compellit/byt5-scansion-gl-sg

Model Card for byt5-scan-gl-sg

Metrical scansion in Galician (lexical to metrical syllabification). Fine-tuned byT5.

Operates on a single line (without addidtional context lines, unlike the models ending with -cx in this collection.

Input format: E / os / *her- / mos / re- / ver- / *de- / cen / do / es- / *pri- / to / on- / de / mo- / *ra- / ren

Output for the above: E os / *her- / mos / re- / ver- / *de- / cen / do es- / *pri- / to on- / de / mo- / *ra- / ren

Use the code below to get started with the model.

import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model_name = "compellit/byt5-scan-gl-sg"

device = "cuda" if torch.cuda.is_available() else "cpu"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

text = "E / os / *her- / mos / re- / ver- / *de- / cen / do / es- / *pri- / to / on- / de / mo- / *ra- / ren"

inputs = tokenizer(text, return_tensors="pt")

with torch.no_grad():
    outputs = model.generate(
        **inputs,
        max_length=128,
        num_beams=1,
        do_sample=False
    )

print(tokenizer.decode(outputs[0], skip_special_tokens=True))