honghuichao opened a new issue, #13387:
URL: https://github.com/apache/tvm/issues/13387
I tested the tvm and found some bugs, as follows：
（1）in mxnet frontend.
def _mx_pad(inputs, attrs):
pad_mode = attrs.get_str("mode", None)
if pad_mode is None:
raise tvm.error.OpAttributeRequired('Attribute "mode" not found in
operator pad.')
if pad_mode not in ["constant", "edge", "reflect"]:
raise tvm.error.OpAttributeInvalid("Value " + mode + ' in attribute
"mode" is not valid')
raise tvm.error.OpAttributeInvalid("Value " + mode + ' in attribute
"mode" is not valid') mode is undeclared.
(2)cudnn.py in /python/tvm/contribute:
idx = -1
if algo_type == "fwd":
idx = _FWD_ALGOS.index(algo_name)
elif algo_type == "bwd_filter":
idx = _BWD_FILTER_ALGOS.index(algo_name)
elif algo_type == "bwd_data":
idx = _BWD_DATA_ALGOS.index(algo_name)
_BWD_DATA_ALGOS and _BWD_FILTER_ALGOS undefine.
(3)
@script
def _mirror_pad_func(data_shape, pad_width):
out = output_tensor((data_shape.shape[0],), "int64")
for i in const_range(data_shape.shape[0]):
out[i] = data_shape[i] + int64(pad_width[i][0]) +
int64(pad_width[i][1])
return out
output_tensor is not imported in this module. I thik tvm can import
tvm.te.hybrid.call /output_tensor/ in releated module for all program using
hybrid.
(4) in te_complier.py
def get_shape(shape):
"""Convert the shape to correct dtype and vars."""
ret = []
for dim in shape:
if isinstance(dim, tvm.tir.IntImm):
if libinfo()["INDEX_DEFAULT_I64"] == "ON":
ret.append(dim)
else:
val = int(dim)
assert val <= np.iinfo(np.int32).max
ret.append(tvm.tir.IntImm("int32", val))
elif isinstance(dim, tvm.tir.Any):
ret.append(te.size_var("any_dim", "int32"))
else:
ret.append(dim)
return ret
np is not define in this module.
(5) in tvm/contrib/mps:
def matmul(lhs, rhs, transa=False, transb=False):
"""Create an extern op that compute matrix mult of A and rhs with CrhsLAS
This function serves as an example on how to calle external libraries.
Parameters
----------
lhs : Tensor
The left matrix operand
rhs : Tensor
The right matrix operand
transa : bool
Whether transpose lhs
transb : bool
Whether transpose rhs
Returns
-------
C : Tensor
The result tensor.
"""
m = lhs.shape[0] if transa is False else lhs.shape[1]
n = rhs.shape[1] if transb is False else rhs.shape[0]
if transa:
m = b
if transb:
n = c
b/c is not define.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
