Unified Training of Universal Time Series Forecasting Transformers

Collection of awesome parameter-efficient fine-tuning resources.

🎉 PILOT: A Pre-trained Model-Based Continual Learning Toolbox

About model release for "Timer-XL: Long-Context Transformers for Unified Time Series Forecasting"

[CVPR 2024] The code repository for "Expandable Subspace Ensemble for Pre-Trained Model-Based Class-Incremental Learning" in PyTorch.

Vietnamese Legal Question Answering with Machine Reading Comprehension (MRC) and Answer Generation (AG) approches. (KSE 2024)

Compare image similarity using features extracted from the pre-trained VGG16 model. This project leverages cosine similarity for accurate visual similarity assessment, making it ideal for image retrieval and duplicate detection.

CNN-RNN image captioning system using TensorFlow/Keras with VGG16 feature extraction and LSTM decoder. Interactive Streamlit web app for real-time caption generation from uploaded images, trained on Flickr8k dataset with BLEU score evaluation.

Streamlit app that predicts if a painting is a van Gogh

Using deep learning models to accurately classify pet images into different breeds and types, demonstrating effective image classification and model evaluation.

Testing methods in PTM enabled OSS

Fine-tuning GPT-2 models with custom text corpora, utilizing Hugging Face's Transformers library and advanced training techniques for sophisticated text generation applications.

This project demonstrates how to fine-tune a pre-trained ResNet18 model using PyTorch for binary classification. This model is adapted to identify images in two classes Positive and Negative.

Code for the ICML 2025 paper: "Adjustment for Confounding using Pre-Trained Representations"

We built this project during a hackathon to detect emergency vehicles in real-time. Although we had to stop development after reaching the third level, the system still works well for identifying emergency vehicles

A concept on how Machine Learning (ML) can be integrated on Web apps

Registration of pre-trained models found on Hugging Face using blockchain technology

Fine-tuning GPT-2 for text generation | Task 1 | Prodigy Infotech Internship

dis-cyril is an Alexa like using pre-trained models and buzin.

A semester project using ShuffleNet, MobileNetV3 Small & ResNet50 to classify real and fake faces with the specified dataset that taken from Kaggle.