We have one horrible disjuncture, between layers 6 → 2. I have one more hypothesis: A little bit of fine-tuning on those two layers is all we really need. Fine-tuned RYS models dominate the Leaderboard. I suspect this junction is exactly what the fine-tuning fixes. And there’s a great reason to do this: this method does not use extra VRAM! For all these experiments, I duplicated layers via pointers; the layers are repeated without using more GPU memory. Of course, we do need more compute and more KV cache, but that’s a small price to pay for a verifiably better model. We can just ‘fix’ an actual copies of layers 2 and 6, and repeat layers 3-4-5 as virtual copies. If we fine-tune all layer, we turn virtual copies into real copies, and use up more VRAM.
March 10, 2026 at 9:22 a.m. PT
。新收录的资料对此有专业解读
我的瓶颈在于我的创造力，或者说取决于我能想到多少可以做的事情，以及我能为客户交付多少价值。这些才是我获取效率提升并且产出更多内容的地方，而不是仅仅在范围内限制输入来让公司盈利。
也就是：豆腐要像人一样有个性。