Preview
Preview shortened for the browser. The exported file keeps everything.
It opens with: “There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing…” — a public X post by Simon Willison (@simonw), posted 2026-11-03. Verified for export on this site: 321 characters, 1 paragraphs. Use the button above to save it as Word, PDF or Markdown, images included.
The source link for this entry is already filled in — press Read the article to export it.
Untick anything you do not want in the document. The order of the rest is unchanged.
Preview shortened for the browser. The exported file keeps everything.
It opens with: “There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing…” — a public X post by Simon Willison (@simonw), posted 2026-11-03. Verified for export on this site: 321 characters, 1 paragraphs. Use the button above to save it as Word, PDF or Markdown, images included.
Figures above were collected on 2026-11-04 and may have changed since.