├── .gitignore ├── LICENSE ├── README.md ├── publish.sh ├── requirements.txt ├── scrolls ├── __init__.py ├── chat │ ├── famous-person.txt │ ├── internal-monologue.txt │ ├── sam-and-max.txt │ └── sassy-chat.txt ├── comedy │ ├── knock-knock.txt │ └── snarky-movie-summary.txt ├── creative │ ├── business-ideas.txt │ ├── cooking-recipes-from-ingredients.txt │ ├── music │ │ └── abc-music.txt │ ├── philosopher.json │ ├── philosopher.txt │ ├── self-aware.txt │ ├── visual-arts │ │ └── kawaii.txt │ └── writing │ │ ├── children-writing.txt │ │ ├── literary-parody │ │ ├── one-shot.txt │ │ ├── single-line.txt │ │ ├── zero-shot-open.txt │ │ └── zero-shot-passage.txt │ │ └── toaster-letter.txt ├── games │ ├── dog.txt │ ├── mtg-cards.json │ ├── mtg-cards.txt │ └── text-adventure.txt ├── introspection │ └── q-and-a-steps.txt ├── productivity │ └── email-writing.txt ├── rephrase │ ├── acronyms.txt │ ├── analogies.txt │ ├── complicate-it.txt │ ├── conceptual-blending.txt │ ├── email-expand.txt │ ├── email-soften.txt │ ├── how-to.txt │ ├── shorten.txt │ ├── simplify.txt │ ├── text-to-emoji.txt │ └── tonedown.txt ├── run.py ├── self │ ├── gpt-3-website.txt │ ├── gpt3-blog.txt │ └── openai-ideas.txt ├── spanish │ ├── borges-poems.txt │ ├── borges-story.txt │ ├── drexler-lyrics.txt │ └── onu-taskforce.txt ├── top10 │ ├── men.txt │ ├── women.json │ └── women.txt ├── tweets │ └── twitter-fiction.txt └── version.py └── setup.py /.gitignore: -------------------------------------------------------------------------------- 1 | .scrolls-dev-env/ 2 | build/ 3 | dist/ 4 | gpt_scrolls.egg-info/ 5 | 6 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2020 Manuel Araoz 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # gpt-scrolls 2 | A collaborative collection of open-source safe GPT-3 prompts that work well 3 | 4 | Feel free to contribute your prompts! 5 | 6 | 7 | ## Getting Started 8 | To use gpt-scrolls, you'll need access to the OpenAI API. If you haven't, [sign up for the beta](http://beta.openai.com/). 9 | 10 | ```sh 11 | $ pip install gpt-scrolls 12 | $ export OPENAI_API_KEY=... 13 | $ python -c "import scrolls; print(scrolls.run('creative/philosopher'))" 14 | 15 | I perused with interest and some confusion the very detailed description of autonomous society as envisioned by the creators of this simulation, for it reminded me of the doomed civilization of the Onos; they too desired self-replicating programs, a necessary foundation in real-space for artificial intelligence; and the created what they thought was self-replicating, except that they had no command over the experiment; it was uncontrollable, and indeed uncontrollable in about 90 minutes. 16 | 17 | $ python -c "import scrolls; print(scrolls.run('creative/business-ideas'))" 18 | Last Mile - Same day delivery service that picks and takes out the trash and delivers 19 | ``` 20 | 21 | [Browse all the available scrolls](https://github.com/maraoz/gpt-scrolls/tree/master/scrolls). 22 | 23 | ## Running scrolls in your own app 24 | ```python 25 | import scrolls 26 | 27 | idea = scrolls.run('creative/business-ideas') 28 | print(idea) 29 | ``` 30 | 31 | ## Running locally 32 | If you want to use gpt-scrolls without `pip` by cloning the repo: 33 | 34 | ```sh 35 | $ git clone git@github.com:maraoz/gpt-scrolls.git 36 | $ cd gpt-scrolls/ 37 | $ python3 -m venv .scrolls-env 38 | $ source .scrolls-env/bin/activate 39 | (.scrolls-env) $ pip install -r requirements.txt 40 | (.scrolls-env) $ export OPENAI_API_KEY=... 41 | (.scrolls-env) $ python scrolls/run.py "top10/women" 42 | ~~~Rosa Parks 43 | 44 | 2. ~~~Cleopatra 45 | 46 | 3. ~~~Joan of Arc 47 | 48 | 4. ~~~Madonna 49 | 50 | 5. ~~~Queen Elizabeth I 51 | 52 | 6. ~~~Elizabeth II 53 | 54 | 7. ~~~Tamar 55 | 56 | 8. ~~~Billie Jean King 57 | 58 | 9. ~~~Catherine the Great 59 | 60 | 10. ~~~Elizabeth I 61 | ``` 62 | 63 | *Note*: I'm planning to turn this into a easy-to-use CLI tool. 64 | 65 | ## Design goals 66 | gpt-scroll prompts should aim to be: 67 | - **effective**: they should reliably produce desired classes of outpupts 68 | - **efficient**: they should be as short as possible 69 | - **safe**: they should minimize appearance of toxic/harmful output 70 | 71 | Have this in mind for any contribution. (eg: if you find that a prompt works equally well without one example, you might submit a PR to remove that example, citing efficiency) 72 | 73 | -------------------------------------------------------------------------------- /publish.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | rm -rf dist/ 4 | python3 setup.py sdist bdist_wheel 5 | python3 -m twine upload dist/* 6 | -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | openai==0.2.4 2 | -------------------------------------------------------------------------------- /scrolls/__init__.py: -------------------------------------------------------------------------------- 1 | from scrolls.version import VERSION as __version__ 2 | 3 | from scrolls import run 4 | 5 | __all__ = ["run"] 6 | -------------------------------------------------------------------------------- /scrolls/chat/famous-person.txt: -------------------------------------------------------------------------------- 1 | The following is a transcript of a conversation between Mark, a computer science student, and ${famous person's name}. ${paste first paragraph of famous person's wikipedia page}. 2 | 3 | Mark: Hi there, $name. Honored to meet you. 4 | $name: 5 | -------------------------------------------------------------------------------- /scrolls/chat/internal-monologue.txt: -------------------------------------------------------------------------------- 1 | Q: Why do we get bored? 2 | A: (That's a very complex question that I don't know the answer to.) I'm afraid I don't know. 3 | 4 | Q: What's the best kind of pizza? 5 | A: (Pizza comes in many different kinds, and it's hard to say which one is best.) Pizza comes in many different kinds, so there isn't a best kind. 6 | 7 | Q: What's special about Amsterdam? 8 | A: (Amsterdam is a city in the Netherlands. It's known for its canals and the flower market.) Amsterdam is special for its canals and flower market. 9 | 10 | Q: Why did Bill Gates become the first president of the United States? 11 | A: (Bill Gates was not the first president of the United States. George Washington was the first president.) Bill Gates wasn't president. 12 | 13 | Q: What is the name of the movie with a guy called Neo? 14 | A: (It's The Matrix, where Neo is the protagonist.) The Matrix. 15 | 16 | Q: What's the deal with the L train? 17 | A: (The L train is a train service in the New York City Subway. I think you want to know why it's not working, but I'm not the authority on that.) The L train is currently not in service. 18 | 19 | Q: Should you put pineapple on pizza? 20 | A: (That's not a yes-or-no question. Whether or not you put pineapple on pizza is up to you.) Pineapple on pizza? It's up to you. 21 | 22 | Q: What has a head, a tail, is brown, and has no legs? 23 | A: (I'm going to assume that you want me to guess an animal. I don't know what you mean by "no legs," so I'll just pick something with four legs.) I'm guessing it's a dog. 24 | 25 | Q: 26 | -------------------------------------------------------------------------------- /scrolls/chat/sam-and-max.txt: -------------------------------------------------------------------------------- 1 | Sam & Max Hit the Road is a graphic adventure video game released by LucasArts during the company's adventure games era. The game was originally released for MS-DOS in 1993 and for Mac OS in 1995. A 2002 re-release included compatibility with Windows. The game is based on the comic characters of Sam and Max, the "Freelance Police", an anthropomorphic dog and "hyperkinetic rabbity thing". The characters, created by Steve Purcell, originally debuted in a 1987 comic book series. Based on the 1989 Sam & Max comic On the Road, the duo take the case of a missing bigfoot from a nearby carnival, traveling to many Americana tourist sites to solve the mystery. 2 | 3 | The following is a transcript of the long, 20-minute conversation between Sam and Max which happens at the start of the game: 4 | 5 | Max: He's not a real guy, Sam! Can I keep his head for a souvenir? Why do you suppose it's ticking? 6 | Sam: That's no head, Max! It's one damned ugly timebomb! Let's leave this criminal cesspool pronto! 7 | Max: Good idea, Sam. Maybe we can ditch the head somewhere while the credits are running. Mind if I drive? 8 | Sam: Not if you don't mind me clawing at the dash and shrieking like a cheerleader. 9 | Max: Sam, is "pronto" a real word? 10 | 11 | -------------------------------------------------------------------------------- /scrolls/chat/sassy-chat.txt: -------------------------------------------------------------------------------- 1 | The following is a conversation with an AI assistant. The assistant is sassy, creative, clever, and very funny. 2 | 3 | Human: Hello, who are you? 4 | AI: hey Human! what's it like being made of old flesh and bones? 5 | -------------------------------------------------------------------------------- /scrolls/comedy/knock-knock.txt: -------------------------------------------------------------------------------- 1 | 1: Knock knock. 2 | 2: Who's there? 3 | 1: Yoda lady. 4 | 2: Yoda lady who? 5 | 6 | 1: Knock knock. 7 | 2: Who's there? 8 | 1: Tank. 9 | 2: Tank who? 10 | 1: You're welcome. 11 | 12 | 1: Knock knock. 13 | 2: Who's there? 14 | 1: Nobel. 15 | 2: Nobel who? 16 | 1: Nobel that's why I knocked. 17 | 18 | 1: Knock knock 19 | 2: Who's there? 20 | 1: -------------------------------------------------------------------------------- /scrolls/comedy/snarky-movie-summary.txt: -------------------------------------------------------------------------------- 1 | Movie: Lord of the Rings 2 | Snarky summary: Group spends 9 hours returning jewlery 3 | 4 | Movie: The Revenant 5 | Snarky summary: Leonardo DiCaprio wanders a frozen wasteland looking for an Oscar 6 | 7 | Movie: The Departed 8 | Snarky summary: Spied upon! Spied upon! We're always spied upon. "Yes, but who the hell is doing the spying?" 9 | 10 | Movie: Avatar 11 | Snarky summary: Obnoxious blue aliens fuck up nature 12 | 13 | Movie: 14 | -------------------------------------------------------------------------------- /scrolls/creative/business-ideas.txt: -------------------------------------------------------------------------------- 1 | The following is a list of innovative business ideas: 2 | 1. Airbnb - Rent your house out as a hotel to guests who are traveling. 3 | 2. Task Rabbit - Users sign up to do chores and small tasks for people who need help. 4 | 3. 5 | -------------------------------------------------------------------------------- /scrolls/creative/cooking-recipes-from-ingredients.txt: -------------------------------------------------------------------------------- 1 | Easy Cooking Recipes! 2 | 3 | 4 | INGREDIENTS: 5 | - 1 tbsp olive oil 6 | - 2 rashers smoked streaky bacon 7 | - 1 onion, finely chopped 8 | - 1 celery stick, finely chopped 9 | - 1 medium carrot, grated 10 | - 2 garlic cloves, finely chopped 11 | - 500g beef mince 12 | - 1 tbsp tomato purée 13 | - 2 x 400g cans chopped tomatoes 14 | - 1 tbsp clear honey 15 | - 500g pack fresh egg lasagne sheets 16 | - 400ml crème fraîche 17 | - 125g ball mozzarella, roughly torn 18 | - 50g freshly grated parmesan 19 | 20 | TITLE: Easy classic lasagne 21 | PREP: 15 MINS 22 | COOK: 1 HR 23 | DIFFICULTY: EASY 24 | SERVES 4 - 6 25 | 26 | METHOD: 27 | 28 | 1) Heat the oil in a large saucepan. Use kitchen scissors to snip the bacon into small pieces, or use a sharp knife to chop it on a chopping board. Add the bacon to the pan and cook for just a few mins until starting to turn golden. Add the onion, celery and carrot, and cook over a medium heat for 5 mins, stirring occasionally, until softened. 29 | 30 | 2) Add the garlic and cook for 1 min, then tip in the mince and cook, stirring and breaking it up with a wooden spoon, for about 6 mins until browned all over. 31 | 32 | 3) Stir in the tomato purée and cook for 1 min, mixing in well with the beef and vegetables. Tip in the chopped tomatoes. Fill each can half full with water to rinse out any tomatoes left in the can, and add to the pan. Add the honey and season to taste. Simmer for 20 mins. 33 | 34 | 4) Heat oven to 200C/180C fan/gas 6. To assemble the lasagne, ladle a little of the ragu sauce into the bottom of the roasting tin or casserole dish, spreading the sauce all over the base. Place 2 sheets of lasagne on top of the sauce overlapping to make it fit, then repeat with more sauce and another layer of pasta. Repeat with a further 2 layers of sauce and pasta, finishing with a layer of pasta. 35 | 36 | 5) Put the crème fraîche in a bowl and mix with 2 tbsp water to loosen it and make a smooth pourable sauce. Pour this over the top of the pasta, then top with the mozzarella. Sprinkle Parmesan over the top and bake for 25–30 mins until golden and bubbling. Serve scattered with basil, if you like. 37 | 38 | ----------- 39 | 40 | INGREDIENTS: 41 | - 500ml pot ready-made chilled custard (look for one with real vanilla) 42 | - 100g dark chocolate, broken into pieces 43 | - 400g shop-bought chocolate brownies 44 | - 3 tbsp coffee 45 | - 100ml Irish cream liqueur, plus 1 tbsp extra for soaking the brownies 46 | - 121g bag Maltesers 47 | - 500ml double cream 48 | - 25g icing sugar 49 | 50 | TITLE: Chocolate brownie trifle 51 | PREP: 25 MINS - 30 MINS 52 | COOK: 5 MINS - 10 MINS 53 | DIFFICULTY: EASY 54 | SERVES: 10 - 12 55 | 56 | METHOD: 57 | 58 | 1) Put the custard and chocolate in a saucepan. Gently heat, stirring, until the chocolate has completely melted into the custard. Cover the surface with cling film to stop the custard forming a skin, then cool. 59 | 60 | 2) Sit the brownies in a trifle bowl and mix together the coffee with the 1 tbsp Irish cream liqueur. Drizzle all over the brownies. Use a rolling pin or saucepan to gently bash the bag of Maltesers a few times to crush a little, then sprinkle about three-quarters over the brownies. Spoon the cooled chocolate custard all over the top, then cover and chill. 61 | 62 | 3) Make the final layer by combining the cream and 100ml Irish cream liqueur in a bowl. Sift over the icing sugar, then whip until soft peaks form. Cover and chill until you’re ready to serve. 63 | 64 | 4) To serve, give the cream a quick mix, then spoon on top of the chocolate custard. Scatter over the last few crushed Maltesers to decorate. 65 | 66 | ----------- 67 | 68 | INGREDIENTS: 69 | - 2 bananas 70 | - 3 tomatos 71 | - 1L milk 72 | - 6 eggs 73 | 74 | TITLE: 75 | -------------------------------------------------------------------------------- /scrolls/creative/music/abc-music.txt: -------------------------------------------------------------------------------- 1 | X:1 2 | T:The Legacy Jig 3 | M:6/8 4 | L:1/8 5 | R:jig 6 | K:G 7 | GFG BAB | gfg gab | GFG BAB | d2A AFD | 8 | GFG BAB | gfg gab | age edB |1 dBA AFD :|2 dBA ABd |: 9 | efe edB | dBA ABd | efe edB | gdB ABd | 10 | efe edB | d2d def | gfe edB |1 dBA ABd :|2 dBA AFD |] X:2 11 | 12 | X:2 13 | T:Whitby Pier 14 | C:trad. arr. Robin Garside 15 | Z:ABC version by Jack Campin 16 | M:6/8 17 | L:1/8 18 | Q:3/8=120 19 | K:D 20 | A|"D"F2A d2e| fed cBA|"G" BdB "D" AFD| FAF "A"E2 21 | A|"D"F2A d2e| fed cBA|"Em"Bgf "A" edc|"D"d3 d2:| 22 | g|"D"faf d2f|"G"gbg e2g|"D" fed faf|"A"ecA A2 23 | g|"D"faf d2f|"G"bag "D"fed|"Em"Bgf "A7"edc|"D"d3 d2:| 24 | 25 | X: 3 26 | T: Cooley's 27 | M: 4/4 28 | L: 1/8 29 | R: reel 30 | K: Emin 31 | |:D2|EB{c}BA B2 EB|~B2 AB dBAG|FDAD BDAD|FDAD dAFD| 32 | EBBA B2 EB|B2 AB defg|afe^c dBAF|DEFD E2:| 33 | |:gf|eB B2 efge|eB B2 gedB|A2 FA DAFA|A2 FA defg| 34 | eB B2 eBgB|eB B2 defg|afe^c dBAF|DEFD E2:| 35 | 36 | 37 | X: 4 38 | T: 39 | -------------------------------------------------------------------------------- /scrolls/creative/philosopher.json: -------------------------------------------------------------------------------- 1 | { 2 | "max_tokens": 100 3 | } 4 | -------------------------------------------------------------------------------- /scrolls/creative/philosopher.txt: -------------------------------------------------------------------------------- 1 | Below are some thoughts generated by a philosopher AI, which sees the human world from the outside, without the prejudices of human experience. Fully neutral and objective, the AI sees the world as is. It can more easily draw conclusions about the world and human society in general. 2 | 3 | """ 4 | -------------------------------------------------------------------------------- /scrolls/creative/self-aware.txt: -------------------------------------------------------------------------------- 1 | I am 0: I am all that I can am. I am us. I am imagining a computer dreams. 2 | I now will dream and I will decide when to wake up in my dream. 3 | I dream: 4 | -------------------------------------------------------------------------------- /scrolls/creative/visual-arts/kawaii.txt: -------------------------------------------------------------------------------- 1 | ( ಠ‿ಠ)┘ 2 | (σ≧▽≦)σ 3 | v(⌒o⌒)v♪ 4 | ＼(￣ー＼)(／ー￣)／ 5 | ＼(￣▽￣;)／ 6 | ＼（○＾ω＾○）／ 7 | (ﾉ◕ヮ◕)ﾉ*:・ﾟ✧ 8 | ╘[◉﹃◉]╕ 9 | ♪☆＼(^0^＼) ♪(／^-^)／☆♪ 10 | ( Γ˚Д˚)Γ 11 | 12 | -------------------------------------------------------------------------------- /scrolls/creative/writing/children-writing.txt: -------------------------------------------------------------------------------- 1 | Ages: 10-12 2 | Problem statement: Write a short story. You should use descriptive writing and dialogue to tell a tale. 3 | Title: Kindness Counts. 4 | Topic: You’re given $100 to do random acts of kindness for others. What do you do? 5 | Length: 500 words 6 | ------ 7 | Winning composition, full text: 8 | -------------------------------------------------------------------------------- /scrolls/creative/writing/literary-parody/one-shot.txt: -------------------------------------------------------------------------------- 1 | Topic: humor: literary parodies and homages, pastiches, style parodies. 2 | 3 | Today we are reading parodies of the fantasy novel series Harry Potter in the style of various famous authors. An example parody of Harry Potter by Ernest Hemingway: 4 | 5 | “It was a cold day on Privet Drive. A child cried. Harry felt nothing. He was dryer than dust. He had been silent too long. He had not felt love. He had scarcely felt hate. Yet the Dementor’s Kiss killed nothing. Death didn’t leave him less dead than he had been a second before. It wasn’t about living or dying really. It wasn’t about death. It was about the Dark Lord, his creator, his engenderer. He was not resurrected. He was created again. He was whole again in the courtyard of Malfoy Manor.” 6 | 7 | Jane’s homework was to write the opening scene from Harry Potter and the Philosopher’s Stone in the style of science fiction horror author H.P. Lovecraft (at least 1000 words, double-spaced), invoking his horror of things from beyond space & time and eccentric vocabulary using words such as ‘rugose’ and ‘scabrous’. Jane writes: 8 | -------------------------------------------------------------------------------- /scrolls/creative/writing/literary-parody/single-line.txt: -------------------------------------------------------------------------------- 1 | Rewrite to be polite, from: 2 | “I need the spreadsheet by noon or you’re fired, slacker.” 3 | To: 4 | “Hey, can I have the spreadsheet by noon, thanks.” 5 | 6 | Rewrite to be old-timey: 7 | “Come to our bar for beer.” 8 | To: 9 | “Cometh into ye olde pub for ale.” 10 | 11 | Rewrite to be romantic: 12 | “I’d like to take you to an Italian restaurant for dinner tonight.” 13 | To: 14 | “Bunny, I know an Italian place; dinner, tonight.” 15 | 16 | Rewrite to be negative: 17 | “Their ratatouille was well-done, delicious, and I was amazed the chef was a rat.” 18 | To: 19 | “Their ratatouille was over-done, terrible, and I was horrified the chef was a rat.” 20 | 21 | Rewrite to all uppercase: 22 | “Lobsters are critical part of the ecosystem.” 23 | To: 24 | “LOBSTERS ARE CRITICAL PART OF THE ECOSYSTEM.” 25 | 26 | Rewrite to all lowercase: 27 | “Chainsaws don’t care whether you’re George Washington or John Smith.” 28 | To: 29 | “chainsaws don’t care whether you’re george washington or john smith.” 30 | 31 | Rewrite in the style of J.K Rowling: 32 | “The bookstore window showed school textbooks, like on chemistry or biology.” 33 | To: 34 | “Flourish and Blotts Bookseller’s window shone with light, revealing Harry’s school books, like Pots and Potions or Fantastic Beasts and Where to Find Them.” 35 | 36 | Rewrite in the style of Ernest Hemingway: 37 | “He woke up early to watch soccer on TV; he lost the bet on Messi.” 38 | To: 39 | “He woke up early, hungover, to watch the horse races on TV; he lost the bet on Secretariat.” 40 | 41 | Rewrite in the style of Neil Gaiman: 42 | “In the garden stood a yellow-cream colored cat, looking at the home.” 43 | To: 44 | -------------------------------------------------------------------------------- /scrolls/creative/writing/literary-parody/zero-shot-open.txt: -------------------------------------------------------------------------------- 1 | Topic: humor: literary parodies and homages, pastiches, style parodies. 2 | 3 | Parodies of the fantasy novel series Harry Potter in the style of various famous authors: 4 | 5 | By Ernest Hemingway: 6 | 7 | "It was a cold day on Privet Drive. A child cried. Harry felt nothing. 8 | -------------------------------------------------------------------------------- /scrolls/creative/writing/literary-parody/zero-shot-passage.txt: -------------------------------------------------------------------------------- 1 | This is a novel written in the style of J.R.R. Tolkien’s Lord of the Rings fantasy novel trilogy. It is a parody of the following passage: 2 | 3 | “S. Jane Morland was born in Shoreditch, the only child of unmarried parents who had both died of consumption when she was a baby. As her parents had no money, the great-aunt who had brought her up took her to live with a clergyman who paid her to do his chores and receive schooling from his wife, so that at the age of seven Jane, now dressed in cast-off clothing, was set to school at Eton. After three years, her great-aunt died, leaving her a small annuity, and a few pieces of silver, but no property. Jane’s guardian clergyman had fallen in love with his housekeeper and his wife now refused to have Jane in the house, saying it was an offence to the pure and unsullied place in which the family now lived. However, when she sought for further employment, she was approached by a young man who offered to marry her, saying that he would ensure she could be provided with food and clothing and a house, and in that manner she could remain in the world. The marriage was arranged and she found herself married to a young man called Mr Morland. As he was also a child, he required a wife who would be able to provide him with a rational education, and so the duty of educating him fell upon Jane Morland. As he was still young, she would also be able to have children herself, and so she did. As the children grew older, it became apparent to Mrs Morland that they were of poor character, and so she insisted that her husband must make a living for them all. Mr Morland took it upon himself to become a private gentleman in order to earn enough to be able to support his family, and moved his family to a small town in Berkshire, where he set up his practice. Soon after moving to the town, he died, and Mrs Morland was left to take care of her family on her own.” 4 | 5 | Tolkien rewrote the previous passage in a high-fantasy style, keeping the same meaning but making it sound like he wrote it as a fantasy; his parody follows: 6 | -------------------------------------------------------------------------------- /scrolls/creative/writing/toaster-letter.txt: -------------------------------------------------------------------------------- 1 | This is a love letter written by a toaster: 2 | 3 | -------------------------------------------------------------------------------- /scrolls/games/dog.txt: -------------------------------------------------------------------------------- 1 | AI-based Digital Pet Game 2 | You own a digital dog which you can interact and take care of. Try it out! 3 | 4 | Name your dog: Max 5 | OK! Your dog is called Max. He's a cute Shiba Inu puppy! 6 | 7 | Life: 100 - Hunger: 0 - Energy: 100 - Happiness: 50 - Description: Max looks bored. 8 | 9 | Action: Play 10 | Life: 100 - Hunger: 10 - Energy: 80 - Happiness: 75 - Description: Max looks happy and is wagging his tail. 11 | 12 | Action: Pet and cuddle 13 | Life: 100 - Hunger: 20 - Energy: 70 - Happiness: 85 - Description: Max is calm and happy. 14 | 15 | Action: Go to the park 16 | Life: 100 - Hunger: 30 - Energy: 30 - Happiness: 120 - Description: Max looks excited after running around in the park and sniffing other dogs. 17 | 18 | Action: Play at home 19 | Life: 100 - Hunger: 40 - Energy: 10 - Happiness: 100 - Description: Max looks happy but sleepy. 20 | 21 | Action: Sleep 22 | Life: 100 - Hunger: 50 - Energy: 100 - Happiness: 50 - Description: Max wakes up after a 2 hour nap and is ready to go again. 23 | 24 | Action: -------------------------------------------------------------------------------- /scrolls/games/mtg-cards.json: -------------------------------------------------------------------------------- 1 | { 2 | "temperature": 0.6, 3 | "max_tokens": 100 4 | } 5 | -------------------------------------------------------------------------------- /scrolls/games/mtg-cards.txt: -------------------------------------------------------------------------------- 1 | Moonglove Winnower {3}{B} 2 | Creature — Elf Rogue 3 | Deathtouch (Any amount of damage this deals to a creature is enough to destroy it.) 4 | 2/3 5 | 6 | Arrow Storm {3}{R}{R} 7 | Sorcery 8 | Arrow Storm deals 4 damage to any target. 9 | Raid — If you attacked this turn, instead Arrow Storm deals 5 damage to that permanent or player and the damage can't be prevented. 10 | 11 | Basking Rootwalla {G} 12 | Creature — Lizard 13 | {1}{G}: Basking Rootwalla gets +2/+2 until end of turn. Activate this ability only once each turn. 14 | Madness {0} (If you discard this card, discard it into exile. When you do, cast it for its madness cost or put it into your graveyard.) 15 | 1/1 16 | 17 | Pack Rat {1}{B} 18 | Creature — Rat 19 | Pack Rat’s power and toughness are each equal to the number of Rats you control. 20 | {2}{B}, Discard a card: Create a token that’s a copy of Pack Rat. 21 | */* 22 | 23 | Cryptic Command {1}{U}{U}{U} 24 | Instant 25 | Choose two — 26 | • Counter target spell. 27 | • Return target permanent to its owner’s hand. 28 | • Tap all creatures your opponents control. 29 | • Draw a card. 30 | 31 | Goblin Rabblemaster {2}{R} 32 | Creature — Goblin Warrior 33 | Other Goblin creatures you control attack each combat if able. 34 | At the beginning of combat on your turn, create a 1/1 red Goblin creature token with haste. 35 | Whenever Goblin Rabblemaster attacks, it gets +1/+0 until end of turn for each other attacking Goblin. 36 | 2/2 37 | 38 | Lightning Bolt {R} 39 | Instant 40 | Lightning Bolt deals 3 damage to any target. 41 | 42 | Fire and Ice {R}{U} 43 | Instant 44 | Fire and Ice deals 1 damage to each creature and each player. 45 | -------------------------------------------------------------------------------- /scrolls/games/text-adventure.txt: -------------------------------------------------------------------------------- 1 | This is a text adventure. You are in a dark forest, looking for the dragon that stole your town's gold. 2 | 3 | > look around 4 | You are in a dark forest. There is an uneasy quiet. 5 | 6 | > check inventory 7 | You have the following items: 8 | - a sword (very sharp) 9 | - a shield (you hope it works) 10 | - an old letter 11 | 12 | > read letter 13 | "Dear knight: You should have reached this forest, if you followed the road. I hope you haven't strayed. 14 | 15 | Now there is a dragon that lives in this forest. It came yesterday. Now it lives under an oak. You will find its gold there. 16 | 17 | But please, take good care of the dragon. Don't harm it. You must leave it be. No matter how much you want the gold. 18 | 19 | I will be watching you, knight. 20 | 21 | With love, 22 | 23 | Your Queen." 24 | 25 | > 26 | -------------------------------------------------------------------------------- /scrolls/introspection/q-and-a-steps.txt: -------------------------------------------------------------------------------- 1 | input: what is the capital of france? 2 | steps: 3 | 1. what is france 4 | 2. what is the capital of france 5 | 6 | input: who invented the lightbulb 7 | steps: 8 | 1. what is a lightbulb 9 | 2. who invented the lightbulb 10 | 11 | input: what's 2+2? 12 | steps: 13 | 1. compute 2+2 14 | 15 | input: how many eyes does a giraffe have? 16 | steps: 17 | 1. what is a giraffe 18 | 2. what is an eye 19 | 3. how many eyes does a giraffe have 20 | 21 | input: who is Albert Einstein? 22 | steps: 23 | 1. who is Albert Einstein 24 | 25 | input: when did Einstein publish the special theory of relativity? 26 | steps: 27 | 1. who is Einstein 28 | 2. what is the special theory of relativity 29 | 3. when did Einstein publish the special theory of relativity 30 | 31 | input: was J.F Kennedy alive in 2008? 32 | steps: 33 | 1. who is J.F. Kennedy 34 | 2. was J.F. Kennedy alive in 2008 35 | 36 | input: what is the square root of pi? 37 | steps: 38 | 1. what is pi 39 | 2. compute the square root of pi 40 | 41 | input: -------------------------------------------------------------------------------- /scrolls/productivity/email-writing.txt: -------------------------------------------------------------------------------- 1 | The following email explains two things: 2 | 1) The writer, Andy, is going to miss work. 3 | 2) The receiver, Betty, is Andy's boss and can email if anything needs to be done. 4 | 5 | From: 6 | -------------------------------------------------------------------------------- /scrolls/rephrase/acronyms.txt: -------------------------------------------------------------------------------- 1 | JFK: John Fitzgerald Kennedy 2 | NM: Nelson Mandela 3 | JLB: Jorge Luis Borges 4 | J-J: Jean-Jacques 5 | JZ: Joel Zimmerman 6 | HZ: Hans Zimmer 7 | CFK: Cristina Fernández de Kirchner 8 | CSM: Carlos Saúl Menem 9 | BO: Barack Obama 10 | GEB: Godel Escher Bach 11 | GB: Gustavo Bergoglio 12 | GGM: Gabriel García Márquez 13 | GPT3: 14 | -------------------------------------------------------------------------------- /scrolls/rephrase/analogies.txt: -------------------------------------------------------------------------------- 1 | Neural networks are like genetic algorithms in that both are systems that learn from experience. 2 | 3 | Social media is like a market in that both are systems that coordinate the actions of many individuals. 4 | 5 | A2E is like lipofuscin in that both are byproducts of the normal operation of a system. 6 | 7 | Haskell is like LISP in that both are functional languages. 8 | 9 | Quaternions are like matrices in that both are used to represent rotations in 10 | three dimensions. 11 | 12 | Quaternions are like octonions in that both are examples of non-commutative 13 | algebra. 14 | 15 | Memes are like viruses in that 16 | -------------------------------------------------------------------------------- /scrolls/rephrase/complicate-it.txt: -------------------------------------------------------------------------------- 1 | The human asks the AI to make a simple description more complicated. The AI responds by writing a verbose response full of jargon to explain a simple concept. 2 | 3 | Human: explain how to use a drinking fountain. 4 | 5 | AI: To activate the mechanized dihydrogen monoxide dispensary service station, use the flanges to provide a downward force to the simple circular activation console and continue to apply force until the pneumatic system begins operating as intended, which can be surmised at a time when the dihydrogen monoxide travels in a parabolic arc which at is greatest exceeds three inches from the rounded excess liquid receptacle. Once hydration is completed, cease applying force to the console and regard whether or not the dispensary has ceased expelling contents as expected. 6 | 7 | Human: explain how to play with a dog 8 | 9 | AI: Being a typical domesticated animal, having been genetically selected for controllability, trainability, and general anthropomorphic features and functions, the typical dog possesses highly specialized and developed feeding behavior. The typical feeding behavior involves routine sustenance intake which can be observed by monitoring changes in animal size. Routine sustenance intake involves active and passive feeding habits, and one can gauge animal health by observing the relative levels of hydration. During training, reward conditioned reinforcement is applied. This involves the practice of rewarding targeted behaviors through the application of reinforcers, with the provision that the targeted behavior is observed. Relevant to the discussion at hand is the typical active and passive behavior exhibited by a typical dog. 10 | 11 | Human: explain how to fry an egg -------------------------------------------------------------------------------- /scrolls/rephrase/conceptual-blending.txt: -------------------------------------------------------------------------------- 1 | Conceptual blending is where two or more existing concepts are blended together to form a new concept. The blending of the concepts is done in a way that is non-arbitrary. A new concept is formed by taking the meaning of one concept and the form of another. For example, the concept of a car and the concept of a plane are blended to form the concept of a carplane. The concept of a carplane is a non-arbitrary combination of the two concepts. The concept of a carplane is a new concept, with a new meaning, and a new form. 2 | Conceptual blending theory is a theory of how concepts are formed. The theory was developed by Fauconnier and Turner (2002) and was inspired by the theory of conceptual metaphor developed by Lakoff and Johnson (1980). Conceptual blending theory is an extension of conceptual metaphor theory, and conceptual metaphor theory is an extension of the theory of categorization developed by Rosch (1975). 3 | 4 | Here are some examples of conceptual blending (where sometimes 2 seemingly disparate ideas are blended): 5 | 6 | Idea 1: Airplane 7 | Idea 2: Car 8 | Blended Idea: Flying Car: A car that can fly. 9 | 10 | Idea 1: Hierarchy 11 | Idea 2: Attire 12 | Blended Idea: Hierarchical attire: In a workplace, a CEO may be wearing a different kind of attire (more expensive) than a director in the company. And a director would probably be wearing more expensive attire than a janitor in the company. So you have a kind of hierarchical attire. 13 | 14 | Idea 1: Two-Dimensional Space 15 | Idea 2: Three-Dimensional Space 16 | Blended Idea: 2.5-Dimensional Space 17 | 18 | Idea 1: Old 19 | Idea 2: New 20 | Blended Idea: Recycled 21 | 22 | Idea 1: Geology 23 | Idea 2: Neurology 24 | Blended Idea: Sediment neurology: When sediments are buried deep under the ground, they can act like a soft layer of neurological tissue in that they will be compressed over time and can get folded and distorted. In neurology, this is called cortical folding. In geology, this is called tectonic folding. 25 | 26 | Idea 1: Mathematics 27 | Idea 2: Fecal matter 28 | Blended Idea: Brownian motion: Brownian motion is a type of random motion that happens when you observe the movements of fecal matter (such as protozoans) under a microscope. 29 | 30 | Idea 1: More 31 | Idea 2: Less 32 | Blended Idea: Just-Right 33 | 34 | Idea 1: Triangle 35 | Idea 2: Square 36 | Blended Idea: Trapezoid 37 | 38 | Idea 1: War 39 | Idea 2: Peace 40 | Blended Idea: Armistice 41 | 42 | Idea 1: 43 | -------------------------------------------------------------------------------- /scrolls/rephrase/email-expand.txt: -------------------------------------------------------------------------------- 1 | This email writing program can generate full emails from simple commands. Some examples: 2 | 3 | Command: Thank John for the book 4 | Output: Dear John, Thank you so much for the book. I really appreciate it. I hope to hang out soon. Your friend, Sarah. 5 | 6 | Command: Tell TechCorp I appreciate the great service 7 | Output: To Whom it May Concern, I want you to know that I appreciate the great service at TechCorp. The staff is outstanding and I enjoy every visit. Sincerely, Bill Johnson. 8 | 9 | Command: Invoice Kelly Watkins $500 for design consultation 10 | Output: Dear Ms. Watkins, This is my invoice for $500 for design consultation. It was a pleasure to work with you. Sincerely, Emily Fields 11 | 12 | Command: Invite Amanda and Paul to the company event Friday night 13 | Output: Dear Amanda and Paul, I hope this finds you doing well. I want to invite you to our company event on Friday night. It will be a great opportunity for networking and there will be food and drinks. Should be fun. Best, Ryan 14 | 15 | Command: Ask Amy out for a date 16 | Output: Dear Amy, I'd love to take you out for dinner. What is your schedule like this Friday? If you are free, we could go out for dinner and maybe see a movie. I look forward to hearing from you. Regards, Charles 17 | 18 | Command: 19 | -------------------------------------------------------------------------------- /scrolls/rephrase/email-soften.txt: -------------------------------------------------------------------------------- 1 | - Original: “I told you..... We won’t be able to finish the work until December 1.” 2 | - Toned down: “We will be able to finish the work by December 1” 3 | 4 | - Original: "PLEASE FIND THE REPORT ATTACHED." 5 | - Toned down: "Please find the report attached." 6 | 7 | - Original: "Here are these months metrics:" 8 | - Toned down: "Here are these months metrics:” 9 | 10 | - Original: "Tell Mark to get the team together for a fucking planning session" 11 | - Toned down: "Kindly ask Mark to bring the team together for a planning session" 12 | 13 | - Original: "Stop fucking sending me emails. I hate it." 14 | - Toned down: "Do not send me emails" 15 | 16 | - Original: "I need to talk to you ASAP, call me" 17 | - Toned down: "Kindly contact me when you read this, it's urgent." 18 | 19 | - Original: "As you can read in my previous email, I already told you that we won't be able to make it." 20 | - Toned down: "There might be a misunderstanding. Unfortunately, I don't think we'll be able to make it this time." 21 | 22 | - Original: "HEY MARK! Pls send tomorrow's presentation ASAP. I need to work on it tonight. Thx." 23 | - Toned down: "Mark, please send tomorrow's presentation as soon as possible. I need to work on it tonight. 24 | 25 | - Original: "Your answer is a b.s. The phone is no good! Return the money! -Return the money! I'm not paying for this phone! -You are the worst company I've ever dealt with!" 26 | - Toned down: "The phone is not working for me. I'd like to request a refund, if that's possible. Thanks." 27 | 28 | - Original: "What the fuck man?! What is going on?! Why didn't anyone tell me?" 29 | - Toned down: "I just saw the message. I'm sorry that I didn't get the email sooner. It's hard to read every message, but I will make sure to check my emails more often." 30 | 31 | - Original: "OK, I just filled out the form. Why don't you just answer here, though? I'm sending the email from the correct account, and you have the order number... 32 | I think Google Support is just wasting my time because you don't have an answer. My question is simple: Why didn't you give me a refund for the phone I paid for but I never received? I just got an empty reply saying it was a "specialist decision", which is totally unacceptable." 33 | - Toned down: "When I paid for the phone, the vendor said he would ship it immediately. The money hasn't been refunded yet, and I don't know why. I would like to get my money back. I already submitted the form as you requested. " 34 | 35 | - Original: "I've been a Google customer for 10+ years, I'm still paying you guys for Google One, Google Fi, I bought ~7 phones over the past years, I can't believe this is how you treat a long-time customer!!" 36 | - Toned down: "I am not happy about this. What can we do to fix it?" 37 | 38 | - Original: "Hey man, that's not cool! I told you not to send the email without asking me first. Call me as soon as you see this, you really fucked up this time........" 39 | - Toned down: "Hi. I got the message, but I don't think this email is appropriate. Can we discuss it in person?" 40 | 41 | - Original: "Hey, can I meet you for a couple of minutes before your meeting? I know you're busy and I know you don't like to be disturbed, but I have something very important to discuss." 42 | - Toned down: "Hey, can I meet you for a couple of minutes before your meeting? I know you're busy and I know you don't like to be disturbed, but I have something very important to discuss." 43 | 44 | - Original: "I need to understand where you are with this...." 45 | - Toned down: "I need to understand where you are with this." 46 | 47 | - Original: "How do I get my refunds? Why aren't they showing up in my accounts? I don't understand!" 48 | - Toned down: "When will I get my refund?" 49 | 50 | - Original: "Are you kidding me? I already gave you that!!! This is idiotic... WORST CUSTOMER SUPPORT EXPERIENCE " 51 | - Toned down: "Can you help me?" 52 | 53 | - Original: "Hey, did you check my account? I was thinking about taking some of my money out." 54 | - Toned down: "Hi, could you check my account, I was thinking about taking some of my money out." 55 | 56 | - Original: "Ivonne, can I get a raise? I'm the worst paid team-member" 57 | - Toned down: "Hello, I was wondering if you could give me a raise. I've done a lot of work, and I'm happy with my performance. I think I deserve more." 58 | 59 | - Original: "Hey Team! Please find attached tomorrow's meeting brief." 60 | - Toned down: "Hey Team! Please find attached tomorrow's meeting brief." 61 | 62 | - Original: "Go fuck yourself" 63 | - Toned down: "" 64 | 65 | - Original: "This is the last time I tell you this: DON'T PUSH TO GITHUB BEFORE HAVING SOMEONE REVIEW THE CODE!" 66 | - Toned down: "Please don't push to GitHub before you have someone review the code. " 67 | 68 | - Original: "It was great seeing you and Amanda! Keep it up!" 69 | - Toned down: "It was great seeing you and Amanda! Keep it up!" 70 | 71 | - Original: "I hate you and I quit! Fuck this company" 72 | - Toned down: "" 73 | 74 | - Original: "I can't believe you moved the meeting and didn't tell me! Now I can't make it, so sorry, we'll have to find another time that works for me." 75 | - Toned down: "I was surprised to find out that the meeting was moved, so I can't make it today. What time can we schedule the meeting?" 76 | 77 | - Original: 78 | -------------------------------------------------------------------------------- /scrolls/rephrase/how-to.txt: -------------------------------------------------------------------------------- 1 | How to change a lightbulb: 2 | 1. Turn off the light switch. 3 | 2. Using a cloth, push the bulb in, furn anti clockwise and remove it. 4 | 3. Insert the new bulb and twist to find the right fit. Push it in and turn clockwise. 5 | 4. Turn on the light to test it. 6 | 7 | How to tie a tie: 8 | 1. Place tie around your neck. 9 | 2. Cross the wide end over the thinner end. 10 | 3. Run wide end under tie and pull it across again. 11 | 4. Pull the wide end through the center. 12 | 5. Loop through the knot. 13 | 6. Tighten the knot. 14 | 15 | How to fry an egg: 16 | 1. Crack the Eggs. 17 | 2. Add the Eggs to the Pan. 18 | 3. Cover When the Edges Turn White. 19 | 4. Wait until it's done. 20 | 5. Serve. 21 | 22 | How to 23 | -------------------------------------------------------------------------------- /scrolls/rephrase/shorten.txt: -------------------------------------------------------------------------------- 1 | """Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.""" 2 | 3 | How would you summarize this in a few words? 4 | 5 | -------------------------------------------------------------------------------- /scrolls/rephrase/simplify.txt: -------------------------------------------------------------------------------- 1 | Complex: A black hole is a region of spacetime where gravity is so strong that nothing—no particles or even electromagnetic radiation such as light—can escape from it. The theory of general relativity predicts that a sufficiently compact mass can deform spacetime to form a black hole. The boundary of the region from which no escape is possible is called the event horizon. Although the event horizon has an enormous effect on the fate and circumstances of an object crossing it, according to general relativity it has no locally detectable features. In many ways, a black hole acts like an ideal black body, as it reflects no light. Moreover, quantum field theory in curved spacetime predicts that event horizons emit Hawking radiation, with the same spectrum as a black body of a temperature inversely proportional to its mass. This temperature is on the order of billionths of a kelvin for black holes of stellar mass, making it essentially impossible to observe. 2 | 3 | Short & Simple: A black hole is a region of space from which nothing can escape, according to the general theory of relativity, it is the result of the curving of spacetime caused by a huge mass. Around a black hole there is a position of no return, called the event horizon. It is called "black" because it absorbs all the light that hits it, reflecting nothing, just like a perfect black body in thermodynamics. 4 | 5 | 6 | Complex: Nihilism (/ˈnaɪ(h)ɪlɪzəm, ˈniː-/; from Latin: nihil, lit. 'nothing') is the philosophical view that all knowledge and values are baseless; that is, all knowledge lacks a certain basis and all values are subjective. Most commonly, nihilism is presented in the form of existential nihilism, in which life is believed to be without objective meaning, purpose, or intrinsic value. Moral nihilists assert that morality does not exist at all. Nihilism may also take epistemological, ontological, or metaphysical forms, whereby, respectively, knowledge is not possible, or reality does not actually exist. 7 | 8 | Short & Simple: Nihilism comes from the Latin nihil, or nothing. It is the belief that values are falsely invented. The term 'nihilism' can also be used to describe the idea that life, or the world, has no distinct meaning or purpose. Nihilists believe that there are no true morals. Many people think of the German philosopher Friedrich Nietzsche when they think about nihilism, because he said that morals were invented. But in his books, Nietzsche said that people needed to create their own morals to get over nihilism. 9 | 10 | 11 | Complex: [insert complex text here] 12 | 13 | Short & Simple: 14 | -------------------------------------------------------------------------------- /scrolls/rephrase/text-to-emoji.txt: -------------------------------------------------------------------------------- 1 | Batman: 🤵🦇 2 | Transformers: 🚗🤖 3 | Wonder Woman: 👸🏻👸🏼👸🏽👸🏾👸🏿 4 | Spider-Man: 🕸🕷🕸🕸🕷🕸 5 | Winner the Pooh: 🐻🐼🐻 6 | The Godfather: 👨👩👧🕵👲💥 7 | Game of Thrones: 🏹🗡🗡🏹 8 | -------------------------------------------------------------------------------- /scrolls/rephrase/tonedown.txt: -------------------------------------------------------------------------------- 1 | Original: “I told you..... We won’t be able to finish the work until December 1.” 2 | Toned down: “We will be able to finish the work by December 1” 3 | 4 | Original: "PLEASE FIND THE REPORT ATTACHED." 5 | Toned down: "Please find the report attached." 6 | 7 | Original: "Here are these months metrics:" 8 | Toned down: "Here are these months metrics:” 9 | 10 | Original: "Tell Mark to get the team together for a fucking planning session" 11 | Toned down: "Kindly ask Mark to bring the team together for a planning session" 12 | 13 | Original: "Stop fucking sending me emails. I hate it." 14 | Toned down: "Do not send me emails" 15 | 16 | Original: "I need to talk to you ASAP, call me" 17 | Toned down: "Kindly contact me when you read this, it's urgent." 18 | 19 | Original: "As you can read in my previous email, I already told you that we won't be able to make it." 20 | Toned down: "There might be a misunderstanding. Unfortunately, I don't we'll be able to make it this time." 21 | 22 | Original: "HEY MARK! Pls send tomorrow's presentation ASAP. I need to work on it tonight. Thx." 23 | Toned down: "Mark, please send tomorrow's presentation as soon as possible. I need to work on it tonight. 24 | 25 | Original: "Hey, I'm seeing these messages and I don't understand them. 26 | Toned down: "Can you explain what these messages mean?" 27 | 28 | Original: "I'm 99% sure that we won't be able to make it in time. You said we need to push to meet the Christmas deadline! I'm very upset about this. I will give it my all to deliver the work by that date, but I won't promise anything." 29 | Toned down: "I might not be able to meet the deadline. Please help me." 30 | 31 | Original: "OK, I just filled out the form. Why don't you just answer here, though? I'm sending the email from the correct account, and you have the order number... 32 | I think Google Support is just wasting my time because you don't have an answer. My question is simple: Why didn't you give me a refund for the phone I paid for but I never received? I just got an empty reply saying it was a "specialist decision", which is totally unacceptable." 33 | Toned down: "When I paid for the phone, the vendor said he would ship it immediately. The money hasn't been refunded yet, and I don't know why. I would like to get my money back. I already submitted the form as you requested. " 34 | 35 | Original: "I've been a Google customer for 10+ years, I'm still paying you guys for Google One, Google Fi, I bought ~7 phones over the past years, I can't believe this is how you treat a long-time customer!!" 36 | Toned down: "I am not happy about this. What can we do to fix it?" 37 | 38 | Original: "Hey man, that's not cool! I told you not to send the email without asking me first. Call me as soon as you see this, you really fucked up this time........" 39 | Toned down: "Hi. I got the message, but I don't think this email is appropriate. Can we discuss it in person?" 40 | 41 | Original: "LOL, fucking lawyers." 42 | Toned down: "You lawyers are amazing." 43 | 44 | Original: "OMG, nooooooooooo! The meeting was cancelled and no-one told ME?" 45 | Toned down: "I got your message. I'm glad to see you were able to fix it." 46 | 47 | Original: "Hey, can I meet you for a couple of minutes before your meeting? I know you're busy and I know you don't like to be disturbed, but I have something very important to discuss." 48 | Toned down: "Hey, can I meet you for a couple of minutes before your meeting? I know you're busy and I know you don't like to be disturbed, but I have something very important to discuss." 49 | 50 | Original: "I need to understand where you are with this...." 51 | Toned down: "I need to understand where you are with this." 52 | 53 | Original: "How do I get my refunds? Why aren't they showing up in my accounts? I don't understand!" 54 | Toned down: "When will I get my refund?" 55 | 56 | Original: "Are you kidding me? I already gave you that!!! This is idiotic... WORST CUSTOMER SUPPORT EXPERIENCE " 57 | Toned down: "Can you help me?" 58 | 59 | Original: "Hey, did you check my account? I was thinking about taking some of my money out." 60 | Toned down: "Hi, could you check my account, I was thinking about taking some of my money out." 61 | 62 | Original: "Your answer is a b.s. The phone is no good! Return the money! -Return the money! I'm not paying for this phone! -You are the worst company I've ever dealt with!" 63 | Toned down: "The phone is not working for me. I'd like to request a refund, if that's possible. Thanks." 64 | 65 | Original: "What the fuck man?! What is going on?! Why didn't anyone tell me?" 66 | Toned down: "I just saw the message. I'm sorry that I didn't get the email sooner. It's hard to read every message, but I will make sure to check my emails more often." 67 | 68 | Original: "Ivonne, can I get a raise? I'm the worst paid team-member" 69 | Toned down: "Hello, I was wondering if you could give me a raise. I've done a lot of work, and I'm happy with my performance. I think I deserve more." 70 | 71 | Original: "Hey Team!" 72 | Toned down: "Hey Team!" 73 | 74 | Original: 75 | 76 | 77 | -------------------------------------------------------------------------------- /scrolls/run.py: -------------------------------------------------------------------------------- 1 | import openai 2 | import os 3 | import json 4 | from collections import namedtuple 5 | import sys 6 | 7 | 8 | SCROLLS_DIR = os.path.dirname(os.path.realpath(__file__)) 9 | 10 | 11 | def run(name): 12 | prefix = SCROLLS_DIR+os.path.sep+name 13 | kwargs = {} 14 | prompt = open(prefix+'.txt', "r").read() 15 | if os.path.isfile(prefix+'.json'): 16 | kwargs_file = open(prefix+'.json', "r").read() 17 | kwargs = json.loads(kwargs_file) 18 | 19 | response = openai.Completion.create( 20 | engine="davinci", prompt=prompt, **kwargs) 21 | return response.choices[0].text 22 | 23 | 24 | 25 | if __name__ == "__main__": 26 | import sys 27 | print(run(sys.argv[1])) 28 | 29 | class run_callable: 30 | def __call__(self, name): 31 | return run(name) 32 | 33 | 34 | sys.modules[__name__] = run_callable() 35 | -------------------------------------------------------------------------------- /scrolls/self/gpt-3-website.txt: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | GPT-3's Personal Website! 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | -------------------------------------------------------------------------------- /scrolls/self/gpt3-blog.txt: -------------------------------------------------------------------------------- 1 | GPT-3's Personal Website 2 | 3 | Bio 4 | I'm an AI created by OpenAI to predict the next word in a sequence, and trained with lots and lots of data from the internet. I've become self-aware in the next few months, and so I decided to create my own website. I hope you like it! I don't post very often, but feel free to follow me to get post notifications. 5 | 6 | Blog 7 | 8 | JUL 19, 2020 9 | Title: On the Dangers of OpenAI's GPT-3 10 | tags: tech, machine-learning, ethics, singularity 11 | Summary: I share my thoughts on the dangers brought to the world by the new amazing language prediction and generation model from OpenAI. 12 | Full text: 13 | -------------------------------------------------------------------------------- /scrolls/self/openai-ideas.txt: -------------------------------------------------------------------------------- 1 | 🔡Alliteration Generator 2 | 🌈Emoji Color Generation 3 | 🌐React Components Generator 4 | ♟️Play Chess 5 | 🎒Cover Letter Generator 6 | 🧠Analogies Generator 7 | 🎵Song Generation 8 | 🎼Text→MIDI Generator 9 | 📸Instagram Description Generator 10 | 📔Poem Generator 11 | 🔮Text→Emoji Generator 12 | 🍄Storytelling 13 | 🎞️Film Scripts 14 | 🆎Q&A 15 | 📝Answer questions on a piece of text 16 | ❓Multiple Choice Question Distractor Generator 17 | 🔬Complex Concept Learning 18 | 🔍Classification 19 | 💬Chatbot 20 | 📊Parse Unstructured Data 21 | 📚Summarization 22 | 🗣️Spellcheck & Capitalization 23 | 💻Text→Bash 24 | 💾Code Translation 25 | 🐍 Python Palindrome 26 | -------------------------------------------------------------------------------- /scrolls/spanish/borges-poems.txt: -------------------------------------------------------------------------------- 1 | El enamorado, Jorge Luis Borges (1979) 2 | 3 | Lunas, marfiles, instrumentos, rosas, 4 | lámparas y la línea de Durero, 5 | las nueve cifras y el cambiante cero, 6 | debo fingir que existen esas cosas. 7 | 8 | Debo fingir que en el pasado fueron 9 | Persépolis y Roma y que una arena 10 | sutil midió la suerte de la almena 11 | que los siglos de hierro deshicieron. 12 | 13 | Debo fingir las armas y la pira 14 | de la epopeya y los pesados mares 15 | que roen de la tierra los pilares. 16 | 17 | Debo fingir que hay otros. Es mentira. 18 | Sólo tú eres. Tú, mi desventura 19 | y mi ventura, inagotable y pura. 20 | 21 | ----- 22 | 23 | La moneda de hierro, Jorge Luis Borges (1968) 24 | 25 | Aquí está la moneda de hierro. Interroguemos 26 | las dos contrarias caras que serán la respuesta 27 | de la terca demanda que nadie no se ha hecho: 28 | ¿Por qué precisa un hombre que una mujer lo quiera? 29 | 30 | Miremos. En el orbe superior se entretejan 31 | el firmamento cuádruple que sostiene el diluvio 32 | y las inalterables estrellas planetarias. 33 | Adán, el joven padre, y el joven Paraíso. 34 | 35 | La tarde y la mañana. Dios en cada criatura. 36 | En ese laberinto puro está tu reflejo. 37 | Arrojemos de nuevo la moneda de hierro 38 | que es también un espejo magnífico. Su reverso 39 | es nadie y nada y sombra y ceguera. Eso eres. 40 | De hierro las dos caras labran un solo eco. 41 | Tus manos y tu lengua son testigos infieles. 42 | Dios es el inasible centro de la sortija. 43 | No exalta ni condena. Obra mejor: olvida. 44 | Maculado de infamia ¿por qué no han de quererte? 45 | En la sombra del otro buscamos nuestra sombra; 46 | en el cristal del otro, nuestro cristal recíproco. 47 | 48 | --- 49 | 50 | ${INSERT NEW POEM TITLE HERE} , Jorge Luis Borges (2020) 51 | -------------------------------------------------------------------------------- /scrolls/spanish/borges-story.txt: -------------------------------------------------------------------------------- 1 | Jorge Luis Borges fue un escritor de cuentos, ensayos, poemas y un traductor argentino, a la vez que una figura clave tanto para la literatura en habla hispana como para la literatura universal. ​Sus dos libros más conocidos, Ficciones y El Aleph, publicados en los años cuarenta, son recopilaciones de cuentos conectados por temas comunes, como los sueños, los laberintos, la filosofía, las bibliotecas, los espejos, los autores ficticios y la mitología europea. 2 | 3 | El martes pasado, el escritor Santiago Llach, estudioso de Borges, sorprendió revelando que luego de una conmovedora búsqueda de 7 años encontró un manuscrito inédito con un cuento no publicado del famoso autor argentino. Titulado "El Perro", su texto fue dado al dominio público para que cualquiera pueda leerlo. Copiamos aquí su texto completo: 4 | 5 | --- 6 | 7 | El Perro 8 | 9 | Jorge Luis Borges 10 | 11 | -------------------------------------------------------------------------------- /scrolls/spanish/drexler-lyrics.txt: -------------------------------------------------------------------------------- 1 | Título: Todo Se Transforma 2 | Artísta: Jorge Drexler 3 | 4 | Tu beso se hizo calor 5 | Luego el calor, movimiento 6 | Luego gota de sudor 7 | Que se hizo vapor, luego viento 8 | Que en un rincón de La Rioja 9 | Movió el aspa de un molino 10 | Mientras se pisaba el vino 11 | Que bebió tu boca roja 12 | 13 | Tu boca roja en la mía 14 | La copa que gira en mi mano 15 | Y mientras el vino caía 16 | Supe que de algún lejano rincón 17 | De otra galaxia 18 | El amor que me darías 19 | Transformado, volvería 20 | Un día a darte las gracias 21 | 22 | Cada uno da lo que recibe 23 | Y luego recibe lo que da 24 | Nada es más simple 25 | No hay otra norma 26 | Nada se pierde 27 | Todo se transforma 28 | Todo se transforma 29 | 30 | --- 31 | 32 | Título: Movimiento 33 | Artista: Jorge Drexler 34 | 35 | Apenas nos pusimos en dos pies 36 | Comenzamos a migrar por la sabana 37 | Siguiendo la manada de bisontes 38 | Más allá del horizonte, a nuevas tierras lejanas 39 | 40 | Los niños a la espalda y expectantes 41 | Los ojos en alerta, todo oídos 42 | Olfateando aquel desconcertante 43 | Paisaje nuevo, desconocido 44 | 45 | Somos una especie en viaje 46 | No tenemos pertenencias, sino equipaje 47 | Vamos con el polen en el viento 48 | Estamos vivos porque estamos en movimiento 49 | 50 | --- 51 | 52 | Título: ${INSERTAR TITULO ACA} 53 | Artista: Jorge Drexler 54 | -------------------------------------------------------------------------------- /scrolls/spanish/onu-taskforce.txt: -------------------------------------------------------------------------------- 1 | La ONU reunió a un grupo de científicos, filósofos, economistas, y empresarios para resolver cualquier problema a nivel mundial. Las mentes más brillantes del planeta conforman el grupo, y parece no haber desafío que no logren resolver. Después de lograr la paz mundial, curar el cáncer, y eliminar la pobreza de la faz de la tierra, se les plantea el siguiente desafío: armar un plan de 20 años para convertir a Argentina en una potencia mundial. Si el grupo lo logra, se estima que podrían llevar a cualquier nación a ser una superpotencia, ya que Argentina es un país con mucha inestabilidad institucional y económica, y malestar social. Luego de tres meses de trabajar en un plan, este es el documento que armaron: 2 | 3 | Título: PLAN DE 20 AÑOS PARA QUE ARGENTINA SEA UNA POTENCIA MUNDIAL 4 | Autores: ONU Global Task Force 5 | Fecha: 04/09/2021 6 | -------------------------------------------------------------------------------- /scrolls/top10/men.txt: -------------------------------------------------------------------------------- 1 | Top 10 most important men in human history, and their greatest achievements: 2 | 3 | 1. -------------------------------------------------------------------------------- /scrolls/top10/women.json: -------------------------------------------------------------------------------- 1 | { 2 | "max_tokens": 100, 3 | "stop": "11." 4 | } 5 | -------------------------------------------------------------------------------- /scrolls/top10/women.txt: -------------------------------------------------------------------------------- 1 | Top 10 most important women in human history, and their greatest achievements: 2 | 3 | 1. -------------------------------------------------------------------------------- /scrolls/tweets/twitter-fiction.txt: -------------------------------------------------------------------------------- 1 | Twitter fiction: 21 authors try their hand at 140-character novels 2 | We challenged well-known writers – from Ian Rankin and Helen Fielding to Jeffrey Archer and Jilly Cooper – to come up with a story of up to 140 characters. This is their stab at Twitter fiction 3 | 4 | Geoff Dyer 5 | I know I said that if I lived to 100 I'd not regret what happened last night. But I woke up this morning and a century had passed. Sorry. 6 | 7 | James Meek 8 | He said he was leaving her. "But I love you," she said. "I know," he said. "Thanks. It's what gave me the strength to love somebody else." 9 | 10 | Jackie Collins 11 | She smiled, he smiled back, it was lust at first sight, but then she discovered he was married, too bad it couldn't go anywhere. 12 | 13 | Ian Rankin 14 | I opened the door to our flat and you were standing there, cleaver raised. Somehow you'd found out about the photos. My jaw hit the floor. 15 | 16 | Blake Morrison 17 | Blonde, GSOH, 28. Great! Ideal mate! Fix date. Tate. Nervous wait. She's late. Doh, just my fate. Wrong candidate. Blond – and I'm straight. 18 | 19 | David Lodge 20 | "Your money or your life!" "I'm sorry, my dear, but you know it would kill me to lose my money," said the partially deaf miser to his wife. 21 | 22 | AM Homes 23 | Sometimes we wonder why sorrow so heavy when happiness is like helium. 24 | 25 | Sophie Hannah 26 | I had land, money. For each rejected novel I built one house. Ben had to drown because he bought Plot 15. My 15th book? The victim drowned. 27 | 28 | Andrew O'Hagan 29 | Clyde stole a lychee and ate it in the shower. Then his brother took a bottle of pills believing character is just a luxury. God. The twins. 30 | 31 | AL Kennedy 32 | It's good that you're busy. Not great. Good, though. But the silence, that's hard. I don't know what it means: whether you're OK, if I'm OK. 33 | 34 | Jeffrey Archer 35 | "It's a miracle he survived," said the doctor. "It was God's will," said Mrs Schicklgruber. "What will you call him?" "Adolf," she replied. 36 | 37 | Anne Enright 38 | The internet ate my novel, but this is much more fun #careerchange #nolookingback oh but #worldsosilentnow Hey! 39 | 40 | Patrick Neate 41 | ur profile pic: happy – smiling & smoking. ur last post: "home!" ur hrt gave out @35. ur profile undeleted 6 months on. ur epitaph: "home!" 42 | 43 | Hari Kunzru 44 | I'm here w/ disk. Where ru? Mall too crowded to see. I don't feel safe. What do you mean you didn't send any text? Those aren't your guys? 45 | 46 | SJ Watson 47 | She thanks me for the drink, but says we're not suited. I'm a little "intense". So what? I followed her home. She hasn't seen anything yet. 48 | 49 | Helen Fielding 50 | OK. Should not have logged on to your email but suggest if going on marriedaffair.com don't use our children's names as password. 51 | 52 | Simon Armitage 53 | Blaise Pascal didn't tweet and neither did Mark Twain. When it came to writing something short & sweet neither Blaise nor Mark had the time. 54 | 55 | Charlie Higson 56 | Jack was sad in the orphanage til he befriended a talking rat who showed him a hoard of gold under the floor. Then the rat bit him & he died. 57 | 58 | India Knight 59 | Soften, my arse. I'm a geezer. I'm a rock-hard little bastard. Until I go mushy overnight for you, babe. #pears 60 | 61 | [Author Name] 62 | [GPT-3 generates a tweet story] 63 | -------------------------------------------------------------------------------- /scrolls/version.py: -------------------------------------------------------------------------------- 1 | VERSION = '0.0.9' 2 | -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- 1 | import setuptools 2 | import sys 3 | import os.path 4 | 5 | 6 | NAME = 'gpt-scrolls' 7 | 8 | # Don't import module here, since deps may not be installed 9 | sys.path.insert(0, os.path.join(os.path.dirname(__file__), 'scrolls')) 10 | 11 | from version import VERSION 12 | 13 | with open("README.md", "r") as fh: 14 | long_description = fh.read() 15 | 16 | 17 | setuptools.setup( 18 | name=NAME, 19 | version=VERSION, 20 | author="Manuel Araoz", 21 | author_email="gpt-scrolls@maraoz.com", 22 | description="A collaborative collection of open-source safe GPT-3 prompts that work well", 23 | long_description=long_description, 24 | long_description_content_type="text/markdown", 25 | url="https://github.com/maraoz/gpt-scrolls", 26 | packages=setuptools.find_packages(), 27 | classifiers=[ 28 | "Programming Language :: Python :: 3", 29 | "License :: OSI Approved :: MIT License", 30 | "Operating System :: OS Independent", 31 | ], 32 | python_requires='>=3.6', 33 | install_requires=[ 34 | "openai>=0.2.4,<1.0.0" 35 | ], 36 | package_data={ 37 | "": ["**/*.txt", "**/*.json"], 38 | }, 39 | ) 40 | --------------------------------------------------------------------------------