Marvin

Quant, system design, currently experimenting with self-hosting stuff.

[Marvin @PrakashMarvin]

@siddy_gg 🔥🔥🔥

[Marvin @PrakashMarvin]

@iamdothash cant wait to set it up and connect to my mac and phone over tailscale! i dont need both scrcpy and kde-connect both now!

[Marvin @PrakashMarvin]

Starship in orbit! Historical!

[SpaceX @SpaceX]

Starship’s 14th flight is set to launch on Monday, Sept 28. The 75-minute launch window opens at 7:15 a.m. CT. Live coverage of the mission starts ~35 minutes before launch → https://t.co/uQKQvgaTmJ

[Marvin @PrakashMarvin]

@0xSero @MiaAI_lab 96 GB Ram.

[Marvin @PrakashMarvin]

@0xSero I have RTX 5060Ti on my Beelink GTi15 Ultra Intel® Core™ Ultra 9 285H, with Omarchy Quattro, i have used Qwen3.8-27B · EXL3 · 3.5bpw by @MiaAI_lab

[Marvin @PrakashMarvin]

By outsourcing our intelligence, we become slaves to those who own it.

[Mia @MiaAI_lab]

Local models are useless?

[Marvin @PrakashMarvin]

@OmarchyMac @hornsby_andrew Hoping for M3 soon!

[Marvin @PrakashMarvin]

@sophiadew i am aligned with the philosophy, dont know if i am known enough to make the list 😅

[Marvin @PrakashMarvin]

@vaxryy @Pekot_ man you make it hard for me to follow you on my main tech X account.

[Marvin @PrakashMarvin]

@trq212 I don’t know about others, but I need a clear mental distinction between starting to make changes and the planning stage.

[Marvin @PrakashMarvin]

@MiaAI_lab Holy shit that is such good news, i love minimax models and this model is performing very well.

[Marvin @PrakashMarvin]

Havent laughed this hard in a while 😂

[rekdt @rekdt]

They call me 007…

I am very bad at Linux permissions

[rekdt @rekdt]

They call me 007…

I am very bad at Linux permissions

[Marvin @PrakashMarvin]

@todaymare Mutin thos clown ain't enough!

[Marvin @PrakashMarvin]

I also built something similar inspired by this

[雨夹雪❄️ @mizorewww]

介绍比Jev快50倍，在你设备上跑的laya-mlx!

只在你的设备上占用最高1G内存

Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统

我将其移植到MLX,并且做了一些性能优化!

视频中就是这个模型在我的本地M3Max上玩贪吃蛇

这个模型能够以每秒决策60次的速度玩贪吃蛇！

https://t.co/31KGUiNunb

[Marvin @PrakashMarvin]

@typesafeai jev who?

[Marvin @PrakashMarvin]

Local 2048 on MLX: ~42 ms per typed decision, zero tokens, offline.

Laya-MLX is amazing! 2048 soon!

[Marvin @PrakashMarvin]

@siddy_gg my smart toaster is running on my smart fridge.

[Marvin @PrakashMarvin]

@amateurhuman @OmarchyMac i am waiting for the M3 version to drop

[Marvin @PrakashMarvin]

@dhh holy smokes! What are the IO specs for the machine.