I get a parameter error when I use a pretrained model

[coobMagicX]

I get a parameter problem when using a pretrained model under pytorch, codebase and codevecs length mismatch in search.py.

Traceback (most recent call last):
File "search.py", line 150, in
assert len(codebase)==len(codevecs), "inconsistent number of chunks, check whether the specified files for codebase and code vectors are correct!"
AssertionError: inconsistent number of chunks, check whether the specified files for codebase and code vectors are correct!

[guxd]

This is probably because you did not specify the --chunk_size argument.
The default number (2M) is set for our provided dataset. If you use your own dataset, you need to set an appropriate chunk size.

[coobMagicX]

Yes, because I used the dataset downloaded from Google Drive, I didn't modify the chunk_size at first, but they didn't match.
Now I have switched to using the project under the keras version, could you please provide the raw code datasets used for project training. Or tell me where I can get the raw code datasets used by the project. Thank you so much.

[guxd]

The raw code datasets are available at /pytorch/train.rawcode.rar