Running 16_transformer.yaml: AssertionError: cannot pass dy.ParameterCollection to a Serializable class.

[nico3865]

I'm trying to run the 16_transformer.yaml, (untouched, except for paths to the data).

$ xnmt --dynet-gpu --settings=debug 16_transformer.yaml

I get the following error: xnmt_error_running_16_transformer_yaml.txt

I am on version 0.1.0, at commit: d9e227b, (this version is required for the "conala-baseline model").

Is this a known issue on this version, and I should upgrade xnmt? Or am I doing something obviously wrong?

N.B. The conala-baseline model runs great, on different datasets, btw, on this version. The 01_standard.yaml runs without error, also.

... Great framework aside from that, very convenient to set up experiments! I appreciate the hard work!

[SergioOyaga]

The same problem here,
Any improvement in the latest version?
my line:

python -m xnmt.xnmt_run_experiments --settings=debug xnmt/test/config/transformer.yaml

Thanks in advance :)