update embedding

Dockerfile

@@ -16,13 +16,13 @@ COPY ./start_service.sh /code/start_service.sh
 # 
 COPY ./app /code/app
 
-RUN useradd -m docker && echo "docker:docker" | chpasswd && adduser docker sudo
+# RUN useradd -m docker && echo "docker:docker" | chpasswd && adduser docker sudo
 
-RUN curl -fsSL https://ollama.com/install.sh | sh
+# RUN curl -fsSL https://ollama.com/install.sh | sh
 
-USER docker
+# USER docker
 
-RUN nohup ollama serve & sleep 5
+# RUN nohup ollama serve & sleep 5
 
 #
 # RUN chmod +x /code/start_service.sh

app/rag.py

@@ -1,3 +1,6 @@
+import os
+import logging
+
 from llama_index.core import (
     SimpleDirectoryReader,
     # VectorStoreIndex,
@@ -16,12 +19,14 @@ from llama_index.core.vector_stores import VectorStoreQuery
 from llama_index.core.indices.vector_store.base import VectorStoreIndex
 from llama_index.vector_stores.qdrant import QdrantVectorStore
 from qdrant_client import QdrantClient
-import logging
 
 from llama_index.llms.llama_cpp import LlamaCPP
 from llama_index.embeddings.fastembed import FastEmbedEmbedding
 
 
+QDRANT_API_URL = os.getenv('QDRANT_API_URL')
+QDRANT_API_KEY = os.getenv('QDRANT_API_KEY')
+
 class ChatPDF:
     logging.basicConfig(level=logging.INFO)
     logger = logging.getLogger(__name__)
@@ -56,11 +61,18 @@ class ChatPDF:
 
         self.logger.info("initializing the vector store related objects")
         # client = QdrantClient(host="localhost", port=6333)
+        # client = QdrantClient(url=QDRANT_API_URL, api_key=QDRANT_API_KEY)
         client = QdrantClient(":memory:")
-        self.vector_store = QdrantVectorStore(client=client, collection_name="rag_documents", enable_hybrid=True)
+        self.vector_store = QdrantVectorStore(
+            client=client,
+            collection_name="rag_documents",
+            # enable_hybrid=True
+        )
 
         self.logger.info("initializing the FastEmbedEmbedding")
-        self.embed_model = FastEmbedEmbedding(model_name="BAAI/bge-small-en")
+        self.embed_model = FastEmbedEmbedding(
+            # model_name="BAAI/bge-small-en"
+        )
 
         llm = LlamaCPP(
             model_url=self.model_url,
@@ -124,7 +136,7 @@ class ChatPDF:
         retriever = VectorIndexRetriever(
             index=index,
             similarity_top_k=6,
-            vector_store_query_mode="hybrid"
+            # vector_store_query_mode="hybrid"
         )
 
         self.logger.info("configure response synthesizer")

start_service.sh

@@ -1,16 +1,16 @@
 #!/bin/sh
 
-# Start Ollama in the background
-ollama serve &
+# # Start Ollama in the background
+# ollama serve &
 
-# Wait for Ollama to start
-sleep 5
+# # Wait for Ollama to start
+# sleep 5
 
-#
-ollama pull mxbai-embed-large
+# #
+# ollama pull mxbai-embed-large
 
-# Pull and run <YOUR_MODEL_NAME>
-ollama pull qwen:1.8b
+# # Pull and run <YOUR_MODEL_NAME>
+# ollama pull qwen:1.8b
 
 #
 fastapi run /code/app/main.py --port 7860