[LoRA] Support LoRA under the breakable/full prefill CUDA graph (#30988)

2026-07-26 22:10:03 -07:00
parent c6a6200a1a
commit ee1736f39a
13 changed files with 582 additions and 13 deletions
@@ -19,6 +19,7 @@ class TestPrefillCudaGraphPadding(CustomTestCase):
    def _make_runner(self):
        runner = PrefillCudaGraphRunner.__new__(PrefillCudaGraphRunner)
        runner._is_full_backend = False
+        runner.enable_lora = False
        runner.prefill_backend_name = Backend.TC_PIECEWISE
        runner.has_mha_companion_layers = False
        runner.capture_hidden_mode = CaptureHiddenMode.NULL