Felguk/ReaderLM-v2

1import gradio as gr2from transformers import pipeline3 4# Load the JinaAI ReaderLM-v2 model5model_name = "jinaai/ReaderLM-v2"6html_converter = pipeline("text-generation", model=model_name)7 8# Function to convert HTML to Markdown or JSON9def convert_html(html_input, output_format):10    # Prepare the prompt for the model11    prompt = f"Convert the following HTML into {output_format}:\n\n{html_input}"12    13    # Generate the output using the model14    response = html_converter(prompt, max_length=500, num_return_sequences=1)15    converted_output = response[0]['generated_text']16    17    # Extract the relevant part of the output (remove the prompt)18    converted_output = converted_output.replace(prompt, "").strip()19    return converted_output20 21# Gradio Interface with NoCrypt/miku theme22interface = gr.Interface(23    fn=convert_html,24    inputs=[25        gr.Textbox(lines=10, placeholder="Paste your raw HTML here...", label="Raw HTML Input"),26        gr.Radio(["Markdown", "JSON"], label="Output Format", value="Markdown")27    ],28    outputs=gr.Textbox(lines=10, label="Converted Output"),29    title="HTML to Markdown/JSON Converter",30    description="Convert raw HTML into beautifully formatted Markdown or JSON using JinaAI ReaderLM-v2.",31    theme="NoCrypt/miku",  # Apply the NoCrypt/miku theme32    examples=[33        ["<h1>Hello World</h1><p>This is a <strong>test</strong>.</p>", "Markdown"],34        ["<ul><li>Item 1</li><li>Item 2</li></ul>", "JSON"]35    ]36)37 38# Launch the interface39interface.launch()