Hai, I am facing this issue for a very long time, but i have been clearing the errors somehow and make it work properly. But this time, i am unable to solve the issue. I create the code in y vscode, it works all fine. and then i am pushing the exact same code to the github(workflow has been written for pushing these commits to the azure server, so whenever the code is pushed to the github, it gets updated in the server automatically and i will be seeing the chat in my webpage). whenever i upload it there, the whole flow gets collapsed. the code exactly loos the same, but the stories and steps are randomly getting picked. can someone help me with what i can do for this issue. i am really struggling to solve this.
Hi Shruti!
Your case is intriguing. Each time we train our data, it is created a slightly different model from the previous one, cause of the nature of how AI works(basicly randomicity).
But this little difference should not collapse your flow!
Try to check some hyperparameters in your config.yml: epochs training, fallback thresholds, Policies configuration…
And also try to run Rasa in “debug mode”, to check the logs to understand your bots behavior.
Hi, Thanks, I will look into it.