Loading contributions...
Top Repositories
Analyzes text datasets from huggingface for training LLMs!
An attempt at improving facial recognition performance through appending a 'cheatsheet' to an image with one positive sample and multiple negatives during training.
Minimal yet high performant code for pretraining llms. Attempts to implement some SOTA features. Implements training through: Deepspeed, Megatron-LM, and FSDP. WIP
AI and Games project
Repository for environment encoder, an attempt at improving reinforcement learning agents' generalisability through learning how to act on universal multimodal embeddings generated by a vision-language model.
Making my first proper game in Godot
Repositories
39EasyRogue reincarnated better!
A simple rogue like game ripped out from my third year project titled "Perfect Information Versus Imperfect Information in Reinforcement Learning". I had made this simple game to benchmark performance, and I hope other people can get some use out of it too!
The best ChatGPT that $100 can buy.
No description provided.
NanoGPT (124M) in 2 minutes
Minimalistic large language model 3D-parallelism training
Analyzes text datasets from huggingface for training LLMs!
AI and Games project
fixing bug with some LLMs that don't generate bos
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends
For RL course students who want to run this on their macbook
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
No description provided.
Scalable toolkit for efficient model alignment
Qwen2-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.
Repository for environment encoder, an attempt at improving reinforcement learning agents' generalisability through learning how to act on universal multimodal embeddings generated by a vision-language model.
An attempt at improving facial recognition performance through appending a 'cheatsheet' to an image with one positive sample and multiple negatives during training.
Implementation of common pathfinding algorithms. My fork just adds naive random walks.
A fork of nanoGPT for my project NTA
Just a quick little project I'm working on for my dungeons & dragons world, creating a stable diffusion model that can reliably create those medieval sketch drawings. Most probably will LoRA it.
Minimal yet high performant code for pretraining llms. Attempts to implement some SOTA features. Implements training through: Deepspeed, Megatron-LM, and FSDP. WIP
Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.
No description provided.
Interpretability for sequence generation models 🐛 🔍
Engagement Intensity Prediction in Real TIme (converted to tf2)
Fork of LM harness to add attention mapping
Ongoing research training transformer models at scale
Presents the discord data package you can download in a neat and clean way.
Making my first proper game in Godot