From 83460fe6e0dd86bdf30f3036a3572e89167d8a55 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=EC=A0=95=EC=8B=9C=EC=9B=90?= Date: Thu, 7 May 2026 18:21:14 +0900 Subject: [PATCH] Cross-reference torch.compile experiment to batched-traversal priority Make the link from the failed torch.compile experiment to Optimization Priorities #5 explicit, so a future revisit happens at the right time (once compile is on the dominant phase, not just trainer optimization steps). Co-Authored-By: Claude Opus 4.7 (1M context) --- docs/performance.md | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/docs/performance.md b/docs/performance.md index b547553..c6a1559 100644 --- a/docs/performance.md +++ b/docs/performance.md @@ -353,5 +353,7 @@ Net result: regression. Two reasons: call dispatch overhead exceeds the kernel-fusion benefit. Implementation preserved on branch `experiments/torch-compile` for -revisiting if the trainer model grows substantially or traversal moves -to GPU-batched inference. Not enabled on `main`. +revisiting if the trainer model grows substantially or after the +batched-traversal-inference work in Optimization Priorities #5 lands — +that is the change that would put compile on the dominant phase, not +just on the trainer's optimization steps. Not enabled on `main`.