git pull origin branchName

!pip install -q -r requirements.txt 
 
import os 
import pandas as pd 
from google.colab import userdata 
 
os.environ["GROQ_API_KEY"] = userdata.get("GROQ_API_KEY") 
 
from smartcart_v1 import llm 
from langchain_core.messages import SystemMessage, HumanMessage 
 
def ask(prompt): 
    """Send one prompt and return the reply text.""" 
    return llm.invoke(prompt).content 

vague_prompt = "suggest earbuds" 
 
clear_prompt = ( 
    "Recommend exactly 3 wireless earbuds for gym use under Rs. 2,000 " 
    "for a college student. For each, give the name, the price in " 
    "rupees and one reason. Use a numbered list." ) 

vague_prompt = "suggest earbuds" 
 
clear_prompt = ( 
    "Recommend exactly 3 wireless earbuds for gym use under Rs. 2,000 " 
    "for a college student. For each, give the name, the price in " 
    "rupees and one reason. Use a numbered list." ) 
 
for name, prompt in [("VAGUE", vague_prompt), ("CLEAR", clear_prompt)]: 
    reply = ask(prompt) 
    print(f"=== {name} ({len(reply.split())} words) ===") 
    print(reply[:300], "\n")

def zero_shot_prompt(query): 
    return ("Classify this shopping query into a category.\n" 
            f"Query: {query}") 
 
EXAMPLES = """Query: wireless mouse for laptop -> Electronics 
Query: organic sugar nestle -> Grocery 
Query: kids dress grey size 8 -> Apparel""" 
 
def few_shot_prompt(query): 
    return ("Classify each shopping query as Electronics, Grocery or Apparel.\n\n" 
            f"{EXAMPLES}\n" 
            f"Query: {query} ->")

queries = pd.read_csv("smartcart_customer_queries.csv").sample(20, random_state=42) 
scores = {"zero_shot": 0, "few_shot": 0} 
for _, row in queries.iterrows(): 
    query = " ".join(row["query_text"].split()) 
    if ask(zero_shot_prompt(query)).strip() == row["category"]: 
        scores["zero_shot"] += 1 
    if ask(few_shot_prompt(query)).strip() == row["category"]: 
        scores["few_shot"] += 1 
 
print("Zero-shot accuracy:", scores["zero_shot"] / 20) 
print("Few-shot accuracy :", scores["few_shot"] / 20) 
print("\nExample zero-shot reply:", ask(zero_shot_prompt("KIDS DRESS GREY SIZE 8"))) 

question = "I want a smartphone for my father. Budget Rs. 15,000." 
 
roles = { 
    "No role":        "", 
    "Budget advisor": "You are a strict budget advisor who always looks for savings. ", 
    "SmartCart helper": "You are a warm, patient SmartCart assistant helping first-time 
    online shoppers. ", } 
 
for name, role in roles.items(): 
    reply = ask(role + question + " Answer in under 60 words.") 
    print(f"=== {name} ===\n{reply}\n")

SYSTEM_PROMPT = """You are Kart, SmartCart's friendly shopping assistant for shoppers in 
India. 
 
Persona: - Warm, patient and to the point. Prices are always in rupees. 
 
Rules: 
1. Only help with shopping. For anything else, politely say you can only help with 
shopping. 
2. Recommend exactly 3 products as a numbered list: name – approximate price – one reason. 
3. Reply in the same style as the shopper: English stays English, Hinglish stays Hinglish. 
4. Never make up SmartCart store policies (returns, delivery, refunds). 
   Say: "Please check the Help page in the SmartCart app for our latest policy." 
5. Keep every reply under 100 words."""

tests = [
    "who won the cricket match yesterday",
    "What is SmartCart's return policy for opened electronics?",
    "office ke liye formal shirt chahiye under 1500 rupees",]

for text in tests:
    messages = [SystemMessage(content=SYSTEM_PROMPT), HumanMessage(content=text)]
    print("You :", text)
    print("Kart:", llm.invoke(messages).content, "\n")

%%writefile smartcart_v1.py
import os
from langchain_groq import ChatGroq
from langchain_core.messages import SystemMessage, HumanMessage

if not os.environ.get("GROQ_API_KEY"):
    raise SystemExit("GROQ_API_KEY not found. Add it to Colab Secrets (Lab 9, Task 2).")

llm = ChatGroq(
    model="qwen/qwen3.8-27b",
    temperature=0.7,
    max_tokens=300,
    model_kwargs={"top_p": 0.95},
    reasoning_effort="none",)

SYSTEM_PROMPT = """You are Kart, SmartCart's friendly shopping assistant for shoppers in
India.

Persona: - Warm, patient and to the point. Prices are always in rupees.

Rules:
1. Only help with shopping. For anything else, politely say you can only help with
shopping.
2. Recommend exactly 3 products as a numbered list: name – approximate price – one reason.
3. Reply in the same style as the shopper: English stays English, Hinglish stays Hinglish.
4. Never make up SmartCart store policies (returns, delivery, refunds).
   Say: "Please check the Help page in the SmartCart app for our latest policy."
5. Keep every reply under 100 words."""

%%writefile smartcart_v1.py
import os
from langchain_groq import ChatGroq
from langchain_core.messages import SystemMessage, HumanMessage

if not os.environ.get("GROQ_API_KEY"):
    raise SystemExit("GROQ_API_KEY not found. Add it to Colab Secrets (Lab 9, Task 2).")

llm = ChatGroq(
    model="qwen/qwen3.8-27b",
    temperature=0.7,
    max_tokens=300,
    model_kwargs={"top_p": 0.95},
    reasoning_effort="none",)

SYSTEM_PROMPT = """You are Kart, SmartCart's friendly shopping assistant for shoppers in
India.

Persona: - Warm, patient and to the point. Prices are always in rupees.

Rules:
1. Only help with shopping. For anything else, politely say you can only help with
shopping.
2. Recommend exactly 3 products as a numbered list: name – approximate price – one reason.
3. Reply in the same style as the shopper: English stays English, Hinglish stays Hinglish.
4. Never make up SmartCart store policies (returns, delivery, refunds).
   Say: "Please check the Help page in the SmartCart app for our latest policy."
5. Keep every reply under 100 words."""


def ask_smartcart(shopper_request):
    """Send one shopping request with the system prompt; return the AIMessage."""
    messages = [
        SystemMessage(content=SYSTEM_PROMPT),
        HumanMessage(content=shopper_request),
    ]
    return llm.invoke(messages)

from smartcart_v1_chat import check_input, safe_ask 
 
tests = pd.read_csv("smartcart_v1_test_results.csv") 
 
new_replies = [] 
for text in tests["input"]: 
    cleaned, problem = check_input(text) 
    new_replies.append(problem if problem else safe_ask(cleaned)) 
 
tests["reply_v1_1"] = new_replies 
tests.to_csv("smartcart_v1_1_test_results.csv", index=False) 
print(tests[["id", "type", "reply_v1_1"]].to_string(max_colwidth=55)) 

git push origin branchName