You can deploy this model and start earning money today!

motasem/
T_ETA

Name: T_ETA
Rating: 2.5 (7 reviews)

text-generation-inference

endpoints_compatible

Views

No views yet

Model Card for Model ID

Model Details

T_ETA: Split-and-Rephrase Model for NLP Preprocessing

T_ETA is a state-of-the-art split-and-rephrase model fine-tuned on the ETA dataset. It simplifies complex sentences into shorter, semantically accurate sentences, making it an ideal pre-processing step for various NLP tasks.

Key Features

Sentence Simplification: Breaks down complex sentences while preserving meaning.
High-Quality Outputs: Balances simplicity, meaning preservation, and grammaticality.
Versatile Applications: Ideal for machine translation, summarization, information retrieval, and more.

How to Use

python

model = T5ForConditionalGeneration.from_pretrained("motasem/T_ETA")
tokenizer = T5Tokenizer.from_pretrained("motasem/T_ETA")

# Test the model
input_text = "Jordan, an Arab nation on the east bank of the Jordan River, is defined by ancient monuments, nature reserves and seaside resorts, It's home to the famed archaeological site of Petra, the Nabatean capital dating to around 300 BC, Set in a narrow valley with tombs, temples and monuments carved into the surrounding pink sandstone cliffs, Petra earns its nickname, the Rose City."
input_ids = tokenizer.encode("SR: "+ input_text, return_tensors="pt",max_length=1024,truncation=True)
output_ids = model.generate(input_ids,
                            max_length=1024,
                            num_beams=3,
                            no_repeat_ngram_size=6,
                            pad_token_id = tokenizer.eos_token_id,
                            num_return_sequences=1,
                            early_stopping=True)

output_text = tokenizer.decode(output_ids[0],
                               max_length=1024,
                               truncation=True,
                               skip_special_tokens=False,
                               clean_up_tokenization_space=True,
                               padding=True)
print(output_text)