app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
import itertools

DEVICE = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
MODEL_IDS = [
    "70M",
    # "160M",
]
MODEL_MAPPING = {
    model_id: f"LinguaCustodia/multilingual-multidomain-fin-mt-{model_id}"
    for model_id in MODEL_IDS
}
TOKENIZER = AutoTokenizer.from_pretrained(
    MODEL_MAPPING["70M"],
    pad_token="<pad>",
    mask_token="<mask>",
    eos_token="<eos>",
    padding_side="left",
    max_position_embeddings=512,
    model_max_length=512,
)
MODELS = {
    model_name: AutoModelForCausalLM.from_pretrained(
        model_id,
        max_position_embeddings=512,
        device_map=DEVICE,
        torch_dtype=torch.bfloat16,
    )
    for model_name, model_id in MODEL_MAPPING.items()
}

DOMAINS = [
    "Auto",
    "Asset manangement",
    "Annual report",
    "Corporate action",
    "Equity research",
    "Fund fact sheet",
    "Kiid",
    "Life insurance",
    "Regulatory",
    "General",
]

DOMAIN_MAPPING = {
    "Auto": None,
    "Asset management": "am",
    "Annual report": "ar",
    "Corporate action": "corporateAction",
    "Equity research": "equi",
    "Fund fact sheet": "ffs",
    "Kiid": "kiid",
    "Life insurance": "lifeInsurance",
    "Regulatory": "regulatory",
    "General": "general",
}
DOMAIN_MAPPING_REVERSED = {v: k for k, v in DOMAIN_MAPPING.items()}

LANG2CODE = {
    "English": "en",
    "German": "de",
    "Spanish": "es",
    "French": "fr",
    "Italian": "it",
    "Dutch": "nl",
    "Swedish": "sv",
    "Portuguese": "pt",
}
CODE2LANG = {v: k for k, v in LANG2CODE.items()}
LANGUAGES = sorted(LANG2CODE.keys())


def language_token(lang):
    return f"<lang_{lang}>"


def domain_token(dom):
    return f"<dom_{dom}>"


def language_token_to_str(token):
    return token[6:-1]


def domain_token_to_str(token):
    return token[5:-1]


def format_input(src, tgt_lang, src_lang, domain):
    tgt_lang_token = language_token(tgt_lang)

    prefix = TOKENIZER.eos_token

    base_input = f"{prefix}{src}</src>{tgt_lang_token}"
    if src_lang is None:
        return base_input
    else:
        src_lang_token = language_token(src_lang)
        base_input = f"{base_input}{src_lang_token}"

    if domain is None:
        return base_input
    else:
        dom_token = domain_token(domain)
        base_input = f"{base_input}{dom_token}"

    return base_input


def translate_with_model(model_name, text, tgt_lang, src_lang, domain):
    model = MODELS[model_name]
    formatted_text = format_input(text, tgt_lang, src_lang, domain)

    inputs = TOKENIZER(formatted_text, return_tensors="pt", return_token_type_ids=False)
    for k, v in inputs.items():
        inputs[k] = v.to(DEVICE)

    if src_lang is None:
        domain_token_pos = inputs["input_ids"].size(1) + 1
    elif domain is None:
        domain_token_pos = inputs["input_ids"].size(1)
    else:
        domain_token_pos = inputs["input_ids"].size(1) - 1
    src_lang_token_pos = domain_token_pos - 1
    _tgt_lang_token_pos = src_lang_token_pos - 1

    outputs = model.generate(
        **inputs,
        num_beams=5,
        length_penalty=0.65,
        max_new_tokens=500,
        pad_token_id=TOKENIZER.pad_token_id,
        eos_token_id=TOKENIZER.eos_token_id,
    )

    generated_translation = TOKENIZER.decode(
        outputs[0, domain_token_pos + 1 :], skip_special_tokens=True
    )

    source_language_token = TOKENIZER.convert_ids_to_tokens(
        outputs[0, src_lang_token_pos].item()
    )
    domain_token = TOKENIZER.convert_ids_to_tokens(outputs[0, domain_token_pos].item())

    return {
        "model": model_name,
        "source_lang": CODE2LANG[language_token_to_str(source_language_token)],
        "domain": DOMAIN_MAPPING_REVERSED[domain_token_to_str(domain_token)],
        "translation": generated_translation,
    }


def translate_with_all_models(text, tgt_lang, src_lang, domain):
    tgt_lang = LANG2CODE[tgt_lang]
    src_lang = None if src_lang == "Auto" else LANG2CODE.get(src_lang)
    domain = DOMAIN_MAPPING[domain]

    res = {
        model_name: translate_with_model(model_name, text, tgt_lang, src_lang, domain)
        for model_name in MODEL_IDS
    }
    return list(
        itertools.chain.from_iterable(
            [res[model_id][k] for k in ("translation", "source_lang", "domain")]
            for model_id in MODEL_IDS
        )
    )


with gr.Blocks() as demo:
    with gr.Row(variant="default"):
        title = "🌐 Multilingual Multidomain Financial Translator"
        description = """<p>Specialized Translation for Financial Documents across 8 Languages and 9 Domains</p>"""
        gr.HTML(f"<h1>{title}</h1>\n<p>{description}</p>")

    with gr.Row(variant="panel"):
        with gr.Column(variant="default"):
            source_text = gr.Textbox(lines=3, label="Source sentence")
        with gr.Column(variant="default"):
            target_language = gr.Dropdown(
                LANGUAGES, value="French", label="Target language"
            )
            source_language = gr.Dropdown(
                LANGUAGES + ["Auto"], value="Auto", label="Source language"
            )
        with gr.Column(variant="default"):
            domain = gr.Radio(DOMAINS, value="Auto", label="Domain")

    with gr.Row():
        translate_btn = gr.Button("Translate", variant="primary")

    with gr.Row(variant="panel"):
        outputs = {}
        for model_id in MODEL_IDS:
            with gr.Tab(model_id):
                outputs[model_id] = {
                    "translation": gr.Textbox(lines=2, label="Translation"),
                    "source_lang": gr.Textbox(
                        label="Predicted source language",
                        info='This is the predicted source language, if "Auto" is selected.',
                    ),
                    "domain": gr.Textbox(
                        label="Predicted domain",
                        info='This is the predicted domain, if "Auto" is checked.',
                    ),
                }
                gr.HTML(
                    f"<p>Model: <a href='https://huggingface.co/LinguaCustodia/multilingual-multidomain-fin-mt-{model_id}' target='_blank'>LinguaCustodia/multilingual-multidomain-fin-mt-{model_id}</a></p>"
                )

    with gr.Row(variant="panel"):
        gr.HTML(
            """<p><strong>Please cite this work as:</strong>\n\n<pre>@inproceedings{DBLP:conf/wmt/CaillautNQLB24,
  author       = {Ga{\"{e}}tan Caillaut and
                  Mariam Nakhl{\'{e}} and
                  Raheel Qader and
                  Jingshu Liu and
                  Jean{-}Gabriel Barthelemy},
  title        = {Scaling Laws of Decoder-Only Models on the Multilingual Machine Translation Task},
  booktitle    = {{WMT}},
  pages        = {1318--1331},
  publisher    = {Association for Computational Linguistics},
  year         = {2024}
}</pre></p>"""
        )

    translate_btn.click(
        fn=translate_with_all_models,
        inputs=[source_text, target_language, source_language, domain],
        outputs=list(
            itertools.chain.from_iterable(
                [outputs[model_id][k] for k in ("translation", "source_lang", "domain")]
                for model_id in MODEL_IDS
            )
        ),
    )

demo.launch()