Separate feature configurations (#42)

* Updated config to use Multistep and updated rasa-pro version.

* Added e2e test multistep folder with symlinks to relevant test.

* Added config folder with multistep config.

* Revert "Updated config to use Multistep and updated rasa-pro version."

This reverts commit 4a094ba.

* Updated rasa-pro to 3.9.3.

* Added seperate qdrant config and updated makefile.

* Updated the readme with the changes to config and new rasa version.

* Updated the poetry lock file.

* Correct typo.

Makefile

@@ -20,7 +20,13 @@ rasa-run: .EXPORT_ALL_VARIABLES
 	rasa inspect --debug
 
 rasa-train: .EXPORT_ALL_VARIABLES
-	rasa train -c config.yml -d domain --data data
+	rasa train -c config/config.yml -d domain --data data
+
+rasa-train-multistep: .EXPORT_ALL_VARIABLES
+	rasa train -c config/multistep-config.yml -d domain --data data
+
+rasa-train-qdrant: .EXPORT_ALL_VARIABLES
+	rasa train -c config/qdrant-config.yml -d domain --data data
 
 rasa-actions:
 	rasa run actions
@@ -34,6 +40,9 @@ rasa-test-flaky: .EXPORT_ALL_VARIABLES
 rasa-test-failing: .EXPORT_ALL_VARIABLES
 	rasa test e2e e2e_tests/failing
 
+rasa-test-multistep: .EXPORT_ALL_VARIABLES
+	rasa test e2e e2e_tests/multistep
+
 rasa-test-one: .EXPORT_ALL_VARIABLES
 	rasa test e2e $(target) --debug
 
@@ -54,7 +63,13 @@ run: .EXPORT_ALL_VARIABLES run-duckling
 	poetry run rasa inspect --debug
 
 train: .EXPORT_ALL_VARIABLES
-	poetry run rasa train -c config.yml -d domain --data data
+	poetry run rasa train -c config/config.yml -d domain --data data
+
+train-multistep: .EXPORT_ALL_VARIABLES
+	poetry run rasa train -c config/multistep-config.yml -d domain --data data
+
+train-qdrant: .EXPORT_ALL_VARIABLES
+	poetry run rasa train -c config/qdrant-config.yml -d domain --data data
 
 actions:
 	poetry run rasa run actions
@@ -68,6 +83,9 @@ test-flaky: .EXPORT_ALL_VARIABLES
 test-failing: .EXPORT_ALL_VARIABLES
 	poetry run rasa test e2e e2e_tests/failing
 
+test-multistep: .EXPORT_ALL_VARIABLES
+	poetry run rasa test e2e e2e_tests/multistep
+
 test-one: .EXPORT_ALL_VARIABLES
 	poetry run rasa test e2e $(target) --debug
 

README.md

@@ -9,7 +9,7 @@ This demo showcases a chatbot built with Rasa's LLM-native approach: [CALM](http
 > CALM's current stage of development.
 
 > [!NOTE]
-> This demo bot is currently compatible with `3.9.1`.
+> This demo bot is currently compatible with `3.9.3`.
 
 ## Terms of Use
 
@@ -280,7 +280,7 @@ After you cloned the repository, follow these installation steps:
    RASA_DUCKLING_HTTP_URL=<url to the duckling server>
    ```
 
-5. Set up the extractive search:
+5. If using qdrant for extractive search:
    - Setup a local docker instance of Qdrant
       ```
       docker pull qdrant/qdrant
@@ -314,8 +314,11 @@ vector_store:
 
 This configuration refers to `addons/qdrant.py` file and the class `Qdrant_Store`. This class is also an example that information retrievers can use a custom query, note that in `search()` function the query is rewritten using the chat transcript by `prepare_search_query` function.
 
+### Configuration
 
-Check `config.yml` to make sure the configuration is appropriate before you train and run the bot.
+Check `config/config.yml` to make sure the configuration is appropriate before you train and run the bot.
+There are some alternative configurations available in the config folder. 
+Theses can be used via the appropriate `make` command during training.
 
 ### Training the bot
 
@@ -325,9 +328,20 @@ make rasa-train
 ```
 which is a shortcut for:
 ```commandline
-rasa train -c config.yml -d domain --data data
+rasa train -c config/config.yml -d domain --data data
 ```
 
+Alternative configurations can be accessed for Multistep command generation:
+```commandline
+make rasa-train-multistep
+```
+or for Enterprise search with qdrant for extractive search:
+```commandline
+make rasa-train-qdrant
+```
+
+
+
 The trained model is stored in `models` directory located in the project root.
 
 ### Starting the assistant
@@ -418,3 +432,16 @@ or
 ```commandline
 rasa test e2e e2e/tests/path/to/a/target/test.yml
 ```
+
+------
+
+To run only the tests which are relevant to multistep command generation, 
+you can use the make command:
+
+```commandline
+make rasa-test-multistep
+```
+or
+```commandline
+run rasa test e2e e2e_tests/multistep
+```

config/config.yml

@@ -0,0 +1,44 @@
+recipe: default.v1
+language: en
+pipeline:
+- name: LLMBasedRouter
+  calm_entry:
+    sticky: "handles everything else, including finance and pizza orders"
+  nlu_entry:
+    sticky: "handles everything around health advice and hotel or restaurant bookings"
+    non_sticky: "handles chitchat"
+- name: WhitespaceTokenizer
+- name: CountVectorsFeaturizer
+- name: CountVectorsFeaturizer
+  analyzer: char_wb
+  min_ngram: 1
+  max_ngram: 4
+- name: LogisticRegressionClassifier
+  max_iter: 100
+  solver: lbfgs
+  tol: 0.0001
+  random_state: 42
+  ranking_length: 10
+- name: CRFEntityExtractor
+- name: DucklingEntityExtractor
+  url: "http://localhost:8000"
+  dimensions: [ "time", "number" ]
+  timezone: "Europe/Berlin"
+  timeout: 3
+- name: NLUCommandAdapter
+- name: SingleStepLLMCommandGenerator
+  llm:
+    model_name: gpt-4
+    request_timeout: 7
+    temperature: 0.0
+    top_p: 0.0
+
+policies:
+- name: RulePolicy
+- name: TEDPolicy
+  constrain_similarities: true
+- name: FlowPolicy
+- name: EnterpriseSearchPolicy
+- name: IntentlessPolicy
+
+assistant_id: 20240418-073244-narrow-archive

config/multistep-config.yml

@@ -0,0 +1,42 @@
+recipe: default.v1
+language: en
+pipeline:
+- name: LLMBasedRouter
+  calm_entry:
+    sticky: "handles everything else, including finance and pizza orders"
+  nlu_entry:
+    sticky: "handles everything around health advice and hotel or restaurant bookings"
+    non_sticky: "handles chitchat"
+- name: WhitespaceTokenizer
+- name: CountVectorsFeaturizer
+- name: CountVectorsFeaturizer
+  analyzer: char_wb
+  min_ngram: 1
+  max_ngram: 4
+- name: LogisticRegressionClassifier
+  max_iter: 100
+  solver: lbfgs
+  tol: 0.0001
+  random_state: 42
+  ranking_length: 10
+- name: CRFEntityExtractor
+- name: DucklingEntityExtractor
+  url: "http://localhost:8000"
+  dimensions: [ "time", "number" ]
+  timezone: "Europe/Berlin"
+  timeout: 3
+- name: NLUCommandAdapter
+- name: MultiStepLLMCommandGenerator
+  llm:
+    model_name: gpt-3.5-turbo-0125
+    request_timeout: 7
+    temperature: 0.0
+    top_p: 0.0
+
+policies:
+- name: RulePolicy
+- name: TEDPolicy
+  constrain_similarities: true
+- name: FlowPolicy
+
+assistant_id: 20240418-073244-narrow-archive

e2e_tests/multistep/cancellations

@@ -0,0 +1 @@
+../passing/cancellations

e2e_tests/multistep/corrections

@@ -0,0 +1 @@
+../passing/corrections

e2e_tests/multistep/disambiguation

@@ -0,0 +1 @@
+../passing/disambiguation

e2e_tests/multistep/flow_guards

@@ -0,0 +1 @@
+../passing/flow_guards

e2e_tests/multistep/happy_path

@@ -0,0 +1 @@
+../passing/happy_path

e2e_tests/multistep/intermittent_flow_switch.yml

@@ -0,0 +1 @@
+../passing/digressions/intermittent_flow_switch.yml

e2e_tests/multistep/invalid_path

@@ -0,0 +1 @@
+../passing/invalid_path

e2e_tests/multistep/invalid_user_inputs

@@ -0,0 +1 @@
+../passing/invalid_user_inputs

e2e_tests/multistep/negations

@@ -0,0 +1 @@
+../passing/negations

e2e_tests/multistep/potential_bugs

@@ -0,0 +1 @@
+../passing/potential_bugs

e2e_tests/multistep/user_starts_multiple_flows.yml

@@ -0,0 +1 @@
+../passing/digressions/user_starts_multiple_flows.yml

e2e_tests/multistep/user_starts_with_a_very_long_message.yml

@@ -0,0 +1 @@
+../passing/digressions/user_starts_with_a_very_long_message.yml

e2e_tests/multistep/validation_too_little_money.yml

@@ -0,0 +1 @@
+../passing/digressions/validation_too_little_money.yml

e2e_tests/passing/happy_path/user_fills_resetting_slot.yml

@@ -19,4 +19,3 @@ test_cases:
       - slot_was_not_set:
         - based_in_california
       - utter: utter_can_do_something_else
-