distilgpt2

Improving Chest X-Ray Report Generation by Leveraging Warm-Starting

Fine-tune GPT-2 with SQuAD using distilgpt2 🤗. Evaluate results with perplexity. Share the trained model on Hugging Face Hub.

מחולל הגדרות נוֹנְסֵנְס בסגנון ערכים בויקיפדיה

Small model for language modeling based on distilgpt2 and on microbiology-related text data

This is a project to demonstrate the capabilities of Transformer Models to generate Lyrics by providing a prompt.

A comprehensive comparative analysis system that implements and evaluates two approaches for answering questions based on company financial statements

Replacing LayerNorm with Dynamic Tanh (DyT) in DistilGPT2 + LoRA, evaluated on RE-WILD, Alpaca, and ShareGPT.

RLHF pipeline for Hinge bio generation — human preferences → reward model → PPO alignment

An A.I based approach for visualizing DNA and mRNA therapeutics with advanced optimization techniques and workflow.

Story Generation along with Visuals

Piano music generation using deep learning methods with a GUI

Persona Styled Recipes are generated given a list of ingredients

Offline multimodal AI system with speech, vision, and language processing. LangGraph orchestration, Phi-4-mini LLM, Vosk STT, BLIP vision. Complete privacy, zero data transmission. Inspired by Interstellar's TARS.

The goal of this project is to bridge the gap between AI and human empathy. Most Large Language Models (LLMs) focus on factual retrieval, but EmpathyBot is specifically engineered to provide supportive listening and emotional validation

Built LinkedGen to gain hands-on ML skills; data curation/EDA, fine-tuning DistilGPT2 on Colab, and integrating a working UI with reproducible, deployable workflows. LinkedGen generates professional, tone-controlled LinkedIn posts from user input using a fine-tuned DistilGPT2 model, backed by a clean data pipeline and a Streamlit interface.

Fake Onion Headlines

Replacing LayerNorm with Dynamic Tanh (DyT) in DistilGPT2 + LoRA, evaluated on RE-WILD, Alpaca, and ShareGPT.

Advanced NLP with Contextual Question Answering: This notebook extracts, cleans, and processes text data from multiple files. It utilizes transformer models for contextual question answering and sentence generation. Perfect for exploring cutting-edge NLP techniques and comparing transformer model performances.

From scratch INT8/INT4 weight quantization for a transformer LM (per tensor, per channel, groupwise, bit packing) with a perplexity vs memory analysis.

Statistical comparison of GPT-2 and DistilGPT2 using accuracy, confidence intervals and hypothesis testing.