Empty (or some spaces) input text causes Runtime error

[whustan]

Rasa Open Source version

3.0.0

Rasa SDK version

3.0.0

Rasa X version

No response

Python version

3.8

What operating system are you using?

Linux

What happened?

In NLU shell, when I input text="" or text=" "(some spaces), it gets error as follow (full details in 'Relevant log output') :
rasa.engine.exceptions.GraphComponentException: Error running graph component for node run_DIETClassifier4.

But in RASA-2.8.6, everything goes well and gets following response:
{ "text": "", "intent": { "name": null, "confidence": 0.0 }, "entities": [] }

I suppose this is a new bug.

Command / Request

No response

Relevant log output

Traceback (most recent call last):
File "~/venv/lib/python3.8/site-packages/rasa/engine/graph.py", line 461, in __call__
output = self._fn(self._component, **run_kwargs)
File "~/venv/lib/python3.8/site-packages/rasa/nlu/classifiers/diet_classifier.py", line 1003, in process
out = self._predict(message)
File "~/venv/lib/python3.8/site-packages/rasa/nlu/classifiers/diet_classifier.py", line 932, in _predict
return self.model.run_inference(model_data)
File "~/venv/lib/python3.8/site-packages/rasa/utils/tensorflow/models.py", line 318, in run_inference
] = self._rasa_predict(batch_in)
File "~/venv/lib/python3.8/site-packages/rasa/utils/tensorflow/models.py", line 280, in _rasa_predict
outputs = tf_utils.sync_to_numpy_or_python_type(self.predict_step(batch_in))
File "~/venv/lib/python3.8/site-packages/rasa/utils/tensorflow/models.py", line 228, in predict_step
return self.batch_predict(batch_in)
File "~/venv/lib/python3.8/site-packages/rasa/nlu/classifiers/diet_classifier.py", line 1721, in batch_predict
tf_batch_data = self.batch_to_model_data_format(
File "~/venv/lib/python3.8/site-packages/rasa/utils/tensorflow/models.py", line 473, in batch_to_model_data_format
tensor, idx = RasaModel._convert_sparse_features(
File "~/venv/lib/python3.8/site-packages/rasa/utils/tensorflow/models.py", line 518, in _convert_sparse_features
shape = [batch[idx + 2][i] for i in range(number_of_dimensions - 1)] + [
File "~/venv/lib/python3.8/site-packages/rasa/utils/tensorflow/models.py", line 518, in <listcomp>
shape = [batch[idx + 2][i] for i in range(number_of_dimensions - 1)] + [
IndexError: index 1 is out of bounds for axis 0 with size 1

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "~/venv/bin/rasa", line 8, in <module>
sys.exit(main())
File "~/venv/lib/python3.8/site-packages/rasa/__main__.py", line 121, in main
cmdline_arguments.func(cmdline_arguments)
File "~/venv/lib/python3.8/site-packages/rasa/cli/shell.py", line 91, in shell_nlu
rasa.nlu.run.run_cmdline(model)
File "~/venv/lib/python3.8/site-packages/rasa/nlu/run.py", line 25, in run_cmdline
result = asyncio.run(agent.parse_message(message))
File "/usr/lib/python3.8/asyncio/runners.py", line 44, in run
return loop.run_until_complete(main)
File "uvloop/loop.pyx", line 1456, in uvloop.loop.Loop.run_until_complete
File "~/venv/lib/python3.8/site-packages/rasa/core/agent.py", line 419, in parse_message
return await self.processor.parse_message(message)
File "~/venv/lib/python3.8/site-packages/rasa/core/processor.py", line 635, in parse_message
parse_data = self._parse_message_with_graph(message, only_output_properties)
File "~/venv/lib/python3.8/site-packages/rasa/core/processor.py", line 659, in _parse_message_with_graph
results = self.graph_runner.run(
File "~/venv/lib/python3.8/site-packages/rasa/engine/runner/dask.py", line 106, in run
dask_result = dask.get(run_graph, run_targets)
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 560, in get_sync
return get_async(
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 503, in get_async
for key, res_info, failed in queue_get(queue).result():
File "/usr/lib/python3.8/concurrent/futures/_base.py", line 437, in result
return self.__get_result()
File "/usr/lib/python3.8/concurrent/futures/_base.py", line 389, in __get_result
raise self._exception
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 545, in submit
fut.set_result(fn(*args, **kwargs))
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 237, in batch_execute_tasks
return [execute_task(*a) for a in it]
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 237, in <listcomp>
return [execute_task(*a) for a in it]
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 228, in execute_task
result = pack_exception(e, dumps)
File "~/venv/lib/python3.8/site-packages/dask/local.py", line 223, in execute_task
result = _execute_task(task, data)
File "~/venv/lib/python3.8/site-packages/dask/core.py", line 121, in _execute_task
return func(*(_execute_task(a, cache) for a in args))
File "~/venv/lib/python3.8/site-packages/rasa/engine/graph.py", line 468, in __call__
raise GraphComponentException(
rasa.engine.exceptions.GraphComponentException: Error running graph component for node run_DIETClassifier4.

Definition of done

• Fix bug

[sara-tagger]

Thanks for raising this issue, @sebwarnke will get back to you about it soon✨

Please also check out the docs and the forum in case your issue was raised there too 🤗

[whustan]

@sara-tagger any progress??

[jupyterjazz]

Hi @whusunny,
I couldn't reproduce the error.
Here's what I did:

rasa init
rasa train
rasa shell nlu

and when I write an empty string or some spaces I get the expected output with no errors.
here's my configuration:

Rasa Version      :         3.0.0
Minimum Compatible Version: 3.0.0
Rasa SDK Version  :         3.0.0
Python Version    :         3.8.12
Operating System  :         Linux-5.11.0-43-generic-x86_64-with-glibc2.17

Can you help me reproduce it? What should I do differently?

[whustan]

Hi, @jupyterjazz
Thanks for your feedback.

Using rasa init project, it indeed runs well. When I change the pipline config as follow then training, the error comes back.

language: zh
pipeline:
  - name: "MitieNLP"
    model: "total_word_feature_extractor_zh.dat"
  - name: "JiebaTokenizer"
  - name: "RegexFeaturizer"
  - name: "MitieFeaturizer"
  - name: "DIETClassifier"
    epochs: 30
  - name: "RegexEntityExtractor"
  - name: "EntitySynonymMapper"
  - name: "FallbackClassifier"
    threshold: 0.2
    ambiguity_threshold: 0.05

Yes, I am using rasa for Chinese:)