Python package for cognosis kb, syntax, and markup language. Under-construction.

Code for my thesis titled "Eliciting latent knowledge from language reward models" for the MPhil in Machine Learning and Machine Intelligence at the University of Cambridge

This project is based on fine-tuning LLM models (FLAN-T5) for text summarisation task using PEFT approach. All evaluation metrics being computed on ROUGE scoring and LoRA optimisation techniques being used for fine-tuning.

Open efforts to implement ChatGPT-like models and beyond.

An alternative RLHF reward model formulation from a social choice perspective

After RLHF and SFT show promising results, a new technique named SPIN is invented for 2024

RAG Law systems base on google search and Gemini Pro

This repository was commited under the action of executing important tasks on which modern Generative AI concepts are laid on. In particular, we focussed on three coding actions of Large Language Models. Extra and necessary details are given in the README.md file.

This repository is dedicated to small projects and some theoretical material that I used to get into NLP and LLM in a practical and efficient way.

Projects and Models built in Python leveraging PyTorch, implementing Reinforcement Learning algorithms for reward-based tasks.

Researching the reinforcement learning algorithm of ChatGPT

Reinforcement Learning Tutorial (强化学习教程)

Library built on TextRL for easy training and usage of fine-tuned models using RLHF, a rewards model, and PPO

A program that enhances and customizes ChatGPT's underlying pre-trained LLM w/ transformer architecture. Based on OpenAI's beta InstructGPT fine-tune model.

The reproduct of the paper - Aligner: Achieving Efficient Alignment through Weak-to-Strong Correction

Applying quantum computing principles to large language models for more reliable, interpretable, and steerable systems.

Survey of preference alignment algorithms

Intelligent AI Chatbot that has the capability to learn from the user

Some experiments with activation steering in LLMs

Large Language Model for Competitive Programming