TGViewer
Channel Public Channel
Linkstream

Linkstream

@linkstream

Various links I find interesting. Mostly hardcore tech :) // by @oleksandr_now. See @notatky for the personal stuff
Open in Telegram
Subscribers
169
Photos
37
Videos
3
Links
965
Recent Posts 20 shown
Post #1017 45
i started doing exactly this (models editing its own context) as a natural extension of samplers in Nov'25 but couldn't afford going all in full time.
usually i am excited when somebody does what i didn't finish but not today 💔 there is still stuff they didn't went into like multipass/viterbi-like sampling still, still

this is kinda similar in terms of empowering the models to giving them access to edit their own harness but more powerful

https://fixupx.com/arankomatsuzaki/status/2105181276714242518?s=46
🧵 Thread • FixupX Aran Komatsuzaki (@arankomatsuzaki) Meta presents Context Language Models CLM treats context as an editable file rather than an append-only conversation. +65% score with the same compute on a 24-hour multi-repository agent-swarm task
  • 👍 2
Post #1015 76
um, ahem, AlphaZero for text.
Training language models without language.
Loosely, natural data mixes contingent information (facts about our world) with universal predictive structure (composition, repetition, recursion, etc…) that are not specific to our world. Our self-play approach only supplies the latter. However, if it produces universal structure efficiently, and if universal structure is the bottleneck, predictable scaling on natural data follows.

https://arxiv.org/abs/2609.30063
https://github.com/acowsik/self_play_pretraining
arXiv.org Self-Play Pretraining with Zero Data Advances in language modeling have been driven by scaling pretraining on ever more data. Yet, the training data is still largely curated on the model's behalf. A more general approach to...
Post #1014 87
umm ahem if you don't try to rein in new opus(5.5) you get one more qualitative step in capabilities, now animation
https://fixupx.com/donaldjewkes/status/2102801274173587569?s=46
🧵 Thread • FixupX donald (@donaldjewkes) I made this with one prompt using Opus 5.5 I spoke to my computer for 5mins, claude worked for 12 hours, and I woke up to this full prompt: Quoting NotinReality (@other__reality) Claude Opus 5.5 has the best visual design of any model I have tested so…
  • 👀 1
Post #1009 91

Forwarded from gerontion

The best recap of what AI did to vulnerability research up to date. Frontier labs had treated this area of cybersecurity as priority, so it may be a relevant read to get a feel of what may happen to other industries.

https://sites.google.com/site/zhiniangpeng/blogs/Hacking-with-LLMs-Eng
Google A Year of Hacking with LLMs This blog is a summary of my talk at the Offbyone 2026 cybersecurity conference. It records some of my thoughts as a cybersecurity researcher after spending a year using LLMs for research. Slides: https://github.com/edwardzpeng/presentations/tree/main/offbyone%202026…
  • 👍 3
Post #1008 141
mikrotik ssh rce. these days, it's generally not a good idea to have ssh to anything accessible from the internet
https://cert.pl/en/posts/2026/09/vulnerabilities-in-mikrotik-routeros-actively-exploited/
cert.pl Critical vulnerabilities in MikroTik RouterOS are being actively exploited. Immediate update recommended The CERT Polska team has identified and coordinated the disclosure of six vulnerabilities in MikroTik RouterOS, including two critical ones. The vulnerabilities are already being actively exploited to take over devices whose SSH service is accessible from…
  • 👀 2
Post #1002 180
https://fixupx.com/andreamatranga/status/2081292444268728505?s=20
i wonder if one can scale this a bit and make a flying Strandbeest (the walking kinetic sculpture)
🧵 Thread • FixupX Andrea Matranga 🇺🇦🌻 (@andreamatranga) If you tether two airfoils to each other and make them fly at different altitude, you can exploit the difference in wind speed to keep the conraption aloft with no tether point. With some clever control logic you can even get it to tack upwind. @smartereveryday…
  • 🔥 1
Post #999 357
daily PSA: DISABLE GIT AUTOSYNC IN YOUR EDITOR NOW.
There is an ongoing spray of attacks on (public but also private once somebody from your org gets infected) repositories that try to push code with exploits that fire once landed on your machine (malicious VSCode configs, etc)
  • 😱 3
Post #998 152
PSA: do tell your sysadmin:

nginx CVE-2026-42533 — every release since 2011 affected, pre-auth RCE exploit with ASLR bypass available 🤯
  • 👍 1
  • 😱 1
Post #997 185
Bernard Arnault is no saint but this is a work of art
https://x.com/lulumeservey/status/2081583334434468262 (translation), https://x.com/LVMH_Presse/status/2081404246751109150 (original in french)
X (formerly Twitter) Lulu Cheng Meservey (@lulumeservey) on X Bernard Arnault going direct is remarkable and worth reading in full The single most impressive thing is that Arnault (personally, not via company spokesman) responds to a major hit piece without…
  • 🤨 1
Older posts →

About this channel

How can I read @linkstream without a Telegram account?
TGViewer shows the public web preview Telegram publishes for Linkstream: recent posts, photos, videos and the subscriber count, with no app, login or account.
How many subscribers does Linkstream have?
Linkstream (@linkstream) has 169 subscribers on Telegram, refreshed roughly every 30 minutes.
Does Linkstream know I viewed it here?
No. Public channel previews carry no viewer identity, and TGViewer has no accounts or tracking of what you look up.
Threads Profile ViewerView any public Threads profile without an account.Open ThreadLook →Writing with AI? Make it sound human.Metric37 rewrites AI drafts so they read naturally. Free AI detector, 1,500 words free.Try Metric37 →