implementation of llama-agents based RAG

bootstraprag/cli.py

@@ -31,7 +31,13 @@ def create(project_name, framework, template, observability):
         choices=framework_choices
     ).execute()
     if framework == 'llamaindex' or framework == 'langchain' or framework == 'haystack':
-        template_choices = ['simple-rag', 'rag-with-react', 'rag-with-hyde', 'rag-with-flare']
+        template_choices = [
+            'simple-rag',
+            'rag-with-react',
+            'rag-with-hyde',
+            'rag-with-flare',
+            'llama-agents-with-simpleq'
+        ]
     elif framework == 'None':
         framework = 'qdrant'
         template_choices = ['simple-search']

bootstraprag/templates/llamaindex/llama_agents_simpleq/agents_core.py → bootstraprag/templates/llamaindex/llama_agents_with_simpleq/agents_core.py

@@ -10,25 +10,29 @@
 from llama_index.core.tools import FunctionTool
 from llama_index.llms.ollama import Ollama
 from llama_index.embeddings.ollama import OllamaEmbedding
+from rag_operations import RAGOperations
+from dotenv import load_dotenv, find_dotenv
+import os
 
 
+_ = load_dotenv(find_dotenv())
 llm = Ollama(model=os.environ['OLLAMA_LLM_MODEL'], base_url=os.environ['OLLAMA_BASE_URL'], request_timeout=300)
 embed_model = OllamaEmbedding(model_name=os.environ['OLLAMA_EMBED_MODEL'], base_url=os.environ['OLLAMA_BASE_URL'])
 
 Settings.llm = llm
 Settings.embed_model = embed_model
 
+rag_ops = RAGOperations()
 
 # create an agent
-def get_the_secret_fact() -> str:
-    """Returns the secret fact."""
-    return "The secret fact is: A baby llama is called a 'Cria'."
+def retrieve_information(query: str) -> str:
+    """used to retrieve the information for user asked questions"""
+    return rag_ops.query_engine.query(query)
 
 class SimpleQAgent:
-    tool = FunctionTool.from_defaults(fn=get_the_secret_fact)
+    tool = FunctionTool.from_defaults(fn=retrieve_information)
 
     agent1 = ReActAgent.from_tools([tool], llm=Settings.llm)
-    agent2 = ReActAgent.from_tools([], llm=Settings.llm)
 
     # create our multi-agent framework components
     message_queue = SimpleMessageQueue(port=8000)
@@ -40,14 +44,7 @@ class SimpleQAgent:
     agent_server_1 = AgentService(
         agent=agent1,
         message_queue=message_queue,
-        description="Useful for getting the secret fact.",
-        service_name="secret_fact_agent",
+        description="Useful for getting the answer from user query.",
+        service_name="question_answer_agent",
         port=8002,
     )
-    agent_server_2 = AgentService(
-        agent=agent2,
-        message_queue=message_queue,
-        description="Useful for getting random dumb facts.",
-        service_name="dumb_fact_agent",
-        port=8003,
-    )

bootstraprag/templates/llamaindex/llama_agents_with_simpleq/main.py

@@ -0,0 +1,28 @@
+from llama_agents import LocalLauncher
+import nest_asyncio
+from agents_core import SimpleQAgent
+
+
+simple_agent = SimpleQAgent()
+
+# needed for running in a notebook
+nest_asyncio.apply()
+
+# launch it
+launcher = LocalLauncher(
+    [simple_agent.agent_server_1],
+    simple_agent.control_plane,
+    simple_agent.message_queue,
+)
+
+# Start a loop to continually get input from the user
+while True:
+    # Get a query from the user
+    user_query = input("Enter your query [type 'bye' to 'exit']: ")
+
+    # Check if the user wants to terminate the loop
+    if user_query.lower() == "bye" or user_query.lower() == "exit":
+        break
+
+    result = launcher.launch_single(user_query)
+    print(result)

bootstraprag/templates/llamaindex/llama_agents_with_simpleq/rag_operations.py

@@ -0,0 +1,64 @@
+from llama_index.core import SimpleDirectoryReader, Settings, StorageContext, VectorStoreIndex
+from llama_index.core.base.base_query_engine import BaseQueryEngine
+from llama_index.core.node_parser import SentenceSplitter
+from llama_index.vector_stores.qdrant import QdrantVectorStore
+from llama_index.embeddings.ollama import OllamaEmbedding
+# enable if you are using openai
+# from llama_index.embeddings.openai import OpenAIEmbedding
+from llama_index.llms.ollama import Ollama
+# enable if you are using openai
+# from llama_index.llms.openai import OpenAI
+from dotenv import load_dotenv, find_dotenv
+from typing import Union
+import qdrant_client
+import logging
+import os
+
+_ = load_dotenv(find_dotenv())
+logging.basicConfig(level=int(os.environ['INFO']))
+logger = logging.getLogger(__name__)
+
+
+class RAGOperations:
+
+    def __init__(self, data_path: str = './data', show_progress: bool = True, chunk_size: int = 512,
+                 chunk_overlap: int = 200):
+        self.documents = SimpleDirectoryReader(input_dir=data_path).load_data(show_progress=show_progress)
+        self.text_parser = SentenceSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
+        self.client = qdrant_client.QdrantClient(url=os.environ['DB_URL'], api_key=os.environ['DB_API_KEY'])
+        self.vector_store = QdrantVectorStore(client=self.client, collection_name=os.environ['COLLECTION_NAME'])
+
+        # use your prefered vector embeddings model
+        logger.info("initializing the OllamaEmbedding")
+        embed_model = OllamaEmbedding(model_name=os.environ['OLLAMA_EMBED_MODEL'],
+                                      base_url=os.environ['OLLAMA_BASE_URL'])
+        # openai embeddings, embedding_model_name="text-embedding-3-large"
+        # embed_model = OpenAIEmbedding(embed_batch_size=10, model=embedding_model_name)
+
+        # use your prefered llm
+        llm = Ollama(model=os.environ['OLLAMA_LLM_MODEL'], base_url=os.environ['OLLAMA_BASE_URL'], request_timeout=600)
+        # llm = OpenAI(model="gpt-4o")
+
+        logger.info("initializing the global settings")
+        Settings.embed_model = embed_model
+        Settings.llm = llm
+
+        Settings.transformations = [self.text_parser]
+        self.query_engine: BaseQueryEngine = None
+
+        self._index_and_create_query_engine()
+
+    def _index_and_create_query_engine(self):
+        logger.info("initializing the storage context")
+        storage_context = StorageContext.from_defaults(vector_store=self.vector_store)
+        logger.info("indexing the nodes in VectorStoreIndex")
+        if not self.client.collection_exists(collection_name=os.environ['COLLECTION_NAME']):
+            vector_index = VectorStoreIndex.from_documents(
+                documents=self.documents,
+                storage_context=storage_context,
+                transformations=Settings.transformations,
+            )
+        else:
+            vector_index = VectorStoreIndex.from_vector_store(vector_store=self.vector_store)
+
+        self.query_engine = vector_index.as_query_engine()

bootstraprag/templates/llamaindex/llama_agents_with_simpleq/readme.md

@@ -0,0 +1,17 @@
+## llama-agents
+
+### install qdrant
+- `docker pull qdrant/qdrant`
+- `docker run -p 6333:6333 -p 6334:6334 \
+  -v $(pwd)/qdrant_storage:/qdrant/storage:z \
+  qdrant/qdrant`
+
+### install ollama
+- navigate to [https://ollama.com/download](https://ollama.com/download)
+
+### how to run llama-agents
+- open `.env`
+- change the `DB_URL`, `DB_API_KEY` and `COLLECTION_NAME` according to you
+- point the right right LLMs (if not local)
+- `pip install -r requirements.txt`
+- `python main.py`

bootstraprag/templates/llamaindex/llama_agents_simpleq/requirements.txt → bootstraprag/templates/llamaindex/llama_agents_with_simpleq/requirements.txt

@@ -6,5 +6,5 @@ llama-index-embeddings-openai==0.1.11
 llama-index-embeddings-ollama==0.2.0
 llama-index-vector-stores-qdrant==0.2.16
 qdrant-client==1.11.0
-fastapi==0.112.1
+fastapi==0.109.2
 uvicorn==0.30.6

pyproject.toml

@@ -0,0 +1,3 @@
+[build-system]
+requires = ["setuptools>=64", "wheel"]
+build-backend = "setuptools.build_meta"