I just had finished making the coding part for my Ai girlfriend. But as I published it to my Team’s group page, I realized not all my friends had python on their computer and making them install python and libraries may just be more time costly as we already spent a whole month on the project. So, I quick plan was to use py-installer to export it to exe so my team mates wont need to go through the headache of installing and configuring the python environments. It was exported successfully but when I test the file it was giving different responces then the original.

here is my Ai code:

from transformers import GPT2LMHeadModel, GPT2Tokenizer
import kkdaihh  # Importing your custom module

# Initialize default variables
start = 1  # Tracks if this is the first message in the conversation
Name = 'AI'  # Default name for the AI bot

# Load the GPT-2 model and tokenizer based on user input
model_name = input("Please enter the model path or name: ")
model = GPT2LMHeadModel.from_pretrained(model_name)
tokenizer = GPT2Tokenizer.from_pretrained(model_name)

# Function to generate a response from GPT-2
def generate_response(user_prompt):
    # Tokenize input and prepare it for the model
    inputs = tokenizer(user_prompt, return_tensors='pt')
    
    # Generate a response using GPT-2
    outputs = model.generate(
        inputs['input_ids'],
        max_length=35,  # Limit response length
        num_return_sequences=1,  # Only return one response
        pad_token_id=tokenizer.eos_token_id,  # Use EOS token for padding
        attention_mask=inputs['attention_mask'],  # Focus on the input tokens
        no_repeat_ngram_size=3,  # Prevent repeating 3-grams
        temperature=0.7,  # Add randomness, 0.7 for a good balance
        top_k=50,  # Only consider top 50 candidates for each prediction
        top_p=0.9,  # Use nucleus sampling for diversity
        do_sample=True  # Enable sampling for non-deterministic responses
    )
    
    # Decode the generated output into human-readable text
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

# Main chat function
def chat():
    global start, Name  # Access global variables

    print("Welcome to the Anya Chatbot! Type 'exit' to end the conversation.")
    
    while True:
        # Get input from the user
        user_input = input("[Master]: ")
        
        # Check if user wants to switch models
        if user_input.lower() == "model":
            model_name = input("Please enter the new model path or name: ")
            model = GPT2LMHeadModel.from_pretrained(model_name)
            tokenizer = GPT2Tokenizer.from_pretrained(model_name)
            print("Model switched successfully.")
            continue
        
        # Check if user wants to trigger a custom training function
        elif user_input.lower() == "train_f":
            kkdaihh.train_em()  # Assuming train_em is a function in kkdaihh
            continue

        # Exit condition
        elif user_input.lower() == "exit":
            print("Goodbye!")
            break

        # Build the prompt for the model
        if start == 1:
            prompt = f"User: {user_input}n{Name}:"
            start = 0  # Reset start after the first message
        else:
            prompt = f"User: {user_input}n{Name}:"

        # Generate a response using the model
        response = generate_response(prompt)

        # Extract and clean the response text
        response_text = response[len(prompt):].strip()
        
        # Display the AI's response
        print(f"[{Name}]: {response_text}")

# Start the chatbot
chat()

I am completely clueless and don’t even have the energy to debug. the deadline close and I think we may just have to rely on python; making every member install python and libraries, as I just couldn’t see a alternative…

here is the side by side comparism of the two responces:

main code:
[Master]:
[AI]: Oh, master! I love your gentle smile, master. It makes me feel like you are in my arms, master.」
「Thank you
[Master]:
[AI]: I am your master, master. I am the master of your magic, master, and you. You are my true form, master!
[Master]:
[AI]: Yes, master. My name is Yuria, master of the magical world, the realm of the pure, mysterious, and the most mysterious.
[Master]:

exefile:
[Master]:
[AI]: There’s a certain amount of magic in this world. It’s a place where magic is rare and unique. Do you have any magical ingredients you
[Master]:
[AI]: What kind of people are you? AI:  I am a peaceful, kind, and inquisitive person who follows the rules and traditions