update

ai/knowledge_base.py

@@ -1,58 +1,37 @@
+from phi.embedder.openai import OpenAIEmbedder
 from phi.knowledge.combined import CombinedKnowledgeBase
 from phi.knowledge.pdf import PDFUrlKnowledgeBase, PDFKnowledgeBase
 from phi.knowledge.website import WebsiteKnowledgeBase
 from phi.vectordb.pgvector import PgVector2
 
+from ai.settings import ai_settings
 from db.session import db_url
 
-url_pdf_knowledge_base = PDFUrlKnowledgeBase(
-    urls=["https://www.family-action.org.uk/content/uploads/2019/07/meals-more-recipes.pdf"],
-    # Store this knowledge base in ai.url_pdf_documents
-    vector_db=PgVector2(
-        schema="ai",
-        db_url=db_url,
-        collection="url_pdf_documents",
-    ),
-    # 2 references are added to the prompt
-    num_documents=2,
-)
-
-local_pdf_knowledge_base = PDFKnowledgeBase(
-    path="data/pdfs",
-    # Store this knowledge base in ai.local_pdf_documents
-    vector_db=PgVector2(
-        schema="ai",
-        db_url=db_url,
-        collection="local_pdf_documents",
-    ),
-    # 3 references are added to the prompt
-    num_documents=3,
-)
-
 pdf_knowledge_base = CombinedKnowledgeBase(
     sources=[
-        url_pdf_knowledge_base,
-        local_pdf_knowledge_base,
+        PDFUrlKnowledgeBase(urls=["https://phi-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"]),
+        PDFKnowledgeBase(path="data/pdfs"),
     ],
-    # Store this knowledge base in ai.pdf_documents
     vector_db=PgVector2(
-        schema="ai",
         db_url=db_url,
+        # Store the embeddings in ai.pdf_documents
         collection="pdf_documents",
+        embedder=OpenAIEmbedder(model=ai_settings.embedding_model),
     ),
     # 2 references are added to the prompt
     num_documents=2,
 )
 
 website_knowledge_base = WebsiteKnowledgeBase(
-    urls=["https://docs.phidata.com/introduction"],
+    # Add URLs to the knowledge base
+    # urls=["https://docs.phidata.com/introduction"],
     # Number of links to follow from the seed URLs
     max_links=15,
-    # Store this knowledge base in ai.website_documents
     vector_db=PgVector2(
-        schema="ai",
         db_url=db_url,
+        # Store the embeddings in ai.website_documents
         collection="website_documents",
+        embedder=OpenAIEmbedder(model=ai_settings.embedding_model),
     ),
     # 3 references are added to the prompt
     num_documents=3,

ai/settings.py

@@ -12,7 +12,7 @@ class AISettings(BaseSettings):
     gpt_3_5: str = "gpt-3.5-turbo-1106"
     dall_e: str = "dall-e-3"
     whisper: str = "whisper-1"
-    embedding_model: str = "text-embedding-ada-002"
+    embedding_model: str = "text-embedding-3-small"
     default_max_tokens: int = 1024
     default_temperature: float = 0
 

ai/storage.py

@@ -3,19 +3,16 @@
 from db.session import db_url
 
 pdf_assistant_storage = PgAssistantStorage(
-    schema="ai",
     db_url=db_url,
     table_name="pdf_assistant",
 )
 
 image_assistant_storage = PgAssistantStorage(
-    schema="ai",
     db_url=db_url,
     table_name="image_assistant",
 )
 
 website_assistant_storage = PgAssistantStorage(
-    schema="ai",
     db_url=db_url,
     table_name="website_assistant",
 )

pyproject.toml

@@ -46,7 +46,7 @@ dependencies = [
   # Linting and Formatting
   "ruff",
   # phidata
-  "phidata==2.3.19",
+  "phidata==2.3.35",
 ]
 
 [build-system]

recipe/assistant_with_storage.py

@@ -10,7 +10,7 @@
 from db.session import db_url
 
 knowledge_base = PDFUrlKnowledgeBase(
-    urls=["https://www.family-action.org.uk/content/uploads/2019/07/meals-more-recipes.pdf"],
+    urls=["https://phi-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"],
     vector_db=PgVector2(collection="recipes", db_url=db_url),
 )
 

recipe/auto_assistant.py

@@ -5,7 +5,7 @@
 from db.session import db_url
 
 knowledge_base = PDFUrlKnowledgeBase(
-    urls=["https://www.family-action.org.uk/content/uploads/2019/07/meals-more-recipes.pdf"],
+    urls=["https://phi-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"],
     vector_db=PgVector2(collection="recipes", db_url=db_url),
 )
 knowledge_base.load(recreate=False)

recipe/rag_assistant.py

@@ -5,7 +5,7 @@
 from db.session import db_url
 
 knowledge_base = PDFUrlKnowledgeBase(
-    urls=["https://www.family-action.org.uk/content/uploads/2019/07/meals-more-recipes.pdf"],
+    urls=["https://phi-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"],
     vector_db=PgVector2(collection="recipes", db_url=db_url),
 )
 knowledge_base.load(recreate=False)

recipe/recipe_assistant_openai.py

@@ -5,7 +5,7 @@
 
 # Load the recipe book from s3
 recipe_book_s3 = UrlFile(
-    url="https://www.family-action.org.uk/content/uploads/2019/07/meals-more-recipes.pdf"
+    url="https://phi-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
 ).get_or_create()
 # Load the recipe book locally
 # recipe_book_local = LocalFile(path="data/pdfs/meals-more-recipes.pdf").get_or_create()

workspace/dev_resources.py

@@ -34,7 +34,6 @@
     pg_database="app",
     # Connect to this db on port 5432
     host_port=5432,
-    debug_mode=True,
 )
 
 # -*- Build container environment