{
  "id": 13622772,
  "title": "My weather app said the sky would be clear. I taught an open model to check, then went outside",
  "url": "https://urgent.news/2026/10/11/my-weather-app-said-the-sky-would-be-clear-i-taught-an-open-model-to",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-11T02:54:00.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/uptimearchitect/my-weather-app-said-the-sky-would-be-clear-i-taught-an-open-model-to-check-then-went-outside-ebb"
  },
  "original_language": "en",
  "account": "Cloud cover is the deciding factor for whether going outside at night is worth it, and weather apps often get this wrong. The author, a systems engineer, decided to build a solution to address this issue. He created Clear Tonight, an open-source tool that compares the cloud forecast from a weather app with a model trained on the forecast's historical accuracy for a specific location. This model, called TabPFN v2, reduces the error in the forecast by 28-41% over a year's worth of data from six different cities. The tool provides a verdict for tonight, an hour-by-hour comparison of the app's prediction and the actual chance of clear skies, and a moon phase indicator. Additionally, the author built a local version of the tool that can be trained on personal location data and generate a short plan based on the results. The project includes a live page for six cities and a sky-photo log that uses Gemma 4 to rate the cloud cover in user-provided photos. The code is available on GitHub, and a demo video showcases the tool's functionality.",
  "summary": "This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass Every stargazer knows this evening. The app says clear . You find the warm jacket, drive out past the streetlights, let your eyes adjust for twenty minutes… and look up at a flat grey lid. Cloud cover is the one number that decides whether going outside at night is worth it, and it's the number weather apps…",
  "key_points": [
    "TabPFN v2 model reduces forecast error by 28-41% over a year's data from six cities.",
    "Local version of the tool can be trained on personal location data and generate a short plan."
  ],
  "editors_take": "This development gives individuals a more accurate forecast of clear skies for their specific location, potentially changing how they plan nighttime activities, by combining weather app data with a locally trained model.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}