first commit

Dockerfile

@@ -0,0 +1,11 @@
+FROM python:3.9
+RUN useradd -m -u 1000 user
+USER user
+ENV HOME=/home/user \
+    PATH=/home/user/.local/bin:$PATH
+WORKDIR $HOME/app
+COPY --chown=user . $HOME/app
+COPY ./requirements.txt ~/app/requirements.txt
+RUN pip install -r requirements.txt
+COPY . .
+CMD ["chainlit", "run", "app.py", "--port", "7860"]

README.md

@@ -1,8 +1,8 @@
 ---
-title: Llmops Final Earnings
-emoji: 📚
-colorFrom: purple
-colorTo: green
+title: Barbie RAQA Application Chainlit Demo
+emoji: 🔥
+colorFrom: red
+colorTo: red
 sdk: docker
 pinned: false
 license: apache-2.0

app.py

@@ -0,0 +1,35 @@
+import sys
+
+sys.path.append(".")
+
+import chainlit as cl
+
+from earnings_app import extract_information
+
+@cl.author_rename
+def rename(orig_author: str):
+    diamond_char = u'\U0001F537'
+    phrase = diamond_char + " Diamond Hands " + diamond_char
+    rename_dict = {"RetrievalQA": phrase}
+    return rename_dict.get(orig_author, orig_author)
+
+@cl.on_chat_start
+async def start():
+    """
+    This is called when the Chainlit chat is started!
+    """
+    chain = await extract_information()
+    cl.user_session.set("chain", chain)
+    await cl.Message("Welcome to the information extraction chat!").send()
+
+@cl.on_message
+async def main(message: cl.Message):
+    """
+    This is called when a message is received!
+    """
+    chain = cl.user_session.get("chain")
+    res = await chain.achat(message.content)
+    # res = await chain.aiinvoke({"input": message})
+    # res = res["text"]
+    out = str(res)
+    await cl.Message(content=out).send()

chainlit.md

@@ -0,0 +1,11 @@
+# Assignment Part 2: Deploying Your Model to a Hugging Face Space
+
+Now that you've done the hard work of setting up the RetrievalQA chain and sourcing your documents - let's tie it together in a ChainLit application.
+
+### Duplicating the Space
+
+Since this is our first assignment, all you'll need to do is duplicate this space and add your own `OPENAI_API_KEY` as a secret in the space.
+
+### Conclusion
+
+Now that you've shipped an LLM-powered application, it's time to share! 🚀

earnings_app.py

@@ -0,0 +1,170 @@
+# -*- coding: utf-8 -*-
+# Imports
+import asyncio
+import os
+import openai
+
+from typing import List, Optional
+# from pydantic import BaseModel, Field
+
+# from langchain.prompts import ChatPromptTemplate
+# from langchain.pydantic_v1 import BaseModel
+# from langchain.utils.openai_functions import convert_pydantic_to_openai_function
+from llama_index.tools import FunctionTool
+from llama_index.vector_stores.types import (
+    VectorStoreInfo,
+    MetadataInfo,
+    ExactMatchFilter,
+    MetadataFilters,
+)
+from llama_index.agent import OpenAIAgent
+from llama_index.retrievers import VectorIndexRetriever
+from llama_index.query_engine import RetrieverQueryEngine
+
+from typing import List, Tuple, Any
+from pydantic import BaseModel, Field
+from llama_index import load_index_from_storage
+from llama_index import set_global_handler
+import llama_index
+from llama_index.embeddings import OpenAIEmbedding
+from llama_index import ServiceContext
+from llama_index.llms import OpenAI
+from llama_index import GPTVectorStoreIndex
+
+set_global_handler("wandb", run_args={"project": "final-project-v1"})
+wandb_callback = llama_index.global_handler
+
+from dotenv import load_dotenv
+load_dotenv()
+
+openai.api_key = os.environ['OPENAI_API_KEY']
+
+top_k = 3
+
+vector_store_info = VectorStoreInfo(
+    content_info="transcripts of earnings calls",
+    metadata_info=[MetadataInfo(
+        name="title",
+        type="str",
+        description="Title of the earnings call",
+    ),
+    MetadataInfo(
+        name="period",
+        type="str",
+        description="Period of the earnings call"
+    ),
+    MetadataInfo(
+        name="ticker",
+        type="str",
+        description="Ticker of the company"
+    ),
+    MetadataInfo(
+        name="year",
+        type="str",
+        description="Year of the earnings call"
+    ),
+    MetadataInfo(
+        name="quarter",
+        type="str",
+        description="Quarter of the earnings call"
+    ),
+    MetadataInfo(
+        name="path",
+        type="str",
+        description="Path to the earnings call"
+    ),
+])
+
+class AutoRetrieveModel(BaseModel):
+    query: str = Field(..., description="natural language query string")
+    filter_key_list: List[str] = Field(
+        ..., description="List of metadata filter field names"
+    )
+    filter_value_list: List[str] = Field(
+        ...,
+        description=(
+            "List of metadata filter field values (corresponding to names specified in filter_key_list)"
+        )
+    )
+
+embed_model = OpenAIEmbedding()
+chunk_size = 500
+
+llm = OpenAI(
+    temperature=0,
+    model="gpt-4" ### YOUR CODE HERE
+)
+
+service_context = ServiceContext.from_defaults(
+    llm=llm,
+    chunk_size=chunk_size,
+    embed_model=embed_model,
+)
+
+index = GPTVectorStoreIndex.from_documents([], service_context=service_context)
+
+
+# Main function to extract information
+async def extract_information():
+    # Make sure to use a recent model that supports tools
+
+    storage_context = wandb_callback.load_storage_context(
+        artifact_url="llmop/final-project-v1/earnings-index:v1"
+    )
+
+    index = load_index_from_storage(storage_context, service_context=service_context)
+
+    def auto_retrieve_fn(
+        query: str, filter_key_list: List[str], filter_value_list: List[str]
+    ):
+        """Auto retrieval function.
+
+        Performs auto-retrieval from a vector database, and then applies a set of filters.
+
+        """
+        query = query or "Query"
+
+        exact_match_filters = [
+            ExactMatchFilter(key=k, value=v)
+            for k, v in zip(filter_key_list, filter_value_list)
+        ]
+        retriever = VectorIndexRetriever(
+            index, filters=MetadataFilters(filters=exact_match_filters), top_k=top_k
+        )
+        query_engine = RetrieverQueryEngine.from_args(retriever, service_context=service_context)
+
+        response = query_engine.query(query)
+        return str(response)
+    
+    description = f"""
+        Who is the CEO of MSFT
+        The vector database schema is given below:
+        {vector_store_info.json()}
+    """
+    auto_retrieve_tool = FunctionTool.from_defaults(
+        fn=auto_retrieve_fn,
+        name="earnings-transcripts",
+        description=description,
+        fn_schema=AutoRetrieveModel
+    )
+
+    agent = OpenAIAgent.from_tools(
+        tools=[auto_retrieve_tool],
+    )
+
+    return agent
+
+
+# if __name__ == "__main__":
+#     text = "Who is the CEO of MSFT."
+#     chain = extract_information()
+#     print(str(chain.chat(text)))
+
+#     async def extract_information_async(message: str):
+#         return str(chain.chat(text))
+
+#     async def main():
+#         res = await extract_information_async(text)
+#         print(res)
+
+    # asyncio.run(main())

requirements.txt

@@ -0,0 +1,12 @@
+chainlit
+langchain
+tiktoken
+openai
+faiss-cpu
+llama-index==0.9.19
+llama-hub
+unstructured
+lxml
+cohere
+wandb
+pydantic