Jun-Jun Wan

475 posts

Jun-Jun Wan banner
Jun-Jun Wan

Jun-Jun Wan

@_zenzoen

entrepreneur. generalist. contrarian. polyglot. a curious mind. building the future of spontaneous agentic orders. ex-autonomous-driving.

digital Katılım Mayıs 2020
232 Takip Edilen53 Takipçiler
Sabitlenmiş Tweet
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
A good reminder back the days: Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt
Deutsch
0
0
5
3.9K
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
@0xTykoo 之前參加過一個OpenAI舉辦的5個小時限時的黑客松
中文
0
0
0
46
Tykoo
Tykoo@0xTykoo·
我觉得应该办一个全世界最健康的黑客松比赛,谁都他妈不许熬夜,一天不许工作超过8小时,定时离开电脑,让你的agent自己跑一晚上任务跑满且不浪费也是种能力, 每天强迫健身一次,大家一起吃健康餐(有点形式主义了)然后全都带上whoop手环,每天都要记录分数,算入成绩里, 我总感觉黑客松摄影老师们非要找角度抓拍大家四仰八叉、疲惫不堪的样子,其实挺病态的。
中文
73
44
906
90.1K
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
Do you think there are more normal people or abnormal people among successful investors and founders? @pmarca @paulg
English
0
0
0
30
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
„Logik unserer Sprache“. Who is 'our' here? What is the logic?
English
0
0
0
22
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
LLMs exhibit a tendency toward recency bias.
English
0
0
0
23
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
going to run 100+ successful companies
English
0
0
1
21
Jun-Jun Wan retweetledi
kepano
kepano@kepano·
New policy from @Atlassian: Unless you opt out by August 17th 2026, data from Jira and Confluence will automatically be used for AI training. Some data cannot be opted out at all on some plans. x.com/kepano/status/…
kepano tweet mediakepano tweet media
kepano@kepano

if your data is stored in a database that a company can freely read and access (i.e. not end-to-end encrypted), the company will eventually update their ToS so they can use your data for AI training — the incentives are too strong to resist

English
59
319
2.4K
538.6K
Viviennn
Viviennn@0xViviennn·
在用 Hermes Agent 的时候,一直感觉跨 session 的中文搜索不太好用。Agent明明说过的话,session_search 就是搜不到,像完全失忆了一样.... 比如昨天搞了一天的A2A通信,但今天再搜"和其他Agent的聊天记录"的时候,返回 0 条结果。但数据确实在数据库里——用 LIKE子串匹配能搜出 20 多条 然后我让 Claude Code 帮忙查了一下,原因是 SQLite FTS5的默认分词器。英文有空格,天然能按词分 中文没有空格,FTS5把每个汉字当成独立 token。搜"记忆断裂"实际执行的是 记 & 忆 & 断 & 裂——四个独立的字分别匹配,丢失了词的连续性,搜不到原本存在的内容 日文、韩文也一样,都没有空格分隔,都会中招,但英文用户完全不受影响。 目前的解决方案是 FTS5 搜不到结果时,检测到查询包含 CJK 字符就自动降级到 LIKE子串匹配。比 FTS5 慢一点(全表扫描),但对 agent 的数据量完全够用。 已提 PR: github.com/NousResearch/h… @NousResearch @Teknium
Viviennn tweet media
中文
44
10
240
136.6K
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
poem, colonization, computer
English
0
0
0
30
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
Why are European founders still chasing Forbes 30 under 30?
English
0
0
0
30
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
Guten Tag Herr Prof. Kant, sapere aude!
Jun-Jun Wan tweet media
Deutsch
0
0
0
33
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
The Enlightenment movement of AI Agents
English
0
0
0
25
Hubert Thieblot
Hubert Thieblot@hthieblot·
pitch me your company in 1 word.
English
2.5K
15
1.1K
285K
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
Handling the “chicken-and-egg” question
English
0
0
0
22
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
All trades, pursuits, and professions are becoming more and more overstocked; and multitudes of persons, of all degrees and ages, are moving about, without employment, useless to themselves, and a burden to the public
English
0
0
0
31
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
Your startup is not AI agent native enough!
English
0
0
0
39
Jun-Jun Wan
Jun-Jun Wan@_zenzoen·
How much can you internalize and how fast retrieve with association?
English
0
0
0
32