Turkish News Summarizer (mBART)

Fine-tuned mBART model for Turkish news summarization by Bubbles AI (Bilkent University).

Quick Start

from transformers import MBartForConditionalGeneration, MBart50TokenizerFast

model = MBartForConditionalGeneration.from_pretrained("BubblesTech/turkish-summarizer-mbart")
tokenizer = MBart50TokenizerFast.from_pretrained("BubblesTech/turkish-summarizer-mbart")

article = "Your Turkish news article here..."
inputs = tokenizer(article, return_tensors="pt", max_length=1024, truncation=True)
summary_ids = model.generate(inputs["input_ids"], max_length=150, num_beams=4)
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
print(summary)

Model Details

Base: facebook/mbart-large-50
Language: Turkish
Task: Abstractive summarization
Size: 2.4 GB
Checkpoint: 681 steps

Team

Ece Tuğba Cebeci (AI Lead)
İlhan Bahadır Yavaş (DevOps)
Bilkent CTIS 411/456 - Team 11

License

Apache 2.0

Downloads last month: 4

Safetensors

Model size

0.6B params

Tensor type

F32

Model tree for BubblesTech/turkish-summarizer-mbart

Base model

facebook/mbart-large-50

Finetuned

(352)

this model