Update README.md

README.md

@@ -21,7 +21,7 @@ With this fine-tune Mistral 7B learns:
 Test it with Fireworks.ai: https://colab.research.google.com/drive/1hGIw6N20Hv8YyNTLBoYpll7AqjzTyRVp?usp=sharing
 
 
-The training data for this model included adult content.
+The training data for the underlying model Mistral 7B included adult content, this model does not add additional guardrails and is not suitable for raw use in all enviornments. 
 
 ## Purpose of the model
 
@@ -38,8 +38,8 @@ Outputting in a structured data format also allows for adding chain of thought s
 
     You must return valid JSON with the keys **scenarioType**, content, and continuationOptions. **'scenarioType' is based on the reader instruction. If the reader instruction is romantic choose intimacy, otherwise choose adventure. Write as if "Scenario Type:" is the scenarioType.**
 
-    <quote>I'm hungry</quote> =>  "scenarioType": "adventure"
-    <action>I hug her</action> => "scenarioType": "intimacy"
+    <quote>I'm hungry, let's go eat.</quote> =>  "scenarioType": "adventure"
+    <quote>Where were you?</quote> => "scenarioType": "discussion"
     
 This demonstrates that during training the model has picked up on the meanings of various scenario types in a way that allows it to change its style on the fly in response to user input
 
@@ -122,7 +122,7 @@ You'll get a response in the format:
 If you get nonsense outputs make sure you haven't missed the indentations and newlines in the prompt format (`[/INST]` should have no whitespace before it and be on a seperate line)
 
 
-Scenario Type can be "adventure","counsel","intimacy", or "confrontation", to tweak the method of replying
+Scenario Type can be "adventure","counsel","discussion", or "confrontation", to tweak the method of replying