Initial commit: Customer Support Chatbot with DialoGPT-medium

car_service_data.jsonl

@@ -0,0 +1,102 @@
+{"input": "I need an oil change", "response": "Sure, when would you like to schedule your oil change?"}
+{"input": "Book a tire rotation for Friday", "response": "I can book your tire rotation for Friday. What time works for you?"}
+{"input": "Can you check my brakes?", "response": "Yes, we can check your brakes. When would you like to come in?"}
+{"input": "I want a car wash tomorrow", "response": "Great, I can schedule a car wash for tomorrow. What time suits you?"}
+{"input": "How much for an oil change?", "response": "An oil change costs $40-$60, depending on your car. Want to book one?"}
+{"input": "Schedule a tune-up next week", "response": "I’ll book a tune-up for next week. Which day and time work best?"}
+{"input": "My car needs new tires", "response": "We can replace your tires. When would you like to bring your car in?"}
+{"input": "Can I get a service today?", "response": "Let me check availability. What service do you need today?"}
+{"input": "I need a battery replacement", "response": "We can replace your battery. When would you like to schedule it?"}
+{"input": "What’s a full service?", "response": "A full service includes oil change, tire check, and more. Want to book one?"}
+{"input": "Fix my AC on Monday", "response": "I’ll schedule an AC repair for Monday. What time works?"}
+{"input": "How much for new brakes?", "response": "Brake replacement is $100-$200. Want to book it?"}
+{"input": "I need an alignment", "response": "We can do an alignment. When would you like to come in?"}
+{"input": "Book a service for Saturday", "response": "What service would you like to book for Saturday?"}
+{"input": "Check my engine light", "response": "We can check your engine light. When are you available?"}
+{"input": "How long for an oil change?", "response": "An oil change takes about 30 minutes. Want to schedule one?"}
+{"input": "I need new wipers", "response": "We can replace your wipers. When would you like to stop by?"}
+{"input": "Can you do a detailing?", "response": "Yes, we offer detailing. When would you like to schedule it?"}
+{"input": "Book a brake job for next Tuesday", "response": "I’ll book your brake job for next Tuesday. What time?"}
+{"input": "What’s included in a tune-up?", "response": "A tune-up includes spark plugs, filters, and more. Want to book one?"}
+{"input": "Replace my spark plugs", "response": "We can replace your spark plugs. When would you like to come in?"}
+{"input": "Book a transmission check for Wednesday", "response": "I’ll book a transmission check for Wednesday. What time?"}
+{"input": "I need a headlight fixed", "response": "We can fix your headlight. When would you like to schedule it?"}
+{"input": "How much for a tire rotation?", "response": "A tire rotation costs $20-$30. Want to book one?"}
+{"input": "Schedule an inspection for this weekend", "response": "I’ll book an inspection for this weekend. Saturday or Sunday?"}
+{"input": "My car’s overheating", "response": "We can check that for you. When can you bring it in?"}
+{"input": "Can you top up my fluids?", "response": "Yes, we can top up your fluids. When would you like to come by?"}
+{"input": "Book an oil change for next Thursday", "response": "I’ll book your oil change for next Thursday. What time?"}
+{"input": "How long for a brake job?", "response": "A brake job takes about 1-2 hours. Want to schedule one?"}
+{"input": "I need a wheel balancing", "response": "We can balance your wheels. When would you like to schedule it?"}
+{"input": "Fix my muffler next Friday", "response": "I’ll schedule a muffler repair for next Friday. What time?"}
+{"input": "How much for a battery replacement?", "response": "A battery replacement costs $80-$120. Want to book it?"}
+{"input": "I need a belt replaced", "response": "We can replace your belt. When would you like to come in?"}
+{"input": "Book a service for Monday morning", "response": "What service would you like to book for Monday morning?"}
+{"input": "Check my tire pressure", "response": "We can check your tire pressure. When are you available?"}
+{"input": "How long for a car wash?", "response": "A car wash takes about 20 minutes. Want to schedule one?"}
+{"input": "I need a new air filter", "response": "We can replace your air filter. When would you like to stop by?"}
+{"input": "Can you do an oil change this afternoon?", "response": "Let me check availability for this afternoon. Any specific time?"}
+{"input": "Book a tire change for next Wednesday", "response": "I’ll book your tire change for next Wednesday. What time?"}
+{"input": "What’s included in an inspection?", "response": "An inspection covers brakes, tires, and more. Want to book one?"}
+{"input": "I need my brakes fixed", "response": "We can fix your brakes. When would you like to schedule it?"}
+{"input": "Book an AC check for tomorrow", "response": "I’ll book an AC check for tomorrow. What time works for you?"}
+{"input": "How much for a wheel alignment?", "response": "A wheel alignment costs $50-$80. Want to book one?"}
+{"input": "Schedule a detailing for Saturday", "response": "I’ll book a detailing for Saturday. What time suits you?"}
+{"input": "My engine’s making a noise", "response": "We can check your engine. When can you bring it in?"}
+{"input": "Can you replace my headlights?", "response": "Yes, we can replace your headlights. When would you like to come by?"}
+{"input": "Book a tune-up for next Monday", "response": "I’ll book your tune-up for next Monday. What time?"}
+{"input": "How long for a tire rotation?", "response": "A tire rotation takes about 30 minutes. Want to schedule one?"}
+{"input": "I need a new battery tomorrow", "response": "We can replace your battery tomorrow. What time works?"}
+{"input": "Can you fix my wipers today?", "response": "Let me check availability for today. Any preferred time?"}
+{"input": "Book an oil change for this Friday", "response": "I’ll book your oil change for this Friday. What time?"}
+{"input": "What does a brake job cost?", "response": "A brake job costs $100-$200. Want to schedule one?"}
+{"input": "I need my tires rotated", "response": "We can rotate your tires. When would you like to come in?"}
+{"input": "Schedule a car wash for next Tuesday", "response": "I’ll book a car wash for next Tuesday. What time?"}
+{"input": "My AC isn’t cooling", "response": "We can check your AC. When can you bring it in?"}
+{"input": "Can you check my fluids?", "response": "Yes, we can check your fluids. When would you like to stop by?"}
+{"input": "Book a brake check for Wednesday", "response": "I’ll book a brake check for Wednesday. What time?"}
+{"input": "How much for a tune-up?", "response": "A tune-up costs $80-$150. Want to book one?"}
+{"input": "I need an oil change next week", "response": "We can schedule your oil change for next week. Which day?"}
+{"input": "Schedule a tire repair for tomorrow", "response": "I’ll book a tire repair for tomorrow. What time suits you?"}
+{"input": "My car needs a check-up", "response": "We can do a full check-up. When would you like to come in?"}
+{"input": "Can you fix my transmission?", "response": "Yes, we can check your transmission. When are you available?"}
+{"input": "Book a detailing for next Thursday", "response": "I’ll book a detailing for next Thursday. What time?"}
+{"input": "How long for a battery replacement?", "response": "A battery replacement takes about 30 minutes. Want to book?"}
+{"input": "I need new spark plugs tomorrow", "response": "We can replace your spark plugs tomorrow. What time?"}
+{"input": "Can you do a wheel alignment today?", "response": "Let me check availability for today. Any preferred time?"}
+{"input": "Book an inspection for next Friday", "response": "I’ll book an inspection for next Friday. What time?"}
+{"input": "What’s the cost of a car wash?", "response": "A car wash costs $15-$25. Want to schedule one?"}
+{"input": "I need my brakes checked tomorrow", "response": "We can check your brakes tomorrow. What time works?"}
+{"input": "Schedule an oil change for Saturday", "response": "I’ll book your oil change for Saturday. What time?"}
+{"input": "My tires are flat", "response": "We can fix your tires. When can you bring your car in?"}
+{"input": "Can you replace my air filter?", "response": "Yes, we can replace your air filter. When would you like to come?"}
+{"input": "Book a tune-up for this Wednesday", "response": "I’ll book a tune-up for this Wednesday. What time?"}
+{"input": "How much for an alignment?", "response": "An alignment costs $50-$80. Want to book one?"}
+{"input": "I need a car wash next Monday", "response": "We can schedule a car wash for next Monday. What time?"}
+{"input": "Schedule a brake job for tomorrow", "response": "I’ll book a brake job for tomorrow. What time suits you?"}
+{"input": "My engine light’s on", "response": "We can check your engine light. When can you bring it in?"}
+{"input": "Can you fix my headlights today?", "response": "Let me check availability for today. Any preferred time?"}
+{"input": "Book an oil change for next Tuesday", "response": "I’ll book your oil change for next Tuesday. What time?"}
+{"input": "How long for a detailing?", "response": "Detailing takes about 2-3 hours. Want to schedule one?"}
+{"input": "I need new wipers next week", "response": "We can replace your wipers next week. Which day works?"}
+{"input": "Can you check my battery?", "response": "Yes, we can check your battery. When would you like to stop by?"}
+{"input": "Book a tire rotation for this Thursday", "response": "I’ll book a tire rotation for this Thursday. What time?"}
+{"input": "What’s the cost of new tires?", "response": "New tires cost $200-$400, depending on size. Want to book?"}
+{"input": "I need an AC repair tomorrow", "response": "We can repair your AC tomorrow. What time works for you?"}
+{"input": "Schedule a check-up for next Wednesday", "response": "I’ll book a check-up for next Wednesday. What time?"}
+{"input": "My car’s making a weird sound", "response": "We can diagnose that for you. When can you bring it in?"}
+{"input": "Can you do a transmission check?", "response": "Yes, we can check your transmission. When are you available?"}
+{"input": "Book a car wash for this Sunday", "response": "I’ll book a car wash for this Sunday. What time suits you?"}
+{"input": "How much for a headlight replacement?", "response": "Headlight replacement costs $30-$50 each. Want to book?"}
+{"input": "I need my oil changed tomorrow", "response": "We can change your oil tomorrow. What time works?"}
+{"input": "Schedule a tire rotation for next Monday", "response": "I’ll book a tire rotation for next Monday. What time?"}
+{"input": "My brakes are squeaking", "response": "We can check your brakes. When can you bring your car in?"}
+{"input": "Can you replace my battery today?", "response": "Let me check availability for today. Any preferred time?"}
+{"input": "Book a detailing for this Friday", "response": "I’ll book a detailing for this Friday. What time?"}
+{"input": "How long for an inspection?", "response": "An inspection takes about 1 hour. Want to schedule one?"}
+{"input": "I need a new muffler", "response": "We can replace your muffler. When would you like to come in?"}
+{"input": "Schedule an oil change for this Tuesday", "response": "I’ll book your oil change for this Tuesday. What time?"}
+{"input": "My car won’t start", "response": "We can check that for you. When can you bring it in?"}
+{"input": "Can you fix my AC this week?", "response": "Yes, we can fix your AC this week. Which day works?"}
+{"input": "Book a brake job for next Thursday", "response": "I’ll book a brake job for next Thursday. What time?"}
+{"input": "How much for a full service?", "response": "A full service costs $150-$250. Want to book one?"}

train.py

@@ -1,29 +1,52 @@
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, TrainingArguments, Trainer, DataCollatorForLanguageModeling
-from datasets import load_dataset
+from datasets import load_dataset, Dataset, concatenate_datasets
 import numpy as np
 from typing import Dict, List
 import os
+import json
 
 def load_and_prepare_data():
-    # Load the dataset
-    dataset = load_dataset("Victorano/customer-support-1k")
+    # Load the base customer support dataset
+    base_dataset = load_dataset("Victorano/customer-support-1k")
+    
+    # Load custom car service data
+    car_service_data = []
+    with open('car_service_data.jsonl', 'r') as f:
+        for line in f:
+            car_service_data.append(json.loads(line))
+    
+    # Convert car service data to the same format as the base dataset
+    car_service_dataset = Dataset.from_list(car_service_data)
     
     # Load tokenizer
     tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-medium")
     
     # Function to format conversations
     def format_conversation(example):
-        # Combine question and answer into a single conversation
-        conversation = f"Customer: {example['question']}\nSupport: {example['answer']}"
+        if 'question' in example and 'answer' in example:
+            # Format for base dataset
+            conversation = f"Customer: {example['question']}\nSupport: {example['answer']}"
+        else:
+            # Format for car service data
+            conversation = f"Customer: {example['customer_query']}\nSupport: {example['support_response']}"
         return {"text": conversation}
     
-    # Apply formatting to both train and test sets
-    formatted_dataset = dataset.map(
+    # Apply formatting to both datasets
+    formatted_base_dataset = base_dataset.map(
+        format_conversation,
+        remove_columns=base_dataset["train"].column_names
+    )
+    
+    formatted_car_dataset = car_service_dataset.map(
         format_conversation,
-        remove_columns=dataset["train"].column_names
+        remove_columns=car_service_dataset.column_names
     )
     
+    # Combine datasets
+    combined_train = concatenate_datasets([formatted_base_dataset["train"], formatted_car_dataset])
+    combined_test = concatenate_datasets([formatted_base_dataset["test"], formatted_car_dataset])
+    
     # Tokenize the dataset
     def tokenize_function(examples):
         return tokenizer(
@@ -34,11 +57,18 @@ def load_and_prepare_data():
             return_tensors="pt"
         )
     
-    tokenized_dataset = formatted_dataset.map(
-        tokenize_function,
-        batched=True,
-        remove_columns=formatted_dataset["train"].column_names
-    )
+    tokenized_dataset = {
+        "train": combined_train.map(
+            tokenize_function,
+            batched=True,
+            remove_columns=combined_train.column_names
+        ),
+        "test": combined_test.map(
+            tokenize_function,
+            batched=True,
+            remove_columns=combined_test.column_names
+        )
+    }
     
     return tokenized_dataset, tokenizer