Closed ioana-blue closed 5 years ago
Starting from observations from #10 I think I found a bug in infer function:
https://github.com/CoderPat/structured-neural-summarization/blob/master/train_and_eval.py#L419
Look at the parameters that are passed:
features_file=args.train_source_file, labels_file=args.train_target_file,
I think those files should be the inference files not the train files. This code does inference on train files. That's why I was getting thousands of outputs for an inference file with 2 samples.
Starting from observations from #10 I think I found a bug in infer function:
https://github.com/CoderPat/structured-neural-summarization/blob/master/train_and_eval.py#L419
Look at the parameters that are passed:
I think those files should be the inference files not the train files. This code does inference on train files. That's why I was getting thousands of outputs for an inference file with 2 samples.