GraphRuntime Module Run Failed With Some Strange Log Results

Hi: The workloads that failed

*********** workloads failed *********
('dense_small_batch.cuda', ('TENSOR', (3444736, 1), 'float32'), ('TENSOR', (4, 1), 'float32'), None, 'float32')
('dense_large_batch.cuda', ('TENSOR', (3444736, 1), 'float32'), ('TENSOR', (4, 1), 'float32'), None, 'float32')
('dense_small_batch.cuda', ('TENSOR', (861184, 1), 'float32'), ('TENSOR', (4, 1), 'float32'), None, 'float32')
('dense_large_batch.cuda', ('TENSOR', (861184, 1), 'float32'), ('TENSOR', (4, 1), 'float32'), None, 'float32')
('dense_small_batch.cuda', ('TENSOR', (215296, 1), 'float32'), ('TENSOR', (4, 1), 'float32'), None, 'float32')
('dense_large_batch.cuda', ('TENSOR', (215296, 1), 'float32'), ('TENSOR', (4, 1), 'float32'), None, 'float32')

The environment:

GPU: Tesla-T4,

CUDA version: CUDA10

OS: ubuntu-16.04

And I am wondering that even any valid config for some workloads can not be found, is there at least a default schedule which can be used to generate a workable runtime module?