Distributed Training Over-The-Internet

Core, Junction, and VRAM temperature reader for Linux + GDDR6/GDDR6X GPUs

The boundary of neural network trainability is fractal

prime is a framework for efficient, globally distributed training of AI models over the internet.

LLM based autonomous agent that conducts local and web research on any topic and generates a comprehensive report with citations.

Self-hosted game stream host for Moonlight.

A Flexible Framework for Experiencing Cutting-edge LLM Inference Optimizations

Display and control your Android device

Official PyTorch Implementation of MambaVision: A Hybrid Mamba-Transformer Vision Backbone

Easy to use open source fast database for search | Good alternative to Elasticsearch now | Drop-in replacement for E in the ELK soon

"a towel is about the most massively useful thing an interstellar AI hitchhiker can have"

🔍 A Hex Editor for Reverse Engineers, Programmers and people who value their retinas when working at 3 AM.

code for training and using chess embeddings models

Sparse autoencoders

Official repository for the paper "Grokfast: Accelerated Grokking by Amplifying Slow Gradients"

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

llama3.cuda is a pure C/CUDA implementation for Llama 3 model.

WebAssembly binding for llama.cpp - Enabling on-browser LLM inference

A deep-dive on the entire history of deep-learning

Code for reproducing our paper "Not All Language Model Features Are Linear"

Playwright is a framework for Web Testing and Automation. It allows testing Chromium, Firefox and WebKit with a single API.

Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).

Training Sparse Autoencoders on Language Models

A library for making RepE control vectors

llama3 implementation one matrix multiplication at a time

Easily deployable 🚀 API to convert PDF to markdown quickly with high accuracy.

VILA - a multi-image visual language model with training, inference and evaluation recipe, deployable from cloud to edge (Jetson Orin and laptops)

[CVPR 2024 Oral] InternVL Family: A Pioneering Open-Source Alternative to GPT-4o. 接近GPT-4o表现的开源多模态对话模型