Paddle test_fuse_optimizer_pass fails on accuracy

js81xvg6  于 2021-12-07  发布在  Java
关注(0)|答案(0)|浏览(197)

Error

True or garbage collection strategy is disabled, which is not recommended
342: test_fuse_optimizer_pass failed
342:  F...F.....
342: ======================================================================
342: FAIL: test_batchnorm_fc_with_fuse_op (test_fuse_optimizer_pass.TestFuseAdamOps)
342: ----------------------------------------------------------------------
342: Traceback (most recent call last):
342:   File "/home/luotao/Paddle/cpu_build/python/paddle/fluid/tests/unittests/test_fuse_optimizer_pass.py", line 79, in test_batchnorm_fc_with_fuse_op
342:     fc_with_batchnorm, False, optimizer=self.optimizer)
342:   File "/home/luotao/Paddle/cpu_build/python/paddle/fluid/tests/unittests/test_fuse_optimizer_pass.py", line 68, in _decorate_compare_fused_optimizer_ops
342:     optimizer=optimizer)
342:   File "/home/luotao/Paddle/cpu_build/python/paddle/fluid/tests/unittests/test_fuse_optimizer_pass.py", line 61, in _compare_fused_optimizer_ops
342:     self.assertAlmostEquals(loss[0], loss[1], delta=1e-6)
342: AssertionError: 1e+20 != 0.018401487 within 1e-06 delta
342:
342: ======================================================================
342: FAIL: test_batchnorm_fc_with_fuse_op (test_fuse_optimizer_pass.TestFuseSGDOps)
342: ----------------------------------------------------------------------
342: Traceback (most recent call last):
342:   File "/home/luotao/Paddle/cpu_build/python/paddle/fluid/tests/unittests/test_fuse_optimizer_pass.py", line 79, in test_batchnorm_fc_with_fuse_op
342:     fc_with_batchnorm, False, optimizer=self.optimizer)
342:   File "/home/luotao/Paddle/cpu_build/python/paddle/fluid/tests/unittests/test_fuse_optimizer_pass.py", line 68, in _decorate_compare_fused_optimizer_ops
342:     optimizer=optimizer)
342:   File "/home/luotao/Paddle/cpu_build/python/paddle/fluid/tests/unittests/test_fuse_optimizer_pass.py", line 61, in _compare_fused_optimizer_ops
342:     self.assertAlmostEquals(loss[0], loss[1], delta=1e-6)
342: AssertionError: 1e+20 != 0.81384623 within 1e-06 delta
342:
342: ----------------------------------------------------------------------
342: Ran 10 tests in 21.841s
342:
342: FAILED (failures=2)
342:
342: [3.1740608 2.7822773 3.4848244 2.4133034] [1.e+20 1.e+20 1.e+20 1.e+20]
342: [3.1740608 2.7822773 3.4848244 2.4133034] [0.01840149 0.02286847 0.01914792 0.01896859]
342: [3.1740608 2.7822773 3.4848244 2.4133034] [0.71334237 0.61988837 0.6804775  0.54394   ]
342: [3.1740608 2.7822773 3.4848244 2.4133034] [0.71334237 0.61988837 0.6804775  0.54394   ]
342: [3.1740608 2.7822773 3.4848244 2.4133034] [1.e+20 1.e+20 1.e+20 1.e+20]
342: [3.1740608 2.7822773 3.4848244 2.4133034] [0.81384623 0.7007753  0.7889953  0.6134242 ]
342: [0.69706124 0.7727882  0.6920713  0.7056082 ] [0.00463198 0.00515448 0.00489672 0.00468711]
342: [0.69706124 0.7727882  0.6920713  0.7056082 ] [0.00463198 0.00515448 0.00489672 0.00468711]
342: [0.6950333  0.64699537 0.72892547 0.7796856 ] [0.5862186  0.5587176  0.60925025 0.64154977]
342: [0.6950333  0.64699537 0.72892547 0.7796856 ] [0.5862186  0.5587176  0.60925025 0.64154977]
342: [0.77840513 0.7429636  0.7347602  0.7789564 ] [0.6642952 0.6306646 0.6273594 0.6834079]
342: [0.77840513 0.7429636  0.7347602  0.7789564 ] [0.6642952 0.6306646 0.6273594 0.6834079]
1/1 Test #342: test_fuse_optimizer_pass .........***Failed   27.12 sec

0% tests passed, 1 tests failed out of 1

Total Test time (real) =  27.23 sec

The following tests FAILED:
	342 - test_fuse_optimizer_pass (Failed)

How to reproduce

  • Machine: Intel(R) Xeon(R) CPU E5-2620 v2 @ 2.10GHz

  • I test it on CI machine and Intel(R) Xeon(R) CPU E5-2620 v3. It works well.

  • CMake: cmake .. -DWITH_GPU=OFF -DWITH_MKLDNN=ON -DCMAKE_BUILD_TYPE=Release -DWITH_TESTING=ON

  • Develop branch, commitid: a710ccc

暂无答案!

目前还没有任何答案,快来回答吧!

相关问题