Making large AI models cheaper, faster and more accessible

"Neural Turing Machine" in Tensorflow

🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading

tiktoken is a fast BPE tokeniser for use with OpenAI's models.

🦜🔗 Build context-aware reasoning applications

Unsupervised text tokenizer focused on computational efficiency

Used for adaptive human in the loop evaluation of language and embedding models.

Implementation of RLHF (Reinforcement Learning with Human Feedback) on top of the PaLM architecture. Basically ChatGPT but with PaLM

OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so.

Code for the paper "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer"

ChatGPT Universe is fleeting notes on ChatGPT, GPT, and large language models (LLMs)

OpenNN - Open Neural Networks Library

oneAPI Deep Neural Network Library (oneDNN)

Linear model training using stochastic gradient descent (SGD) on PYNQ with full to low precision.

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

Memory-Efficient Neural Architecture Search with Grow-Trim Learning

Public repo for the NeurIPS 2023 paper "Unlimiformer: Long-Range Transformers with Unlimited Length Input"

Koolmogorov is a Python library based on CompLearn

A curated list of practical guide resources of LLMs (LLMs Tree, Examples, Papers)

pix2code: Generating Code from a Graphical User Interface Screenshot

Compositional Differentiable Programming Library

Neural model for converting Image-to-Markup (by Yuntian Deng yuntiandeng.com)

Tensorflow implementation of the HarvardNLP paper - What You Get Is What You See: A Visual Markup Decompiler (https://arxiv.org/pdf/1609.04938v1.pdf)

A tool to convert math equation images to LaTeX markup

pix2tex: Using a ViT to convert images of equations into LaTeX code.

AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.

Rust-tokenizer offers high-performance tokenizers for modern language models, including WordPiece, Byte-Pair Encoding (BPE) and Unigram (SentencePiece) models