minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

CI: synchronize prefill graph test fixtures (#31665)

Browse Source
This commit is contained in:
Mick
2026-07-18 18:48:05 +08:00
committed by GitHub
parent faf6894093
commit 573c075fef
2 changed files with 4 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/registered/unit/configs/test_multimodal_piecewise_cuda_graph.py
+1
View File
@@ -33,6 +33,7 @@ class TestMultimodalPiecewiseCudaGraph(CustomTestCase):
runner.prefill_backend_name = backend runner.prefill_backend_name = backend
runner.has_mha_companion_layers = backend == Backend.BREAKABLE runner.has_mha_companion_layers = backend == Backend.BREAKABLE
runner.capture_hidden_mode = CaptureHiddenMode.NULL runner.capture_hidden_mode = CaptureHiddenMode.NULL
runner.capture_num_tokens = [4, 16]
runner.max_num_tokens = 16 runner.max_num_tokens = 16
return runner return runner
test/registered/unit/model_executor/runner/test_prefill_cuda_graph_padding.py
+3
View File
@@ -1,6 +1,7 @@
import unittest import unittest
from types import SimpleNamespace from types import SimpleNamespace
from sglang.srt.model_executor.cuda_graph_config import Backend
from sglang.srt.model_executor.forward_batch_info import ( from sglang.srt.model_executor.forward_batch_info import (
CaptureHiddenMode, CaptureHiddenMode,
ForwardMode, ForwardMode,
@@ -18,6 +19,8 @@ class TestPrefillCudaGraphPadding(CustomTestCase):
def _make_runner(self): def _make_runner(self):
runner = PrefillCudaGraphRunner.__new__(PrefillCudaGraphRunner) runner = PrefillCudaGraphRunner.__new__(PrefillCudaGraphRunner)
runner._is_full_backend = False runner._is_full_backend = False
runner.prefill_backend_name = Backend.TC_PIECEWISE
runner.has_mha_companion_layers = False
runner.capture_hidden_mode = CaptureHiddenMode.NULL runner.capture_hidden_mode = CaptureHiddenMode.NULL
runner.capture_num_tokens = [4, 16] runner.capture_num_tokens = [4, 16]
runner.max_num_tokens = 16 runner.max_num_tokens = 16