Log inSign up
Log inSign up
TNG Technology Consulting GmbH
1,752 posts
TNG Technology Consulting GmbH profile banner
@tngtech

TNG Technology Consulting GmbH

@tngtech
TNG, aka "The Nerd Group", is a consulting partnership focused on high end information technology, particularly AI. 936 employees, 99.9% academics, ~53% PhDs.
Unterföhring, Deutschland
tngtech.com/en
Joined December 2010
177 Following
2,257 Followers
RepliesRepliesRepostsRepostsMediaMediaArticlesArticles

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • Pinned
    @tngtech
    TNG Technology Consulting GmbH
    @tngtech
    Aug 13
    Sleeper Agent and How to Tame Them
    Article cover image
    Article
    Sleeper Agents and How to Tame Them
    First, they learn what to look for. Then, what to do when they find it. Lastly, how to hide it. TL;DR: Can an LLM be manipulated to carry a hidden, malicious payload? Unfortunately, yes: We trained a...
    1
  • @tngtech
    TNG Technology Consulting GmbH
    @tngtech
    Sep 16
    We think that @OpenRouter's "Union Alpha" is a brain of brains, an AutoRouter. First of all, the riddle pattern echoes "Say 'friend' and enter", as it contains "union" in its name and it behaves like one: When faced with probing prompts that bypass caching... - the timeouts
    4
  • @tngtech
    TNG Technology Consulting GmbH
    @tngtech
    Aug 25
    We get 10 tps with Kimi K3 on our @AIatAMD MI325X*8 node, thanks to the 4bit conversion recipe by @shakudo_io. This is the speed that they also reported. But hey, 10 tps... Does anyone have an idea how to speed this up by A LOT? Else we shard the AMD into 2*4 GPUs and put a
  • @tngtech
    TNG Technology Consulting GmbH
    @tngtech
    Aug 21
    We think that the stealth model "Ox Alpha" on @OpenRouter is a GLM-5.3 variant, as the mean absolute differences (MAD) in their traits are very small. But it appears that not just a vision encoder was added: The three bigger trait differences, namely in "clinical register" (=
    3
  • @tngtech
    TNG Technology Consulting GmbH
    @tngtech
    Aug 19
    Extremely positive results for @Zai_org's GLM-5.3 in @sam_paech's EQ-Bench v4 benchmark: It appears to be the best model by a wide margin, not just compared to the previous open-weights champion K3, but also compared to the best commercial models such as Fable. It's personality
Edit with