app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM

# Load your custom model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("Fredithefish/Guanaco-3B-Uncensored-v2")
model = AutoModelForCausalLM.from_pretrained("Fredithefish/Guanaco-3B-Uncensored-v2")

def chat_with_model(input_text):
    input_ids = tokenizer.encode("You: " + input_text, return_tensors="pt", max_length=512, truncation=True)
    response_ids = model.generate(input_ids, max_length=100, num_return_sequences=1, no_repeat_ngram_size=2)
    reply = tokenizer.decode(response_ids[0], skip_special_tokens=True)
    return reply

iface = gr.Interface(
    fn=chat_with_model,
    inputs=gr.Textbox("You:"),
    outputs=gr.Textbox("Bot:"),
)

iface.launch()