Error saving HumpbackNorthAtlantic

[alpha-tango]

f08472f in acodet/funcs.py around L446 reverted fixes I made in an earlier commit, a436d05. I am now getting this error (and can fix it by reimplementing the .item() calls).

 File "mydir/acodet/run.py", line 70, in main
    run_training(**kwargs)
  File "mydir/acodet/acodet/train.py", line 334, in run_training
    save_model_results(checkpoint_dir, result)
  File "mydir/acodet/acodet/funcs.py", line 446, in save_model_results
    json.dump(result, f)
  File "/homedir/.local/share/uv/python/cpython-3.11.14-linux-x86_64-gnu/lib/python3.11/json/__init__.py", line 179, in dump
    for chunk in iterable:
  File "/homedir/.local/share/uv/python/cpython-3.11.14-linux-x86_64-gnu/lib/python3.11/json/encoder.py", line 432, in _iterencode
    yield from _iterencode_dict(o, _current_indent_level)
  File "/homedir/.local/share/uv/python/cpython-3.11.14-linux-x86_64-gnu/lib/python3.11/json/encoder.py", line 406, in _iterencode_dict
    yield from chunks
  File "/homedir/.local/share/uv/python/cpython-3.11.14-linux-x86_64-gnu/lib/python3.11/json/encoder.py", line 326, in _iterencode_list
    yield from chunks
  File "/homedir/.local/share/uv/python/cpython-3.11.14-linux-x86_64-gnu/lib/python3.11/json/encoder.py", line 439, in _iterencode
    o = _default(o)
        ^^^^^^^^^^^
  File "/homedir/.local/share/uv/python/cpython-3.11.14-linux-x86_64-gnu/lib/python3.11/json/encoder.py", line 180, in default
    raise TypeError(f'Object of type {o.__class__.__name__} '
TypeError: Object of type ndarray is not JSON serializable

@vskode I assume you made the change because you were getting an error there? What was it?

[alpha-tango]

Also the accuracy stats in training printouts look funky -- I'm seeing values >> 100 for accuracy etc.

[vskode]

that's strange, it's working for me for all model types. could you maybe just do a list(n) instead? That way it should still work for all of them and turn the ndarrays into lists which json wants?

[alpha-tango]

@vskode I had to make it a tad more complicated than that, please reopen if my fix errors for you.

[alpha-tango]

Perhaps my numpy version change is what was making the difference, My fix is now breaking for me. I will look into it more.

Traceback (most recent call last):
  File "mydir/acodet/run.py", line 83, in <module>
    main()
  File "mydir/acodet/run.py", line 70, in main
    run_training(**kwargs)
  File "mydir/acodet/acodet/train.py", line 341, in run_training
    save_model_results(checkpoint_dir, result)
  File "mydir/acodet/acodet/funcs.py", line 441, in save_model_results
    result["fbeta"] = [[float(i) for i in n] for n in result["fbeta"]]
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "mydir/acodet/acodet/funcs.py", line 441, in <listcomp>
    result["fbeta"] = [[float(i) for i in n] for n in result["fbeta"]]
                       ^^^^^^^^^^^^^^^^^^^^^
TypeError: 'float' object is not iterable