bug: thai2rom errors #1088
Description
Description
Encountered a couple errors when attempting to run the thai2rom engine on my GPU.
File "/home/james/projects/sandbox/.venv/lib/python3.10/site-packages/pythainlp/transliterate/thai2rom.py", line 151, in forward
sequences_output = sequences_output.index_select(
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument index in method wrapper_CUDA__index_select)
File "/home/james/projects/sandbox/.venv/lib/python3.10/site-packages/torch/nn/utils/rnn.py", line 341, in pack_padded_sequence
data, batch_sizes = _VF._pack_padded_sequence(input, lengths, batch_first)
RuntimeError: 'lengths' argument should be a 1D CPU int64 tensor, but got 1D cuda:0 Long tensor
I've fixed it locally and will be opening a PR.
Expected results
Expected to have thai2rom use my GPU with no errors.
Current results
Errors out, cannot use GPU for thai2rom
Steps to reproduce
Run thai2rom engine. Example code:
from pythainlp.transliterate import romanize
thai = "สวัสดีครับ"
engines = ["royin", "thai2rom", "thai2rom_onnx", "tltk", "lookup"]
for engine in engines:
print("Engine:", engine)
result_romanized = romanize(thai, engine=engine)
print(result_romanized + "\n")PyThaiNLP version
5.1.0
Python version
3.10.12
Operating system and version
Windows 10
More info
No response
Possible solution
No response
Files
No response