Christopher Owen retweeted
What it feels like when you start using a Chinese model and it actually just does what you want without a fucking condescending moral lecture.
Christopher Owen retweeted
All employees come in 5 levels, choose wisely:
Level 1 - tells you there's a problem
Level 2 - tells you there's a problem and what caused it
Level 3 - tells you there's a problem, what caused it, and options to fix it
Level 4 - tells you there's a problem, what caused it, options to fix it, and which one they'd pick
Level 5 - tells you there was a problem, what caused it, how they fixed it, and future proof’d it
My name's on the same line as Mark Dowd.
I still have a signed copy of his book.
oreilly.com/library/view/the…
What an incredibly cool visualization!
moved my 4 sparks from tensorfold to vllm today, @christopherowen's spark-ds41f on the 64k-page kernel. tf stays installed as fallback. ran a few tests tonight, everything in the video is replayed from the logs
¯\_(ツ)_/¯
... works all the way to PXE
github.com/christopherowen/d…
Nvidia intentionally broke my dgx-spark-fan-control with their latest firmware release.
Why can't we set fan curves?
Why is this a protected firmware-only function?
Looks like one of you did some research!!
forums.developer.nvidia.com/…
hmm... looks like Google is doing spark owners a solid with their latest disclosure.
nitter.cf/IntCyberDigest/status/…
‼️ Anthropic and Google researchers found a flaw that lets one crafted packet hijack a server during network boot, before Secure Boot checks the OS.
Almost all servers on earth are affected.
CVE-2026-97631 sits in TianoCore EDK II, code in the UEFI firmware of most servers.
Readers added context they thought people might want to know
The vulnerability is limited to servers actively PXE network-booting with vulnerable EDK II firmware; most servers boot from local disks and are unaffected in normal use, with Intel noting limited impact on supported products.
github.com/tianocore/edk2…
arstechnica.com/security/2026/…
This is essentially a programming language in Neuralese.
1 byte = 1 token.
the brainchild of @mdrakestapleton
aienos.com/omega/
asking chatgpt if we can control the fans on the spark with the latest firmware...
apparently I am the source of this on the internet
there is a tiny computer inside your spark.
it's encrypted and hidden from you.
this is the 'firmware' you keep updating.
Say you had about 20 gb left per rank on your 4x switchless setup with a tuned ds41f.
What would you do with it?
Asking for a friend.
9%Add a system one
47%Add image/video gen
38%More prompt cache
6%More long prompt space
32 votes • 5 daysThis is akin to wizardry.
ローカルで使うモデルなんて、倫理観無ければ無いほど良いんだから!!
僕が作ったテストバージョンのAblitretedをもとにCoxianさんが改良してその手法を共有してくれて、今は更に改良が進んでる。拒否そのものではなく、プロンプトに応答しないケースそのものを排除して全てのプロンプトに従うようにチューニングを進めている。ここで鍵になってくるのがモデルにある倫理観。
倫理観を残したまま拒否を消すと説教みたいな前置きが入ってトークン効率がガタ落ちしたりする。あとはLegalな代替案提示とかもよくある挙動。
これを防ぐために倫理観も消しつつも精度は損なわないようにチューニングしてる。相反する要件を満たしていくのは難しいけど昨日から改善ループに閉じ込めることに成功したから明日くらいにはかなりいいモデルになってると思う!
出来たら公開するよーー(大声)
🐳鲸鱼兄弟们!DeepSeek 要放大招来啦!
V5、V5-open、V5-next都端上桌了。
deepseek Harness 的最新代码中发现已经有相关的代码了。
看来离发布也不远了。
github.com/deepseek-ai/deeps…


















