研究人員稱 LLM 可原生讀寫神經網絡權重；GPT-6 Astra 修改 Qwen3.5-0.8B 權重以提升數學能力
Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math (Rikiya Takehi)
研究人員表示，LLM 可原生讀取和寫入神經網絡權重。GPT-6 Astra 修改 Qwen3.5-0.8B 的權重，以提升其數學能力。
Featured Podcasts
The Difference Between 1% Improvements and 10x Growth | Immad Akhund
Grit explores what it takes to create, build and scale world-class organizations.
Subscribe to Grit.
How a 19-year-old intern built the growth engine behind this $2B AI company | Hector Serrano
Interviews with world-class product leaders and growth experts to uncover actionable advice to help you build, launch, and grow your own product.
Subscribe to Lenny's Podcast.
Big Tech's Ugly Week, Getting Rich At Anthropic, AI Costs Plunge
The Big Technology Podcast takes you behind the scenes in the tech world featuring interviews with plugged-in insiders and outside agitators.
Subscribe to Big Technology Podcast.
OpenAI's Latest Discoveries Spark a Backlash + Help! My Kid Hates AI
The official podcast of the singularity, from Casey Newton and Kevin Roose. Twice-weekly news, analysis, and interviews with the people who are building the future.
Subscribe to Machine Gods.
The Accidental YC Founder Solving AI's GPU Crisis | SF Compute's Evan Conrad
Veteran tech reporter Alex Konrad sits down with breakout entrepreneurs taking on the status quo to shake up their fields in AI, design, nuclear energy, space, and more.
Subscribe to The Upstarts Podcast.
Why Sam Altman is against a single superintelligence
Stripe cofounder John Collison interviews founders, builders, and leaders over a pint.
Subscribe to Cheeky Pint.
來源：Techmeme · techmeme.com