Interactive learning issue

Rasa 1.0.1 Rasa nlu 0.15.1

Hi, when I try to use interactive learning with the command: “rasa interactive --verbose” (see the post: rasa interactive training is failing · Issue #3631 · RasaHQ/rasa · GitHub), I get that error message: Bot loaded. Visualisation at http://localhost:5006/visualization.html. Type a message and press enter (press ‘Ctr-c’ to exit). Processed Story Blocks: 100%|█| 123/123 [00:00<00:00, 1018.71it/s, # trackers=1] ? Your input -> hi
2019-07-01 22:49:43 ERROR rasa.core.training.interactive - An exception occurred while recording messages. Traceback (most recent call last): File “/home/flo/env/sotav/lib/python3.6/site-packages/rasa/core/training/interactive.py”, line 1347, in record_messages await _validate_nlu(intents, endpoint, sender_id) File “/home/flo/env/sotav/lib/python3.6/site-packages/rasa/core/training/interactive.py”, line 1124, in _validate_nlu valid = await _validate_user_text(latest_message, endpoint, sender_id) File “/home/flo/env/sotav/lib/python3.6/site-packages/rasa/core/training/interactive.py”, line 1086, in _validate_user_text text = _as_md_message(parse_data) File “/home/flo/env/sotav/lib/python3.6/site-packages/rasa/core/training/interactive.py”, line 1060, in _as_md_message return MarkdownWriter()._generate_message_md(parse_data) File “/home/flo/env/sotav/lib/python3.6/site-packages/rasa/nlu/training_data/formats/markdown.py”, line 277, in _generate_message_md entities = sorted(message.get(“entities”, []), key=lambda k: k[“start”]) File “/home/flo/env/sotav/lib/python3.6/site-packages/rasa/nlu/training_data/formats/markdown.py”, line 277, in entities = sorted(message.get(“entities”, []), key=lambda k: k[“start”]) KeyError: ‘start’ 2019-07-01 22:49:43 ERROR asyncio - Task exception was never retrieved future: <Task finished coro=<_serve_application..run_interactive_io() done, defined at /home/flo/env/sotav/lib/python3.6/site-packages/rasa/core/training/interactive.py:1391> exception=KeyError(‘start’,)>

Add-on: the interactive learning only works if I feed it with the names of intents, for example: /greet.

But if I type the full messages inside the intents, for example, “hi”, or “how are you”, then the interactive learning crashes with the message above. Any idea how to fix that? Thanks.

I don’t know the error code, however, your addon:

the interactive learning only works if I feed it with the names of intents, for example: /greet.

Makes it pretty clear that your NLU model isn’t loading correctly. You’ll either have to retrain your NLU model and point the interactive learning to its location or there’s something else going on with your NLU files.