Edited Qwen3.5-0.8B released after GPT-6 Astra rewrites weights without training
Rikiya Takehi and collaborators published an edited Qwen3.5-0.8B on Hugging Face, reporting that GPT-6 Astra rewrote the model's weights directly with no training and boosted math scores by 40%.
Read the full story at AlphaSignal ↗
Timeline · 1 report
- 2026-11-04 12:18 · AlphaSignal
GPT-6 Astra Rewrites Qwen3.5 and Boosts Math Scores by 40%