Added Grok models support

correctiveRag.py

@@ -17,6 +17,7 @@ from dotenv import load_dotenv
 load_dotenv(verbose=True)
 assert os.getenv("PINECONE_API_KEY") is not None
 assert os.getenv("HUGGINGFACEHUB_EMBEDDINGS_MODEL") is not None
+assert os.getenv("TAVILY_API_KEY") is not None
 
 
 logger = logging.getLogger(__name__)        # Child logger for this module
@@ -28,9 +29,22 @@ PINECONE_API_KEY = {os.getenv("PINECONE_API_KEY")[:5]}
 """)
 
 # Prepare the LLM
-from langchain_groq import ChatGroq
-assert os.getenv('GROQ_MODEL') is not None, "GROQ_MODEL not set"
-llm = ChatGroq(model_name=os.getenv('GROQ_MODEL'), temperature=0, verbose=True)
+# from langchain_groq import ChatGroq
+# assert os.getenv('GROQ_MODEL') is not None, "GROQ_MODEL not set"
+# assert os.getenv('GROQ_API_KEY') is not None, "GROQ_API_KEY not set"
+# llm = ChatGroq(model_name=os.getenv('GROQ_MODEL'), temperature=0, verbose=True)
+
+# For using Grok
+from langchain_openai import ChatOpenAI
+assert os.getenv('XAI_API_KEY') is not None, "XAI_API_KEY not set"
+assert os.getenv('XAI_MODEL') is not None, "XAI_MODEL not set"
+assert os.getenv('XAI_BASE_URL') is not None, "XAI_BASE_URL not set"
+llm = ChatOpenAI(
+    api_key=os.getenv("XAI_API_KEY"), 
+    base_url=os.getenv("XAI_BASE_URL"), 
+    model=os.getenv("XAI_MODEL"), 
+    temperature=0.1)
+
 
 # from langchain_openai import ChatOpenAI
 # assert os.getenv('OPENAI_MODEL_NAME') is not None, "GROQ_MODEL not set"
@@ -44,8 +58,9 @@ llm = ChatGroq(model_name=os.getenv('GROQ_MODEL'), temperature=0, verbose=True)
 from langchain_huggingface import HuggingFaceEmbeddings
 from langchain_pinecone import PineconeVectorStore
 index_name, namespace = 'courses', 'dsa'
+
 # Simple RAG
-# embeddings = HuggingFaceEmbeddings(model_name=EMBEDDINGS_MODEL)     # os.getenv("HUGGINGFACEHUB_EMBEDDINGS_MODEL")
+# embeddings = HuggingFaceEmbeddings(model_name=os.getenv("HUGGINGFACEHUB_EMBEDDINGS_MODEL"))
 # docsearch = PineconeVectorStore.from_existing_index(embedding=embeddings, index_name=index_name, namespace=namespace)
 # retriever = docsearch.as_retriever(search_type="mmr", search_kwargs={ 'k': 5 })