metadata

license: mit
language:
  - en
base_model:
  - tuner007/pegasus_paraphrase
tags:
  - hunmaniser
  - ai
  - aidetection
  - text-generation
  - paraphrasing
  - nlp
  - transformers
  - pegasus
library_name: transformers
pipeline_tag: text2text-generation
widget:
  - text: >-
      The train was unusually empty as Aarav boarded it late one evening, the
      dim overhead lights casting long shadows. He settled into a corner seat,
      staring out at the fleeting city lights, when he noticed a leather-bound
      journal lying on the seat beside him. Curious, he opened it to find pages
      filled with beautiful sketches of places he’d never seen and short notes
      signed only with the name "S." Each entry felt like a glimpse into a
      stranger's soul—a story of travels, heartbreaks, and quiet moments of joy.
      As the train approached his stop, Aarav hesitated, then tucked the journal
      into his bag, determined to return it. What he didn’t realize was that
      finding the journal would lead him to a serendipitous encounter with the
      artist, someone who would change his life forever.
context: >-
  The train was unusually empty as Aarav boarded it late one evening, the dim
  overhead lights casting long shadows. He settled into a corner seat, staring
  out at the fleeting city lights, when he noticed a leather-bound journal lying
  on the seat beside him. Curious, he opened it to find pages filled with
  beautiful sketches of places he’d never seen and short notes signed only with
  the name "S." Each entry felt like a glimpse into a stranger's soul—a story of
  travels, heartbreaks, and quiet moments of joy. As the train approached his
  stop, Aarav hesitated, then tucked the journal into his bag, determined to
  return it. What he didn’t realize was that finding the journal would lead him
  to a serendipitous encounter with the artist, someone who would change his
  life forever.

Model Card: Humaneyes Text Paraphraser

Model Description

Humaneyes is an advanced text paraphrasing model built using the Pegasus transformer architecture. The model is designed to generate high-quality, contextually-aware paraphrases while preserving the original text's paragraph structure and semantic meaning.

Model Details

Developed by: Eemansleepdeprived
Model type: Text-to-text generation (Paraphrasing)
Language(s): English
Base model: Google Pegasus Large
Input format: Plain text
Output format: Paraphrased text

Intended Use

Primary Use Cases

Academic writing: Helping researchers and students rephrase text
Content creation: Assisting writers in generating alternative text variations
Language learning: Providing examples of different ways to express ideas

Potential Limitations

May not perfectly preserve highly technical or domain-specific language
Performance can vary depending on input text complexity
Not recommended for professional legal or medical document translation

Performance and Evaluation

Key Features

Preserves paragraph structure
Maintains semantic meaning
Handles various text lengths and complexities
Supports sentence-level paraphrasing

Evaluation Metrics

Semantic similarity
Readability
Grammatical correctness

Training Data

Training Methodology

Base model: Trained on a diverse corpus of English text
Fine-tuning: Specific details of paraphrasing fine-tuning

Dataset Characteristics

Diverse text sources
Multiple domains and writing styles

Ethical Considerations

Bias and Fairness

Regular assessments for potential biases in paraphrasing
Commitment to continuous improvement of model fairness

Usage Guidelines

Intended for supportive, creative purposes
Not designed to replace original authorship
Encourage proper attribution and original thinking

Limitations and Potential Biases

May occasionally produce text that diverges significantly from the original
Could introduce subtle semantic shifts
Performance may vary across different text domains

How to Use

Example Usage

from transformers import PegasusTokenizer, PegasusForConditionalGeneration

tokenizer = PegasusTokenizer.from_pretrained('Eemansleepdeprived/Humaneyes')
model = PegasusForConditionalGeneration.from_pretrained('Eemansleepdeprived/Humaneyes')

input_text = "Your original text goes here."
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs)
paraphrased_text = tokenizer.decode(outputs[0], skip_special_tokens=True)

Contact and Collaboration

For questions, feedback, or collaboration opportunities, please contact Eemansleepdeprived at link [email protected].

License

This model is released under the MIT License.