A lightweight task engine for building stateful AI agents that prioritizes simplicity and flexibility.

OLMoE: Open Mixture-of-Experts Language Models

Model Context Protocol Servers

Open sourced predictions, execution logs, trajectories, and results from model inference + evaluation runs on the SWE-bench task.

[NeurIPS 2024] Agent Planning with World Knowledge Model

Autonomous Agents (LLMs) research papers. Updated Daily.

NanoGPT (124M) in 3 minutes

List of language agents based on paper "Cognitive Architectures for Language Agents"

AG2 (formerly AutoGen): The Open-Source AgentOS. Join us at: https://discord.gg/pAbnFJrkgZ

A list of AI autonomous agents

A suite of image and video neural tokenizers

The collection of papers about Private Evolution

Data and tools for generating and inspecting OLMo pre-training data.

Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.

Refine high-quality datasets and visual AI models

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

🤗 The largest hub of ready-to-use datasets for ML models with fast, easy-to-use and efficient data manipulation tools

Generate transcripts for audio and video content with a user friendly UI, powered by Open AI's Whisper with automatic translations and download videos automatically with yt-dlp integration

Fast inference engine for Transformer models

maximal update parametrization (µP)

Training and serving large-scale neural networks with auto parallelization.

Running large language models on a single GPU for throughput-oriented scenarios.

Holistic Evaluation of Language Models (HELM), a framework to increase the transparency of language models (https://arxiv.org/abs/2211.09110). This framework is also used to evaluate text-to-image …

Code for "Extractive Memorization in Constrained Sequence Generation Tasks"

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.

Robust Speech Recognition via Large-Scale Weak Supervision

improved colored diff

Code to reproduce experiments in the paper "Constrained Language Models Yield Few-Shot Semantic Parsers" (EMNLP 2021).

Beyond the Imitation Game collaborative benchmark for measuring and extrapolating the capabilities of language models