Revert to llama3 8b

rag.py

@@ -22,7 +22,8 @@ NEO4J_PASSWORD = os.getenv("NEO4J_PASSWORD")
 vector_index = os.getenv("VECTOR_INDEX")
 
 chat_llm = HuggingFaceEndpoint(
-    repo_id="HuggingFaceH4/zephyr-7b-beta",
+    # repo_id="HuggingFaceH4/zephyr-7b-beta",
+    repo_id="meta-llama/Meta-Llama-3-8B-Instruct",
     task="text-generation",
     max_new_tokens=4096,
     do_sample=False,
@@ -218,7 +219,7 @@ def global_retriever(query: str, level: int, response_type: str):
 
     where 1, 2, 3, 7, 34, 46, and 64 represent the id (not the index) of the relevant data record.
 
-    Do not include information where the supporting evidence for it is not provided. 
+    Do not include information where the supporting evidence for it is not provided. Style the response in markdown.
 
 
     ---Target response length and format---