VectorV17 commited on
Commit
3e15df8
·
verified ·
1 Parent(s): da4279b

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +199 -0
app.py ADDED
@@ -0,0 +1,199 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import datetime
3
+ from dotenv import load_dotenv
4
+ import gradio as gr
5
+
6
+ # LangChain & Chroma Vector DB Components
7
+ from langchain_chroma import Chroma
8
+ from langchain_huggingface import HuggingFaceEmbeddings, HuggingFaceEndpoint
9
+ from langchain_core.prompts import PromptTemplate
10
+ from langchain_core.runnables import RunnablePassthrough
11
+ from langchain_core.output_parsers import StrOutputParser
12
+
13
+ # Load Environment Variables (HUGGINGFACEHUB_API_TOKEN)
14
+ load_dotenv()
15
+
16
+ # -------------------------------------------------------------------------
17
+ # 1. Mock Data Setup (Simplilearn MOOC FAQs)
18
+ # -------------------------------------------------------------------------
19
+ # In a full app, you would load these from a JSON or CSV text file.
20
+ faq_data = [
21
+ {
22
+ "question": "What is the refund policy for Simplilearn courses?",
23
+ "answer": "Simplilearn offers a 7-day money-back guarantee for its online self-paced learning and blended learning programs. Raise a refund request within 7 days of purchase, provided you haven't accessed more than 25% of the content."
24
+ },
25
+ {
26
+ "question": "Can I access my Simplilearn courses on a smartphone?",
27
+ "answer": "Yes, Simplilearn has an official mobile app available on both Android and iOS platforms. You can download your video lessons to watch offline, complete quizzes, and track your learning progress on the go."
28
+ },
29
+ {
30
+ "question": "Do I get a certificate after completing a course?",
31
+ "answer": "Yes, upon successfully fulfilling the course completion criteria (which typically includes finishing 85% of the course material and clearing the simulation projects/exams), you will receive an industry-recognized digital certificate."
32
+ },
33
+ {
34
+ "question": "Are there any prerequisites for enrolling in the AI and ML Bootcamps?",
35
+ "answer": "While basic programming knowledge (preferably in Python) and a fundamental understanding of mathematics/statistics are helpful, Simplilearn's bootcamps include foundational electives to bring beginners up to speed."
36
+ }
37
+ ]
38
+
39
+ # Extract text chunks and metadata for the Vector Database
40
+ documents = [item["answer"] for item in faq_data]
41
+ metadatas = [{"question": item["question"]} for item in faq_data]
42
+ ids = [f"faq_{i}" for i in range(len(faq_data))]
43
+
44
+ # -------------------------------------------------------------------------
45
+ # 2. RAG Pipeline Initialization (CPU-Optimized)
46
+ # -------------------------------------------------------------------------
47
+ # Initialize a lightweight, highly accurate sentence encoder for local CPU execution
48
+ embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
49
+
50
+ # Create an in-memory Chroma DB Vector Store
51
+ vector_store = Chroma.from_texts(
52
+ texts=documents,
53
+ embedding=embeddings,
54
+ metadatas=metadatas,
55
+ ids=ids
56
+ )
57
+ retriever = vector_store.as_retriever(search_kwargs={"k": 2})
58
+
59
+ # Setup an Encoder-Decoder generative pipeline via Hugging Face Hosted Inference API
60
+ # Mistral-7B-Instruct handles complex reasoning over text efficiently without running locally
61
+ llm = HuggingFaceEndpoint(
62
+ repo_id="mistralai/Mistral-7B-Instruct-v0.3",
63
+ temperature=0.3,
64
+ max_new_tokens=256,
65
+ huggingfacehub_api_token=os.getenv("HUGGINGFACEHUB_API_TOKEN")
66
+ )
67
+
68
+ # Explicitly craft a strict contextual RAG prompt template
69
+ prompt_template = """You are Lane, an intelligent, helpful, and friendly Agentic AI Assistant for Simplilearn MOOC.
70
+ Use the following pieces of retrieved context to answer the user's question accurately.
71
+ If you do not know the answer based on the context, politely state that you do not know.
72
+
73
+ Context:
74
+ {context}
75
+
76
+ Question: {question}
77
+
78
+ Helpful Answer:"""
79
+
80
+ rag_prompt = PromptTemplate.from_template(prompt_template)
81
+
82
+ # Build the complete LangChain Expression Language (LCEL) chain
83
+ def format_docs(docs):
84
+ return "\n\n".join(doc.page_content for doc in docs)
85
+
86
+ rag_chain = (
87
+ {"context": retriever | format_docs, "question": RunnablePassthrough()}
88
+ | rag_prompt
89
+ | llm
90
+ | StrOutputParser()
91
+ )
92
+
93
+ # -------------------------------------------------------------------------
94
+ # 3. Application Core Logic
95
+ # -------------------------------------------------------------------------
96
+ def get_current_time_string():
97
+ """Generates the localized timestamp format highlighted in the UI reference."""
98
+ now = datetime.datetime.now()
99
+ return now.strftime("%dth %B %Y | Time: %I:%M%p").lower()
100
+
101
+ def chat_interface(user_message, history):
102
+ """Processes inbound user chat messages through the RAG framework."""
103
+ if not user_message.strip():
104
+ return "", history
105
+
106
+ # Process query through our tokenized encoder-decoder RAG architecture
107
+ ai_response = rag_chain.invoke(user_message)
108
+
109
+ # Update Gradio Chat history payload format
110
+ history.append({"role": "user", "content": user_message})
111
+ history.append({"role": "assistant", "content": ai_response})
112
+
113
+ return "", history
114
+
115
+ def handle_quick_reply(button_text, history):
116
+ """Bypasses standard empty chat entries to process pre-baked suggestions instantly."""
117
+ return chat_interface(button_text, history)
118
+
119
+ # -------------------------------------------------------------------------
120
+ # 4. Gradio Web Interface Layout
121
+ # -------------------------------------------------------------------------
122
+ # Injecting custom CSS rules to strictly align font colors, themes, and container styles to your visual specs
123
+ custom_css = """
124
+ .header-container { background-color: #f7f9fc; padding: 20px; border-radius: 8px; margin-bottom: 15px; }
125
+ .meta-container { text-align: right; font-family: 'Georgia', serif; font-style: italic; color: #3b5998; }
126
+ .sidebar-panel { background-color: #004d40; color: white; padding: 25px; border-radius: 8px; text-align: center; height: 100%; }
127
+ .sidebar-panel img { border-radius: 50%; max-width: 150px; margin: 0 auto 15px auto; }
128
+ .quick-btn { background-color: #e0f2f1 !important; border: 1px solid #b2dfdb !important; color: #004d40 !important; font-weight: 600; }
129
+ """
130
+
131
+ with gr.Blocks(css=custom_css, title="Simplilearn MOOC FAQ Agent") as demo:
132
+
133
+ # TOP SECTION: Header banner split into Info Branding and Dynamic Meta Controls
134
+ with gr.Row(elem_classes=["header-container"]):
135
+ with gr.Column(scale=3):
136
+ gr.Markdown("""
137
+ # About Us
138
+ At Simplilearn Agentic Hub, we are passionate about the power of structured education and its ability
139
+ to transform careers. We are a leading knowledge processing framework dedicated to helping
140
+ learners unlock the full potential of specialized MOOC educational pathways.
141
+ """)
142
+ with gr.Column(scale=2, elem_classes=["meta-container"]):
143
+ # UI display showcasing timestamping and system state parameters
144
+ time_display = gr.Markdown(value=f"### {get_current_time_string()}")
145
+ gr.Markdown("### ☀️ 🌙 ➔ Change Mode (Day / Night)")
146
+
147
+ # MAIN SECTION: Split layout dividing Interactive Workspace from Identity Card
148
+ with gr.Row():
149
+ # Chat Canvas Engine
150
+ with gr.Column(scale=4):
151
+ chatbot = gr.Chatbot(
152
+ label="Lane AI Conversation",
153
+ type="messages",
154
+ value=[{"role": "assistant", "content": "Hello! I'm Lane, your friendly AI Agent. How can I help you today?"}]
155
+ )
156
+
157
+ # Input Area
158
+ with gr.Row():
159
+ user_input = gr.Textbox(
160
+ show_label=False,
161
+ placeholder="Type your query regarding Simplilearn courses here...",
162
+ scale=5
163
+ )
164
+ submit_btn = gr.Button("Send", variant="primary", scale=1)
165
+
166
+ # Quick Reply Interaction Layer
167
+ with gr.Row():
168
+ btn_refund = gr.Button("Provide refund policy details", elem_classes=["quick-btn"])
169
+ btn_cert = gr.Button("Learn about course certification", elem_classes=["quick-btn"])
170
+
171
+ # Identity Profile Display Card
172
+ with gr.Column(scale=2, elem_classes=["sidebar-panel"]):
173
+ # Placeholder for Avatar matching the style from the mockup
174
+ gr.HTML("""
175
+ <div style='margin-top: 20px;'>
176
+ <img src='https://cdn-icons-png.flaticon.com/512/4712/4712010.png' alt='Lane Avatar' style='width:130px; height:130px; object-fit:cover; margin-bottom:15px; border: 4px solid #ffffff;'>
177
+ <h2 style='color: white; margin-bottom: 5px;'>Lane <span style='background-color:#ffffff; color:#004d40; padding:2px 6px; border-radius:4px; font-size:14px;'>AI</span></h2>
178
+ <p style='color: #e0f2f1; font-style: italic;'>Your Agentic AI Assistant</p>
179
+ <hr style='border-color: rgba(255,255,255,0.2); margin: 20px 0;'>
180
+ <p style='font-size: 14px; text-align: justify; color: #e0f2f1;'>
181
+ Lane runs an optimized RAG vector-space framework pipeline. It targets relevant local enterprise facts instantly without remote workflow friction.
182
+ </p>
183
+ </div>
184
+ """)
185
+
186
+ # -------------------------------------------------------------------------
187
+ # 5. Event Binding Pipeline Connections
188
+ # -------------------------------------------------------------------------
189
+ # Interactive workflow triggers for text submission
190
+ submit_btn.click(chat_interface, inputs=[user_input, chatbot], outputs=[user_input, chatbot])
191
+ user_input.submit(chat_interface, inputs=[user_input, chatbot], outputs=[user_input, chatbot])
192
+
193
+ # Event routing configuration mapping pre-configured quick buttons to context generation
194
+ btn_refund.click(handle_quick_reply, inputs=[btn_refund, chatbot], outputs=[user_input, chatbot])
195
+ btn_cert.click(handle_quick_reply, inputs=[btn_cert, chatbot], outputs=[user_input, chatbot])
196
+
197
+ # Run production runtime thread configuration
198
+ if __name__ == "__main__":
199
+ demo.launch(server_name="0.0.0.0", server_port=7860)