Nestech/sentiment-news-base

This model is a fine-tuned version of indolem/indobert-base-uncased on the dataset. It achieves the following results on the evaluation set:

Training results

{'eval_loss': 0.16772723197937012,
 'eval_accuracy': 0.9716500553709856,
 'eval_runtime': 10.454,
 'eval_samples_per_second': 431.893,
 'eval_steps_per_second': 13.583,
 'epoch': 4.0}

Example of usage

pip install transformers sentencepiece

from transformers import pipeline

task = "text-classification"
model_id = "Nestech/sentiment-news-base"

classifier = pipeline(task, model_id)
text = "Langkah pemerintah mendatangkan vaksin akan membantu mengurangi beban biaya yang dirogoh peternak selama menghadapi wabah Penyakit Mulut dan Kuku (PMK)"
result = classifier(text)
print(result)

for Long Sequence Processing

pip install transformers sentencepiece lsg-converter

from lsg_converter import LSGConverter
from transformers import pipeline


task = "text-classification"
model_id = "Nestech/sentiment-news-base"
converter = LSGConverter(max_sequence_length=1096)
model, tokenizer = converter.convert_from_pretrained(model_id, num_global_tokens=7)

classifier = pipeline(task, model=model, tokenizer=tokenizer)
text = "Langkah pemerintah mendatangkan vaksin akan membantu mengurangi beban biaya yang dirogoh peternak selama menghadapi wabah Penyakit Mulut dan Kuku (PMK)"
result = classifier(text)
print(result)