gen – Performs generative model evaluations and plots results

gen.py

usage: gen.py [-h] [--model MODEL] [--val] [--start START] [--count COUNT]
              [--search SEARCH] [-p PLOT] [--exp EXP]
-h, --help

show this help message and exit

--model <model>

Model to use for pretraining.

--val

Run with the validation dataset instead of the test.

--start <start>

Index of the first audit log to use for the demo.

--count <count>

Number of audit logs to use for the demo.

Search method to use for decoding. If beam, :k is the beam size.

-p <plot>, --plot <plot>
--exp <exp>

Experiment to run.

class gen.GenerationExperiment(config, path_prefix, vocab, model, *args, **kwargs)
Parameters:

  • config (dict)

  • path_prefix (str)

  • vocab (EHRVocab)

  • model (str)

eval_generation(output_df=None, label_df=None, output_tokens=None, label_tokens=None)
examples_seen()
min_size()
on_finish()
plot()
stopping_criteria(context_length=0, total_length=0)
Parameters:

  • context_length (int)

  • total_length (int)

window_size()
class gen.NextActionExperiment(config, path_prefix, vocab, model, *args, **kwargs)
Parameters:

  • config (dict)

  • path_prefix (str)

  • vocab (EHRVocab)

  • model (str)

eval_generation(output_df=None, label_df=None, output_tokens=None, label_tokens=None)
examples_seen()
on_finish()
plot()
stopping_criteria(context_length=0, total_length=0)
Parameters:

  • context_length (int)

  • total_length (int)

window_size()
class gen.ScoringExperiment(config, path_prefix, vocab, model, *args, **kwargs)
Parameters:

  • config (dict)

  • path_prefix (str)

  • vocab (EHRVocab)

  • model (str)

eval_generation(output_df=None, label_df=None, output_tokens=None, label_tokens=None)
examples_seen()
on_finish()
plot()
stopping_criteria(context_length=0, total_length=0)
Parameters:

  • context_length (int)

  • total_length (int)

window_size()