Upload 6 files

Dockerfile

@@ -0,0 +1,27 @@
+# Base image with Python
+FROM python:3.10-slim
+
+# Install system dependencies
+RUN apt-get update && apt-get install -y \
+    git \
+    curl \
+    && rm -rf /var/lib/apt/lists/*
+
+# Set work directory
+WORKDIR /app
+
+# Copy application files
+COPY . /app
+
+# Install Python dependencies
+RUN pip install --no-cache-dir --upgrade pip
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Download SentenceTransformer model (caches on build)
+RUN python -c "from sentence_transformers import SentenceTransformer; SentenceTransformer('all-MiniLM-L6-v2')"
+
+# Expose the app on port 7860 for Hugging Face Spaces
+ENV PORT 7860
+
+# Start FastAPI with uvicorn
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

main.py

@@ -0,0 +1,103 @@
+# main.py
+
+from fastapi import FastAPI, Request, Form
+from fastapi.responses import HTMLResponse
+from fastapi.staticfiles import StaticFiles
+from fastapi.templating import Jinja2Templates
+from dotenv import load_dotenv
+
+# Load environment variables
+load_dotenv()
+from gliner import GLiNER
+from groq import Groq
+from vectordb_utils import search_vectordb, init_qdrant_collection
+GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
+app = FastAPI()
+templates = Jinja2Templates(directory="templates")
+app.mount("/static", StaticFiles(directory="static"), name="static")
+
+# Load models
+gliner_model = GLiNER.from_pretrained("urchade/gliner_medium-v2.1")
+groq_client = Groq(api_key=GROQ_API_KEY)
+
+init_qdrant_collection()
+
+def extract_entities(text):
+    labels = ["PRODUCT", "ISSUE", "PROBLEM", "SERVICE"]
+    return gliner_model.predict_entities(text, labels)
+
+def validate_answer(user_query, retrieved_answer):
+    prompt = f"""
+You are a validator assistant.
+
+Given the user query and the answer retrieved from a knowledge base, decide if the answer is relevant and correctly addresses the query.
+
+Respond ONLY with:
+- "YES" if the answer is appropriate.
+- "NO" if the answer is unrelated, inaccurate, or insufficient.
+
+User Query:
+{user_query}
+
+Retrieved Answer:
+{retrieved_answer}
+
+Is the answer appropriate?
+"""
+    completion = groq_client.chat.completions.create(
+        model="meta-llama/llama-4-scout-17b-16e-instruct",
+        messages=[{"role": "user", "content": prompt}],
+        temperature=0, max_completion_tokens=10, top_p=1
+    )
+    return completion.choices[0].message.content.strip()
+
+def generate_response(user_query, validated_answer):
+    prompt = f"""
+You are a customer support agent.
+
+Using the following validated support answer, respond helpfully and politely to the user's query.
+warning: there should not be [],<> tags in the response
+
+User Query:
+{user_query}
+
+Support Answer:
+{validated_answer}
+
+Compose your response:
+"""
+    completion = groq_client.chat.completions.create(
+        model="meta-llama/llama-4-scout-17b-16e-instruct",
+        messages=[{"role": "user", "content": prompt}],
+        temperature=0.8, max_completion_tokens=1000, top_p=1
+    )
+    return completion.choices[0].message.content.strip()
+
+@app.get("/", response_class=HTMLResponse)
+async def home(request: Request):
+    return templates.TemplateResponse("chat.html", {"request": request, "chat_history": []})
+
+@app.post("/chat", response_class=HTMLResponse)
+async def chat(request: Request, message: str = Form(...)):
+    entities = extract_entities(message)
+    entity_info = [(e['text'], e['label']) for e in entities]
+
+    results = search_vectordb(message)
+    if not results:
+        bot_reply = "Sorry, I couldn't find anything helpful."
+    else:
+        answer = results[0].payload["response"]
+        if validate_answer(message, answer) == "YES":
+            bot_reply = generate_response(message, answer)
+        else:
+            bot_reply = "Sorry, I couldn't find a suitable answer. Please contact support."
+
+    chat_history = [
+        {"sender": "User", "message": message},
+        {"sender": "Bot", "message": bot_reply}
+    ]
+    return templates.TemplateResponse("chat.html", {
+        "request": request,
+        "chat_history": chat_history,
+        "entities": entity_info
+    })

requirements.txt

@@ -0,0 +1,8 @@
+fastapi
+uvicorn
+jinja2
+gliner
+groq
+qdrant-client
+sentence-transformers
+dotenv

static/styles.css

@@ -0,0 +1,38 @@
+body {
+    font-family: Arial, sans-serif;
+    background-color: #f5f7fa;
+    padding: 20px;
+}
+.chat-container {
+    max-width: 600px;
+    margin: auto;
+    background: white;
+    padding: 20px;
+    border-radius: 8px;
+}
+.chat-box {
+    border: 1px solid #ccc;
+    padding: 10px;
+    max-height: 300px;
+    overflow-y: auto;
+    margin-bottom: 10px;
+}
+.user {
+    text-align: right;
+    color: #0066cc;
+}
+.bot {
+    text-align: left;
+    color: #333;
+}
+form {
+    display: flex;
+    gap: 10px;
+}
+input[type="text"] {
+    flex-grow: 1;
+    padding: 8px;
+}
+.entities {
+    margin-top: 20px;
+}

templates/chat.html

@@ -0,0 +1,35 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <title>Support Chatbot</title>
+    <link rel="stylesheet" href="/static/styles.css">
+</head>
+<body>
+    <div class="chat-container">
+        <h2>🧠 Customer Support Chatbot</h2>
+        <div class="chat-box">
+            {% for entry in chat_history %}
+                <div class="{{ 'user' if entry.sender == 'User' else 'bot' }}">
+                    <strong>{{ entry.sender }}:</strong> {{ entry.message }}
+                </div>
+            {% endfor %}
+        </div>
+
+        <form method="post" action="/chat">
+            <input type="text" name="message" placeholder="Type your question..." required>
+            <button type="submit">Send</button>
+        </form>
+
+        {% if entities %}
+        <div class="entities">
+            <h4>🔍 Detected Entities:</h4>
+            <ul>
+                {% for text, label in entities %}
+                    <li>{{ label }}: <em>{{ text }}</em></li>
+                {% endfor %}
+            </ul>
+        </div>
+        {% endif %}
+    </div>
+</body>
+</html>

vectordb_utils.py

@@ -0,0 +1,31 @@
+# vectordb_utils.py
+
+from qdrant_client import QdrantClient
+from qdrant_client.models import VectorParams, Distance, PointStruct
+from sentence_transformers import SentenceTransformer
+import uuid
+
+encoder = SentenceTransformer("all-MiniLM-L6-v2")
+qdrant = QdrantClient(":memory:")
+collection_name = "customer_support_docsv1"
+
+def init_qdrant_collection():
+    qdrant.recreate_collection(
+        collection_name=collection_name,
+        vectors_config=VectorParams(size=384, distance=Distance.COSINE)
+    )
+
+def add_to_vectordb(query, response):
+    vector = encoder.encode(query).tolist()
+    qdrant.upload_points(
+        collection_name=collection_name,
+        points=[PointStruct(
+            id=str(uuid.uuid4()),
+            vector=vector,
+            payload={"query": query, "response": response}
+        )]
+    )
+
+def search_vectordb(query, limit=3):
+    vector = encoder.encode(query).tolist()
+    return qdrant.search(collection_name=collection_name, query_vector=vector, limit=limit)