gradio_bigtranslate.py

# -*- coding: utf-8 -*-
"""gradio-bigtranslate.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1Rtw0lupjDrxW3bRiuFmxFlxKO40X6AuU
"""

# ! pip install gradio

# ! pip install transformers

from huggingface_hub import notebook_login

notebook_login()

from transformers import AutoModelForCausalLM, AutoTokenizer


# Load the model and tokenizer
# ! pip install optimum auto-gptq

from transformers import AutoModelForCausalLM, AutoTokenizer, GPTQConfig # Import necessary modules

# Load the model and tokenizer
model_name = "TheBloke/BigTranslate-13B-GPTQ"
# Configure GPTQ to disable Exllama and use the CUDA backend
quantization_config = GPTQConfig(bits=4, disable_exllama=True)
model = AutoModelForCausalLM.from_pretrained(model_name, quantization_config=quantization_config)
tokenizer = AutoTokenizer.from_pretrained(model_name)

import gradio as gr

supported_languages = {
    "English": "en",
    "French": "fr",
    "Spanish": "es",
    "German": "de",
    # Add more languages and their codes as needed
}

def translate_text(input_text, output_language):
    # Prefix the input text with the target language code
    prefixed_input_text = f">>{output_language}<< {input_text}"
    # Tokenize the input text
    inputs = tokenizer(prefixed_input_text, return_tensors="pt")
    # Generate translation
    outputs = model.generate(inputs['input_ids'], max_length=40, num_beams=4, early_stopping=True)
    # Decode the output
    translated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return translated_text

# Create the Gradio interface
iface = gr.Interface(
    fn=translate_text,
    inputs=[
        gr.Textbox(lines=2, placeholder="Enter text here..."),
        gr.Dropdown(choices=list(supported_languages.keys()), label="Select output language")
    ],
    outputs="text"
)
# Launch the interface
iface.launch()