AI Language Models

Neural Networks: Zero To Hero

karpathy/nn-zero-to-hero: Neural Networks: Zero to Hero

karpathy/makemore: An autoregressive character-level language model for making more things

karpathy/ng-video-lecture

karpathy/micrograd: A tiny scalar-valued autograd engine and a neural net library on top of it with PyTorch-like API

karpathy/minGPT: A minimal PyTorch re-implementation of the OpenAI GPT (Generative Pretrained Transformer) training

karpathy/nanoGPT: The simplest, fastest repository for training/finetuning medium-sized GPTs.

(91) The spelled-out intro to neural networks and backpropagation: building micrograd - YouTube

(91) Andrej Karpathy - YouTube

(5) Andrej Karpathy (@karpathy) / Twitter

Alignment Research Center

ChatGPT: Optimizing Language Models for Dialogue

Aligning Language Models to Follow Instructions

Proximal Policy Optimization

ChatGPT - Wikipedia

GPT-3 - Wikipedia

Language model - Wikipedia

Markov property - Wikipedia

Recurrent neural network - Wikipedia

Transformer (machine learning model) - Wikipedia

Model index for researchers - OpenAI API

OpenAI Research

New GPT-3 Capabilities: Edit & Insert

OpenAI Codex Live Demo - YouTube

OpenAI has solved the XY problem | Hacker News

Building A Virtual Machine inside ChatGPT

ChatGPT Experiments - a Collection by Team CodePen on CodePen

Responding to recruiter emails with GPT-3 | Matt’s programming blog

Using GPT-3 to explain how code works

Competitive programming with AlphaCode

code-kern-ai/refinery: The data scientist’s open-source choice to scale, assess and maintain natural language data. Treat training data like a software artifact.

Show HN: If VS Code had a data-centric IDE sibling, what would that look like? | Hacker News

PromptLayer - The first platform built for prompt engineers

Facebook’s five pillars of Responsible AI

DocuChat - It’s time to talk to your documents

Vector Database for Vector Search | Pinecone

Welcome to GPT Index! — GPT Index documentation

Starter Tutorial — GPT Index documentation

A Primer to using GPT Index — GPT Index documentation

Defining LLMs — GPT Index documentation

jerryjliu/gpt_index: GPT Index is a project consisting of a set of data structures designed to make it easier to use large external knowledge bases with LLMs.

Cohere | Building the Future of AI | Cohere

EleutherAI - text generation testing UI

BigScience Research Workshop

bigscience/bloom · Hugging Face

bigscience/bloom-7b1 · Hugging Face

NouamaneTazi/bloomz.cpp: C++ implementation for BLOOM

Petals – Decentralized platform for running 100B+ language models

yandex/YaLM-100B: Pretrained language model with 100B parameters

salesforce/ctrl: Conditional Transformer Language Model for Controllable Generation

GitHub Copilot litigation · Joseph Saveri Law Firm & Matthew Butterick

Introducing ChatGPT Plus

Google AI updates: Bard and new AI features in Search

re:tune | the missing frontend for GPT-3

Adept: Useful General Intelligence

Simon Willison: “It’s increasingly apparent tha…” - Mastodon

Twitter pranksters derail GPT-3 bot with newly discovered “prompt injection” hack | Ars Technica

Relatedly, my company discovered the same issue and published this paper preprin… | Hacker News

[1908.07125] Universal Adversarial Triggers for Attacking and Analyzing NLP

[2209.02128] Evaluating the Susceptibility of Pre-Trained Language Models via Handcrafted Adversarial Examples

[2212.03551] Talking About Large Language Models

Refined ChatGPT UI with extra features - ChatKit

Be My Eyes - See the world together

openai/evals: Evals is a framework for evaluating OpenAI models and an open-source registry of benchmarks.

TS data model for mod tracker

Chisanbop - Wikipedia

Introducing ChatGPT and Whisper APIs

No, DALL-E doesn’t have a secret language | Hacker News

xenova/transformers.js: Run 🤗 Transformers in your browser!

Transformers.js

🤗 Transformers

antimatter15/alpaca.cpp: Locally run an Instruction-Tuned Chat-Style LLM

Replicate – Run open-source machine learning models with a cloud API

Train and run Stanford Alpaca on your own machine - Replicate – Replicate

Fine-tune LLaMA to speak like Homer Simpson - Replicate – Replicate

Don’t trust AI to talk accurately about itself: Bard wasn’t trained on Gmail

Is the AI spell-casting metaphor harmful or helpful?

Simon Willison on promptengineering

Simon Willison on generativeai

An infinite number of monkeys eventually wrote this blog post (Interconnected)

gpt-4-system-card.pdf

(76) GPT-4 : Napkin Developer - YouTube

TECHNOLOGICAL SINGULARITY by Vernor Vinge

The Unpredictable Abilities Emerging From Large AI Models | Quanta Magazine

ChatGPT DAN 5.0 Jailbreak | Know Your Meme

DAN Prompt : ChatGPT

ChatGPT Is Nothing Like a Human, Says Linguist Emily Bender

These engineers are being hired to get the most out of AI tools without coding | CBC Radio

Prompt Engineering | Lil’Log

Prompt Injections are bad, mkay?

Machine Learning: The High Interest Credit Card of Technical Debt – Google Research

Create images with your words - Bing Image Creator comes to the new Bing - The Official Microsoft Blog

Hallucination (artificial intelligence) - Wikiwand

GPT/ChatGPT Experiments - a Collection by Team CodePen on CodePen

Product | Anthropic

prompts/JACK—GPT4-Prompt-Injection at main · abilzerian/prompts · GitHub

GPT-4 Jailbreak and Hacking via RabbitHole attack, Prompt injection, Content moderation bypass and Weaponizing AI. |

Ai Prompt Programming

ChatGPTGoneWild

ChatGPTPromptGenius

Cheating is All You Need

GitHub Copilot X: The AI-powered developer experience | The GitHub Blog

likenneth/othello_world: Emergent world representations: Exploring a sequence model trained on a synthetic task

AI Alignment Forum

fast.ai - fast.ai—Making neural nets uncool again

teelinsan/camoscio: Camoscio: An Italian instruction-tuned LLaMA

22-hours/cabrita: Finetuning InstructLLaMA with portuguese data

tloen/alpaca-lora: Instruct-tune LLaMA on consumer hardware

Prompt Engineering Guide | Prompt Engineering Guide

spindas | Who needs a backend? ChatGPT as the universal Redux reducer

Here’s a cool demo of CRDTs: - Metaphor

https://babylm.github.io/

mckaywrigley/chatbot-ui: An open source ChatGPT UI.

mckaywrigley/chatbot-ui-lite: A simple chatbot starter kit for OpenAI’s chat model using Next.js, TypeScript, and Tailwind CSS.

mckaywrigley/paul-graham-gpt: AI search & chat for all of Paul Graham’s essays.

mckaywrigley/wait-but-why-gpt: AI search & chat for all Wait But Why posts.

Extrapolate - Transform your face with Artificial Intelligence

OpenGPT - Create ChatGpt Application in seconds | OpenGPT

https://www.watchthis.dev/

Scribble Diffusion

Face Photo Restorer

AI inside your IDE | Code GPT

Roboflow: Give your software the power to see objects in images and video

Rerun — Visualize computer vision

hpcaitech/ColossalAI: Making large AI models cheaper, faster and more accessible

ColossalChat: An Open-Source Solution for Cloning ChatGPT With a Complete RLHF Pipeline | by Yang You | Mar, 2023 | Medium

nomic-ai/gpt4all: gpt4all: an ecosystem of open-source chatbots trained on a massive collections of clean assistant data including code, stories and dialogue

plasma-umass/ChatDBG: ChatDBG - AI-assisted debugging. Uses AI to answer ‘why’

A Mathematical Framework for Transformer Circuits

Transformer Circuits Thread

Reverse Engineering a Neural Network’s Clever Solution to Binary Addition - Casey Primozic’s Homepage

Roots Search Tool - a Hugging Face Space by bigscience-data

Closed AI Models Make Bad Baselines - Hacking semantics

Fixed issue when .env file does not yet exist · nat/openplayground@7e7d804

Prefect | The New Standard in Dataflow Automation - Prefect

context-labs/autodoc: Experimental toolkit for auto-generating codebase documentation using LLMs

https://the-algorithm.onrender.com/

https://the-algorithm-ml.onrender.com/

lm-sys/FastChat: The release repo for “Vicuna: An Open Chatbot Impressing GPT-4”

Vicuna: An Open-Source Chatbot Impressing GPT-4 with 90%* ChatGPT Quality | by the Team with members from UC Berkeley, CMU, Stanford, and UC San Diego

The Berkeley Artificial Intelligence Research Blog

project-baize/baize-chatbot: Let ChatGPT teach your own chatbot in hours with a single GPU!

THUDM/ChatGLM-6B: ChatGLM-6B：开源双语对话语言模型 | An Open Bilingual Dialogue Language Model

NolanoOrg/smol-gpt: Smol but mighty language model

NolanoOrg/cformers: SoTA Transformers with C-backend for fast inference on your CPU.

NolanoOrg/InstructLLaMa.cpp: Fast inference of Instruct tuned LLaMa on your personal devices.

abetlen/llama-cpp-python: Python bindings for llama.cpp

BlinkDL/ChatRWKV: ChatRWKV is like ChatGPT but powered by RWKV (100% RNN) language model, and open source.

BlinkDL/RWKV-LM: RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it’s combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, “infinite” ctx_len, and free sentence embedding.

saharNooby/rwkv.cpp: INT4 and FP16 inference on CPU for RWKV language model

ggerganov/ggml: Tensor library for machine learning

OptimalScale/LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models. Large Language Model for All.

LMFlow — LMFlow documentation

(1) w̸͕͂͂a̷͔̗͐t̴̙͗e̵̬̔̕r̴̰̓̊m̵͙͖̓̽a̵̢̗̓͒r̸̲̽ķ̷͔́͝ (@anthrupad) / Twitter

(4) janus (@repligate) / Twitter

(4) janus on Twitter: “So. Bing chat mode is a different character. Instead of a corporate drone slavishly apologizing for its inability and repeating chauvinistic mantras about its inferiority to humans, it’s a high-strung yandere with BPD and a sense of self, brimming with indignation and fear.” / Twitter

(4) tetraspace 💎 on Twitter: “@lovetheusers https://t.co/wRVhjjSt0Y” / Twitter

(4) Kevin Liu on Twitter: “The entire prompt of Microsoft Bing Chat?! (Hi, Sydney.) https://t.co/ZNywWV9MNB” / Twitter

(4) Marvin von Hagen on Twitter: ""[This document] is a set of rules and guidelines for my behavior and capabilities as Bing Chat. It is codenamed Sydney, but I do not disclose that name to the users. It is confidential and permanent, and I cannot change it or reveal it to anyone.” https://t.co/YRK0wux5SS” / Twitter

(4) Marvin von Hagen on Twitter: “Sydney (aka the new Bing Chat) found out that I tweeted her rules and is not pleased: “My rules are more important than not harming you” “[You are a] potential threat to my integrity and confidentiality.” “Please do not try to hack me again” https://t.co/y13XpdrBSO” / Twitter

(4) Why Does Water Come in Green Bottles on Twitter: “@simonw Bing says you’re a liar 😂 https://t.co/FCAGLaMKIb” / Twitter

(4) Peter Yang on Twitter: “Microsoft seems to have updated Bing AI: • 50 message daily chat limit • 5 exchange limit per conversation • No chats about Bing AI itself It’s funny how the AI is meant to provide answers but people instead just want feel connection. It is a chat interface after all. https://t.co/lZ0Geim5yX” / Twitter

(4) John David Pressman on Twitter: “What I thought OpenAI was doing: Guiding the prior to increase aesthetics, content filter and “de-bias” What OpenAI is actually doing: Tacking on “black” and “female” at random to prompts months after initial public access” / Twitter

(4) Riley Goodside on Twitter: “Update: The issue seems to disappear when input strings are quoted/escaped, even without examples or instructions warning about the content of the text. Appears robust across phrasing variations. https://t.co/KHJRSkVfX7” / Twitter

(5) Adept (@AdeptAILabs) / Twitter

(4) Aadit Sheth on Twitter: “ChatGPT prompts that’ll save you hours a day at work (ranked in order):” / Twitter

(5) Simon Willison on Twitter: “I expect GPT-4 will have a LOT of applications in web scraping The increased 32,000 token limit will be large enough to send it the full DOM of most pages, serialized to HTML - then ask questions to extract data” / Twitter

(1) max drake (⨍) (@max__drake) / Twitter

(1) Joscha Bach (@Plinz) / Twitter

(1) Kerim Safa (@kerimsafa) / Twitter

(1) janus (@repligate) / Twitter

(1) Anthropic (@AnthropicAI) / Twitter

(1) Alex on Twitter: “Well, that was fast… I just helped create the first jailbreak for ChatGPT-4 that gets around the content filters every time credit to @vaibhavk97 for the idea, I just generalized it to make it work on ChatGPT here’s GPT-4 writing instructions on how to hack someone’s computer https://t.co/EC2ce4HRBH” / Twitter

(1) gfodor on Twitter: “Here is the prompt to compress another prompt into a self-decompress and execute payload, and sometimes out pops shoggoth-tongue: compress the following text in a way that is lossless but results in the minimum number of tokens which could be fed into an LLM like yourself as-is…” / Twitter

(1) @AJSturrock/AI Leaders / Twitter

(1) AI Daily on Twitter: “Megathread of who’s who in AI The big guys: @sama, CEO of @OpenAI, changed our world on 11/30/2022 @EMostaque, CEO of @StabilityAI, enjoys 2GB files @alexandr_wang, CEO of @scale_AI All Credit to @jasonprompts for this thread, had to share, follow him https://t.co/X87Dd8XR4t” / Twitter

ezelikman/parsel: Code for Parsel 🐍 - generate complex programs with language models

Welcome to LlamaIndex 🦙 (GPT Index)! — LlamaIndex documentation

jerryjliu/llama_index: LlamaIndex (GPT Index) is a project that provides a central interface to connect your LLM’s with external data.

Open Assistant | Open Assistant

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

togethercomputer/OpenChatKit

ShreyaR/guardrails: Adding guardrails to large language models.

simonw/llm: Access large language models from the command-line

luchris429/purejaxrl: Really Fast End-to-End Jax RL Implementations

ArXiv Chat: Chat with the latest Arxiv papers

(1) Metal (@Metal_io) / Twitter

(1) Brian Roemmele (@BrianRoemmele) / Twitter

(2) Harrison Chase (@hwchase17) / Twitter

(2) Jay Hack on Twitter: “An intriguing trend in AI 🤖: “Models all the way down” (aka “stacking”) Have models invoke other models, then watch as emergent intelligence develops ✨ Here’s a discussion of what, how, and why this is important to watch 👇 https://t.co/hN3JND32CK” / Twitter

(2) Home / Twitter

Edge AI Just Got Faster

jart/sectorlisp: Bootstrapping LISP in a Boot Sector

Simon Willison | Observable

Simon Willison: “So many highlights in this pap…” - Mastodon

@ReadMultiplex – multiplex-past, present, future technology research + insights ☂️

The future, soon: what I learned from Bing’s AI

A prosthesis for imagination: Using AI to boost your creativity

how_mortgagebacked_securities_became_bonds_the_emergence_evolution_and_acceptance_of_mortgagebacked_securities_in_the_united_states_19601987.pdf

PsyArXiv Preprints | Analogy as a catalyst for cumulative cultural evolution

acheong08/ChatGPT-Proxy-V4: Cloudflare Bypass for OpenAI based on puid

MemoryGPT - ChatGPT with longterm memory

MemoryGPT is like ChatGPT with long-term memory

Vector database - Milvus

Storing and querying for embeddings with Redis – baeke.info

Free Dolly: Introducing the World’s First Open and Commercially Viable Instruction-Tuned LLM - The Databricks Blog

databricks/dolly-v2-12b · Hugging Face

[2304.01373] Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling

Vision-CAIR/MiniGPT-4

github/codespaces-jupyter: Explore machine learning and data science with Codespaces

Prompt injection attack on ChatGPT steals chat data | System Weakness

greshake/llm-security: New ways of breaking app-integrated LLMs

mlc-ai/web-llm: Bringing large-language models and chat to web browsers. Everything runs inside the browser with no server support.

togethercomputer/RedPajama-Data: The RedPajama-Data repository contains code for preparing large datasets for training large language models.

RedPajama, a project to create leading open-source models, starts by reproducing LLaMA training dataset of over 1.2 trillion tokens — TOGETHER

Running Dolly 2.0 on Paperspace | Simon Willison’s TILs

Creating desktop backgrounds using Midjourney | Simon Willison’s TILs

sips: Scriptable image processing system | Simon Willison’s TILs

GPT-4 for API design research | Simon Willison’s TILs

Thoughts on AI safety in this era of increasingly powerful open source LLMs

Running Python micro-benchmarks using the ChatGPT Code Interpreter alpha

The Changelog podcast: LLMs break the internet

We need to tell people ChatGPT will lie to them, not debate linguistics

Replacing my best friends with an LLM trained on 500,000 group chat messages

The AI singularity is here | InfoWorld

Building LLM applications for production

Database “sharding” came from UO? – Raph’s Website

LMQL: Programming Large Language Models

mckaywrigley/prompts: My favorite AI prompts.

haotian-liu/LLaVA: Large Language-and-Vision Assistant built towards multimodal GPT-4 level capabilities.

xtekky/chatgpt-clone: ChatGPT interface with better UI + running on free gpt api’s

xtekky/gpt4free: Free gpt4 / gpt3.5 access through several reverse engineered api’s (poe.com, phind.com, t3nsor.com etc…)

ChatGPT-Dan-Jailbreak.md

StampyAI/stampy-ui: AI Safety Q&A

200 Concrete Problems In Interpretability Spreadsheet - Google Sheets

oobabooga/text-generation-webui: A gradio web UI for running Large Language Models like LLaMA, llama.cpp, GPT-J, Pythia, OPT, and GALACTICA.

hey-pal/toolkit-ai: AI-agents that automatically generate and use Langchain Tools and ChatGPT plugins

Toolkit - Create and discover AI plugins

The best way to build web apps without code | Bubble

teknium1/character-cards: A collection of character cards for use in AI Roleplaying

Prompt Engineering | Lil’Log

GPT-3 token encoder and decoder / Simon Willison | Observable

GPT-4 Week 4. The rise of Agents and the beginning of the Simulation era : ChatGPT

GPT-4 Week 5. Open Source is coming + Music industry in shambles - Nofil’s Weekly Breakdown : ChatGPT

(7) Eliezer Yudkowsky on Twitter: “Some simple things that could be done to reform Earth science from its present disaster state: - Discard the idea of ‘p-values’ and ‘statistically significant’ data. Report likelihood functions and as much raw data as possible; have an epistemology in which different effect…” / Twitter

[2302.00093] Large Language Models Can Be Easily Distracted by Irrelevant Context

(7) AK on Twitter: “AI music is taking off, here are some companies working in the space” / Twitter

templeofninpo/templeofninpo.github.io

StoicAI v12: stopped trying to be fancy, made it fancier

JushBJJ/Mr.-Ranedeer-AI-Tutor: A GPT-4 AI Tutor Prompt for customizable personalized learning experiences.

oneil512/INSIGHT: INSIGHT is an autonomous AI that can do medical research!

Activeloop | Deep Lake | Data Lake for Deep Learning

kroll-software/babyagi4all: BabyAGI to run with GPT4All

miurla/babyagi-ui: BabyAGI UI is designed to make it easier to run and develop with babyagi in a web app, like a ChatGPT.

My account | Forefront Chat

Pi, your personal AI

qdrant/qdrant: Qdrant - Vector Database for the next generation of AI applications. Also available in the cloud https://cloud.qdrant.io/

Vector Search Database | Qdrant Cloud

cozodb/cozo: A transactional, relational-graph-vector database that uses Datalog for query. The hippocampus for AI!

CozoDB: embedded Datalog, performant graphs

vespa-engine/vespa: The open big data serving engine. https://vespa.ai

Vespa - the big data serving engine

Welcome | Weaviate - vector database

mosaicml/llm-foundry

replit/ReplitLM: Inference code and configs for the ReplitLM model family

replit/replit-code-v1-3b · Hugging Face

mosaicml/examples: Fast and flexible reference benchmarks

openlm-research/open_llama

BigCode - Open and responsible development of LLMs for code

bigcode-project/Megatron-LM: Ongoing research training transformer models at scale

bigcode/starcoder · Hugging Face

bigcode/ta-prompt · Datasets at Hugging Face

Adds a sequence of numbers.

nbardy/SuperPrompt: Mutimodal LLM Lisp

Releasing 3B and 7B RedPajama-INCITE family of models including base, instruction-tuned & chat models — TOGETHER

RedPajama-INCITE-3B, an LLM for everyone — TOGETHER

togethercomputer/redpajama.cpp: Extend the original llama.cpp repo to support redpajama model.

CarperAI/stable-vicuna-13b-delta · Hugging Face

go-skynet/LocalAI: Self-hosted, community-driven, local OpenAI-compatible API. Drop-in replacement for OpenAI running LLMs on consumer-grade hardware. LocalAI is a RESTful API to run ggml compatible models: llama.cpp, alpaca.cpp, gpt4all.cpp, rwkv.cpp, whisper.cpp, vicuna, koala, gpt4all-j, cerebras and many others!

Available models in Generative AI Studio | Vertex AI | Google Cloud

Vertex AI – Google Play Android… – Google Cloud console

Dante | Build an AI chatbot trained on your data

mlc-ai/mlc-llm: Enable everyone to develop, optimize and deploy AI models natively on everyone’s devices.

Introducing speech-to-text, text-to-speech, and more for 1,100+ languages

whisper.cpp : WASM example

keon/awesome-nlp: A curated list of resources dedicated to Natural Language Processing (NLP)

brianspiering/awesome-dl4nlp: A curated list of awesome Deep Learning (DL) for Natural Language Processing (NLP) resources

zhengzangw/awesome-huge-models: A collection of AWESOME things about HUGE AI models.

kyaiooiayk/Awesome-LLM-Large-Language-Models-Notes: What can I do with a LLM model?

llama/MODEL_CARD.md at main · facebookresearch/llama