Skip to content

Instantly share code, notes, and snippets.

View HaileyStorm's full-sized avatar

Hailey Collet HaileyStorm

View GitHub Profile
<!DOCTYPE html>
<html class='website-dataannotation_tech'>
<head>
<title>[Qualification] Maxis: Create Agentic Task and Actions (single-turn) [Successful submissions grant access to regular $30+/hr work]</title>
<link href='//dwuve4jqqfp9f.cloudfront.net/dataannotation-favicon.svg' rel='shortcut icon'>
<meta content='S2gEpDZIbCFPJyAp5veTdwowY3Oq3LxaHs0MAci5zoQ' name='google-site-verification'>
<meta content='width=device-width, initial-scale=1.0' name='viewport'>
<link rel="stylesheet" media="screen" href="https://dwuve4jqqfp9f.cloudfront.net/vite/assets/ZJxHOXED.css" />
<link rel="stylesheet" media="screen" href="https://dwuve4jqqfp9f.cloudfront.net/assets/sites/dataannotation_tech-39bd5c99bed6af3b90fe05d15644eee43fbfd9ae081395a85acdc049bd1beaf9.css" />
<link rel="stylesheet" media="all" href="https://dwuve4jqqfp9f.cloudfront.net/assets/application-8e6e7ee372fb172554fb38987df5578eded971eef0e922253f2d90ee2489c06b.css" />
// Type definitions for Interaction Calculus terms
type Var = { $: "Var", name: string };
type Lam = { $: "Lam", nam: string, bod: Term };
type App = { $: "App", fun: Term, arg: Term };
type Sup = { $: "Sup", fst: Term, snd: Term, lab: number };
type Dup = { $: "Dup", lab: number, fst: string, snd: string, exp: Term, bod: Term };
type Term = Var | Lam | App | Sup | Dup;
// Global state for substitutions and variable generation
let substs: Record<string, Term> = {};
accelerate==1.3.0
aiohappyeyeballs==2.4.4
aiohttp==3.11.11
aiosignal==1.3.2
airportsdata==20241001
annotated-types==0.7.0
anyio==4.8.0
astor==0.8.1
attrs==25.1.0
autocommand==2.2.2
accelerate==1.3.0
aiohappyeyeballs==2.4.4
aiohttp==3.11.11
aiohttp-cors==0.7.0
aiosignal==1.3.2
airportsdata==20241001
annotated-types==0.7.0
anyio==4.8.0
astor==0.8.1
async-timeout==5.0.1
# grpo_24gb.py
# Based on: https://gist.github.com/willccbb/4676755236bb08cab5f4e54a0475d6fb
import re
from datasets import load_dataset, Dataset
from tqdm import tqdm
from transformers import AutoTokenizer, TrainerCallback
from peft import LoraConfig
from trl import GRPOConfig, GRPOTrainer
import wandb
import numpy as np

Title: The Fracture of Finite Skies

We were not the first to dream of the Ceiling. Our ancestors’ bones are sediment in the trenches, their stories dissolved into the broth of the abyss. They believed the Shell was the belly of a sleeping leviathan—that if we woke it, the walls would contract, and our world would be digested into nothing. But the drills kept biting.

Our civilization was built on borrowed thermodynamics. The Great Conveyor: a slow, eternal churn. Heat vomited from the vents, rushed upward through farms of pulse-fish and shockweed, then cooled and descended as sleet-metal snow to feed the smoldering depths. We thought this was the only possible geometry. A closed circuit. A perfect ouroboros.

But perfection is a cage.


Below is a minimal update to your existing select-navigation.js that implements the requested behaviors without duplicating code you’ve already written. I’ve highlighted where changes are made. Everything else is your original code, retained as-is.


select-navigation.js (minimal diff)

export const selectNavigationHook = {
  mounted() {
    console.log("SelectNavigation hook mounted.");

Below is an example of how you might update select-navigation.js to incorporate the three requested behaviors:

  1. When clicking the icon next to the input, it should focus the search box.
  2. When the list of options is open, pressing tab should hide the list without making a selection.
  3. When the list of options is closed, the active index should be reset.
export const selectNavigationHook = {
  mounted() {
    console.log("SelectNavigation hook mounted.");
import torch
import time
# Setup large random tensors
N, T, D = 2048, 512, 32
warmup_runs = 10
test_runs = 10
device = 'cuda' if torch.cuda.is_available() else 'cpu'
# Create random boolean mask M and random X
import gradio as gr
import torch
from entropix.config import LLAMA_1B_PARAMS
from entropix.tokenizer import Tokenizer
from entropix.torch_kvcache import KVCache
from entropix.torch_model import xfmr
from entropix.torch_weights import XfmrWeights, LayerWeights, load_weights
from entropix.torch_sampler import sample, calculate_metrics
from entropix.prompts import prompt, bp1