←
Jin513 posts
Jin avatar

Jin

@jinfizz

tech lead and research scientist, calmly computing @MegaETH. All opinions are my own.

◷ Joined December 2023▧ 25 Photos and videos
Jin avatar
Jin@jinfizz·4 Aug 2024

Many folks were suprised by how fast I decided to leave my PhD program at Harvard and join MegaETH two months ago. To be honest it was such an easy choice. Any systems researcher would do the same after reading @yilongl_megaeth's analysis of Etheruem performance and chatting with @yangl1996.

Jin avatar
Jin@jinfizz·Nov 17

It’s the future

Just watched @jinfizz drive an ADE from his iPhone Duo and I can't help but think foldables are going to be the new remote workstation.

Jin avatar
Jin@jinfizz·Nov 13

With all those timeouts and untold degradation, gpt is just THE trash. Stable speed matters

“Built for speed” Still garbage speeds compared to anything else on the market. Note that I ran Sol and Opus at regular tier! Didn’t even use fast/ultrafast.

Post attachment
Jin avatar
Jin@jinfizz·Oct 26

我觉得claude desktop的UI改进挺好的,看起来干净了很多,原来总给我一种泛黄陈旧感,让我觉得昨天开发的东西都很旧不想再碰。 我用claude就是因为opus 5.5,不用别的agent是因为怕被封号,就这么简单。 批评说去掉了品牌感,导致用户迁移到其它agent没有心理障碍,怕是根本没想明白大家为什么用claude 不过新设计宣发时候非要强调迭代得多快,挺抽象的。用户只想知道为什么更好了,你越说快用户越不放心。就算设计没有明确的benchmark数字,也是可以讲清楚方向的。

Jin avatar
Jin@jinfizz·Oct 26

when can people understand they need to show evals launching a harness

Introducing Octopus OS, a reinvented agent harness. Agents can edit their own harness in the middle of a task, on their own. Unlike Harness² and similar projects, self-improvement isn’t a separate stage bolted onto the agent. The whole architecture is just the environment they work in. Over 8 months, chasing self-improvement that feels truly native, I dropped the usual agent loop entirely and designed a new architecture that works a lot like a real operating system, with a 45k-line Rust kernel holding it all together. The system state lives in JSON files: goals act as real environments where agents work, along with inputs, outputs, and custom triggers they respond to. Agents can edit all of it while they work inside. It was also built as a product from day one, not a research demo. The full desktop app took 3 months, built on top of the Codex source. The client already works, and you can use it today like any other agent, Codex included. And it’s open source.

0:35
Jin avatar
Jin@jinfizz·Oct 24

就kimi套餐这个b额度,还在这宣发自己便宜,洗洗睡吧你 利益相关:本人拥有两个顶配号

Jin avatar
Jin@jinfizz·Oct 23

为什么我用一个新手机号就能买到kimi 1399/月的套餐,但是旧号持续的699订阅就看不到升级选项?

Jin avatar
Jin@jinfizz·Oct 23

我看最近大家都在玩oppo 心力球,也买了一个。 但这东西居然要求买oppo手机,且数据是被oppo管理的。 下面是一份帮你自主管理心力球数据的sdk,你可以随意让agent写应用来帮你管理硬件。没有任何限制,不需要oppo手机,不需要下载应用。 github.com/kiyoakii/xinliqiu…

Jin avatar
Jin@jinfizz·Oct 20

Has anyone noticed that Claude obviously reduced 5h session token limit? It's been burning so fast for the past few days

Jin avatar
Jin@jinfizz·Oct 3

antigravity is the only agent app that stuck at google login redirection

Post attachment
Jin avatar
Jin@jinfizz·Oct 1

when you pay $500 a month only to try ultrafast good job openai, really good

Post attachment
Jin avatar
Jin@jinfizz·Sep 22

Been running deepswe with deepseek v4.1 flash for fun. Interestingly it recognizes deepswe question format and always tries to guess what hidden tests would be instead of programming as an expert. In its thinkings, keyword "hidden tests" just comes out of nowhere. RL I guess.

Jin avatar
Jin@jinfizz·Sep 21

因为gpt降智被迫和fable讨论一个项目,聊了两天感觉智商有大问题,干活也不认真讲细节,各种擅自作主的错误,导致实验结论都不可信,血压蹭蹭涨

Jin avatar
Jin@jinfizz·Sep 17

OpenAI's fraud.

Jin@jinfizzSep 17

根据我几天的实测,codex会在拥挤时自动把astra降级成5.6 luna,有的账号会直接全局默认降级,没有任何通知,且应该按照astra价格收费。 如果你也有一样的担忧,可以用我写的监控器,默认半小时监控一次。 github.com/kiyoakii/is-gpt-n…

Post attachmentPost attachment
Jin avatar
Jin@jinfizz·Sep 17

根据我几天的实测,codex会在拥挤时自动把astra降级成5.6 luna,有的账号会直接全局默认降级,没有任何通知,且应该按照astra价格收费。 如果你也有一样的担忧,可以用我写的监控器,默认半小时监控一次。 github.com/kiyoakii/is-gpt-n…

Post attachmentPost attachment