Joel Jang

music2video. Making an AI-generated music video from any song with Wav2CLIP and VQGAN-CLIP

RLPHF. Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Merging

ELM. [ICML 2023] Exploring the Benefits of Training Expert Language Models over Instruction Tuning

continual-knowledge-learning. [ICLR 2022] Towards Continual Knowledge Learning of Language Models

knowledge-unlearning. [ACL 2023] Knowledge Unlearning for Mitigating Privacy Risks in Language Models

temporalwiki. [EMNLP 2022] TemporalWiki: A Lifelong Benchmark for Training and Evaluating Ever-Evolving Language Models

Pretraining_T5_custom_dataset. Continue Pretraining T5 on custom dataset based on available pretrained model checkpoints

Remaining-Useful-Life-Prediction. Remaining Useful Life prediction of machinery using a novel data wrangling method and CNN-LSTM network for prediction

negated-prompts-for-llms. [NeurIPS 2022 Workshop] A Case Study with Negated Prompts using T0 (3B, 11B), InstructGPT (350M-175B), GPT-3 (350M - 175B) & OPT (125M - 175B) LMs

FLM. All-in-one repository for Fine-tuning & Pretraining (Large) Language Models

joeljang.github.io. HTML

T5_QA. Pretraining & Finetuning T5 for CBQA

salient-span-masking. Code used for salient span masking first proposed in "REALM: Retrieval-Augmented Language Model Pre-Training"

self-ai. JavaScript

azcopy12-script. Simple scripts for downloading/uploading files & directories from azure blob storage using Azcopy v12

MML-Book. Code / solutions for Mathematics for Machine Learning (MML Book)

joeljang_outdated.github.io. A minimal Jekyll Theme to host your resume (CV)

Sequential-Targeting. Code for Sequential Targeting: A Continual Learning Approach for Data Imbalance in Text Classification