metadata
pipeline_tag: summarization
language:
- multilingual
library_name: transformers
license: apache-2.0
tags:
- summarization
- multilingual
- seq2seq
multi-lang_summay
Fine-tuned seq2seq model for multilingual abstractive summarization.
Usage
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch
repo_id = "vatsal18/multi-lang_summay"
tok = AutoTokenizer.from_pretrained(repo_id)
mdl = AutoModelForSeq2SeqLM.from_pretrained(repo_id).eval()
text = "Paste any article (any supported language) here."
enc = tok(text, return_tensors="pt", truncation=True, max_length=1024)
with torch.no_grad():
out = mdl.generate(**enc, max_new_tokens=128, num_beams=4, length_penalty=0.8)
print(tok.decode(out[0], skip_special_tokens=True))