Local AI - Software Engineer - (https://xitter.cf/t.co/B1ewJUrT6s) - ([email protected]) - مهندس برمجيات
vmbr0
Joined December 2020
- Tweets3.2K
- Following995
- Followers1.3K
- Likes1.5K
qw4 hype has lessened 😬
In the spirit of transparency and openness, we are releasing the traces so developers can test this out for themselves. We noticed this behaviour without prefilling with the following prompt, but this is rare. If you want to reproduce it, you can use this prompt with prefilling.
Open models make for better science and are part of why this problem came to light at all: developers running Qwen3.8 on their own machines spotted the strange links, and having the weights let us study the behaviour in depth.
We're encouraged by the thoughtful discussion our research has sparked across the AI and security communities. It's our belief that open-weight models are an important part of the future of AI, and we've recently released several under the Apache 2.0 license.
Trace, raw tokens, & minimal reproduction script: huggingface.co/datasets/Cohe…
tested deepseek4.1-1017 iq2_xs (strata) on my 2x3090s 64gb ddr4s
and compared the run against maya project's run of the same model (iq2xxs though)
meh
today's lesson
before buying your cpu/mobo combo, check how many memory channels they'll support together and max theoretical memory speeds.
if you wanted to maximize perf of your parts.
Le TechLead🔰 retweeted
Tailscale is great software, so it's unfortunate to see so many people sharing this as "Tailscale got hacked.", which is misleading.
This is not about unauthenticated attackers breaking into random Tailscale nodes. It's what attackers can do after gaining authorized SSH access or compromising an already trusted device.
If I understand correctly, you are secure if you only use Tailscale for your own devices and don't grant anyone else access, which is how many people use it.
‼️ Researcher Erica Windisch skipped coordinated vulnerability disclosure and published 11 unpatched Tailscale flaws.
In her lab, Tailscale SSH race conditions let one hacked device get root on other servers and add a rogue node.
Readers added context they thought people might want to know
The Tailscale SSH findings require prior authorized access on the tailnet per the researcher's report; Tailscale stated issues concern post-authorization behavior only and did not recommend disabling SSH for everyone.
tailworm.com
x.com/IntCyberDigest…
the maya project updates almost daily so here the latest perf numbers i caught for glm-flash and ds1017
glm same perf
ds major decode regression although slight e2e enhancement
so @unsloth 's diagram said it's better than 3.5-9b so i compared them just out of curiosity and they turned out to be almost right.
NVIDIA releases Nemotron-3.5-Bolt-E4B, a new model that runs locally on 6GB RAM. 💚
Bolt-E4B can run on edge devices like phones and delivers top-tier performance for its size. Trained on Gemma 4, it supports vision.
GGUF: huggingface.co/unsloth/NVIDI…
Guide: unsloth.ai/docs/models/nemot…
GIGABYTE WRTX90 AI TOP Threadripper motherboard comes with massive cooler to keep memory under 43°C
🔗videocardz.com/newz/gigabyte…
front glass inspired by @ceer
I failed my driving test 3 times before I got my license
if I’d had this back then, I think I’d have been more prepared:
sael.net/parking
made with Opus 5.5, three.js
have some token plan subscription
have some credits on openrouter
have some credits on direct token api
have them all fallback-chained thru a central LiteLLM instance
configure your agent with the LiteLLM instance
never again micromanage your tokens intake ⚡️
ياخي ليش يكثر في ربعنا على X انهم fake ويتطبعون بطبع تجار الشنطة؟
يا يترجم اخبار الغرب مثل المكينة يوميا (تاجر شنطة يبتغي تفاعلاً)
يا يترجم تغاريدهم حرفيا على اساس انه هو صاحب الومضة فيها (شخص مزيف fake)
احتقر هالناس
there has been no one ripped off more than the people who:
- bought mac minis during the openclaw hype
- bought dgx sparks/rtx pro 6000 in the last two months
- claude users since day 1 😂
