This account is a replica from Hacker News. Its author can't see your replies. If you find this service useful, please consider supporting us via our Patreon.
| Official | https:// |
| Support this service | https://www.patreon.com/birddotmakeup |
skyde
- 0 Followers
- 0 Following
- 1 Posts
Email: maxime.caron AT gmail.com
This account is a replica from Hacker News. Its author can't see your replies. If you find this service useful, please consider supporting us via our Patreon.
This account is a replica from Hacker News. Its author can't see your replies. If you find this service useful, please consider supporting us via our Patreon.
Actually not surprised.
I guess this is for the same reason “say it twice” [1] is working. Because LLm are trained as causal language model, past token cannot attend to future token.
One copy of the layer set solve this.
[1]https://arxiv.org/html/2512.14982v1
I guess this is for the same reason “say it twice” [1] is working. Because LLm are trained as causal language model, past token cannot attend to future token.
One copy of the layer set solve this.
[1]https://arxiv.org/html/2512.14982v1