embedding model was changed from Gemini to one of the Huggingface models

README.md

@@ -27,7 +27,7 @@ This project implements a **Corrective Retrieval-Augmented Generation (CRAG)** s
 
 *   **Orchestration**: LangGraph
 *   **LLM**: DeepSeek (via `DeepSeek-V3` / `deepseek-chat`)
-*   **Embeddings**: Google Gemini (`models/text-embedding-004`)
+*   **Embeddings**: Hugging Face (`sentence-transformers/all-MiniLM-L6-v2`)
 *   **Vector Database**: ChromaDB
 *   **Web Search**: Tavily Search API
 *   **UI**: Streamlit
@@ -54,7 +54,7 @@ Verify the application directly on the web:
     Create a `.env` file in the root directory and add your API keys:
     ```env
     DEEPSEEK_API_KEY=your_deepseek_key
-    GOOGLE_API_KEY=your_google_key
+    HF_TOKEN=your_huggingface_token #for using embedding models from huggingface
     TAVILY_API_KEY=your_tavily_key
     LANGCHAIN_API_KEY=your_langchain_key # Optional: for LangSmith tracing
     LANGCHAIN_TRACING_V2=true           # Optional

ingestion.py

@@ -2,8 +2,7 @@ from dotenv import load_dotenv
 from langchain_text_splitters import  RecursiveCharacterTextSplitter
 from langchain_community.document_loaders import WebBaseLoader
 from langchain_community.vectorstores import Chroma
-from langchain_community.embeddings import OllamaEmbeddings
-from langchain_google_genai import GoogleGenerativeAIEmbeddings
+from langchain_huggingface import HuggingFaceEndpointEmbeddings
 
 load_dotenv()
 
@@ -24,7 +23,7 @@ text_splitter = RecursiveCharacterTextSplitter.from_tiktoken_encoder(
 splits = text_splitter.split_documents(docs_list)
 
 #embedding = OllamaEmbeddings(model="nomic-embed-text")
-embedding = GoogleGenerativeAIEmbeddings(model="models/text-embedding-004")
+embedding = HuggingFaceEndpointEmbeddings(model="sentence-transformers/all-MiniLM-L6-v2")
 
 vectorstore = Chroma.from_documents(
     documents=splits,

requirements.txt

@@ -10,5 +10,5 @@ pytest>=8.2.2
 langchain-deepseek
 pydantic
 langchain_text_splitters
-langchain-google-genai #for embedding
-streamlit
+streamlit
+langchain-huggingface #for embedding