Changed similarity value to work with RAG Settings instead of LLM Settings

private_gpt/server/chat/chat_service.py

@@ -118,7 +118,7 @@ class ChatService:
                 node_postprocessors=[
                     MetadataReplacementPostProcessor(target_metadata_key="window"),
                     SimilarityPostprocessor(
-                        similarity_cutoff=settings.llm.similarity_value
+                        similarity_cutoff=settings.rag.similarity_value
                     ),
                 ],
             )

private_gpt/settings/settings.py

@@ -107,6 +107,12 @@ class LLMSettings(BaseModel):
         description="If set, any documents retrieved from the RAG must meet a certain score. Acceptable values are between 0 and 1.",
     )
 
+class RagSettings(BaseModel):
+    similarity_value: float = Field(
+        None,
+        description="If set, any documents retrieved from the RAG must meet a certain score. Acceptable values are between 0 and 1.",
+    )
+
 
 class VectorstoreSettings(BaseModel):
     database: Literal["chroma", "qdrant", "pgvector"]
@@ -346,6 +352,7 @@ class Settings(BaseModel):
     data: DataSettings
     ui: UISettings
     llm: LLMSettings
+    rag: RagSettings
     embedding: EmbeddingSettings
     llamacpp: LlamaCPPSettings
     huggingface: HuggingFaceSettings

settings.yaml

@@ -40,7 +40,9 @@ llm:
   max_new_tokens: 512
   context_window: 3900
   temperature: 0.1      # The temperature of the model. Increasing the temperature will make the model answer more creatively. A value of 0.1 would be more factual. (Default: 0.1)
-  #similarity_value: 0.4    #If set, any documents retrieved from the RAG must meet a certain score. Acceptable values are between 0 and 1.
+
+rag:
+  #similarity_value: 0.4    #Optional - If set, any documents retrieved from the RAG must meet a certain score. Acceptable values are between 0 and 1.
 
 llamacpp:
   prompt_style: "mistral"