Stories visualization is very slow and error

I want to generate stories visualization, but it takes long time and it eventually failed.

$ python -m rasa_core.visualize -d domain.yml -s data/stories.md -o graph1104.png

/Users/wisionlearning/anaconda3/lib/python3.6/site-packages/pykwalify/core.py:99: UnsafeLoaderWarning: 

The default 'Loader' for 'load(stream)' without further arguments can be unsafe.

Use 'load(stream, Loader=ruamel.yaml.Loader)' explicitly if that is OK.

Alternatively include the following in your code:

import warnings

warnings.simplefilter('ignore', ruamel.yaml.error.UnsafeLoaderWarning)

In most other cases you should consider using 'safe_load(stream)'

data = yaml.load(stream)

2018-11-04 00:20:54 **INFO** __main__ - Starting to visualize stories...

Processed Story Blocks: 100%|██| 43/43 [00:00<00:00, 176.19it/s, # trackers=105]

Processed Story Blocks: 84%|▊| 36/43 [05:41<01:06, 9.49s/it, # trackers=224253Processed Story Blocks: 86%|▊| 37/43 [09:46<01:35, 15.86s/it, # trackers=224253Processed Story Blocks: 86%|▊| 37/43 [09:52<01:36, 16.01s/it, # trackers=224253Processed Story Blocks: 88%|▉| 38/43 [13:21<01:45, 21.10s/it, # trackers=224253Processed Story Blocks: 88%|▉| 38/43 [13:32<01:46, 21.38s/it, # trackers=448506Processed Story Blocks: 91%|▉| 39/43 [26:17<02:41, 40.45s/it, # trackers=448506Processed Story Blocks: 91%|▉| 39/43 [26:32<02:43, 40.82s/it, # trackers=224253Processed Story Blocks: 93%|▉| 40/43 [35:23<02:39, 53.10s/it, # trackers=224253Processed Story Blocks: 93%|▉| 40/43 [35:26<02:39, 53.15s/it, # trackers=37373]Processed Story Blocks: 98%|▉| 42/43 [37:06<00:53, 53.01s/it, # trackers=791924Processed Story Blocks: 100%|█| 43/43 [1:12:27<00:00, 101.10s/it, # trackers=791924]

Processed Story Blocks: 5%| | 2/43 [34:24<11:45:22, 1032.25s/it, # trackers=44Processed Story Blocks: 5%| | 2/43 [37:57<12:58:15, 1138.91s/it, # trackers=2032459]Killed: 9

I use lots of check points because there are blocks in my stories and the users can loop back to beginning if they want.
domain.yml (11.9 KB)stories.md (6.8 KB)

Before I used interactive learning to build my stories. There were 35+ stories. It took a lot of work. But then, when I run the bot, sometimes went to the wrong path using KerasPolicy. Also the visualization from those 35+ stories was not the same as what I constructed the stories. That’s why I tried to use check point. However, it seems not working.

This is still a problem today, and it has to do with the fact that the visualize script actually writes out the stories (correct me if I am wrong), and if you have loops, there’s infinitely many outcomes to the stories, so the script never stops (and eventually just freezes your computer from trying out too many cases). The visualize script should be modified to take into account check points. Can anyone from RASA have a look into this? It would be extremely helpful. Ideally, the visualize script should run within a second… all that data is usually hand-written, so generating a graph out of those files should not take time at all…

Any update on this?