Feat: support cuda graph for LoRA (#4115)

Co-authored-by: Beichen Ma <mabeichen12@gmail.com>
2025-04-29 02:30:44 -04:00
parent 2c3ea29476
commit 8c0cfca87d
13 changed files with 366 additions and 55 deletions
--- a/test/srt/models/lora/test_lora_backend.py
+++ b/test/srt/models/lora/test_lora_backend.py
@@ -24,7 +24,7 @@ from utils import (
    DEFAULT_PROMPTS,
    TORCH_DTYPES,
    LoRAModelCase,
-    run_batch_lora_test,
+    run_lora_test_one_by_one,
 )

 from sglang.test.test_utils import CustomTestCase, is_in_ci
@@ -42,7 +42,7 @@ class TestLoRABackend(CustomTestCase):
            )
            for torch_dtype in TORCH_DTYPES:
                for backend in BACKENDS:
-                    run_batch_lora_test(
+                    run_lora_test_one_by_one(
                        prompts,
                        model_case,
                        torch_dtype,