deploy

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"]

app.py

@@ -0,0 +1,196 @@
+import os
+import chainlit as cl
+from dotenv import load_dotenv
+from operator import itemgetter
+from langchain_huggingface import HuggingFaceEndpoint
+from langchain_community.document_loaders import TextLoader
+from langchain_text_splitters import RecursiveCharacterTextSplitter
+from langchain_community.vectorstores import FAISS
+from langchain_huggingface import HuggingFaceEndpointEmbeddings
+from langchain_core.prompts import PromptTemplate
+from langchain.schema.output_parser import StrOutputParser
+from langchain.schema.runnable import RunnablePassthrough
+from langchain.schema.runnable.config import RunnableConfig
+from tqdm.asyncio import tqdm_asyncio
+import asyncio
+from tqdm.asyncio import tqdm
+
+# GLOBAL SCOPE - ENTIRE APPLICATION HAS ACCESS TO VALUES SET IN THIS SCOPE #
+# ---- ENV VARIABLES ---- # 
+"""
+This function will load our environment file (.env) if it is present.
+
+NOTE: Make sure that .env is in your .gitignore file - it is by default, but please ensure it remains there.
+"""
+load_dotenv()
+
+"""
+We will load our environment variables here.
+"""
+HF_LLM_ENDPOINT = os.environ["HF_LLM_ENDPOINT"]
+HF_EMBED_ENDPOINT = os.environ["HF_EMBED_ENDPOINT"]
+HF_TOKEN = os.environ["HF_TOKEN"]
+
+# ---- GLOBAL DECLARATIONS ---- #
+
+# -- RETRIEVAL -- #
+"""
+1. Load Documents from Text File
+2. Split Documents into Chunks
+3. Load HuggingFace Embeddings (remember to use the URL we set above)
+4. Index Files if they do not exist, otherwise load the vectorstore
+"""
+### 1. CREATE TEXT LOADER AND LOAD DOCUMENTS
+### NOTE: PAY ATTENTION TO THE PATH THEY ARE IN. 
+text_loader = TextLoader("data/paul_graham_essays.txt")
+documents = text_loader.load()
+
+### 2. CREATE TEXT SPLITTER AND SPLIT DOCUMENTS
+text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=30)
+split_documents = text_splitter.split_documents(documents)
+
+### 3. LOAD HUGGINGFACE EMBEDDINGS
+hf_embeddings = HuggingFaceEndpointEmbeddings(
+    model=HF_EMBED_ENDPOINT,
+    task="feature-extraction",
+    huggingfacehub_api_token=os.environ["HF_TOKEN"],
+)
+
+async def add_documents_async(vectorstore, documents):
+    await vectorstore.aadd_documents(documents)
+
+async def process_batch(vectorstore, batch, is_first_batch, pbar):
+    if is_first_batch:
+        result = await FAISS.afrom_documents(batch, hf_embeddings)
+    else:
+        await add_documents_async(vectorstore, batch)
+        result = vectorstore
+    pbar.update(len(batch))
+    return result
+
+async def main():
+    print("Indexing Files")
+    
+    vectorstore = None
+    batch_size = 32
+    
+    batches = [split_documents[i:i+batch_size] for i in range(0, len(split_documents), batch_size)]
+    
+    async def process_all_batches():
+        nonlocal vectorstore
+        tasks = []
+        pbars = []
+        
+        for i, batch in enumerate(batches):
+            pbar = tqdm(total=len(batch), desc=f"Batch {i+1}/{len(batches)}", position=i)
+            pbars.append(pbar)
+            
+            if i == 0:
+                vectorstore = await process_batch(None, batch, True, pbar)
+            else:
+                tasks.append(process_batch(vectorstore, batch, False, pbar))
+        
+        if tasks:
+            await asyncio.gather(*tasks)
+        
+        for pbar in pbars:
+            pbar.close()
+    
+    await process_all_batches()
+    
+    hf_retriever = vectorstore.as_retriever()
+    print("\nIndexing complete. Vectorstore is ready for use.")
+    return hf_retriever
+
+async def run():
+    retriever = await main()
+    return retriever
+
+hf_retriever = asyncio.run(run())
+
+# -- AUGMENTED -- #
+"""
+1. Define a String Template
+2. Create a Prompt Template from the String Template
+"""
+### 1. DEFINE STRING TEMPLATE
+RAG_PROMPT_TEMPLATE = """\
+<|start_header_id|>system<|end_header_id|>
+You are a helpful assistant. You answer user questions based on provided context. If you can't answer the question with the provided context, say you don't know.<|eot_id|>
+
+<|start_header_id|>user<|end_header_id|>
+User Query:
+{query}
+
+Context:
+{context}<|eot_id|>
+
+<|start_header_id|>assistant<|end_header_id|>
+"""
+
+### 2. CREATE PROMPT TEMPLATE
+rag_prompt = PromptTemplate.from_template(RAG_PROMPT_TEMPLATE)
+
+# -- GENERATION -- #
+"""
+1. Create a HuggingFaceEndpoint for the LLM
+"""
+### 1. CREATE HUGGINGFACE ENDPOINT FOR LLM
+hf_llm = HuggingFaceEndpoint(
+    endpoint_url=f"{HF_LLM_ENDPOINT}",
+    max_new_tokens=512,
+    top_k=10,
+    top_p=0.95,
+    typical_p=0.95,
+    temperature=0.01,
+    repetition_penalty=1.03,
+    huggingfacehub_api_token=os.environ["HF_TOKEN"]
+)
+
+@cl.author_rename
+def rename(original_author: str):
+    """
+    This function can be used to rename the 'author' of a message. 
+
+    In this case, we're overriding the 'Assistant' author to be 'Paul Graham Essay Bot'.
+    """
+    rename_dict = {
+        "Assistant" : "Paul Graham Essay Bot"
+    }
+    return rename_dict.get(original_author, original_author)
+
+@cl.on_chat_start
+async def start_chat():
+    """
+    This function will be called at the start of every user session. 
+
+    We will build our LCEL RAG chain here, and store it in the user session. 
+
+    The user session is a dictionary that is unique to each user session, and is stored in the memory of the server.
+    """
+
+    ### BUILD LCEL RAG CHAIN THAT ONLY RETURNS TEXT
+    lcel_rag_chain = {"context": itemgetter("query") | hf_retriever, "query": itemgetter("query")}| rag_prompt | hf_llm
+
+    cl.user_session.set("lcel_rag_chain", lcel_rag_chain)
+
+@cl.on_message  
+async def main(message: cl.Message):
+    """
+    This function will be called every time a message is recieved from a session.
+
+    We will use the LCEL RAG chain to generate a response to the user query.
+
+    The LCEL RAG chain is stored in the user session, and is unique to each user session - this is why we can access it here.
+    """
+    lcel_rag_chain = cl.user_session.get("lcel_rag_chain")
+
+    msg = cl.Message(content="")
+
+    async for chunk in lcel_rag_chain.astream(
+        {"query": message.content},
+        config=RunnableConfig(callbacks=[cl.LangchainCallbackHandler()]),
+    ):
+        await msg.stream_token(chunk)
+
+    await msg.send()

chainlit.md

@@ -0,0 +1 @@
+Ask me anything about Paul Graham essays!

requirements.txt

@@ -0,0 +1,132 @@
+aiofiles==23.2.1
+aiohappyeyeballs==2.4.3
+aiohttp==3.10.8
+aiosignal==1.3.1
+annotated-types==0.7.0
+anyio==3.7.1
+async-timeout==4.0.3
+asyncer==0.0.2
+attrs==24.2.0
+bidict==0.23.1
+certifi==2024.8.30
+chainlit==0.7.700
+charset-normalizer==3.3.2
+click==8.1.7
+dataclasses-json==0.5.14
+Deprecated==1.2.14
+distro==1.9.0
+exceptiongroup==1.2.2
+faiss-cpu==1.8.0.post1
+fastapi==0.100.1
+fastapi-socketio==0.0.10
+filelock==3.16.1
+filetype==1.2.0
+frozenlist==1.4.1
+fsspec==2024.9.0
+googleapis-common-protos==1.65.0
+greenlet==3.1.1
+grpcio==1.66.2
+grpcio-tools==1.62.3
+h11==0.14.0
+h2==4.1.0
+hpack==4.0.0
+httpcore==0.17.3
+httpx==0.24.1
+huggingface-hub==0.25.1
+hyperframe==6.0.1
+idna==3.10
+importlib_metadata==8.4.0
+Jinja2==3.1.4
+jiter==0.5.0
+joblib==1.4.2
+jsonpatch==1.33
+jsonpointer==3.0.0
+langchain==0.3.0
+langchain-community==0.3.0
+langchain-core==0.3.1
+langchain-huggingface==0.1.0
+langchain-openai==0.2.0
+langchain-qdrant==0.1.4
+langchain-text-splitters==0.3.0
+langsmith==0.1.121
+Lazify==0.4.0
+MarkupSafe==2.1.5
+marshmallow==3.22.0
+mpmath==1.3.0
+multidict==6.1.0
+mypy-extensions==1.0.0
+nest-asyncio==1.6.0
+networkx==3.2.1
+numpy==1.26.4
+nvidia-cublas-cu12==12.1.3.1
+nvidia-cuda-cupti-cu12==12.1.105
+nvidia-cuda-nvrtc-cu12==12.1.105
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-cudnn-cu12==9.1.0.70
+nvidia-cufft-cu12==11.0.2.54
+nvidia-curand-cu12==10.3.2.106
+nvidia-cusolver-cu12==11.4.5.107
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nccl-cu12==2.20.5
+nvidia-nvjitlink-cu12==12.6.77
+nvidia-nvtx-cu12==12.1.105
+openai==1.51.0
+opentelemetry-api==1.27.0
+opentelemetry-exporter-otlp==1.27.0
+opentelemetry-exporter-otlp-proto-common==1.27.0
+opentelemetry-exporter-otlp-proto-grpc==1.27.0
+opentelemetry-exporter-otlp-proto-http==1.27.0
+opentelemetry-instrumentation==0.48b0
+opentelemetry-proto==1.27.0
+opentelemetry-sdk==1.27.0
+opentelemetry-semantic-conventions==0.48b0
+orjson==3.10.7
+packaging==23.2
+pillow==10.4.0
+portalocker==2.10.1
+protobuf==4.25.5
+pydantic==2.9.2
+pydantic-settings==2.5.2
+pydantic_core==2.23.4
+PyJWT==2.9.0
+PyMuPDF==1.24.10
+PyMuPDFb==1.24.10
+python-dotenv==1.0.1
+python-engineio==4.9.1
+python-graphql-client==0.4.3
+python-multipart==0.0.6
+python-socketio==5.11.4
+PyYAML==6.0.2
+qdrant-client==1.11.2
+regex==2024.9.11
+requests==2.32.3
+safetensors==0.4.5
+scikit-learn==1.5.2
+scipy==1.13.1
+sentence-transformers==3.1.1
+simple-websocket==1.0.0
+sniffio==1.3.1
+SQLAlchemy==2.0.35
+starlette==0.27.0
+sympy==1.13.3
+syncer==2.0.3
+tenacity==8.5.0
+threadpoolctl==3.5.0
+tiktoken==0.7.0
+tokenizers==0.20.0
+tomli==2.0.1
+# torch==2.4.1
+tqdm==4.66.5
+transformers==4.45.1
+triton==3.0.0
+typing-inspect==0.9.0
+typing_extensions==4.12.2
+uptrace==1.26.0
+urllib3==2.2.3
+uvicorn==0.23.2
+watchfiles==0.20.0
+websockets==13.1
+wrapt==1.16.0
+wsproto==1.2.0
+yarl==1.13.1
+zipp==3.20.2