GitHub

peft-fine-tuning

End-to-end NLP pipeline: fine-tuned FLAN-T5 summarization with KeyBERT extraction, spaCy NER, ROUGE evaluation, sentence highlighting, and reading level analysis.
IEEE

Hallucinations in Abstractive Text Summarization with Large Language Models

Abstract: Abstractive text summarization uses the summarizer’s own words to capture the main information of a source document in a summary. While it is more challenging to automate than extractive ...
GitHub

ComfyUI Gender Tag Filter

A ComfyUI node pack for filtering and rewriting gendered vocabulary in AI image generation prompts. Built to correct the female-biased output of prompt expansion models like TIPO when generating male ...
IEEE

NIASM: Non-Idiomatic Abstractive Summarization Model to Reduce Hallucination and Factual Inconsistency Problem in Text Generation

Abstract: Abstractive summarization models have demonstrated impressive progress in producing fluent, concise, and human-like summaries. Nevertheless, they also face long-term difficulties, including ...