From 2e70b60ea3934e62a49b90a988f362edcdb0a9e0 Mon Sep 17 00:00:00 2001 From: youkaichao Date: Fri, 6 Sep 2024 23:33:44 -0700 Subject: [PATCH] fix documentation format (#60) --- docs/opt_tutorial.rst | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/docs/opt_tutorial.rst b/docs/opt_tutorial.rst index 6f1bb539..21403ada 100644 --- a/docs/opt_tutorial.rst +++ b/docs/opt_tutorial.rst @@ -153,8 +153,9 @@ Conclusion ---------- Optimizing code with ``torch.compile`` involves: + 1. Using ``depyf`` to decompile the bytecode and understand the performance bottlenecks. 2. Identifying and addressing unnecessary checks or other sources of overhead. 3. Using ``depyf`` to directly call the compiled function, bypassing unnecessary steps where appropriate. -By following these steps, you can significantly improve performance, especially in environments where execution time is critical. +By following these steps, we can significantly improve performance, especially in environments where execution time is critical.