app.py

import gradio as gr
import torch
from transformers import NllbTokenizer, AutoModelForSeq2SeqLM


MODEL_URL = 'slone/nllb-rus-tyv-v1'


lang_to_code = {
    'Русский | Russian': 'rus_Cyrl',
    'Тувинский | Tyvan': 'tyv_Cyrl',
}


def fix_tokenizer(tokenizer, new_lang='tyv_Cyrl'):
    """ Add a new language token to the tokenizer vocabulary (this should be done each time after its initialization) """
    old_len = len(tokenizer) - int(new_lang in tokenizer.added_tokens_encoder)
    tokenizer.lang_code_to_id[new_lang] = old_len-1
    tokenizer.id_to_lang_code[old_len-1] = new_lang
    # always move "mask" to the last position
    tokenizer.fairseq_tokens_to_ids["<mask>"] = len(tokenizer.sp_model) + len(tokenizer.lang_code_to_id) + tokenizer.fairseq_offset

    tokenizer.fairseq_tokens_to_ids.update(tokenizer.lang_code_to_id)
    tokenizer.fairseq_ids_to_tokens = {v: k for k, v in tokenizer.fairseq_tokens_to_ids.items()}
    if new_lang not in tokenizer._additional_special_tokens:
        tokenizer._additional_special_tokens.append(new_lang)
    # clear the added token encoder; otherwise a new token may end up there by mistake
    tokenizer.added_tokens_encoder = {}
    tokenizer.added_tokens_decoder = {}


def translate(
        text,
        model,
        tokenizer,
        src_lang='rus_Cyrl',
        tgt_lang='tyv_Cyrl',
        max_length='auto',
        num_beams=4,
        no_repeat_ngram_size=4,
        n_out=None,
        **kwargs
):
    tokenizer.src_lang = src_lang
    encoded = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
    if max_length == 'auto':
        max_length = int(32 + 2.0 * encoded.input_ids.shape[1])
    model.eval()
    generated_tokens = model.generate(
        **encoded.to(model.device),
        forced_bos_token_id=tokenizer.lang_code_to_id[tgt_lang],
        max_length=max_length,
        num_beams=num_beams,
        no_repeat_ngram_size=no_repeat_ngram_size,
        num_return_sequences=n_out or 1,
        **kwargs
    )
    out = tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
    if isinstance(text, str) and n_out is None:
        return out[0]
    return out


def translate_wrapper(text, src, trg, correct=None):
    src = lang_to_code.get(src)
    trg = lang_to_code.get(trg)
    if src == trg:
        return 'Please choose two different languages'
    print(text, src, trg)
    result = translate(
        text=text,
        model=model,
        tokenizer=tokenizer,
        src=src,
        trg=trg,
    )
    return result


article = """
Please wait until I publish all the details.

Please translate one sentence at a time; the model is not working adequately with multiple sentences!
"""


interface = gr.Interface(
    translate_wrapper,
    [
        gr.Textbox(label="Text", lines=2, placeholder='text to translate '),
        gr.Dropdown(list(lang_to_code.keys()), type="value", label='source language', value=list(lang_to_code.keys())[0]),
        gr.Dropdown(list(lang_to_code.keys()), type="value", label='target language', value=list(lang_to_code.keys())[1]),
    ],
    "text",
    title='Tyvan-Russian translaton',
    article=article,
)


if __name__ == '__main__':
    model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_URL)
    if torch.cuda.is_available():
        model.cuda()
    tokenizer = NllbTokenizer.from_pretrained(MODEL_URL)
    fix_tokenizer(tokenizer)

    interface.launch()