transformer icon indicating copy to clipboard operation
transformer copied to clipboard

error "multihead_attention/transpose"

Open tine8899 opened this issue 6 years ago • 0 comments

Traceback (most recent call last):
  File "train.py", line 172, in <module>
    g = Graph("train"); print("Graph loaded")
  File "train.py", line 80, in __init__
    causality=False)
  File "/Users/walnut/Documents/01work/masterthesis/transformer/modules.py", line 211, in multihead_attention
    outputs = tf.matmul(Q_, tf.transpose(K_, [0, 2, 1])) # (h*N, T_q, T_k)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/ops/array_ops.py", line 1409, in transpose
    ret = transpose_fn(a, perm, name=name)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/ops/gen_array_ops.py", line 8636, in transpose
    "Transpose", x=x, perm=perm, name=name)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py", line 787, in _apply_op_helper
    op_def=op_def)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/util/deprecation.py", line 454, in new_func
    return func(*args, **kwargs)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 3155, in create_op
    op_def=op_def)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1731, in __init__
    control_input_ops)
  File "/Users/walnut/anaconda2/envs/masterthesis/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1579, in _create_c_op
    raise ValueError(str(e))
ValueError: Dimension must be 4 but is 3 for 'encoder/num_blocks_0/multihead_attention/transpose' (op: 'Transpose') with input shapes: [256,32,4,512], [3].

tine8899 avatar Feb 10 '19 17:02 tine8899