gpt-3#effective-prompt-programming
Theoretical Limitations of Self-Attention in Neural Sequence Models
Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention
I Think ‘GPT-3 Can’t Do Parity Checking’ Isn’t Quite Right. It Can Clearly Pattern Match the Algorithm, Almost Perfectly. It’s Just a Little Mistake Prone. Here, I Invented a Syntax for Having It Evaluate Parity on Each Pair of Digits. It...almost Gets It Right.
Inspired by an AI Dungeon Example Where Math Is Discussed in Simple Language, I Seem to Be Having Decent Results Here. I Had To... Not Just Say What Parity IS but HOW to Calculate It (‘Count the Number of 1s’) and Then It Sort of Walks Itself through Decently. Tho Kinda Confused
Gpt-Scrolls/scrolls/rephrase/conceptual-Blending.txt at Master · Maraoz/gpt-Scrolls
Experimenting With the Ideas of Linguistic Relativity, Specifically the Weak Version of the Sapir–Whorf Hypothesis. #GPT3 Was Able to Generate New Concepts With Seemingly Relevant Unique Spelling.
Asking GPT-3 to Draw ASCII Images Produces the Same Drawings Frequently
HTLM: Hyper-Text Pre-Training and Prompting of Language Models
FineWeb: Decanting the Web for the Finest Text Data at Scale
EleutherAI/gpt-Neo: An Implementation of Model Parallel GPT-2 and GPT-3-Style Models Using the Mesh-Tensorflow Library.
The Pile: An 800GB Dataset of Diverse Text for Language Modeling
Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data
Two tiny terriers chase very large bear out of California home
https://www.lesswrong.com/posts/L5JSMZQvkBAx9MD5A/to-what-extent-is-gpt-3-capable-of-reasoning#eq6FTwG2yWuBdPofs
Hydrochloric Acid Poisoning: MEDLINEPlus Medical Encyclopedia
Tests Show That the Popular AI Still Has a Poor Grasp of Reality.
Experiments testing GPT-3’s ability at commonsense reasoning: results
A Robot Wrote This Entire Article. Are You Scared Yet, Human? We Asked GPT-3, OpenAI’s Powerful New Language Generator, to Write an Essay for Us from Scratch. The Assignment? To Convince Us Robots Come in Peace | For More about GPT-3 and How This Essay Was Written and Edited, Please Read Our Editor’s Note Below
Q. Why Doesn‘t the API Seem to Have Knowledge about Recent Events? A. The Models’ Training Data That Cuts off in October 2019, so They May Not Have Knowledge of Current Events. We Plan to Add More Continuous Training in the Future.
Ikreymer/cdx-Index-Client: A Command-Line Tool for Using CommonCrawl Index API
Table 2.2: Datasets Used to Train GPT-3. ‘Weight in Training Mix’ Refers to the Fraction of Examples during Training That Are Drawn from a given Dataset, Which We Intentionally Do Not Make Proportional to the Size of the Dataset. As a Result, When We Train for 300 Billion Tokens, Some Datasets Are Seen up to 3.4 times during Training While Other Datasets Are Seen Less Than Once
The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence
https://medium.com/@ElementalCognition/why-does-ai-get-so-confused-by-language-f5f64a9ef6cc
XLNet: Generalized Autoregressive Pretraining for Language Understanding
It’s All about the Prelude Before the Conversation. You Need to Tell It What the AI Is and Is Not Capable. It’s Not Trying to Be Right, It’s Trying to Complete What It Thinks the AI Would Do ツ
GPT-3 Knows Both the Correct and the (plausible) Incorrect Answer to a Question.
GPT-3 (AI Dungeon 2) Is Also Capable of Formulating Some Really Bad Medical Advice. Although so Far I Only Managed to Make It Lie to My Only If It‘s Accompanied by a True Answer. It Doesn’t Want to Lie When It’s the Only Answer It Must to Give. But It’s Capable of Formulating Lies.
GPT-3 Gives Some Interesting True and False Answers to Some Questions. But It’s Important to Note That It Gives opposite Answers Just As Often, I Cheery Picked the Most ‘Sensational’ Ones. Usually It Said the opposite Thing, and It Also Role-Plays Sometimes (eg. As a Spy)
#gpt3 Has Some Reasonably Impressive Ability Not Only to Detect Nonsense, but to Explain Why Something Is Nonsensical:
Given That #gpt3 Obviously Has a Very Detailed Model of Language and Grammar, I Was Curious to See If It Could Both Correct Grammatical Errors and Explain the Corrections. The Answer Is ‘Yes’, Although It Took More Retries Than I Thought It Would for Explanations:
I found that getting GPT-3 to add its own "internal monologue" in parentheses to be a helpful strategy…
Really like the Idea of Getting GPT-3 to Introspect on Character’s Internal States (priming in Bold)
AI Dungeon Players Can Now Translate Their Stories into Emojis by Just Clicking a Button.
Artificial Neural Networks Today Are Not Conscious, according to Douglas Hofstadter
Gödel, Escher, Bach author Douglas Hofstadter on the state of AI today § What about AI terrifies you?
Chris Froome, First Man to Cycle through the Eurotunnel Crossing the English Channel
Sudanese Man Who Walked through Channel Tunnel Granted UK Asylum
Verbal Probability Expressions In National Intelligence Estimates: A Comprehensive Analysis Of Trends From The Fifties Through Post-9/11
CTRL: A Conditional Transformer Language Model For Controllable Generation
Time Vectors: Time is Encoded in the Weights of Finetuned Language Models
This Is a Python Script As Described in XKCD #1263: ‘Reassuring’. It Generates Thousands of Reassuring Parables about Things Humans Are Better Than Computers at Every Second.
GPT-3’s Completion of the Chinese Room Argument from Searle’s Minds, Brains, and Programs (original Text Is in Bold):
Why Computers Don’t Need to Match Human Intelligence: With continuing advances in machine learning, it makes less and less sense to compare AI to the human mind
https://www.reddit.com/r/MachineLearning/comments/iaitpu/d_knowledge_discovery_with_gpt3/g1onprj/
Can GPT-3 produce new ideas? Partially automating Robin Hanson and others § If you never miss a plane…
https://www.reddit.com/r/boardgames/comments/sbkous/does_this_ai_generated_idea_for_a_board_game/
https://boardgamegeek.com/boardgame/37111/battlestar-galactica-the-board-game
https://boardgamegeek.com/boardgame/240980/blood-on-the-clocktower
https://boardgamegeek.com/boardgame/150376/dead-of-winter-a-crossroads-game
https://boardgamegeek.com/boardgame/1678/peloponnesian-war-431-404-bc
https://boardgamegeek.com/boardgame/24396/war-on-terror-the-boardgame
Max Tegmark on How a 'Put-Up-Or-Shut-Up' Resolution Led Him to Work on AI and Algorithmic News Selection
Wikipedia Bibliography:
