Alexson @Alexson_AI

AI + Math @ HKU Former researcher — NLP, RAG & HCI 📷 Usually taking photos when I’m not debugging Joined August 2026

Tweets

12
Followers

1
Following

96
Likes

100

Alexson @Alexson_AI

2 weeks ago

@ai_horn1 @dotey 我也是这么想的，其中一个解释可能是：因为问题已经被解决了，所以已经没有动力让“大众”的数学家去研究这个题目了

0 0 1 169 0

View Details

Alexson @Alexson_AI

2 weeks ago

@ScarletKc 我感觉oai就是想要先满足绝大部分的现实需求…还记得一开始的3.5是纯粹的chat模型吗，现在已经变成了协助解决现实问题的模型了。未来可见的也会在后训练里刻意加入发散性思维

1 0 2 1K 1

View Details

Alexson @Alexson_AI

2 weeks ago

目前只能填form，所以不能完全說是取代了原先的Jev模型。但是這是一個好的開始

Cua @trycua

2 weeks ago

1/ Introducing CUA-S1: a family of System One Models, small, specialized, and built for computer use. Today we're open-sourcing CUA-S1-FORMS, the first in the family: github.com/trycua/cua

117 491 6K 1.3M 7K

0 0 0 14 0

View Details

Alexson @Alexson_AI

2 weeks ago

在很久以前multi-agent decision making的時候，大家都已經在研究用什麼語言讓agent溝通是比較好的了。因此這篇論文屬於是把很久以前的問題拿出來重新研究一次。

AYi @AYi_AInotes

3 weeks ago

清华大学和无问芯穹等团队刚刚开源了一篇注定载入史册的重磅论文（已收录于 ICLR 2026，代码已开源）：他们彻底打破了所有多智能体（Multi-Agent）系统的底层范式—— 大型语言模型之间，从今天起可以彻底不通过人类文字进行直接交流了！这项被称为 C2C（Cache-to-Cache，缓存到缓存）的技术，

547 632 4K 382K 3K

0 0 0 18 0

View Details

They were building in stealth for 2 years, I was building in stealth for 2 hours… Happy to open source Qwen-2.5-1B-RLCD, 5x faster on-device inference for JSON workloads that need to be type-safe. ⚡️Demo below on a M4 MacBook⚡️ every LLM has the ability to efficiently batch