Hello!
I am struggling with this problem for a long time, I would be glad if anyone could help! so I have quite a few story blocks where the same intent appears (mostly “affirm” and “deny”). For example here are two stories:
- story: work yes story
steps:
- action: utter_work
- intent: affirm
- action: utter_something_about_work
- story: life yes story
steps:
- action: utter_life
- intent: affirm
- action: utter_something_about_life
so there are quite a few story blocks using “intent:affirm” like shown above, but when I test, no matter what is the question before, I almost always get the same prediction after I input some words that is detected as “intent:affirm”. I will be happy about any help. Thank you! Here are my policies:
- name: MemoizationPolicy
max_history: 2
epochs: 1000
priority: 6
- name: TEDPolicy
max_history: 3
epochs: 1000
constrain_similarities: True
use_gpu: True
priority: 5
- name: RulePolicy
core_fallback_threshold: 0.2
epochs: 1000
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: True
priority: 4
- name: AugmentedMemoizationPolicy
max_history: 2
epochs: 500
priority: 3
- name: UnexpecTEDIntentPolicy
max_history: 2
epochs: 1000
tolerance: 0.2