benefit from knowing whether torch.compile caches are enabled or not to adjust their own caching behavior. Pull Request resolved: pytorch#166699 Approved by: https://github.com/oulgen, https://github.com/mlazos
diff --git a/torch/compiler/config.py b/torch/compiler/config.py
@@ -35,6 +35,7 @@
     "enable_cpp_symbolic_shape_guards",
     "wrap_top_frame",
     "reorderable_logging_functions",
+    "force_disable_caches",
 ]
 
 

Original file line number	Diff line number	Diff line change
`@@ -35,6 +35,7 @@`
`35`	`35`	`"enable_cpp_symbolic_shape_guards",`
`36`	`36`	`"wrap_top_frame",`
`37`	`37`	`"reorderable_logging_functions",`
	`38`	`+ "force_disable_caches",`
`38`	`39`	`]`
`39`	`40`
`40`	`41`