shingjan opened a new pull request, #11924:
URL: https://github.com/apache/tvm/pull/11924
Hitting this error with `vision_maskrcnn` in torchbench. Stacktrace below:
```
ERROR:torchdynamo.optimizations.backends:tvm error
Traceback (most recent call last):
File "/home/xx/torchdynamo/torchdynamo/optimizations/backends.py", line
640, in tvm_compile_inner
mod, params = relay.frontend.from_pytorch(jit_mod, shape_list)
File "/home/xx/tvm/python/tvm/relay/frontend/pytorch.py", line 4320, in
from_pytorch
outputs =
converter.convert_operators(_get_operator_nodes(graph.nodes()), outputs,
ret_name)
File "/home/xx/tvm/python/tvm/relay/frontend/pytorch.py", line 3694, in
convert_operators
relay_out = relay_op(
File "/home/xx/tvm/python/tvm/relay/frontend/pytorch.py", line 856, in
new_full
dtype = _convert_dtype_value(inputs[2])
File "/home/xx/tvm/python/tvm/relay/frontend/pytorch.py", line 3765, in
_convert_dtype_value
return _convert_data_type(convert_torch_dtype_map[val])
File "/home/xx/tvm/python/tvm/relay/frontend/pytorch.py", line 3806, in
_convert_data_type
raise NotImplementedError("input_type {} is not handled
yet".format(input_type))
NotImplementedError: input_type torch.unit8 is not handled yet
input_type torch.unit8 is not handled yet
```
reference: https://pytorch.org/docs/stable/tensors.html
cc: @masahi
--
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]
