update part 3 to use llama index v0.10.0

part_3_rag/rag.py

@@ -3,20 +3,19 @@
 import sys
 import textwrap
 from pathlib import Path
-
-from llama_index import (
-    Response,
-    set_global_tokenizer,
-    ServiceContext,
-    VectorStoreIndex,
+from llama_index.core import (
     SimpleDirectoryReader,
+    VectorStoreIndex,
     StorageContext,
     load_index_from_storage,
+    set_global_tokenizer,
     set_global_handler,
 )
-from llama_index.core.llms.types import ChatMessage, MessageRole, ChatResponse
-from llama_index.embeddings import HuggingFaceEmbedding
-from llama_index.llms import LlamaCPP
+from llama_index.core.base.llms.types import ChatResponse, ChatMessage, MessageRole
+from llama_index.core.base.response.schema import Response
+from llama_index.embeddings.huggingface import HuggingFaceEmbedding
+from llama_index.llms.llama_cpp import LlamaCPP
+
 from transformers import AutoTokenizer
 
 from shared.settings import DATA_DIR
@@ -44,13 +43,13 @@ def load_embedding_model() -> HuggingFaceEmbedding:
 
 
 def save_or_load_index(
-    index_dir: Path, service_context: ServiceContext
+    index_dir: Path, embed_model: HuggingFaceEmbedding
 ) -> VectorStoreIndex:
     index_exists = any(item for item in index_dir.iterdir() if item.name != ".gitkeep")
     if index_exists:
         storage_context = StorageContext.from_defaults(persist_dir=index_dir)
         return load_index_from_storage(
-            storage_context=storage_context, service_context=service_context
+            storage_context=storage_context, embed_model=embed_model
         )
 
     transcript_files = glob.glob(str(DATA_DIR / "**/*transcript*"), recursive=True)
@@ -64,7 +63,7 @@ def save_or_load_index(
     ).load_data()
 
     index = VectorStoreIndex.from_documents(
-        documents, service_context=service_context, show_progress=True
+        documents, embed_model=embed_model, show_progress=True
     )
     # persist the index
     index.storage_context.persist(persist_dir=index_dir)
@@ -84,14 +83,10 @@ def run_inference(
         AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.2")
     )
 
-    service_context = ServiceContext.from_defaults(
-        llm=llm, embed_model=embedding_model, system_prompt=SYSTEM_PROMPT_TEXT
-    )
-
     index_dir = DATA_DIR / "indices"
-    index = save_or_load_index(index_dir=index_dir, service_context=service_context)
+    index = save_or_load_index(index_dir=index_dir, embed_model=embedding_model)
 
-    query_engine = index.as_query_engine()
+    query_engine = index.as_query_engine(llm=llm)
     return query_engine.query(messages[1].content)
 
 

requirements.txt

@@ -1,5 +1,7 @@
 llama-cpp-python==0.2.38
-llama-index==0.9.39
+llama-index>=0.10.0,<0.11.0
+llama-index-embeddings-huggingface>=0.1.4,<0.2.0
+llama-index-llms-llama-cpp>=0.1.3,<0.2.0
 transformers==4.37.1
 deepeval==0.20.55
 pytest==8.0.0