thesephist

import { cascade } from "../src/cascade.ts";

export const summarize = cascade
  .chatCompletion("summarize", {
    model: "gpt-3.5-turbo",
    renderPrompt: (context: { title: string; content: string }) => [
      {
        type: "user",
        content: [
          `Write a one sentence summary about ${context.title}.\n\n${context.content}.`,

thesephist

; pull-all
;
; Run "git pull" en masse on all git repositories with a default remote under a
; specific directory.

(defn indent (s bit)
  (-> s
      (split '\n')
      (map (fn (line)
             (if (empty? line)

thesephist

import torch
import torch.nn as nn
import torch.nn.functional as F

from transformers import GPT2Tokenizer, GPT2LMHeadModel

ppl_model_name = 'gpt2-xl' if device == 'cuda' else 'gpt2'
ppl_tokenizer = GPT2Tokenizer.from_pretrained(ppl_model_name)
load_opts = {
    'device_map': 'auto',

thesephist

// Code to measure selection rects

// NOTE: this isn't actually JS... it's Oak. But I think the code is readable if you know JS :/

// NOTE: I *think* this event listener only works when attached to `document`.
with document.addEventListener('selectionchange') fn handleSelectionChange {
	if active := document.activeElement {
		? -> ?
		_ -> if active.classList.contains('textarea-itself') {
			false -> ?

thesephist

#!/usr/bin/env python

import os
import json
from tqdm import tqdm
from transformers import GPT2TokenizerFast

os.environ['TOKENIZERS_PARALLELISM'] = 'false'

FILENAME = './thesephist.jsonl'

thesephist

{ name: 'Lighting', options: [
  'golden hour, warm glow'
  'blue hour, twilight, ISO12000'
  'midday, direct lighting, overhead sunlight'
  'overcast, whitebox, flat lighting, diffuse'
  'dreamlike diffuse ethereal lighting'
  'dramatic lighting, dramatic shadows, illumination'
  'studio lighting, professional lighting, well-lit'
  'flash photography'
  'low-key lighting, dimly lit'

thesephist

ask CLI

ask is a little CLI I made to interact with OpenAI's GPT-3 (text-davinci-002) from my shell/terminal. The instruction fine-tuning on that model makes it particularly ideal for just asking questions and making requests.

With this CLI, I can do something like:

$ ask 'Write a haskell function that reverses a string'
reverseString :: String -> String
reverseString = foldl (\acc x -> x : acc) []

thesephist

/**
 * Fancy ID generator that creates 20-character string identifiers with the following properties:
 *
 * 1. They're based on timestamp so that they sort *after* any existing ids.
 * 2. They contain 72-bits of random data after the timestamp so that IDs won't collide with other clients' IDs.
 * 3. They sort *lexicographically* (so the timestamp is converted to characters that will sort properly).
 * 4. They're monotonically increasing.  Even if you generate more than one in the same timestamp, the
 *    latter ones will sort after the former ones.  We do this by using the previous random bits
 *    but "incrementing" them by 1 (only in the case of a timestamp collision).
 */

thesephist

rush mv *.pdf 'o-{{name}}.pdf && ls *.pdf | oak pipe "print(line), exec('/usr/local/bin/gs', ['-sDEVICE=pdfwrite', '-dCompatibilityLevel=1.4', '-dPDFSETTINGS=/ebook', '-dNOPAUSE', '-dQUIET', '-dBATCH', '-sOutputFile=' + str.trimStart(line, 'o-'), line], '').stdout |> str.trim()"

thesephist

"Afterglow" is a speculative science fiction novel by Greg Egan about a world attempting to rebuild after an apocalyptic nuclear attack on semiconductor foundries in Taiwan causes worldwide silicon shortages and geopolitical issues.

1. Synopsis

The novel is set in the year 2040, in the aftermath of a nuclear attack on semiconductor foundries in Taiwan that has left the world without access to silicon. The attack has caused widespread economic and political upheaval, and the novel follows the efforts of a group of scientists, engineers, and entrepreneurs to develop new technologies and rebuild the world.

2. Characters

The novel follows a large cast of characters, including: