prompt engineering

.github/workflows/ci.yml

@@ -0,0 +1 @@
+# GitHub Actions CI workflow

.gitignore

@@ -3,4 +3,4 @@ todo.txt
 /airflow
 .env
 /src/.streamlit/secrets.toml
-/vectorstore/
+/vectorstore

app/Dockerfile.api

@@ -0,0 +1,5 @@
+FROM python:3.10
+WORKDIR /app
+COPY . /app
+RUN pip install --no-cache-dir -r requirements.txt
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8000"]

app/main.py

@@ -0,0 +1,25 @@
+from fastapi import FastAPI
+from langchain_community.vectorstores import FAISS
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain.chains import RetrievalQA
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+from langchain_community.llms import HuggingFacePipeline
+
+app = FastAPI()
+
+vectorstore = FAISS.load_local(
+    "./vectorstore/",
+    embeddings=HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+)
+
+tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")  # Or llama-3
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
+pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512)
+llm = HuggingFacePipeline(pipeline=pipe)
+
+retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
+rag_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever, chain_type="stuff")
+
+@app.get("/query/")
+def query_rag(question: str):
+    return {"response": rag_chain.run(question)}

app/rag.py

@@ -0,0 +1,56 @@
+# source:https://python.langchain.com/api_reference/langchain/chains/langchain.chains.retrieval_qa.base.RetrievalQA.html
+
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
+from langchain.prompts import ChatPromptTemplate
+from langchain.chains import create_retrieval_chain
+from langchain.chains.combine_documents import create_stuff_documents_chain
+from langchain_community.llms import Ollama
+from fastapi import FastAPI
+import requests
+from pydantic import BaseModel
+from langchain.chains import create_retrieval_chain
+from dotenv import load_dotenv
+import os
+
+
+load_dotenv()
+token= os.getenv("TOKEN")
+app = FastAPI()
+
+class QueryInput(BaseModel):
+    query: str
+
+
+# build the retrieval and augmented generator chain here
+
+embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+db = FAISS.load_local("./vectorstore/agriquery_faiss_index", embeddings, allow_dangerous_deserialization=True)
+llm = Ollama(model="llama3", base_url="http://localhost:11434")
+
+retriever = db.as_retriever()
+
+system_prompt = (
+    "You are an agriultural research assistant."
+    "Use the given context to answer the question."
+    "If you don't know the answer, say you don't know."
+    "Context: {context}"
+)
+
+prompt = ChatPromptTemplate.from_messages(
+    [
+        ("system", system_prompt),
+        ("human", "{input}"),
+    ]
+)
+
+question_answer_chain = create_stuff_documents_chain(llm,prompt)
+chain = create_retrieval_chain(retriever, question_answer_chain)
+
+
+@app.post("/query")
+async def query_handler(input: QueryInput):
+    result = chain.invoke({"input": input.query})
+    answer = result['answer'].replace("\\n", "\n").strip()
+
+    return {"answer": answer}

app/requirements.txt

@@ -0,0 +1,7 @@
+fastapi
+uvicorn
+transformers
+sentence-transformers
+faiss-cpu
+langchain
+langchain_community

app/streamlit_app.py

@@ -0,0 +1,73 @@
+import streamlit as st
+from langchain.chains import RetrievalQA
+from langchain_community.llms import HuggingFacePipeline
+from transformers import pipeline
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
+from langchain.prompts import ChatPromptTemplate
+from langchain.chains import create_retrieval_chain
+from langchain.chains.combine_documents import create_stuff_documents_chain
+from langchain_community.llms import Ollama
+import os
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
+
+
+# ----------------------
+system_prompt = (
+    "You are an agriultural research assistant."
+    "Use the given context to answer the question."
+    "If you don't know the answer, say you don't know."
+    "Context: {context}"
+)
+
+prompt = ChatPromptTemplate.from_messages(
+    [
+        ("system", system_prompt),
+        ("human", "{input}"),
+    ]
+)
+
+# Initialize embeddings & documents
+@st.cache_resource
+def load_retriever():
+    embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+    db = FAISS.load_local("./vectorstore/agriquery_faiss_index", embeddings, allow_dangerous_deserialization=True)
+    retriever = db.as_retriever()
+    return retriever
+
+# Load a lightweight model via HuggingFace pipeline
+@st.cache_resource
+def load_llm():
+    # pipe = pipeline("text-generation", model="google/flan-t5-small", max_new_tokens=256)
+    
+    # load the tokenizer and model on cpu/gpu
+    model_name = "meta-llama/Llama-2-7b-chat-hf"
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")
+    pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=256)
+    return HuggingFacePipeline(pipeline=pipe)
+
+# Setup RAG Chain
+@st.cache_resource
+def setup_qa():
+
+    retriever = load_retriever()
+    llm = load_llm()
+    question_answer_chain = create_stuff_documents_chain(llm,prompt)
+    chain = create_retrieval_chain(retriever, question_answer_chain)
+
+    # qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)
+    return chain
+
+
+# Streamlit App UI
+st.title("🌾 AgriQuery: RAG-Based Q&A Assistant")
+
+query = st.text_input("Ask a question related to agriculture:")
+
+if query:
+    qa = setup_qa()
+    with st.spinner("Thinking..."):
+        result = qa.invoke({"input": query})
+    st.success(result['answer'])

app/templates/index.html

@@ -0,0 +1,16 @@
+
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title>AgriQuery</title>
+</head>
+<body>
+    <h1>AgriQuery</h1>
+    <form action="/query" method="get">
+        <label for="q">Enter your question:</label><br>
+        <input type="text" id="q" name="q" required><br><br>
+        <input type="submit" value="Ask">
+    </form>
+</body>
+</html>

docker-compose.yaml

@@ -0,0 +1,188 @@
+x-airflow-common:
+  &airflow-common
+  build:
+    context: ./airflow
+    dockerfile: Dockerfile.airflow
+  environment:
+    &airflow-common-env
+    AIRFLOW__CORE__EXECUTOR: LocalExecutor
+    AIRFLOW__CORE__AUTH_MANAGER: airflow.providers.fab.auth_manager.fab_auth_manager.FabAuthManager 
+    AIRFLOW__DATABASE__SQL_ALCHEMY_CONN: postgresql+psycopg2://airflow:airflow@postgres/airflow 
+    AIRFLOW__CORE__FERNET_KEY: ${AIRFLOW__CORE__FERNET_KEY} # needed for multiuser or production
+    AIRFLOW__API_AUTH__JWT_SECRET: ${JWT_SECRET} 
+    AIRFLOW__WEBSERVER__SECRET_KEY: ${AIRFLOW__WEBSERVER__SECRET_KEY}
+    AIRFLOW__CORE__DAGS_ARE_PAUSED_AT_CREATION: 'true'
+    AIRFLOW__CORE__LOAD_EXAMPLES: 'false'
+    AIRFLOW__WEBSERVER__BASE_URL: http://localhost:8080
+    AIRFLOW__CORE__EXECUTION_API_SERVER_URL: 'http://airflow-apiserver:8080/execution/'
+    AIRFLOW__SCHEDULER__ENABLE_HEALTH_CHECK: 'true'
+    # _PIP_ADDITIONAL_REQUIREMENTS: ${_PIP_ADDITIONAL_REQUIREMENTS:-}
+    AIRFLOW_CONFIG: '/opt/airflow/config/airflow.cfg'
+    # MLFLOW_TRACKING_URI: 'http://mlflow:5000'
+
+  volumes:
+    - ${AIRFLOW_PROJ_DIR:-.}/airflow/dags:/opt/airflow/dags
+    - ${AIRFLOW_PROJ_DIR:-.}/airflow/logs:/opt/airflow/logs
+    - ${AIRFLOW_PROJ_DIR:-.}/airflow/config:/opt/airflow/config
+    - ${AIRFLOW_PROJ_DIR:-.}/airflow/plugins:/opt/airflow/plugins
+    - ${AIRFLOW_PROJ_DIR:-.}/data:/opt/airflow/data
+    - ${AIRFLOW_PROJ_DIR:-.}/scripts:/opt/airflow/scripts
+    - ${AIRFLOW_PROJ_DIR:-.}/model:/opt/airflow/model
+    - ${AIRFLOW_PROJ_DIR:-.}/app:/opt/airflow/app
+  user: "${AIRFLOW_UID:-50000}:0"
+  depends_on:
+    &airflow-common-depends-on
+    postgres:
+      condition: service_healthy
+
+services:
+  postgres:
+    image: postgres:13
+    environment:
+      POSTGRES_USER: ${POSTGRES_USER} # using nginx would provide an added measure of security wnen in production
+      POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
+      POSTGRES_DB: airflow
+    volumes:
+      - postgres-db-volume:/var/lib/postgresql/data
+    healthcheck:
+      test: ["CMD", "pg_isready", "-U", "airflow"]
+      interval: 10s
+      retries: 5
+      start_period: 5s
+    restart: always
+
+  airflow-apiserver:
+    <<: *airflow-common
+    command: api-server
+    ports:
+      - "8080:8080"
+    healthcheck:
+      test: ["CMD", "curl", "--fail", "http://localhost:8080/api/v2/version"]
+      interval: 30s
+      timeout: 10s
+      retries: 5
+      start_period: 30s
+    restart: always
+    depends_on:
+      <<: *airflow-common-depends-on
+      airflow-init:
+        condition: service_completed_successfully
+
+  airflow-scheduler:
+    <<: *airflow-common
+    command: scheduler
+    healthcheck:
+      test: ["CMD", "curl", "--fail", "http://localhost:8974/health"]
+      interval: 30s
+      timeout: 60s
+      retries: 5
+      start_period: 30s
+    restart: always
+    depends_on:
+      airflow-init:
+        condition: service_completed_successfully
+
+  airflow-dag-processor:
+    <<: *airflow-common
+    command: dag-processor
+    healthcheck:
+      test: ["CMD-SHELL", 'airflow jobs check --job-type DagProcessorJob --hostname "$${HOSTNAME}"']
+      interval: 30s
+      timeout: 10s
+      retries: 5
+      start_period: 30s
+    restart: always
+    depends_on:
+      airflow-init:
+        condition: service_completed_successfully
+
+  airflow-triggerer:
+    <<: *airflow-common
+    command: triggerer
+    healthcheck:
+      test: ["CMD-SHELL", 'airflow jobs check --job-type TriggererJob --hostname "$${HOSTNAME}"']
+      interval: 40s
+      timeout: 10s
+      retries: 5
+      start_period: 40s
+    restart: always
+    depends_on:
+      airflow-init:
+        condition: service_completed_successfully
+
+  airflow-init:
+    <<: *airflow-common
+    entrypoint: /bin/bash
+    command:
+      - -c
+      - |
+        [[ -z "${AIRFLOW_UID}" ]] && export AIRFLOW_UID=$(id -u)
+        mkdir -p /opt/airflow/{logs,dags,plugins,config}
+        /entrypoint airflow config list >/dev/null
+        chown -R "${AIRFLOW_UID}:0" /opt/airflow
+        ls -la /opt/airflow/{logs,dags,plugins,config}
+    environment:
+      <<: *airflow-common-env
+      _AIRFLOW_DB_MIGRATE: 'true'
+      _AIRFLOW_WWW_USER_CREATE: 'true'
+      _AIRFLOW_WWW_USER_USERNAME: ${AIRFLOW_WWW_USER_USERNAME}
+      _AIRFLOW_WWW_USER_PASSWORD: ${AIRFLOW_WWW_USER_PASSWORD}
+      # _PIP_ADDITIONAL_REQUIREMENTS: ${_PIP_ADDITIONAL_REQUIREMENTS}
+    user: "0:0"
+
+  airflow-cli:
+    <<: *airflow-common
+    profiles: [debug]
+    command:
+      - bash
+      - -c
+      - airflow
+    environment:
+      <<: *airflow-common-env
+      CONNECTION_CHECK_MAX_COUNT: "0"
+    depends_on:
+      postgres:
+        condition: service_healthy
+
+  rag-api:
+    build: ./app
+    ports:
+      - "8000:8000"
+    volumes:
+      - ./vectorstore:/app/vectorstore
+
+  # streamlit:
+  #   container_name: streamlit_app
+  #   build:
+  #     context: ./app
+  #     dockerfile: Dockerfile.streamlit
+  #   volumes:
+  #     - ./app:/app
+  #     - ./data:/data
+  #     - ./script:/script
+  #   working_dir: /app
+  #   ports:
+  #     - "127.0.0.1:8501:8501"
+
+  # mlflow:
+  #   build:
+  #     context: ./mlflow
+  #     dockerfile: Dockerfile.mlflow
+  #   ports:
+  #     - "127.0.0.1:5000:5000"
+
+  # Optional nginx reverse proxy for shared/internal environments
+  # nginx:
+  #   image: nginx:alpine
+  #   ports:
+  #     - "80:80"
+  #   volumes:
+  #     - ./nginx/default.conf:/etc/nginx/conf.d/default.conf:ro
+  #     - ./nginx/htpasswd:/etc/nginx/.htpasswd:ro
+  #   depends_on:
+  #     - airflow-apiserver
+  #     - mlflow
+  #     - streamlit
+
+volumes:
+  postgres-db-volume:

vectorstore/agriquery_faiss_index/index.faiss

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:4e423526ab5496cb63337855267b62e4218d9a140d4c8ed99c492f9af0a9aba3
-size 6743085

vectorstore/agriquery_faiss_index/index.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:28b483608516c5f1f4f4d101b4bb47194c5de0d19eb9853421dfc6e10374a7f9
-size 1527481