张鹏

[张鹏 @pczzy]

@tomocha0 Fumoto是什么

[张鹏 @pczzy]

@Mid_observatory 听见了乌鸦叫

[张鹏 @pczzy]

@passluo @yangshaobo 北京撤光了

[张鹏 @pczzy]

@rewoIf

[Chris Maathuis @sports8]

Young man you are a one of a kind bird caller. The champ of champs. I would love to hear you do a magpie.

[张鹏 @pczzy]

@thsottiaux 什么时候可以用手机操作服务器上的codex?

[张鹏 @pczzy]

@herock Codex智能又好用，我不常不关注重置，希望能直接跟手机app打通，我就可以手机拌匀服务器上的codex了

[张鹏 @pczzy]

@AYi_AInotes 同样提示词拿seedance2.0 生成的，效果没有预期的好

[张鹏 @pczzy]

剪了个麒麟啤酒罐，做简易手冲 #咖啡

[张鹏 @pczzy]

#周杰伦 #鸟巢 演唱会

[alphaXiv @askalphaxiv]

Here’s a fun comparison between GLM 5.2 and Opus 4.8 on a one-shot reproduction of the SDPO paper

This is a hard task: the model must resolve messy verl issues and then run ablations to completion and confirm the paper’s claims.

- GLM 5.2 costs $6.21 while Opus 4.8 cost us $46.35

- Both models spent a bulk of their tokens resolving initial verl issues. GLM 5.2 attempted 14 failed runs before first success while Opus 4.8 attempted 9 runs.

- GLM 5.2 surprisingly took 2.65M tokens (excl re-reads) compared to 4.53M tokens for Opus 4.8

[张鹏 @pczzy]

北京的云

[张鹏 @pczzy]

@YigeKing 世外桃源

[张鹏 @pczzy]

@kudo_pon @tomocha0 很热闹

[张鹏 @pczzy]

@MrMarket89 @mindmoon_108 贵不贵？

[张鹏 @pczzy]

@tomo_coco73 又地震了？

[张鹏 @pczzy]

@fraveris @daemonzhang6 想去钓鱼