Open madoka109 opened 3 years ago
I use ./cifar_trainval.sh to train，but it occurs some problems，why?
Traceback (most recent call last): File "/environment/python/versions/miniconda3-4.7.12/envs/tf1.15/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1365, in _do_call return fn(*args) File "/environment/python/versions/miniconda3-4.7.12/envs/tf1.15/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1350, in _run_fn target_list, run_metadata) File "/environment/python/versions/miniconda3-4.7.12/envs/tf1.15/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1443, in _call_tf_sessionrun run_metadata) tensorflow.python.framework.errors_impl.InternalError: 2 root error(s) found. (0) Internal: Blas GEMM launch failed : a.shape=(128, 64), b.shape=(64, 10), m=128, n=10, k=64 [[{{node resnet/tower_0/fully_connected/dense/MatMul}}]] [[resnet/tower_0/softmax_cross_entropy_loss/assert_broadcastable/is_valid_shape/has_valid_nonscalar_shape/has_invalid_dims/concat/_1549]] (1) Internal: Blas GEMM launch failed : a.shape=(128, 64), b.shape=(64, 10), m=128, n=10, k=64 [[{{node resnet/tower_0/fully_connected/dense/MatMul}}]] 0 successful operations. 0 derived errors ignored.
I use ./cifar_trainval.sh to train，but it occurs some problems，why?
Traceback (most recent call last): File "/environment/python/versions/miniconda3-4.7.12/envs/tf1.15/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1365, in _do_call return fn(*args) File "/environment/python/versions/miniconda3-4.7.12/envs/tf1.15/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1350, in _run_fn target_list, run_metadata) File "/environment/python/versions/miniconda3-4.7.12/envs/tf1.15/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1443, in _call_tf_sessionrun run_metadata) tensorflow.python.framework.errors_impl.InternalError: 2 root error(s) found. (0) Internal: Blas GEMM launch failed : a.shape=(128, 64), b.shape=(64, 10), m=128, n=10, k=64 [[{{node resnet/tower_0/fully_connected/dense/MatMul}}]] [[resnet/tower_0/softmax_cross_entropy_loss/assert_broadcastable/is_valid_shape/has_valid_nonscalar_shape/has_invalid_dims/concat/_1549]] (1) Internal: Blas GEMM launch failed : a.shape=(128, 64), b.shape=(64, 10), m=128, n=10, k=64 [[{{node resnet/tower_0/fully_connected/dense/MatMul}}]] 0 successful operations. 0 derived errors ignored.