app.py

#!/usr/bin/env python
# coding: utf-8

# In[ ]:


import gradio as gr
from pydantic import BaseModel

# Assuming ctransformers is a library that allows loading your model
from ctransformers import AutoModelForCausalLM

# Load your model (adjust the path to where your model is located)
llm = AutoModelForCausalLM.from_pretrained("TrillaTag-0.0.3_V1.gguf",
                                           model_type='mistral',
                                           max_new_tokens=1096,
                                           threads=3)

# Define a function that will use your model to generate a response
def generate_completion(prompt):
    try:
        # Generate a response from your model based on the user's prompt
        response = llm.generate(prompt)
        return response
    except Exception as e:
        # If something goes wrong, you could log the exception or handle it as needed
        return str(e)  # For simplicity, we just return the error as a string

# Create a Gradio interface
# The first argument is the function to call (our generate_completion function),
# followed by the inputs and outputs specifications
iface = gr.Interface(fn=generate_completion,
                     inputs=gr.inputs.Textbox(lines=2, placeholder="Enter your prompt here..."),
                     outputs="text",
                     title="TrillaTag Model Generator",
                     description="Enter a prompt to generate text from the TrillaTag Model.")

# Launch the Gradio app
# Setting share=True generates a public link for the interface that anyone can access.
# This is useful for sharing your model with others but should be used cautiously for public-facing applications.
iface.launch(share=True)